Daniel Steffens MECHANISMS of LOW BACK PAIN Belo Horizonte

Daniel Steffens

MECHANISMS OF LOW BACK PAIN

Belo Horizonte 2015 Daniel Steffens

MECHANISMS OF LOW BACK PAIN

Tese apresentada ao Programa de

Pósgraduação em Ciências da Reabilitação, da Escola de Educação Física, Fisioterapia e Terapia Ocupacional da Universidade Federal de Minas Gerais como requisito à obtenção do título de Doutor em Ciências da Reabilitação

O rientadora: Profª. Drª. Leani SM Pereira Co -Orientador: Prof. Dr. Chris Maher Co -Orientadora: Profª. Drª. Jane Latimer

Belo Horizonte 2015

S817 Steffens, Daniel 2015 Mechanisms of low back pain. [manuscrito]/.Daniel Steffens – 2015. 245f., enc.: il.

Orientadora: Leani Souza Máximo Pereira Co-orientador: Chris Maher Co-orientadora: Jane Latimer

Tese (doutorado) – Universidade Federal de Minas Gerais, Escola de Educação Física, Fisioterapia e Terapia Ocupacional.

1. Aptidão física - Teses. 2. Dor Lombar - Teses. 3. Incapacidade - Teses. I. Pereira, Leani Souza Máximo. II. Maher, Chris. III. Latimer, Jale. IV. Universidade Federal de Minas Gerais. Escola de Educação Física, Fisioterapia e Terapia Ocupacional. V. Título.

CDU: 612.76 Ficha catalográfica elaborada pela equipe de bibliotecários da Biblioteca da Escola de Educação Física, Fisioterapia e Terapia Ocupacional da Universidade Federal de Minas Gerais.

Supervisors’ Statement

As supervisors of Daniel Steffens’ doctoral work, we certify that we consider his thesis “Mechanisms of Low Back Pain” to be suitable for examination.

Professor Leani de Souza Máximo Pereira

Universidade Federal de Minas Gerais Date: 01.01.2015

Professor Christopher Maher

The George Institute for Global Health Date: 01.01.2015

The University of Sydney

Professor Jane Latimer

The George Institute for Global Health Date: 01.01.2015

The University of Sydney

Candidate’s Statement

I, Daniel Steffens, hereby declare that this submission is my own work and that it contains no material previously published or written by another person except where acknowledged in the text. Nor does it contain material which has been accepted for the award of another degree.

I, Daniel Steffens, understand that if I am awarded a higher degree for my thesis entitled “Mechanisms of low back pain” being lodged herewith for examination, the thesis will be lodged in the Universidade Federal de Minas Gerais and The University of Sydney libraries and be available immediately for use. I agree that the University Librarian (or in the case of a department, the Head of the Department) may supply a photocopy or microform of the thesis to an individual for research or study or to a library.

Daniel Steffens Date: 01.01.2015 iii iv v

Acknowledgments

My thanks go firstly to my supervisors Prof Leani SM Pereira and Prof Chris G Maher and my co-supervisor Prof Jane Latimer. Leani, all this could not be possible without your persistence and dedication. Chris, you are the best supervisor I could have ever asked for. Your competence, dedication, patience and sense of humour made this journey much easier. You are truly an inspiration to all of us. Jane, your kindness and guidance allowed me to always see the light at the end of the tunnel. You were a great co-supervisor for which I have immense respect and admiration.

I would like to acknowledge my appreciation to the Universidade Federal de Minas Gerais and The University of Sydney for establishing the Cotutelle agreement allowing me to pursue my studies. I would also like to thank the George Institute for Global Health for providing the modern infra-structure and support; and the Department of Physiotherapy from the Universidade Federal de Minas Gerais for their support.

I am grateful to several people whose advice and support ensured that this thesis was completed. I am indebted to all my co-authors, who played a major role in the construction of each chapter of this thesis. In particular, I would like to thank Dr Mark Hancock for all his support, dedication, mentorship, transparency and willingness to help. You are one of the reasons I started my PhD.

To all my friends and colleagues, it is impossible to imagine these four years without the BBQs, beers, laughs and advice. In particular, Zamba, Marcinha, Big Mike, Vinicius, Tarci, Bruno, Tie, Patricia, Gustavo, Marina, Amabile, Saad, Aron, Matt and Richard. You have made all the difference.

To all my family, Mum, Dad, brothers and sister, you all have contributed to this. Mum, your unconditional support allowed me to choose my journey. Dad, I miss you and I wish you were here to share this moment with me. Your son will finally become a Doctor. You both were always there for me and I hope I have made you proud.

Finally, to the love of my life. Paula, thank you for your support, encouragement and help at every single stage. This thesis is dedicated to you, the most important person in my life. vi

Publications and Presentations

Parts of the work presented in this thesis have been published and/or presented in the following forms:

Publications

Steffens D, Maher CG, Ferreira ML, Hancock MJ, Glass T, Latimer J. Clinicians’ views on factors that trigger a sudden onset of low back pain. European Spine Journal. 2014; 23:512-519.

Steffens D, Ferreira ML, Latimer J, Ferreira PH, Koes BK, Blyth F, Li Q, Maher CG. What triggers an episode of low back pain? A case-crossover study. Arthritis Care & Research. 2015; 67:403-410.

Steffens D, Maher CG, Li Q, Ferreira ML, Pereira LSM, Koes BK, Latimer J. Effect of weather on back pain: results from a case-crossover study. Arthritis Care & Research. 2014; 66:1867-1872.

Steffens D, Hancock MJ, Maher CG, Williams C, Jensen TS, Latimer J. Does magnetic resonance imaging predict future low back pain? A systematic review. European Journal of Pain. 2014; 18:755-765.

Steffens D, Hancock MJ, Maher CG, Latimer J, Satchell R, Ferreira ML, Ferreira PH, Partington M, Bouvier AL. Prognosis of chronic low back pain in patients presenting to a private community-based group exercise program. European Spine Journal. 2014; 23: 113-119.

Steffens D, Hancock MJ, Pereira LSM, Kent PM, Latimer J, Maher CG. Do magnetic resonance imaging findings identify patients with low back pain who respond better to particular interventions? A systematic review. Submitted for publication to European Journal of Pain on 28th October 2014.

Steffens D, Maher CG, Ferreira ML, Hancock MJ, Pereira LSM, Williams CM, Latimer J. Influence of clinician characteristics and operational factors on recruitment of participants with low back pain: an observational study. Journal of Manipulative Physiological Therapeutics. 2014; 38:151-158. vii

Steffens D, Ferreira ML, Maher CG, Latimer J, Koes BW, Blyth FM, Ferreira PH. Triggers for an episode of sudden onset low back pain: study protocol. BMC Musculoskeletal Disorders. 2012; 24:13-17.

Presentations

Steffens D, Ferreira ML, Latimer J, Ferreira PH, Koes BK, Blyth F, Li Q, Maher CG. What triggers an episode of low back pain? Results of a Case-crossover study. XIII International Back Pain Forum. Campos do Jordão, Brazil, 2014.

Steffens D, Maher CG, Li Q, Ferreira ML, Pereira LSM, Koes BK, Latimer J. Could the weather trigger an episode of acute low back pain? A case cross-over study. XIII International Back Pain Forum. Campos do Jordão, Brazil, 2014.

Steffens D, Ferreira ML, Latimer J, Ferreira PH, Koes BK, Blyth F, Li Q, Maher CG. What triggers an episode of low back pain? Results of a case-crossover study. Australian Pain Society, 34th Annual Meeting. Hobart, Australia, 2014.

Steffens D, Hancock MJ, Maher CG, Williams C, Jensen TS, Latimer J. Does magnetic resonance imaging predict future low back pain? A systematic review. VIII Pain in Europe. Florence, Italy, 2013.

Steffens D, Ferreira ML, Maher CG, Latimer J, Koes BW, Blyth FM, Ferreira PH. Does the method of training of recruiting clinicians influence recruitment to a low back pain case-crossover study. Primary Care Research on Back Pain - XII Odense International Forum. Odense, Denmark, 2012.

Steffens D, Ferreira ML, Maher CG, Latimer J, Koes BW, Blyth F, Ferreira PH. Clinician’s views on triggers for sudden onset low back pain. Primary Care Research on Back Pain - XII Odense International Forum. Odense, Denmark, 2012.

Steffens D, Hancock MJ, Satchill R, Ferreira ML, Ferreira PH, Maher CG, Partington M, Bouvier AL. Prognosis of patients with chronic low back pain presenting to a private functional group exercise program. Australian Physiotherapy Association Biennial Conference. Brisbane, Australia, 2011.

viii

Preface

This thesis is part of a Cotutelle Agreement between the Universidade Federal de Minas Gerais and The University of Sydney made on 1 August 2011.

This thesis is arranged in Ten Chapters, written so that each chapter can be read independently. The Universidade Federal de Minas Gerais and The University of Sydney allow for published papers arising from the candidature to be included in the thesis.

Chapter One is an introduction to the thesis and provides an overview of the relevant low back pain literature with a specific focus on aspects related to the mechanisms of back pain.

Chapter Two is an observational study conducted to describe short- and long-term factors that primary care clinicians consider important in triggering a sudden episode of acute low back pain. This study is presented as published in the European Spine Journal.

Chapter Three is the protocol for the case-crossover study presented in Chapter Four. This protocol is presented as published in BMC Musculoskeletal Disorders.

Chapter Four investigates the increase in risk of an episode of sudden onset, acute low back pain associated with transient exposure to a range of physical (e.g manual tasks, vigorous physical activity) and psychosocial factors (e.g. distraction/fatigue). The paper is presented in the format required by Arthritis Care & Research where it has been accepted for publication.

Chapter Five is a case-crossover study evaluating the influence of various weather conditions on risk of an episode of sudden onset, acute low back pain. This study is presented as published in Arthritis Care & Research.

Chapter Six is a systematic review investigating whether magnetic resonance imaging findings of the lumbar spine predict future low back pain. This study is presented as published in the European Journal of Pain.

Chapter Seven is an observational study investigating the prognosis and prognostic factors for patients with chronic low back pain presenting to a private, community-based, group exercise program. This study is presented as published in the European Spine Journal. ix

Chapter Eight consists of a systematic review investigating if the presence of magnetic resonance imaging findings at baseline identifies patients with low back pain who respond better to particular interventions. The paper is presented in the format required by European Journal of Pain where it was submitted for publication.

Chapter Nine is an observational study conducted to identify factors that influence recruitment to a large observational study. The paper is presented in the format required by Journal of Manipulative and Physiological Therapeutics where it has been accepted for publication.

Finally, Chapter Ten consists of an overview, and discusses the clinical implications and directions for further research.

Each chapter contains its own reference list. Appendices that were published as online supplementary material are included at the end of the relevant chapter. Ethical approval was obtained from the Human Research Ethics Committee of the University of Sydney for all studies prior to commencement. x

Resumo A dor lombar é um grande problema a nível mundial e está associada a elevados custos socioeconômicos e de saúde para o indivíduo e para a sociedade. Nas últimas décadas, apesar do número considerável de pesquisas sobre o tema, o avanço na identificação de abordagens que forneçam resultados significativos no tratamento da dor lombar ainda permanece limitado. A identificação ou a melhor compreensão dos fatores de risco que aumentam a ocorrência da dor lombar, ou que estejam associados com o prognóstico ou a resposta de tratamento, são cruciais para o desenvolvimento de estratégias para sua prevenção. Os estudos desenvolvidos nesta tese focam na avaliação dos mecanismos do desenvolvimento da dor lombar, em relação ao risco, prognóstico e a resposta terapêutica. Até o momento, evidências sobre as causas da dor lombar na população que procura o atendimento em serviços de atenção primária à saúde permanecem limitadas. Os profissionais que atuam em cuidados primários, que comumente atendem um grande número desses pacientes, ainda não conseguem fornecer informações esclarecedoras sobre as causas do aparecimento da dor lombar. O estudo observacional apresentado no Capítulo Dois investigou os fatores de risco de curto e longo prazo que os profissionais em cuidados primários acreditam ser responsáveis por desencadear um episódio súbito de dor lombar aguda. Este estudo baseou- se nas opiniões de 103 profissionais em cuidados de saúde primários que estavam recrutando participantes para um estudo caso-cruzado. Os fatores de risco a curto e a longo prazo mais citados como responsáveis por desencadear um episódio súbito de dor lombar aguda foram fatores biomecânicos (89,3% e 54,2%, respectivamente) e características individuais, tais como episódios anteriores de dor lombar (6,4%e 39,0%,respectivamente). Surpreendentemente, fatores de risco psicológicos/psicossociais e genéticos não foram considerados por esses profissionais, como importantes para desencadeamento de um episódio súbito de dor lombar aguda, apesar da literatura atual considerar esses fatores como relevantes. Os resultados apresentados no Capitulo 2 ajudará a informar aos profissionais de saúde do setor primário sobre a importância de considerar esses fatores em programas de prevenção e tratamento da dor lombar. Uma melhor compreensão dos fatores que aumentam o risco de dor lombar é crucial para o desenvolvimento de estratégias de prevenção. Estudos anteriores concentraram-se em fatores de risco que não são modificáveis tais como a idade dos

indivíduos ou, envolvem hábitos de vida a longo prazo como o tabagismo. Um estudo caso-cruzado para investigar quais os fatores de risco físicos e psicossociais que estavam presentes no período mais próximo para o aparecimento de um episódio súbito de dor lombar aguda foi realizado. Os Capítulos Três e Quatro descrevem o protocolo e os resultados encontrados no estudo, respectivamente. Um total de 999 pacientes que apresentaram um novo episódio de dor lombar aguda foram recrutados por meio de 300 clínicas de atenção primária á saúde em Sidney, Austrália. A exposição a fatores desencadeadores físicos e psicossociais durante as duas horas que precederam o início da dor foi comparada com as mesmas duas horas em dois períodos controles, de 24 e 48 horas antes do surgimento da dor. A exposição à tarefas manuais envolvendo má postura (odds ratio 8,03; intervalo de confiança de 95%: 5,46 a 11,82), manuseio de objetos longe do corpo (odds ratio 6,20; intervalo de confiança de 95%: 2,41 a 15,94), manuseio de pessoas ou animais vivos (odds ratio 5,8; intervalo de confiança de 95%: 2,25 a 14,98) ou manuseio de carga instável, difícil de pegar ou segurar (odds ratio 5,13; intervalo de confiança de 95%:2,40 a 10,93), bem como distrair-se durante uma tarefa (odds ratio 25,0; intervalo deconfiança de 95%: 3,4 a 184,5), ou estar cansado (odds ratio 3,7; intervalo de confiança de95%: 2,2 a 6,3), aumentou significativamente as chances do surgimento de um novo episódio de dor lombar. O consumo de bebidas alcoólicas (odds ratio 1,5; intervalo de confiança de 95%: 0,6 a 3,7) ou a realização de atividade sexual (odds ratio 0,7; intervalo de confiança de 95%: 0,3 a 1,8), não aumentaram o risco no surgimento de dor lombar. Estas associações não foram moderadas por atividade física habitual, índice de massa corporal, episódios anteriores de dor lombar, ou depressão e ansiedade. A idade dos indivíduos moderou o risco associado à exposição à cargas pesadas (odds ratio para pessoas com idades entre 20, 40 ou 60 anos foram de 13,6, 6,0 e 2,7, respectivamente) e de atividade sexual (odds ratio para pessoas com idades entre 20, 40 ou 60 anos foram 0,05, 0,41 e 3,21, respectivamente). Este foi o primeiro estudo a usar o desenho caso-cruzado para avaliar a associação entre exposições físicas e psicossociais e o risco do surgimento de dor lombar aguda. Os resultados apresentados podem ser usados para auxiliar no desenvolvimento de novas abordagens de prevenção dor lombar. Existem muitos fatores responsáveis por desencadear um episódio de dor lombar. Muitos pacientes com dor musculoesquelética comumente relatam que seus sintomas são influenciados pelo clima, no entanto, até o momento esta associação não foi avaliada

xii

para a condição musculoesquelética mais comum, que é a dor lombar. O estudo relatado no Capítulo Cinco teve como objetivo investigar a influência de diferentes condições climáticas no risco de desencadeamento de um episódio de dor lombar aguda. Foi realizado um estudo caso-cruzado envolvendo 993 pacientes que apresentaram-se à clínicas de atenção primária a saúde em Sidney, Austrália. As informações clínicas e demográficas de todos os participantes foram obtidas a partir de uma entrevista. Os parâmetros meteorológicos foram obtidos por meio do Instituto Australiano de Meteorologia. Utilizou-se o método analítico em pares (regr essão logística condicional) para o estudo caso-cruzado para contrastar o clima no momento em que os participantes notaram pela primeira vez a dor lombar (janela caso) com o tempo, ao mesmo tempo, uma semana e um mês antes (janelas controle). A maioria dos participantes eram do sexo masculino (54,2%), com idade média de 45,2 anos. Temperatura, umidade relativa do ar, pressão atmosférica, direção do vento e precipitação não mostraram associação com o início de um novo episódio de dor lombar. Velocidade do vento (odds ratio 1,17; intervalo de confiança de 95%: 1,04 a 1,32; p = 0,01; para um aumento de 11 km/h) e rajada de vento (odds ratio 1,14; intervalo de confiança de 95%:1,02 a 1,28; p = 0,02; para um aumento de 14 km/h) aumentaram as chances para o surgimento da dor. A maioria dos parâmetros climáticos não foram associados com o surgimento da dor lombar. A velocidade do vento e da rajada do vento foram associadas a um pequeno aumento no risco do surgimento de dor lombar, porém essa variável não é considerada como sendo clinicamente importante. A ressonância magnética tem sido muito utilizada pelos médicos para a identificação de patologias responsáveis pela dor lombar. No entanto, a importância dos achados na ressonância magnética permanece controversa, a evidência é limitada e nenhuma revisão sistemática sobre este tema foi conduzida até o presente momento. Portanto, a revisão sistemática apresentada no Capítulo Seis investigou se os achados de ressonância magnética da coluna lombar podem prever o aparecimento de futuros casos de dor lombar em pessoas com e sem dor lombar presente. As buscas foram realizadas em três bancos de dados internacionais (MEDLINE, EMBASE e CINAHL) e 12 estudos de coorte prospectivos foram encontrados. Devido à heterogeneidade dos estudos incluídos, a condução de uma meta-análise não foi possível. Não foram identificadas associações consistentes entre os achados na ressonância magnética e dor ou incapacidade. Três estudos relataram associações significativas para Modic changes tipo 1 com dor (odds ratio 6,2; intervalo de confiança de 95%: 1,9 a 20,2),

xiii

degeneração discal com incapacidade em amostras com a presença de dor lombar (odds ratio 2,2; intervalo de confiança de 95%: 1,2 a 4,0) e hérnia de disco com dor em uma amostra mista, com e sem dor lombar (odds ratio 0,2; intervalo de confiança de 95%: não reportado; p = 0,01). Poucos estudos investigaram se a ressonância magnética pode prever o aparecimento de novos casos de dor lombar. Embora três resultados estatisticamente significativos tenham sido encontrados, esses estudos apenas fornecem evidências limitadas. Há uma clara necessidade de envidar esforços, para que futuros estudos sejam conduzidos adequadamente nesta área. O prognóstico da dor lombar crônica é considerado desfavorável. Uma melhor compreensão dos fatores prognósticos para os pacientes com dor lombar crônica poderia ajudar os profissionais de saúde a tratar e educar os pacientes em relação a essa disfunção. O estudo prospectivo apresentado no Capítulo Sete teve como objetivo analisar o prognóstico e os fatores prognósticos para pacientes com dor lombar crônica que se apresentaram a uma clinica de fisioterapia privada para realização de exercícios em grupo. Este estudo baseou-se em um coorte de 118 pacientes consecutivos com dor lombar crônica atendidos em uma clínica de atenção primária. Análises de regressão linear múltipla foram realizadas para investigar se uma série de variáveis prognósticas (por exemplo, número de episódios anteriores) podem prever dor e incapacidade em 12 meses de acompanhamento. A maioria dos pacientes (95%), foram acompanhados no decorrer dos 12 meses. Aos 3 e 6 meses de acompanhamento dos pacientes, a intensidade da dor, o incômodo causado pela dor, incapacidade e função apresentaram melhoras semelhantes. No entanto, de 6 a 12 meses, a incapacidade e a função continuaram a melhorar, enquanto um progresso marginal foi observado no incômodo causado pela dor e na intensidade da dor. Os modelos finais mostraram evidências de uma associação entre a intensidade da dor avaliada na linha de base com a intensidade da dor aos 12 meses; enquanto que a duração do episódio atual, incapacidade avaliada na linha de base e nível educacional foram associados com a incapacidade aos 12 meses. A maior parte da variância no resultado não foi explicada pelos preditores investigados neste estudo. Nossos resultados sugerem que, na população estudada, a previsão dos resultados aos 12 meses pode ser mais difícil, ou outros preditores não investigados devam ser considerados. Nos últimos anos, pouco ou nenhum progresso ocorreu na identificação de estratégias de intervenção eficazes para a dor lombar. Este fato pode ser explicado pela dificuldade na identificação de uma causa específica de dor lombar na maioria das

xiv

pessoas. Como resultado, uma única intervenção é oferecida para distintos grupos de pacientes com causas de dor potencialmente diferentes. A ressonância magnética tem a capacidade de revelar uma série de degenerações e outras anormalidades anatômicas que acometem a coluna lombo-sacra. No entanto, pouco tem-se focado na identificação de subgrupos baseados em possíveis causas patoanatômicas da dor lombar. Assim, a revisão sistemática apresentada no Capítulo Oito investigou se a presença de achados de ressonância magnética tem a capacidade de identificar pacientes com dor lombar que respondem melhor a certas intervenções terapeuticas. Esta revisão conduziu buscas em três bancos de dados (MEDLINE, EMBASE e CENTRAL) e incluiu ensaios clínicos randomizados que investigaram achados na ressonância magnética como modificadores do efeito de tratamento para pacientes com dor lombar ou ciática. Baseado nos resultados de oito estudos que investigaram interações em 38 subgrupos para combinações de diferentes achados de ressonância magnética, intervenções e resultados, apenas dois subgrupos apresentaram uma interação significativa. Atualmente, existe uma carência de estudos para determinar se os achados de uma ressonância magnética podem modificar o efeito do tratamento para dor lombar ou ciática. A dificuldade de recrutamento dos pacientes para participar em pesquisas na atenção primária á saúde é um fator importante que impede a realização de estudos contendo um adequado e representativo tamanho amostral. Há uma carência de estudos que investigam quais seriam as barreiras e facilitadores para aumentar o recrutamento de pacientes pelos profissionais para participar de estudos observacionais em cuidados primários à saúde. O estudo final desta tese é apresentado no Capítulo Nove e teve como objetivo investigar os fatores associados ao recrutamento de participantes para um estudo observacional. O estudo incluiu uma amostra de 138 profissionais em cuidados primários que identificaram 1.585 pacientes de outubro de 2011 a novembro de 2012. Os dados foram analisados através de uma regressão binomial negativa multivariada para determinar as associações de uma variedade de características clínicas e fatores operacionais, como a taxa de recrutamento. Os profissionais em cuidados primários recrutaram 951 participantes, a uma taxa de 0,99 participantes por mês. Dois fatores operacionais (profissionais que escolheram receber treinamento pelo telefone e o número de pacientes não elegíveis referidos ao estudo) e um fator de clínico (profissionais que não eram membros de sua respectiva associação) foram associados com a taxa de recrutamento. O tamanho amostral necessário para o estudo foi atingido dentro de um prazo razoável. No entanto, os fatores operacionais e clínicos associados à taxa de

recrutamento parecem limitados. Os estudos apresentados nesta tese contribuem para a melhor compreensão dos mecanismos responsáveis pela dor lombar. Foi constatado que fatores de risco biomecânicos e individuais foram considerados pelos profissionais em cuidados primários como sendo fatores importantes para o desenvolvimento de dor lombar aguda e devem ser investigados em estudos futuros. A exposição a uma variedade de fatores físicos e psicossociais aumentou o risco de um episódio de dor lombar aguda e devem ser considerados. Parâmetros meteorológicos têm pouco efeito no surgimento de um episódio de dor lombar aguda. Apesar de a velocidade do vento e de rajadas de vento mostrarem ter um efeito pequeno, a magnitude do aumento não foi clinicamente relevante. Esses resultados contribuem para o conhecimento sobre os fatores de risco da dor lombar e podem orientar futuras estratégias de prevenção. Poucos estudos investigaram a associação entre os achados da ressonância magnética e dor lombar, portanto, futuros estudos nesta área se fazem necessários. O prognóstico de pacientes com dor lombar crônica que se apresentaram a uma clínica de fisioterapia privada para realização de exercícios em grupo é favorável. Esta informação é importante tanto para profissionais da área da saúde quanto para pacientes, uma vez que contribui para uma expectativa realista, podendo ser usado para orientar a tomada de decisões sobre a necessidade de implementação de intervenções adicionais. Além disso, a necessidade de estudos de alta qualidade, com tamanho amostral adequados, investigando achados da ressonância magnética como modificadores de efeito de tratamento é essencial para determinar a importância clínica destes resultados na dor lombar e ciática. Finalmente, esta tese fornece evidências de que é possível recrutar um grande número de participantes para estudos observacionais. No entanto, a identificação de fatores que possam estimular o recrutamento ainda permanece obscura.

Palavras-chave: Dor Lombar. Fatores de Risco. Mecanismo. Incapacidade. Ressonância Magnética.

xvi

Abstract

Low back pain is a very common condition that presents a significant burden to individuals and society, being one of the leading causes of disability globally. Despite considerable research over the past decades, there has been limited progress in identifying management approaches that provide large treatment effects. Progress in identifying effective prevention strategies has been no better. The identification, or better understanding, of factors that increase the risk for the development of back pain, or that are associated with prognosis or response to treatment, is crucial for developing and refining prevention and management strategies. The studies in this thesis focus on evaluating the mechanisms of low back pain in relation to risk, prognosis and response to treatment.

There is limited information about the causes of low back pain in the general population presenting to care. Primary care clinicians who commonly manage a large number of these patients could provide valuable insights into what may be the most important causes for low back pain. The observational study presented in Chapter Two investigated the short- and long-term risk factors that primary care clinicians believe most likely to trigger a sudden episode of acute low back pain. This study was based on the views of 103 primary care clinicians recruiting patients for a large case-crossover study. The most endorsed short- and long-term risk factors to trigger an episode of sudden acute low back pain were biomechanical (89.3% and 54.2%, respectively) and traits of the individual such as previous low back pain episodes (6.4% and 39.0%, respectively). Surprisingly, psychological/psychosocial and genetic risk factors were not considered important risk factors for an episode of low back pain. Primary care clinicians believe that biomechanical and individual risk factors are the most important factors to trigger an episode of low back pain. The reason why the current literature considers psychosocial and genetic risk factors important and primary care clinicians do not should be further investigated. This information will help inform low back pain management and prevention programs.

A better understanding of what factors increase the risk of back pain is crucial to the development of prevention strategies. Previous studies have focused on factors that are either not modifiable (e.g. age) or involve long-term exposure (e.g. smoking). To evaluate factors more proximal to the pain onset, a case-crossover study investigating a range of physical and psychosocial risk factors for an episode of sudden onset, acute back pain was conducted. Chapter Three and Four describe the study protocol and the report of the study, xvii

respectively. A total of 999 patients with a new episode of acute low back pain were recruited from 300 primary care clinics in Sydney. Exposure to physical and psychosocial triggers during the two hours preceding back pain onset was compared to that in the 2-hour periods 24 and 48 hours before the onset. Exposure to manual tasks involving awkward positioning (odds ratio 8.03; 95% confidence interval 5.46 to 11.82), objects far from the body (odds ratio 6.20; 95% confidence interval 2.41 to 15.94), handling of live people or animals (odds ratio 5.8; 95% confidence interval 2.25 to 14.98) or a load that was unstable, unbalanced or difficult to grasp or hold (odds ratio 5.13; 95% confidence interval 2.40 to 10.93); as well as being distracted during a task (odds ratio 25.0; 95% confidence interval 3.4 to 184.5), or being fatigued (odds ratio 3.7; 95% confidence interval 2.2 to 6.3) significantly increased the odds of a new episode of back pain. Exposure to alcohol consumption (odds ratio 1.5; 95% confidence interval 0.6 to 3.7) or sexual activity (odds ratio 0.7; 95% confidence interval 0.3 to 1.8) did not increase risk of back pain onset. These associations were not moderated by habitual physical activity, body mass index, previous low back pain episodes, or depression and anxiety. Age moderated the risk associated with exposure to heavy loads (odds ratios for people aged 20, 40 or 60 years were 13.6, 6.0 and 2.7 respectively) and sexual activity (odds ratios for people aged 20, 40 or 60 years were 0.05, 0.41 and 3.21 respectively). This is the first study to use this robust design to evaluate the association of physical and psychosocial exposures and low back pain onset. These results can inform the development of new prevention approaches for back pain.

There are many factors believed to trigger an episode of low back pain. Many patients with musculoskeletal pain commonly report that their symptoms are influenced by the weather, but this issue has not been evaluated for the most common musculoskeletal condition, back pain. The study reported in Chapter Five aimed to investigate the influence of various weather conditions on risk of acute low back pain. A case-crossover study was performed with 993 patients presenting to primary care in Sydney. All participants’ demographic and clinical information were obtained from an interview. Weather parameters were obtained from the Australian Bureau of Meteorology. We used the pair-matched analytic approach (conditional logistic regression) for the case-crossover design to contrast the weather at the time when participants first noticed their back pain (case window) with the weather at the same time one week and one month prior (control windows). Most participants were male (54.2%), with mean age of 45.2 years. Temperature, relative humidity, air pressure, wind direction and precipitation showed no association with onset of a new episode xviii

of back pain. Higher wind speed (odds ratio 1.17; 95% CI 1.04 to 1.32; p=0.01; for an increase of 11 km/h) and wind gust (odds ratio 1.14; 95% CI 1.02 to 1.28; p=0.02; for an increase of 14 km/h) increased the odds of pain onset. Most weather parameters were not associated with the onset of back pain. Higher wind speed and wind gust speed provided a very small increase in risk of back pain onset that does not seem clinically important.

Magnetic resonance imaging has been increasingly used by clinicians in order to identify pathology responsible for low back pain. However, the importance of findings on magnetic resonance imaging remains controversial, the evidence is limited and no systematic review on this topic has been conducted. Therefore, the systematic review presented in Chapter Six investigated whether magnetic resonance imaging findings of the lumbar spine predict future low back pain in different samples with and without low back pain. Searches performed in three international databases (MEDLINE, EMBASE and CINAHL) located 12 prospective cohort studies. Due to heterogeneity of the included studies, pooling the data for a meta-analysis was not possible. No consistent associations between findings on magnetic resonance imaging and pain or disability were identified. Three studies reported significant associations for Modic changes type 1 with pain (odds ratio 6.2; 95% confidence interval 1.9 to 20.2), disc degeneration with disability in samples with current LBP (odds ratio 2.2; 95% confidence interval 1.2 to 4.0) and disc herniation with pain in a mixed sample (odds ratio 0.2; 95% confidence interval not provided; p = 0.01). Few studies have investigated if magnetic resonance imaging findings predict future low back pain. Although there were three statistically significant results, overall these studies only provide limited evidence. There is a clear need for further appropriately designed research.

The prognosis for chronic low back pain is considered poor. Better understanding of prognostic factors for patients with chronic low back pain may help clinicians to manage and educate patients regarding their future health. The prospective study presented in Chapter Seven aimed to examine the prognosis and prognostic factors for patients with chronic low back pain who presented to a private, community-based, group exercise program. This study was based on the data of a cohort of 118 consecutive patients with chronic low back pain in a primary care setting. Multivariate linear regression analyses were performed to investigate whether a range of prognostic variables (e.g. number of previous episodes) predict pain and disability at 12 months follow up. Most of the patients (95%) were followed up at 12 months. At 3 and 6 months, pain intensity, bothersomeness, disability and function improved similarly, but from 6 to 12 months, disability and function continued to improve while only xix

small further changes in bothersomeness and pain intensity occurred. The final models showed evidence of an association between baseline pain intensity and 12 months pain outcomes; whereas duration of current episode, baseline disability and educational level accounted for 12 months disability outcome. Most of the variance in outcome was not explained by any of the predictors we investigated in this study. Our findings suggest that in this population, predicting 12 month outcome may be more difficult or other predictors may be more important.

There has been little or no progress in identifying effective intervention strategies for low back pain. This lack of progress may be explained by the current inability to identify a specific cause in most people. As a result, a single intervention is provided to heterogeneous groups of patients with potentially different causes of their pain. Magnetic resonance imaging can reveal a range of degenerative findings and anatomical abnormalities affecting the lumbosacral spine. However, very little attention has focussed on identifying subgroups based on possible patho-anatomical causes of low back pain. Thus, the systematic review presented in Chapter Eight investigated if the presence of magnetic resonance imaging findings identifies patients with low back pain who respond better to particular interventions. This review conducted a sensitive search in three databases (MEDLINE, EMBASE and CENTRAL) and included randomised controlled trials investigating findings on magnetic resonance imaging as treatment effect modifiers for patients with low back pain or sciatica. Based on data from eight published reports investigating 38 subgroup interactions for combinations of different magnetic resonance imaging findings, interventions and outcomes, two subgroups displayed a significant interaction. At present, there is a lack of studies to determine whether magnetic resonance imaging features modify effect of treatment for low back pain or sciatica.

Patient recruitment to research studies is often difficult. Problematic recruitment in primary care is one major factor preventing the collection of large representative samples for research. Studies investigating factors that enhance recruitment to observational studies in primary care is lacking. The final study of this thesis is presented in Chapter Nine and aimed to investigate factors associated with recruitment of participants to an observational study. The study included a sample of 138 primary care clinicians that screened 1,585 patients from October 2011 to November 2012. Multivariate negative binomial regression was used to determine associations of a variety of clinician characteristics and operational factors with the recruitment rate. Primary care clinicians recruited 951 participants at a rate of 0.99 xx

participants per month. Two operational factors (clinicians who chose to be trained by telephone and number of ineligible patients referred) and one clinician factor (clinicians who were not member of their association) were associated with recruitment rate. We reached our target sample size in a reasonable time frame. However, the operational and clinician factors associated with recruitment rate seem limited.

Overall, the studies described in this thesis have provided an important contribution to better understanding the mechanisms of back pain. Firstly, biomechanical and individual risk factors were considered by primary care clinicians’ to be important triggers for acute low back pain and should be investigated further in future research. Secondly, exposure to a range of physical and psychosocial triggers increased the risk of an acute low back pain episode. Weather parameters have little effect on triggering an episode of acute low back pain. While wind speed and wind gusts were shown to have a small effect the magnitude of the increase was not clinically relevant. These results are important in increasing our knowledge about risk factors for low back pain and guide future prevention strategies. A systematic review of the literature found only a few studies on the association between findings of magnetic resonance imaging and low back pain, and clearly more studies in this area are needed. Thirdly, the prognosis of patients with chronic low back pain presenting to a private, community-based, group exercise program is favourable. This information is important for clinicians and patients as it helps with realistic expectation and can be used to guide decision making regarding the need for additional interventions. In addition, the need for high quality, adequately powered trials investigating magnetic resonance imaging findings as effect modifiers is essential to determine the clinical importance of these findings in low back pain and sciatica. Finally, this thesis provides evidence that it is reasonably easy to recruit large number of participants to observational studies; however, the identification of consistent factors that increase recruitment remains unclear.

Keywords: Low back pain. Risk Factors. Mechanism. Disability. Magnetic Resonance Imaging. xxi

Table of Contents

Supervisors’ statement ...... i Candidate’s statement ...... ii

Approval letter ...... iii

Record ...... iv

Acknowledgements ...... v

Publications and presentations ...... vi

Preface ...... viii

Abstract ...... x

Resumo (Abstract in Portuguese) ...... xvi

Chapter One: Introduction ...... 26

1. Epidemiology of low back pain ...... 27

2. Economic burden of low back pain ...... 27

3. Definition and classification of low back pain ...... 28

4. Mechanisms of low back pain ...... 29

4.1 Risk factors ...... 29

4.1.1. Primary care clinician’s perception on risk factors for low back pain .. 30

4.1.2. Physical and psychosocial risk factors ...... 30

4.1.3. Environmental factors and low back pain ...... 31

4.1.4. Magnetic resonance imaging ...... 32

4.2. Prognostic factors...... 32

4.2.1. Prognostic factors for chronic low back pain ...... 33

4.3. Subgroup of low back pain ...... 33

4.3.1. Magnetic resonance imaging subgrouping ...... 34

5. Recruitment for large observational studies of low back pain ...... 35

5.1. Factors that influence recruitment to observational studies ...... 35

6. Aims of the thesis ...... 36 xxii

7. References ...... 37

Chapter Two: Clinicians’ views on factors that trigger a sudden onset of low back pain ..... 46

Statement of authorship contribution ...... 47

Abstract ...... 48

Introduction ...... 48

Materials and methods ...... 49

Results ...... 51

Discussion ...... 52

Conclusions ...... 53

References ...... 53

Chapter Three: Triggers for an episode of sudden onset low back pain: study protocol ..... 56

Statement of authorship contribution ...... 57

Abstract ...... 58

Background ...... 58

Methods/Design ...... 59

Discussion ...... 61

References ...... 61

Additional file 1: Clinicians’ questionnaire ...... 63

Additional file 2: Study participants’ questionnaire ...... 64

Chapter Four: What triggers an episode of acute low back pain? A case-crossover study . 79

Statement of authorship contribution ...... 80

Abstract ...... 81

Introduction ...... 81

Patients and methods ...... 81

Results ...... 83

Discussion ...... 86

References ...... 88 xxiii

Appendix Table 1 ...... 89

Appendix Table 2 ...... 91

Appendix Table 3 ...... 93

Appendix Table 4 ...... 95

Appendix 5 ...... 97

Chapter Five: Effect of weather on back pain: Results from a case-crossover study ...... 112

Statement of authorship contribution ...... 113

Abstract ...... 114

Introduction ...... 114

Subjects and methods ...... 115

Results ...... 116

Discussion ...... 116

References ...... 119

Chapter Six: Does magnetic resonance imaging predict future low back pain? A systematic review ...... 120

Statement of authorship contribution ...... 121

Abstract ...... 122

Introduction ...... 122

Methods ...... 123

Results ...... 124

Discussion ...... 130

Conclusions ...... 131

References ...... 131

Appendix S1 ...... 133

Appendix S2 ...... 134

Chapter Seven: Prognosis of chronic low back pain in patients presenting to a private community-based group exercise program ...... 136

Statement of authorship contribution ...... 137 xxiv

Abstract ...... 138

Introduction ...... 138

Material and methods ...... 139

Results ...... 140

Discussion ...... 141

Conclusions ...... 142

References ...... 144

Chapter Eight: Do magnetic resonance imaging findings identify patients with low back pain who respond better to particular interventions? A systematic review ...... 145

Statement of authorship contribution ...... 146

Abstract ...... 148

Introduction ...... 149

Methods ...... 151

Results ...... 154

Discussion ...... 159

Conclusions ...... 162

References ...... 163

Figure 1 ...... 168

Table 1 ...... 169

Table 2 ...... 170

Table 3 ...... 172

Appendix S1 ...... 175

Appendix S2 ...... 176

Appendix S3 ...... 177

Appendix S4 ...... 179

Chapter Nine: Clinician characteristics and operational factors have limited influence on participant recruitment in primary care: Results from an observational study ...... 183

Statement of authorship contribution ...... 184 xxv

Abstract ...... 185

Introduction ...... 185

Methods ...... 186

Results ...... 188

Discussion ...... 189

Conclusion ...... 191

References ...... 191

Chapter Ten: Conclusions ...... 193

10.1. Aim ...... 194

10.2. Overview of principal findings ...... 195

10.3. Implications and suggestions for future research ...... 196

10.3.1. Mechanism: Risk factors for low back pain ...... 196

10.3.2. Management: Prognosis and subgroups for low back pain ...... 199

10.3.3. Factors influencing recruitment rate ...... 200

10.4. References ...... 202

Appendix ...... 205

Appendix A: Media coverage of Chapter Five ...... 206

Appendix B: Curriculum Vitae ...... 237

Chapter One

Introduction

1. Epidemiology of low back pain

Low back pain is an extremely common health problem that affects a large part of the population during their lifetime (HOY et al., 2010; HOY et al., 2012). Until 10 years ago, it was largely thought of as a problem of developed Western countries (VOLINN, 1997). However, since that time an increasing number of studies have demonstrated that low back pain is also a major problem in less developed countries (MARIETTE and MARIETTE, 2003; LOUW, MORRIS and GRIMMER-SOMERS, 2007; HOY, BROOKS et al., 2010; BALAGUE et al., 2012). It is estimated that over one year, the incidence of people who have a first-ever episode of low back pain ranged from 6.3% to 15.4%, and the one year incidence of people who have any episode of low back pain (i.e., first-ever or recurrent) ranged from 1.5% to 36% (HOY, BROOKS et al., 2010). The highest prevalence of low back pain is among females and people aged 40 to 80 years. The mean (SD) global prevalence of activity limiting low back pain lasting for more than 1 day is estimated to be 11.9% (2.0) (HOY, BAIN et al., 2012). Although low back pain is not a life threatening condition, it is the leading cause of activity limitation and work absence throughout much of the world, and it causes an enormous economic burden on individuals, families, communities, industry and governments (THELIN, HOLMBERG and THELIN, 2008; FERREIRA et al., 2011; MURRAY et al., 2012).

2. Economic burden of low back pain

The economic burden of low back pain is very large and appears to be rising. It is estimated that the number of visits to allied health professionals exceeds 30 million per year in the United States alone (ANDERSSON, 1999). This high frequency of annual visits leads to both high direct and indirect costs (HOY et al., 2010). Direct costs includes resources spent on assessing and treating low back pain, such as medications, assistive devices, diagnostic tests, and may also include non-medical costs incurred by the patient and family and other public resources (e.g. transportation). Indirect costs commonly include costs related to employment and household productivity (EKMAN, JOHNELL and LIDGREN, 2005; DAGENAIS, CARO and HALDEMAN, 2008).

In the United States, the economic burden for low back pain is estimated to be over US$85 billion a year (DAGENAIS, CARO and HALDEMAN, 2008). In Australia AU$ 9.17

billion per annum is the total cost spent on care for low back pain (WALKER, MULLER and GRANT, 2003). The total annual cost in Europe has been estimated, for instance, €6.4 billion in the Netherlands (LAMBEEK et al., 2011), €6.6 billion in Switzerland (WIESER et al., 2011), ₤12.3 billion in the United Kingdom (HOY, BROOKS et al., 2010) and €1.9 billion in Sweden (EKMAN, JOHNELL and LIDGREN, 2005). Although the economic burden for low back varies among countries, it is apparent that low back pain represents an enormous public health problem worldwide.

3. Definition and classification of low back pain

Low back pain is defined as pain and discomfort, localised below the costal margin and above the inferior gluteal folds, with or without leg pain (VAN TULDER et al., 2006). Diagnostic triage is a simple and practical classification, which has gained international acceptance, by dividing low back pain into three main categories: (i) serious spinal pathologies; (ii) Nerve root pain (radicular pain)/spinal canal stenosis; and (iii) non-specific low back pain (KOES et al., 2010).

Serious spinal pathologies include spinal tumours, vertebral infections, cauda equina syndrome, vertebral fracture and inflammatory diseases such as ankylosing spondylitis (CHOU et al., 2007). Only a minority of patients (less than 1%) with acute low back pain presenting to primary care are diagnosed with a serious spinal pathology (HENSCHKE et al., 2009). When serious spinal pathology is suspected as a cause of low back pain, further diagnostic investigations are usually required. Guidelines recommend the use of red flags to screen for serious pathology and to identify those patients that may need imaging and laboratory test or specialist referral to establish a definitive diagnosis (MAHER et al., 2011).

Nerve root pain (radicular pain) or spinal canal stenosis represent approximately 5% of the low back pain cases and are characterised by presentations where leg pain is dominant (KONSTANTINOU and DUNN, 2008). In approximately 90% of the case of leg dominant pain the condition is caused by a herniated disc with nerve root compression, but lumbar canal or foraminal stenosis and (less often) tumours or cysts are other possible causes (VALAT et al., 2010). While there are a range of definitions of sciatica, the key clinical features that can help clinicians to distinguish it from nonspecific low back pain include unilateral leg pain that is worse than the low back pain, pain radiating below the knee,

presence of numbness or pins and needles in a dermatomal distribution, positive results on a straight leg raise test, and weakness or reflex changes, or both, in a myotomal distribution (KOES, VAN TULDER and PEUL, 2007).

Non-specific low back pain is a term used synonymously with simple low back pain or mechanical low back pain and is defined as low back pain not attributable to a recognisable, known specific pathology (BALAGUE, MANNION et al., 2012). Due to the inability to identify the source of pain, the vast majority of low back pain patients (approximately 90%) fall into the non-specific low back pain category (HANCOCK et al., 2007).

Low back pain is often classified in three stages (acute, sub-acute and chronic) according to its duration and this provides some information to the clinician with regards to treatment and prognosis. Acute low back pain is usually defined as an episode persisting for less than 6 weeks; sub-acute low back pain as low back pain persisting between 6 to 12 weeks and chronic low back pain as low back pain persisting for 12 weeks or longer (VAN TULDER, BECKER et al., 2006).

4. Mechanisms of low back pain

Low back pain may originate from many anatomical structures in the lumbar spine, including bones, intervertebral discs, ligaments, muscles, neural structures and blood vessels (DEYO and WEINSTEIN, 2001). However, the exact source of low back pain is often difficult to identify using conventional tests available in primary care (ANDERSSON, 1999; HANCOCK, MAHER et al., 2007). Research into risk factors for low back pain is often challenging due to heterogeneity across research methods, case definitions and study populations, it is clear that there are a number of environmental and individual factors that influence the onset and course of low back pain (HOY, BROOKS et al., 2010).

4.1. Risk factors

Low back pain is a complex condition with many factors believed to contribute to its onset (LATZA et al., 2000). Investigations into risk factors for low back pain is in a developing stage when compared with other common conditions, such as cardiovascular

disease and cancer (MANEK and MACGREGOR, 2005). Broadly, the risk factors associated with low back pain can be classified as individual, psychosocial, occupational, genetic and biomechanical (HAMBERG-VAN REENEN et al., 2007; LANG et al., 2012; TAYLOR et al., 2014). There are a reasonable number of recognised risk factors for low back pain, however, most of these risk factors are not robust or replicable, and many are not modifiable (TAYLOR, GOODE et al., 2014). Identifying factors that may increase the risk for or predispose individuals to the development of back pain is critical in attempting to reduce the prevalence and ultimately the social impact of this condition (RUBIN, 2007).

4.1.1. Primary care clinician’s perceptions on risk factors for low back pain

Risk factors for the development of low back pain in the general population who present to primary health clinics are poorly understood (VAN TULDER, BECKER et al., 2006), despite the high volume of patients seeking care (DEYO and PHILLIPS, 1996). One reason for this is that most work up-to-date has focused on samples of specific occupational groups or working conditions (BURDORF, NAAKTGEBOREN and DE GROOT, 1993; MURTEZANI et al., 2011; FERGUSON et al., 2012; VANDERGRIFT et al., 2012). Although studies conducted in occupational settings may reveal important risk factors for work-related back pain, these risks may not be relevant to other populations, such as those drawn from primary care. Primary care clinicians who frequently manage patients with low back pain may provide an important understanding into the most common risk factors. The identification of putative risk factors in primary care is essential to strengthen future research and help inform low back pain prevention programs. Chapter Two presents the most likely risk factors involving short and long-term exposure that primary care clinicians believe could trigger a sudden episode of acute low back pain.

4.1.2. Physical and psychosocial risk factors

Many studies have attempted to identify and evaluate the contribution of multiple different demographic, physical, socioeconomic, psychological, and occupational factors to the development of back pain (VINDIGNI et al., 2005; RUBIN, 2007; TAYLOR, GOODE et al., 2014). Known non-modifiable risk factors for low back pain include increasing age, number of children, previous episode of low back pain and major scoliosis. Those that are

modifiable are poor health, obesity, smoking and prolonged sitting (VINDIGNI, WALKER et al., 2005). However, there is a significant limitation in interpreting the literature on risk factors for low back pain. Most of the recognised risk factors were assessed only in single or small studies, with weak and non-reproducible evidence to support a definite association, are not robust or replicable across studies, and were not modifiable (HURWITZ and MORGENSTERN, 1997; KOPEC, SAYRE and ESDAILE, 2004; RUBIN, 2007; TAYLOR, GOODE et al., 2014). Another problem with previous studies on risk factors for low back pain is the focus on long-term exposure. Little is known about short-term exposure to physical and psychological risk e.g. distraction while lifting. Modifying such risk factors may be extremely important in preventing back pain recurrence.

One of the best designs to investigate transient risk factors is the case-crossover design. This design is more efficient than cohort designs because it samples only cases and may be less exposed to selection bias than case-control designs because cases provide their own control data (MACLURE, 1991). One of the main advantages of the case-crossover design is that each case serves as its own control. Consequently, case-crossover studies are not confounded by time-invariant risk factors. Another advantage is that the case-crossover design is immune to one of the main causes of bias in case-control studies: the selection of control that is not representative of the population that produced the cases. A further advantage is the ability to assess short-term reversible exposures (MACLURE, 1991; MACLURE and MITTLEMAN, 2000). Chapter Three represents the published study protocol of a case-crossover study investigating the association between an episode of sudden onset, acute back pain with transient exposure to a range of physical and psychosocial factors. The results of this study are described in Chapter Four.

4.1.3. Environmental factors and low back pain

There are an increasing number of studies examining the association between environmental factors (e.g. temperature, humidity, air pressure, wind and precipitation) and the onset of musculoskeletal conditions. Most of the efforts to measure climatic effects have been directed at individuals with rheumatoid arthritis and chronic pain (JAMISON, ANDERSON and SLATER, 1995; PATBERG and RASKER, 2004; ABASOLO et al., 2013). Patients with musculoskeletal pain (e.g. low back pain) commonly report that certain weather conditions influence their symptoms (SMEDSLUND et al., 2009). Despite the high

frequency with which this belief is reported, there are few robust studies that have investigated this potential association (TENIAS et al., 2009; ABASOLO, TOBIAS et al., 2013). Chapter Five investigates the influence of various weather conditions on risk of low back pain.

4.1.4. Magnetic resonance imaging

Most low back pain sufferers, around 90%, are classified as having non-specific low back pain (VAN TULDER, BECKER et al., 2006), reflecting the inability to identify a clear anatomical source for the pain (WANG, VIDEMAN and BATTIE, 2012). If the source of pain could be identified in at least some of these patients, then it is possible that more logical and effective interventions could be found (HANCOCK, MAHER et al., 2007).

Magnetic resonance imaging is the preferred investigation for most spinal diseases and is increasingly requested for people with low back pain (SHEEHAN, 2010). Various abnormalities can be identified on lumbar magnetic resonance imaging, including disc herniation, nerve root impingement, disc degeneration, and high intensity zone or annular tear (ENDEAN, PALMER and COGGON, 2011). However, the importance of findings on magnetic resonance imaging remains controversial (WASSENAAR et al., 2012). Many studies have documented a high prevalence of disc abnormalities on imaging in asymptomatic subjects (JARVIK et al., 2001). Magnetic resonance imaging findings in currently asymptomatic people may represent markers of early pre-symptomatic disease that is later characterized by episodes of pain and/or disability. Chapter Six reports on a systematic review that sought to investigate whether magnetic resonance imaging findings of the lumbar spine predict future low back pain in different samples with and without low back pain.

4.2. Prognostic factors

Prognosis is a description of the course or prediction of the outcome of a health condition over time (HAYDEN et al., 2010). Important to prognosis is consideration and assessment of characteristics or factors that are associated with or determine the course of a condition. Clinicians may use prognostic information to educate or inform the management

of their patients (CROFT, DUNN and RASPE, 2006). The likely prognosis of low back pain varies according to the duration of symptoms. The prognosis for acute low back pain is inconsistently reported, with estimates of recovery ranging from 39% to 90% (KOES, VAN TULDER et al., 2010; DA et al., 2012). Patients presenting with a longer duration with low back pain or with recurrent low back pain the prognosis may be less favourable (KOES, VAN TULDER et al., 2010).

4.2.1. Prognostic factors for chronic low back pain

For those patients with symptoms persisting for longer than 3 months, approximately one third will recover within 1-2 years after initial onset (VON KORFF et al., 1993; COSTA LDA et al., 2009). According to a recent systematic review, the prognosis of patients suffering from chronic pain is less favourable for those who have taken previous sick leave for low back pain, have high disability levels or high pain intensity at onset of chronic low back pain, have lower education, perceive themselves as having a high risk of persistent pain, and were born outside Australia (COSTA LDA, MAHER et al., 2009). Patients presenting for care in settings where these adverse prognostic factors are uncommon may have a more favourable prognosis than widely reported. Until now no study has investigated the prognosis of people with chronic low back pain attending a private, community-based, group exercise program. Chapter Seven reports the prognosis and prognostic factors for this population.

4.3. Subgroups of low back pain

To date, there are clear trends in recent high quality randomised clinical trials, that show that the sorts of interventions that primary care clinicians have to offer, on average, have small (sometimes insignificant) to moderate effects (LITTLE et al., 2008; LAMB et al., 2010), and often there is little or no difference between the effectiveness of different interventions (CAIRNS, FOSTER and WRIGHT, 2006; JOHNSON et al., 2007). One explanation for this lack of progress may be that a single intervention is usually provided to heterogeneous groups of patients with potentially different causes of their pain. Many clinicians and researchers believe that there are subgroups of people with spinal pain who respond differently to treatment and have different prognosis (KENT, KEATING and BUCHBINDER, 2009). The Identification of subgroups of low back pain patients has been

identified as a key research priority in the field (COSTA LDA et al., 2013), which may lead to improve low back pain patient’s outcome.

Two recent reviews have investigated subgrouping of low back pain treatment. Most previous research in this area has focussed on identifying clinical and psychosocial variables associated with patients who respond better to different interventions (KENT, MJOSUND and PETERSEN, 2010; KENT and KJAER, 2012). However, very little attention has focussed on identifying subgroups based on possible patho-anatomical causes of low back pain.

4.3.1. Magnetic resonance imaging subgrouping

There has been considerable interest in magnetic resonance imaging subgrouping recently (KJAER et al., 2006). Modic changes were first described by Modic et al. 1988 (MODIC et al., 1988; MODIC and ROSS, 2007), who identified three types (Type I, II and III). Based on the histological studies, Type I was characterised by fissured endplates and vascular granulation tissue adjacent to the endplate, whereas Type II was characterised as disruption of the endplates as well as fatty degeneration of the adjacent bone marrow (MODIC et al., 1988). Type III appeared to involve sclerosis of the bone marrow as seen on radiographs (MODIC, MASARYK et al., 1988). Up to date, one review has investigated if Modic changes constitute a specific subgroup of low back pain (JENSEN and LEBOEUF- YDE, 2011). However, the inclusion of single subgroup designs (e.g. studies including all people with Modic changes and no people without Modic changes) prevents a robust evaluation of whether effect modification occurred in this subgroup (KENT et al., 2010).

If subgroups of patients, based on magnetic resonance imaging findings, who respond best to specific interventions could be identified, the potential exists to significantly improve patient outcomes and healthcare system efficiency. Chapter Eight reports on a systematic review that sought to investigate if the presence of magnetic resonance imaging findings identifies patients with low back pain who respond better to particular interventions.

5. Recruitment for large observational studies of low back pain

There is a need for large-scale well-designed studies evaluating risk factors for low back pain. Universally, research studies need sufficient participants to ensure statistical power and validity, but recruitment remains problematic (FAIRHURST and DOWRICK, 1996). Participant recruitment is considered the most difficult aspect of the research process (BLANTON et al., 2006; BOWER, WILSON and MATHERS, 2007). Previous research has reported that problems with recruitment are a major reason for the failure of research studies, leading to wasted research funding (HUNNINGHAKE, DARBY and PROBSTFIELD, 1987). One major challenge to collecting large quality samples of data is efficient recruitment of participants to a study (SPAAR et al., 2009).

5.1. Factors that influence recruitment to observational studies

There are many factors that could potentially contribute to successful recruit participants to an observational study. Previous research has examined methods to increase the participation of both patients and healthcare professionals to primary care studies, with much of the research to date focused on randomised controlled trials (RENDELL, MERRITT and GEDDES, 2007; ROLLMAN et al., 2008). There is limited evidence available to inform researchers about factors that can influence recruitment to observational studies conducted in primary care (HAYWARD et al., 2013). To improve the availability of information on participant recruitment to observational studies, Chapter Nine of this thesis investigates factors that influence recruitment to an observational study for low back pain.

6. Aims of the thesis

The specific aims of this thesis were to:

1. Describe the short and long-term factors that primary care clinicians consider important in triggering a sudden episode of acute low back pain (Chapter 2)

2. Investigate the increase in risk of an episode of sudden onset, acute back pain associated with transient exposure to a range of physical and psychosocial factors (Chapter 3 and 4).

3. Investigate the influence of various weather conditions on risk of low back pain (Chapter 5).

4. Systematically review whether magnetic resonance imaging findings of the lumbar spine predict future low back pain in different samples with and without low back pain (Chapter 6).

5. Examine the prognosis and prognostic factors for patients with chronic low back pain presenting to a private, community-based, group exercise program (Chapter 7).

6. Investigate if the presence of magnetic resonance imaging findings identifies patients with low back pain who respond better to particular interventions (Chapter 8).

7. Identify factors that influence recruitment to an observational study (Chapter 9).

7. References

ABASOLO, L., A. TOBIAS, L. LEON, L. CARMONA, J. L. FERNANDEZ-RUEDA, A. B. RODRIGUEZ, B. FERNANDEZ-GUTIERREZ and J. A. JOVER, Weather conditions may worsen symptoms in rheumatoid arthritis patients: the possible effect of temperature. Reumatol Clin, 9, 4, p. 226-228, 2013. ANDERSSON, G. B., Epidemiological features of chronic low-back pain. Lancet, 354, 9178, p. 581-585, 1999. BALAGUE, F., A. F. MANNION, F. PELLISE and C. CEDRASCHI, Non-specific low back pain. Lancet, 379, 9814, p. 482-491, 2012. BLANTON, S., D. M. MORRIS, M. G. PRETTYMAN, K. MCCULLOCH, S. REDMOND, K. E. LIGHT and S. L. WOLF, Lessons learned in participant recruitment and retention: the EXCITE trial. Phys Ther, 86, 11, p. 1520-1533, 2006. BOWER, P., S. WILSON and N. MATHERS, Short report: how often do UK primary care trials face recruitment delays? Fam Pract, 24, 6, p. 601-603, 2007. BURDORF, A., B. NAAKTGEBOREN and H. C. DE GROOT, Occupational risk factors for low back pain among sedentary workers. J Occup Med, 35, 12, p. 1213-1220, 1993. CAIRNS, M. C., N. E. FOSTER and C. WRIGHT, Randomized controlled trial of specific spinal stabilization exercises and conventional physiotherapy for recurrent low back pain. Spine (Phila Pa 1976), 31, 19, p. E670-681, 2006. CHOU, R., A. QASEEM, V. SNOW, D. CASEY, J. T. CROSS, JR., P. SHEKELLE and D. K. OWENS, Diagnosis and treatment of low back pain: a joint clinical practice guideline from the American College of Physicians and the American Pain Society. Ann Intern Med, 147, 7, p. 478-491, 2007. COSTA LDA, C., B. W. KOES, G. PRANSKY, J. BORKAN, C. G. MAHER and R. J. SMEETS, Primary care research priorities in low back pain: an update. Spine (Phila Pa 1976), 38, 2, p. 148-156, 2013. COSTA LDA, C., C. G. MAHER, J. H. MCAULEY, M. J. HANCOCK, R. D. HERBERT, K. M. REFSHAUGE and N. HENSCHKE, Prognosis for patients with chronic low back pain: inception cohort study. BMJ, 339, p. b3829, 2009. CROFT, P. R., K. M. DUNN and H. RASPE, Course and prognosis of back pain in primary care: the epidemiological perspective. Pain, 122, 1-2, p. 1-3, 2006.

DA, C. M. C. L., C. G. MAHER, M. J. HANCOCK, J. H. MCAULEY, R. D. HERBERT and L. O. COSTA, The prognosis of acute and persistent low-back pain: a meta-analysis. CMAJ, 184, 11, p. E613-624, 2012. DAGENAIS, S., J. CARO and S. HALDEMAN, A systematic review of low back pain cost of illness studies in the United States and internationally. Spine J, 8, 1, p. 8-20, 2008. DEYO, R. A. and W. R. PHILLIPS, Low back pain. A primary care challenge. Spine (Phila Pa 1976), 21, 24, p. 2826-2832, 1996. DEYO, R. A. and J. N. WEINSTEIN, Low back pain. N Engl J Med, 344, 5, p. 363-370, 2001. EKMAN, M., O. JOHNELL and L. LIDGREN, The economic cost of low back pain in Sweden in 2001. Acta Orthop, 76, 2, p. 275-284, 2005. ENDEAN, A., K. T. PALMER and D. COGGON, Potential of magnetic resonance imaging findings to refine case definition for mechanical low back pain in epidemiological studies: a systematic review. Spine (Phila Pa 1976), 36, 2, p. 160-169, 2011. FAIRHURST, K. and C. DOWRICK, Problems with recruitment in a randomized controlled trial of counselling in general practice: causes and implications. J Health Serv Res Policy, 1, 2, p. 77-80, 1996. FERGUSON, S. A., W. G. ALLREAD, D. L. BURR, C. HEANEY and W. S. MARRAS, Biomechanical, psychosocial and individual risk factors predicting low back functional impairment among furniture distribution employees. Clin Biomech (Bristol, Avon), 27, 2, p. 117-123, 2012. FERREIRA, G. D., M. C. SILVA, A. J. ROMBALDI, E. D. WREGE, F. V. SIQUEIRA and P. C. HALLAL, Prevalence and associated factors of back pain in adults from southern Brazil: a population-based study. Rev Bras Fisioter, 15, 1, p. 31-36, 2011. HAMBERG-VAN REENEN, H. H., G. A. ARIENS, B. M. BLATTER, W. VAN MECHELEN and P. M. BONGERS, A systematic review of the relation between physical capacity and future low back and neck/shoulder pain. Pain, 130, 1-2, p. 93-107, 2007. HANCOCK, M. J., C. G. MAHER, J. LATIMER, M. F. SPINDLER, J. H. MCAULEY, M. LASLETT and N. BOGDUK, Systematic review of tests to identify the disc, SIJ or facet joint as the source of low back pain. Eur Spine J, 16, 10, p. 1539-1550, 2007. HAYDEN, J. A., K. M. DUNN, D. A. VAN DER WINDT and W. S. SHAW, What is the prognosis of back pain? Best Pract Res Clin Rheumatol, 24, 2, p. 167-179, 2010.

HAYWARD, R. A., M. PORCHERET, C. D. MALLEN and E. THOMAS, Recruiting patients and collecting data for an observational study using computerised record pop-up prompts: the PROG-RES study. Prim Health Care Res Dev, 14, 1, p. 21-28, 2013. HENSCHKE, N., C. G. MAHER, K. M. REFSHAUGE, R. D. HERBERT, R. G. CUMMING, J. BLEASEL, J. YORK, A. DAS and J. H. MCAULEY, Prevalence of and screening for serious spinal pathology in patients presenting to primary care settings with acute low back pain. Arthritis Rheum, 60, 10, p. 3072-3080, 2009. HOY, D., C. BAIN, G. WILLIAMS, L. MARCH, P. BROOKS, F. BLYTH, A. WOOLF, T. VOS and R. BUCHBINDER, A systematic review of the global prevalence of low back pain. Arthritis Rheum, 64, 6, p. 2028-2037, 2012. HOY, D., P. BROOKS, F. BLYTH and R. BUCHBINDER, The Epidemiology of low back pain. Best Pract Res Clin Rheumatol, 24, 6, p. 769-781, 2010. HOY, D., L. MARCH, P. BROOKS, A. WOOLF, F. BLYTH, T. VOS and R. BUCHBINDER, Measuring the global burden of low back pain. Best Pract Res Clin Rheumatol, 24, 2, p. 155-165, 2010. HUNNINGHAKE, D. B., C. A. DARBY and J. L. PROBSTFIELD, Recruitment experience in clinical trials: literature summary and annotated bibliography. Control Clin Trials, 8, 4 Suppl, p. 6S-30S, 1987. HURWITZ, E. L. and H. MORGENSTERN, Correlates of back problems and back-related disability in the United States. J Clin Epidemiol, 50, 6, p. 669-681, 1997. JAMISON, R. N., K. O. ANDERSON and M. A. SLATER, Weather changes and pain: perceived influence of local climate on pain complaint in chronic pain patients. Pain, 61, 2, p. 309-315, 1995. JARVIK, J. J., W. HOLLINGWORTH, P. HEAGERTY, D. R. HAYNOR and R. A. DEYO, The Longitudinal Assessment of Imaging and Disability of the Back (LAIDBack) Study: baseline data. Spine (Phila Pa 1976), 26, 10, p. 1158-1166, 2001. JENSEN, R. K. and C. LEBOEUF-YDE, Is the presence of modic changes associated with the outcomes of different treatments? A systematic critical review. BMC Musculoskelet Disord, 12, p. 183, 2011. JOHNSON, R. E., G. T. JONES, N. J. WILES, C. CHADDOCK, R. G. POTTER, C. ROBERTS, D. P. SYMMONS, P. J. WATSON, D. J. TORGERSON and G. J. MACFARLANE, Active exercise, education, and cognitive behavioral therapy for persistent disabling low back pain: a randomized controlled trial. Spine (Phila Pa 1976), 32, 15, p. 1578-1585, 2007.

KENT, P., M. HANCOCK, D. H. PETERSEN and H. L. MJOSUND, Clinimetrics corner: choosing appropriate study designs for particular questions about treatment subgroups. J Man Manip Ther, 18, 3, p. 147-152, 2010. KENT, P. and P. KJAER, The efficacy of targeted interventions for modifiable psychosocial risk factors of persistent nonspecific low back pain - a systematic review. Man Ther, 17, 5, p. 385-401, 2012. KENT, P., H. L. MJOSUND and D. H. PETERSEN, Does targeting manual therapy and/or exercise improve patient outcomes in nonspecific low back pain? A systematic review. BMC Med, 8, p. 22, 2010. KENT, P. M., J. L. KEATING and R. BUCHBINDER, Searching for a conceptual framework for nonspecific low back pain. Man Ther, 14, 4, p. 387-396, 2009. KJAER, P., L. KORSHOLM, T. BENDIX, J. S. SORENSEN and C. LEBOEUF-YDE, Modic changes and their associations with clinical findings. Eur Spine J, 15, 9, p. 1312- 1319, 2006. KOES, B. W., M. VAN TULDER, C. W. LIN, L. G. MACEDO, J. MCAULEY and C. MAHER, An updated overview of clinical guidelines for the management of non-specific low back pain in primary care. Eur Spine J, 19, 12, p. 2075-2094, 2010. KOES, B. W., M. W. VAN TULDER and W. C. PEUL, Diagnosis and treatment of sciatica. BMJ, 334, 7607, p. 1313-1317, 2007. KONSTANTINOU, K. and K. M. DUNN, Sciatica: review of epidemiological studies and prevalence estimates. Spine (Phila Pa 1976), 33, 22, p. 2464-2472, 2008. KOPEC, J. A., E. C. SAYRE and J. M. ESDAILE, Predictors of back pain in a general population cohort. Spine (Phila Pa 1976), 29, 1, p. 70-77; discussion 77-78, 2004. LAMB, S. E., Z. HANSEN, R. LALL, E. CASTELNUOVO, E. J. WITHERS, V. NICHOLS, R. POTTER and M. R. UNDERWOOD, Group cognitive behavioural treatment for low-back pain in primary care: a randomised controlled trial and cost-effectiveness analysis. Lancet, 375, 9718, p. 916-923, 2010. LAMBEEK, L. C., M. W. VAN TULDER, I. C. SWINKELS, L. L. KOPPES, J. R. ANEMA and W. VAN MECHELEN, The trend in total cost of back pain in The Netherlands in the period 2002 to 2007. Spine (Phila Pa 1976), 36, 13, p. 1050-1058, 2011. LANG, J., E. OCHSMANN, T. KRAUS and J. W. LANG, Psychosocial work stressors as antecedents of musculoskeletal problems: a systematic review and meta-analysis of stability- adjusted longitudinal studies. Soc Sci Med, 75, 7, p. 1163-1174, 2012.

LATZA, U., W. KARMAUS, T. STURMER, M. STEINER, A. NETH and U. REHDER, Cohort study of occupational risk factors of low back pain in construction workers. Occup Environ Med, 57, 1, p. 28-34, 2000. LITTLE, P., G. LEWITH, F. WEBLEY, M. EVANS, A. BEATTIE, K. MIDDLETON, J. BARNETT, K. BALLARD, F. OXFORD, P. SMITH, L. YARDLEY, S. HOLLINGHURST and D. SHARP, Randomised controlled trial of Alexander technique lessons, exercise, and massage (ATEAM) for chronic and recurrent back pain. BMJ, 337, p. a884, 2008. LOUW, Q. A., L. D. MORRIS and K. GRIMMER-SOMERS, The prevalence of low back pain in Africa: a systematic review. BMC Musculoskelet Disord, 8, p. 105, 2007. MACLURE, M., The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol, 133, 2, p. 144-153, 1991. MACLURE, M. and M. A. MITTLEMAN, Should we use a case-crossover design? Annu Rev Public Health, 21, p. 193-221, 2000. MAHER, C., WILLIAMS C, LIN C and L. J., Managing low back pain in primary care. Aust Prescr, 34, p. 128-132, 2011. MANEK, N. J. and A. J. MACGREGOR, Epidemiology of back disorders: prevalence, risk factors, and prognosis. Curr Opin Rheumatol, 17, 2, p. 134-140, 2005. MARIETTE, S. and X. MARIETTE, Low back pain in rural Tibet. Lancet, 361, 9369, p. 1654, 2003. MODIC, M. T., T. J. MASARYK, J. S. ROSS and J. R. CARTER, Imaging of degenerative disk disease. Radiology, 168, 1, p. 177-186, 1988. MODIC, M. T. and J. S. ROSS, Lumbar degenerative disk disease. Radiology, 245, 1, p. 43- 61, 2007. MODIC, M. T., P. M. STEINBERG, J. S. ROSS, T. J. MASARYK and J. R. CARTER, Degenerative disk disease: assessment of changes in vertebral body marrow with MR imaging. Radiology, 166, 1 Pt 1, p. 193-199, 1988. MURRAY, C. J., T. VOS, R. LOZANO, M. NAGHAVI, A. D. FLAXMAN, C. MICHAUD, M. EZZATI, K. SHIBUYA, J. A. SALOMON, S. ABDALLA, V. ABOYANS, J. ABRAHAM, I. ACKERMAN, R. AGGARWAL, S. Y. AHN, M. K. ALI, M. ALVARADO, H. R. ANDERSON, L. M. ANDERSON, K. G. ANDREWS, C. ATKINSON, L. M. BADDOUR, A. N. BAHALIM, S. BARKER-COLLO, L. H. BARRERO, D. H. BARTELS, M. G. BASANEZ, A. BAXTER, M. L. BELL, E. J. BENJAMIN, D. BENNETT, E. BERNABE, K. BHALLA, B. BHANDARI, B. BIKBOV, A. BIN ABDULHAK, G. BIRBECK, J. A. BLACK, H. BLENCOWE, J. D. BLORE, F. BLYTH, I. BOLLIGER, A.

BONAVENTURE, S. BOUFOUS, R. BOURNE, M. BOUSSINESQ, T. BRAITHWAITE, C. BRAYNE, L. BRIDGETT, S. BROOKER, P. BROOKS, T. S. BRUGHA, C. BRYAN- HANCOCK, C. BUCELLO, R. BUCHBINDER, G. BUCKLE, C. M. BUDKE, M. BURCH, P. BURNEY, R. BURSTEIN, B. CALABRIA, B. CAMPBELL, C. E. CANTER, H. CARABIN, J. CARAPETIS, L. CARMONA, C. CELLA, F. CHARLSON, H. CHEN, A. T. CHENG, D. CHOU, S. S. CHUGH, L. E. COFFENG, S. D. COLAN, S. COLQUHOUN, K. E. COLSON, J. CONDON, M. D. CONNOR, L. T. COOPER, M. CORRIERE, M. CORTINOVIS, K. C. DE VACCARO, W. COUSER, B. C. COWIE, M. H. CRIQUI, M. CROSS, K. C. DABHADKAR, M. DAHIYA, N. DAHODWALA, J. DAMSERE-DERRY, G. DANAEI, A. DAVIS, D. DE LEO, L. DEGENHARDT, R. DELLAVALLE, A. DELOSSANTOS, J. DENENBERG, S. DERRETT, D. C. DES JARLAIS, S. D. DHARMARATNE, M. DHERANI, C. DIAZ-TORNE, H. DOLK, E. R. DORSEY, T. DRISCOLL, H. DUBER, B. EBEL, K. EDMOND, A. ELBAZ, S. E. ALI, H. ERSKINE, P. J. ERWIN, P. ESPINDOLA, S. E. EWOIGBOKHAN, F. FARZADFAR, V. FEIGIN, D. T. FELSON, A. FERRARI, C. P. FERRI, E. M. FEVRE, M. M. FINUCANE, S. FLAXMAN, L. FLOOD, K. FOREMAN, M. H. FOROUZANFAR, F. G. FOWKES, M. FRANSEN, M. K. FREEMAN, B. J. GABBE, S. E. GABRIEL, E. GAKIDOU, H. A. GANATRA, B. GARCIA, F. GASPARI, R. F. GILLUM, G. GMEL, D. GONZALEZ-MEDINA, R. GOSSELIN, R. GRAINGER, B. GRANT, J. GROEGER, F. GUILLEMIN, D. GUNNELL, R. GUPTA, J. HAAGSMA, H. HAGAN, Y. A. HALASA, W. HALL, D. HARING, J. M. HARO, J. E. HARRISON, R. HAVMOELLER, R. J. HAY, H. HIGASHI, C. HILL, B. HOEN, H. HOFFMAN, P. J. HOTEZ, D. HOY, J. J. HUANG, S. E. IBEANUSI, K. H. JACOBSEN, S. L. JAMES, D. JARVIS, R. JASRASARIA, S. JAYARAMAN, N. JOHNS, J. B. JONAS, G. KARTHIKEYAN, N. KASSEBAUM, N. KAWAKAMI, A. KEREN, J. P. KHOO, C. H. KING, L. M. KNOWLTON, O. KOBUSINGYE, A. KORANTENG, R. KRISHNAMURTHI, F. LADEN, R. LALLOO, L. L. LASLETT, T. LATHLEAN, J. L. LEASHER, Y. Y. LEE, J. LEIGH, D. LEVINSON, S. S. LIM, E. LIMB, J. K. LIN, M. LIPNICK, S. E. LIPSHULTZ, W. LIU, M. LOANE, S. L. OHNO, R. LYONS, J. MABWEIJANO, M. F. MACINTYRE, R. MALEKZADEH, L. MALLINGER, S. MANIVANNAN, W. MARCENES, L. MARCH, D. J. MARGOLIS, G. B. MARKS, R. MARKS, A. MATSUMORI, R. MATZOPOULOS, B. M. MAYOSI, J. H. MCANULTY, M. M. MCDERMOTT, N. MCGILL, J. MCGRATH, M. E. MEDINA-MORA, M. MELTZER, G. A. MENSAH, T. R. MERRIMAN, A. C. MEYER, V. MIGLIOLI, M. MILLER, T. R. MILLER, P. B. MITCHELL, C. MOCK, A. O. MOCUMBI, T. E. MOFFITT, A. A.

MOKDAD, L. MONASTA, M. MONTICO, M. MORADI-LAKEH, A. MORAN, L. MORAWSKA, R. MORI, M. E. MURDOCH, M. K. MWANIKI, K. NAIDOO, M. N. NAIR, L. NALDI, K. M. NARAYAN, P. K. NELSON, R. G. NELSON, M. C. NEVITT, C. R. NEWTON, S. NOLTE, P. NORMAN, R. NORMAN, M. O'DONNELL, S. O'HANLON, C. OLIVES, S. B. OMER, K. ORTBLAD, R. OSBORNE, D. OZGEDIZ, A. PAGE, B. PAHARI, J. D. PANDIAN, A. P. RIVERO, S. B. PATTEN, N. PEARCE, R. P. PADILLA, F. PEREZ-RUIZ, N. PERICO, K. PESUDOVS, D. PHILLIPS, M. R. PHILLIPS, K. PIERCE, S. PION, G. V. POLANCZYK, S. POLINDER, C. A. POPE, 3RD, S. POPOVA, E. PORRINI, F. POURMALEK, M. PRINCE, R. L. PULLAN, K. D. RAMAIAH, D. RANGANATHAN, H. RAZAVI, M. REGAN, J. T. REHM, D. B. REIN, G. REMUZZI, K. RICHARDSON, F. P. RIVARA, T. ROBERTS, C. ROBINSON, F. R. DE LEON, L. RONFANI, R. ROOM, L. C. ROSENFELD, L. RUSHTON, R. L. SACCO, S. SAHA, U. SAMPSON, L. SANCHEZ-RIERA, E. SANMAN, D. C. SCHWEBEL, J. G. SCOTT, M. SEGUI-GOMEZ, S. SHAHRAZ, D. S. SHEPARD, H. SHIN, R. SHIVAKOTI, D. SINGH, G. M. SINGH, J. A. SINGH, J. SINGLETON, D. A. SLEET, K. SLIWA, E. SMITH, J. L. SMITH, N. J. STAPELBERG, A. STEER, T. STEINER, W. A. STOLK, L. J. STOVNER, C. SUDFELD, S. SYED, G. TAMBURLINI, M. TAVAKKOLI, H. R. TAYLOR, J. A. TAYLOR, W. J. TAYLOR, B. THOMAS, W. M. THOMSON, G. D. THURSTON, I. M. TLEYJEH, M. TONELLI, J. A. TOWBIN, T. TRUELSEN, M. K. TSILIMBARIS, C. UBEDA, E. A. UNDURRAGA, M. J. VAN DER WERF, J. VAN OS, M. S. VAVILALA, N. VENKETASUBRAMANIAN, M. WANG, W. WANG, K. WATT, D. J. WEATHERALL, M. A. WEINSTOCK, R. WEINTRAUB, M. G. WEISSKOPF, M. M. WEISSMAN, R. A. WHITE, H. WHITEFORD, N. WIEBE, S. T. WIERSMA, J. D. WILKINSON, H. C. WILLIAMS, S. R. WILLIAMS, E. WITT, F. WOLFE, A. D. WOOLF, S. WULF, P. H. YEH, A. K. ZAIDI, Z. J. ZHENG, D. ZONIES, A. D. LOPEZ, M. A. ALMAZROA and Z. A. MEMISH, Disability-adjusted life years (DALYs) for 291 diseases and injuries in 21 regions, 1990-2010: a systematic analysis for the Global Burden of Disease Study 2010. Lancet, 380, 9859, p. 2197-2223, 2012. MURTEZANI, A., Z. IBRAIMI, S. SLLAMNIKU, T. OSMANI and S. SHERIFI, Prevalence and risk factors for low back pain in industrial workers. Folia Med (Plovdiv), 53, 3, p. 68-74, 2011. PATBERG, W. R. and J. J. RASKER, Weather effects in rheumatoid arthritis: from controversy to consensus. A review. J Rheumatol, 31, 7, p. 1327-1334, 2004.

RENDELL, J. M., R. D. MERRITT and J. R. GEDDES, Incentives and disincentives to participation by clinicians in randomised controlled trials. Cochrane Database Syst Rev, 2, p. MR000021, 2007. ROLLMAN, B. L., G. S. FISCHER, F. ZHU and B. H. BELNAP, Comparison of electronic physician prompts versus waitroom case-finding on clinical trial enrollment. J Gen Intern Med, 23, 4, p. 447-450, 2008. RUBIN, D. I., Epidemiology and risk factors for spine pain. Neurol Clin, 25, 2, p. 353-371, 2007. SHEEHAN, N. J., Magnetic resonance imaging for low back pain: indications and limitations. Ann Rheum Dis, 69, 1, p. 7-11, 2010. SMEDSLUND, G., P. MOWINCKEL, T. HEIBERG, T. K. KVIEN and K. B. HAGEN, Does the weather really matter? A cohort study of influences of weather and solar conditions on daily variations of joint pain in patients with rheumatoid arthritis. Arthritis Rheum, 61, 9, p. 1243-1247, 2009. SPAAR, A., M. FREY, A. TURK, W. KARRER and M. A. PUHAN, Recruitment barriers in a randomized controlled trial from the physicians' perspective: a postal survey. BMC Med Res Methodol, 9, p. 14, 2009. TAYLOR, J. B., A. P. GOODE, S. Z. GEORGE and C. E. COOK, Incidence and risk factors for first-time incident low back pain: a systematic review and meta-analysis. Spine J, p. 2014. TENIAS, J. M., M. ESTARLICH, V. FUENTES-LEONARTE, C. INIGUEZ and F. BALLESTER, Short-term relationship between meteorological variables and hip fractures: an analysis carried out in a health area of the Autonomous Region of Valencia, Spain (1996- 2005). Bone, 45, 4, p. 794-798, 2009. THELIN, A., S. HOLMBERG and N. THELIN, Functioning in neck and low back pain from a 12-year perspective: a prospective population-based study. J Rehabil Med, 40, 7, p. 555- 561, 2008. VALAT, J. P., S. GENEVAY, M. MARTY, S. ROZENBERG and B. KOES, Sciatica. Best Pract Res Clin Rheumatol, 24, 2, p. 241-252, 2010. VAN TULDER, M., A. BECKER, T. BEKKERING, A. BREEN, M. T. DEL REAL, A. HUTCHINSON, B. KOES, E. LAERUM and A. MALMIVAARA, Chapter 3. European guidelines for the management of acute nonspecific low back pain in primary care. Eur Spine J, 15 Suppl 2, p. S169-191, 2006.

VANDERGRIFT, J. L., J. E. GOLD, A. HANLON and L. PUNNETT, Physical and psychosocial ergonomic risk factors for low back pain in automobile manufacturing workers. Occup Environ Med, 69, 1, p. 29-34, 2012. VINDIGNI, D., B. F. WALKER, J. R. JAMISON, C. DA COSTA, L. PARKINSON and S. BLUNDEN, Low back pain risk factors in a large rural Australian Aboriginal community. An opportunity for managing co-morbidities? Chiropr Osteopat, 13, p. 21, 2005. VOLINN, E., The epidemiology of low back pain in the rest of the world. A review of surveys in low- and middle-income countries. Spine (Phila Pa 1976), 22, 15, p. 1747-1754, 1997. VON KORFF, M., R. A. DEYO, D. CHERKIN and W. BARLOW, Back pain in primary care. Outcomes at 1 year. Spine (Phila Pa 1976), 18, 7, p. 855-862, 1993. WALKER, B. F., R. MULLER and W. D. GRANT, Low back pain in Australian adults: the economic burden. Asia Pac J Public Health, 15, 2, p. 79-87, 2003. WANG, Y., T. VIDEMAN and M. C. BATTIE, ISSLS prize winner: Lumbar vertebral endplate lesions: associations with disc degeneration and back pain history. Spine (Phila Pa 1976), 37, 17, p. 1490-1496, 2012. WASSENAAR, M., R. M. VAN RIJN, M. W. VAN TULDER, A. P. VERHAGEN, D. A. VAN DER WINDT, B. W. KOES, M. R. DE BOER, A. Z. GINAI and R. W. OSTELO, Magnetic resonance imaging for diagnosing lumbar spinal pathology in adult patients with low back pain or sciatica: a diagnostic systematic review. Eur Spine J, 21, 2, p. 220-227, 2012. WIESER, S., B. HORISBERGER, S. SCHMIDHAUSER, C. EISENRING, U. BRUGGER, A. RUCKSTUHL, J. DIETRICH, A. F. MANNION, A. ELFERING, O. TAMCAN and U. MULLER, Cost of low back pain in Switzerland in 2005. Eur J Health Econ, 12, 5, p. 455- 467, 2011.

Chapter Two

Clinicians’ views on factors that trigger a sudden onset of low back pain

Chapter Two is published as:

Steffens D, Maher CG, Ferreira ML, Hancock MJ, Glass T, Latimer J. Clinicians’ views on factors that trigger a sudden onset of low back pain. European Spine Journal. 2014; 23:512- 519. 47

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “Clinicians’ views on factors that trigger a sudden onset of low back pain”, we confirm that Daniel Steffens has made the following contributions:

 Conception and design of the research  Data collection  Analysis and interpretation of the findings  Writing of the manuscript and critical appraisal of the content

Christopher G Maher Date: 01.01.2015

Manuela L Ferreira Date: 01.01.2015

Mark J Hancock Date: 01.01.2015

Timothy Glass Date: 01.01.2015

Jane Latimer Date: 01.01.2015

48 Eur Spine J (2014) 23:512–519 DOI 10.1007/s00586-013-3120-y

ORIGINAL ARTICLE

Clinicians’ views on factors that trigger a sudden onset of low back pain

Daniel Steffens • Chris G. Maher • Manuela L. Ferreira • Mark J. Hancock • Timothy Glass • Jane Latimer

Received: 8 August 2013 / Revised: 24 November 2013 / Accepted: 24 November 2013 / Published online: 8 December 2013 Ó Springer-Verlag Berlin Heidelberg 2013

Abstract sudden episode of acute LBP, based on their experience. Purpose Little is known about what triggers an episode of Descriptive statistics and frequency distributions were used low back pain (LBP) in those presenting to primary care. to describe clinician’s characteristics and the frequencies of Previous studies of risk factors have focused on specific the main risk factor categories were reported. occupational settings and work conditions. No study has Results Based on the views of 103 primary care clini- asked primary care clinicians to consider what triggers an cians, biomechanical risk factors appear to be the most episode of sudden-onset LBP in patients presenting to them important short-term triggers (endorsed by 89.3 % of cli- for care. The purpose of this study, therefore, was to nicians) and long-term triggers (endorsed by 54.2 % of describe the short- and long-term factors that primary care clinicians) for a sudden episode of acute LBP. Individual clinicians consider important in triggering a sudden epi- risk factors were endorsed by 39 % of clinicians as sode of acute LBP. important long-term triggers, while only 6.4 % of clini- Methods One hundred and thirty-one primary care clinicians considered them important short-term triggers. Other cians who were recruiting patients with LBP to a large risk factors, such as psychological/psychosocial and observational study were invited to participate. A ques- genetic factors, were not commonly endorsed as risk fac- tionnaire was designed to obtain information about the tors for an episode of LBP by primary care clinicians. clinician’s characteristics, profession and clinical experi- Conclusions This study shows that primary care clini- ence. We also asked clinicians to nominate the five short- cians believe that biomechanical risk factors are the most and five long-term exposure factors, most likely to trigger a important short-term triggers, while biomechanical and individual risk factors are the most important long-term triggers for a sudden onset of LBP. However, other risk D. Steffens Á C. G. Maher Á M. L. Ferreira Á J. Latimer factors, such as psychological/psychosocial and genetic, Musculoskeletal Division, The George Institute for Global Health, Sydney Medical School, The University of Sydney, were not commonly endorsed as risk factors for an episode Level 13, 321 Kent Street, Sydney, NSW 2000, Australia of LBP by primary care clinicians. Results of this study are based on primary care clinicians’ views and further & D. Steffens ( ) investigation is needed to test the validity of these sug- Musculoskeletal Division, The George Institute for Global Health, Sydney Medical School, The University of Sydney, gested risk factors. Missenden Road, PO Box M201, Sydney, NSW 2050, Australia e-mail: [email protected] Keywords Low back pain Á Primary care Á Risk factors Á Observational Á Epidemiology M. J. Hancock Discipline of Physiotherapy, Faculty of Human Sciences, Macquarie University, 2 Technology Place, Macquarie Park, Sydney, NSW 2113, Australia Introduction

T. Glass Warwick Medical School, University of Warwick, Low back pain (LBP) is the most common musculoskeletal Coventry CV4 7AL, UK condition affecting approximately 80 % of the adult 123 49 Eur Spine J (2014) 23:512–519 513 population at least once during their lifetime [1, 2]. LBP Materials and methods troubles are the fifth most frequent reason for a visit to a primary health care clinician in the USA and the seventh Data for this study were provided by clinicians recruiting most frequent reason in Australia [3, 4]. In Australia back participants to the TRIGGERS case crossover study [35]. pain is the health condition that carries the greatest burden TRIGGERS is an observational study and was designed when considering lives lived with disability [5]. The global to quantify the transient increase in risk of a sudden burden of the condition is enormous and includes high episode of LBP associated with acute exposure to a costs with medical care, loss of productivity and indemnity range of common physical and psychological factors payments [6]. Despite this, investigation into risk factors [35]. for LBP is in a developing stage when compared with other common conditions such as cardiovascular disease and Study sample cancer [7]. A better understanding of factors that trigger an episode of LBP may provide important insights into the Participants in the study were 131 primary care clinicians prevention and management of this condition. (physiotherapists and chiropractors) who were recruiting LBP is a complex condition with many factors believed patients for the TRIGGERS study from October 2011 to to contribute to its onset [8]. These factors can be aggre- November 2012 in NSW, Australia. gated into a smaller number of categories [9] including biomechanical factors (regular lifting, exposure to vibra- Recruitment tion, physically demanding jobs, bending and twisting, pushing and pulling heavy loads, awkward posture) [2, 7– Clinicians were contacted and invited to participate by 20], psychological/psychosocial factors (job satisfaction, e-mail. In the e-mail, we attached the invitation letter, local support in the workplace, depression, job control, consent form and a one-page questionnaire. The invitation stress) [2, 7, 9, 16, 19–25] and individual risk factors letter consisted of a brief explanation of the study methods (sedentary lifestyle, age, smoking, gender, obesity, poor and aims. Clinicians who consented to participate were general health, marital status, pregnancy) [2, 7, 9, 12, 13, instructed to answer the one-page questionnaire based on 20, 21, 23, 26–31]. Past studies have investigated many of their clinical experience and send it back to the research these factors and their relationship to LBP. A few studies team. Clinicians who had not responded within 2 weeks have also investigated genetic risk factors [7, 29, 32]. were contacted by a study researcher and again invited to Interestingly, only a small number of studies have exam- participate. Ethical approval for the study was granted by ined risk factors across all four risk factor categories. the University of Sydney Human Research Ethics While there are many studies of risk factors for LBP [9, Committee. 33, 34], most knowledge stems from studies focused on specific occupational groups and working conditions [8– Data collection 10, 12–19, 23]. Although studies conducted in occupational settings may reveal important risk factors for work-related A one-page self-administered questionnaire was designed back pain, these risks may not be relevant to other popu- to obtain information about the clinician’s characteristics lations such as those drawn from primary care. In addition, (gender, age and address), profession (physiotherapist or previous studies of LBP have been criticised as being too chiropractor) and clinical experience (years as practising narrowly focused on only one or perhaps two of the cate- clinician and years managing LBP). Two items were used gories of individual, biomechanical, psychological/psy- to measure their beliefs about triggers for LBP. These chosocial and genetic aspects of the problem [12]. were: Little is known about the causes of LBP in the general 1. Based on your clinical experience, list what you population who present to primary care. The starting point consider to be the five most likely factors involving in evaluating risk factors for LBP in primary care is the short-term exposure that are triggers for a sudden identification of putative risk factors that need to be episode of acute low back pain? investigated. Interviewing primary care clinicians who 2. Based on your clinical experience, list what you frequently manage patients with LBP can provide insight consider to be the five most likely factors involving into what may be the most common triggers for this con- long-term exposure that increase the risk of a sudden dition. The aim of this study, therefore, was to describe the episode of acute low back pain? most likely risk factors involving short- and long-term exposure that primary care clinicians believe could trigger All questionnaires were returned by e-mail and the a sudden episode of acute LBP. answers entered in the study database.

123 50 514 Eur Spine J (2014) 23:512–519

Fig. 1 Diagram of risk factors for low back pain

Risk factors categories Coding short- and long-term risk factors

We developed descriptive categories to code the free text Two researchers individually coded the free text answers responses based upon published risk factor studies. A about short- and long-term exposures that may be a search from the earliest record to January 2013 was trigger for a sudden episode of acute LBP. The free text undertaken on PubMed to identify relevant studies using responses were ﬁrst coded into ﬁve main categories the following keywords: low back pain, backache, lum- (Fig. 1): bago, risk factors, causality, aetiology and epidemiology. 1. Individual risk factors Abstracts were retrieved and examined. Reported risk 2. Biomechanical risk factors factors were extracted and entered into a table. Figure 1 3. Psychological/psychosocial risk factors describes the categories we developed including main 4. Genetic risk factors category (or aggregate label) and the sub-categories 5. Other risk factors (Table 1).

123 51 Eur Spine J (2014) 23:512–519 515

Table 1 Clinicians’ characteristics (n = 103) Table 3 Frequency of main risk factors stratiﬁed by short and long term Characteristics n (%) or mean ± SD Main categories risk factors Short terma Long terma Gender (male) 56 (55 %) N % N % Age (years) 43 ± 10 Profession Individual 33 6.4 201 39 Physiotherapist 102 (99 %) Biomechanical 460 89.3 279 54.2 Chiropractor 1 (1 %) Psychological/psychosocial 3 0.6 16 3.1 Current position Genetic 0 0.0 1 0.2 Principal/owner 45 (44 %) Other risk factors 9 1.7 8 1.6 Clinician/senior clinician 58 (56 %) Missing data 10 1.9 10 1.9 Clinical experience (years) 19 ± 10 Total 515 100 515 100 Clinical experience managing LBP (years) 18 ± 9 a v2 = 180.70, p \ 0.001

version 21 (SPSS, Inc., Chicago, IL) with a signiﬁcance Table 2 Reliability between researchers A and B level set at 0.05. Risk factors categories Observed agreement (%) ja

Short-term main categories 99 0.95 Results Short-term sub-categories 96 0.89 Long-term main categories 94 0.88 Of the 131 clinicians invited to participate in the study, 103 Long-term sub-categories 89 0.71 (79 %) completed the questionnaire. The characteristics of a Interpretation: 0.01–0.20 slight agreement; 0.21–0.40 fair agree- the participating clinicians are described in Table 2. The ment; 0.41–0.60 moderate agreement; 0.61–0.80 substantial agree- mean age was 43 years (range 23–65), clinical experience ment; 0.81–0.99 almost perfect agreement [36] managing LBP was 18 years (range 1–40) and slightly more were male (55 %). The same researchers then coded the free text responses In general, the reliability between the two independent into sub-categories. Each nominated risk factor was only researchers who coded the short- and long-term risk factor classified once into a main category and a sub-category categories was very good (Table 2). The observed agree- (e.g. 1: Smoking was first categorised as an ‘‘individual ment ranged from 89 to 99 % and the inter-rater reliability risk factor’’ and then sub-categorised into ‘‘smoking’’. e.g. ranged from substantial (j = 0.71) to almost perfect 2: Golf was first categorised as a ‘‘biomechanical risk agreement (j = 0.95) [36]. factor’’ and then sub-categorised into ‘‘sport injuries’’). In case of disagreement, a third researcher was consulted and Short- and long-term main categories risk factors a decision was made by consensus. Table 3 lists the frequency of endorsement of the main Statistical analysis short- and long-term risk factor categories. Based on the views of 103 primary care clinicians, biomechanical risk To check the reliability of the coding of free text data, we factors appear to be the most important short-term triggers compared the independent coding of the two researchers. (endorsed by 89.3 % of clinicians) and long-term triggers Inter-rater reliability was expressed using percentage exact (endorsed by 54.2 % of clinicians) for a sudden episode of agreement and Cohen’s kappa [36] for the main categories acute LBP. Individual risk factors were endorsed by 39 % and sub-categories. of clinicians as important long-term triggers, while only Descriptive statistics and frequency distributions were 6.4 % of clinicians considered them important short-term calculated to describe the characteristics of the clinicians triggers. Only one clinician reported genetics (0.2 %) as a and the frequencies that the main risk factor categories long-term risk factor for acute LBP. Chi-square tests con- were reported (individual, biomechanical, psychological/ firmed a significant difference in proportions of the main psychosocial, genetic and other risk factors). short- and long-term risk factor categories (v2 = 180.70, The clinicians’ endorsement of various categories of p \ 0.001). While biomechanical factors were considered triggers (main categories and sub-categories) was exam- the most important short- and long-term risk factors, ined using the Chi-squared (v2) statistic. All analyses were individual factors were endorsed far more commonly as performed using the statistical package SPSS for Windows long-term risk factors.

123 52 516 Eur Spine J (2014) 23:512–519

Table 4 Frequency of risk factor sub-categories stratified by short risk factors, prolonged sitting (13.4 %) was most frequently and long term endorsed followed by lifting (10.9 %) and physical inactiv- Risk factors sub-categories Short terma Long terma ity (9.1 %). Similarly to the analysis of the main categories, the Chi-square test (v2 = 360.456, p \ 0.001) suggested N % N % that there was a difference between short- and long-term sub- Lifting 90 17.5 56 10.9 category risk factors. Obvious sub-category differences Prolonged sitting 47 9.1 69 13.4 included physical trauma and physical inactivity for short Physical trauma 46 8.9 3 0.6 and long term; physical trauma was endorsed more fre- Bending 41 8.0 24 4.7 quently as a short-term risk factor, while inactivity was Unaccustomed activity 38 7.4 2 0.4 endorsed more commonly as a long-term risk factor. Other biomechanical risk factors 27 5.2 28 5.4 Sport injuries 24 4.7 9 1.7 Gardening 20 3.9 2 0.4 Discussion Bending and twisting 17 3.3 6 1.2 Twisting 14 2.7 4 0.8 Statement of principal findings Coughing/sneezing 12 2.3 1 0.2 Lifting and twisting 12 2.3 0 0.0 Based on the opinions of 103 primary care clinicians, the Lifting and bending 12 2.3 8 1.6 most important main short- and long-term risk factors to Driving 11 2.1 15 2.9 trigger an episode of sudden acute LBP are biomechanical Bending and twisting while lifting 10 1.9 2 0.4 (89.3 and 54.2 %, respectively) and individual risk factors Sudden movements 8 1.6 1 0.2 (6.4 and 39 %, respectively). Surprisingly, other risk factors, such as psychological/psychosocial and genetic factors, Other risk factors 9 1.7 8 1.6 were not commonly endorsed as risk factors for an episode of Prolonged standing 6 1.2 16 3.1 LBP by primary care clinicians. Some of the most frequently Unexpected load 6 1.2 0 0.0 reported short- and long-term sub-categories to trigger a Repetitive movements 5 1.0 8 1.6 sudden episode of acute LBP are lifting (17.5 and 10.9 %, Other individual risk factors 4 0.8 30 5.8 respectively) and prolonged sitting (9.1 and 13.4 %, Pulling/pushing 5 1.0 2 0.4 respectively). Pregnancy/childbirth 3 0.6 13 2.5 Previous LBP episodes 2 0.4 15 2.9 Strengths and weaknesses of the study Diminished trunk muscle strength and 2 0.4 22 4.3 fatigability We were able to interview a considerable number of Spine/pelvis/lower limb impairment 2 0.4 23 4.5 experienced primary care clinicians currently treating Sedentary work 2 0.4 14 2.7 patients with LBP. In addition, we identified a range of Stress 2 0.4 9 1.7 factors that are considered triggers involving short- and Overweight 1 0.2 25 4.9 long-term exposures. The reliability between the two Spine/pelvis/lower limb pathology 1 0.2 11 2.1 independent researchers who coded the short- and long- Physical inactivity 0 0.0 47 9.1 term risk factor categories was very good with the inter- Less frequent risk factorsb 26 5.0 32 6.2 rater reliability ranging from substantial to almost perfect Missing data 10 1.9 10 1.9 agreement [36]. However, the results of the study are based Total 515 100 515 100 on primary care clinicians’ views and further investigation Bold top five short and long risk factor sub-categories is needed to test the validity of these suggested risk factors. a v2 = 360.456, p \ 0.001 b Risk factors with a grouped frequency count B5 were combined Comparison to others studies

To our knowledge, this is the first observational study that Short- and long-term sub-categories risk factors has interviewed primary care clinicians to determine their views on short- and long-term triggers for a sudden episode Table 4 lists the frequency of endorsement of short- and of acute LBP in the general population. Most of the pre- long-term sub-category risk factors for acute LBP. Lifting vious studies on risk factors for LBP tended to focus on (17.5 %) was the most frequently endorsed short-term sub- specific triggers and included samples from occupational category risk factor, followed by prolonged sitting (9.1 %) settings. There is a research focus on biomechanical [2, 7– and physical trauma (8.9 %). For long-term sub-category 20], psychological/psychosocial [2, 7, 9, 16, 19–25] and 123 53 Eur Spine J (2014) 23:512–519 517 individual risk factors [2, 7, 9, 12, 13, 20, 21, 23, 26–31] investigate all main categories of risk factors found to for LBP; however, only a few studies investigated risk trigger an episode of LBP (individual, biomechanical, factors from all domains together [7]. psychosocial/psychological, genetic and others risk fac- Findings from our study suggest that the views of cli- tors). There is also a need to investigate the combination of nicians are as variable as in the evidence found in the two or more risk factors. Previous studies have confirmed literature. Our study found that biomechanical risk factors, that a combination of two or more risk factors (i.e. poor such as lifting, prolonged sitting and bending, were con- posture, repetitive lifting and high job strain) are com- sidered by primary care clinicians to be important risk monly identified in the same individual [23]. Increasing our factors for the onset of LBP, in accordance with previous understanding of what triggers an episode of LBP will studies [10, 15, 26, 37]. On the other hand, individual risk enable us to design better prevention programs. Future factors, such as physical inactivity, have only weak support research could involve the modification of endorsed trig- in the literature as a risk factor for developing LBP [27, gers in a clinical trial to determine the effectiveness in 29]. This is in contrast to the study clinicians who con- reducing future episodes of LBP. In addition, there is a sidered physical inactivity as a common long-term risk need to determine whether psychological/psychosocial factor for LBP. Although psychological/psychosocial and features are important risk factors in the primary care set- genetic risk factors did not appear to be endorsed strongly ting using rigorous prospective cohort designs. by the primary care clinicians interviewed in this study, there is strong evidence in the literature suggesting the opposite [12, 22, 37]. However, this inconsistency between Conclusions the present study and the literature may be due to the different settings in which the studies were performed and/or Based on primary care clinicians’ opinion, biomechanical clinician’s lack of awareness about the importance of risk factors are the most important short-term triggers, psychological risk factors. Psychological/psychosocial while biomechanical and individual risk factors are the factors may be more important in occupational settings most important long-term triggers for sudden-onset LBP. than in primary care. Conversely, individual risk factors Psychological/psychosocial and genetic risk factors were may be more important in primary care settings than not considered important risk factors by primary care cli- occupational settings. nicians. Findings from this study should be further investigated to better understand short- and long-term exposures Meaning of the study: possible mechanisms that are triggers for an acute sudden episode of LBP. This and implications for clinicians information will help inform LBP prevention programs.

Based on the opinion of primary care clinicians, the most Conﬂict of interest None. likely short-term risk factors for triggering an episode of LBP are, by far, the biomechanical risk factors and the most likely causes involving long-term exposure are indi- References vidual and biomechanical risk factors. This is somewhat inconsistent with the literature [27, 29]. The reasons why 1. Britt H, Miller G, Knox S, Charles J, Pan Y, Henderson J, experienced clinicians think such factors are important may Baryram C, Valenti L, Ng A, O’Halloran J (2005) General be due to the fact that they are seeing a general sample of practice activity in Australia 2004–2005. General practice series No. 18. Australian Institute of Health and Welfare, Canberra the population, while previous studies tend to focus on 2. Rubin DI (2007) Epidemiology and risk factors for spine pain. occupational settings. One other reason may be that pri- Neurol Clin 25(2):353–371. doi:10.1016/j.ncl.2007.01.004 mary care clinicians have limited expertise in assessing 3. Williams C, Maher C, Hancock M, McAuley J, McLachlan A, psychosocial/psychological and genetic factors as likely Britt H, Fahridin S, Harrison C, Latimer J (2010) Low back pain and best practice care: a survey of general practice physicians. triggers for an onset of LBP. JAMA 170(3):271–277 A better understanding of the most endorsed risk factors 4. Deyo R (1998) Low-back pain. Sci Am 279(2):48–53 for LBP in primary care will help clinicians to improve 5. Murray CJ, Vos T, Lozano R, Naghavi M, Flaxman AD, Michaud patient treatment and also provide important advice in C, Ezzati M, Shibuya K, Salomon JA, Abdalla S, Aboyans V, Abraham J, Ackerman I, Aggarwal R, Ahn SY, Ali MK, Alva- prevention of future LBP episodes. rado M, Anderson HR, Anderson LM, Andrews KG, Atkinson C, Baddour LM, Bahalim AN, Barker-Collo S, Barrero LH, Bartels Unanswered questions and future research DH, Basanez MG, Baxter A, Bell ML, Benjamin EJ, Bennett D, Bernabe E, Bhalla K, Bhandari B, Bikbov B, Bin Abdulhak A, Birbeck G, Black JA, Blencowe H, Blore JD, Blyth F, Bolliger I, Future studies should investigate risk factors for LBP in Bonaventure A, Boufous S, Bourne R, Boussinesq M, Braithwaite samples of the general population and simultaneously T, Brayne C, Bridgett L, Brooker S, Brooks P, Brugha TS, Bryan- 123 54 518 Eur Spine J (2014) 23:512–519

Hancock C, Bucello C, Buchbinder R, Buckle G, Budke CM, 7. Manek NJ, MacGregor AJ (2005) Epidemiology of back disor- Burch M, Burney P, Burstein R, Calabria B, Campbell B, Canter ders: prevalence, risk factors, and prognosis. Curr Opin Rheu- CE, Carabin H, Carapetis J, Carmona L, Cella C, Charlson F, matol 17:134–140 Chen H, Cheng AT, Chou D, Chugh SS, Coffeng LE, Colan SD, 8. Latza U, Karmaus W, Stu¨rmer T, Steiner M, Neth A, Rehder U Colquhoun S, Colson KE, Condon J, Connor MD, Cooper LT, (2000) Cohort study of occupational risk factors of low back pain Corriere M, Cortinovis M, de Vaccaro KC, Couser W, Cowie BC, in construction workers. Occup Environ Med 57:28–34 Criqui MH, Cross M, Dabhadkar KC, Dahiya M, Dahodwala N, 9. Ferguson S, Allread W, Burr D, Heaney C, Marras W (2012) Damsere-Derry J, Danaei G, Davis A, De Leo D, Degenhardt L, Biomechanical, psychosocial and individual risk factors predict- Dellavalle R, Delossantos A, Denenberg J, Derrett S, Des Jarlais ing low back functional impairment among furniture distribution DC, Dharmaratne SD, Dherani M, Diaz-Torne C, Dolk H, Dorsey employees. Clin Biomech 27:117–123 ER, Driscoll T, Duber H, Ebel B, Edmond K, Elbaz A, Ali SE, 10. Burdorf A, Naaktgeboren B, de Groot HCWM (1993) Occupa- Erskine H, Erwin PJ, Espindola P, Ewoigbokhan SE, Farzadfar F, tional risk factors for low back pain among sedentary workers. Feigin V, Felson DT, Ferrari A, Ferri CP, Fevre EM, Finucane J Occup Med 35(12):1213–1220 MM, Flaxman S, Flood L, Foreman K, Forouzanfar MH, Fowkes 11. Frymoyer JW, Pope MH, Clements JH, Wilder DG, MacPherson FG, Fransen M, Freeman MK, Gabbe BJ, Gabriel SE, Gakidou E, B, Ashikaga T, Vermont B (1983) Risk factors in low-back pain: Ganatra HA, Garcia B, Gaspari F, Gillum RF, Gmel G, Gonzalez- An epidemiological survey. J B Join Surg 65-A(2):213–218 Medina D, Gosselin R, Grainger R, Grant B, Groeger J, Guille- 12. Kerr MS, Frank JW, Shannon HS, Norman RWK, Wells RP, min F, Gunnell D, Gupta R, Haagsma J, Hagan H, Halasa YA, Neumann WP (2001) Biomechanical and psychosocial risk fac- Hall W, Haring D, Haro JM, Harrison JE, Havmoeller R, Hay RJ, tors for low back pain at work. Am J Pub Health 91(7): Higashi H, Hill C, Hoen B, Hoffman H, Hotez PJ, Hoy D, Huang 1069–1075 JJ, Ibeanusi SE, Jacobsen KH, James SL, Jarvis D, Jasrasaria R, 13. Lee P, Helewa A, Goldsmith CH, Smythe HA, Stitt LW (2000) Jayaraman S, Johns N, Jonas JB, Karthikeyan G, Kassebaum N, Low back pain: prevalence and risk factors in an industrial set- Kawakami N, Keren A, Khoo JP, King CH, Knowlton LM, ting. J Rheumatol 28(2):346–351 Kobusingye O, Koranteng A, Krishnamurthi R, Laden F, Lalloo 14. Marras W, Lavender S, Leurgans S, Fathallah F, Ferguson S, R, Laslett LL, Lathlean T, Leasher JL, Lee YY, Leigh J, Lev- Allread W, Rajulu S (1995) Biomechanical risk factors for inson D, Lim SS, Limb E, Lin JK, Lipnick M, Lipshultz SE, Liu occupationally related low back disorders. Ergonomics 38(2): W, Loane M, Ohno SL, Lyons R, Mabweijano J, MacIntyre MF, 377–410 Malekzadeh R, Mallinger L, Manivannan S, Marcenes W, March 15. Murtezani A, Ibraimi Z, Sllamniku S, Osmani T, Sherifi S (2011) L, Margolis DJ, Marks GB, Marks R, Matsumori A, Matzopoulos Prevalence and risk factors for low back pain in industrial R, Mayosi BM, McAnulty JH, McDermott MM, McGill N, workers. Folia Med 53(3):68–74 McGrath J, Medina-Mora ME, Meltzer M, Mensah GA, Merri- 16. Nieuwenhuyse AV, Fatkhutdinova L, Verbeke G, Pirenne D, man TR, Meyer AC, Miglioli V, Miller M, Miller TR, Mitchell Johannik K, Somville PR, Mairiaux P, Moens GF, Masschelein R PB, Mock C, Mocumbi AO, Moffitt TE, Mokdad AA, Monasta L, (2004) Risk factors for first-ever low back pain among workers in Montico M, Moradi-Lakeh M, Moran A, Morawska L, Mori R, their first employment. Occup Med 54:513–519 Murdoch ME, Mwaniki MK, Naidoo K, Nair MN, Naldi L, 17. Smedley J, Egger P, Cooper C, Coggon D (1995) Manual han- Narayan KM, Nelson PK, Nelson RG, Nevitt MC, Newton CR, dling activities and risk of low back pain in nurses. Occup Nolte S, Norman P, Norman R, O’Donnell M, O’Hanlon S, Environ Med 52:160–163 Olives C, Omer SB, Ortblad K, Osborne R, Ozgediz D, Page A, 18. Troup JDG (1984) Causes, prediction and prevention of back pain Pahari B, Pandian JD, Rivero AP, Patten SB, Pearce N, Padilla at work. Scand J Work Environ Health 10:419–428 RP, Perez-Ruiz F, Perico N, Pesudovs K, Phillips D, Phillips MR, 19. Vandergrift JL, Gold JE, Hanlon A, Punnett L (2012) Physical and Pierce K, Pion S, Polanczyk GV, Polinder S, Pope CA, 3rd, psychosocial ergonomic risk factors for low back pain in automo- Popova S, Porrini E, Pourmalek F, Prince M, Pullan RL, Ramaiah bile manufacturing workers. Occup Environ Med 69:29–34 KD, Ranganathan D, Razavi H, Regan M, Rehm JT, Rein DB, 20. Vindigni D, Walker BF, Jamison JR, Da Costa C, Parkinson L, Remuzzi G, Richardson K, Rivara FP, Roberts T, Robinson C, De Blunden S (2005) Low back pain risk factors in a large rural Leon FR, Ronfani L, Room R, Rosenfeld LC, Rushton L, Sacco Australian Aboriginal community. An opportunity for managing RL, Saha S, Sampson U, Sanchez-Riera L, Sanman E, Schwebel co-morbidities? Chiropr Osteopathy 13(1):21 DC, Scott JG, Segui-Gomez M, Shahraz S, Shepard DS, Shin H, 21. Griffith LE, Shannon HS, Wells RP, Walter SD, Cole DC, Cote P, Shivakoti R, Singh D, Singh GM, Singh JA, Singleton J, Sleet Frank J, Hogg-Johnson S, Langlois LE (2012) Individual par- DA, Sliwa K, Smith E, Smith JL, Stapelberg NJ, Steer A, Steiner ticipant data meta-analysis of mechanical workplace risk factors T, Stolk WA, Stovner LJ, Sudfeld C, Syed S, Tamburlini G, and low back pain. Am J Public Health 102:309–318 Tavakkoli M, Taylor HR, Taylor JA, Taylor WJ, Thomas B, 22. Hoogendoorn WE, van Poppel MNM, Bongers PM, Koes BW, Thomson WM, Thurston GD, Tleyjeh IM, Tonelli M, Towbin JA, Bouter LM (2000) Systematic review of psychosocial factors at Truelsen T, Tsilimbaris MK, Ubeda C, Undurraga EA, van der work and private life as risk factors for back pain. Spine Werf MJ, van Os J, Vavilala MS, Venketasubramanian N, Wang 25(16):2114–2125 M, Wang W, Watt K, Weatherall DJ, Weinstock MA, Weintraub 23. Janwantanakul P, Sitthipornvorakul E, Paksaichol A (2012) Risk R, Weisskopf MG, Weissman MM, White RA, Whiteford H, factors for the onset of nonspecific low back pain in office Wiebe N, Wiersma ST, Wilkinson JD, Williams HC, Williams workers: a systematic review of prospective cohort studies. SR, Witt E, Wolfe F, Woolf AD, Wulf S, Yeh PH, Zaidi AK, J Manip Physiol Ther 35:568–577 Zheng ZJ, Zonies D, Lopez AD, AlMazroa MA, Memish ZA 24. Linton SJ (2000) A review of psychological risk factors in back (2012) Disability-adjusted life years (DALYs) for 291 diseases and neck pain. Spine 25(9):1148–1156 and injuries in 21 regions, 1990–2010: a systematic analysis for 25. Ramond A, Bouton C, Richard I, Roquelaure Y, Baufreton C, the Global Burden of Disease Study 2010. Lancet 380 Legrand E, Huez J (2011) Psychosocial risk factors for chronic (9859):2197–2223. doi:10.1016/s0140-6736(12)61689-4 low back pain in primary care—a systematic review. Fam Pract 6. Hoy D, March L, Brooks P, Woolf A, Blyth F, Vos T, Buchbinder 28:12–21 R (2010) Measuring the global burden of low back pain. Best 26. Adams M, Mannion AF, Dolan P (1999) Personal risk factors for Pract Res Clin Rheumatol 24:155–165 first-time low back pain. Spine 24(23):2497–2505

123 55 Eur Spine J (2014) 23:512–519 519

27. Chen S, Liu M, Cook J, Bass S, Lo SK (2009) Sedentary lifestyle women: the UK Twin Spine Study. Ann Rheum Dis as a risk factor for low back pain: a systematic review. Int Arch 70:1740–1745 Occup Environ Health 82:797–806 33. Ferguson SA, Marras WS (1997) A literature review of low back 28. Leboeuf-Yde C (2000) Body weight and low back pain: a sys- disorder surveillance measures and risk factors. Clin Biomech tematic literature review of 56 journal articles reporting on 65 12(4):211–226 epidemiologic studies. Spine 25(2):226–237 34. Davis K, Heaney C (2000) The relationship between psychosocial 29. Leboeuf-Yde C (2004) Back pain—individual and genetic fac- work characteristics and low back pain: underlying methodo- tors. J Electromyogr Kinesiol 14:129–133 logical issues. Clin Biomech 15(6):389–406 30. Reisbord LS, Greenland S (1985) Factors associated with self- 35. Steffens D, Ferreira ML, Maher CG, Latimer J, Koes BW, Blyth reported back-pain prevalence: a population-based study. FM, Ferreira PH (2012) Triggers for an episode of sudden onset J Chronic Dis 38(8):691–702 low back pain: study protocol. BMC Musculoskelet Disord 31. Wang S, Dezinno P, Maranets I, Berman M, Caldwell-Andrews 13(1):7. doi:10.1186/1471-2474-13-7 AA, Kain ZN (2004) Low back pain during pregnancy: preva- 36. Viera AJ, Garrett JM (2005) Understanding interobserver lence, risk factors, and outcomes. Am Coll Obstet Gynecol agreement: the kappa statistic. Fam Med 37(5):360–363 104(1):65–70 37. Hoy D, Brooks P, Blyth F, Buchbinder R (2010) The epidemi- 32. Livshits G, Popham M, Malkin I, Sambrook PN, MacGregor AJ, ology of low back pain. Best Pract Res Clin Rheumatol Spector T, Williams FMK (2011) Lumbar disc degeneration and 24:769–781 genetic factors are the main risk factors for low back pain in

123 56

Chapter Three

Triggers for an episode of sudden onset low back pain: study protocol

Chapter Three is published as:

Steffens D, Ferreira ML, Maher CG, Latimer J, Koes BW, Blyth FM, Ferreira PH. Triggers for an episode of sudden onset low back pain: study protocol. BMC Musculoskeletal Disorders. 2012; 24:13-17. 57

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “Triggers for an episode of sudden onset low back pain: study protocol”, we confirm that Daniel Steffens has made the following contributions:

 Conception and design of the research  Writing of the manuscript and critical appraisal of the content

Manuela L Ferreira Date: 01.01.2015

Christopher G Maher Date: 01.01.2015

Jane Latimer Date: 01.01.2015

Bart W Koes Date: 01.01.2015

Fiona M Blyth Date: 01.01.2015

Paulo H Ferreira Date: 01.01.2015

58 Steffens et al. BMC Musculoskeletal Disorders 2012, 13:7 http://www.biomedcentral.com/1471-2474/13/7

STUDYPROTOCOL Open Access Triggers for an episode of sudden onset low back pain: study protocol Daniel Steffens1, Manuela L Ferreira1, Christopher G Maher1*, Jane Latimer1, Bart W Koes2, Fiona M Blyth3 and Paulo H Ferreira4

Abstract Background: Most research on risk factors for low back pain has focused on long term exposures rather than factors immediately preceding the onset of low back pain. The aim of this study is to quantify the transient increase in risk of a sudden episode of low back pain associated with acute exposure to a range of common physical and psychological factors. Methods/design: This study uses a case-crossover design. One thousand adults with a sudden onset of low back pain presenting to primary care clinicians will be recruited. Basic demographic and clinical information including exposure to putative triggers will be collected using a questionnaire. These triggers include exposure to hazardous manual tasks, physical activity, a slip/trip or fall, consumption of alcohol, sexual activity, being distracted, and being fatigued or tired. Exposures in the case window (0-2 hours from the time when participants first notice their back pain) will be compared to exposures in two control time-windows (one 24-26 hours and another 48-50 hours before the case window). Discussion: The completion of this study will provide the first-research based estimates of the increase in risk of a sudden episode of acute low back pain associated with transient exposure to a range of common factors thought to trigger low back pain.

Background fruitful: Cochrane reviews of workplace interventions Nearly 4 million people in Australia suffer from back [5], insoles [6] and lumbar supports [7] have failed to pain at any one time [1], with total treatment costs support these traditional back pain prevention exceeding $1 billion a year [2]. In the US, the figure is approaches. Prevention strategies have to date been lar- an astonishing US$32 billion a year [3]. Back complaints gely based on controlling long-term exposure to risk are the seventh most common condition in patients factors, for example, modifying seats to control vibration consulting general practitioners in Australia, and the in truck drivers. However it is likely that the full poten- most common musculoskeletal condition [4]. It is also tial of prevention will not be reached unless we also the most common health problem for which an imaging consider commonly occurring, modifiable risk factors test is ordered by a general practitioner [4]. that happen just before the onset of back pain. In this A potential solution to managing the problem of low regard we see our proposed research as complementary back pain is the identification and control of modifiable to, rather than in conflict with, research evaluating long risk factors. This approach is appealing and seemingly term risk factors. logical and there are notable examples where such an The existence of short term risk factors or ‘triggers’ is approach has provided major improvements in public consistent with the time course of back pain. It is well health. For back pain this approach has not yet been established that most people will experience low back pain in their lifetime [8], that pain is typically recurrent [9] and that episodes are usually of sudden onset [10]. * Correspondence: [email protected] 1Musculoskeletal division, The George Institute for Global Health, Sydney For example research conducted by our group demon- Medical School, The University of Sydney, PO Box M201, Missenden Road, strated that in an inception cohort of 969 subjects, 82% Sydney, New South Wales 2050, Australia reported that their onset of low back pain was sudden Full list of author information is available at the end of the article

© 2012 Steffens et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. 59 Steffens et al. BMC Musculoskeletal Disorders 2012, 13:7 Page 2 of 5 http://www.biomedcentral.com/1471-2474/13/7

[10]. This pattern of low back pain suggests that rather Cases will be identified from patients presenting to pri- than solely looking at long term exposure to risk factors mary care seeking treatment for an episode of sudden it would be instructive to also look closely at events onset, acute, low back pain. In the case crossover design occurring immediately prior to the episode to identify the time of the onset of low back pain is identified and modifiable triggers to the episode. This information is then data are obtained on exposure to a series of possi- routinely sought by health practitioners when a patient ble risk factors in the two hour period prior to the with low back pain seeks care. The treating clinician onset of low back pain (case window). Additional data commonly asks the patient what activity they were per- are obtained on exposures to the same set of possible forming just prior to the onset of pain, in essence, “was risk factors in an earlier period (24-26 and 48-50 hours the episode triggered by something unusual that hap- prior to the case window) that did not precede an epi- pened just before?” The scientific method best suited to sode of low back pain (these are referred to as the con- answer this question is the case-crossover design [11]. trol windows). We will use the case-crossover design to provide the The study has been approved by the Human Research first accurate estimates of the transient increase in risk Ethics Committee at the University of Sydney (protocol of low back pain associated with transient exposure to number 05-2011/13742) and has received funding from various triggers. It is possible that we will identify sev- Australia’s National Health and Medical Research Coun- eral factors that are not modifiable but this information cil (application ID APP1003608). will be extremely important to our understanding and explanation of the causes of low back pain. Study Participants One thousand consecutive patients (study participants) Study Aims presenting to primary care clinicians (general medical 1) To quantify the transient increase in risk of an epi- practitioners, physiotherapists, chiropractors and phar- sode of sudden onset, acute, low back pain associated macists) for treatment of an episode of sudden onset, with exposure to a range of common physical and psy- acute, low back pain will be recruited in Sydney, Austra- chological factors listed in Table 1. lia. Primary care clinicians will be trained individually or 2) To determine if habitual physical activity moderates in small groups on the study methods and procedures. the transient increase in risk of an episode of sudden Study participants must be 18 years of age (or older) to onset, acute, low back pain associated with exposure to participate. the physical and psychological factors listed above. To be eligible to enter the study participants must meet the criteria below: Methods/Design The study will use the case-crossover design. The case- ▪ Comprehends spoken English; crossover design enables quantification of the risk asso- ▪ Primary complaint of pain in the area between the ciated with transient exposures [12]. It is more efficient 12th rib and buttock crease, with or without leg pain; than cohort designs because it samples only cases, and ▪ Pain at least moderate intensity during the first 24 may be less exposed to selection bias than case-control hours of the episode (assessed using a modified ver- designs because cases provide their own control data. sion of item 7 of the SF36);

Table 1 Factors that may trigger an episode of low back pain to be evaluated in the study Physical Factors Hazardous manual tasks: - tasks involving heavy loads; - tasks involving awkward postures; - tasks involving objects that could not be positioned close to the body; - tasks involving live people or animals; - tasks involving loads that are unstable, unbalanced or difficult to grasp or hold; Vigorous physical activity Moderate physical activity A slip/trip or fall Consumption of alcohol Sexual activity Psychological Being distracted Factors Being fatigued or tired 60 Steffens et al. BMC Musculoskeletal Disorders 2012, 13:7 Page 3 of 5 http://www.biomedcentral.com/1471-2474/13/7

▪ Presentation for treatment within 7 days from the pain. Where possible, using a diary, calendar and/or time of pain onset; smartphone, we will then ask them to recall what they ▪ Not have known or suspected serious spinal were doing in the three days leading up to the onset of pathology (eg metastatic, inflammatory or infective their back pain and also on the day of their back pain. diseases of spine, cauda equina syndrome, spinal Following this we will ask about exposure to the pre- fracture); viously mentioned putative triggers. Where subjects respond affirmatively we collect detailed information on An episode of acute low back pain will be defined as the trigger, time and duration in free text. We will also an episode preceded by a period of at least one month ask the study participant to consider what they think without low back pain where the participant was not mayhavetriggeredtheirLBPandsimilarlyrecord consulting a health care practitioner or continuing with detailed information on the nominated trigger, time and medication for their low back pain (in accordance with duration in free text. the De Vet et al. [13] definition of an ‘episode’ of acute When asking the study participant about exposure to low back pain). A sudden onset episode of low back specific triggers we have developed a script to lead the pain will be defined as pain of at least moderate inten- interview (see additional file 2). sity that developed over the first 24 hours (assessed using a modified version of item 7 of the SF36). Blinding To describe further the cohort of study clinicians, we Clinicians and study participants will be blinded to the will collect descriptive data, including the clinician’sage, case and control periods. The study questionnaire is contact details, current position and past clinical experi- designed to investigate exposure to triggers over a ence. Secondly, we are collecting information regarding longer time period than will be used in the analysis so what clinicians in general consider as possible triggers that participants in the trial remain blind to the dura- for a new episode of back pain. Based on their clinical tion of the case and control windows. For example par- experience, they are asked to list the five most likely ticipants will be asked about their exposures for three triggers for a sudden onset episode of low back pain. days preceding their back pain and also on the day of They will consider both (i) short term and (ii) long term their back pain. A random sample of telephone calls will exposures (see additional file 1). These data will be used be audited and the congruency of the log and telephone to inform the categorisation of putative risk factors in call checked by the investigators. Data entry into the the analyses of our participant data, and to assess database will be conducted by a separate person who whether opinions of the study participants regarding will be blinded to all putative risk factors. Blinding may possible triggers for their low back pain are analogous be less important in case-crossover designs than case- with their clinicians’ perceptions. control studies because in the case-crossover design participants report exposure to triggers in both the case and Participant recruitment control windows. Recall bias can only occur if there is Patients seeking care for acute low back pain that fulfil differential mis-reporting in the case and control win- the inclusion criteria and agreeing to participate will be dows. In our opinion this is unlikely. referred to the study and their details (screening form and consent form) will be sent by fax to the study office. Statistical analysis A study researcher will receive the fax and contact the The analyses follows standard methods for stratified participant as soon as possible to perform the study analyses [12] with the individual subject the stratifying interview. Patients not abletoanswerthestudyques- variable in a case-crossover design. The estimates of tionnaire in seven days from the time their clinician relative risk are based on the ratio of the observed fre- referred them to the study will be excluded. quency of exposure to each of the transient triggers dur- Prior to the study interview, the researchers will dou- ing the case period, to the expected frequency of ble-check the eligibility criteria and explain the nature exposure during the two control periods. This is known of the study to the participant. Participants are able to as a matched-pair interval approach where contrasts are withdrawn from the study at any time. made between a pair of case control periods contributed by the same subject. In our proposed study there will be Study interview two matched-pair intervals. The interview is divided into two parts. In the first part To analyse the matched-pair interval data we will use we will collect basic demographic and clinical data and standard methods for case-control data (Mantel-Haens- in the second part, we will collect information on puta- zel estimator). Instead of having concordant and discor- tive triggers (see additional file 2). We will record the dant pairs of subjects, the pairs will consist of two date and time when the patient first noticed their back intervals for each subject, a case period (2 hours prior 61 Steffens et al. BMC Musculoskeletal Disorders 2012, 13:7 Page 4 of 5 http://www.biomedcentral.com/1471-2474/13/7

to the event) and a control period (24-26 hours prior to Additional material the event). A subject’s pair of intervals will either be concordant or discordant with respect to each of the Additional file 1: Clinicians’ questionnaire. Questionnaire to be triggers nominated on the item list. Ninety-five percent applied to describe further the cohort of study clinicians. confidence intervals will be computed by exact methods Additional file 2: Study Participants’ Questionnaire. Questionnaire to be applied to the study participants. based on the binomial distribution. Comparison with the first control period will form the primary analyses. Secondary analyses will be performed as described above but using the second control period (48-50 hours Acknowledgements The TRIGGERS study team includes Anurina Das. We would like to prior to the event) as the control data for the matched- acknowledge the valuable contribution of Qiang Li to the study statistical pair interval. analysis plan. The National Health and Medical Research Council (NHMRC), Australia, provides funding for this study.

Sample size Author details The study was designed to be adequately powered for 1Musculoskeletal division, The George Institute for Global Health, Sydney the primary analysis, which involves estimation of the Medical School, The University of Sydney, PO Box M201, Missenden Road, Sydney, New South Wales 2050, Australia. 2Department of General Practice, risk associated with transient exposure to the different Erasmus MC, PO Box 2040, 3000 CA Rotterdam, The Netherlands. 3Centre for types of triggers. We calculated the sample size neces- Education and Research on Ageing, The University of Sydney, C 22 - 4 sary for a paired case-control study using the procedures Concord Hospital, Sydney, New South Wales 2006, Australia. Discipline of Physiotherapy, Faculty of Health Sciences, The University of Sydney, PO Box described by Dupont [14]. This showed that in a con- 170, Lidcombe 1825, Syndey, New South Wales, Australia. ventional paired case-control design with alpha set at 0.05 we would need a sample of 1,000 cases to provide Authors’ contributions CGM, JL, MLF, BWK, FMB and PHF are the principal investigators - together an 80% probability of detecting an odds ratio of 1.5 or they conceived and designed the trial and procured funding. DS and MLF greater across the plausible range of exposure preva- drafted the first version of the manuscript. All authors contributed to the lence’s in control windows (0.2 to 0.8) and plausible writing of the manuscript. All authors read and approved the final version of the manuscript. range of correlations between exposure in case and control windows (0.0 to 0.5). Competing interests The authors declare that they have no competing interests. Prof Maher and A/Prof Latimer’s fellowships are funded by the Australian Research Council. Discussion This case cross-over study will provide the first-research Received: 24 December 2011 Accepted: 24 January 2012 based estimates of the transient increase in risk of a Published: 24 January 2012 sudden onset, acute, episode of low back pain associated References with transient exposures to a range of physical and psy- 1. Australian Institute of Health and Welfare: Arthritis and musculoskeletal chological factors. We anticipate that we will identify conditions in Australia. Canberra: Australian Institute of Health and Welfare; several modifiable factors that are triggers for an episode 2005. 2. Walker B, Muller R, Grant W: Low back pain in Australian adults: the of low back pain. This information will be invaluable in economic burden. Asia Pac J Public Health 2003, 15(2):79-87. designing future prevention strategies, and enabling clin- 3. Agency for Healthcare Research and Quality: Total Expenses and Percent icians to give evidence based advice to patients keen to Distribution for Selected Conditions by Type of Service: United States. Medical Expenditure Panel Survey Household Component Data; 2005. avoid future episodes of low back pain. 4. Britt H, Miller G, Knox S, Charles J, Pan Y, Henderson J, Baryram C, Valenti L, Recall bias is a major limitation of retrospective stu- Ng A, O’Halloran J: General practice activity in Australia 2004-2005. dies. Participants may under or overestimate the usual Canberra: Australian Institute of Health and Welfare; 2005. 5. Van Oostrom S, Driessen M, de Vet H, Franche R, Schonstein E, Loisel P, Van frequency of exposures to the time of the injury (case Mechelen W, Anema J: Workplace interventions for preventing work window). They may also under or overestimate exposure disability. Cochrane Database of Systematic Reviews 2009. in the case control windows because of memory lapse or 6. Sahar T, Cohen MJ, Ne’eman V, Kandel L, Odebiyi D, Lev I, Brezis M, Lahad A: Insoles for prevention and treatment of back pain. Cochrane difficulty in estimating exposure. In this study, partici- Database of Systematic Reviews 2007, , 4: CD005275. pants must have presented for care within seven days of 7. Van Duijvenbode I, Jellema P, Van Poppel MN, Van Tulder MW: Lumbar the onset of the injury to facilitate recall of activities. In supports for prevention and treatment of low back pain. Cochrane Database of Systematic Reviews 2008, 2. addition, our questionnaire asks participants to use 8. Waddell G: The Back Pain Revolution. Edinburgh: Churchill Livingstone;, 2 prompts such as referring to their agenda, calendar and/ 2004. or smartphones to stimulate their memory of the activ- 9. Von Korff M, Saunders K: The course of back pain in primary care. Spine 1996, 21(24):2833-2839. ities they performed in the days prior to the onset of 10. Henschke N, Maher CG, Refshauge KM: A systematic review identifies five their low back pain. red flags to screen for vertebral fracture in patients with low back pain. The completion of this trial is expected by early 2014. J Clin Epidemiol 2008, 61(2):110-118. 11. Maclure M, Mittleman M: Should we use a case-crossover design? Annu Rev Public Health 2000, 21:193-221. 62 Steffens et al. BMC Musculoskeletal Disorders 2012, 13:7 Page 5 of 5 http://www.biomedcentral.com/1471-2474/13/7

12. Maclure M: The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol 1991, 133:144-153. 13. de vet H, Heymans M, Dunn K, Pope D, van der Beek A, Macfarlane G, Bouter L, Croft P: Episodes of low back pain. A proposal for uniform definitions to be used in research. Spine 2002, 27(21):2409-2416. 14. Dupont W: Power calculations for matched case-control studies. Biometrics 1988, 44:1157-1168.

Pre-publication history The pre-publication history for this paper can be accessed here: http://www.biomedcentral.com/1471-2474/13/7/prepub

doi:10.1186/1471-2474-13-7 Cite this article as: Steffens et al.: Triggers for an episode of sudden onset low back pain: study protocol. BMC Musculoskeletal Disorders 2012 13:7.

Submit your next manuscript to BioMed Central and take full advantage of:

• Convenient online submission • Thorough peer review • No space constraints or color ﬁgure charges • Immediate publication on acceptance • Inclusion in PubMed, CAS, Scopus and Google Scholar • Research which is freely available for redistribution

Submit your manuscript at www.biomedcentral.com/submit 63 Additional file 1 – Clinicians’ Questionnaire

Name: ______

Gender: □ Female □ Male Date of Birth: ______

Address: ______

Phone:______FAX: ______

Email:______

Profession (tick one)

 physiotherapist  medical practitioner  pharmacist Current Position:______

Clinical experience

Years as practicing clinician ______

Years managing low back pain ______

1. Based on your clinical experience, list what you consider to be the five most likely factors involving short term exposure that are triggers for a sudden episode of acute low back pain? (E.g. in my clinical experience, running 15km on the road with poor shoes can trigger an episode of shin splints).

1)______2)______3)______4)______5)______

2. Based on your clinical experience, list what you consider to be the five most likely factors involving long term exposure that increase the risk of a sudden episode of acute low back pain (e.g. in my clinical experience working with a ‘poke neck’ posture increases the risk of neck pain and headaches).

1)______2)______3)______4)______5)______64 Appendix 5. Study Questionnaire

Triggers Specific Questions Tick the boxes where the participant reports pain on the mannequin below:

Height (cm): ______Weight (Kg): ______

How much back pain have you had during the first 24 hours of this episode? [1] None Very mild Mild Moderate Severe Very Severe      

During the first 24 hours of this episode how much did back pain interfere with your normal work (including both work outside the home and housework)? [1] Not at all A little bit Moderately Quite a bit Extremely     

How tense or anxious have you felt in the past week? Circle one. [2]

0 1 2 3 4 5 6 7 8 9 10 Calm and As relaxed tense/anxious as I’ve ever felt

Physical Activity [3]

1. In the last week, how many times have you walked continuously, for at least 10 minutes, for recreation, exercise or to get to or from places? ______times; And in the week before your low back pain started? ______times.

What do you estimate was the total time that you spent walking in this way in the last week? ______hours ______minutes; And in the week before your low back pain started? ______hours ______minutes.

2. In the last week, how many times did you do any vigorous physical activity which made you breathe harder or puff and pant? (e.g. jogging, cycling, aerobics, competitive tennis)______times; And in the week before your low back pain started? ______times.

What do you estimate was the total time that you spent doing this vigorous physical activity in the last week? ______hours ______minutes; And in the week before your low back pain started? ______hours ______minutes.

3. In the last week, how many times did you do any moderate physical activities that you have not already mentioned? (e.g. gentle swimming, social tennis, golf)______times; And in the week before your low back pain started? ______times.

What do you estimate was the total time that you spent doing these activities in the last week? ______hours ______minutes; And in the week before your back pain started? ______hours ______minutes.

4. Was your level of physical activity last week typical for you?  Yes  No

Triggers Exposure

Please write the date and time you first noticed your low back pain on the table below:

Day Fri Sat Sun Mon Tues Wed Thurs Fri Sat Sun

Date

Time

I am going to ask you to recall what you were doing in the three days leading up to your back pain and also on the day of your back pain. To help your memory I would like you to sit down with your diary and smartphone. To help make sure we have the right days I want you to tell me the day, weather and a key thing you did on each day.

Example: Tuesday: cold and wet; visited parents

Day of back pain: ______

Day before: ______

2 days earlier: ______

3 days earlier: ______

If yes, precisely describe manual task (type of task, 1a. MANUAL TASKS … load, duration and time). e.g. Lifted 50 large boxes, HEAVY LOADS [4] Day one at a time, (~15kg each box – perceived as heavy) from the floor and placed them on a bench at waist height. 8:00am - 20min. The first group of questions is about manual tasks. Manual tasks include lifting, lowering, pushing, Day of back carrying or otherwise moving, pain holding or restraining any person, animal or item.

Firstly we are interested in manual tasks involving HEAVY LOADS.  Yes  No So on the day of your back pain did you engage in any manual tasks involving a heavy load?

That was the (restate their description of the day). Day before

Now what about the day before…. That was the (restate their description of the day).  Yes  No

2 days earlier

OK now two days back That was the (restate their description of the day).  Yes  No

3 days earlier

OK finally three days back That was the (restate their description of the  Yes  No day)

If yes, precisely describe manual task and posture (type 1b. MANUAL TASKS… Day of task, load, duration, time & body position). e.g. knelt AWKWARD POSTURE [4] down while gardening. 2:00pm – 40min. Now I want you to think about manual tasks involving an AWKWARD POSTURE. So on the day of your back pain did Day of back you engage in any manual tasks pain involving an awkward position?

That was the (restate their description of the day).  Yes  No

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back That was the 2 days (restate their description of the day). earlier

 Yes  No

OK finally three days back That was 3 days the (restate their description of the earlier day)

 Yes  No

If yes, precisely describe manual task and posture (type 1c. MANUAL TASKS… of task, load, duration, time and body position). e.g. AN OBJECT THAT COULD Lifted large box (~7Kg – perceived as light) out of the NOT BE POSITIONED CLOSE Day car boot and placed it on the floor. 11:00am – 10 sec. TO THE BODY [4] Now I want you to think about manual tasks involving AN OBJECT THAT COULD NOT BE POSITIONED CLOSE TO THE Day of back BODY. pain So on the day of your back pain did you engage in any manual tasks involving an object that could not be positioned close to the body?  Yes  No That was the (restate their description of the day).

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back That was the (restate their description of the day). 2 days earlier

 Yes  No

OK finally three days back That was 3 days the (restate their description of the earlier day)

 Yes  No

If yes, precisely describe manual task and posture (type of task, load, duration, time and body position). e.g. 1d. MANUAL TASKS… Lifted 2 year-old child (~12kg – perceived as moderate) Day LIVE PEOPLE OR ANIMALS [4] from the floor onto the bed. 8:00pm once. Lifted and carried baby (5 kg perceived as light), 3 to 4 times from one room to another. 8:10pm – 30min Now I want you to think about manual tasks involving LIVE PEOPLE OR ANIMALS. So on the day of your back pain did Day of back you engage in any manual tasks pain involving live people or animals? That was the (restate their description of the day).  Yes  No

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

2 days OK now two days back… earlier That was the (restate their description of the day).

 Yes  No

3 days OK finally three days back… earlier That was the (restate their description of the day)

 Yes  No

1e. MANUAL TASKS… If yes, precisely describe manual task and posture (type A LOAD THAT WAS of task, load, duration, time and body position). e.g. UNSTABLE, UNBALANCED OR Day lifted a 4 meter extension ladder (~12 Kg – perceived as DIFFICULT TO GRASP OR moderate) from the car roof racks, carried to garage HOLD [4] and hung on wall. 3:00pm - 5min. Now I want you to think about manual tasks involving A LOAD THAT WAS UNSTABLE, UNBALANCED OR DIFFICULT Day of back TO GRASP OR HOLD. pain So on the day of your back pain did you engage in any manual tasks involving a load that was unstable, unbalanced or difficult to grasp or hold?  Yes  No That was the (restate their description of the day).

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back… That was the (restate their 2 days description of the day). earlier

 Yes  No

3 days OK finally three days back… earlier That was the (restate their description of the day)

 Yes  No

If yes, precisely describe activity/task, time and 2a. VIGOROUS Day duration. e.g. ran 10km at fast pace (5 mins per km) PHYSICAL ACTIVITY 10:00am - 50min. The next questions are about VIGOROUS PHYSICAL ACTIVITY. This could be sports or hobbies, paid or volunteer work, work outside the home and Day of back housework. pain

Examples of vigorous physical activity include: running, rope skipping, axe chopping, using heavy  Yes  No tools, canoeing and truck driving. So on the day of your back pain did you engage in any manual tasks involving VIGOROUS PHYSICAL ACTIVITIES? Day before That was the (restate their description of the day).

Now what about the day before…  Yes  No That was the (restate their description of the day).

OK now two days back… That was 2 days the (restate their description of the earlier day).

 Yes  No

OK finally three days back… That was the (restate their description of 3 days the day) earlier

 Yes  No

2b. MODERATE If yes, precisely describe activity/task, time and Day PHYSICAL ACTIVITY duration. e.g. Mowed the lawn. 11:00am -12:00 pm. The next questions are about MODERATE PHYSICAL ACTIVITY. This could be sports or hobbies, paid or volunteer work, work outside the home and Day of back housework. pain

Examples of moderate physical activity include: leisure cycling, fishing, general home repairs, music  Yes  No playing, golf, surfing and painting. So on the day of your back pain did you engage in any manual tasks involving MODERATE PHYSICAL ACTIVITIES? That was the (restate their Day before description of the day).

Now what about the day before…. That was the (restate their description of the day).  Yes  No

OK now two days back… That was 2 days the (restate their description of the earlier day).

 Yes  No

OK finally three days back… That 3 days was the (restate their description of earlier the day).

 Yes  No

If yes, precisely describe incident and time. e.g. 3. SLIP, TRIP OR FALL [5] Day Descending stairs, missed bottom step and jarred back. 8:30am – one occasion. Now I want you to think about SLIP, TRIP OR FALL. So on the day of your back pain did you have a slip, trip or fall? That Day of back was the (restate their description of pain the day).

 Yes  No

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back… That was the (restate their 2 days description of the day). earlier

 Yes  No

OK finally three days back…. That was the (restate their 3 days description of the day) earlier

 Yes  No

If yes, specify amount (refer to standard drink table at 4. CONSUMED ALCOHOL [6] Day the end of booklet), time and duration. E.g. 2x red wine glasses (180ml). 8:20pm – 1h. Now I want you to think about ALCOHOL CONSUMPTION. Day of back So on the day of your back pain did pain you consume alcohol? That was the (restate their description of the day).  Yes  No

Now what about the day before…. That was the (restate their Day before description of the day).  Yes  No OK now two days back… That was the (restate their description of the 2 days day). earlier

OK finally three days back… That  Yes  No was the (restate their description of the day) 3 days earlier

 Yes  No

5. SEXUAL ACTIVITY [7] Day If yes, specify time. E.g. 11:00pm Now I want you to think about SEXUAL ACTIVITY. So on the Day of back day of your back pain did you pain engage in sexual activity? That was the (restate their description of the  Yes  No day). Day before Now what about the day before…. That was the (restate their description of the day).  Yes  No

OK now two days back… That was 2 days the (restate their description of the earlier day).

 Yes  No OK finally three days back… That was the (restate their description of the day) 3 days earlier

 Yes  No

If yes, specify time and duration, distraction and task. e.g. Distracted by child crying while lifting a box from 6. DISTRACTION [8] Day car boot and it slipped from his/her hands. 8:30am – one occasion. Now I want you to think about being DISTRACTED. Day of back So on the day of your back pain were pain you DISTRACTED for any reason while engaged in a task or activity?  Yes  No That was the (restate their description of the day). Day before Now what about the day before…. That was the (restate their  Yes  No description of the day). 2 days OK now two days back… That was earlier the (restate their description of the day).  Yes  No

OK finally three days back… That 3 days was the (restate their description of earlier the day)  Yes  No

If yes, specify time and duration. e.g. Disrupted and 7. FATIGUE OR TIREDNESS [9] Day poor sleep the night before as youngest child kept waking due to earache. 3:00am – 24h. Now I want you to think about feeling FATIGUED or TIRED. So Day of back on the day of your back pain did you pain feel FATIGUED OR TIRED? That was the (restate their description of  Yes  No the day). Day before Now what about the day before…. That was the (restate their  Yes  No description of the day).

OK now two days back… That was 2 days the (restate their description of the earlier day).  Yes  No

OK finally three days back… That was the (restate their description of 3 days the day) earlier  Yes  No

8. What do you think may have triggered your low back pain? (Record what the patient thinks may have triggered his/her episode of back pain. eg. Bent down once to pick up newspaper from lawn. 7:00am - immediately).

______

So on the day of your back pain did you Day of back pain do the activity described above? That was the (restate their description of the day).  Yes  No

Day before Now what about the day before…. That was the (restate their description of the day).  Yes  No

OK now two days back… That was the 2 days earlier (restate their description of the day).

 Yes  No

OK finally three days back… That was the (restate their description of the day) 3 days earlier

 Yes  No

Reference

1. Ware J, Sherbourne C: The MOS 36-item short-form health survey (SF-36). 1.

Conceptual framework and item selection. Medical Care 1992, 30:473-483.

2. Linton S, Hellsing A, Bergstrom G: Exercise for workers with musculoskeletal

pain: does enhancing compliance decrease pain? J Occup Med 1996, 6(3):177-190.

3. Armstrong T, Bauman A, Davies J: Physical activity patterns of Australian Adults.

In. Edited by Welfare AIoHa. Canberra; 2000.

4. Australian Safety Compensation Council: National code of practice for the

prevention of musculoskeletal disorders from performing manual tasks at work.

In. Canberra; 2007.

5. Verma S, Lombardi D, Chang W, Courtney T, Huang Y, Brennan M, Mittleman M,

Ware J, Perry M: Rushing, distraction, walking on contaminated floors and risk

of slipping in limited-service restaurants: a case--crossover study. Occupational &

Environmental Medicine 2010, 68(8):575-581.

6. Vinson DC, Mabe N, Leonard LL, Alexander J, Becker J, Boyer J, Moll J: Alcohol

and injury. A case-crossover study. Arch Fam Med 1995, 4(6):505-511.

7. Dahabreh IJ, Paulus JK, Dahabreh IJ, Paulus JK: Association of episodic physical

and sexual activity with triggering of acute cardiac events: systematic review and

meta-analysis. JAMA, 305(12):1225-1233.

8. Sorock G, Lombardi D, Peng D, Hauser R, Eisen E, Herrick R, Mittleman M: Glove

use and the relative risk of acute hand injury: a case-crossover study. Journal of

Occupational & Environmental Hygiene 2004, 1(3):182-190.

9. Chen S, Fong P, Lin S, Chang C, Chan C: A case-crossover study on transient risk

factors of work-related eye injuries. Occupational & Environmental Medicine

2009, 66(8):517-522.

Chapter Four

What triggers an episode of low back pain? A case-crossover study

Chapter Four is published as:

Steffens D, Ferreira ML, Latimer J, Ferreira PH, Koes BK, Blyth F, Li Q, Maher CG. What triggers an episode of low back pain? A case-crossover study. Arthritis Care & Research. 2015; 67:403-410. 80

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “What triggers an episode of low back pain? A case-crossover study”, we confirm that Daniel Steffens has made the following contributions:

 Conception and design of the research  Data collection  Analysis and interpretation of the findings  Writing of the manuscript and critical appraisal of the content

Manuela L Ferreira Date: 01.01.2015

Jane Latimer Date: 01.01.2015

Paulo H Ferreira Date: 01.01.2015

Bart W Koes Date: 01.01.2015

Fiona Blyth Date: 01.01.2015

Qiang Li Date: 01.01.2015

Christopher G Maher Date: 01.01.2015

What Triggers an Episode of Acute Low Back Pain? A Case–Crossover Study

DANIEL STEFFENS,1 MANUELA L. FERREIRA,2 JANE LATIMER,2 PAULO H. FERREIRA,2 3 2 2 2 BART W. KOES, FIONA BLYTH, QIANG LI, AND CHRISTOPHER G. MAHER

Objective. To investigate a range of transient risk factors for an episode of sudden-onset, acute low back pain (LBP). Methods. This case–crossover study recruited 999 subjects with a new episode of acute LBP between October 2011 and November 2012 from 300 primary care clinics in Sydney, Australia. Each participant was asked to report exposure to 12 putative triggers over the 96 hours preceding the onset of back pain. Conditional logistic regression was used to estimate odds ratios (ORs) expressing the magnitude of increased risk with exposure to each trigger. Results. Exposure to a range of physical and psychosocial triggers significantly increased the risk of a new onset of LBP; ORs ranged from 2.7 (moderate or vigorous physical activity) to 25.0 (distracted during an activity or task). Age moderated the effect of exposure to heavy loads and sexual activity. The ORs for heavy loads for people ages 20, 40, or 60 years were 13.6, 6.0, and 2.7, respectively. The risk of developing back pain was greatest between 7:00 AM and noon. Conclusion. Transient exposure to a number of modifiable physical and psychosocial triggers substantially increases risk for a new episode of LBP. Triggers previously evaluated in occupational injury studies, but never in LBP, have been shown to significantly increase risk. These results aid our understanding of the causes of LBP and can inform the development of new prevention approaches.

INTRODUCTION whereas the role of modifiable factors more proximal to the onset of back pain is yet to be investigated. Controlling Back pain affects approximately 10% of the world’s pop- exposure to these factors may be extremely important in ulation at any point in time (1–3). When disease burden is preventing back pain. measured by disability-adjusted life years, back pain is one The case–crossover design provides an ideal method for of the 10 leading causes of disease burden globally. Back quantifying the increased risk due to transient exposure to pain is, however, unique among this list of 10 diseases triggers (7,8). In this design a person acts as their own because there has been little or no progress in identifying control and hence the potential for between-person con- effective prevention strategies (4–6). founding is reduced. This perfect matching is important in Understanding what modifiable factors increase the risk back pain research as it eliminates potential effects of of back pain is a crucial first step in prevention. Existing unmeasured confounders such as genetic and lifestyle in- research has focused on factors that are not modifiable fluences (9). The TRIGGERS study for low back pain aimed (e.g., age) or involve long-term exposure (e.g., smoking), to investigate a number of transient physical and psychosocial risk factors for an episode of sudden-onset, acute Supported by the National Health and Medical Research low back pain (LBP). Physical factors included heavy Council of Australia. 1 loads; awkward positioning; handling of objects far from Daniel Steffens, BPT: The University of Sydney, Sydney, the body; handling people or animals and unstable load- Australia, and Federal University of Minas Gerais, Minas Gerais, Brazil; 2Manuela L. Ferreira, PhD, Jane Latimer, ing; a slip, trip, or fall; engagement in moderate or vigorous PhD, Paulo H. Ferreira, PhD, Fiona Blyth, PhD, Qiang Li, physical activity; and sexual activity. Psychosocial factors PhD, Christopher G. Maher, PhD: The George Institute for included alcohol consumption and being distracted and Global Health, The University of Sydney, Sydney, Australia; fatigued. 3Bart W. Koes, PhD: Erasmus Medical Center, Rotterdam, The Netherlands. Address correspondence to Manuela L. Ferreira, PhD, The George Institute for Global Health, Sydney Medical School, The University of Sydney, PO Box M201 Missenden PATIENTS AND METHODS Road, Sydney, 2015, New South Wales, Australia. E-mail: [email protected]. Study design. The TRIGGERS study employed a case– Submitted for publication August 21, 2014; accepted in revised form December 9, 2014. crossover design to quantify the risk associated with transient exposure to modifiable triggers for back pain. Trig-

403 82

404 Steffens et al

subjects with back pain and adjustments made to improve Significance & Innovations clarity and participant recall. ● Back pain is the leading cause of disability glob- During the interview, participants were asked to identify ally, yet there has been little or no progress in the date and time of pain onset with the assistance of identifying effective prevention strategies. recommended recall aids such as a diary, calendar, and/or ● To date, no studies have examined the role of smartphone. Each participant was then asked to report exposure to transient risk factors in triggering an exposure, including its time and duration, to each of the 12 episode of acute low back pain. putative triggers in the 96 hours preceding the onset of back pain. For example, for questions about manual tasks ● The results of this study demonstrate for the first involving a heavy load would be as follows: so on the day time that brief exposure to a range of physical and of your back pain did you engage in any manual tasks psychosocial factors can considerably increase the involving a heavy load? That was the...(the interviewer risk of an episode of acute back pain. would restate the participant’s description of the day, i.e., ● These results will have significant clinical and that was the quite warm day you went to church). Now policy implications for the control of a disease that what about the day before...(the interviewer would reis a major problem worldwide. state the description of that day, i.e., the windy day when you did your weekly shopping). What about two days back. That was the...(the interviewer would restate the description of that day, i.e., the day you visited your parents and it rained all day). Participants, clinicians, and gers from the list of hazardous tasks provided in a national interviewers were blinded to the time and length of the code of practice were included (10). Additionally a num- predetermined case and control windows. ber of factors that had been previously identified as trig- Physical triggers included manual tasks (heavy loads, gers in occupational injury studies (11–14), but never eval- awkward positioning, handling of objects far from the uated in the area of back pain, were included. The body, and handling people or animals and unstable load- exposure to each trigger during the 2 hours preceding the ing); a slip, trip, or fall; engagement in moderate or vigor- onset of back pain (case window) was compared to expo- ous physical activity; and sexual activity. Physical activi- sure in two 2-hour periods ending 24 and 48 hours before the onset of back pain (control windows). The study was ties were coded as moderate or vigorous physical activity approved by the Human Research Ethics Committee of the considering their energy cost (18). Psychosocial triggers University of Sydney (protocol number 05-2011/13742). A included alcohol consumption and being distracted or fa- detailed protocol has been previously published (15). tigued. Recruiting clinicians were not aware of the specific triggers evaluated in the study. Although data were obtained for exposure to the 12 putative triggers over the 96 Participants. A total of 999 consecutive patients, age Ն18 years and with a new episode of acute back pain, were hours prior to back pain onset, the primary analysis only recruited from 300 primary care clinics in New South used data for the 2-hour period immediately preceding the Wales, Australia between October 2011 and November back pain onset and for the two 2-hour periods ending 24 2012. A new episode of back pain was defined as a primary and 48 hours before the pain onset. complaint of pain between the 12th rib and the buttock Participants were also questioned regarding their habit- crease, with or without leg pain, causing the patient to ual physical activity. This was assessed using the Active seek health care or take medication and preceded by a Australia questionnaire, which estimates the total number period of at least 1 month without back pain (16). Patients of hours of light, moderate, and vigorous physical activity needed to present to primary care within 7 days from pain performed by the participant in the past week (19). For the onset and report pain of at least moderate intensity in purpose of this study, the questionnaire was modified to the first 24 hours of the current episode (measured using include the recall period of interest, i.e., the week before item 7 of the Short Form 36 questionnaire). This inclusion the onset of back pain. Supplementary Appendix A, avail- means that we did not study patients with an insidious able in the online version of this article at http://online onset, a less common presentation for nonspecific back library.wiley.com/doi/10.1002/acr.22533/abstract, pre- pain that would be less suited to study with a case– sents the study questionnaire. crossover design (17). Patients were excluded from the study if they presented with known or suspected serious Statistical analysis. The frequency of exposure to each spinal pathology (e.g., metastatic, inflammatory, or infec- trigger was calculated for the case window (2 hours prior tive diseases of the spine). All participants gave written to onset of back pain) and 2 control windows (24–26 hours informed consent for participation. and 48–50 hours prior to onset of back pain, respectively). A frequency distribution graph of pain onset by time of Study interview. Participants were interviewed by day (in hours) was calculated. Conditional logistic regres- phone within 7 days following referral to the study. sion models were constructed to quantify the risk of back Trained research staff used an interview script to collect pain onset associated with each trigger, where each par- sociodemographic and clinical characteristics of the back ticipant represented a matched set of data for case and pain episode as well as data on exposure to a variety of control exposures (20). Odds ratios (ORs) and 95% confi- possible triggers. The interview script was piloted on 20 dence intervals (95% CIs) were derived comparing expo- 83

Triggers for Low Back Pain 405 sure in the case window with each of 2 control windows. *(999 ؍ Table 1. Characteristics of the participants (n Conditional logistic regression was also used to assess validity of the control windows by comparing their expo- Characteristics Value sure data. Secondary analyses evaluated interaction be- Ϯ tween exposure to triggers and habitual physical activity, Age, years 45.3 13.4 Male sex 541 (54.2) age, body mass index (BMI), number of previous LBP Height, cm 172.4 Ϯ 10.4 episodes, depression, and anxiety scores. Weight, kg 78.9 Ϯ 18.1† It is plausible that physical and psychosocial factors Body mass index, kg/m2‡ 26.4 Ϯ 5.2† could cause an episode of back pain that does not become Duration of current episode, days 4.9 Ϯ 2.7 Ϯ apparent in the 2 hours after the participant was exposed. Number of previous episodes 5.9 14.0 Ϯ Therefore, sensitivity analyses were conducted using case Days to seek care 3.0 2.1 Days from presentation to health care and 2.0 Ϯ 2.0 and control windows of 4- and 6-hour duration. Sensitiv- interview ity analyses using case and control windows of 1-hour Days of reduced activity 2.3 Ϯ 2.2 duration were also performed. STATA 12 software was Pain scores (0–10) 5.3 Ϯ 2.1 used for all analyses. Currently taking medication 452 (45.3) Currently employed 836 (83.7) Workers compensation 89 (8.9) Sample size. The study was designed to be adequately If in paid employment, what is done for a powered for the primary analysis, which involves estima- living tion of the risk associated with transient exposure to the Not employed 163 (16.3) different types of triggers by comparing exposures in the Clerical/administrative worker 103 (10.3) case window to the first control window. We calculated Community/personal service worker 47 (4.7) Laborer 30 (3.0) the sample size necessary for a paired case–control study Machinery operator/driver 27 (2.7) using the procedures described by Dupont (21). Sample Manager 159 (15.9) size was estimated using procedures for a paired case– Professional 341 (34.1) control design (21). With alpha set at 0.05, a sample of Sales worker 52 (5.2) 1,000 cases would provide an 80% probability of detecting Technician/trade worker 77 (7.7) an OR of 1.5 or greater across the plausible range of expo- Pain location§ Upper back 59 (5.9) sure prevalence in control windows (0.2–0.8) and the Lower back 999 (100) plausible range of correlations between exposure in case Left thigh (back) 97 (9.7) and control windows (0.0–0.5). Left leg (back) 44 (4.40) Right thigh (back) 107 (10.7) Right leg (back) 48 (4.8) RESULTS Right thigh (front) 29 (2.9) Right leg (front) 11 (1.1) Primary care clinicians referred 1,639 patients with a new Left thigh (front) 27 (2.7) episode of back pain, with 999 being interviewed and Left leg (front) 8 (0.8) contributing data to the study (Figure 1). Pain severity in first 24 hours Moderate 373 (37.3) Ϯ Severe 494 (49.5) Patterns of pain onset. The mean SD duration of the Very severe 132 (13.2) episode was 4.9 Ϯ 2.7 days, with a mean Ϯ SD time Pain interfering with work in first 24 hours between pain onset and presentation to primary care of Not at all 21 (2.1) 2.0 Ϯ 2.1 days, and 1.9 Ϯ 2.0 days from presentation to A little bit 101 (10.1) interview (Table 1). Approximately half the participants Moderately 250 (25.0) Quite a bit 389 (38.9) (49.5%) reported having severe pain in the first 24 hours of Extremely 238 (23.8) Habitual physical activity in last week¶ Sedentary 542 (54.3) Insufficient activity 164 (16.4) Sufficient activity 293 (29.3) Habitual physical activity during week before Sedentary 360 (36.0) Insufficient activity 174 (17.4) Sufficient activity 465 (46.6) Tense/anxious scores 4.0 Ϯ 2.6 Depression scores 2.7 Ϯ 2.7

* Values are the number (percentage) or the mean Ϯ SD. †Nϭ 998. ‡ Body mass index ϭ weight in kilograms divided by the square of the height in meters. § Pain location was assessed using a pain mannequin provided to the participant by the referring clinician. ¶ Habitual physical activity: moderate activity time ϩ (2 ϫ vigorous activity time). Sedentary (zero), insufficient activity (Ն1toՅ149), and sufficient activity (Ն150). Figure 1. Study recruitment flow chart. LBP ϭ low back pain. 84

406 Steffens et al

Figure 2. Frequency of back pain onset by time of day for 999 participants. Panel shows the percentage of episodes that commenced in each 1-hour time epoch across the day. the current episode, and 87.8% reported that pain inter- 2 control windows (Table 2). The highest exposure in the fered at least moderately with daily activities. Mornings case window was to manual tasks involving an awkward were the most frequent time of day for back pain onset, posture (27.4% of case windows; 7% for first control win- with 35.2% of participants (n ϭ 352) reporting pain onset dow and 5.4% for second control window), followed by between 7:00 AM and 10:00 AM (Figure 2). Only 3.7% of manual tasks involving heavy loads (17.9% in case win- participants (n ϭ 37) reported pain onset between mid- dows; 6.4% and 5.9% in the first and second control night and 5:00 AM, with a large increase in reports from windows, respectively). A total of 37 participants reported 6:00 AM. being exposed to a slip, trip, or fall in the 2 hours before pain onset, compared to only 1 participant in the first Exposure to potential triggers. Exposure to all physical control window and none in the second control window. triggers was more frequent in the case window than in the This suggests a strong association between this trigger and

Table 2. Exposure frequency and ORs for each trigger: primary analysis (2-hour window)*

Case window First control window Second control window (0–2 hours) (24–26 hours) (48–50 hours) Triggers no. (%) no. (%) no. (%) OR (95% CI)† P

Physical factors Manual tasks involving… Heavy loads 179 (17.9) 64 (6.4) 59 (5.9) 5.0 (3.3–7.4) Ͻ 0.001 Awkward posture 274 (27.4) 70 (7.0) 54 (5.4) 8.0 (5.5–11.8) Ͻ 0.001 Objects not close to the body 40 (4.0) 14 (1.4) 10 (1.0) 6.2 (2.4–15.9) Ͻ 0.001 People or animals 86 (8.6) 62 (6.2) 63 (6.3) 5.8 (2.3–15.0) Ͻ 0.001 Unstable, unbalanced, 52 (5.2) 19 (1.9) 13 (1.3) 5.1 (2.4–10.9) Ͻ 0.001 difﬁcult to grasp Moderate or vigorous physical 225 (22.5) 129 (12.9) 112 (11.2) 2.7 (2.0–3.6) Ͻ 0.001 activity Vigorous physical activity only 105 (10.5) 44 (4.4) 38 (3.8) 3.9 (2.4–6.3) Ͻ 0.001 Slip/trip/fall‡ 37 (3.7) 1 (0.1) 0 (0.0) – – Sexual activity 8 (0.8) 11 (1.1) 11 (1.1) 0.7 (0.3–1.8) 0.49 Psychosocial factors Consumption of alcohol 13 (1.3) 9 (0.9) 12 (1.2) 1.5 (0.6–3.7) 0.37 Distracted during an activity 30 (3.0) 6 (0.6) 8 (0.8) 25.0 (3.4–184.5) 0.002 or task Fatigued/tired 118 (11.8) 69 (6.9) 60 (6.0) 3.7 (2.2–6.3) Ͻ 0.001

* Results of the primary analyses based on case and control windows of 2 hours duration. OR ϭ odds ratio; 95% CI ϭ 95% conﬁdence interval. † For the primary analysis, ORs and 95% CIs were derived by comparing exposure in the case window (0–2 hours) to exposure in the ﬁrst control window (24–26 hours). ‡ Due to small frequencies of exposure in the control windows, this trigger could not be included in the conditional logistic regression analyses. 85

Triggers for Low Back Pain 407

Table 3. Inﬂuence of time of onset on risk: ORs for back pain that developed in the morning versus later in the day*

Time of onset, 7 AM to 12 PM Time of onset, 1 PM to 6 AM Interaction (558 ؍ n) (441 ؍ n) analysis, Triggers OR (95% CI) P OR (95% CI) P P†

Physical factors Manual tasks involving Heavy loads 5.3 (2.70–10.4) Ͻ 0.0001 4.8 (2.9–7.9) Ͻ 0.0001 0.813 Awkward posture 13.4 (6.5–27.4) Ͻ 0.0001 6.0 (3.8–9.5) Ͻ 0.0001 0.066 Objects not close to the body 7.5 (1.7–32.8) 0.007 5.3 (1.6–18.3) 0.008 0.728 People/animals 7.0 (1.59–30.8) 0.01 5.0 (1.5–17.3) 0.011 0.733 Unstable/unbalanced/difﬁcult to grasp 9.5 (2.2–40.8) 0.002 3.7 (1.5–9.0) 0.005 0.276 Moderate or vigorous physical activity 1.9 (1.2–3.0) 0.008 3.3 (2.2–5.0) Ͻ 0.0001 0.072 Vigorous physical activity 2.5 (1.2–5.2) 0.014 5.2 (2.7–9.9) Ͻ 0.0001 0.144 Slip/trip/fall‡ Sexual activity 1.0 (0.4–2.9) 1.0 0.3 (0.0–2.2) 0.215 0.263 Psychosocial factors Consumption of alcohol – – 1.5 (0.6–3.7) 0.374 – Distracted during an activity or task 12.0 (1.6–92.3) 0.017 Fatigued/tired 4.5 (1.9–10.9) 0.001 3.3 (1.8–6.4) Ͻ 0.0001 0.591

* Results of post hoc analyses based on case and control windows of 2-hours’ duration. OR ϭ odds ratio; 95% CI ϭ 95% confidence interval. † Interaction test between risk factors (7:00 AM to 12:00 PM and 1:00 PM to 6:00 AM). ‡ Due to small frequencies of exposure, this trigger could not be included in the conditional logistic regression analyses. onset of back pain; however, exposure frequencies were and 5.1 (95% CI 2.4–10.9) for manual tasks involving very small in the control windows and slip, trip, or fall unstable or unbalanced objects. Being exposed to physical could not be sensibly included in the regression analyses. activity of at least moderate intensity (i.e., participants Exposure frequency for being fatigued and tired was reporting being exposed to either moderate or vigorous higher in the case window (11.8%) than control windows physical activity) increased the odds of developing back (6.90% and 6.00% in the first and second control win- pain in the following 2 hours by 2.7 (95% CI 2.0–3.6) dows, respectively). However, for sexual activity and al- when compared to no exposure to physical activity. The cohol consumption, exposure frequency was similar odds increased further when participants were exposed to across case and control windows (Table 2). vigorous physical activity (i.e., only participants reporting being exposed to vigorous physical activity) compared to Association of exposure and risk of back pain. All no physical activity (OR 3.9, 95% CI 2.4–6.3). physical triggers included in the analysis were strongly Among the psychosocial triggers, being distracted dur- associated with an increased risk of back pain (Table 2). ing a task or activity (OR 25.0, 95% CI 3.4–184.5) or Exposure to manual tasks involving awkward positioning fatigued (OR 3.7, 95% CI 2.2–6.3) increased significantly was associated with 8.0 (95% CI 5.5–11.8) times greater the odds of a new onset of back pain, but alcohol consump- odds of the onset of back pain. Likewise, the OR associated tion and sexual activity showed no association with onset with exposure to manual tasks involving objects not close of back pain. to the body was 6.2 (95% CI 2.4–16.0), 5.80 (95% CI No significant change in OR estimates was observed 2.3–15.0) for manual tasks involving people or animals, when window duration was increased to 4 hours (Supple-

Table 4. ORs for combined triggers: secondary analysis (2-hour window)*

Case window First control window Second control window (0–2 hours) (24–26 hours) (48–50 hours) OR Combined Triggers no. (%) no. (%) no. (%) (95% CI)† P

Heavy loads ϩ awkward posture 89 (8.9) 22 (2.2) 11 (1.1) 6.2 (3.4–11.1) Ͻ 0.0001 Heavy loads ϩ unstable/unbalanced/ 43 (4.3) 15 (1.5) 8 (0.8) 5.0 (2.2–11.3) Ͻ 0.0001 difﬁcult to grasp Heavy loads ϩ fatigued 23 (2.3) 6 (0.6) 3 (0.3) 5.3 (1.8–15.3) 0.002 Heavy loads ϩ object not close to 25 (2.5) 12 (1.2) 6 (0.6) 3.2 (1.3–7.9) 0.014 body Moderate or vigorous physical 24 (2.4) 4 (0.4) 5 (0.5) 7.7 (2.3–25.5) 0.001 activity ϩ fatigued

*ORϭ odds ratio; 95% CI ϭ 95% conﬁdence interval. † ORs describe the increase in odds of an episode of low back pain when exposed to the combined trigger compared to not being exposed to the combined trigger. 86

408 Steffens et al mentary Table 1, available in the online version of this creased the odds of developing back pain by 6.2 (95% CI article at http://onlinelibrary.wiley.com/doi/10.1002/acr. 3.4–11.1, P Ͻ 0.00001) and by 5.3 (95% CI 1.8–15.3, P ϭ 22533/abstract) or 6 hours (Supplementary Table 2, avail- 0.002) if associated with feeling fatigued when compared able in the online version of this article at http://online to no exposure. Similarly, engaging in physical activity library.wiley.com/doi/10.1002/acr.22533/abstract). The seems to increase further the risk of back pain if associated OR estimates observed when window duration was de- with feeling fatigued or tired (OR 7.7, 95% CI 2.3–25.5, creased to 1 hour (Supplementary Table 3, available in the P ϭ 0.001). online version of this article at http://onlinelibrary.wiley. com/doi/10.1002/acr.22533/abstract) were similar to the primary analysis (window duration of 2 hours) (Table 2). DISCUSSION

Interactions. No interaction was observed between ex- To date, back pain risk studies have only examined long- posure to any trigger and habitual participation in physical term exposure to factors such as smoking and inactivity activity, BMI, number of previous LBP episodes, depres- (22). Our study adds to the knowledge of risk of back pain sion, and anxiety score with P values greater than 0.05 for by demonstrating for the first time that brief exposure to a all triggers (Supplementary Table 4, available in the online range of modifiable physical and psychosocial factors inversion of this article at http://onlinelibrary.wiley.com/ creases the risk of an episode of back pain. The ORs for the doi/10.1002/acr.22533/abstract). However, age moderated triggers we evaluated ranged from 2.7 to 25.0, confirming the effect of exposure to manual tasks involving heavy that short-term exposure may substantially increase risk of loads (P ϭ 0.01) and sexual activity (P ϭ 0.04). To illus- back pain. Older age decreased the risk associated with trate the moderating effect of age on exposure to these 2 manual tasks involving heavy loads and increased the risk triggers, we used the coefficients from the regression mod- in those exposed to sexual activity. Habitual physical ac- els to calculate the ORs for a subject age 20, 40, or 60 years. tivity, age, BMI, previous number of LBP episodes, depression, and anxiety scores did not significantly change the For manual tasks involving heavy loads these were as risk associated with exposure to the other investigated follows: at 20 years: OR 13.6, 95% CI 5.4–34.5, P Ͻ 0.01; triggers. Notably, we also demonstrated that the onset of at 40 years: OR 6.0, 95% CI 3.8–9.5, P Ͻ 0.01; and at 60 back pain is not evenly distributed across the day, with years: OR 2.7, 95% CI 1.5–4.7, P Ͻ 0.01). For sexual mornings being the most frequent time of day for back pain activity these were as follows: at 20 years: OR 0.05, 95% CI onset. 0.03–0.97, P ϭ 0.04; at 40 years: OR 0.41, 95% CI 0.12– A major strength of our study was the large representa- 1.43, P ϭ 0.16; and at 60 years: OR 3.21, 95% CI 0.57– tive sample of patients with a moderate to severe episode 18.22, P ϭ 0.19). of back pain recruited at inception. Moreover, the self- Given the unexpected diurnal pattern of pain onset matching in case–crossover designs addresses some of the observed in our study, with more episodes commencing important limitations of previous risk studies in the back in the morning, exploratory post hoc analyses were con- pain field (23,24). Self-matching eliminates potential ef- ducted to evaluate if the back was more vulnerable to fects of unmeasured confounders, such as genetic and triggers in the morning. We evaluated the risk associated lifestyle factors, and minimizes selection bias (9). A poten- with being exposed to triggers in the period between tial limitation of case–crossover studies is the potential for 7:00 AM and 12:00 PM to that between 1:00 PM and 6:00 AM. recall bias, especially if the recall period is long; however, While there was no statistically significant interaction be- in our study the mean time between episode onset and tween time of exposure and risk of developing back pain, interview was only 5 days. Further, in an attempt to max- when compared to the afternoon or night, exposure to imize recall, participants were asked to link the day of awkward posture (OR 13.4, 95% CI 6.5–27.4) and manual onset of back pain and each of the 3 previous control days tasks involving unstable loading (OR 9.5, 95% CI 2.2–40.8) to significant events using a calendar and/or personal di- in the morning was more strongly associated with risk of ary. If participants were more likely to remember what had back pain onset (Table 3). Unfortunately our study lacked happened during the case window than in the control sufficient power to resolve this issue. windows (i.e., differential recall bias), the effect of a trigger would be overestimated. To minimize this potential prob- Multiple triggers. A theory-driven approach was ad- lem research assistants and participants were blinded to opted to select combinations of triggers that would in- the case and control windows. We also acknowledge that crease the risk of back pain onset. Five combinations for some people recall may have been difficult, so our were chosen by consensus among investigators as follows: interview script asked participants to refer to their diary/ 1) manual tasks involving heavy loads and awkward pos- smartphone and to nominate key aspects of each day. ture; 2) manual tasks involving heavy loads and loads that Participants were also blinded to the time and length of the are unstable, unbalanced, or difficult to grasp; 3) manual case and control windows and were asked to describe their tasks involving heavy loads and being fatigued; 4) manual engagement in each trigger over a period of 96 hours. tasks involving heavy loads and objects far from the body; These strategies have been used successfully in other and 5) being engaged in moderate or vigorous physical case–crossover studies and suggest that participants are as activity and feeling fatigued (Table 4). These analyses were able to recall data on the preceding 2 days as they are on not prespecified in the study protocol. Manual tasks in- the day of the event (25). Furthermore, the analysis of the volving heavy loads associated with awkward posture in- study did not control for time-variant confounders beyond 87

Triggers for Low Back Pain 409 the triggers included in the study. We acknowledge, there- The challenge for future research is to develop and eval- fore, that additional time-variant confounders may have uate prevention programs that aim to reduce exposure to influenced our study findings. the triggers identified in this study by thoughtfully con- A recent systematic review has identified multiple psy- sidering each trigger. Exposure to triggers such as manual chosocial and physically related risk factors for LBP (26). tasks involving heavy loads could be completely avoided No consistent risk factor emerged as predictive of first-time by redesigning the workplace so workers are no longer LBP, although prior LBP was a consistent predictor of required to lift heavy loads. Exposure to other triggers future incident LBP. However, many of the risk factors could be reduced by education, for example, through investigated are not robust or replicable, and many are not population-based public health messaging or onsite train- modifiable. Our results demonstrate that the onset of back ing of workers. Some triggers such as slips, trips, and falls pain may be triggered by brief exposure to physical and are probably more difficult to avoid, but we would note psychological factors. Past research has linked long-term that there are successful falls prevention strategies di- exposure to physical risk factors, such as heavy loads and rected at the elderly, suggesting that even these triggers are awkward posture, and future occurrence of back pain with potentially avoidable. It may not be sensible to aim to the accumulation of exposure over time holding stronger avoid the trigger “moderate or vigorous physical activity” associations (23). Our results provide the first accurate because, while transient exposure increases the risk of estimates to confirm that even brief exposure to these LBP, long-term exposure has many health benefits against physical factors may trigger moderate to severe back pain. many chronic diseases. The importance of psychosocial factors has also been high- We acknowledge that changing human behavior is far lighted as our results have identified that transient expo- from simple; however, the burden of back pain around the sure to stress and fatigue triples the odds of developing globe does provide a compelling case that something immediate back pain, whereas distraction increases the should be done. For example, the burden of disease due odds by a factor of 25. to road traffic injury is far less than that for back pain, Results from our interaction analysis suggested that age yet many countries devote considerable resources to con- moderates the effect of exposure to manual tasks involving trolling behaviors that increase the risk of road crashes. heavy loads. Interestingly, the odds associated with man- For example, mobile phone use has been shown in case– ual tasks involving heavy loads at age 60 years is more crossover studies to increase the risk for road crashes, than 5 times smaller than at age 20 years. While previous which has led to media campaigns to change driver be- studies have reported the association between long-term havior. exposure to heavy loads and LBP (23), this is the first study An important unanswered question from our study is that demonstrates a decrease in risk with age. One poten- whether long-term exposure to risk factors, such as tial reason for this may be that older people have learned smoking and driving, moderate the risk associated with to lift correctly or are more careful when handling heavy transient exposure to the triggers we studied. We also loads. Future research is required to evaluate this hypoth- acknowledge that our secondary analyses evaluating esis. whether attributes of the person (such as BMI and habitual There is little published information about the time of physical activity) moderate the risk of developing low onset of back pain. Biomechanists have theorized that the back pain, were underpowered, and we would encourage risk of back pain may be higher in the morning as inter- larger studies to evaluate this important issue. Future re- vertebral discs imbibe fluid overnight, leaving them more search directed at determining how best to modify the susceptible to stresses when loaded (27,28). Sprains and triggers identified in this study, and whether incorporating strains (an injury category that includes back pain) have this knowledge into prevention programs leads to reduced been previously associated with a similar diurnal pattern episodes of LBP, should be also conducted. Moreover, we to our study with ϳ40% of sprains and strains occurring did not collect data on the place of occurrence of LBP. from 8:00 AM to 11:00 AM (29). The strong diurnal pattern Future studies could investigate whether the trigger stud- for back pain onset would suggest that the morning may be ied occurred in the work place, during sport, or at home. a key time to intervene in order to prevent back pain. This Our results have significant clinical and policy implica- approach was adopted in a sham controlled trial (30) that tions for the control of a disease that is a major problem showed that advising patients with persistent LBP to re- globally. We offer robust estimates for the increase in risk strict lumbar flexion in the morning was effective in re- of back pain following exposure to modifiable physical ducing pain. Unfortunately, our post hoc analyses pro- and psychosocial triggers. Future research should evaluate vided imprecise estimates of the effect of time of exposure the success of programs that modify the triggers identified on the triggers evaluated in the study. Nonetheless the in this study in preventing episodes of back pain. OR point estimates for morning exposure (i.e., 7:00 AM to 12:00 PM) to manual tasks involving awkward posture and AUTHOR CONTRIBUTIONS those involving unstable, unbalanced, and difficult to All authors were involved in drafting the article or revising grasp objects were meaningfully different to the point es- it critically for important intellectual content, and all authors timates for exposure at other times. While larger studies approved the final version to be submitted for publication. are still clearly required, our study provides preliminary Dr. M. Ferreira had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy evidence that both the nature of the trigger and the time of of the data analysis. day of exposure to the trigger may influence the risk of Study conception and design. Steffens, M. Ferreira, Latimer, developing back pain. P. Ferreira, Koes, Blyth, Li, Maher. 88

410 Steffens et al

Acquisition of data. Steffens, M. Ferreira, Latimer, Blyth, Li, uniform definitions to be used in research. Spine 2002;27: Maher. 2409–16. Analysis and interpretation of data. Steffens, M. Ferreira, 17. Henschke N, Maher CG, Refshauge KM, Herbert RD, Cum- Latimer, P. Ferreira, Koes, Blyth, Li, Maher. ming RG, Bleasel J, et al. Prevalence of and screening for serious spinal pathology in patients presenting to primary care settings with acute low back pain. Arthritis Rheum 2009; REFERENCES 60:3072–80. 18. Ainsworth BE, Haskell WL, Herrmann SD, Meckes N, Bassett 1. Murray CJ, Lopez AD. Measuring the global burden of disease. DR Jr, Tudor-Locke C, et al. 2011 compendium of physical New Engl J Med 2013;369:448–57. activities: a second update of codes and MET values. Med Sci 2. Murray CJ, Vos T, Lozano R, Naghavi M, Flaxman AD, Sports Exerc 2011;43:1575–81. Michaud C, et al. Disability-adjusted life years (DALYs) for 19. Armstrong T, Bauman A, Davies J. Physical activity patterns 291 diseases and injuries in 21 regions, 1990-2010: a system- of Australian adults. Canberra: Australian Institute of Health atic analysis for the Global Burden of Disease Study 2010. and Welfare; 2000. Lancet 2013;380:2197–223. 20. Mittleman MA, Maclure M, Robins JM. Control sampling 3. Vos T, Flaxman AD, Naghavi M, Lozano R, Michaud C, Ezzati strategies for case-crossover studies: an assessment of relative M, et al. Years lived with disability (YLDs) for 1160 sequelae efficiency. Am J Epidemiol 1995;142:91–8. of 289 diseases and injuries 1990-2010: a systematic analysis 21. Dupont WD. Power calculations for matched case control for the Global Burden of Disease Study 2010. Lancet 2013; studies. Biometrics 1988;44:1157–68. 380:2163–96. 22. Gatchel RJ, Polatin PB, Mayer TG. The dominant role of 4. Deyo RA. Treatments for back pain: can we get past trivial psychosocial risk factors in the development of chronic low effects? [letter]. Ann Intern Med 2004;141:957–8. back pain disability. Spine 1995;20:2702–9. 5. Jellema P, van Tulder MW, van Poppel MN, Nachemson AL, 23. Heneweer H, Staes F, Aufdemkampe G, van Rijn M, Vanhees Bouter LM. Lumbar supports for prevention and treatment of L. Physical activity and low back pain: a systematic review of low back pain: a systematic review within the framework of recent literature. Euro Spine J 2011;20:826–45. the Cochrane Back Review Group. Spine 2001;26:377–86. 24. Hoogendoorn WE, van Poppel MN, Bongers PM, Koes BW, 6. Sahar T, Cohen M, Ne’eman V, Kandel L, Odebiyi D, Lev I, Bouter LM. Systematic review of psychosocial factors at work et al. Insoles for prevention and treatment of back pain. and private life as risk factors for back pain. Spine (Phila Pa Cochrane Database Syst Review 2007;4:CD005275. 1976) 2000;25:2114–25. 7. Maclure M. The case-crossover design: a method for studying transient effects on the risk of acute events. Am J Epidemiol 25. Broderick CR, Herbert RD, Latimer J, Barnes C, Curtin JA, 1991;133:144–53. Mathieu E, et al. Association between physical activity and 8. Maclure M, Mittlemann MA. Should we use a case-crossover risk of bleeding in children with hemophilia. JAMA 2012;308: design? Ann Rev Public Health 2000;21:193–221. 1452–9. 9. Mittleman MA, Mostofsky E. Exchangeability in the case- 26. Taylor JB, Goode AP, George SZ, Cook CE. Incidence and risk crossover design. Int J Epidemiol 2014;43:1645–55. factors for first-time incident low back pain: a systematic 10. Australian Safety and Compensation Council. National code review and meta-analysis. Spine J 2014;14:2299–319. of practice for the prevention of musculoskeletal disorders 27. Adams M, Dolan P, Hutton W. Diurnal variations in the from performing manual tasks at work. Canberra: Common- stresses on the lumbar spine. Spine (Phila Pa 1976) 1987;12: wealth of Australia; 2007. 130–7. 11. Sorock GS, Lombardi DA, Hauser R, Eisen EA, Herrick RF, 28. Adams M, Dolan P, Hutton W, Porter R. Diurnal changes in Mittleman MA. A case-crossover study of transient risk fac- spinal mechanics and their clinical significance. J Bone Joint tors for occupational acute hand injury. J Occup Environ Med Surg Br 1990;72:266–70. 2004;61:305–11. 29. Choi B, Levitsky M, Lloyd R, Stones I. Patterns and risk 12. Verma SK, Lombardi DA, Chang WR, Courtney TK, Huang factors for sprains and strain in Ontario, Canada 1990: an YH, Brennan MJ, et al. Rushing, distraction, walking on con- analysis of the Workplace Health and Safety Agency data taminated floors and risk of slipping in limited-service base. J Occup Environ Med 1996;38:379–89. restaurants: a case–crossover study. Occup Environ Med 30. Snook S, Webster B, McGorry R, Fogleman M, McCann K. The 2011;68:575–81. reduction of chronic nonspecific low back pain through the 13. Vinson DC, Mabe N, Leonard LL, Alexander J, Becker J, Boyer control of early morning lumbar flexion: a randomized con- J, et al. Alcohol and injury: a case-crossover study. Arch Fam trolled trial. Spine (Phila Pa 1976) 1998;23:2601–7. Med 1995;4:505–11. 31. Ware J, Sherbourne C. The MOS 36-item short-form health 14. Chen SY, Fong PC, Lin SF, Chang CH, Chan CC. A case- survey (SF-36). 1. Conceptual framework and item selection. crossover study on transient risk factors of work-related eye Med Care 1992;30:473–83. injuries. Occup Environ Med 2009;66:517–22. 32. Linton S, Hellsing A, Bergstrom G. Exercise for workers with 15. Steffens D, Ferreira ML, Maher CG, Latimer J, Koes BW, Blyth musculoskeletal pain: does enhancing compliance decrease FM, et al. Triggers for an episode of sudden onset low back pain? J Occup Med 1996;6:177–90. pain: study protocol. BMC Musculoskelet Disord 2012;13:7. 33. Dahabreh IJ, Paulus JK. Association of episodic physical and 16. De Vet H, Heymans M, Dunn K, Pope D, van der Beek A, sexual activity with triggering of acute cardiac events: system- Macfarlane G, et al. Episodes of low back pain: a proposal for atic review and meta-analysis. JAMA 2011;305:1225–33. Appendix Table 1. Exposure frequency and odds ratios for each trigger - sensitivity analysis (4 hour window)

Triggers Case Window First Control Window (24- Second Control Window Odds Ratio (95% CI) P

(0-4 hours), No. (%) 28 hours), No. (%) (48-52 hours), No. (%)

Physical factors

Manual tasks involving

Heavy loads 202 (20.2) 79 (7.9) 67 (6.7) 5.0 (3.4 to 7.3) <0.0001

Awkward posture 290 (29.0) 82 (8.2) 61 (6.1) 7.3 (5.1 to 10.5) <0.0001

Objects not close to the body 44 (4.4) 19 (1.9) 13 (1.3) 4.1 (1.9 to 8.9) <0.0001

Live people/ animals 92 (9.2) 71(7.1) 65 (6.5) 3.1 (1.5 to 6.3) 0.002

Unstable/ unbalance/ difficult to grasp or hold 61 (6.1) 22 (2.2) 16 (1.6) 4.9 (2.5 to 9.7) <0.0001

Moderate or vigorous physical activity 250 (25.0) 160 (16.0) 142 (14.2) 2.2 (1.7 to 2.9) <0.0001

Vigorous physical activity only 113 (11.3) 59 (5.9) 47 (4.7) 2.6 (1.7 to 3.8) <0.0001

Slip/ trip/ fall* 38 (3.8) 1 (0.1) 0 (0.0) -- --

Sexual activity 11 (1.1) 19 (1.9) 13 (1.3) 0.6 (0.3 to 1.2) 0.136

Psychosocial factors

Consumption of alcohol 19 (1.9) 15 (1.5) 16 (1.6) 1.3 (0.6 to 2.8) 0.451

Distracted during an activity or task 31 (3.1) 6 (0.6) 9 (0.9) 26.0 (3.5 to 191.6) 0.001

Fatigued/ tired 132 (13.2) 78 (7.8) 69 (6.9) 3.8 (2.3 to 6.4) <0.0001

Results of sensitivity analyses based on case and control windows of 4 hours duration.

89 *Due to small frequencies of exposure in the control windows, this trigger could not be included in conditional logistic regression analyses.

90 Appendix Table 2. Exposure frequency and odds ratios for each triggers - sensitivity analysis (6 hour window)

Triggers Case Window First Control Window (24-28 Second Control Window Odds Ratio (95% CI) P

(0-4 hours), No. (%) hours), No. (%) (48-52 hours), No. (%)

Physical factors

Manual tasks involving

Heavy loads 210 (21.0) 88 (8.8) 72 (7.2) 4.4 (3.1 to 6.3) <0.0001

Awkward posture 301 (30.1) 86 (8.6) 63 (6.3) 7.1 (5.0 to 10.2) <0.0001

Objects not close to the body 44 (4.4) 21 (2.1) 13 (1.3) 3.3 (1.6 to 6.7) 0.001

Live people/ animals 97 (9.7) 74 (7.4) 72 (7.2) 3.3 (1.6 to 6.7) 0.001

Unstable/ unbalance/ difficult to grasp or hold 64 (6.4) 25 (2.5) 16 (1.6) 4.0 (2.2 to 7.4) <0.0001

Moderate or vigorous physical activity 263 (26.3) 173 (17.3) 153 (15.3) 2.1 (1.6 to 2.8) <0.0001

Vigorous physical activity only 119 (11.9) 64 (6.4) 53 (5.3) 2.5 (1.7 to 3.7) <0.0001

Slip/ trip/ fall* 38 (3.8) 1 (0.1) 0 (0.0) -- --

Sexual activity 13 (1.3) 22 (2.2) 19 (1.9) 0.6 (0.3 to 1.2) 0.122

Psychosocial factors

Consumption of alcohol 22 (2.2) 20 (2.0) 21 (2.1) 1.1 (0.6 to 2.3) 0.724

Distracted during an activity or task 31 (3.1) 6 (0.6) 9 (0.9) 26.0 (3.5 to 191.6) 0.001

Fatigued/ tired 138 (13.8) 87 (8.7) 75 (7.5) 3.2 (2.0 to 5.1) <0.0001

91 Results of sensitivity analyses based on case and control windows of 6 hours duration.

*Due to small frequencies of exposure in the control windows, this trigger could not be included in conditional logistic regression analyses.

92 Appendix Table 3. Interaction analysis between baseline variables and physical and psychological factors

Habitual physical Age* BMI* Previous episodes* Depression* Anxiety*

Physical factors activity*

OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P OR (95% CI) P

Heavy loads 0.95 (0.75-1.20) 0.67 0.59 (0.39-0.88) 0.01 1.38 (0.82-2.30) 0.22 0.93 (0.78-1.10) 0.40 0.89 (0.58-1.37) 0.59 0.91 (0.56-1.48) 0.71

Awkward posture 1.21 (0.81-1.79) 0.36 1.13 (0.76-1.68) 0.53 1.08 (0.73-1.60) 0.69 1.01 (0.91-1.13) 0.82 0.83 (0.54-1.26) 0.38 0.81 (0.51-1.28) 0.36

Objects not close to the 1.48 (0.44-4.99) 0.53 0.24 (0.05-1.17) 0.08 1.20 (0.46-3.10) 0.71 1.24 (0.78-1.98) 0.36 0.40 (0.15-1.12) 0.08 0.70 (0.23-2.17) 0.54 body

Live people/ animals 0.36 (0.09-1.47) 0.15 0.85 (0.29-2.47) 0.76 0.79 (0.22-2.80) 0.71 0.77 (0.50-1.18) 0.23 0.40 (0.11-1.50) 0.17 0.64 (0.15-2.77) 0.55

Unstable/ unbalance/ 4.06 (0.82- 0.89 (0.57-1.39) 0.62 0.42 (0.14-1.24) 0.12 0.09 0.85 (0.62-1.17) 0.33 0.63 (0.28-1.42) 0.27 0.75 (0.31-1.82) 0.52 difficult to grasp or hold 20.06)

Vigorous physical activity 1.03 (0.73-1.45) 0.86 0.56 (0.30-1.02) 0.06 1.96 (0.99-3.87) 0.05 0.92 (0.74-1.15) 0.48 0.95 (0.55-1.64) 0.86 1.37 (0.73-2.58) 0.33

Moderate physical activity 0.89 (0.69-1.16) 0.40 0.87 (0.59-1.29) 0.48 0.98 (0.69-1.40) 0.92 1.14 (0.92-1.41) 0.23 0.79 (0.52-1.20) 0.27 0.80 (0.52-1.24) 0.33

Slip/ trip/ fall ------

Consumption of alcohol 1.14 (0.37-3.50) 0.82 1.16 (0.54-2.47) 0.71 0.86 (0.31-2.42) 0.78 0.90 (0.42-1.91) 0.78 2.13 (0.46-9.79) 0.33 1.15 (0.35-3.82) 0.81

Sexual activity 0.89 (0.08-9.93) 0.93 3.82 (1.00-14.52) 0.04 1.00 (0.29-3.39) 1.00 1.18 (0.83-1.66) 0.36 1.21 (0.44-3.31) 0.71 0.67 (0.23-1.94) 0.46

Psychological factors

Distracted 0.52 (0.14-1.90) 0.32 0.50 (0.06-4.33) 0.53 -- -- 0.85 (0.18-4.07) 0.84 -- -- 0.16 (0.01-5.41) 0.31

93 Fatigued/ tired 1.29 (0.66-2.55) 0.46 0.99 (0.58-1.71) 0.99 0.92 (0.54-1.58) 0.77 0.86 (0.70-1.06) 0.16 1.23 (0.69-2.20) 0.49 0.93 (0.50-1.73) 0.83

*Entered as continuous variables, increase for 1 SD. Habitual physical activity=350**; Age=13; BMI=5; Number of previous episodes=5; Depression=3; Anxiety=3.

** HPA: Habitual physical activity is not normal distributed and we used IQR=350 mins.

Bold= p<0.05.

94 Appendix Table 4. Exposure frequency and odds ratios for each trigger - secondary analysis (1hour window)

Case window First control window Second control window

Triggers (0-1 hours), (24-25 hours), (48-49 hours), Odds Ratio (95% CI) † P

No. (%) No. (%) No. (%)

Physical factors

Manual tasks involving

Heavy loads 166 (16.6) 52 (5.2) 50 (5.0) 6.2 (3.9 to 9.7) <0.0001

Awkward posture 259 (25.9) 65 (6.5) 47 (4.7) 7.9 (5.4 to 11.8) <0.0001

Objects not close to the body 35 (3.5) 13 (1.3) 9 (0.9) 6.5 (2.3 to 18.6) <0.0001

Live people or animals 73 (7.3) 55 (5.5) 55 (5.5) 5.5 (1.9 to 16.0) 0.002

Unstable/ unbalanced/ difficult to grasp or hold loads 49 (4.9) 16 (1.6) 12 (1.2) 7.6 (3.0 to 19.3) <0.0001

Moderate or vigorous physical activity 204 (20.4) 107 (10.7) 87 (8.7) 2.9 (2.1 to 4.0) <0.0001

Vigorous physical activity only 96 (9.6) 37 (3.7) 28 (2.8) 4.3 (2.6 to 7.2) <0.0001

Slip/ trip/ fall* 34 (3.4) 1 (0.1) 0 (0.0) -- --

Sexual activity 5 (0.5) 8 (0.8) 9 (0.9) 0.6 (0.2 to 1.9) 0.410

Psychosocial factors

Consumption of alcohol 11 (1.1) 9 (0.9) 9 (0.9) 1.2 (0.5 to 3.0) 0.655

Distracted during an activity or task 26 (2.6) 6 (0.6) 8 (0.8) 21.0 (2.8 to 156.10 0.003

95 Fatigued/ tired 108 (10.8) 63 (6.3) 57 (5.7) 3.4 (2.0 to 5.6) <0.0001

Results of the secondary analyses based on case and control windows of 1 hour duration. *Due to small frequencies of exposure in the control windows, this trigger could not be included in the conditional logistic regression analyses. †Odds ratios and 95% confidence interval were derived comparing exposure in the case window (0-1 hours) with the control window 1 (24-25 hours).

96 97 Appendix 5. Study Questionnaire

Triggers Specific Questions Tick the boxes where the participant reports pain on the mannequin below:

Height (cm): ______Weight (Kg): ______

How much back pain have you had during the first 24 hours of this episode? [1] None Very mild Mild Moderate Severe Very Severe      

How tense or anxious have you felt in the past week? Circle one. [2]

0 1 2 3 4 5 6 7 8 9 10 Calm and As relaxed tense/anxious as I’ve ever felt

Physical Activity [3]

What do you estimate was the total time that you spent walking in this way in the last week? ______hours ______minutes; And in the week before your low back pain started? ______hours ______minutes.

What do you estimate was the total time that you spent doing these activities in the last week? ______hours ______minutes; And in the week before your back pain started? ______hours ______minutes.

4. Was your level of physical activity last week typical for you?  Yes  No

Triggers Exposure

Please write the date and time you first noticed your low back pain on the table below:

Day Fri Sat Sun Mon Tues Wed Thurs Fri Sat Sun

Date

Time

Example: Tuesday: cold and wet; visited parents

Day of back pain: ______

Day before: ______

2 days earlier: ______

3 days earlier: ______

100

Firstly we are interested in manual tasks involving HEAVY LOADS.  Yes  No So on the day of your back pain did you engage in any manual tasks involving a heavy load?

That was the (restate their description of the day). Day before

Now what about the day before…. That was the (restate their description of the day).  Yes  No

2 days earlier

OK now two days back That was the (restate their description of the day).  Yes  No

3 days earlier

OK finally three days back That was the (restate their description of the  Yes  No day)

101

That was the (restate their description of the day).  Yes  No

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back That was the 2 days (restate their description of the day). earlier

 Yes  No

OK finally three days back That was 3 days the (restate their description of the earlier day)

 Yes  No

102

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back That was the (restate their description of the day). 2 days earlier

 Yes  No

OK finally three days back That was 3 days the (restate their description of the earlier day)

 Yes  No

103

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

2 days OK now two days back… earlier That was the (restate their description of the day).

 Yes  No

3 days OK finally three days back… earlier That was the (restate their description of the day)

 Yes  No

104

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back… That was the (restate their 2 days description of the day). earlier

 Yes  No

3 days OK finally three days back… earlier That was the (restate their description of the day)

 Yes  No

105

Now what about the day before…  Yes  No That was the (restate their description of the day).

OK now two days back… That was 2 days the (restate their description of the earlier day).

 Yes  No

OK finally three days back… That was the (restate their description of 3 days the day) earlier

 Yes  No

106

Now what about the day before…. That was the (restate their description of the day).  Yes  No

OK now two days back… That was 2 days the (restate their description of the earlier day).

 Yes  No

OK finally three days back… That 3 days was the (restate their description of earlier the day).

 Yes  No

107

 Yes  No

Now what about the day before…. Day before That was the (restate their description of the day).

 Yes  No

OK now two days back… That was the (restate their 2 days description of the day). earlier

 Yes  No

OK finally three days back…. That was the (restate their 3 days description of the day) earlier

 Yes  No

108

OK finally three days back… That  Yes  No was the (restate their description of the day) 3 days earlier

 Yes  No

OK now two days back… That was 2 days the (restate their description of the earlier day).

 Yes  No OK finally three days back… That was the (restate their description of the day) 3 days earlier

 Yes  No

109

OK finally three days back… That 3 days was the (restate their description of earlier the day)  Yes  No

OK now two days back… That was 2 days the (restate their description of the earlier day).  Yes  No

OK finally three days back… That was the (restate their description of 3 days the day) earlier  Yes  No

110

______

So on the day of your back pain did you Day of back pain do the activity described above? That was the (restate their description of the day).  Yes  No

Day before Now what about the day before…. That was the (restate their description of the day).  Yes  No

OK now two days back… That was the 2 days earlier (restate their description of the day).

 Yes  No

OK finally three days back… That was the (restate their description of the day) 3 days earlier

 Yes  No

111

Reference

1. Ware J, Sherbourne C: The MOS 36-item short-form health survey (SF-36). 1.

Conceptual framework and item selection. Medical Care 1992, 30:473-483.

2. Linton S, Hellsing A, Bergstrom G: Exercise for workers with musculoskeletal

pain: does enhancing compliance decrease pain? J Occup Med 1996, 6(3):177-190.

3. Armstrong T, Bauman A, Davies J: Physical activity patterns of Australian Adults.

In. Edited by Welfare AIoHa. Canberra; 2000.

4. Australian Safety Compensation Council: National code of practice for the

prevention of musculoskeletal disorders from performing manual tasks at work.

In. Canberra; 2007.

5. Verma S, Lombardi D, Chang W, Courtney T, Huang Y, Brennan M, Mittleman M,

Ware J, Perry M: Rushing, distraction, walking on contaminated floors and risk

of slipping in limited-service restaurants: a case--crossover study. Occupational &

Environmental Medicine 2010, 68(8):575-581.

6. Vinson DC, Mabe N, Leonard LL, Alexander J, Becker J, Boyer J, Moll J: Alcohol

and injury. A case-crossover study. Arch Fam Med 1995, 4(6):505-511.

7. Dahabreh IJ, Paulus JK, Dahabreh IJ, Paulus JK: Association of episodic physical

and sexual activity with triggering of acute cardiac events: systematic review and

meta-analysis. JAMA, 305(12):1225-1233.

8. Sorock G, Lombardi D, Peng D, Hauser R, Eisen E, Herrick R, Mittleman M: Glove

use and the relative risk of acute hand injury: a case-crossover study. Journal of

Occupational & Environmental Hygiene 2004, 1(3):182-190.

9. Chen S, Fong P, Lin S, Chang C, Chan C: A case-crossover study on transient risk

factors of work-related eye injuries. Occupational & Environmental Medicine

2009, 66(8):517-522.

112

Chapter Five

Effect of weather on back pain: results from a case-crossover study

Chapter Five is published as:

Steffens D, Maher CG, Li Q, Ferreira ML, Pereira LSM, Koes BK, Latimer J. Effect of weather on back pain: results from a case-crossover study. Arthritis Care & Research. 2014; 66:1867-1872. 113

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “Effect of weather on back pain: results from a case-crossover study”, we confirm that Daniel Steffens has made the following contributions:

 Conception and design of the research  Data collection  Analysis and interpretation of the findings  Writing of the manuscript and critical appraisal of the content

Christopher G Maher Date: 01.01.2015

Qiang Li Date: 01.01.2015

Manuela L Ferreira Date: 01.01.2015

Leani SM Pereira Date: 01.01.2015

Bart W Koes Date: 01.01.2015

Jane Latimer Date: 01.01.2015

114

Effect of Weather on Back Pain: Results From a Case-Crossover Study

DANIEL STEFFENS,1 CHRIS G. MAHER,2 QIANG LI,2 MANUELA L. FERREIRA,2 3 4 2 LEANI S. M. PEREIRA, BART W. KOES, AND JANE LATIMER

Objective. To investigate the influence of various weather conditions on risk of low back pain. Methods. We conducted a case-crossover study in primary care clinics in Sydney, Australia. A total of 993 consecutive patients with a sudden, acute episode of back pain were recruited from October 2011 to November 2012. Following the pain onset, demographic and clinical data about the back pain episode were obtained for each participant during an interview. Weather parameters (temperature, relative humidity, air pressure, wind speed, wind gust, wind direction, and precipitation) were obtained from the Australian Bureau of Meteorology for the entire study period. Weather exposures in the case window (time when participants first noticed their back pain) were compared to exposures in 2 control time windows (same time duration, 1 week and 1 month before the case window). Results. Temperature, relative humidity, air pressure, wind direction, and precipitation showed no association with for 0.01 ؍ onset of back pain. Higher wind speed (odds ratio [OR] 1.17 [95% confidence interval (95% CI) 1.04–1.32], P for an increase of 14 km/hour) increased 0.02 ؍ an increase of 11 km/hour) and wind gust (OR 1.14 [95% CI 1.02–1.28], P the odds of pain onset. Conclusion. Weather parameters that have been linked to musculoskeletal pain such as temperature, relative humidity, air pressure, and precipitation do not increase the risk of a low back pain episode. Higher wind speed and wind gust speed provided a small increase in risk of back pain, and although this reached statistical significance, the magnitude of the increase was not clinically important.

INTRODUCTION with which this belief is reported, there are few robust studies that have investigated this potential association. Patients with musculoskeletal pain commonly report that The key problems are that study participants are not certain weather conditions influence their symptoms, the blinded to the study hypotheses, the studies have no con- pain from rheumatoid arthritis being a clear example of trol period, and data are mainly based on subjective recall this (1–3). Previous studies have reported that cold or of both weather and symptoms. humid weather conditions (4,5) and changes in weather conditions (6) negatively influence symptoms in patients It is a methodological challenge to assess the effect of the experiencing chronic pain. Despite the high frequency weather on pain onset using traditional study designs. To our knowledge, only 2 studies have assessed whether aspects of the weather influence musculoskeletal pain using 1Daniel Steffens, BPhty: The George Institute for Global a case-crossover methodology (3,7). The case-crossover ap- Health, Sydney Medical School, and The University of Syd- proach was specifically designed to study exposures, such ney, Sydney, New South Wales, Australia, and Federal Uni- versity of Minas Gerais, Minas Gerais, Brazil; 2Chris G. as the weather, that have a short induction time and tran- Maher, PhD, Qiang Li, MBiostat, Manuela L. Ferreira, PhD, sient effect. With this design, it would be possible to Jane Latimer, PhD: The George Institute for Global Health, evaluate the increased risk associated with aspects of the Sydney Medical School, and The University of Sydney, Syd- weather by comparing exposure to meteorological vari- ney, New South Wales, Australia; 3Leani S. M. Pereira, PhD: Federal University of Minas Gerais, Minas Gerais, Brazil; ables at the time of the pain onset or exacerbation (defined 4Bart W. Koes, PhD: Erasmus MC, University Medical Cen- as the case window) and in earlier periods when the per- ter Rotterdam, Rotterdam, The Netherlands. son was pain free (defined as the control windows) (8). Address correspondence to Daniel Steffens, BPhty, The George Institute for Global Health, Sydney Medical School, Subsequent to completing a case-crossover study evalu- The University of Sydney, PO Box M201, Missenden Road, ating physical and psychosocial triggers for an episode of Sydney, 2050, New South Wales, Australia. E-mail: low back pain (LBP), we became aware of the limited data [email protected]. on weather and musculoskeletal pain. We took the oppor- Submitted for publication March 19, 2014; accepted in revised form May 27, 2014. tunity to link back pain data from the original data set and historical weather data obtained from meteorological re-

1867 115

1868 Steffens et al

onset of their LBP were included. In total, 993 people with Significance & Innovations LBP participated. All participants were ages Ն18 years. ● Patients with musculoskeletal pain commonly report that their symptoms are influenced by the Meteorological data. Sydney is the state capital of New weather, but this issue has not been evaluated in South Wales and the most populous city in Australia. It robust research or for the most common musculo- has a temperate climate with warm summers and mild skeletal condition, back pain. winters, and rainfall spread throughout the year. Meteoro- ● There was no association between temperature, logical data were obtained from the Australian Bureau of relative humidity, air pressure, wind direction, Meteorology for the entire study period from 5 weather and precipitation and risk of back pain. monitoring stations in Sydney (www.bom.gov.au). The weather stations were located in 3 main regions: Sydney ● Higher wind speeds slightly increased the odds of Central (Sydney Airport-066037), Sydney North West back pain onset, but the effect is not important. (Penrith Lakes-067113 and Badgerys Creek-067108), and Sydney South West (Mount Annan-068257 and Camden Airport-068192). Two weather stations (Penrith Lakes and Mount Annan) did not provide data on air pressure; therefore, air pressure data were used from 2 neighboring cords. The aim of this study was to quantify the transient weather stations (Badgerys Creek and Camden Airport). increase in risk of sudden onset of acute LBP associated For each participant enrolled in the study, we used data with the following weather parameters: temperature (°C), from the weather station closest to the region where they relative humidity (%), air pressure (hPa), wind speed (km/ lived. The following hourly weather parameters were ob- hour), wind gust (km/hour), wind direction (degrees true), tained: temperature (°C), relative humidity (%), air pres- and precipitation (mm). sure (hPa), wind speed (sustained wind speed averaged over 10 minutes leading up to the time of the observation; km/hour), wind gust (short burst of high-speed wind SUBJECTS AND METHODS averaged over 3 seconds leading up to the time of the The present study is a reanalysis of original case-crossover observation; km/hour), wind direction (direction where study data (9) linked to historical weather data obtained the wind is coming from; degrees true), and precipitation from meteorological records. Since the present study was (mm). conceived after completion of the original case-crossover study (9), the weather exposure data and back pain history Study design. To determine whether there is an associ- data are independent, and participants and staff were ation between weather parameters and LBP onset, we used blinded to the study hypotheses during data collection. a case-crossover design. This design compares exposure to Ethical approval for the study was granted by the Univer- weather parameters at the time of back pain onset (defined sity of Sydney Human Research Ethics Committee (proto- as the case window) with exposure at the same time 1 col 05-2011/13742). week and 1 month prior to the pain onset (defined as control windows 1 and 2, respectively) for each partici- Study participants. Consecutive patients presenting to pant. The time periods for exposure were defined as fol- primary care clinicians (general medical practitioners, lows: 1) within 1 hour (average value at 1 hour immedi- physiotherapists, chiropractors, and pharmacists) for treat- ately before the pain onset), 2) at 24 hours (average value at ment of an episode of sudden-onset, acute LBP were re- 24 hours immediately before the pain onset), and 3) aver- cruited in Sydney, Australia, from October 2011 to No- age value within 24 hours (average value from 0–24 hours vember 2012. To be eligible to enter the study, participants immediately before the pain onset). must have met the following criteria: 1) comprehends spo- To determine whether change in the weather parameters ken English; 2) primary symptom of pain in the area be- is associated with LBP onset, we computed a change score tween the 12th rib and buttock crease, with or without leg using this formula: average value over 0–24 hours imme- pain; 3) pain of at least moderate intensity during the first diately prior to the pain onset minus average value over 24 hours of the episode (assessed using a modified version 25–48 hours immediately prior to the pain onset for each of item 7 of the Short Form 36 [SF-36]); 4) presentation for participant. treatment within 7 days from the time of pain onset; and 5) no known or suspected serious spinal pathology (e.g., met- Statistical analysis. First, a descriptive analysis was astatic, inflammatory, or infective diseases of the spine; performed. Characteristics of the study subjects and dis- cauda equina syndrome; spinal fracture). A sudden-onset tribution of weather parameters were reported. Second, episode of LBP was defined as pain of at least moderate the analysis followed standard methods for stratified ana- intensity that developed over the first 24 hours (assessed lyses. In the case-crossover design, the individual subject using a modified version of item 7 of the SF-36) (10). is the stratifying variable (8,11). We used the matched-pair analytical approach (conditional logistic regression) to Participant interview. Basic demographic and clinical contrast exposures (meteorological variables) for the case data were collected by telephone interview (9). Only those period with exposures for the control period. For each patients who were interviewed within 14 days from the subject, 1 case period was matched to 2 control periods 116

Influence of Weather on Back Pain 1869 exactly 1 week and 1 month before the date and time of the the potential problems associated with the case-crossover pain onset (11). Odds ratios (ORs) and 95% confidence design. We avoided the problems associated with recall, intervals (95% CIs) were derived comparing exposure in since exposure data were objectively measured and ob- the case window with each of the 2 control windows. All tained independently of the back pain data. Since both weather parameters were treated as continuous variables participants and assessors were blinded to the study hy- and we calculated the OR associated with a 1-SD increase potheses, we avoided bias associated with people’s beliefs in the weather parameter. The analyses were performed about weather and pain. Lastly, we enrolled a large and using Stata, version 12 (12). well-defined cohort of consecutive patients from primary care clinics. This study has some limitations that should be taken RESULTS into account. First, our data did not include potentially important individual data, such as time spent outdoors, Primary care clinicians screened 1,639 consecutive pa- characteristics of housing or work, and air conditioning, tients from October 2011 to November 2012, where 993 which could modify a participant’s vulnerability to met the inclusion criteria and consented to enter the study. weather conditions. Second, we used meteorological data The characteristics of the study participants are shown in obtained from 3 main regions in Sydney and assumed that Table 1. The mean Ϯ SD age was 45.2 Ϯ 13.4 years. Most the LBP onset occurred in the individual while in a region participants were male (54.2%) and professional workers close to their home. This may have introduced misclassi- (34.2%), and had a mean Ϯ SD number of previous epi- fication for some patients’ exposure. The effect of this sodes of back pain of 5.9 Ϯ 14.0. nondifferential bias would be to change the present find- During the study period of 13 months, the mean weather ings toward the null (13). Using data from 3 distinct parameters were 1.4 mm of precipitation (range 0.0–115.4), weather regions, however, helped minimize the spatial temperature of 16.7°C (range Ϫ0.7 to 37.5), relative humid- variations in the weather parameters that exist within re- ity of 71.6% (range 6.0–100.0%), wind speed of 11.2 km/ gions (14). Third, participants’ time of back pain onset was hour (range 0.0–74.0), wind gust of 16.2 km/hour (range based on their recall, which is a potential limitation of 0.0–100.0), wind direction of 164.6 degrees true (range retrospective studies. Therefore, participants were asked 360.0 to 0.0), and 1,017.3 hPa of air pressure (range 994.7– to use their diary, calendar, or smartphone to help recall 1,035.8) (Table 2). the onset time. Also, to avoid time recall bias, interviews Descriptive data for the meteorological parameters in the were performed as soon as possible after the onset of back case and control windows are shown in Table 3. Estimates pain, with the mean Ϯ SD time between pain onset and of fixed parameters from conditional logistic regression presentation to primary care of 3.0 Ϯ 2.1 days and from models for each weather parameter are also shown in presentation to interview of 1.9 Ϯ 1.9 days. Table 3. Only 2 of the 28 analyses were significant: wind There is little published research investigating the effect speed 24 hours prior to onset (OR 1.17 [95% CI 1.04–1.32], of the weather on musculoskeletal pain. Of the 2 previous P ϭ 0.01 for an increase of 11 km/hour) and wind gust 24 case-crossover studies, one found no effect of relative hu- hours prior to onset (OR 1.14 [95% CI 1.02–1.28], P ϭ 0.02 midity, pressure, rain, and hours of sun and cloud cover for an increase of 14 km/hour) increased the risk of back on symptoms of rheumatoid arthritis (3), while the other pain. None of the other weather parameters investigated found that higher wind speed slightly increased the risk of was associated with back pain onset. hip fracture (7), but only in a subgroup of participants. Typically, the research cited to support a relationship DISCUSSION between weather and LBP uses very weak designs. For example, many studies simply survey patients about their This study provides the first evaluation of the influence of opinion on the effect of weather on their symptoms (4–6). the weather on the most common musculoskeletal condi- Sometimes the belief that weather affects musculoskeletal tion, back pain. Contrary to popular belief, weather param- pain is supported by reviews of studies that report higher eters, such as temperature, precipitation, air pressure, prevalence of musculoskeletal pain in studies conducted wind direction, and humidity, were not associated with in cooler settings; however, there are other between-study the onset of back pain. Unexpectedly, heavier wind speed factors that also could have contributed to this finding 24 hours prior to an episode increased the risk of back (15). At present, there is no evidence derived from robust pain, but the magnitude of the effect was very small and research that supports the widespread belief that the unlikely to be clinically important. Additionally, we did weather affects musculoskeletal pain. There is, however, not adjust the critical P value for multiple comparisons; if some evidence for other health conditions. Previous case- this was done, the obtained P values of 0.01 and 0.02 for crossover studies have shown that exposure to lower tem- wind parameters would no longer be statistically signifi- peratures increases the risk of myocardial infarction (16), cant. whereas higher temperatures and lower pressures lead to The use of a case-crossover design is a strength of this an increase in risk of headaches (17). study. In case-crossover studies, cases act as their own Our study provides clear evidence that weather does not controls; consequently, case-crossover studies are not con- have an important effect on LBP onset. Only a trivial founded by time-invariant risk factors, since exposure in- increase in the risk was observed with higher wind speed formation is collected from the same individual (11). The 24 hours prior to the onset of pain in this population of timing and nature of our study allowed us to avoid some of Australian adults. One possible explanation for the lack of 117

1870 Steffens et al

Table 1. Characteristics of the participants*

Sydney Sydney Sydney Central North West South West Overall (993 ؍ n) (71 ؍ n) (256 ؍ n) (666 ؍ n)

Male sex 356 (53.5) 138 (53.9) 44 (61.9) 538 (54.2) Age, mean Ϯ SD years 45.0 Ϯ 14.2 45.3 Ϯ 11.8 47.2 Ϯ 11.6 45.2 Ϯ 13.4 Height, mean Ϯ SD cm 172.9 Ϯ 10.3 170.6 Ϯ 9.9 173.9 Ϯ 12.2 172.4 Ϯ 10.4 Weight, mean Ϯ SD kg 78.8 Ϯ 17.7† 78.7 Ϯ 19.4 80.2 Ϯ 17.7 78.8 Ϯ 18.1‡ BMI, mean Ϯ SD kg/m2 26.2 Ϯ 5.0† 26.9 Ϯ 5.9 26.4 Ϯ 4.4 26.4 Ϯ 5.2‡ Duration of current episode, mean Ϯ SD days 5.1 Ϯ 2.7 4.5 Ϯ 2.8 4.9 Ϯ 2.9 4.9 Ϯ 2.7 No. of previous episodes, mean Ϯ SD 6.3 Ϯ 15.9 4.7 Ϯ 7.7 6.2 Ϯ 12.6 5.9 Ϯ 14.0 Days to seek care, mean Ϯ SD 3.0 Ϯ 2.1 2.7 Ϯ 2.1 2.9 Ϯ 2.0 3.0 Ϯ 2.1 Days of reduced activity, mean Ϯ SD 2.4 Ϯ 2.2 1.9 Ϯ 1.9 3.0 Ϯ 2.3 2.3 Ϯ 2.2 Depression status, mean Ϯ SD 2.7 Ϯ 2.6 2.7 Ϯ 2.8 2.9 Ϯ 2.8 2.7 Ϯ 2.7 Pain, mean Ϯ SD 5.1 Ϯ 2.1 5.7 Ϯ 2.1 5.4 Ϯ 2.1 5.3 Ϯ 2.1 GPES, mean Ϯ SD 1.8 Ϯ 1.8 1.7 Ϯ 1.7 1.7 Ϯ 1.9 1.8 Ϯ 1.8 Tense/anxious, mean Ϯ SD 4.0 Ϯ 2.5 4.1 Ϯ 2.7 4.1 Ϯ 2.6 4.0 Ϯ 2.5 Presence of leg pain 64 (9.6) 27 (10.5) 10 (14.1) 101 (10.2) Compensation 47 (7.1) 26 (10.2) 15 (21.1) 88 (8.9) Medication 320 (48.1) 98 (38.3) 32 (45.1) 450 (45.3) What do you do for a living? Not employed 124 (18.6) 29 (11.3) 9 (12.7) 162 (16.3) Clerical and administrative worker 69 (10.4) 30 (11.7) 4 (5.6) 103 (10.4) Community and personal service worker 31 (4.7) 15 (5.9) 1 (1.4) 47 (4.7) Laborer 13 (2.0) 10 (3.9) 7 (9.9) 30 (3.0) Machinery operator and driver 14 (2.1) 10 (3.9) 3 (4.2) 27 (2.7) Manager 106 (15.9) 39 (15.2) 11 (15.5) 156 (15.7) Professional 234 (35.1) 83 (32.4) 23 (32.4) 340 (34.2) Sales worker 33 (5.0) 17 (6.6) 2 (2.8) 52 (5.2) Technician and trade worker 42 (6.3) 23 (9.0) 11 (15.5) 76 (7.7) Pain location§ Upper back 39 (5.9) 18 (7.0) 2 (2.8) 59 (5.9) Lower back 666 (100.0) 256 (100.0) 71 (100.0) 993 (100.0) Left thigh (back) 65 (9.8) 19 (7.4) 11 (75.5) 95 (9.6) Left leg (back) 22 (3.3) 15 (5.9) 5 (7.0) 42 (4.2) Right thigh (back) 72 (10.8) 23 (9.0) 12 (16.9) 107 (10.8) Right leg (back) 31 (4.7) 12 (4.7) 5 (7.0) 48 (4.8) Right thigh (front) 20 (3.0) 8 (3.1) 1 (1.4) 29 (2.9) Right leg (front) 8 (1.2) 3 (1.2) 0 (0.0) 11 (1.1) Left thigh (front) 20 (3.0) 3 (1.2) 3 (4.2) 26 (2.6) Left leg (front) 5 (0.8) 2 (0.8) 0 (0.0) 7 (0.7) Pain severity Moderate 251 (37.7) 98 (38.3) 22 (31.0) 371 (37.4) Severe 328 (49.3) 127 (49.6) 36 (50.7) 491 (49.5) Very severe 87 (13.1) 31 (12.1) 13 (18.3) 131 (13.2) Pain interfering work Not at all 16 (2.4) 5 (2.0) 0 (0.0) 21 (2.1) A little bit 75 (11.3) 24 (9.4) 2 (2.8) 101 (10.2) Moderately 161 (24.2) 72 (28.1) 15 (21.1) 248 (25.0) Quite a bit 259 (38.9) 97 (37.9) 30 (42.3) 386 (38.9) Extremely 155 (23.3) 58 (22.7) 24 (33.8) 237 (23.9) Habitual physical activity in the last week¶ Sedentary 346 (51.9) 149 (58.2) 43 (60.5) 538 (54.2) Insufficient activity 114 (17.1) 34 (13.2) 15 (21.1) 163 (16.4) Sufficient activity 206 (30.9) 73 (28.5) 13 (18.3) 292 (29.4) Habitual physical activity in the week before¶ Sedentary 212 (31.8) 112 (43.7) 33 (46.5) 357 (35.9) Insufficient activity 122 (18.3) 41 (16.0) 11 (15.5) 174 (17.5) Sufficient activity 332 (49.8) 103 (40.2) 27 (38.0) 462 (45.5)

* Values are the number (percentage) unless indicated otherwise. BMI ϭ body mass index; GPES ϭ Global Perceived Effect Score. †Nϭ 665. ‡Nϭ 992. § Pain location was assessed using a pain manikin provided to participants by the referring clinician. ¶ Habitual physical activity ϭ moderate activity time ϩ (2 ϫ vigorous activity time). Sedentary ϭ 0 minutes, insufﬁcient activity ϭ Ն1toՅ149 minutes, and sufﬁcient activity ϭ Ն150 minutes. 118

Inﬂuence of Weather on Back Pain 1871

Table 2. Features of weather parameters in 3 Sydney conurbations from October 2011 to November 2012

Sydney Central Sydney South West Sydney North West Mean ؎ SD* Min Max Mean ؎ SD* Min Max Mean ؎ SD* Min Max

Precipitation, mm 1.4 Ϯ 4.9 0.0 75.4 1.5 Ϯ 5.9 0.0 115.4 1.3 Ϯ 5.0 0.0 80.0 Temperature, °C 17.9 Ϯ 4.6 6.0 37.5 16.8 Ϯ 6.0 1.1 37.1 15.5 Ϯ 6.1 Ϫ0.7 37.4 Relative humidity, % 65.8 Ϯ 18.0 6.0 100.0 75.2 Ϯ 23.1 11.0 100.0 73.8 Ϯ 21.0 10.0 99.0 Wind speed, km/hour 19.9 Ϯ 9.9 0.0 74.0 6.9 Ϯ 5.9 0.0 50.0 7.0 Ϯ 5.3 0.0 33.0 Wind gust, km/hour 25.6 Ϯ 12.9 0.0 100.0 11.0 Ϯ 9.2 0.0 74.0 12.0 Ϯ 9.0 0.0 61.0 Wind direction, degrees true 190.9 Ϯ 106.0 0.0 360.0 163.1 Ϯ 111.3 0.0 360.0 139.6 Ϯ 108.1 0.0 360.0 Air pressure, hPa 1,017.2 Ϯ 6.4 994.7 1,035.3 1,017.3 Ϯ 6.5 995.6 1,035.7 1,017.3 Ϯ 6.6 995.0 1,035.8

* Values are the mean Ϯ SD of hourly measures for the study period. effect in our results may be the temperate climate of the with more extreme weather conditions may present a dif- Sydney region where the study was conducted. Regions ferent result, but further research is needed. Interestingly,

*(993 ؍ Table 3. Exposure and estimates of ﬁxed parameters for included weather conditions (n

Case window Control window Control window (onset day) 1 (1 week ago) 2 (1 month ago) OR (95% CI)† P 1 SD†

Precipitation, mm Within 1 hour‡ 1.12 Ϯ 4.36 1.24 Ϯ 5.45 1.21 Ϯ 4.75 0.98 (0.89–1.07) 0.59 5 At 24 hours§ 1.14 Ϯ 5.02 1.20 Ϯ 4.56 1.25 Ϯ 4.89 0.98 (0.89–1.09) 0.76 5 Average value over 24 hours¶ 1.20 Ϯ 3.42 1.36 Ϯ 4.15 1.35 Ϯ 3.77 0.95 (0.87–1.05) 0.32 4 Change from 0–24 to 25–48 hours# 0.15 Ϯ 4.05 Ϫ0.23 Ϯ 4.58 0.03 Ϯ 4.90 1.08 (1.00–1.18) 0.05 4 Temperature, °C Within 1 hour‡ 17.72 Ϯ 5.36 17.64 Ϯ 5.44 17.26 Ϯ 5.72 1.05 (0.90–1.22) 0.52 5 At 24 hours§ 17.65 Ϯ 5.46 17.61 Ϯ 5.52 17.21 Ϯ 5.56 1.03 (0.88–1.19) 0.75 5 Average value over 24 hours¶ 16.91 Ϯ 4.28 16.95 Ϯ 4.33 16.55 Ϯ 4.60 0.96 (0.82–1.13) 0.63 4 Change from 0–24 to 25–48 hours# Ϫ0.03 Ϯ 2.32 Ϫ0.13 Ϯ 2.17 Ϫ0.17 Ϯ 2.12 1.04 (0.96–1.13) 0.30 2 Relative humidity, % Within 1 hour‡ 62.92 Ϯ 20.74 63.94 Ϯ 20.36 64.84 Ϯ 20.53 0.91 (0.81–1.03) 0.14 21 At 24 hours§ 63.29 Ϯ 21.05 64.16 Ϯ 20.99 64.32 Ϯ 20.72 0.93 (0.82–1.04) 0.20 21 Average value over 24 hours¶ 66.36 Ϯ 14.37 67.27 Ϯ 13.90 67.78 Ϯ 13.96 0.92 (0.83–1.01) 0.09 14 Change from 0–24 to 25–48 hours# 0.67 Ϯ 12.19 0.72 Ϯ 11.88 0.38 Ϯ 11.55 1.00 (0.91–1.09) 0.93 12 Wind speed, km/hour Within 1 hour‡ 16.56 Ϯ 10.37 16.55 Ϯ 10.78 16.32 Ϯ 10.67 1.00 (0.89–1.13) 0.99 11 At 24 hours§ 17.26 Ϯ 10.90 16.30 Ϯ 10.60 16.45 Ϯ 11.04 1.17 (1.04–1.32) 0.01 11 Average value over 24 hours¶ 16.11 Ϯ 8.38 15.77 Ϯ 8.44 15.74 Ϯ 8.82 1.09 (0.96–1.23) 0.19 8 Change from 0–24 to 25–48 hours# Ϫ0.15 Ϯ 6.14 Ϫ0.10 Ϯ 5.89 Ϫ0.68 Ϯ 6.19 0.99 (0.91–1.08) 0.85 6 Wind gust, km/hour Within 1 hour‡ 22.41 Ϯ 13.16 22.53 Ϯ 13.92 22.05 Ϯ 13.53 0.99 (0.88–1.11) 0.81 14 At 24 hours§ 23.22 Ϯ 13.88 22.08 Ϯ 13.43 22.38 Ϯ 14.20 1.14 (1.02–1.28) 0.02 14 Average value over 24 hours¶ 21.54 Ϯ 10.24 21.16 Ϯ 10.39 21.14 Ϯ 11.00 1.07 (0.95–1.20) 0.26 10 Change from 0–24 to 25–48 hours# Ϫ0.16 Ϯ 8.43 Ϫ0.16 Ϯ 8.08 Ϫ0.96 Ϯ 8.49 1.00 (0.92–1.09) 1.00 8 Wind direction, degrees true Within 1 hour‡ 186.01 Ϯ 107.96 182.36 Ϯ 107.00 194.28 Ϯ 104.60 1.05 (0.95–1.16) 0.39 107 At 24 hours§ 191.75 Ϯ 102.40 194.45 Ϯ 106.80 189.42 Ϯ 108.09 0.97 (0.87–1.07) 0.51 107 Average value over 24 hours¶ 186.67 Ϯ 59.92 184.78 Ϯ 59.87 187.39 Ϯ 59.66 1.05 (0.94–1.18) 0.36 60 Change from 0–24 to 25–48 hours# Ϫ0.06 Ϯ 53.94 0.25 Ϯ 54.38 0.79 Ϯ 55.69 0.99 (0.91–1.08) 0.90 54 Air pressure, hPa Within 1 hour‡ 1,017.65 Ϯ 6.49 1,017.26 Ϯ 6.53 1,017.34 Ϯ 6.40 1.06 (0.98–1.16) 0.16 6 At 24 hours§ 1,017.62 Ϯ 6.39 1,017.23 Ϯ 6.31 1,017.29 Ϯ 6.49 1.07 (0.98–1.17) 0.14 6 Average value over 24 hours¶ 1,017.61 Ϯ 6.03 1,017.12 Ϯ 6.08 1,017.23 Ϯ 6.04 1.09 (1.00–1.19) 0.06 6 Change from 0–24 to 25–48 hours# Ϫ0.13 Ϯ 4.85 0.09 Ϯ 4.91 0.08 Ϯ 4.63 0.95 (0.86–1.04) 0.28 5

* Values are the mean Ϯ SD unless indicated otherwise. OR ϭ odds ratio; 95% CI ϭ 95% confidence interval. † Per 1-SD increase. ‡ Exposure defined as the value at 1 hour immediately before the pain onset. § Exposure defined as the value at 24 hours immediately before the pain onset. ¶ Exposure defined as the average value from 0–24 hours immediately before the pain onset. # Exposure defined as the average value from 0–24 hours immediately before the pain onset minus the average value from 25–48 hours the day before pain onset. 119

1872 Steffens et al the popular belief about temperature, precipitation, air Analysis and interpretation of data. Steffens, Maher, Li, Ferreira, pressure, wind direction, and humidity and their associa- Pereira, Koes, Latimer. tion with back pain seems to be stronger than the data would support. It should be noted, however, that there may be musculoskeletal conditions other than LBP that REFERENCES may be affected by weather parameters, and this is an 1. Patberg WR, Rasker JJ. Weather effects in rheumatoid arthritis: important area for further research. from controversy to consensus. A review. J Rheumatol 2004; Further studies are needed to confirm our findings in 31:1327–34. wider populations and also to determine whether there is 2. Smedslund G, Mowinckel P, Heiberg T, Kvien TK, Hagen KB. a subgroup of people in whom weather is more strongly Does the weather really matter? A cohort study of influences of weather and solar conditions on daily variations of joint associated with back pain onset. Case-crossover designs pain in patients with rheumatoid arthritis. Arthritis Rheum could be conducted in other musculoskeletal pain condi- 2009;61:1243–7. tions. The importance of indoor temperatures, character- 3. Abasolo L, Tobias A, Leon L, Carmona L, Fernandez-Rueda JL, istics of housing or work, and air conditioning use should Rodriguez AB, et al. Weather conditions may worsen symp- be taken into account, since if a majority of the events toms in rheumatoid arthritis patients: the possible effect of temperature. Reumatol Clin 2013;9:226–8. occur within the home, the results obtained in relation to 4. Shutty MS Jr, Cundiff G, DeGood DE. Pain complaint and the meteorological variables may be biased toward the null weather: weather sensitivity and symptom complaints in value. The small association found with higher wind chronic pain patients. Pain 1992;49:199–204. speed may be better explained in regions where summer 5. Jamison RN, Anderson KO, Slater MA. Weather changes and pain: perceived influence of local climate on pain complaint and winter are generally relatively extreme, rather than the in chronic pain patients. Pain 1995;61:309–15. moderate hot and cold of temperate regions. 6. Roth-Isigkeit A, Thyen U, Stoven H, Schwarzenberger J, In addition, co-exposure to multiple triggers (e.g., phys- Schmucker P. Pain among children and adolescents: restric- ical and meteorological factors) may increase risk of back tions in daily living and triggering factors. Pediatrics 2005; pain more than simple exposure to one meteorological 115:e152–62. 7. Tenias JM, Estarlich M, Fuentes-Leonarte V, Iniguez C, Ball- trigger. We are unaware of any study that has investigated ester F. Short-term relationship between meteorological vari- this for musculoskeletal conditions. Additionally, using ables and hip fractures: an analysis carried out in a health area the case-crossover design to investigate whether exposure of the Autonomous Region of Valencia, Spain (1996-2005). to weather parameters is associated with pain exacerbation Bone 2009;45:794–8. 8. Maclure M. The case-crossover design: a method for studying or flares, in a sample of people with chronic back pain, transient effects on the risk of acute events. Am J Epidemiol may provide useful explanations for disease etiology and 1991;133:144–53. how to improve quality of life. 9. Steffens D, Ferreira ML, Maher CG, Latimer J, Koes BW, Blyth In conclusion, this study shows that common weather FM, et al. Triggers for an episode of sudden onset low back parameters previously believed to influence musculoskel- pain: study protocol. BMC Musculoskelet Disord 2012;13:7. 10. De Vet HC, Heymans MW, Dunn KM, Pope DP, van der Beek etal pain do not increase the risk of an episode of LBP. This AJ, Macfarlane GJ, et al. Episodes of low back pain: a proposal study did, however, find a weak association between ex- for uniform definitions to be used in research. Spine 2002;27: posure to higher wind speed, wind gust, and back pain 2409–16. onset, but the magnitude of this effect was small and 11. Mittleman MA, Maclure M, Robins JM. Control sampling strategies for case-crossover studies: an assessment of relative therefore not clinically important. efficiency. Am J Epidemiol 1995;142:91–8. 12. StataCorp. Stata statistical software. College Station (TX): StataCorp; 2013. ACKNOWLEDGMENTS 13. Tsuchihashi Y, Yorifuji T, Takao S, Suzuki E, Mori S, Doi H, The authors would like to thank all of the clinicians who et al. Environmental factors and seasonal influenza onset in participated in the TRIGGERS for low back pain study. Okayama city, Japan: case-crossover study. Acta Med Okayama 2011;65:97–103. 14. Vaneckova P, Bambrick H. Cause-specific hospital admissions AUTHOR CONTRIBUTIONS on hot days in Sydney, Australia. PloS One 2013;8:e55459. 15. Pienimaki T. Cold exposure and musculoskeletal disorders All authors were involved in drafting the article or revising it and diseases: a review. Int J Circumpolar Health 2002;61:173– critically for important intellectual content, and all authors ap- 82. proved the final version to be published. Mr. Steffens had full 16. Madrigano J, Mittleman MA, Baccarelli A, Goldberg R, Melly access to all of the data in the study and takes responsibility for S, von Klot S, et al. Temperature, myocardial infarction, and the integrity of the data and the accuracy of the data analysis. mortality: effect modification by individual- and area-level Study conception and design. Steffens, Maher, Li, Ferreira, characteristics. Epidemiology 2013;24:439–46. Pereira, Koes, Latimer. 17. Mukamal KJ, Wellenius GA, Suh HH, Mittleman MA. Acquisition of data. Steffens, Maher, Li, Ferreira, Pereira, Koes, Weather and air pollution as triggers of severe headaches. Latimer. Neurology 2009;72:922–7. 120

Chapter Six

Does magnetic resonance imaging predict future low back pain? A systematic review

Chapter Six is published as:

Steffens D, Hancock MJ, Maher CG, Williams C, Jensen TS, Latimer J. Does magnetic resonance imaging predict future low back pain? A systematic review. European Journal of Pain. 2014; 18:755-765. 121

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “Does magnetic resonance imaging predict future low back pain? A systematic review”, we confirm that Daniel Steffens has made the following contributions:

 Data extraction, analysis and interpretation of the findings  Writing of the manuscript and critical appraisal of the content

Mark J Hancock Date: 01.01.2015

Christopher G Maher Date: 01.01.2015

Ciaran Williams Date: 01.01.2015

Tue S Jensen Date: 01.01.2015

Jane Latimer Date: 01.01.2015

122

REVIEW ARTICLE Does magnetic resonance imaging predict future low back pain? A systematic review D. Steffens1, M.J. Hancock2, C.G. Maher1, C. Williams3, T.S. Jensen4,5, J. Latimer1

1 The George Institute for Global Health, Sydney Medical School, The University of Sydney, Australia 2 Discipline of Physiotherapy, Faculty of Human Sciences, Macquarie University, Sydney, Australia 3 Active Physiotherapy Newtown, Sydney, Australia 4 Research Department, The Spine Centre of Southern Denmark, Middelfart, Denmark 5 Institute of Regional Health Services Research, Faculty of Health Sciences, University of Southern Denmark, Odense, Denmark

Correspondence Abstract Daniel Steffens E-mail: [email protected] Background and Objective: Magnetic resonance imaging (MRI) has the potential to identify pathology responsible for low back pain (LBP). Funding sources However, the importance of findings on MRI remains controversial. We None. aimed to systematically review whether MRI findings of the lumbar spine predict future LBP in different samples with and without LBP. Conflict of interest None declared. Databases and Data Treatment: MEDLINE, CINAHL and EMBASE databases were searched. Included were prospective cohort studies investi- Accepted for publication gating the relationship between baseline MRI abnormalities of the lumbar 22 October 2013 spine and clinically important LBP outcome at follow-up. We excluded cohorts with specific diseases as the cause of their LBP. Associations doi:10.1002/j.1532-2149.2013.00427.x between MRI findings and LBP pain outcomes were extracted from eligible studies. Results: A total of 12 studies met the inclusion criteria. Six studies presented data on participants with current LBP; one included a sample with no current LBP, three included a sample with no history of LBP and two included mixed samples. Due to small sample size, poor overall quality and the heterogeneity between studies in terms of participants, MRI findings and clinical outcomes investigated, it was not possible to pool findings. No consistent associations between MRI findings and outcomes were identified. Single studies reported significant associations for Modic changes type 1 with pain, disc degeneration with disability in samples with current LBP and disc herniation with pain in a mixed sample. Conclusions: The limited number, heterogeneity and overall quality of the studies do not permit definite conclusions on the association of MRI findings of the lumbar spine with future LBP (PROSPERO: CRD42012002342).

1. Introduction LBP is likely to be a major contributor to the lack of progress in management. After excluding people with Despite the enormous costs of low back pain (LBP) nerve root pain and serious pathologies (e.g., fracture and thousands of clinical trials, little progress has been and cancer), around 90–95% of LBP sufferers are clas- made in the management of LBP with most treat- sified as having non-specific low back pain (NSLBP) ments having only small effects (Deyo, 2004; Keller reflecting the inability to identify a clear source for the et al., 2007). Limited understanding of the aetiology of pain (van Tulder et al., 2006). If the source of pain

MRI predicting future LBP D. Steffens et al.

or otherwise of MRI findings to the development or Databases • We identified relevant studies by conducting course of LBP (Endean et al., 2011). Longitudinal electronic searches of MEDLINE, CINAHL and studies provide the possibility to investigate if MRI EMBASE, and by examining the reference lists of findings are associated with important outcomes such identified papers. as the development of future LBP in currently asymptomatic people or the course of LBP in people with current LBP. We are unaware of any previous system- What does this study add? atic review of prospective longitudinal studies investi- • No consistent associations were identified across gating the association between MRI findings of the multiple studies. lumbar spine and future LBP. Therefore, the specific • Only Modic changes and disc degeneration at review questions were: baseline predicted poor outcome from low back (1) Do MRI findings predict future LBP in people with pain at follow-up in single studies. no history of LBP? • The importance of magnetic resonance imaging (2) Do MRI findings predict future LBP in people with (MRI) findings in predicting clinical outcomes is no current LBP, but a previous history of LBP? unclear due to the limited number and quality of (3) Do MRI findings predict the course of LBP in studies and the heterogeneity between studies in people with current LBP? terms of the participants, MRI findings and clini- (4) Do MRI findings predict future LBP in a mixed cal outcomes investigated. sample of participants with and without current LBP? could be identified in at least some of these patients 2. Methods then it is possible that more targeted and effective A review protocol was specified in advance and registered treatments could be found. on PROSPERO: International prospective register of sys- Magnetic resonance imaging (MRI) has the poten- tematic reviews (http://www.crd.york.ac.uk/PROSPERO/ tial to identify pathology responsible for LBP display_record.asp?ID=CRD42012002342). (Schwarzer et al., 1995); however, the importance of findings on MRI remains controversial (Modic and 2.1 Search strategy Ross, 2007). Previous studies reveal high rates of abnormalities on MRI in people without LBP (Boden A systematic search of the literature from the earliest record et al., 1990; Boos et al., 1995, 2000; Jarvik et al., to the first week of May 2012 was undertaken using a highly 2001). Consequently, it can be difficult to determine sensitive search strategy suggested by the Cochrane Back Review Group together with a strategy for searching whether abnormalities seen on MRI are truly the MEDLINE for prognosis studies. We combined text and, cause of LBP since morphological changes are where appropriate, Medical Subject Headings terms for LBP, common in asymptomatic subjects. The lack of a back pain or backache; and inception, survival, life tables, log widely accepted gold standard test contributes to the rank, prospective or follow-up studies; and magnetic reso- difficulty in assessing the diagnostic accuracy of MRI nance imaging. The complete search strategies from all data- findings (Hancock et al., 2012). bases are included in Supporting Information Appendix S1. The presence of pathology on MRI in people We identified relevant studies by electronic searches of without LBP does not necessarily mean MRI findings general biomedical and science databases (MEDLINE, are not important to the aetiology of LBP (Hancock CINAHL and EMBASE), as well as examined the reference et al., 2012). In many other health conditions (e.g., lists of identified papers. A final list of included studies was cardiovascular disease), pathology (e.g., atherosclero- sent to experts in the field who reviewed the list for possible omissions. This search had no language restrictions. sis) exists in people without current symptoms; however, this pathology has been shown to be very important to the aetiology of the disorder (Duncan 2.2 Study selection et al., 2007). MRI findings in currently asymptomatic To be included, studies were required to meet all of the people may represent markers of early pre- following criteria: symptomatic disease that is later characterized by epi- (a) Prospective cohort study. This included secondary analy- sodes of pain and/or disability. sis of randomized controlled trials (RCTs), but only where Most previous research investigating the association the treatment provided was conservative, or data on the between MRI and LBP has been cross sectional. These conservative arm were presented separately to the surgical studies provide only weak evidence of the importance arm.

D. Steffens et al. MRI predicting future LBP

(b) Participants underwent baseline MRI reporting any reviewers were discussed and resolved during a consensus abnormality of the lumbar spine (e.g., but not limited to disc meeting. Each criteria was scored positive ‘Yes’ or negative degeneration, disc herniation, facet joint arthropathy, Modic ‘No’. A positive score indicates sufficient information and a changes, high intensity zone). positive assessment. The same criteria have been used in (c) Reported a clinically important LBP outcome at previous systematic reviews on the prognosis of acute LBP follow-up (e.g., pain, disability or a global measure of (Pengel et al., 2003; Costa et al., 2012). Methodological recovery). quality was not an inclusion criterion. (d) The association between MRI findings (baseline or change) and LBP outcomes (or sufficient raw data to calcu- 2.4 Analysis late a measure of association) was reported. Simply reporting a p-value was insufficient to meet this criterion. Our intention was to pool results, but due to the heterogene- We excluded studies that included patients with specific ity between the studies in terms of the participants, MRI diseases such as tumours, fractures, inflammatory arthritis findings and clinical outcomes, it was not possible or appro- and cauda equina syndrome, but not sciatica. One reviewer priate to pool findings. The results are presented descriptively. screened the titles and abstracts to exclude clearly irrelevant Studies of participants with current LBP were used to articles. For each potentially eligible study, the full article investigate if MRI findings predicted the course of LBP. was obtained and independently assessed for inclusion by Studies of participants with no current LBP were used to two review authors (D.S., M.J.H., C.G.M., J.L., C.W. or assess if MRI findings were risk factors for future LBP. T.S.J.). Any discrepancies were resolved by discussion. 3. Results 2.3 Data extraction 3.1 Selection of studies 2.3.1 Study characteristics Our search identified 6666 citations (2304 MEDLINE, Data extraction was completed independently by two 265 CINAHL, 4635 EMBASE) after removing all dupli- reviewers. Included studies were categorized into four main cates (n = 538). After review of title and abstract, 6608 groups: (1) sample with current LBP; (2) mixed sample of records were excluded. Five additional studies that people with and without current LBP; (3) sample with no met the inclusion criteria were identified after consul- current LBP, but previous LBP; and (4) sample with no tation with experts in the field, resulting in a total of history of LBP. Data were extracted from the selected studies 63 full-text articles eligible for assessment. When regarding number of subjects, sample source, age, follow-up reviewing full-text articles for the 63 articles, a further duration, MRI findings, clinical outcomes and strength of 23 were excluded as they were not prospective association between MRI finding and clinical outcome. studies, 16 did not assess a LBP outcome at follow-up, When sufficient raw data were available, odds ratios (OR) and 95% confidence intervals (95% CI) were calculated. 8 did not present an association between MRI findings and LBP outcome (or provide raw data to enable cal- culation of this) and 4 did not perform MRI at base- 2.3.2 Methodological quality line. A table of excluded full-text articles and the The methodological quality of each of the studies was primary reason for exclusion is included in Supporting assessed independently by two reviewers (D.S., M.J.H., Information Appendix S2. Therefore, 13 studies met C.G.M., J.L., C.W. or T.S.J.) using a standardized checklist of all the inclusion criteria (Borenstein et al., 2001; pre-defined criteria (Pengel et al., 2003; Costa et al., 2012). Elfering et al., 2002; Carragee et al., 2005, 2006a,b; The checklist is a modified version based on theoretical con- Jarvik et al., 2005; Modic et al., 2005; Kleinstuck siderations and methodological aspects described by Altman et al., 2006; McNee et al., 2011; Hellum et al., 2012; (2001). These criteria comprised (1) definition of study Jensen et al., 2012; Keller et al., 2012) (Fig. 1). For sample (description of participant source and inclusion and one study, results from the same cohort were identi- exclusion criteria); (2) representative sample of the target fied in two different publications (Carragee et al., sample (participants selected by random selection or as con- 2006a,b), but only the published report with the full secutive cases); (3) follow-up rate >80% (outcome data were available for at least 80% of participants at 3-month results was included in our analysis (Carragee et al., follow-up or later); (4) adequate follow-up time (at least one 2006a). Consequently, the methodological quality and prognostic outcome was followed up at 3 months or later); results are based on 12 studies. (5) interpretable prognostic outcomes available (raw data, percentages, survival rates or continuous outcome reported); 3.2 Methodological quality and (6) blinding (assessor unaware of at least one prognostic factor, used to predict prognostic outcome, at the time prog- Most studies defined the study sample (91.5%). Only nostic outcome was measured). Disagreements among the four studies (33%) described methods for assembling a

MRI predicting future LBP D. Steffens et al.

Records identified 2012; Keller et al., 2012); two included a mixed through searches MEDLINE = 2304 sample of people with and without LBP (Carragee CINAHL = 265 et al., 2005, 2006a), one included a sample with no EMBASE = 4635 (n = 7204) current LBP, but previous LBP symptoms (Jarvik et al., Duplicate records excluded (n = 538) 2005), and three studies included a sample of patients Records screened after with no history of LBP (Boos et al., 2000; Borenstein duplicates removed Records excluded (n = 6666) •Not related to low back pain; et al., 2001; Elfering et al., 2002). •Ineligible designs; •No MRI at baseline. The samples were recruited from primary health (n = 6608) Full-text articles Full-text articles assessed care (Elfering et al., 2002; Jarvik et al., 2005; identified by for eligibility experts in the field (n =63) Kleinstuck et al., 2006), secondary health care (Boos (n = 5) Records excluded • 23 not a prospective study; et al., 2000; Carragee et al., 2005, 2006a; Modic et al., • 8 no association between MRI findings and LBP 2005; McNee et al., 2011; Hellum et al., 2012; Jensen Studies included - outcome; 12 different cohorts • 16 no LBP outcome at et al., 2012; Keller et al., 2012) or volunteers from the (n = 13) follow-up; community (Borenstein et al., 2001). The majority of • 4 no MRI at baseline. (n = 51) the studies focused largely or completely on men and women of working age, but some also included older Figure 1 Flowchart of search strategy. participants (ranging from 18 to 84 years old). The follow-up period ranged from 12 to 84 months. representative sample of the target sample. Nine studies (75%) had a follow-up of at least 80% and all studies had a follow-up for at least one prognostic 3.4 Association of MRI findings with outcome at 3 months or longer and quantified progno- clinical outcomes sis. Eleven studies used blinded assessment (91.5%). Due to the heterogeneity of samples, MRI findings and Data for individual studies are presented in Table 1. clinical outcomes, it was not possible to combine the results of the studies included. The findings of all 12 included studies are presented in Tables 3 and Figs. 2 3.3 Study characteristics and 3. A comprehensive description of each study is provided ORs and 95% CI were calculated from the extracted in Table 2. Six studies included a sample of current raw data for the following studies (Boos et al., 2000; LBP (Modic et al., 2005; Kleinstuck et al., 2006; Elfering et al., 2002; Carragee et al., 2006a; McNee McNee et al., 2011; Hellum et al., 2012; Jensen et al., et al., 2011; Hellum et al., 2012; Jensen et al., 2012).

Table 1 Methodological quality assessment of included studies. Deﬁnition of Representative Follow-up rate Follow-up >3 Outcomes Blinded Study study samplea sampleb >80%c monthsd reportede outcomef

Borenstein et al., 2001 Yes No No Yes Yes Yes Carragee et al., 2006a Yes Yes Yes Yes Yes Yes Carragee et al., 2005 Yes Yes Yes Yes Yes Yes Elfering et al., 2002 No No Yes Yes Yes No Jarvik et al., 2005 Yes Yes Yes Yes Yes No Keller et al., 2012 Yes No No Yes Yes Yes McNee et al., 2011 Yes Yes No Yes Yes Yes Jensen et al., 2012 Yes No Yes Yes Yes Yes Kleinstuck et al., 2006 Yes No Yes Yes Yes Yes Modic et al., 2005 Yes No Yes Yes Yes Yes Boos et al., 2000 Yes No Yes Yes Yes Yes Hellum et al., 2012 Yes No Yes Yes Yes Yes aDescription of participant source and inclusion and exclusion criteria. bParticipants selected by random selection or as consecutive cases. cOutcome data were available for at least 80% of participants at 3-month follow-up or later. dAt least one prognostic outcome was followed up at 3 months or later. eRaw data, percentages, survival rates or continuous outcome reported. fAssessor unaware of at least one prognostic factor, used to predict prognostic outcome, at time prognostic outcome was measured.

Mean age Outcome scoring Follow-up, Study Sample source (range) MRI ﬁndings MRI scoring (threshold) Clinical outcomes (threshold) duration (%)

Population with current LBP Keller et al., 2012 269 patients referred to university 49.7 (20–60) Modic changes type 1 Yes/No (present) Patient global impression of 1–7 (≥3 not recovered) 12 months (40%) spine clinic improvement (PGI-I) Modic changes type 2 Yes/No (present) Patient global impression of 1–7 (≥3 not recovered) improvement (PGI-I) Hellum et al., 66 trial participants recruited from 41.5 (?) Modic changes type 1 Yes/No (present) Disability (ODI) 0–100 (≥15points) 12 months (100%) 2012 university hospitals (only Modic changes type 2 Yes/No (present) participants in conservative Disc degeneration (height % reduction (≥40%) ® group) reduction) Disc degeneration (signal intensity) Graded 1 to 4 (>3) Facet joint arthopathy No or slight to ≥moderate (≥moderate) High-intensity zone Yes/No (present) McNee et al., 323 patients of hospital radiology ?(20–64) Number of MRI abnormalities 0 to 4 (all levels) Pain >14 days past 4 weeks Yes/No (?) 22.2 months (mean) 2011 department Disc degeneration ? (present) Disability past 4 weeks (RM) 0–24 (≥11 points) (74%) Jensen et al., 96 patients of specialist outpatient 46 (21–60) Modic changes type 1 Yes/No (present) Pain (NPRS) 0–10 (no improvement, 14 months (100%) 2012 spine clinic Modic changes type 1 (change in Size 0–4 (increase) change score ≤0) size) Kleinstuck et al., 53 participants recruited from 44 (?) High intensity zone Yes/No (present) Pain last 2 weeks (VAS) 0–10 (NA) 12 months (90%) 2006 local media advertisements Disc bulge Yes/No (present) Disability (RM) 0–24 (NA) Disc degeneration Graded 5–25 (> 16) Modic change type 1 and type 2 Graded 0–15 (?) Modic et al., 246 patients non-responders to a 43 (?) Canal stenosis ? (present) Disability (RM) 0–24 (<50% improvement) 24 months (80%) 2005 preoperative intensive Nerve root compression ? (present) conservative in-/outpatient Disc herniation ? (present) treatment programme Populations of people with and without current LBP Carragee et al., 200 patients from university 39.4 (?) Disc degeneration Graded 1–5 (grade ≥3) Pain (NPRS) 0–10 (≥6for≥1 week) 60 months (100%) 2006a hospital Endplate changes Mild to severe (>moderate) Canal stenosis Mild to severe (>moderate) Disability (ODI) 0–100 (?) Carragee et al., 100 patients from university 42 (?) Disc herniation Yes/No (present) Pain (remission) Yes/No (6 months) 63 months (mean) 2005 hospital (100%) Population with no current LBP, but previous LBP Jarvik et al., 2005 148 Veterans Affairs outpatients 54 (36–71) Disc herniation Protrusion or extrusion (present) Pain (PFI) 1–6 (pain >2 or any of the 36 months (88%) Nerve root contact Contact/ deviation/ compression other three symptoms (present) as >1) Canal stenosis Mild to severe (> moderate) Populations with no history of LBP

Elfering et al., 41 trauma patients presenting to a ? (20–50) Disc degeneration (change) Same or worse (worsening) Pain (NQ) Days with pain last month 62 months (mean) LBP future predicting MRI u Pain J Eur 2002 university clinic (>8) (100%) Boos et al., 2000 46 patients of a university clinic ?(?) Disc herniation Yes/No (present) Pain (NQ) Days with pain last month 62 months (mean) Nerve root contact Contact/deviation/ compression (>8) (100%)

•• (present) 21)••–•• (2013) Disc degeneration Grade 1 to 5 (present) Borenstein et al., 67 volunteers recruited through 42 (?) MRI abnormalities (change) ? (worsening) Pain (ordinal scale) 0–5 (?) 84 months (75%) 2001 advertising or by word of month

?, not reported; LBP, low back pain; MRI, magnetic resonance imaging; NA, not applicable (continuous outcome); NPRS, numerical pain rating scale; NQ, Nordic questionnaire; ODI, Oswestry disability questionnaire; PFI, pain frequency

5 index; PGI-I, patient global impression of improvement scale; RM, Roland Morris disability questionnaire; VAS, visual analogue scores. 126 127

MRI predicting future LBP D. Steffens et al.

Table 3 Association between presence of MRI ﬁndings and poor clinical outcome. MRI ﬁndings present Study Clinical outcome (follow-up duration) OR (95% CI) unless indicated

Populations with current LBP Modic changes type 1 Jensen et al., 2012 Pain (14 months) 6.2 (1.9–20.2) Modic changes type 1 (change in size) Jensen et al., 2012 Pain (14 months) 1.0 (0.4–2.4) Modic changes type 1 Keller et al., 2012 Impression of improvement (12 months) 1.1 (0.5–2.7)a Modic changes type 1 Hellum et al., 2012 Disability (12 months) 1.4 (0.5–4.1) Modic changes type 2 Hellum et al., 2012 Disability (12 months) 0.5 (0.2–1.4) Modic changes type 2 Keller et al., 2012 Impression of improvement (12 months) 1.4 (0.7–2.6)a Modic changes type 1 or type 2 Hellum et al., 2012 Disability (12 months) 0.3 (0.1–1.6) Modic changes type 1 and type 2 Kleinstuck et al., 2006 Pain (12 months) −2(p = 0.1)b Modic changes type 1 and type 2 Hellum et al., 2012 Disability (12 months) 1.0 (0.3–3.3) Modic changes type 1 and type2 Kleinstuck et al., 2006 Disability (12 months) 0.4 (p = 0.2)b Disc degeneration McNee et al., 2011 Pain (22.2 months, mean) 1.6 (0.9–2.8) Disc degeneration Kleinstuck et al., 2006 Pain (12 months) −1(p = 0.4)b Disc degeneration McNee et al., 2011 Disability (22.2 months, mean) 2.2 (1.2–4.0) Disc degeneration Kleinstuck et al., 2006 Disability (12 months) −0.8 (p = 0.1)b Disc degeneration (height reduction) Hellum et al., 2012 Disability (12 months) 0.7 (0.2–2.1) Disc degeneration (signal intensity) Hellum et al., 2012 Disability (12 months) 0.7 (0.2–3.0) High-intensity zone Kleinstuck et al., 2006 Pain (12 months) −2(p = 0.1)b High-intensity zone Hellum et al., 2012 Disability (27 months) 1.5 (0.5–4.5) High-intensity zone Kleinstuck et al., 2006 Disability (12 months) 0.3 (p = 0.5)b Disc bulge Kleinstuck et al., 2006 Pain (12 months) 1 (p = 0.4)b Disc bulge Kleinstuck et al., 2006 Disability (12 months) 0.8 (p = 0.1)b Facet joint arthopathy (≥moderate) Hellum et al., 2012 Disability (27 months) 0.7 (0.2–3.0) Disc herniation Modic et al., 2005 Disability (24 months) 0.4 (0.2–0.7) Canal stenosis Modic et al., 2005 Disability (24 months) 2.4 (0.9–6.7) Nerve root compression Modic et al., 2005 Disability (24 months) 0.8 (0.4–1.5) ≥1 MRI abnormality McNee et al., 2011 Pain (22.2 months, mean) 0.7 (0.3–1.6) ≥2 MRI abnormalities McNee et al., 2011 Pain (22.2 months, mean) 0.8 (0.4–1.8) ≥3 MRI abnormality McNee et al., 2011 Pain(22.2 months, mean) 1.0 (0.4–2.2) ≥1 MRI abnormality McNee et al., 2011 Disability (22.2 months, mean) 1.2 (0.5–2.9) ≥2 MRI abnormalities McNee et al., 2011 Disability (22.2 months, mean) 1.3 (0.5–3.2) ≥3 MRI abnormality McNee et al., 2011 Disability (22.2 months, mean) 1.5 (0.6–3.7) Populations of people with and without current LBP Disc degeneration (grade 5) Carragee et al., 2006a Pain (60 months) 4.4 (p = 0.08)c Disc degeneration (graded ≥3) Carragee et al., 2006a Disability (60 months) 1.0 (0.4–2.3) Canal stenosis Carragee et al., 2006a Pain (60 months) 2.9 (p = 0.09)c Canal stenosis (≥moderate) Carragee et al., 2006a Disability (60 months) 2.1 (0.8–5.1) Endplate changes (≥moderate) Carragee et al., 2006a Pain (60 months) 2.5 (p = 0.1)c Disc herniation Carragee et al., 2005 Pain (63 months, mean) 0.2 (p = 0.01)c Population with no current LBP, but previous LBP Disc herniation (extrusion) Jarvik et al., 2005 Pain (36 months) 1.2 (0.4–3.4)a Disc herniation (protrusion) Jarvik et al., 2005 Pain (36 months) 0.5 (0.3–0.9)a Nerve root contact Jarvik et al., 2005 Pain (36 months) 2.2 (0.6–8.0)a Canal stenosis (>moderate) Jarvik et al., 2005 Pain (36 months) 1.9 (0.8–4.8)a Populations with no history of LBP Disc degeneration Boos et al., 2000 Pain (62 months, mean) 2.1 (0.1–26.9) Disc degeneration (change) Elfering et al., 2002 Pain (62 months, mean) 4.8 (0.3–69.3) Disc herniation Boos et al., 2000 Pain (62 months, mean) 0.7 (0.1–9.3) Nerve root contact Boos et al., 2000 Pain (62 months, mean) 8.8 (0.6–117.2) MRI abnormalities (change) Borenstein et al., 2001 Pain (84 months) 3.5d

Odds ratios greater than 1 indicate greater odds of poor outcome in those with MRI feature than those without. LBP, low back pain; MRI, magnetic resonance imaging. aHazard ratios (95% confidence interval). Hazard ratios greater than 1 indicate greater incidence of outcome in those with MRI feature than those without. A positive association was defined as CI limits above 1. bβ (p-value) positive values indicate that the presence of the given MRI finding at baseline was associated with a poorer outcome. cOdds ratios (p-value), no confidence interval reported. dRelative risk (no confidence interval provided). Relative risk greater than 1 indicates that the poor outcome is more likely to develop in people with the MRI feature.

D. Steffens et al. MRI predicting future LBP

Number of events MRI Outcome MRI findings present Study Outcome (time) Odds ratio (95% CI) OR (95% CI) + / - + / - Modic changes type 1 Jensen et al., 2012 Pain (14 months) 74 / 22 47 / 49 6.2 (1.9–20.2) Modic changes type 1 (change in size) Jensen et al., 2012 Pain (14 months) ? / ? ? / ? 1.0 (0.4–2.4) Modic changes type 1 Hellum et al., 2012 Disability (12 months) 44 / 22 31 / 35 1.4 (0.5–4.1) Modic changes type 2 Hellum et al., 2012 Disability (12 months) 44 / 22 31 / 35 0.5 (0.2–1.4) Modic changes type 1 or 2 Hellum et al., 2012 Disability (12 months) 9 / 59 31 / 37 0.3 (0.1–1.6) Modic changes type 1 and 2 Hellum et al., 2012 Disability (12 months) 53 / 13 30 / 36 1.0 (0.3–3.3) Disc degeneration McNee et al., 2011 Pain (22.2 months, mean) 11 / 13 85 / 155 1.6 (0.9–2.8) Disc degeneration McNee et al., 2011 Disability (22.2 months, mean) 11 / 130 70 / 170 2.2 (1.2–4.0) Disc degeneration (height reduction) Hellum et al., 2012 Disability (12 months) 20 / 45 30 / 35 0.7 (0.2–2.1) Disc degeneration (signal intensity) Hellum et al., 2012 Disability (12months) 10 / 55 30 / 35 0.7 (0.2–3.0) High intensity zone Hellum et al., 2012 Disability (27 months) 42 / 23 30 / 35 1.5 (0.5–4.5) Facet joint arthopathy (≥moderate) Hellum et al., 2012 Disability (27 months) 10 / 55 30 / 35 0.7 (0.2–3.0) Disc herniation Modic et al., 2005 Disability (24 months) ? / ? ? / ? 0.4 (0.2–0.7) Canal stenosis Modic et al., 2005 Disability (24 months) ? / ? ? / ? 2.4 (0.9–6.7) Nerve root compression Modic et al., 2005 Disability (24 months) ? / ? ? / ? 0.8 (0.4–1.5) ≥1 MRI abnormality McNee et al., 2011 Pain (22.2 months, mean) 208 / 32 85 / 155 0.7 (0.3–1.6) Figure 2 Forest plot presenting association ≥2 MRI abnormalities McNee et al., 2011 Pain (22.2 months, mean) 173 / 32 75 / 130 0.8 (0.4–1.8) ≥3 MRI abnormality McNee et al., 2011 Pain (22.2 months, mean) 106 / 32 56 / 82 1.0 (0.4–2.2) between magnetic resonance imaging (MRI) ≥1 MRI abnormality McNee et al., 2011 Disability (22.2 months, mean) 208 / 32 70 / 170 1.2 (0.5–2.9) ≥2 MRI abnormalities McNee et al., 2011 Disability (22.2 months, mean) 173 / 32 61 / 144 1.3 (0.5–3.2) ﬁndings and low back pain (LBP) outcomes on ≥3 MRI abnormality McNee et al., 2011 Disability (22.2 months, mean) 106 / 32 44 / 94 1.5 (0.6–3.7) a sample with current LBP. 0.1 0.2 0.5 1 2 5 10

CI = 1.9–20.2) over a 14-month period. There was no 3.4.1 Current LBP association between the change in size of Modic type 1 Six studies investigated the association between MRI changes and change in LBP intensity (OR = 1.0; 95% findings and a range of clinical outcomes in a sample CI = 0.4–2.4), dichotomized into improvement with current LBP [five with chronic LBP (Kleinstuck (decrease in LBP intensity) or no improvement (no et al., 2006; McNee et al., 2011; Hellum et al., 2012; change or increase of LBP intensity). Hellum et al. Jensen et al., 2012; Keller et al., 2012) and one with (2012) reported data from a randomized trial con- acute LBP (Modic et al., 2005)]. Of the six studies, four ducted at five hospitals. One hundred and fifty-five reported on associations of Modic changes [type 1 patients were randomized to receive either conserva- (Kleinstuck et al., 2006; Jensen et al., 2012) and/or tive treatment or disc replacement surgery. In the con- type 2 (Kleinstuck et al., 2006; Hellum et al., 2012; servatively treated group (n = 66), neither type 1 or Keller et al., 2012)], three on disc degeneration type 2 Modic changes (OR = 0.3; 95% CI = 0.1–1.6) (Kleinstuck et al., 2006; McNee et al., 2011; Hellum nor type 1 or type 2 Modic changes (OR = 1.0; 95% et al., 2012), two on high-intensity zone (HIZ) CI = 0.3–3.3) were significantly associated with dis- (Kleinstuck et al., 2006; Hellum et al., 2012) and only ability. Kleinstuck et al. (2006) investigated a total of one reported on each of disc bulge (Kleinstuck et al., 53 patients who participated in a randomized clinical 2006), disc herniation (Modic et al., 2005), canal trial of active therapy for chronic LBP. Baseline end- stenosis (Modic et al., 2005), facet joint arthropathy plate changes (defined as Modic changes type 1 and (Hellum et al., 2012), the number of MRI abnormali- type 2) were not significantly associated with pain ties (≥ 1to≥ 3) (McNee et al., 2011) and nerve root intensity or disability at 12-month follow-up. compression (Modic et al., 2005). Only one study reported recovery rate at follow-up The association of Modic changes with LBP (Keller et al., 2012) in a sample of patients with (Kleinstuck et al., 2006; Jensen et al., 2012; Keller current LBP (n = 269). After 1 year, 40% of patients et al., 2012) and with disability (Kleinstuck et al., rated themselves as recovered. Neither type 1 Modic 2006; Hellum et al., 2012; Keller et al., 2012) was changes [hazard ratio (HR) = 1.1; 95% CI = 0.5–2.7] investigated by four studies. Jensen et al. (2012) nor type 2 Modic changes (HR = 1.4; 95% CI = 0.7– investigated 96 patients recruited from a specialized 2.6) were significantly associated with not recovering outpatient spine clinic. This sample was drawn from over 12 months. an RCT that reported no effect of treatment, and Disc degeneration was reported in three studies reported a significant association of Modic change type (Kleinstuck et al., 2006; McNee et al., 2011; Hellum 1 with worsening of LBP intensity (OR = 6.2; 95% et al., 2012). McNee et al. (2011) investigated 323

Number of events MRI finding present Study Outcome (time) MRI Outcome Odds ratio (95% CI) OR (95% CI) Sample with no history of LBP + / - + / - Disc degeneration Boos et al., 2000 Pain (62 months, mean) 15 / 15 3 / 27 2.1 (0.1–26.9) Disc degeneration (change) Elfering et al., 2002 Pain (62 months, mean) 19 / 21 3 / 32 4.8 (0.3–69.3) Disc herniation Boos et al., 2000 Pain (62 months, mean) 22 / 8 3 / 27 0.7 (0.1–9.3) Nerve root contact Boos et al., 2000 Pain (62 months, mean) 7 / 23 3 / 27 8.8 (0.6–117.2) Figure 3 Forest plot presenting association Sample with mixed LBP Disc degeneration (graded ≥3) Carragee et al, 2006a Pain (60 months) 153 / 47 44 / 156 1.0 (0.4–2.3) between magnetic resonance imaging (MRI) Canal stenosis (≥moderate) Carragee et al, 2006a Pain (60 months) 26 / 174 44 / 156 2.1 (0.8–5.1) ﬁndings and low back pain (LBP) outcomes. 0.01 0.1 1 10 100

MRI predicting future LBP D. Steffens et al. patients with mechanical LBP, and found positive Carragee et al. (2006a) investigated 200 patients associations between disc degeneration with LBP from a University Hospital and found that none of the (OR = 1.6; 95% CI = 0.9–2.8) and disability (OR = 2.2; baseline MRI findings significantly predicted serious 95% CI = 1.2–4.0). In the study by Kleinstuck et al. LBP episodes nor disability after 60 months. Grade 5 (2006), disc degeneration was not associated with disc degeneration (OR = 4.40; p = 0.08), moderate/ pain, or disability at 12-month follow-up. Hellum severe endplate changes (OR = 2.5; p = 0.1) and canal et al. (2012) found no association between baseline stenosis (OR = 2.9; p = 0.09) were weakly, but not sig- findings of disc degeneration (signal intensity) and nificantly, associated with serious LBP episodes. 12-month disability in participants treated conserva- Another study by Carragee et al. (2005) reported that tively (OR = 0.7; 95% CI = 0.2–3.0). Similarly, disc the presence of disc herniation was associated with degeneration (height reduction) was not associated pain (OR = 0.2; p = 0.01) at follow-up, in 100 patients with 12-month disability (OR = 0.7; 95% CI = 0.2– from a University Hospital. 2.1). One study investigated the association between the number of MRI abnormalities (≥ 1to≥ 3) at baseline 3.4.3 No current LBP, but previous LBP and future LBP and disability (McNee et al., 2011). One study investigated the association between base- Neither one nor more MRI abnormalities were associ- line MRI findings and LBP in 128 Veterans Affairs ated with pain or disability at follow-up. patients who were initially asymptomatic (Jarvik Kleinstuck et al. (2006) found no association et al., 2005). Baseline MRI findings of nerve root between the presence of baseline disc bulge and HIZ contact (HR = 2.2; 95% CI = 0.6–8.0) and central with LBP or disability outcomes. Hellum et al. (2012) spinal stenosis (HR = 1.9; 95% CI = 0.8–4.8) produced also investigated the association between HIZ and dis- non-significant HRs for future new LBP. The study ability. HIZ in the conservatively treated group found that having a disc herniation (protrusion) (OR = 1.5; 95% CI = 0.5–4.5) was not significantly (HR = 0.5; 95% CI = 0.3–0.9) reduced the likelihood associated with disability at 12 months. of future LBP, whereas a disc herniation (extrusion) Disc herniation, severe canal stenosis and nerve root (HR = 1.2; 95% CI = 0.4–3.4) did not predict future compression were investigated in one study (Modic new LBP episodes. et al., 2005). A total of 246 patients were randomized to either the early information arm of the study, with MRI results provided within 48 h, or the second arm 3.4.4 No history of LBP of the study, where both patients and physicians were blinded to MRI results. Disability at 24-month Three studies provided estimates of the association follow-up occurred 0.4 times (95% CI = 0.2–0.7) as between MRI findings [disc degeneration (Boos et al., often among patients with disc herniation at baseline 2000; Elfering et al., 2002), disc herniation (Boos as among patients without disc herniation. Severe et al., 2000), neural compromise (Boos et al., 2000) canal stenosis (OR = 2.4; 95% CI = 0.9–6.7) and nerve and worsening abnormalities on MRI (Borenstein root compression (OR = 0.8; 95% CI = 0.4–1.5) were et al., 2001)] with future LBP in a sample of asymp- not significant predictors of future disability. tomatic individuals with no history of LBP. Similarly, Hellum et al. (2012) reported that base- Borenstein et al. (2001) followed up 50 subjects for line findings of facet joint arthropathy did not predict 84 months and reported a relative risk that LBP would disability at 12 months (OR = 0.7; 95% CI = 0.2–3.0) develop in individuals with worsening abnormalities in the conservatively treated group. on MRI scans of 3.5 (no CIs or p-values reported). Disc degeneration was investigated by two studies. Elfering et al. (2002) reported the association between 3.4.2 Mixed samples (sample with and without disc degeneration and LBP after 60 months in 41 current LBP) patients. There was no significant association with LBP Two studies investigated a mixed sample (subjects in those with disc degeneration (OR = 4.8; 95% without or mild LBP symptoms and/or with chronic CI = 0.3–69.3), although the very wide CIs indicate non-lumbar pain), reporting associations between disc lack of statistical power. Boos et al. (2000) investigated degeneration, endplate changes, canal stenosis 46 asymptomatic individuals and did not find a signifi- (Carragee et al., 2006a) and disc herniation (Carragee cant association between disc degeneration and LBP et al., 2005) with disability (Carragee et al., 2006a) (OR = 2.1; 95% CI = 0.1–26.9). Similarly, disc hernia- and LBP (Carragee et al., 2005, 2006a) outcomes. tion and neural compromise were not predictors of

D. Steffens et al. MRI predicting future LBP

LBP (OR = 0.7; 95% CI = 0.1–9.3 and OR = 8.8; 95% reviewed particularly the different samples, MRI find- CI = 0.6–117.2, respectively). ings investigated and clinical outcomes. For this reason, we were unable to pool study results. Some 4. Discussion studies included participants from RCTs (Modic et al., 2005; Kleinstuck et al., 2006; Hellum et al., 2012; Jensen et al., 2012) and it is not possible to determine 4.1 Statement of principal findings how the interventions may have impacted on the out- Twelve studies met the inclusion criteria and were comes. The association between MRI finding and included. Six studies investigated participants with outcome may vary as a result of confounding due to current LBP, two investigated mixed samples including treatment. The association between MRI findings and people with and without LBP, one included a sample outcomes in a sample that underwent surgery with no current LBP, but previous LBP symptoms, and (Hellum et al., 2012) were not reported in this review. three studies included a sample of participants with no Another factor that could have influenced the results history of LBP. No consistent associations were iden- of the included studies was that most studies used a tified across multiple studies. Single studies reported different MRI protocol, sequences and training of MRI significant associations for Modic changes type 1 readers. The current review only investigated MRI (OR = 6.2; 95% CI = 1.9–20.2) with pain, and disc findings and does not provide evidence on the value of degeneration with disability (OR = 2.2; 95% CI = 1.2– plain radiographs or computed tomography scans in 4.0) in samples with current LBP, and disc herniation predicting future LBP. (OR = 0.2; p = 0.01) with pain in a mixed sample with and without LBP. This systematic review reveals that there are rela- 4.3 Comparison with other studies tively few studies that have investigated MRI findings One previous systematic review has investigated the as predictors of future LBP. Not only are there few cross-sectional association of vertebral endplate signal studies, but these studies are mostly small (number of changes (Modic changes) with current NSLBP (Jensen participants ranging from 41 to 323) and investigate et al., 2008). A relatively strong association between different MRI findings in a range of different samples vertebral signal changes and NSLBP was found in 7 of (i.e., current LBP, no current LBP or mixed) and use 10 studies with ORs ranging from 2.0 to 19.9. In our different outcome measures. As a result, it is not pos- search, we chose to include articles that investigated sible to draw firm conclusions about the ability of MRI the association of a variety of baseline MRI findings findings to predict future LBP. Some MRI findings (i.e., disc degeneration, HIZ, herniation) with pain and were statistically associated with future LBP in single disability outcomes in a range of different samples. studies, but comparable studies were not identified to Furthermore, despite other systematic reviews in this enable confirmation of these findings. The small field including cross-sectional studies, we chose to sample size of most studies meant that some poten- exclude these studies as they cannot logically deter- tially clinically important associations may have been mine if baseline MRI findings predict future LBP. missed. It remains unclear whether the MRI findings have important associations with LBP outcomes or whether no important associations truly exist. 4.4 Meaning of the study While single studies reported significant associations 4.2 Strengths and weaknesses of the study for Modic changes type 1, disc herniation and disc To our knowledge, this is the first systematic review to degeneration for future LBP, there remains consider- summarize the available evidence of MRI findings pre- able uncertainty about the importance of MRI find- dicting future LBP. We used a very sensitive search ings. Definitive conclusions are not possible as the strategy previously used in other high-quality LBP available studies typically enrolled small non- prognosis studies (Pengel et al., 2003; Costa et al., representative samples and the results were inconsis- 2012) and MRI systematic reviews (Chou et al., 2011; tent between studies. Perhaps the only clear result to Endean et al., 2011), making it very likely that all emerge from this review is that there is a paucity of eligible studies were included in our review. We also high-quality studies in this important area. This consulted experts in the field to reduce the risk of absence of evidence is in contrast with the rapidly missing any important articles. A limitation of the increasing use of MRI in patients with LBP (Chou present study is the heterogeneity of the studies et al., 2012).

MRI predicting future LBP D. Steffens et al.

Many of the MRI findings are truly continuous mea- J.L. designed the study, extracted data and edited the manu- sures with different degrees of severity. Despite this, script. All authors approved the final version. included studies tend to dichotomize these (i.e., present or no present). This may result in loss of important data, and therefore, more research is References needed to investigate if the relationships are linear or Altman, D.G. (2001). Systematic reviews of evaluations of prognostic if important thresholds exist. variables. BMJ 323, 224–228. Boden, S.D., Davis, D.O., Dina, T.S., Patronas, N.J., Wiesel, S.W. (1990). Abnormal magnetic-resonance scans of the lumbar spine in asymptom- 4.5 Recommendations for future research atic subjects. A prospective investigation. J Bone Joint Surg 72, 403– 408. Further large, high-quality studies are clearly needed Bombardier, C. (2000). Outcome assessments in the evaluation of treatment of spinal disorders: Summary and general recommendations. to help determine the clinical meaningfulness of MRI Spine 25, 3100–3103. findings in relation to LBP. Additionally, a number of Boos, N., Rieder, R., Schade, V., Spratt, K.F., Semmer, N., Aebi, M. (1995). factors related to reporting of included studies could be The diagnostic accuracy of magnetic resonance imaging, work perception, and psychosocial factors in identifying symptomatic disc hernia- improved. Firstly, all of the included studies had some tions. Spine 20, 2613–2625. methodological shortcomings. Future studies should Boos, N., Semmer, N., Elfering, A., Schade, V., Gal, I., Zanetti, M., Kissling, describe the selection of participants in more detail. R., Buchegger, N., Hodler, J., Main, C.J. (2000). Natural history of individuals with asymptomatic disc abnormalities in magnetic reso- Assessment and reporting of clinical outcomes was not nance imaging: Predictors of low back pain-related medical monsulta- well standardized. For example, pain intensity was tion and work incapacity. Spine 25, 1484–1492. measured in nine studies with four different scales. Borenstein, D.G., O’Mara, J.W.J., Boden, S.D., Lauerman, W.C., Jacobson, A., Platenberg, C., Schellinger, D., Wiesel, S.W. (2001). The value of Standardized methods for reporting outcomes, based magnetic resonance imaging of the lumbar spine to predict low-back on published recommendations, would greatly pain in asymptomatic subjects: A seven-year follow-up study. J Bone improve future studies (Bombardier, 2000). The use of Joint Surg 83, 1306–1311. Carragee, E., Alamin, T., Cheng, I., Franklin, T., Hurwitz, E. (2006a). Does a standardized MRI protocol, same sequences and minor trauma cause serious low back illness? Spine 31, 2942–2949. training of MRI readers is also highly recommended. Carragee, E., Alamin, T., Cheng, I., Franklin, T., van den Haak, E., The investigation of association between MRI find- Hurwitz, E. (2006b). Are first-time episodes of serious LBP associated with new MRI findings? Spine J 6, 624–635. ings and clinical outcomes is complicated. All reviewed Carragee, E.J., Alamin, T.F., Miller, J.L., Carragee, J.M. (2005). Disco- studies focused on the association between a single graphic, MRI and psychosocial determinants of low back pain disability imaging findings and pain and/or disability. Future and remission: A prospective study in subjects with benign persistent back pain. Spine J 5, 24–35. studies should investigate which MRI findings typically Chou, D., Samartzis, D., Bellabarba, C., Patel, A., Luk, K.D.K., Kisser, occur together and whether clusters of findings are J.M.S., Skelly, A.C. (2011). Degenerative magnetic resonance imaging more predictive of outcome than single findings. changes in patients with chronic low back pain: A systematic review. Spine 36, S43–S53. Chou, R., Deyo, R., Jarvik, J. (2012). Appropriate use of lumbar imaging for evaluation of low back pain. Radiol Clin North Am 50, 569–585. 5. Conclusion Costa, L.C.M., Maher, C.G., Hancock, M.J., McAuley, J.H., Herbert, R.D., Costa, L.O.P. (2012). The prognosis of acute and persistent low-back This review shows that there are few (heterogeneous) pain: A meta-analysis. CMAJ 184, 613–624. Deyo, R.A. (2004). Treatments for back pain: Can we get past trivial longitudinal studies that have investigated the associa- effect? Ann Intern Med 141, 957–958. tion of lumbar spine MRI findings and LBP outcomes, Duncan, R., Peat, G., Thomas, E., Hay, E., McCall, I., Croft, P. (2007). which indicates the need for further research. Symptoms and radiographic osteoarthritis: Not as discordant as they are made out to be? Ann Rheum Dis 66, 86–91. Elfering, A., Semmer, N., Birkhofer, D., Zanetti, M., Hodler, J., Boos, N. (2002). Risk factors for lumbar disc degeneration: A 5-year prospective Acknowledgement MRI study in asymptomatic individuals. Spine 27, 125–134. Endean, A., Palmer, K.T., Coggon, D. (2011). Potential of magnetic reso- We thank Prof Jeffrey Jarvik for reviewing included studies nance imaging findings to refine case definition for mechanical low and suggesting possible additional studies. back pain in epidemiological studies: A systematic review. Spine 36, 160–169. Hancock, M., Maher, C., Macaskill, P., Latimer, J., Kos, W., Pik, J. (2012). MRI findings are more common in selected patients with acute low Author contributions back pain than controls? Eur Spine J 21, 240–246. Hellum, C., Johnsen, L.G., Gjertsen, O., Berg, L., Necklmann, G., D.S. designed and managed the study, planned analysis, Grundnes, O., Rossvoll, I., Skouen, J.S., Brox, J.I., Storheim, K. (2012). drafted manuscript, extracted and analysed data. M.J.H. Predictors of outcome after surgery with disc prothesis and rehabilitation in patients with chronic low back pain and degenerative disc: designed and managed the study, planned analysis, extracted 2-year follow-up. Eur Spine J 21, 681–690. and analysed data. C.G.M. designed and managed the study, Jarvik, J.G., Hollingworth, W., Heagerty, P.J., Haynor, D.R., Boyko, E.J., extracted data and edited the manuscript. C.W., T.S.J. and Deyo, R.A. (2005). Three-year incidence of low back pain in an initially

D. Steffens et al. MRI predicting future LBP

asymptomatic cohort: Clinical and imaging risk factors. Spine 30, 1541– Modic, M.T., Obuchowski, N.A., Ross, J.S., Brant-Zawadzki, M.N., Grooff, 1548. P.N., Mazanec, D.J., Benzel, E.C. (2005). Acute low back pain and Jarvik, J.J., Hollingworth, W., Heagerty, P., Haynor, D.R., Deyo, R.A. radiculopathy: MR imaging findings and their prognostic role and effect (2001). The longitudinal assessment of imaging and disability of the on outcome. Radiology 237, 597–604. back (LAIDBack) study. Spine 26, 1158–1166. Modic, M.T., Ross, J.S. (2007). Lumbar degenerative disk disease. Radiol- Jensen, R.K., Leboeuf-Yde, C., Wedderkopp, N., Sorensen, J.S., Jensen, ogy 245, 43–61. T.S., Manniche, C. (2012). Is the development of Modic changes asso- Pengel, L.H.M., Herbert, R.D., Maher, C.G., Refshauge, K.M. (2003). ciated with clinical symptoms? A 14-month cohort study with MRI. Eur Acute low back pain: Systematic review of its prognosis. BMJ 327, Spine J 21, 2271–2279. 323. Jensen, T.S., Karppinen, J., Sorensen, J.S., Niinimaki, J., Leboeuf-Yde, C. Schwarzer, A.C., Aprill, C.N., Derby, R., Fortin, J., Kine, G., Bogduk, N. (2008). Vertebral endplate signal changes (Modic change): A systematic (1995). The prevalence and clinical features of internal disc disruption literature review of prevalence and association with non-specific low in patients with chronic low back pain. Spine 20, 1878–1883. back pain. Eur Spine J 17, 1407–1422. van Tulder, M., Becker, A., Bekkering, T., Breen, A., Del Real, M.T.G., Keller, A., Boyle, E., Skog, T.A., Cassidy, J.D., Bautz-Holter, E. Hutchinson, A., Koes, B., Laerum, E., Malmivaara, A. (2006). Chapter (2012). Are Modic changes prognostic for recovery in a cohort 3. European guidelines for the management of chronic nonspecific low of patients with non-specific low back pain? Eur Spine J 21, 418– back pain. Eur Spine J 15, S169–S191. 424. Keller, A., Hayden, J., Bombardier, C., van Tulder, M. (2007). Effect sizes of non-surgical treatments of non-specific low-back pain. Eur Spine J 16, 1776–1788. Supporting Information Kleinstuck, F., Dvorak, J., Mannion, A.F. (2006). Are ‘structural abnormalities’ on magnetic resonance imaging a contraindication to the suc- Additional Supporting Information may be found in the cessful conservative treatment of chronic nonspecific low back pain? online version of this article at the publisher’s web-site: Spine 31, 2250–2257. McNee, P., Shambrook, J., Harris, E.C., Kim, M., Sampson, M., Palmer, Appendix S1. Search strategy. K.T., Coggon, D. (2011). Predictors of long-term pain and disability in patients with low back pain investigated by magnetic resonance Appendix S2. List of excluded full-text articles and the imaging: A longitudinal study. BMC Musculoskelet Disord 12, 234. primary reason for exclusion.

Appendix S1. Search strategy

MEDLINE 19 13 AND 18 1 exp cohort studies/ 20 (‘Magnetic resonance Imaging’/exp AND [embase]/lim) 2 incidence/ 21 (‘MRI’/exp AND [embase]/lim) 3 follow-up studies.mp. 22 (‘Magnetic resonance’/exp AND [embase]/lim) 4 prognos$.mp. 23 (‘NMR’/exp AND [embase]/lim) 5 predict$.mp. 24 (‘Nuclear magnetic resonance’/exp AND [embase]/lim) 6 course.mp. 25 (‘Disc degeneration’/exp AND [embase]/lim) 7 inception.mp. 26 (‘Desiccation’/exp AND [embase]/lim) 8 survival.mp. 27 (‘Loss of disc height’/exp AND [embase]/lim) 9 logistic.mp. 28 (‘Bulge’/exp AND [embase]/lim) 10 Cox.mp. 29 (‘Protrusion’/exp AND [embase]/lim) 11 life tables.mp. 30 (‘Extrusion’/exp AND [embase]/lim) 12 log rank.mp. 31 (‘Nerve root compromise’/exp AND [embase]/lim) 13 or/1–12 32 (‘Annular tear’/exp AND [embase]/lim) 14 low back pain.mp. 33 (‘Endplate changes’/exp AND [embase]/lim) 15 back pain.mp. 34 (‘Stenosis’/exp AND [embase]/lim) 16 Lumbago.mp. 35 (‘Facet degeneration’/exp AND [embase]/lim) 17 Back injuries.mp. 36 (‘High intensity zone’/exp AND [embase]/lim) 18 Backache.mp. 37 (‘Modic changes’/exp AND [embase]/lim) 19 or/14–18 38 (‘Degenerative disc disease’/exp AND [embase]/lim) 20 Magnetic resonance Imaging/ 39 (‘Spondylolisthesis’/exp AND [embase]/lim) 21 MRI.mp. 40 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29 or 30 or 31 22 Magnetic adj5 resonance.mp. or 31 or 33 or 34 or 35 or 36 or 37 or 38 or 39 or 40 23 NMR.mp. 19 AND 40 24 Nuclear magnetic resonance.mp. CINAHL 25 Disc degeneration.mp. 1 (MH ‘Prospective Studies+’) 26 Desiccation.mp. 2 (MH ‘incidence+’) 27 Loss of disc height.mp. 3 ‘predic*’ 28 Bulge.mp. 4 (MH ‘prognosis+’) 29 Protrusion.mp. 5 ‘course’ 30 Extrusion.mp. 6 ‘Inception’ 31 Nerve root compromise.mp. 7 (MH ‘Survival Analisis+’) or (MH ‘Cox Proportional Hazards Model’) 32 Annular tear.mp. 8 (MH ‘Logistic Regression+’) 33 Endplate changes.mp. 9 (MH ‘Life Table Methods’) 34 Stenosis.mp. 10 (MH ‘Log-Rank Test’) 35 Facet degeneration.mp. 11 1or2or3or4or5or6or7or8or9or10 36 High intensity zone.mp. 12 (MH ‘Low Back Pain’) or (MH ‘Back Pain+’) 37 Modic changes.mp. 13 (MH ‘Lumbago+’) 38 Degenerative disc disease.mp. 14 (‘Back injuries+’) 39 Spondylolisthesis.mp. 15 (‘Backache+’) 40 or/20–39 16 12 or 13 or 14 or 15 41 13 AND 19 AND 40 17 11 AND 16 EMBASE 18 (MH ‘MRI’) or (MH ‘Magnetic Resonance’) 1 (‘cohort analysis’/exp AND [embase]/lim) 19 (MH ‘NRI’) or (MH ‘Nuclear Magnetic Resonance’) 2 (‘incidence’/exp AND [embase]/lim) 20 (MH ‘Disc degeneration+’) 3 (‘follow up’/exp OR ‘follow up’ AND [embase]/lim) 21 (MH ‘Desiccation+’) 4 (‘prognosis’/exp OR prognos* AND [embase]/lim) 22 (MH ‘Loss of disc heigh+’) 5 (‘prediction’/exp OR predict* AND [embase]/lim) 23 (MH ‘Bulge+’) 6 (‘disease course’/exp OR ‘course’ AND [embase]/lim) 24 (MH ‘Protrusion+’) 7 (‘inception’/exp AND [embase]/lim) 25 (MH ‘Extrusion+’) 8 (‘survival’/exp OR ‘survival’ AND [embase]/lim) 26 (MH ‘Nerve root compromise+’) 9 (‘logistic regression analysis’/exp OR ‘logistic’ AND 27 (MH ‘Annular tear+’) [embase]/lim) 28 (MH ‘Endplate changes+’) 10 (‘proportional hazards model’/exp OR ‘cox’ AND [embase]/lim) 29 (MH ‘Stenosis+’) 11 (‘life table’/exp OR ‘life table’ OR ‘life tables’/exp OR ‘life tables’ AND 30 (MH ‘Facet degeneration+’) [embase]/lim) 31 (MH ‘High intensity zone+’) 12 (‘log rank test’/exp OR ‘log rank’ AND [embase]/lim) 32 (MH ‘Modic changes+’) 13 1or2or3or4or5or6or7or8or9or10or11or12 33 (MH ‘Degenerative disc disease+’) 14 (‘low back pain’/exp OR ‘low back pain’ AND [embase]/lim) 34 (MH ‘Spondylolisthesis+’) 15 (‘backache’/exp OR ‘low back pain’/exp AND [embase]/lim) 35 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26 or 27 or 28 or 29 16 (‘Lumbago’/exp AND [embase]/lim) or 30 or 31 or 32 or 33 or 34 17 (‘Back injuries’/exp AND [embase]/lim) 36 17 AND 35 18 14 or 15 or 16 or 17 134

Appendix S2. List of excluded full-text articles and the primary reason for exclusion

# Study Title First reason for excluding

#1 Albert, H. B. et al. 2007 Modic changes following lumbar disc herniation No association #2 Ash, L. M. et al. 2008 Effects of diagnostic information, per se, on patient outcomes in acute radiculopathy No LBP outcome at and low back pain follow-up #3 Baranto, A. et al. 2006 Back pain and degenerative abnormalities in the spine of young elite divers: A 5-year No LBP outcome at follow-up magnetic resonance imaging study follow-up #4 Baranto, A., M. et al. 2009 Back pain and MRI changes in the thoraco-lumbar spine of top athletes in four different No LBP outcome at sports: A 15-year follow-up study follow-up #5 Bartolozzi, C. et al. 1991 The incidence of disk changes in volleyball players. The magnetic resonance findings Not a prospective study #6 Beattie, P. F. et al. 2000 Associations between patient report of symptoms and anatomic impairment visible on Not a prospective study lumbar magnetic resonance imaging #7 Bennett, A. N. et al. 2008 Severity of baseline magnetic resonance imaging-evident sacroiliitis and HLA-B27 No association status in early inflammatory back pain predict radiographically evident ankylosing spondylitis at eight years #8 Bennett, D. L. et al. 2006 Lumbar spine MRI in the elite-level female gymnast with low back pain No LBP outcome at follow-up #9 Boden, S. D. et al. 1990 Abnormal magnetic-resonance scans of the lumbar spine in asymptomatic subjects. A No LBP outcome at prospective investigation follow-up #10 Borthakur, A. et al. 2011 T1(rho) magnetic resonance imaging and discography pressure as novel biomarkers for Not a prospective study disc degeneration and low back pain #11 Braithwaite, I. et al. 1998 Vertebral endplate (Modic) changes on lumbar spine MRI: correlation with pain Not a prospective study reproduction at lumbar discography #12 Buirski, G. 1992 Magnetic resonance signal patterns of lumbar discs in patients with low back pain. A Not a prospective study prospective study with discographic correlation #13 Buirski, G. and M. The symptomatic lumbar disc in patients with low-back pain. Magnetic resonance Not a prospective study Silberstein 1993 imaging appearances in both a symptomatic and control population #14 Buttermann, G. R. et al. Pain and disability correlated with disc degeneration via magnetic resonance imaging Not a prospective study 2008 in scoliosis patients #15 Buttermann, G.R. 2004 The effect of spinal steroid injections for degenerative disc disease No association #16 Carragee, E. J. et al. 2000 2000 Volvo Award winner in clinical studies: Lumbar high-intensity zone and Not a prospective study discography in subjects without low back problem. #17 Carragee, E. J. et al. 2004 Prospective controlled study of the development of lower back pain in previously No association asymptomatic subjects undergoing experimental discography #18 Chen, B., et al. 2001 The magnetic resonance imaging of the lumbar spine in out-patients with low back Not a prospective study pain #19 Cheung, K. M. et al. 2009 Prevalence and pattern of lumbar magnetic resonance imaging changes in a Not a prospective study population study of one thousand forty-three individuals #20 de Schepper, E. I. et al. The association between lumbar disc degeneration and low back pain: the influence of Not a prospective study 2010 age, gender, and individual radiographic features #21 Erkintalo, M. O. et al. 1995 Development of degenerative changes in the lumbar intervertebral disk: Results of a No association prospective MR imaging study in adolescents with and without low-back pain #22 Esposito, P. et al 2006 Predictive value of MRI vertebral end-plate signal changes (Modic) on outcome of No LBP outcome at surgically treated degenerative disc disease. Results of a cohort study including 60 follow-up patients #23 Fayad, F. et al. 2007 Relation of inflammatory modic changes to intradiscal steroid injection outcome in Not a prospective study chronic low back pain #24 Fayad, F. et al. 2009 Reliability of a modified Modic classification of bone marrow changes in lumbar spine Not a prospective study MRI. #25 Grable, H. R. 1993 Abnormal findings on magnetic resonance imaging in a group of motor vehicle Not a prospective study accident patients with low back pain #26 Graves, J. M. et al. 2012 Early imaging for acute low back pain: One-year health and disability outcomes among No MRI at baseline Washington state workers #27 Haig, A. J. et al. 2006 Predictors of pain and function in persons with spinal stenosis, low back pain, and no Not a prospective study back pain #28 Hollingworth, W. et al. Self reported health status and magnetic resonance imaging findings in patients with No association 1998 low back pain 135

Appendix S2. (continued)

# Study Title First reason for excluding

#29 Iwamoto, J. et al. 2005 Relationship between radiographic abnormalities of lumbar spine and incidence of low No MRI at Baseline back pain in high school rugby players: A prospective study #30 Jensen, O. K. et al. 2010 One-year prognosis in sick-listed low back pain patients with and without radiculopathy. No MRI at Baseline Prognostic factors influencing pain and disability #31 Jensen, R.K. et al 2011 Is the presence of Modic changes associated with the outcomes of different Not a prospective study treatments? A systematic critical review #32 Jensen, T. S. et al. 2009 Characteristics and natural course of vertebral endplate signal (Modic) changes in the No LBP outcome at Danish general population follow-up #33 Jensen, T. S. et al. 2010 Predictors of new vertebral endplate signal (Modic) changes in the general population No LBP outcome at follow-up #34 Jensen, T.S. et al 2007 Magnetic resonance imaging findings as predictors of clinical outcome in patients with No LBP outcome at sciatica receiving active conservative treatment follow-up #35 Kanayama, M. et al. 2009 Cross-sectional magnetic resonance imaging study of lumbar disc degeneration in 200 Not a prospective study healthy individuals: Clinical article #36 Kerttula, L. et al. 2012 Modic type I change may predict rapid progressive, deforming disc degeneration: a No LBP outcome at prospective 1-year follow-up study follow-up #37 Kuisma, M. et al 2006 A three-year follow-up of lumbar spine endplate (Modic) changes No LBP outcome at follow-up #38 Kujala, U. M. et al. 1996 Low-back pain in adolescent athletes No LBP outcome at follow-up #39 Kujala, U. M. et al. 1999 Prolonged low-back pain in young athletes: a prospective case series study of findings No association and prognosis #40 Luoma, K. et al. 2008 MRI follow-up of subchondral signal abnormalities in a selected group of chronic low Not a prospective study back pain patients #41 Luoma, K. et al. 2009 Relationship of Modic type 1 change with disc degeneration: A prospective MRI study No LBP outcome at follow-up #42 Marzo-Ortega, H. et al. Baseline and 1-year magnetic resonance imaging of the sacroiliac joint and lumbar No LBP outcome at 2009 spine in very early inflammatory back pain. Relationship between symptoms, follow-up HLA-B27 and disease extent and persistence #43 Matsui, H. et al. 1998 Familial predisposition for lumbar degenerative disc disease Not a prospective study #44 Mitra, D. et al. 2004 Longitudinal study of vertebral type-1 end-plate changes on MR of the lumbar spine Not a prospective study #45 Siepe, C.J. et al 2006 Clinical results of total lumbar disc replacement with prodisc II No association #46 Symmons, D. P. et al. A longitudinal study of back pain and radiological changes in the lumbar spines of No MRI at Baseline 1991 middle aged women. II. Radiographic findings #47 Takatalo, J. et al. 2011 Does lumbar disc degeneration on magnetic resonance imaging associate with low Not a prospective study back symptom severity in young Finnish adults? #48 Tung, G. A. et al. 1999 Spinal epidural abscess: Correlation between MRI findings and outcome Not a prospective study #49 Videman, T. et al. 2003 Associations between back pain history and lumbar MRI findings Not a prospective study #50 Waris, E. et al. 2007 Disc degeneration in low back pain: a 17-year follow-up study using magnetic No LBP outcome at resonance imaging follow-up #51 Williams, F. M. K. et al. Progression of lumbar disc degeneration over a decade: A heritability study No LBP outcome at 2011 follow-up 136

Chapter Seven

Prognosis of chronic low back pain in patients presenting to a private community- based group exercise program

Chapter Seven is published as:

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “Prognosis of chronic low back pain in patients presenting to a private community-based group exercise program”, we confirm that Daniel Steffens has made the following contributions:

 Conception and design of the research  Data collection  Analysis and interpretation of the findings  Writing of the manuscript and critical appraisal of the content

Mark J Hancock Date: 01.01.2015

Christopher G Maher Date: 01.01.2015

Jane Latimer Date: 01.01.2015

Rob Satchell Date: 01.01.2015

Manuela L Ferreira Date: 01.01.2015

Paulo H Ferreira Date: 01.01.2015

Melissa Partington Date: 01.01.2015

Anna L Bouvier Date: 01.01.2015

138 Eur Spine J (2014) 23:113–119 DOI 10.1007/s00586-013-2846-x

ORIGINAL ARTICLE

Prognosis of chronic low back pain in patients presenting to a private community-based group exercise program

Daniel Steffens • Mark J. Hancock • Chris G. Maher • Jane Latimer • Robert Satchell • Manuela Ferreira • Paulo H. Ferreira • Melissa Partington • Anna-Louise Bouvier

Received: 10 October 2012 / Revised: 8 May 2013 / Accepted: 1 June 2013 / Published online: 23 June 2013 Ó Springer-Verlag Berlin Heidelberg 2013

Abstract intensity improved markedly during the first 6 months Purpose To examine the prognosis and prognostic factors (35 %) with further minimal reductions up to 12 months for patients with chronic low back pain presenting to a (39 %). Interestingly, disability improved to a greater private, community-based, group exercise program. degree than pain (48 % improvement at 6 months) and Methods A total of 118 consecutive patients with chronic continued to improve throughout the 12 months (60 %). LBP were recruited. Baseline assessments included socio- Baseline pain intensity accounted for 10 % of the variance demographic characteristics, back pain history and clinical in the 1 year pain outcomes. Duration of current episode, examination findings. Primary outcome measures were pain baseline disability and educational level accounted for intensity and disability at 3, 6 and 12 months. Potential 15 % of the variation in disability at 12 months. prognostic factors to predict pain intensity and disability at Conclusions During a period of 12 months, patients with 12 months were assessed using a multivariate regression chronic LBP presenting to a private, community-based, model. group exercise program improved markedly, with greater Results 112 (95 %) participants were followed up at improvements in disability than pain. The predictors 12 months. The majority of participants were female investigated accounted for only 10 and 15 % of pain and (73 %), had high educational levels (82 %) and resided in disability outcomes, respectively. suburbs with a high socio-economic status (99 %). Pain Keywords Chronic low back pain Á Prognosis Á Disability Á D. Steffens (&) Á C. G. Maher Á J. Latimer Á M. Ferreira Musculoskeletal Division, The George Institute for Global Outcomes Health, University of Sydney, Level 13, 321 Kent Street, Sydney, NSW 2000, Australia e-mail: [email protected] Introduction M. J. Hancock Discipline of Physiotherapy, Faculty of Human Sciences, Clinical guidelines report the prognosis of chronic back Macquarie University, 2 Technology Place, Macquarie Park, pain to be poor [1]. A recent systematic review of the Sydney, NSW 2113, Australia course of acute and chronic LBP [2] found moderate R. Satchell within-study and between-study variability in both pain Coast Allied Health, 2/171 Prince Edward Ave, and disability outcomes. This strongly suggests that while Culburra Beach, NSW 2540, Australia the average prognosis may be poor this is not the case for a substantial proportion of people. The moderate levels of P. H. Ferreira Discipline of Physiotherapy, Faculty of Health Sciences, between-study variability suggest that factors related to The University of Sydney, Lidcombe, study design, setting and participants may significantly Sydney, NSW 2141, Australia impact on the reported prognoses. Previous studies have identified factors associated with M. Partington Á A.-L. Bouvier Physiocise, Movement for Life, Suite 14, 77 Penshurst Street, prognosis in chronic LBP [3–6]; however, the lack of Willoughby, Sydney, NSW 2068, Australia quality studies means prognostic factors for chronic LBP 123 139 114 Eur Spine J (2014) 23:113–119 remain unclear [7]. A large Australian cohort study [8] participate if they met all the following inclusion criteria; found previous sick leave due to LBP, high disability levels (1) chronic non-specific LBP (symptoms greater than or high pain intensity at onset of chronicity, low levels of 3 months duration, according to the classification proposed education, greater perceived risk of persistent pain, and by de Vet et al. [4]) with or without leg pain (2) pain on a being born outside Australia (immigrants) were associated numerical pain rating scale equal or greater than two out of with delayed recovery. Patients presenting for care in set- ten (3) adequate English communication for all data to be tings where these adverse prognostic factors are uncommon recorded and (4) baseline assessment performed less than may have a more favorable prognosis than widely reported. 4 weeks prior to exercise classes starting. Exclusion cri- Until now no study has investigated the prognosis of teria were (1) LBP that was not attributable to a recog- people with chronic low back pain attending a private, nizable, known specific pathology (e.g., infection, tumor, community-based, group exercise program. fracture, inflammatory disorder, cauda equina, radicular It is reasonable to expect that evidence-based treatments syndrome) [1], or (2) nerve root compromise (with at least which are endorsed in clinical guidelines will impact on the two of the following signs: myotomal weakness, derma- prognosis reported and may be an important source of the tomal sensory loss or hyporeflexia of the lower limb) or (3) between-study variability [2]. The European guideline for currently pregnant. The University of Sydney’s Human the management of chronic non-specific LBP recommends Research Ethics Committee approved the study. supervised exercise therapy as a first-line treatment in the management of chronic LBP. Furthermore, group exercise Baseline data collection and the use of cognitive behavioral therapy are also recommended [1]. Community-based group exercise programs Baseline data were collected by a physiotherapist as part of based on guideline recommendations for chronic LBP have the standard 90-min assessment for all patients prior to become increasingly common in the management of starting the group exercise program. Data collected inclu- chronic LBP. Many of these programs are private, and ded demographic information, baseline measures of out- given the requirement that patients must pay for the ser- comes and potential prognostic factors. vice, attract patients without work related injuries with higher average socioeconomic status and level of educa- Group exercise program tion. However, the prognosis of this particular group of private paying patients presenting to individualized group All participants were enrolled in a group exercise program, exercise classes is not well reported in the literature. at one of two private clinics in Sydney. The program In assessing the prognosis of chronic LBP it is important involved a strong educational component combined with to consider at a minimum both pain and disability out- physical retraining. Patients paid up front for a 10 week comes. A recent review found greater changes in pain than foundation program. Patients with private health insurance disability but lower absolute scores for disability than pain (87 %) were able to claim up to 50 % of the total cost of at 1 year, as the baseline pain levels were higher than the classes from their insurer. The foundation program disability [2]. A better understanding of the relative prog- focuses on movement and posture re-education, combined nosis in terms of pain and disability for people paying for with very specific pain behavior education, Acceptance and and receiving high quality group exercise programs is Commitment Therapy principles (focusing on awareness, important. Therefore, the aim of the present study was to acceptance, defusing of negative thoughts) [9] and CBT examine the prognosis and prognostic factors for private specifically in the area of decreasing catastrophizing. paying patients with chronic LBP who presented to a pri- Educational and physical elements of the program focused vate, community-based, group exercise program. on practicing activities of daily living, in particular sitting, standing, walking and bending. These were performed in front of large mirrors which gave patients maximum visual Materials and methods feedback about their physical movement patterns compared to an idealized goal (e.g., using neutral spine and hip and We conducted a prospective study of participants with knee flexion for bending). The environment was light, non- chronic LBP presenting to a private, community-based, threatening, interactive and involved music as well as group exercise program. education. Participants were also given a book of exercises and educational materials related to back pain and posture Setting and participants [10, 11]. At the end of the 10-week program, some participants enrolled in more advanced classes while other Consecutive patients with LBP who presented to a private, participants ceased participation in the group exercise community-based, group exercise program were invited to classes. Advanced classes focused on physical aspects such 123 140 Eur Spine J (2014) 23:113–119 115 as developing lumbo-pelvic stability, endurance with glu- instance neck/shoulders; (3) CT/MRI/disc bulges/degener- teal strength, and incorporating upper body strengthening ative changes on X-Ray; (4) high kinesiophobia (defined as in functional positions. Whether participants continued in scores [40 on the TAMPA Scale for Kinesiophobia); (5) classes was recorded, but did not impact on their poor response to hands-on treatment in the last few years; involvement in the study. (6) degree of constant pain (scored as yes/no); (7) unable to perform certain physical activities because of pain; Outcome variables (8) difficulty standing for prolonged periods; (9) difficulty sitting for prolonged periods; (10) trauma, such as Outcome measures were assessed at baseline, 3, 6 and MVA/fall. 12 months. Follow-up measures were collected when patients attended the exercise classes. Participants who did Statistical analysis not attend the classes during the week their follow-up assessment was due, were contacted by telephone and a Analyses were performed using SPSS version 20 (SPSS, telephone interview was conducted by a trained researcher. Inc., Chicago, IL). Descriptive statistics (mean ± SD) were used to summarize the baseline characteristics of the Primary outcomes patients and prognostic outcomes at baseline, 3, 6 and 12 months. Comparison between mean baseline and 3, 6 or The primary outcomes were average pain over the past 12 month outcomes were performed using Paired–samples week measured using the numerical pain rating scale [12] t test. P \ 0.05 was considered significant. scored from 0 (no pain) to 10 (worst pain possible) and To evaluate prognostic factors, univariate associations disability, measured using the Roland Morris Disability between each of the prognostic variables and pain scores Questionnaire [13], scored from 0 to 24, with higher scores and disability (RMDQ) at 12 months were assessed using indicating worse disability status. Pain was assessed as linear regression. Variables with significant univariate both change in pain and also the proportion of patients who associations (p \ 0.2) were entered into a backwards fully recovered from pain (score 0 or 1 for 1 month at time multivariate regression model. of assessment). For the regression analyses, the number of previous episodes was dichotomized (0–1 episodes/2 or more epi- Secondary outcomes sodes), duration of the current episode was recorded in days and log transformed, and educational level was Secondary outcomes were global impression of recovery dichotomized (0 = school, high school certificate and (Global Perceived Effect Scale) [12, 14], bothersomeness trade/diploma and 1 = advanced diploma, bachelor degree of pain (Bothersome of Pain Scale) [15], function (Patient- and postgraduate degree). specific Functional Scale) [16] and kinesiophobia (TAMPA Scale for Kinesiophobia) [17]. Results Prognostic factors From January 2010 to January 2011 consecutive patients A limited number of baseline prognostic factors were from two private physiotherapy clinics in Sydney, Aus- investigated. We limited the number of prognostic factors tralia were screened. One hundred and eighteen met the to one per ten patients [18]. The prognostic factors were inclusion criteria and all consented to enter the study. measures of pain intensity, previous episodes of LBP, Follow-up rates were high, with 116 (98 %), 114 (97 %) duration of current episode, disability, global perceived and 112 (95 %) participants being followed up at 3, 6 effect of improvement, function, kinesiophobia, educa- and 12 months, respectively. Baseline characteristics of tional level, location of pain (absence of pain below the the 118 participants are shown in Table 1. The majority knee), catastrophizing, patients enrolled to the second term of participants were female (73 %), had high educational (scored as 0 = patients that only attended the first term of levels (82 %), did not smoke (96 %) and resided in the classes or 1 = patients that attended two or more suburbs with a high socioeconomic status (99 %). All terms) and an 11-point prognosis scale. The Physiocise participants were private paying and not covered by prognosis scale consists of ten items and is scored by workers compensation. adding up the number of positive items. The score can Participants’ outcome data at baseline, 3, 6 and therefore vary from 0 to 10 with higher values indicating 12 months are presented in Table 2 and Fig. 1. Pain worse prognosis. The items are (1) greater than 5 years intensity, bothersomeness, disability and function reduced since first acute episode; (2) other painful areas, for during the follow-up period, with mean pain intensity, 123 141 116 Eur Spine J (2014) 23:113–119

Table 1 Baseline characteristics of study population score 0 or 1 for 1 month at time of assessment) from their Characteristics No (%) or mean LBP, respectively. (±SD) The results of the univariate analysis are presented in Table 3. Table 4 presents the results of the multivariate Gender (female) 86 (73) backward regression analysis. Baseline pain intensity was Age (years) 46.5 ± 12.5 the only independent predictor of 12-month pain scores and Weight (kg) 72.5 ± 15.5 accounted for only 10 % (R2 = 0.102) of the outcome Height (cm) 171 ± 8.5 variance (Table 4). Duration of current episode, disability Smoker 5 (4) and educational level were independent predictors of dis- Socioeconomic status (n = 118) ability accounting for 15 % of the variation on disability at Resides in suburb with mean household income 117 (99) 12 months (Table 4). above Australian mean Previous episodes of LBP (n = 113) Never 9 (7.5) Discussion 1–5 26 (22.5) 6–10 11 (9.5) Summary of main findings 11 or over 71 (60.5) Previous sick leave due to LBP (n = 116) 65 (55) The primary finding of this study is that patients with Previous treatment for LBP (n = 118) 114 (96.5) chronic LBP who presented to a private, community-based, a Duration of current episode of LBP , median 365 (172-730) group exercise program incorporating cognitive behavior (IQR) (n = 113) therapy improved substantially over the course of 1 year. b Leg pain (n = 112) 23 (19.5) During the first 6 months disability, function and pain c Highest level of education diploma or higher 97 (82) intensity improved similarly; however between 6 and (n = 109) 12 months disability and function continued to improve Employment status before episode of LBP (n = 113) while only small further improvements in pain occurred. At Full time/full duties 82 (69.5) 12 months pain intensity had reduced on average by 39 % Part time/full duties 14 (12) while disability and function had improved by 60 and Not seeking employment (retired/child care) 12 (10) 72 %, respectively. n Work status changed as result of LBP ( = 113) 10 (8.5) The predictors investigated did not explain a substantial Current employment status (n = 113) amount of the variation in outcome for either pain or dis- Full time/full duties 62 (52.5) ability. The only independent predictor of 12-month pain Full time/selected duties 6 (5) outcomes was baseline pain intensity, but this accounted Part time/full duties 19 (16) for only 10 % of the variation. The independent predictors Part time/selected duties 2 (1.7) of 12-month disability were duration of current episode, Not seeking employment (retired/child care) 24 (20) baseline disability and educational level, accounting for Private health insurance 103 (87) 15 % of the variation at 12 months. Pain catastrophizing scale (n = 112) 16 ± 10 IQR interquartile range Strengths and limitations of the study a Measured in days b Defined as back pain extending below the knee A strength of this study is the inclusion of a clearly defined c University study based cohort of consecutive patients with chronic LBP ([3 months) that enrolled in a private paying group exercise program. The high 1-year follow-up rate (95 %) and bothersomeness, disability and function improving by 39, the use of validated outcome measures are further 41, 60 and 72 % of the initial levels, respectively. Pain strengths. Nevertheless, we understand that an even longer intensity and bothersomeness improved markedly during follow-up period with more regular sampling may give a the first 6 months (35 and 37 %, respectively) with mini- more comprehensive assessment of the course of chronic mal further reductions up to 12 months (39 and 41 %, LBP. respectively). Conversely, disability and function contin- We limited the number of baseline prognostic factors ued to improve considerably throughout the 12 months. investigated to one per ten patients, which could have ruled After 3, 6 and 12 months, 19 (16 %), 30 (25.5 %) and out some important prognostic factors already investigated 29 (25 %) participants had completely recovered (pain by others studies, although inclusion of too many variables

123 142 Eur Spine J (2014) 23:113–119 117

Table 2 Primary and secondary outcomes Outcomes Baseline 3 months 3 months 6 months 6 months 12 months 12 months 12 months mean ± SD mean ± SD percentage of mean ± SD percentage of mean ± SD absolute change percentage change change scoresg ± SD of change

Bothersomenessa 5.4 ± 2.5 4.1 ± 2.2* 24 3.4 ± 2.3* 37 3.2 ± 2.4* 2.2 ± 3.2 41 Global 0.25 ± 2.4 1.5 ± 1.9* – 2.1 ± 1.7* – 2.5 ± 2* – – impression of recoveryb Painc 4.4 ± 2.1 3.7 ± 2.1* 16 2.9 ± 2* 35 2.7 ± 2.2* 1.7 ± 2.6 39 Functiond 12.3 ± 5.3 15.5 ± 6.1* 26 18 ± 6* 46 21.1 ± 6* 8.8 ± 9.4 72 Disabilitye 7.8 ± 4.2 5.2 ± 4.2* 34 4.1 ± 4* 48 3.1 ± 3.2* 4.7 ± 4.5 60 Kinesiophobiaf 36 ± 7.5 34.8 ± 7.3 4 32.8 ± 7* 9 31 ± 7* 5 ± 10.7 14 * Signiﬁcant mean difference between baseline data and 3, 6 or 12 months (p \ 0.05) a Rated on scale from 0 = not at all bothersome to 10 = extremely bothersome b Rated on scale from -5 = vastly worse, 0 = unchanged to 5 = completely recovered c Rated on scale from 0 = no pain to 10 = worst pain possible d Sum of the activities scores on a scale from 0 = unable to perform activity to 10 = able to perform activity at pre-injury level, divided by the number of activities e Rated from 0 to 24, with higher scores indicating a higher level of disability f A total score was calculated after inversion of the individual scores of items 4, 8, 12 and 16. Rated from 17 to 68, with higher scores indicating a high degree of kinesiophobia g 12 months absolute scores are based on the difference between baseline scores and 12 months follow-up

literature, it is important to realize we aimed to recruit a cohort lacking many of the previously reported adverse prognostic factors and who were receiving recommended care. As such we hypothesized that patients presenting with a high educational and socioeconomic status and drawn from a private paying, community-based, group exercise program incorporating cognitive behavioral therapy, would have a more favorable prognosis than widely reported for chronic LBP. The results of the study support our hypothesis. The prognosis found in this study may not generalize well to patients in other settings where these favorable prognostic factors are not present. Fig. 1 Percentage improvement in function, disability, pain intensity An interesting finding in our study was that disability and bothersomeness and function improved more than pain, especially between 6 and 12 months. These findings are consistent with results from a previous trial [20] where patients receiving a pro- into the model makes it less stable and less generalizable gram based on cognitive behavioral principles, similar to [18]. our cohort, improved by 49 % for disability but only 29 % for pain after 12 months. Similar findings of greater Comparison with existing literature changes in disability than pain were also reported by a systematic review of multidisciplinary rehabilitation for Our study found a more favorable prognosis for people chronic LBP [21]. Conversely a recent systematic review with chronic LBP than widely reported in the literature. of the prognosis of persistent LBP [2] found greater Van Tulder et al. [19] reported only small improvements in changes in pain than disability over 12 months. The studies pain intensity and disability after 12 months (14.2 and included in this review did not typically include exercise 14.7 %, respectively). Grotle and colleagues [5], found programs with cognitive behavioral principles and may only moderate changes in disability after 1 year (25 % explain these differences in findings. reduction) for people with chronic LBP. When comparing Our study found few important predictors of 12 month the prognosis of our study participants to previous pain or disability. Multivariate models predicted only 10

123 143 118 Eur Spine J (2014) 23:113–119

Table 3 Univariate regression analysis with pain intensity and disability at 12 months follow-up as the dependent variable Prognostic factors Pain intensity Disability NR2 Unstandardized coefﬁcient p value R2 Unstandardized coefﬁcient p value (95 % CI) (95 % CI)

Number of previous episodesa 113 0.010 -0.650 (-1.889 to 0.589) 0.589 0.011 -0.016 (-2.798 to 0.766) 0.261 Pain intensityb 113 0.101 0.342 (0.150 to 0.534) 0.001 0.039 0.307 (0.021 to 0.592) 0.035 Duration of current episodec 113 0.006 0.148 (-0.211 to 0.508) 0.416 0.047 0.601 (0.094 to 1.108) 0.021 Disabilityd 113 0.013 0.061 (-0.039 to 0.160) 0.231 0.106 0.249 (0.113 to 0.386) \0.001 Functione 107 0.001 -0.013 (-0.095 to 0.069) 0.753 0.003 -0.035 (-0.154 to 0.083) 0.556 Global impression of recoveryf 113 0.054 -0.218 (-0.390 to -0.046) 0.014 0.007 -0.116 (-0.369 to 0.138) 0.369 Kinesiophobiag 113 0.034 0.055 (0.001 to 0.110) 0.049 0.040 0.085 (0.006 to 0.164) 0.034 Educational Levelh 109 0.006 0.485 (-0.743 to 1.714) 0.435 0.028 1.536 (-0.210 to 3.281) 0.084 Leg Paini 112 0.008 0.500 (-0.551 to 1.550) 0.348 0.006 0.617 (-0.894 to 2.128) 0.420 Catastrophizingj 112 0.044 0.047 (0.006 to 0.089) 0.027 0.031 0.057 (-0.003 to 0.118) 0.063 Patients enrolled for the 2nd termk 113 0.014 0.553 (-0.320 to 1.425) 0.212 0.001 -0.253 (-1.517 to 1.010) 0.692 11 point prognosis scalel 113 0.001 -0.040 (-0.278 to 0.198) 0.739 0.017 0.234 (-0.105 to 0.574) 0.174 a An episode of LBP was classified as pain lasting for more than 24 h. Scored as 0 (from 0 to 1 episode) or 1 (from two or more episodes) b Rated on a scale from 0 = no pain to 10 = worst pain possible c Duration of the current episode was recorded in days and log transformed for the analyses d Rated from 0 to 24, with higher scores indicating a higher level of disability e Sum of the activities scores on a scale from 0 = unable to perform activity to 10 = able to perform activity at pre-injury level divided by the number of activities f Rated on scale from -5 = vastly worse, 0 = unchanged to 5 = completely recovered g A total score was calculated after inversion of the individual scores of items 4, 8, 12 and 16. Rated from 17 to 78, with higher scores indicating a high degree of kinesiophobia h Educational level was score from 0 = school and technical college (school, high school certificate and trade) to 1 = University study based (diploma, advanced diploma, bachelor degree and postgraduate degree) i Defined as back pain extended below the knee. Scored as 0 = no or 1 = yes j The Pain Catastrophizing Scale sum score was calculated from all items (range, 13–65), with a higher scores indicating a higher level of pain catastrophizing k Patients who continues classes after the first term. Scored as 0 = patients enrolled for one term only or 1 = patients enrolled for two or more terms l Rated from 0 to 10, with higher values indicating worse prognosis and 15 % of pain and disability outcomes, respectively. disability the effect size for most predictors is relatively These findings are somewhat lower than previous studies low. A possible reason predictors in our study failed to although most previous studies have failed to identify explain much of the variance may be that we deliberately strong predictors of outcome in chronic LBP. Bekkering included a cohort lacking most of the previously described et al. [22] followed a mixed group (acute/chronic LBP) of prognostic factors. In this population, predicting outcome 500 patients with non-specific LBP for 12 months. This may be more difficult or other predictors may be more study evaluated prognostic factors associated with pain and important. disability at 12 months follow-up. The final model consisted of two factors for pain (duration of the current episode and pain intensity at baseline) and three for disability Conclusions (paid job, episode duration and disability), these factors explained 10 and 28 %, respectively, of the variance. Over the course of 1 year, patients with chronic LBP who Grotle et al. [5], found five prognostic factors associated presented to a private paying, community-based, group with 12-month disability in people with chronic LBP exercise program incorporating cognitive behavioral ther- (being not employed widespread pain, chronic pain grade, apy, improved markedly. During the first 6 months, pain fear of pain and catastrophizing), representing 52.7 % of intensity, bothersomeness, disability and function improved variance. However, they note that beyond baseline similarly, but from 6 to 12 months, disability and function

123 144 Eur Spine J (2014) 23:113–119 119

Table 4 Backward regression analyses with pain intensity (n = 112) 3. Burton AK, McClune TD, Clarke RD, Main CJ (2004) Long-term and disability (n = 108) at 12 months follow up as dependent follow-up of patients with low back pain attending for manipu- variable lative care: outcomes and predictors. Man Ther 9(1):30–35 4. de Vet HCW, Heymans MW, Dunn KM, Pope DP, van der Beek 2 Predictors R Regression coefficient AJ, Macfarlane GJ, Bouter LM, Croft OR (2002) Episodes of low (95 % CI) back pain: a proposal for uniform definitions to be used in research. Spine 27:2409–2416 Pain intensity at 12 months 0.102 5. Grotle M, Foster NE, Dunn KM, Croft P (2010) Are prognostic (n = 112) indicators for poor outcome different for acute and chronic low Pain intensity at baselinea 0.344 (0.050 to 0.534) back pain consulters in primary care? Pain 151:790–797 Constant 1.244 (0.323 to 2.195) 6. Hayden JA, Dunn KM, van der Windt DA, Shaw WS (2010) What is the prognosis of back pain? Best Pract Res Clin Rheu- Disability at 12 months 0.154 matol 24(2):167–179 (n = 109) 7. Hayden JA, Chou R, Hogg-Johnson S, Bombardier C (2009) b Duration of current episode 0.457 (-0.048 to 0.961) Systematic reviews of low back pain prognosis had variable Disabilityc 0.234 (0.095 to 0.372) methods and results: guidance for future prognosis reviews. Educational leveld 1.408 (-0.260 to 3.075) J Clin Epidemiol 62:781–796 8. Costa Lda C, Maher CG, McAuley JH, Hancock MJ, Herbert RD, Constant -2.505 (-5.825 to 0.815) Refshauge KM, Henschke N (2009) Prognosis for patients with a Rated on a scale from 0 = no pain to 10 = worst pain possible chronic low back pain: inception cohort study. BMJ 339:b3829 b 9. Melzack R (2001) Pain and the Neuromatrix in the Brain. J Dent Duration of the current episode was recorded in days and log Educ 65(12):1378–1382 transformed for the analyses 10. Bouvier AL (2008) Physiocise moviment for life: Backs, brains, c Rated from 0 to 24, with higher scores indicating a higher level of breathing. Physiocise, Australia disability 11. Bouvier AL, Fleming J (2010) The feel good body: 7 steps to d Educational level was scored from 0 = school and technical col- easing aches and looking great. Harper Collins, Australia lege (school, high school certificate and trade) to 1 = University 12. Pengel LHM, Refshauge KM, Maher CG (2004) Responsiveness study based (diploma, advanced diploma, bachelor degree and post- of pain, disability, and physical impairment outcomes in patients graduate degree) with low back pain. Spine 29(8):879–883 13. Roland M, Morris R (1983) A study of the natural history of back pain. Part I: development of a reliable and sensitive measure of continued to improve while only small further changes in disability in low-back pain. Spine 8(2):141–144 bothersomeness and pain intensity occurred. 14. Scrimshaw SV, Maher CG (2001) Responsiveness of Visual Pain intensity at baseline was the only independent Analogue and McGill Pain Scale Measures. J Manipulative Physiol Ther 24(8):501–504 predictor of 12-month pain scores, accounting for 10 % of 15. Deyo RA, Battie M, Beurskens AJ, Bombardier C, Croft P, Koes the variation. There were three independent predictors of B, Malmivaara A, Roland M, Von Korff M, Waddell G (1998) 12 months disability; duration of current episode, baseline Outcome measures for low back pain research. A proposal for disability and educational level, accounting for 15 % of the standardized use. Spine 23(18):2003–2012 16. Westaway MD, Stratford PW, Binkley JM (1998) The patient- variation. Most of the variance in outcome was not specific functional scale: validation of lts use in persons with explained by any of the predictors we investigated in this neck dysfunction. J Orthop Sports Phys Ther 27(5):331–338 study. 17. Miller RP, Kori S, Todd D (1991) The Tampa Scale: a measure of kinesiophobia. Clin J Pain 7(1):51–52 Acknowledgments We thank all staff at the Willoughby and Sta- 18. Harrell FE, Lee KL (1984) Regression modelling strategies for dium Physiocise clinics for their valuable assistance and support of improved prognostic prediction. Stat Med 3:143–152 this study. 19. van Tulder MW, Koes BW, Metsemakers JFM, Bouter LM (1998) Chronic low back pain in primary care: a prospective Conflict of interest None. study on the management and course. Fam Pract 15:126–132 20. Lambeek LC, van Mechelen W, Knol DL, Loisel P, Anema JR (2010) Randomised controlled trial of integrated care to reduce disability from chronic low back pain in working and private life. BMJ 340:c1035 References 21. Guzman J, Esmail R, Karjalainen K, Malmivaara A, Irvin E, Bombardier C (2001) Multidisciplinary rehabilitation for chronic 1. Airaksinen O, Brox JI, Cedraschi C, Hildebrandt J, Klaber- low back pain: systematic review. BMJ 322:1511–1516 Moffett J, Kovacs F, Mannion AF, Reis S, Staal JB, Ursin H, 22. Bekkering GE, Hendriks HJM, van Tulder MW, Knol LD, Zanoli G (2006) European guidelines for the management of Simmonds MJ, Oostendorp RAB, Bouter LM (2005) Prognostic chronic nonspecific low back pain. Eur Spine J 15:192–300 factors for low back pain in patients referred for physiotherapy: 2. LdCM Costa, Maher CG, Hancock MJ, McAuley JH, Hebert RD, comparing outcomes and varying modeling techniques. Spine Costa LOP (2012) The prognosis of acute and persistent low-back 30:1881–1886 pain: a meta-analysis. CMAJ 184(11):E613–E624

123 145

Chapter Eight

Do magnetic resonance imaging findings identify patients with low back pain who respond better to particular interventions? A systematic review

Chapter Eight published as:

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “Do magnetic resonance imaging findings identify patients with low back pain who respond better to particular interventions? A systematic review”, we confirm that Daniel Steffens has made the following contributions:

 Data extraction, analysis and interpretation of the findings  Writing of the manuscript and critical appraisal of the content

Mark J Hancock Date: 01.01.2015

Leani SM Pereira Date: 01.01.2015

Peter M Kent Date: 01.01.2015

Jane Latimer Date: 01.01.2015

Christopher G Maher Date: 01.01.2015

147

Title page

Title: Do magnetic resonance imaging findings identify patients with low back pain who respond better to particular interventions? A systematic review.

Running head: MRI findings as effect modifiers for specific interventions.

Authors: D. Steffens1,3, M.J. Hancock2, L.S.M. Pereira3, P.M. Kent4,5, J. Latimer1, C.G. Maher1

1 Musculoskeletal division, The George Institute for Global Health, Sydney Medical School, The University of Sydney, Australia.

2 Discipline of Physiotherapy, Faculty of Human Sciences, Macquarie University, Sydney, Australia.

3 Department of Physiotherapy, Federal University of Minas Gerais, Minas Gerais, Brazil.

4 Department of Sports Science and Clinical Biomechanics, University of Southern Denmark, Odense, Denmark. 5 Research Department, The Spine Centre of Southern Denmark, Institute of Regional Health Services Research, University of Southern Denmark, Middelfart, Denmark.

Correspondence: Daniel Steffens, The George Institute for Global Health, Sydney Medical School, The University of Sydney, P.O. Box M201 Missenden Rd, Sydney, 2050, New South Wales, Australia, phone 61 2 8238-24 34, fax 61 2 9657-0301, email address [email protected]

Article category: Review article.

Funding sources: None declared.

Conflict of interest: None declared.

148

Databases:

 This systematic review identified relevant studies by electronic searches of MEDLINE,

EMBASE, The Cochrane Central Register of Controlled Trials and by examination of

the reference lists of identified papers.

What does this review add?

 The included studies investigated 38 interactions for combinations of different MRI

findings, interventions and outcomes.

 Individual trials suggested some MRI findings might be effect modifiers for specific

interventions.

 The limited number of suitable studies and the heterogeneity between them did not

permit definitive conclusions about effect modification.

Abstract

Background and Objective: Magnetic resonance imaging (MRI) can reveal a range of

degenerative findings and anatomical abnormalities; however, the clinical importance of

these remains uncertain and controversial. We aimed to investigate if the presence of MRI

findings identifies patients with LBP who respond better to particular interventions.

Databases and data treatment: MEDLINE, EMBASE and CENTRAL databases were

searched. We included RCTs investigating MRI findings as treatment effect modifiers for

patients with LBP or sciatica. We excluded studies with specific diseases as the cause of

LBP. Risk of bias was assessed using the criteria of the Cochrane Back Review Group.

Each MRI finding was examined for its individual capacity for effect modification.

149

Results: Eight published trials met the inclusion criteria. The methodological quality of trials

was inconsistent. Substantial variability in MRI findings, treatments and outcomes across

the eight trials prevented pooling of data. Patients with Modic Changes type 1 when

compared with patients with Modic Changes type 2 had greater improvements in function

when treated by Diprospan (steroid) injection, compared with saline. Patients with central

disc herniation when compared with patients without disc herniation had greater

improvements in pain when treated by surgery, compared with rehabilitation.

Conclusions: Although individual trials suggested some MRI findings might be effect

modifiers for specific interventions, none of these interactions were investigated in more

than a single trial. High quality, adequately powered trials investigating MRI findings as

effect modifiers are essential to determine the clinical importance of MRI findings in LBP

and sciatica (PROSPERO: CRD42013006571).

Systematic Review Registration Number: PROSPERO: CRD42013006571.

1. Introduction

Low back pain (LBP) is an extremely common health problem (Hoy et al., 2010), with an enormous global burden (Buchbinder et al., 2013). Limited progress has been made in the management of LBP with most treatments showing little or no effect (Keller et al., 2007; van

Tulder et al., 2006). One explanation for this lack of progress might be the current inability to identify a specific cause for LBP in most people (van Tulder et al., 2006). As a result, a single intervention is usually provided to heterogeneous groups of patients with potentially different causes of their pain. Identifying more homogenous subgroups of LBP patients has been identified as a key research priority in the field (Costa Lda et al., 2013). Most previous

150

research in this area has focussed on identifying clinical and psychosocial variables

associated with patients who respond better to different interventions (Kent and Kjaer, 2012;

Kent et al., 2010b). However, very little attention has focussed on identifying subgroups

based on biological mechanisms or anatomical structures. Some early work has investigated

subgroups based on different pain mechanisms (Smart et al., 2012; Rabey et al., 2014; Vibe

Fersum et al., 2013) due to increasing evidence for the role of central mechanisms in the development of chronic LBP (O’Sullivan 2005). Subgrouping based on possible peripheral patho-anatomical causes of LBP has received little attention and its value is unknown.

The importance of magnetic resonance imaging (MRI) findings such as disc herniation, facet joint arthropathy and modic changes (bone marrow and endplate lesions visible on MRI) in

identifying the source of an individual patient’s LBP remains unclear and controversial.

Many MRI findings are common in people without LBP, yet these findings are typically

more common in people with LBP than those without (Cheung et al., 2009; Hancock et al.,

2012; Steffens et al., 2013). Research into the importance or otherwise of MRI findings has

been frustrated by the lack of a widely accepted gold standard (Hancock et al., 2012). An

alternative approach in such cases is to investigate if the presence of MRI findings predicts

different response to interventions (Rutjes et al., 2007). If this were the case, it would provide

evidence for the importance of such findings and a logical rationale for selecting specific

interventions for individual patients.

To our knowledge, there has been no review of diverse MRI findings as effect modifiers for

LBP interventions. Therefore, the aim of this systematic review was to investigate if the

presence of MRI findings at baseline identifies patients with LBP who respond better to

particular interventions.

151

2. Methods

The review protocol was specified in advance and registered on PROSPERO: International prospective register of systematic reviews (refer to this link for full access of the protocol, http://www.crd.york.ac.uk/PROSPERO/display_record.asp?ID=CRD42013006571). The

PRISMA statement was used to guide the conduct and reporting of the study (Moher et al.,

2009).

2.1 Search strategy

A sensitive search was performed of MEDLINE, EMBASE and The Cochrane Central

December, 2013. We used a search strategy based on the recommendations of the Cochrane

Back Review Group (Furlan et al., 2009) for randomised controlled trials (RCTs) and LBP, combined with Medical Subject Headings and keywords related to ‘MRI’ and ‘effect modification/ subgroups’. After piloting the search strategy we decided to use two different searches and then combine the results.

Search 1 included terms from each of the following domains (i) RCTs, (ii) LBP and (iii)

MRI. Search 2 included terms from each of the following domains (i) RCTs, (ii) LBP and

(iii) effect modification/ subgroup. Searches 1 and 2 were merged to generate the final search strategy (Refer to Appendix S1 and S2 for the full search strategy). Reference and citation tracking of relevant articles were performed. A final list of the included studies was sent to two experts in the field who reviewed the list for possible omissions.

2.2 Study selection

To be included, studies were required to meet all of the following criteria:

152

(i) Participants: Recruited samples of populations with current LBP or sciatica, who were not diagnosed with serious disease (e.g. cancer, spinal infection, spinal fracture, inflammatory arthritis or cauda equina syndrome) as the source of LBP.

(ii) Interventions: Investigated any type of intervention for LBP, including conservative, surgical, or placebo. Included studies needed to have compared any intervention for LBP or sciatica, with any type of intervention, placebo or no treatment control.

(iii) Outcome: Reported for either pain (e.g. measured by the Visual Analogue Scale,

Numerical Rating Scale) or disability (e.g. measured by the Roland Morris Disability Scale,

Oswestry Disability Scale). In studies that included participants with a primary complaint of

LBP, self-reported LBP was considered the primary outcome while in trials of sciatica self- reported leg pain was considered the primary outcome.

(iv) Study design: Included studies needed to be an RCT which had used methods capable of identifying whether patients with a specific MRI finding had a different treatment effect than those without the MRI finding or with a different MRI finding. Studies were required to have included and reported a patient’s results separately for either (a) sample with and without a

particular MRI finding (i.e. disc herniation) or (b) people with a different type or severity of

MRI finding (i.e. mild vs. severe disc degeneration).

One reviewer screened titles and abstracts of each citation and excluded clearly irrelevant

studies. For each potentially eligible study the full text was retrieved and two reviewers

independently assessed whether the study fulfilled the inclusion criteria. In cases of

disagreement, a third reviewer was consulted and a decision made by consensus. The search

had no language restrictions.

153

2.3 Data extraction

Relevant data were independently extracted by two reviewers using a standardised form. In cases of disagreement, a joint review of the original article was performed until consensus was reached. The extraction form included the following criteria: clinical settings, sample, age, treatment groups, MRI findings and point estimates and measures of variability for outcomes. Outcome data were extracted for short-term outcomes (0 to ≤ 6 months) and long- term outcomes (>6 months). When multiple time points fell within the same category, we used the one closest to 3 months for short-term and closest to 12 months for long-term.

2.4 Risk of bias

Risk of bias was assessed using criteria recommended by the Cochrane Back Review Group

(Furlan et al., 2009). Two reviewers independently assessed the criteria of all included studies. In cases of disagreement, a third reviewer was consulted and a decision made by consensus (refer to Appendix S3 for further details on the criteria list for the methodological quality assessment). Data pooling was appropriate only if the studies were considered homogeneous with regard to population sample, MRI measure, clinical outcomes and treatment.

2.5 Analysis

It eventuated that we were unable to undertake the pre-specified meta-analysis due to the small number of included trials and the heterogeneity between them in terms of MRI findings, treatment and clinical outcomes. Therefore, each MRI finding of the lumbar spine was examined for its individual capacity for effect modification and interaction. The results are presented descriptively for LBP and sciatica populations.

154

We extracted (i) mean difference and 95% confidence intervals (95% CI) from studies that reported continuous outcomes, (ii) hazard ratios (HR) and 95% CI from studies that reported time-to-event categorical outcomes, and (iii) contingency table data to calculate Odds Ratios

(OR) for categorical outcomes. If not reported or provided, the effect modification and subgroup interaction was calculated using the method suggested by Kent et al (Kent et al.,

2010b) for continuous outcomes and the method suggested by Hancock et al (Hancock et al.,

2013) for categorical outcomes.

Four studies had key information not available from published manuscripts and additional information was requested (Arts et al., 2010; Pearson et al., 2012; Pearson et al., 2008; Peul et al., 2009). Two studies reported combined RCT and observational cohort data (Pearson et al., 2012; Pearson et al., 2008). The separated RCT data for the intention-to-treat analysis was requested. The effect modification and/or the subgroup interaction were calculated by the current review authors, for six studies (Buttermann, 2004; Cao et al., 2011; Hellum et al.,

2012; Pearson et al., 2012; Pearson et al., 2008; Tafazal et al., 2009).

In this review, the term subgroup interaction refers to how much more effective (compared with the control intervention) the intervention is in the subgroup (MRI positive) than for those not in the subgroup (MRI negative).

3. Results

3.1 Study Selection

The search identified 6239 papers. After review of titles and abstracts, we excluded 6186

(Figure 1). Based on full-text review of 53 papers, we excluded a further 45 and included eight trials in the review (Arts et al., 2010; Buttermann, 2004; Cao et al., 2011; Hellum et al.,

2012; Pearson et al., 2012; Pearson et al., 2008; Peul et al., 2009; Tafazal et al., 2009). The

155

primary reasons for the exclusion of trials retrieved in full-text are noted in Appendix S4. No

additional studies were identified after contacting two experts in the field of MRI and LBP.

3.2 Risk of bias

The risk of bias assessments for the included studies are shown in Table 1. Randomisation,

drop-out rate, co-interventions and outcome timing were the only criteria scored ‘yes’ in all

trials. Participant blinding, outcome assessor blinding and absence of selective outcome reporting were the criteria most commonly scored ‘no’.

3.3 Study Characteristics

The characteristics of the included studies are shown in Table 2. Three trials studied patients

with LBP (Buttermann, 2004; Cao et al., 2011; Hellum et al., 2012) and five studied patients

with sciatica (Arts et al., 2010; Pearson et al., 2012; Pearson et al., 2008; Peul et al., 2009;

Tafazal et al., 2009). The samples were recruited from secondary health care (Arts et al.,

2010; Buttermann, 2004; Pearson et al., 2012; Pearson et al., 2008; Tafazal et al., 2009), and

tertiary health care (Cao et al., 2011; Hellum et al., 2012; Peul et al., 2009) settings. The

number of participants varied from 120 to 472 and most studies sampled predominantly

adults in their middle age. The treatments evaluated in the trials included surgery, injections

and rehabilitation. No study had the primary aim of investigating MRI effect modifiers. LBP

duration was categorised as acute (less than 6 weeks), sub-acute (6-12 weeks) and chronic

(greater than 12 weeks) (Furlan et al., 2009).

3.4 Results of the review

Due to the heterogeneity of samples, MRI findings, clinical outcomes and treatment, it was not possible to perform meta-analysis of the results for any of the included studies. For ease of interpretation, the studies were grouped into LBP population (Buttermann, 2004; Cao et

156

al., 2011; Hellum et al., 2012) or sciatica population (Arts et al., 2010; Pearson et al., 2012;

Pearson et al., 2008; Peul et al., 2009; Tafazal et al., 2009) as the importance of MRI findings

might be quite different in these two populations. Detailed findings of all included studies are presented in Table 3.

3.4.1 Low back pain population samples

One study reported on a population with sub-acute LBP (symptoms ≥ 6 weeks) (Cao et al.,

2011) and two reported on populations with chronic LBP (symptoms ≥ 1 year) (Buttermann,

2004; Hellum et al., 2012). All three studies investigated Modic changes (Modic changes type

1 corresponding to vertebral body edema and hyper-vascularity; Modic changes type 2

reflecting fatty replacements of the red bone marrow; and Modic changes type 3 consisting of

subchondral bone sclerosis (Modic et al., 1988a; Modic et al., 1988b)) as effect modifiers

(Buttermann, 2004; Cao et al., 2011; Hellum et al., 2012), while one study investigated disc

herniation and facet joint arthritis (Hellum et al., 2012).

Cao et al (Cao et al., 2011) investigated various intradiscal injection regimens for patients

with Modic changes (n=120). Patients with Modic changes type 1, when compared with

patients with Modic changes type 2, had greater improvements in disability in the short-term

(3 months) when treated by Diprospan (steroid) injection, compared with saline (mean

difference 8.30; 95% CI, 1.01 to 15.59, on a 0 to 100 disability scale). Other subgroup

interactions for pain and disability with Modic changes were not significant.

Hellum et al (Hellum et al., 2012) investigated whether features of degenerative disc were

effect modifiers for disc prosthesis compared with multidisciplinary rehabilitation at two-year

follow up (n=154). The presence of Modic changes type 1 and/or 2 was not a significant

effect modifier for improvements in disability (percentage of patients improved ≥15 points on

a 0 to 100 scale, categorized by yes/no), OR ranging from 0.63 (95% CI, 0.15 to 2.65) to 2.96

157

(95% CI, 0.65 to 13.52). Similarly, disc herniation, facet joint arthropathy and high intensity

zone were not significant effect modifiers for improvement in disability when treated with

surgery, compared with rehabilitation (Hellum et al., 2012).

Buttermann (Buttermann, 2004) investigated whether Modic changes type 1 was an effect

modifier for spinal injection and steroid, compared with discography alone at 1-3 and 12-24

months (n=171). Presence of Modic changes type 1 was not a significant effect modifier for

injection success (coded as ‘yes’ if the overall opinion about their injection was considered

successful) at short (OR, 7.94; 95% CI, 0.40 to 156.46) or long-term follow up (OR, 2.20;

95% CI, 0.11 to 45.98).

3.4.2 Sciatica population samples

Three studies reported potential MRI effect modifiers in one population sample with sub-

acute sciatica (symptoms ≥6 weeks) (Arts et al., 2010; Peul et al., 2009; Tafazal et al., 2009)

and two with chronic sciatica (symptom ≥12 weeks) (Pearson et al., 2012; Pearson et al.,

2008). Three studies investigated disc herniation (Arts et al., 2010; Pearson et al., 2008; Peul et al., 2009), two investigated spinal stenosis (Arts et al., 2010; Pearson et al., 2012), one investigated disc height (Arts et al., 2010) and one investigated different types of MRI findings (disc prolapse vs. spinal stenosis) (Tafazal et al., 2009) as effect modifiers.

Pearson et al (Pearson et al., 2008) studied whether features of disc herniation were effect

modifiers for discectomy, compared with conservative rehabilitation at three and 12 months

follow up (n=472). Patients with central disc herniation, when compared with patients

without central disc herniation, had better response to surgery at long-term follow up (24

months), mean difference 1.60; 95% CI, 0.17 to 3.03 (0 to 6 point Likert scale). In patients

with central herniation, one-year pain outcomes were better (mean difference 1.60; 95% CI,

0.10 to 3.10; 0 to 6 point Likert scale) for those receiving surgery compared with

158

rehabilitation. In those without central herniation, surgery was no better than rehabilitation

(mean difference 0.00; 95% CI, -0.40 to 0.40; 0 to 6 point Likert scale). Other disc herniation characteristics (e.g. posterolateral and protrusion) were not associated with significant treatment interactions.

Peul et al (Peul et al., 2009) investigated if disc herniation was an effect modifier for response to early surgery compared with prolonged conservative care (n=283). Sequestrated disc herniation (Hazard ratio, 0.94; 95% CI, 0.56 to 1.57) and disc herniation enhancement

(Hazard ratio, 0.85; 95% CI, 0.47 to 1.54) did not have any significant interaction with treatment at 12 months (very much improved and much improved were coded as recovered).

Arts et al (Arts et al., 2010) investigated if disc herniation, spinal stenosis and disc height were effect modifiers for response to tubular discectomy, compared with conventional microdiscectomy, at one-year follow up (n=325). None of the MRI findings produced significant interactions with treatment for long-term recovery outcomes.

Pearson et al (Pearson et al., 2012) investigated whether features of spinal stenosis were effect modifiers for response to surgery, compared with rehabilitation, in 278 patients at three and 24 months follow up. Spinal stenosis did not produce any significant interactions with treatment for short- and long-term disability outcomes.

Tafazal et al (Tafazal et al., 2009) investigated whether features of disc herniation (disc prolapse) or lumbar spinal stenosis were effect modifiers for the efficacy of corticosteroids injection in 150 patients. Neither MRI features produced significant interactions with bupivacaine (a local anaesthetic) and steroid or bupivacaine alone at short-term follow up.

159

4. Discussion

4.1 Statement of principal findings

This review could only identify eight studies, which provided adequate data to assess if MRI

findings were treatment effect modifiers. The included studies investigated 38 interactions for

combinations of different MRI findings, interventions and outcomes. No effect modifiers

were consistently identified across more than one study. Individual trials suggested some

MRI findings might be effect modifiers for specific interventions. However, these are single

study results and caution should be taken when interpreting the findings. Some other subgroup interactions presented trends and confidence intervals that included potentially important interactions; however, these trials were underpowered due to their small sample sizes.

4.2 Strengths and weaknesses of the study

We believe this is the first systematic review of RCTs to investigate if diverse MRI findings

are effect modifiers for interventions in people with LBP and/or sciatica. The strength of this

review is the use of a pre-specified protocol and the comprehensive approach to identifying

all suitable RCTs. We also provide data for all included trials on the interaction effect as well

as the subgroup effects for those with and without the MRI finding of interest. We used a

sensitive search strategy and contacted experts in the field, reducing the risk of missing any

important trial. Despite this, we could have missed important studies because of human error

or because they were contained in databases that were not searched. A limitation of our

review is that the inconsistency of MRI findings, interventions and outcomes investigated

across the studies, inhibited our ability to perform meta-analysis. Furthermore, most trials

were not powered for subgroup interaction analysis, as it was not the primary aim of the

study. As a result, some non-significant findings may include a potentially important

160

interaction (e.g. OR, 7.94; 95% CI, 0.40 to 156.46) (Buttermann, 2004). Another limitation of our review is the possibility of publication bias as, beyond contacting two content experts, we did not attempt to identify unpublished trials that might have been found in other clinical trials registries and in conference proceedings.

4.3 Comparison with other studies

Three previous reviews have investigated effect modifiers for LBP treatments. Two of these reviews investigated effect modifiers for specific interventions (manual therapy/exercise and psychosocial intervention) (Kent and Kjaer, 2012; Kent et al., 2010b). These reviews did not include MRI findings as potential effect modifiers. One review specifically investigated

Modic changes as effect modifiers (Jensen and Leboeuf-Yde, 2011). Interestingly, all reviews found a limited number of suitable studies, which had inconsistent findings, had small sample sizes, and provided limited evidence for strong effect modifiers. These results corroborate our findings. The review investigating Modic changes as an effect modifier for different LBP treatments had several method limitations (Jensen and Leboeuf-Yde, 2011); for example, the inclusion of single subgroup designs ( i.e. studies including all people with Modic changes and no people without Modic changes) as these types of studies cannot robustly test if effect modification occurred (Kent et al., 2010a).

4.4 Meaning of the study

From 38 subgroup interactions investigated, one presented a significant effect modifier for

LBP and one for sciatica populations. These positive findings could represent spurious findings. However, the lack of statistically significant interactions may also be partly due to most studies being underpowered for this type of analysis. Consequently, it remains unclear whether MRI findings are important effect modifiers for interventions for LBP and sciatica populations. What is clear is that there are very few trials and most of these are

161

underpowered, reinforcing the need for more and larger trials in this potentially important and

evolving area.

4.5 Recommendations for future research

Studies on subgroup interaction are a research priority in LBP (Costa Lda et al., 2013) and

well conducted trials provide the possibility to answer the important and controversial

question about the importance or otherwise of MRI findings. The need for larger, high-

quality trials is evident. Due to the nature of subgroup and interaction analyses, such trials

need a larger sample size than if their only interest was the main effect of treatment. Perhaps

one way to gain statistical power would be to combine several sets of individual patient data,

to acquire an adequate number of individuals with and/or without an MRI finding of interest.

Furthermore, future trials should adopt comprehensive and standardised methods for

measuring pain (i.e. pain rating scale).

A key finding from our review was that only trials including surgery or injections had investigated MRI findings as effect modifiers for LBP interventions. We could find no evidence for the importance or otherwise of MRI findings for conservative interventions.

While we recommend the need for larger, high-quality trials, it is important to note that limited evidence exists for the use of surgery in most patients with LBP (Chou et al., 2009).

Therefore future trials investigating the importance of pathoanatomic findings in improving outcomes from surgery should be limited to patient groups with indications for surgery, such as those with sciatica or degenerative spondylolisthesis. The role of central pain processing is known to be important in many people with chronic LBP (O’Sullivan 2005) and where this predominates rather than peripheral nociceptive mechanisms it is unlikely that surgical

interventions will be effective regardless of pathoanatomic changes identified on MRI.

162

5. Conclusions

This review identified eight studies that investigated if MRI findings identify patients with

LBP and/ or sciatica who respond better to a variety of interventions. While two statistically significant interactions were found between specific MRI findings and response to treatment, the limited number of suitable studies and the heterogeneity between them did not permit definitive conclusions about effect modification. Further well-designed, adequately powered studies are required.

Acknowledgements

We thank Professor Michele Crites-Battie and Professor Jeffrey G. Jarvik for reviewing included studies and suggesting possible additional studies. We also thank authors from the included studies for providing additional information.

Author’s contributions

D.S., M.J.H. and C.G.M. contributed to the conception and design. All authors participated in data acquisition, analysis and interpretation. D.S. drafted the article and all authors revised it critically and gave the final approval of the version to be published.

163

References

Arts, M.P., Brand, R., Koes, B.W., and Peul, W.C. (2010). Effect modifiers of outcome of

surgery in patients with herniated disc related sciatica? A subgroup analysis of a

randomised clinical trial. J Neurol Neurosurg Psychiatry 81, 1265-1274.

Buchbinder, R., Blyth, F.M., March, L.M., Brooks, P., Woolf, A.D., and Hoy, D.G. (2013).

Placing the global burden of low back pain in context. Best Pract Res Clin Rheumatol 27,

575-589.

Buttermann, G.R. (2004). The effect of spinal steroid injections for degenerative disc disease.

Spine J 4, 495-505.

Cao, P., Jiang, L., Zhuang, C., Yang, Y., Zhang, Z., Chen, W., and Zheng, T. (2011).

Intradiscal injection therapy for degenerative chronic discogenic low back pain with end

plate Modic changes. Spine J 11, 100-106.

Cheung, K.M., Karppinen, J., Chan, D., Ho, D.W., Song, Y.Q., Sham, P., Cheah, K.S.,

Leong, J.C., and Luk, K.D. (2009). Prevalence and pattern of lumbar magnetic resonance

imaging changes in a population study of one thousand forty-three individuals. Spine

(Phila Pa 1976) 34, 934-940.

Chou R., Baisden J., Carragee E.J., Resnick D.K., Shaffer W.O., Loeser J.D. (2009). Surgery

for low back pain: a review of the evidence for an American Pain Society Clinical Practice

Guideline. Spine (Phila Pa 1976) 34, 1094-1109.

Costa Lda, C., Koes, B.W., Pransky, G., Borkan, J., Maher, C.G., and Smeets, R.J. (2013).

Primary care research priorities in low back pain: an update. Spine (Phila Pa 1976) 38,

148-156.

Furlan, A.D., Pennick, V., Bombardier, C., and van Tulder, M. (2009). 2009 updated method

guidelines for systematic reviews in the Cochrane Back Review Group. Spine (Phila Pa

1976) 34, 1929-1941.

164

Hancock, M., Maher, C., Macaskill, P., Latimer, J., Kos, W., and Pik, J. (2012). MRI findings

are more common in selected patients with acute low back pain than controls? Eur Spine J

21, 240-246.

Hancock, M.J., Kjaer, P., Korsholm, L., and Kent, P. (2013). Interpretation of subgroup

effects in published trials. Phys Ther 93, 852-859.

Hellum, C., Johnsen, L.G., Gjertsen, O., Berg, L., Neckelmann, G., Grundnes, O., Rossvoll,

I., Skouen, J.S., Brox, J.I., and Storheim, K. (2012). Predictors of outcome after surgery

with disc prosthesis and rehabilitation in patients with chronic low back pain and

degenerative disc: 2-year follow-up. Eur Spine J 21, 681-690.

Hoy, D., Brooks, P., Blyth, F., and Buchbinder, R. (2010). The Epidemiology of low back

pain. Best Pract Res Clin Rheumatol 24, 769-781.

Jensen, R.K., and Leboeuf-Yde, C. (2011). Is the presence of modic changes associated with

the outcomes of different treatments? A systematic critical review. BMC Musculoskelet

Disord 12, 183.

Keller, A., Hayden, J., Bombardier, C., and van Tulder, M. (2007). Effect sizes of non-

surgical treatments of non-specific low-back pain. Eur Spine J 16, 1776-1788.

Kent, P., Hancock, M., Petersen, D.H., and Mjosund, H.L. (2010a). Clinimetrics corner:

choosing appropriate study designs for particular questions about treatment subgroups. J

Man Manip Ther 18, 147-152.

Kent, P., and Kjaer, P. (2012). The efficacy of targeted interventions for modifiable

psychosocial risk factors of persistent nonspecific low back pain - a systematic review.

Man Ther 17, 385-401.

Kent, P., Mjosund, H.L., and Petersen, D.H. (2010b). Does targeting manual therapy and/or

exercise improve patient outcomes in nonspecific low back pain? A systematic review.

BMC Med 8, 22.

165

Modic M.T., Masaryk T.J., Ross J.S., Carter J.R. (1988a). Imaging of degenerative disk

disease. Radiol 168, 177–186.

Modic M.T., Steinberg P.M., Ross J.S., Masaryk T.J., Carter J.R. (1988b). Degenerative disk

disease: assessment of changes in vertebral body marrow with MR imaging. Radiol 166,

193–199.

Moher, D., Liberati, A., Tetzlaff, J., and Altman, D.G. (2009). Preferred reporting items for

systematic reviews and meta-analyses: the PRISMA statement. Ann Intern Med 151, 264-

269, W264.

O'Sullivan P. (2005). Diagnosis and classification of chronic low back pain disorders:

maladaptive movement and motor control impairments as underlying mechanism. Man

Ther 10, 242-255.

Pearson, A., Lurie, J., Tosteson, T., Zhao, W., Abdu, W., and Weinstein, J.N. (2012). Who

should have surgery for spinal stenosis? Treatment effect predictors in SPORT. Spine

(Phila Pa 1976) 37, 1791-1802.

Pearson, A.M., Blood, E.A., Frymoyer, J.W., Herkowitz, H., Abdu, W.A., Woodward, R.,

Longley, M., Emery, S.E., Lurie, J.D., Tosteson, T.D., and Weinstein, J.N. (2008). SPORT

lumbar intervertebral disk herniation and back pain: does treatment, location, or

morphology matter? Spine (Phila Pa 1976) 33, 428-435.

Peul, W.C., Arts, M.P., Brand, R., and Koes, B.W. (2009). Timing of surgery for sciatica:

subgroup analysis alongside a randomized trial. Eur Spine J 18, 538-545.

Rabey M., Beales D., Slater H., O'Sullivan P. (2014). Multidimensional pain profiles in four

cases of chronic non-specific axial low back pain: An examination of the limitations of

contemporary classification systems. Man Ther 14, S1356-689x.

166

Rutjes, A.W., Reitsma, J.B., Coomarasamy, A., Khan, K.S., and Bossuyt, P.M. (2007).

Evaluation of diagnostic tests when there is no gold standard. A review of methods. Health

Technol Assess 11, iii, ix-51.

Smart K.M., Blake C., Staines A., Thacker M., Doody C. (2012). Mechanisms-based

classifications of musculoskeletal pain: part 1 of 3: symptoms and signs of central

sensitisation in patients with low back (± leg) pain. Man Ther 17, 336-344.

Steffens, D., Hancock, M.J., Maher, C.G., Williams, C., Jensen, T.S., and Latimer, J. (2013).

Does magnetic resonance imaging predict future low back pain? A systematic review. Eur

J Pain.

Tafazal, S., Ng, L., Chaudhary, N., and Sell, P. (2009). Corticosteroids in peri-radicular

infiltration for radicular pain: a randomised double blind controlled trial. One year results

and subgroup analysis. Eur Spine J 18, 1220-1225.

van Tulder, M.W., Koes, B., and Malmivaara, A. (2006). Outcome of non-invasive treatment

modalities on back pain: an evidence-based review. Eur Spine J 15 Suppl 1, S64-81.

Vibe Fersum K., O'Sullivan P., Skouen J.S., Smith A., Kvåle A. (2013). Efficacy of

classification-based cognitive functional therapy in patients with non-specific chronic low

back pain: a randomized controlled trial. Eur J Pain 17, 916-928.

167

Figure and Tables’ legends

Figure 1. Flow chart diagram of review process.

Table 1. Assessment of risk of bias of the included studies.

Table 2. Individual characteristics of the included studies.

Table 3. Subgroup treatment effect and interaction for low back pain and sciatica population.

Appendix S1. Search strategy 1 (MEDLINE, EMBASE AND CENTRAL).

Appendix S2. Search strategy 2 (MEDLINE, EMBASE AND CENTRAL).

Appendix S3. Criteria list for the risk of bias assessment.

Appendix S4. List of excluded full-text articles and the primary reason for exclusion.

168

Records identified through Additional records identified database searching through other sources MEDLINE = 2118 (n=1) EMBASE = 5228

Identification Cochrane Central = 438

Records after duplicates Records excluded

removed (n=6239) (n=6186)

Screening

Full-text articles Full-text articles assessed for excluded eligibility (n=53) - Not an RCT (n=10) Eligibility - No evaluation of MRI findings (n=9)

- Not possible to RCTs included in review elucidate association

(n=8) between MRI and

Included outcome (n=26)

Figure 1. Flow diagram of review process

Table 1. Risk of bias of the included studiesa Method criteriab Study Randomisation Concealed Participan Clinicians Outcome Acceptable Analysed Free of Baseline Co-interventions Compliance Outcome allocation t blinding blinding assessor drop-out according selective similarit timing blinding rate to outcomes y treatment allocation Arts et al. 2010 Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes NA Yes Buttermann 2004 Yes Yes No No No Yes Yes No Yes Yes NA Yes Cao et al. 2011 Yes ? Yes Yes Yes Yes Yes No Yes Yes NA Yes Hellum et al. 2012 Yes Yes No Yes No Yes Yes No Yes Yes Yes Yes Pearson et al. 2012 Yes Yes No No No Yes yes No No Yes Yes Yes Pearson et al. 2008 Yes Yes No No No Yes yes No Yes Yes Yes Yes Peul et al. 2009 Yes Yes No No No Yes ? Yes Yes Yes Yes Yes Tafazal et al. 2009 Yes Yes Yes Yes Yes Yes Yes No Yes Yes NA Yes a Each criterion was scored as yes, unclear (?), not applicable (NA) or no, where yes indicates the criterion has been met. b Quality of included studies based on the Cochrane Back Review Group method (Furlan et al. 2009).

169 Table 2. Individual study characteristics Follow-up, Study Clinical setting Sample Age, mean (SD) Outcomes (threshold) Treatment groups duration (%)a Secondary care 325 patients with - Recovery („complete recovery‟ and 1) Surgery: conventional micro- Arts et al. 12 months (neurosurgical sub-acute sciatica 41.45 (10.75) „almost complete recovery were coded discectomy; 2010 (100%) outpatient clinic). (>6 to 8 weeks). as recovered). 2) Surgery: Tubular discectomy. - Success (was coded „yes‟ or „no‟ 171 patients with based on overall opinion as to whether 1-3 (100%) and Buttermann Secondary care (spine 1) Injection: discography and steroid; chronic LBP (>1 42.83 (8.66) patients thought their injection was 12-24 months et al. 2004 institute). 2) Injection: discography only. year). successful in the treatment of their (100%) symptoms). - Pain (VAS, ranges from 0 to 10, with 1) Injection: 3mL Diprospan; 120 patients with 0 corresponding to no pain); Cao et al. Tertiary care 2) Injection: 1mL Diprospan and 2mL sub-acute LBP (≥6 42.30 (8.72) - Disability (ODI ranges from 0% to 3 months (100%) 2011 (hospital). Songmeile; weeks). 100%, with 0% corresponding to no 3) Injection: 3mL normal saline. disability). 1) Surgery: replacement of the - Disability (ODI, ranges from 0% to degenerative lumbar disc with an artificial Tertiary care (five 154 patients with 100%, with 0% corresponding to no Hellum et al. lumbar disc; 24 months Norwegian university chronic LBP (≥1 41.20 (7.00) disability. Percentage of patients 2012 2) Rehabilitation: multidisciplinary (100%) hospitals). year). improved ≥15 ODI points categorized treatment consisted of a cognitive by yes/no). approach and supervised physical activity. 1) Surgery: standard open decompressive Secondary care (11 laminectomy; 278 patients with - Disability (ODI, ranges from 0% to 3 (88.49%) and Pearson et multidisciplinary spine 2) Rehabilitation: usual care – at least chronic sciatica ? (?) 100%, with 0% corresponding to no 12 months al. 2012 physical therapy, education and practices). (≥12 weeks). disability). (87.41%) counselling with home exercises, and non- steroidal anti-inflammatory drugs. 1) Surgery: standard open discectomy with Secondary care (11 examination and decompression of nerve 472 patients with - Pain (Pain bothersomeness, ranges root; 3 (86.02%) and Pearson et multidisciplinary spine chronic sciatica ? (?) from 0 to 6, with 0 corresponding to 2) Rehabilitation: usual care – at least 12 months al. 2008 practices). (≥12 weeks). not bothersome). physical therapy, education and (87.5%) counselling with home exercises, and non- steroidal anti-inflammatory drugs. Tertiary care (9 283 patients with - Recovery („very much improved‟ and 1) Surgery: discectomy; Peul et al. 12 months sub-acute sciatica 42.55 (9.30) „much improved‟ were coded as 2) Rehabilitation: Education, pain 2009 hospitals). (100%) (>6 to 8 weeks). recovered). medication, physiotherapy if necessary.

170 Secondary care - Pain (VAS, ranges from 0 to 100, 1) Injection: 2mL of 0.25% bupivacaine 150 patients with with 0 corresponding to no pain); and 40mg of methylprednisolone Tafazal et al. (specialist spine 3 months sub-acute sciatica 51.90 (?) - Disability (ODI, ranges from 0% to (Depomedrone); 2009 (82.66%) clinic). (≥6 weeks). 100%, with 0% corresponding to no 2) Injection: 2mL of 0.25% bupivacaine disability). alone. ?= data not available; VAS=Visual analogue scale; ODI=Oswestry disability index; aPercentage based on the sample available for the subgroup interaction.

171 Table 3. Subgroup treatment effect and interaction for low back pain and sciatica population MRI threshold Treatment Subgroup Clinical Interaction, mean Treatment effect Treatment effect outcome MRI feature Study difference (95% Positive (+) Negative (-) Treatment 1 Treatment 2 for MRI + for MRI - (time), CI), unless threshold otherwise indicated Low back pain populations Pain (ST), 0 to Modic type Type 1 Type 2 Cao et al. 2011 Diprospan Saline 5.20 (4.44 to 5.96) 5.20 (4.60 to 5.80) 0.00 (-0.98 to 0.98) 10 Diprospan + Pain (ST), 0 to Modic type Type 1 Type 2 Cao et al. 2011 Saline 5.00 (4.29 to 5.71) 5.20 (4.60 to 5.80) -0.20 (-0.74 to 1.14) songmeile 10 Diprospan+ Pain (ST), 0 to Modic type Type 1 Type 2 Cao et al. 2011 Diprospan 0.20 (-0.40 to 0.80) 0.00 (-0.54 to 0.54) 0.20 (-0.62 to 1.02) songmeile 10 28.90 (22.52 to 20.60 (15.69 to Disability Modic type Type 1 Type 2 Cao et al. 2011 Diprospan Saline 8.30 (1.01 to 15.59)a 35.28) 25.51) (ST), 0 to 100 Diprospan + 28.40 (21.95 to 20.20 (15.17 to Disability Modic type Type 1 Type 2 Cao et al. 2011 Saline 8.20 (-0.11 to 16.51) songmeile 34.85) 25.23) (ST), 0 to 100 Diprospan + Disability Modic type Type 1 Type 2 Cao et al. 2011 Diprospan 0.50 (-1.21 to 2.21) 0.40 (-1.35 to 2.15) 0.10 (-2.39 to 2.59) songmeile (ST), 0 to 100 Disability Hellum et al. Modic type 1 Present Absent Surgery Rehabilitation 6.07 (1.66 to 22.12)b 2.05 (0.92 to 4.55)b (LT), 0 to 100, 2.96 (0.65 to 13.52)b 2012 ≥15points Disability Modic type 1 Hellum et al. Present Absent Surgery Rehabilitation 3.21 (0.66 to 15.59)b 2.94 (1.40 to 6.16)b (LT), ODI 0 to 1.10 (0.19 to 6.27)b and 2 2012 100, ≥15points Disability Hellum et al. Modic type 2 Present Absent Surgery Rehabilitation 2.16 (0.67 to 6.93)b 3.43 (1.48 to 7.93)b (LT), 0 to 100, 0.63 (0.15 to 2.65)b 2012 ≥15points Buttermann Discography + 76.85 (9.47 to Success (ST), Modic type 1 Present Absent Discography 9.68 (1.15 to 80.91)b 7.94 (0.40 to 156.46)b 2004 steroid 623.52)b yes Buttermann Discography + 12.33 (1.49 to Success (LT), Modic type 1 Present Absent Discography 5.61 (0.63 to 50.02)b 2.20 (0.11 to 45.98)b 2004 steroid 101.86)b yes Disability Disc Height No height Hellum et al. Surgery Rehabilitation 2.61 (1.17 to 5.82)b 3.64 (1.04 to 12.78)b (LT), 0 to 100, 0.72 (0.16 to 3.18)b herniation reduction reduction 2012 ≥15points Disability Disc Signal No signal Hellum et al. 12.00 (1.05 to Surgery Rehabilitation 2.51 (1.23 to 5.13)b (LT), 0 to 100, 0.21 (0.02 to 2.64)b herniation intensity intensity 2012 136.79)b ≥15points

172 Disability Facet joint Hellum et al. ≥moderate 1/3 of spinal ≤1/3 of spinal Tubular Conventional (LT), ≥ almost Arts et al. 2010 0.93 (0.70 to 1.24)c 1.00 (0.66 to 1.49)c 0.94 (0.57 to 1.53)c herniation canal canal discectomy microdiscectomy complete recovery Recovery Disc Mediolateral Tubular Conventional (LT), ≥ almost Median Arts et al. 2010 0.91 (0.67 to 1.24)c 0.98 (0.68 to 1.40)c 1.07 (0.67 to 1.72)c herniation and lateral discectomy microdiscectomy complete recovery Spinal Central No central Pearson et al. Surgery Rehabilitation 0.60 (-5.40 to 6.60) -11.00 (-25.70 to Disability 11.60 (-4.79 to 27.99)

173 stenosis 20012 3.70) (ST), 0 to 100 Spinal Pearson et al. -2.40 (-16.90 to Disability Central No central Surgery Rehabilitation 2.30 (-3.40 to 7.90) 4.70 (-10.55 to 19.95) stenosis 20012 12.10) (LT), 0 to 100 Spinal Pearson et al. 2.80 (-11.50 to Disability -4.60 (-20.33 to Lateral recess No lateral recess Surgery Rehabilitation -1.80 (-7.80 to 4.20) stenosis 20012 17.10) (ST), 0 to 100 11.13) Spinal Pearson et al. -3.80 (-17.60 to Disability Lateral recess No lateral recess Surgery Rehabilitation 2.50 (-3.20 to 8.20) 6.30 (-8.13 to 20.73) stenosis 20012 9.90) (LT), 0 to 100 Spinal Neuroforamin No Pearson et al. Disability Surgery Rehabilitation -3.20 (-12.80 to 6.30) -0.40 (-6.90 to 6.10) -2.80 (-14.33 to 8.73) stenosis al neuroforaminal 20012 (ST), 0 to 100 Spinal Neuroforamin No Pearson et al. Disability Surgery Rehabilitation 0.50 (-8.80 to 9.90) 2.00 (-4.30 to 8.20) -1.50 (-12.85 to 9.85) stenosis al neuroforaminal 20012 (LT), 0 to 100 Spinal Pearson et al. -5.60 (-13.40 to Disability Severe Mild/ moderate Surgery Rehabilitation 3.40 (-4.10 to 10.90) 9.00 (-1.87 to 19.87) stenosis 20012 2.20) (ST), 0 to 100 Spinal Pearson et al. Disability Severe Mild/ moderate Surgery Rehabilitation 3.40 (-3.80 to 10.70) -0.30 (-7.90 to 7.30) 3.70 (-6.85 to 14.25) stenosis 20012 (LT), 0 to 100 Recovery Spinal Tubular Conventional (LT), ≥ almost Lateral recess No lateral recess Arts et al. 2010 0.63 (0.34 to 1.15)2 1.03 (0.80 to 1.32)2 1.64 (0.85 to 3.15)2 stenosis discectomy microdiscectomy complete recovery Recovery Tubular Conventional (LT), ≥ almost Disc height ≥7mm <7mm Arts et al. 2010 0.92 (0.71 to 1.18)2 1.24 (0.70 to 2.20)2 1.35 (0.73 to 2.52)2 discectomy microdiscectomy complete recovery Herniation/ Tafazal et al. Bupivacaine + 3.10 (-11.23 to -1.30 (-15.21 to Pain (ST), 0 to Disc prolapse Spinal stenosis Bupivacaine 4.40 (-18.13 to 26.93) stenosis 2009 steroid 17.43) 17.81) 100 Herniation/ Tafazal et al. Bupivacaine + -5.00 (-3.73 to Disability Disc prolapse Spinal stenosis Bupivacaine -0.20 (-9.34 to 9.47) 4.80 (-9.06 to 18.66) stenosis 2009 steroid 13.73) (ST), 0 to 100 ST=short-term (0 to ≤6months); LT=long-term (>6months). Mean difference and 95% CI, positive values favors treatment effect for MRI positive (+). a Statistically significant. b Values are represented as odds ratios and 95% confidence intervals. An odd ratio greater than 1 favors treatment effect for MRI positive (+). c Values are represented as hazard ratio and 95% confidence intervals. A hazard ratio greater than 1 favors treatment effect for MRI positive (+).

174 175

Appendix S1. Search strategy 1 MEDLINE via Ovid and Cochrane Central of Controlled trials via The Cochrane Library 1. (randomized controlled trial or controlled clinical trial or comparative study or clinical trial or clinical trials or randomized or placebo$ or random allocation or random$ or double-blind method or single-blind method).mp. [mp=title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier] 2. animal/ not human/ 3. 1 not 2 4. (low back pain or back pain or back strain or simple back pain or non-specific back pain or low back syndrome or low back dysfunction or lumbar pain or backache or lumbago or sciatica or radiculopathy).mp. [mp=title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier] 5. 3 and 4 6. (magnetic resonance imaging or mri or magnetic resonance or nmr or nuclear magnetic resonance or disc degeneration or desiccation or loss of disc height or bulge or protrusion or extrusion or nerve root compromise or annular tear or endplate changes or stenosis or facet degeneration or high intensity zone or modic changes or degenerative disc disease or spondylolisthesis).mp. [mp=title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier] 7. 5 and 6 EMBASE (www.embase.com) 1. ‘randomized controlled trial’/exp OR ‘randomized controlled trial’ OR ‘controlled study’/exp OR ‘controlled study’ OR ‘double blind procedure’/exp OR ‘double blind procedure’ OR ‘placebo’/exp OR ‘placebo’ OR ‘random allocation’/exp OR ‘random allocation’ OR ‘clinical trial’/exp OR ‘clinical trial’ OR ‘clinical trials’/exp OR ‘clinical trials’ OR ‘double blind’ OR ‘single blind’ 2. ‘animal’/exp OR ‘animal’ OR ‘not human’ 3. #1 NOT #2 4. ‘low back pain’/exp OR ‘low back pain’ OR ‘back pain’/exp OR ‘back pain’ OR ‘lumbar pain’/exp OR ‘lumbar pain’ OR ‘backache’/exp OR ‘backache’ OR ‘lumbago’/exp OR ‘lumbago’ OR ‘radiculopathy’/exp OR ‘radiculopathy’ Or ‘sciatic$’ 5. #3 AND #4 6. 'magnetic resonance imaging'/exp OR 'magnetic resonance imaging' OR 'mri'/exp OR 'mri' OR 'nuclear magnetic resonance'/exp OR 'nuclear magnetic resonance' OR 'nmr'/exp OR 'nmr' OR 'disc degeneration'/exp OR 'disc degeneration' OR 'desiccation'/exp OR 'desiccation' OR 'loss of disc height' OR 'bulge' OR 'protrusion' OR 'extrusion' OR 'nerve root compression'/exp OR 'nerve root compression' OR 'annular tear' OR 'endplate changes' OR 'stenosis'/exp OR 'stenosis' OR 'facet degeneration' OR 'high intensity zone' OR 'modic changes' OR 'degenerative disc disease' OR 'spondylolisthesis'/exp OR 'spondylolisthesis' 7. #5 AND #6

176

Appendix S2. Search strategy 2 MEDLINE via Ovid and Cochrane Central of Controlled trials via The Cochrane Library 1. (randomized controlled trial or controlled clinical trial or comparative study or clinical trial or clinical trials or randomized or placebo$ or random allocation or random$ or double-blind method or single-blind method).mp. [mp=title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier] 2. animal/ not human/ 3. 1 not 2 4. (low back pain or back pain or back strain or simple back pain or non-specific back pain or low back syndrome or low back dysfunction or lumbar pain or backache or lumbago or sciatica or radiculopathy).mp. [mp=title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier] 5. 3 and 4 6. (target intervent$ or targeted treatment$ or subgroup$ or treatment effect or effect mod$ or effect med$ or subgroup anal$).mp. [mp=title, abstract, original title, name of substance word, subject heading word, keyword heading word, protocol supplementary concept, rare disease supplementary concept, unique identifier] 7. 5 and 6 EMBASE (www.embase.com) 1. ‘randomized controlled trial’/exp OR ‘randomized controlled trial’ OR ‘controlled study’/exp OR ‘controlled study’ OR ‘double blind procedure’/exp OR ‘double blind procedure’ OR ‘placebo’/exp OR ‘placebo’ OR ‘random allocation’/exp OR ‘random allocation’ OR ‘clinical trial’/exp OR ‘clinical trial’ OR ‘clinical trials’/exp OR ‘clinical trials’ OR ‘double blind’ OR ‘single blind’ 2. ‘animal’/exp OR ‘animal’ OR ‘not human’ 3. #1 NOT #2 4. ‘low back pain’/exp OR ‘low back pain’ OR ‘back pain’/exp OR ‘back pain’ OR ‘lumbar pain’/exp OR ‘lumbar pain’ OR ‘backache’/exp OR ‘backache’ OR ‘lumbago’/exp OR ‘lumbago’ OR ‘radiculopathy’/exp OR ‘radiculopathy’ Or ‘sciatic$’ 5. #3 AND #4 6. 'target intervent$' OR 'targeted treatment$' OR 'subgroup$' OR 'treatment effect' OR 'effect mod$' OR 'effect med$' OR 'subgroup anal$' 7. #5 AND #6

177

Appendix S3. Criteria list for the risk of bias assessment 1) Randomization: A random (unpredictable) assignment sequence. Examples of adequate methods are coin toss (for studies with 2 groups), rolling a dice (for studies with 2 or more groups), drawing of balls of different colors, drawing of ballots with the study group labels from a dark bag, computer-generated random sequence, pre-ordered sealed envelopes, sequentially-ordered vials, telephone call to a central office, and pre-ordered list of treatment assignments Examples of inadequate methods are: alternation, birth date, social insurance/ security number, date in which they are invited to participate in the study, and hospital registration number. 2) Concealed allocation: Assignment generated by an independent person not responsible for determining the eligibility of the patients. This person has no information about the persons included in the trial and has no influence on the assignment sequence or on the decision about eligibility of the patient. 3) Participant blinding: This item should be scored “yes” if the index and control groups are indistinguishable for the patients or if the success of blinding was tested among the patients and it was successful. 4) Clinicians blinding: This item should be scored “yes” if the index and control groups are indistinguishable for the care providers or if the success of blinding was tested among the care providers and it was successful. 5) Outcome assessor blinding: Adequacy of blinding should be assessed for the primary outcomes. This item should be scored “yes” if the success of blinding was tested among the outcome assessors and it was successful or: –for patient-reported outcomes in which the patient is the outcome assessor (e.g., pain, disability): the blinding procedure is adequate for outcome assessors if participant blinding is scored “yes”; –for outcome criteria assessed during scheduled visit and that supposes a contact between participants and outcome assessors (e.g., clinical examination): the blinding procedure is adequate if patients are blinded, and the treatment or adverse effects of the treatment cannot be noticed during clinical examination; –for outcome criteria that do not suppose a contact with participants (e.g., radiography, magnetic resonance imaging): the blinding procedure is adequate if the treatment or adverse effects of the treatment cannot be noticed when assessing the main outcome; –for outcome criteria that are clinical or therapeutic events that will be determined by the interaction between patients and care providers (e.g., co-interventions, hospitalization length, treatment failure), in which the care provider is the outcome assessor: the blinding procedure is adequate for outcome assessors if item “4” (caregivers) is scored “yes”; –for outcome criteria that are assessed from data of the medical forms: the blinding procedure is adequate if the treatment or adverse effects of the treatment cannot be noticed on the extracted data. 6) Acceptable drop-out rate: The number of participants who were included in the study but did not complete the observation period or were not included in the analysis must be described and reasons given. If the percentage of withdrawals and drop outs does not exceed 20% for short-term follow up and 30% for long-term follow up and does not lead to substantial bias a “yes” is scored. (N.B. these percentages are arbitrary, not supported by literature). 7) Analysed according to treatment allocation: All randomized patients are reported/analyzed in the group they were allocated to by randomization for the most important moments of effect measurement (minus missing values) irrespective of non-compliance and co-interventions. 8) Free of selective outcomes: In order to receive a “yes”, the review author determines if all the results from all pre-specified outcomes have been adequately reported in the published report of the trial. This information is either obtained by comparing the protocol and the report, or in the absence of the protocol, assessing that the published report includes enough information to make this judgment. 9) Baseline similarity: In order to receive a “yes”, groups have to be similar at baseline regarding demographic factors, duration and severity of complaints, percentage of patients with neurological

178

symptoms, and value of main outcome measure(s). 10) Co-interventions: This item should be scored “yes” if there were no co-interventions or they were similar between the index and control groups. 11) Compliance: The reviewer determines if the compliance with the interventions is acceptable, based on the reported intensity, duration, number and frequency of sessions for both the index intervention and control intervention(s). For example, physiotherapy treatment is usually administered over several sessions; therefore it is necessary to assess how many sessions each patient attended. For single session interventions (e.g., surgery), this item is irrelevant. 12) Outcome timing: Timing of outcome assessment should be identical for all intervention groups and for all important outcome assessments.

Appendix S4. List of excluded full-text articles and the primary reason for exclusion. Study Title First reason for excluding Ackerman, S. J., et al. Persistent low back pain in patients suspected of having herniated nucleus pulposus: Not an RCT 1997 Radiologic predictors of functional outcome - Implications for treatment selection Comparison of clinical outcomes and natural morphologic changes between Ahn, S. H., et al. 2002 Not an RCT sequestered and large central extruded disc herniations Antibiotic treatment in patients with chronic low back pain and vertebral bone Not possible to elucidate Albert, H. B., et al. 2013 edema (Modic type 1 changes): a double-blind randomized clinical controlled trial association between MRI and of efficacy outcome Tubular diskectomy vs conventional microdiskectomy for the treatment of lumbar Not possible to elucidate Arts, M. P., et al. 2011 disk herniation: 2-Year results of a double-blind randomized controlled trial association between MRI and outcome Effectiveness of percutaneous laser disc decompression versus conventional open Brouwer, P. A., et al. discectomy in the treatment of lumbar disc herniation; design of a prospective Not an RCT 2009 randomized controlled trial Browder, D. A., et al. Effectiveness of an extension-oriented treatment approach in a subgroup of subjects No evaluation of MRI 2007 with low back pain: a randomized clinical trial findings A double-blind, randomized, prospective study of epidural steroid injection vs. the Not possible to elucidate Brown, L. L. 2012 mild procedure in patients with symptomatic lumbar spinal stenosis association between MRI and outcome Four-year follow up of surgical versus non-surgical therapy for chronic low back Not possible to elucidate Brox, J. I., et al. 2010 pain association between MRI and outcome Randomized clinical trial of lumbar instrumented fusion and cognitive intervention Not possible to elucidate Brox, J. I., et al. 2003 and exercises in patients with chronic low back pain and disc degeneration association between MRI and outcome A clinical prediction rule to identify patients with low back pain most likely to No evaluation of MRI Childs, J. D., et al. 2004 benefit from spinal manipulation: a validation study findings Comparative prospective randomized study comparing conservative treatment and Not possible to elucidate Erginousakis, D., et al. percutaneous disk decompression for treatment of intervertebral disk herniation association between MRI and 2011 outcome Filiz, M., et al. 2005 The effectiveness of exercise programmes after lumbar disc surgery: a randomized Not possible to elucidate

179 controlled study association between MRI and outcome Freburger, J. K., et al. Effectiveness of physical therapy for the management of chronic spine disorders: A Not an RCT 2006 propensity score approach Is there a subgroup of patients with low back pain likely to benefit from mechanical No evaluation of MRI Fritz, J. M., et al. 2007 traction? Results of a randomized clinical trial and subgrouping analysis findings Lumbar fusion versus nonsurgical treatment for chronic low back pain. A Not possible to elucidate Fritzell, P., et al. 2001 multicenter randomized controlled trial from the Swedish Lumbar Spine Study association between MRI and Group outcome Chronic low back pain and fusion: A comparison of three surgical techniques: A Not possible to elucidate Fritzell, P., et al. 2002 prospective multicenter randomized study from the Swedish Lumbar Spine Study association between MRI and Group outcome No difference in long-term trunk muscle strength, cross-sectional area, and density Not possible to elucidate Froholdt, A., et al. 2011 in patients with chronic low back pain 7 to 11 years after lumbar fusion versus association between MRI and cognitive intervention and exercises outcome Efficacy of exercise and ultrasound in patients with lumbar spinal stenosis: a Not possible to elucidate Goren, A., et al. 2010 prospective randomized controlled trial association between MRI and outcome A randomized clinical trial and subgroup analysis to compare flexion-distraction Not possible to elucidate Gudavalli, M. R., et al. with active exercise for chronic low back pain association between MRI and 2006 outcome Surgery with disc prosthesis versus rehabilitation in patients with low back pain and Not possible to elucidate Hellum, C., et al. 2011 degenerative disc: two year follow-up of randomised study association between MRI and outcome Long-term effectiveness of bone-setting, light exercise therapy, and physiotherapy Not possible to elucidate Hemmila, H. M., et al. for prolonged back pain: a randomized controlled trial association between MRI and 2002 outcome The efficacy of epidural depo-methylprednisolone and triamcinolone acetate in Not possible to elucidate Huda, N., et al. 2010 relieving the symptoms of lumbar canal stenosis: A comparative study association between MRI and outcome Lumbar spinal stenosis: assessment of long-term outcome 12 years after operative Hurri, H., et al. 1998 Not an RCT and conservative treatment

180 Rest versus exercise as treatment for patients with low back pain and Modic Not possible to elucidate Jensen, R. K., et al. 2012 changes. A randomized controlled clinical trial association between MRI and outcome Correlation of size and type of Modic types 1 and 2 lesions with clinical symptoms: Kaapa, E., et al. 2012 a descriptive study in a subgroup of patients with chronic low back pain on the basis Not an RCT of a university hospital patient sample Facet joints infiltration: A viable alternative treatment to physiotherapy in patients Kawu, A. A., et al. 2011 Not an RCT with low back pain due to facet joint arthropathy Multicenter randomized controlled trial comparing particulate versus nonparticulate Not possible to elucidate Kennedy, D. J., et al. corticosteroids via lumbar transforaminal epidural injection for acute unilateral, association between MRI and 2013 unilevel radicular pain due to herniated nucleus pulposus outcome Effectiveness of physical therapy and epidural steroid injections in lumbar spinal Not possible to elucidate Koc, Z., et al. 2009 stenosis association between MRI and outcome A randomized clinical trial of manual therapy and physiotherapy for persistent back No evaluation of MRI Koes, B. W., et al. 1993 and neck complaints: Subgroup analysis and relationship between outcome measures findings Does it matter which exercise? A randomized control trial of exercise for low back No evaluation of MRI Long, A., et al. 2004 pain findings Surgical or nonoperative treatment for lumbar spinal stenosis? A randomized Not possible to elucidate Malmivaara, A., et al. controlled trial association between MRI and 2007 outcome Intravenous lidocaine, amantadine, and placebo in the treatment of sciatica: A Not possible to elucidate Medrik-Goldberg, T., et double-blind, randomized, controlled study association between MRI and al. 1999 outcome Effectiveness of microdiscectomy for lumbar disc herniation: a randomized Not possible to elucidate Osterman, H., et al. 2006 controlled trial with 2 years of follow-up association between MRI and outcome Peng, B., et al. 2009 Diagnosis and surgical treatment of back pain originating from endplate Not an RCT Does opioid pain medication use affect the outcome of patients with lumbar disc No evaluation of MRI Radcliff, K., et al. 2011 herniation? A subgroup analysis of the SPORT study findings Rajasekaran, S., et al. Lumbar spinous process splitting decompression provides equivalent outcomes to Not possible to elucidate 2013 conventional midline decompression in degenerative lumbar canal stenosis: A association between MRI and

181 prospective, randomised controlled study of 51 patients outcome Chiropractic manipulation in the treatment of acute back pain and sciatica with disc Not possible to elucidate Santilli, V., et al. 2006 protrusion: a randomized double-blind clinical trial of active and simulated spinal association between MRI and manipulations outcome Sherman, K. J., et al. Characteristics of patients with chronic back pain who benefit from acupuncture No evaluation of MRI 2009 findings Long-term results of surgery for lumbar spinal stenosis: a randomised controlled Not possible to elucidate Slatis, P., et al. 2011 trial association between MRI and outcome What works best for whom? An exploratory, subgroup analysis in a randomized, Not possible to elucidate Steenstra, I. A., et al. controlled trial on the effectiveness of a workplace intervention in low back pain association between MRI and 2009 patients on return to work outcome Styczynski, T., et al. The effect of the grade of degenerative changes in the spine on the outcomes of Not an RCT 2007 surgery for lumbar discopathy with a radicular syndrome Underwood, M. R., et al. Do baseline characteristics predict response to treatment for low back pain? No evaluation of MRI 2007 Secondary analysis of the UK BEAM dataset findings Differences in outcome of a multidisciplinary treatment between subgroups of Vollenbroek-Hutten, M. No evaluation of MRI chronic low back pain patients defined using two multiaxial assessment instruments: M., et al. 2004 findings the multidimensional pain inventory and lumbar dynamometry No effect of 6-month intake of glucosamine sulfate on Modic changes or high Not possible to elucidate Wilkens, P., et al. 2012 intensity zones in the lumbar spine: sub-group analysis of a randomized controlled association between MRI and trial outcome Zhuo, X., et al. 2010 Effectiveness comparison of two surgical procedures on lumbar disc protrusion Not an RCT

182 183

Chapter Nine

Influence of clinician characteristics and operational factors on recruitment of participants with low back pain: an observational study

Chapter Nine published as:

Statement from co-authors confirming authorship contribution of the PhD candidate

As co-authors of the paper “Influence of clinician characteristics and operational factors on recruitment of participants with low back pain: an observational study”, we confirm that Daniel Steffens has made the following contributions:

 Conception and design of the research  Data collection  Analysis and interpretation of the findings  Writing of the manuscript and critical appraisal of the content

Christopher G Maher Date: 01.01.2015

Manuela L Ferreira Date: 01.01.2015

Mark J Hancock Date: 01.01.2015

Leani SM Pereira Date: 01.01.2015

Christopher M Williams Date: 01.01.2015

Jane Latimer Date: 01.01.2015

185

INFLUENCE OF CLINICIAN CHARACTERISTICS AND OPERATIONAL FACTORS ON RECRUITMENT OF PARTICIPANTS WITH LOW BACK PAIN:AN OBSERVATIONAL STUDY

Daniel Steffens, BPT (Hons), a, b Chris G. Maher, PhD, c Manuela L. Ferreira, PhD, d Mark J. Hancock, PhD, e Leani S.M. Pereira, PhD, f Christopher M. Williams, PhD, g and Jane Latimer, PhD c

ABSTRACT

Objective: The purpose of this study was to identify factors that influence recruitment of patients to an observational study of low back pain (LBP). Methods: From 1147 primary care (first health contact) clinicians initially contacted, 138 (physiotherapists and chiropractors) agreed to participate in a large observational study of LBP and were the focus of the current study. Data were collected pertaining to clinicians' characteristics, operational factors, and the number of patients recruited. The association of a variety of clinician characteristics and operational factors with recruitment rate was determined using a multivariate negative binomial regression analysis. Results: From October 2011 to November 2012, 1585 patients were screened by 138 study clinicians with 951 eligible patients entering the observational study. Clinicians who were members of their professional association had a recruitment rate less than half that of those who were nonmembers (P b .0001). Clinicians who were trained by telephone had a recruitment rate 4.01 times higher than those trained face to face (P b .0001). Similarly, clinicians who referred a larger number of ineligible participants had a slightly higher recruitment rate with an incident rate ratio of 1.04 per ineligible patient (P b .0001). Other clinicians' characteristics and operational factors were not associated with recruitment. Conclusion: This study provides evidence that it is feasible to recruit participants from primary care practices to a simple observational study of LBP. Factors identified as influencing recruitment were professional association (negative association), training by telephone, and referring a higher number of ineligible participants. (J Manipulative Physiol Ther 2015;xx:1-8) Key Indexing Terms: Patient Selection; Back Pain; Primary Health Care; Physical Therapy; Chiropractic

articipant recruitment is one of the most challenging studies do not conclude on schedule due to low participation, phases of the research process and may cause studies 60% to 80% of studies do not meet their chronological Pto become unfeasible.1,2 It is estimated that 85% of endpoint because of challenges in recruitment, and 30% of

a PhD Student, Musculoskeletal Division, The George Institute for g Research Fellow, Hunter Medical Research Institute, School Global Health, Sydney Medical School, The University of Sydney, of Medicine and Public Health, University of Newcastle, Sydney, Australia. Newcastle, Australia. b PhD Student, Department of Physiotherapy, Federal University Submit requests for reprints to: Daniel Steffens, BPT of Minas Gerais, Belo Horizonte, Brazil. (Hons), PhD Student, Musculoskeletal Division, The George c Professor, Musculoskeletal Division, The George Institute for Institute for Global Health, Sydney Medical School, The Global Health, Sydney Medical School, The University of Sydney, University of Sydney, PO Box M201, Missenden Road, Sydney, Sydney, Australia. New South Wales, 2050, Australia. (e-mail: dsteffens@george d Senior Research Fellow, Musculoskeletal Division, The George institute.org.au). Institute for Global Health, Sydney Medical School, The University Paper submitted April 11, 2014; in revised form October 1, of Sydney, Sydney, Australia. 2014; accepted October 10, 2014. e Senior Lecturer, Discipline of Physiotherapy, Faculty of 0161-4754 Human Sciences, Macquarie University, Sydney, Australia. Copyright © 2014 by National University of Health f Professor, Department of Physiotherapy, Federal University of Sciences. Minas Gerais, Belo Horizonte, Brazil. http://dx.doi.org/10.1016/j.jmpt.2014.10.016 186 2 Steffens et al Journal of Manipulative and Physiological Therapeutics Factors Predicting Recruitment of Participants Month 2015

1,3,4 study sites fail to recruit even a single participant. METHODS Unsatisfactory and/or untimely participant recruitment has Design serious consequences, leading to an underpowered study, This observational study investigated primary care increased resource use and higher costs.5-7 Importantly, the clinicians enrolling patients with acute LBP to a case-cross- integrity and validity of the study also rely on obtaining an over study (TRIGGERS). Participants were recruited from adequate sample size, and failure to achieve this may cause a October 2011 to November 2012. The methods and study with inconclusive findings.8 procedures for recruitment of patients to the TRIGGERS Most previous studies have focused on investigating factors study have been published elsewhere.22 Participants in the that increase recruitment to randomized controlled trials TRIGGERS study (n = 999) were also eligible to enroll in the (RCTs).3-5,7,9-12 Although RCTs are considered the “gold PACE clinical trial.23 The PACE clinical trial is a standard” of study design,13 not all scientific questions can be double-blind placebo-controlled trial assessing the effect answered with this design. Researchers are often interested in that paracetamol has on recovery from acute nonspecific LBP. questions regarding etiology and prognosis, which may be The inclusion criteria for the TRIGGERS and PACE studies better answered using an observational study design. Many of were similar; therefore, patients recruited for the PACE the barriers encountered when recruiting participants to RCTs clinical trial could also be enrolled in the TRIGGERS study. may be similar to those encountered when conducting However, data collected from recruiting clinicians (eg, observational studies; however, factors affecting recruitment personal information) and the study operational procedures to observational studies have not been carefully evaluated.14 were different for both studies. Therefore, we reported the Previous studies have identified reasons clinicians do not data collected from participants who enrolled in the enroll eligible patients into clinical trials.15,16 Although these TRIGGERS study only (n = 951). Ethical approval for the reasons have been identified predominantly from studies study was granted by the University of Sydney Human evaluating general practitioners, it is likely that many also Research Ethics Committee (protocol no. 05-2011/13742). apply to allied health practitioners (physiotherapists and chiropractors) who are operating as first contact practitioner for patients presenting with back pain. These reasons include Participants difficulty for practitioners in following the study protocol and TRIGGERS recruited patients seeking care for LBP in completing the recruitment process and patient preference for primary care clinics across Sydney, Australia. Eligible a certain therapy and difficulties obtaining informed consent participants met the following inclusion criteria: (1) compre- from patients. In primary care, these recruitment barriers are hends spoken English, (2) main complaint of LBP with or often heightened by the clinician's lack of time, which 17 without leg pain (pain between 12th rib and buttock crease), significantly affects their ability to recruit participants. (3) current episode of back pain less than or equal to 7 days Other factors reported to influence recruitment of patients duration, (4) new episode (preceded by at least 1 month include the importance of the research question, the simplicity without LBP), (5) pain of at least moderate intensity during the of the research design, and ease of access to treatment. first 24 hours of this episode (scored on a 6-point scale from Financial reimbursement has been suggested as a possible none to very severe). The exclusion criterion was confirmed 10,18,19 factor ; however, a recent systematic review found that, or suspected serious spinal pathology (ie, cancer, fracture, in randomized controlled trials, reimbursement for time spent and infection). on recruitment is not associated with better recruitment.20 In addition, it is possible that recruiting from health professionals other than general practitioners such as physiotherapists and Clinician Recruitment chiropractors may produce a different outcome. Regardless, Primary care clinicians were recruited for this study. In recruitment of patients in primary care remains a significant Australia, primary care clinicians are those registered to issue.5,7 Therefore, studies that use simple recruitment provide the first health contact for patients presenting from strategies, minimal clinical involvement, and health profes- the community and include general practitioners, practice sionals other than general practitioners may have an nurses, psychologists, physiotherapists, chiropractors, and advantage in recruiting patients in primary care settings. pharmacists.24 According to the original protocol, general The reasons certain studies recruit successfully while practitioners and pharmacists would be contacted to aid others do not remain unclear.21 A better understanding of recruitment. However, no attempts were made to recruit clinicians' characteristics and the study operational charac- general practitioners or pharmacists as adequate numbers of teristics (eg, method of training and type and number of patients were recruited through physiotherapists and chiro- contacts) may lead researchers to identify study strategies practors. In this study, the recruiting primary care clinicians associated with recruitment of a larger number of were physiotherapists and chiropractors. participants. Therefore, the aim of this study was to identify Lists of physiotherapists working in Sydney were acquired factors that influence recruitment to an observational study from their association's Web site. All physiotherapists drawn of triggers for low back pain (LBP). from the Australian Physiotherapy Association database were 187 Journal of Manipulative and Physiological Therapeutics Steffens et al 3 Volume xx, Number x Factors Predicting Recruitment of Participants

members of the association. Thirty-five physiotherapists who were instructed to fax screening forms for both eligible and participated in the study were not members of the Australian ineligible patients to the study researchers as soon as the Physiotherapy Association and were identified by their forms were completed. For those participants agreeing to be colleagues (who were members and received our invitation). involved, consent forms were signed by the participants and The chiropractors were selected from a Google search. A total clinicians at the participating sites. of 1147 clinicians (39 chiropractors and 1108 physiotherapists) were invited to participate by letter. The letter outlined the study aims, sample size, inclusion/exclusion criteria, and Reimbursement the benefits to the clinician and patient of participation. Clinicians were reimbursed AU $99 per eligible patient Interested clinicians were invited to contact the study research referred to the study. This sum was used to cover the team to obtain further information. A study researcher phoned clinician's time spent in recruiting participants, explaining the those who responded to further explain the study procedures. study to them, and liaising with the study staff. Clinicians Additional training was provided for clinicians interested in were also reimbursed AU $10 per ineligible patient screened, recruiting to the study. A total of 138 clinicians (135 to cover clerical and administration costs. Participants were physiotherapists and 3 chiropractors) agreed to participate in reimbursed with AU $50 gift card for the time spent the study. answering the questions and to cover the cost of the mobile telephone calls to the researcher. The typical duration of the interview was approximately 30 minutes. Training Methods After confirming their interest in recruiting, clinicians were offered 2 methods of training: (1) face-to-face training at Clinician's Characteristics and Data Collection their clinics or (2) training by a telephone call. The method of Personal information from the recruiting clinicians was training was selected by the clinician. collected, including sex, date of birth, practice details (location/postcode), profession, current position, years of practice, years managing LBP, and whether the clinician was Face-to-Face Training a member of their professional association. All contacts made Clinicians who chose face-to-face training were visited between the study researchers and the clinicians were entered and trained by an experienced research assistant at their own into a database. Contacts were classified as phone call, letter, practice. Face-to-face training was supplemented with or e-mails. distribution of a paper copy of the study protocol. Training took around 30 minutes and was done in a group of up to 5 clinicians working at the same practice. Recruitment Outcome and Recruitment Predictor Variables The recruitment outcome was the total number of eligible patients recruited by each individual clinician by the end of Training by Telephone Call the study. Eligible patients were patients referred by the Clinicians who chose telephone training received their hard clinician successfully enrolled in the TRIGGERS study. copy of the study protocol by post approximately 1 week Although the prediction of recruitment study was before their telephone training call was scheduled. The calls conceived after the main TRIGGERS study commenced, were made by a research assistant and covered all the topics in the recruitment predictor variables and analysis were defined the face-to-face training. This training took around 30 minutes a priori as presented in the manuscript. Clinicians' characte- and was done individually. ristics included (1) sex (male/female), (2) age (years), (3) suburb socioeconomic status (determined by comparing the Features Covered in Both Training Methods clinic postcode to Australia Bureau of Statistics data on Irrespective of whether the clinician chose face-to-face or economic advantage and disadvantage by postcode and ≥ telephone training, the following topics were covered during dichotomized into high AU $577 or low socioeconomic b training: study background, aims, screening form (inclusion/ status AU $577, based on the average individual weekly exclusion criteria—refer to supplementary material for income), (4) profession (physiotherapist or chiropractor), (5) further details), informed consent form, referring patients clinical experience as practicing clinician (years), (6) clinical (providing patient contact information to the research team), experience managing LBP (years), (7) current position terms and conditions, human ethics, and study benefits. (employee or business owner), and (8) professional associa- Clinicians could opt out at any time during the study period. tion membership status (member or not member). Operational factors included (1) training method (face-to-face or telephone call), (2) number of letters (total number of letters sent to the Screening Patients clinician by end of the study), (3) number of telephone calls Clinicians were asked to screen for eligibility all (ie, (total number of phone calls made to the clinician by end of consecutive) patients who presented with LBP. Clinicians the study), (4) total number of e-mails (total number of e-mails 188 4 Steffens et al Journal of Manipulative and Physiological Therapeutics Factors Predicting Recruitment of Participants Month 2015

Table 1. Characteristics of Recruiting Clinicians Stratified by Table 2. Clinician's Descriptive Data (n = 138) Recruitment Rate Mean ± SD or Recruitment No. of No. of Eligible No. of Ineligible Variables n (%) Rate per Clinicians Participants per Participants per a Sex, male 73 (53) Month (%) Strata (%) Strata (%) Age 42 ± 10 0 32 (23.2) 0 (0) 8 (1.3) Profession, physiotherapist 135 (98) N0-0.5 44 (31.9) 112 (11.6) 69 (10.9) Current position N0.5-1 27 (19.6) 162 (17.1) 235 (37.0) Employee 67 (48.5) N1-2 20 (14.5) 197 (20.7) 158 (24.9) Business owner 71 (51.5) N2 15 (10.9) 480 (50.6) 164 (25.9) Clinical experience (y) 19.5 ± 10 Total 138 (100) 951 (100) 634 (100) Clinical experience managing LBP (y) 18.5 ± 9.5 a a SES of suburb of clinician clinic (high) 104 (75.5) Recruitment rate: total number of participants recruited divided by Member of their respective association 103 (74.5) the number of months in the study. Training method (telephone) 79 (57.2) No. of letters b 1.1 ± 0.3 No. of telephone call b 0.4 ± 0.3 No. of e-mails b 0.3 ± 0.2 Total no. of contacts (letter/e-mail/telephone call) b 1.8 ± 0.6 sent to the clinician by end of the study), (5) total number of b contacts (total number of phone calls, letters, or e-mails made No. of ineligible patients 4.6 ± 13.9 and/or sent to the clinician by end of the study), and (6) LBP, low back pain; SES, socioeconomic status. a number of ineligible patients referred (ineligible patients were Suburb socioeconomic status—determined by comparing the clinic defined as patients not willing to participate or not fulfilling postcode to Australia Bureau of Statistics data on economic advantage and disadvantage. study inclusion criterion). b Number of contacts divided by the total time (months) participating in the study. Data Analysis Analyses were performed using STATA version 12 25 (College Station, TX). Descriptive statistics were per- eligible patients who entered the study (943 referred by formed to describe the clinician's characteristics and the physiotherapists and 8 referred by chiropractors). Table 1 recruitment rate for the study (defined as the number of shows participant recruitment rate per month. The overall patients per month). To evaluate factors that influenced recruitment rate per clinician was 0.99 patients per month of clinicians' recruitment rate, a negative binomial regression participation. A minority of study clinicians (n = 15 and all analysis was conducted where recruitment rate was the physiotherapists) recruited more than 50% of the participants dependent variable and the predictors described above (n = 480). Thirty-two clinicians (23.2%) did not recruit a (clinicians and operational characteristics) were independent single participant during the study period. The top 15 variables. We used the negative binomial regression analysis clinicians (clinicians with recruitment rate N2 patients per because the outcome data were overdispersed (tested by month) recruited a median of 24 participants to the study and comparing the variance of the data to the mean patient count determined that 164 patients were ineligible. For the low recruited by clinicians with the likelihood ratio test). Variables recruiters (clinicians with recruitment rate ≤2 patients per b with significant univariate associations (P .2) were entered month), the median was 3 and determined that 472 patients into a backward stepwise multivariate regression model. were ineligible. b Statistical significance was defined as P .05. As clinicians Clinician's descriptive data are presented in Table 2.Most started the study on different dates, this was accounted for in of the study clinicians were physiotherapists (98%) and had a the analysis by including the number of days in the study as an mean clinical experience managing LBP of 18.5 years. More offset variable in the model. For continuous variables, the than half of the clinicians preferred the training to be performed incident rate ratio (IRR) can be interpreted as the rate ratio in by telephone (57.2%) rather than face to face (42.8%). which the total number of participants is expected to change with a 1-unit increase in the exposure variable. For binary variables, the IRR indicates the expected change in rate of Factors That Influenced Recruitment Rate: Univariate and Multivariate patient recruitment when the variable is positive. Analyses Five clinician factors (sex, age, clinical experience as practicing clinician, clinical experience managing LBP, and RESULTS whether clinicians were members of their respective associa- From 1147 clinicians initially contacted, 135 physiothera- tions) and 5 operational factors (training method, number of pists (12.2%) and 3 chiropractors (7.7%) agreed to letters, number of telephone calls, number of contacts—letter/ participate. Between October 2011 and November 2012, e-mail/telephone call, and number of ineligible patients study clinicians screened 1585 patients. There were 951 referred) revealed a significant association (P b .2) with 189 Journal of Manipulative and Physiological Therapeutics Steffens et al 5 Volume xx, Number x Factors Predicting Recruitment of Participants

Table 3. Characteristics Associated With Recruitment of DISCUSSION Participants, Univariate and Multivariate Analysis (n = 138) Main Findings Univariate Multivariate Analysis Analysis Although 41.3% of the clinicians referred 2 or less eligible participants during the study period, we successfully recruited Factors IRR (95% CI) IRR (95% CI) our target sample (n = 951) in a reasonable period of time Clinicians factors (13.8 months). The overall recruitment rate was 0.99 patients, Sex, male 2.02 (1.23-3.31) a – Age 1.04 (1.01-1.06) a – per clinician, per month of participation. This provides Profession, 1.32 (0.80-2.18) – evidence that, in relatively simple observational studies for physiotherapist LBP, where clinicians are reimbursed for their time, it should Current position, 0.49 (0.08-2.95) – be relatively easy to recruit large numbers of participants from c employee primary care. The recruitment success of this study was Clinical experience, y 1.03 (1.00-1.05) a – Clinical experience 1.03 (1.01-1.06) a – achieved mainly because 15 primary care clinicians recruited managing LBP, y 50.6% of the sample. SES of suburb of 1.13 (0.63-2.03) – Among the clinician and operational characteristics clinician clinic, high d investigated, 3 of 14 factors increased recruitment. However, a b Member of their 0.41 (0.24-0.72) 0.42 (0.25-0.71) these factors must be considered carefully as they are respective association Operational factors unsurprising or uninterpretable and the practical implications Training method, 2.98 (1.84-4.81) a 4.01 (2.38-6.79) b seem limited. Clinicians that were members of their telephone respective associations had a recruitment rate less than No. of letters 0.95 (0.89-1.01) a – nonmembers conflicts with the view that members who a – No. of telephone call 0.92 (0.82-1.03) engage in continuing education are more likely to be No. of e-mails 0.94 (0.79-1.12) – Total no. of contacts, 0.97 (0.93-1.01) a – interested in research. Even in studies that recruit the required letter/e-mail/telephone call sample size in a reasonable time frame, identifying factors No. of ineligible patients 1.04 (1.01-1.08) a 1.03 (1.02-1.06) b that increase recruitment seems challenging, providing a With continuous variables, the IRR can be interpreted as the rate ratio in which strong case for the urgent need for more studies. the total number of participants is expected to change with a 1-unit increase in This study investigated clinician and operational charac- the exposure variable. With binary variables, the IRR indicates the expected teristics and did not assess the characteristics of patients; change in rate of patient recruitment when the variable is positive. investigation would require a different study design, that is, CI, confidence interval; IRR, Incident rate ratio; LBP, low back pain. one where the characteristics of patients not recruited to the a Candidate variables with significant univariate association (P b .2) that entered the multivariate analysis. trial are also determined. Patients may decline participation b P b .0001. for a variety of reasons including lack of time, lack of c Employee compared with business owner. understanding of relevance of research question, already d Suburb socioeconomic status—determined by comparing the clinic anxious about their disease, and others. Understanding better postcode to Australia Bureau of Statistics data on economic advantage and the patient characteristics that predict participation in clinical disadvantage, defined as high, greater than or equal to AU $577, or low, less than AU $577. trials of back pain is an important area for future research. patient recruitment in the univariate analyses (Table 3)and Comparison With Other Studies were candidates for the multivariate analysis. Almost universally, recruitment is a challenge.26 To date, After the backward stepwise regression, 3 variables were most of the studies investigating factors that influence remaining in the model. These variables are presented in recruitment of patients in primary care have focused on Table 3 with incident rate ratios. From the clinician's RCTs.27 Although many of the challenges encountered with characteristics, only whether clinicians were members of patient recruitment to RCTs are also applicable to observational their respective associations was associated (inversely) with studies, there may be important differences.28 There is a recruitment. Clinicians that were members of their significant lack of research on observational study designs. We respective associations had a recruitment rate less than identified no previous observational studies that reported half that of nonmembers (P b .001). The other 2 variables recruitment rates in primary care and, therefore, could not associated with recruitment were operational factors compare our research findings with previous studies in the field. (training method and number of ineligible patients Findings from our observational study show that clinicians referred). Clinicians that were trained over the telephone trained by telephone and those who refer ineligible patients had a recruitment rate 4.01 times greater than those trained throughout the study are likely to have higher recruitment face to face. Similarly, clinicians that referred a higher rates. Clinicians who are a member of their professional number of ineligible participants had a greater recruitment association are less likely to recruit. These factors have not rate, with incident rate ratio of 1.03 (P b .0001). previously been identified in earlier studies as important to 190 6 Steffens et al Journal of Manipulative and Physiological Therapeutics Factors Predicting Recruitment of Participants Month 2015

recruitment. The latest Cochrane review on strategies to recruiting participants, explaining the study to them, and influence recruitment for RCTs found that using telephone liaising with the study staff and clerical and administration reminders, opt-out procedures requiring potential partici- costs. The time involved in this process took from 30 to 45 pants to contact the trial team if they did not want to be minutes, and the reimbursement valued the clinician's time contacted about a trial, making the trial open rather than according to prevailing physiotherapy consultation fees. This blinded, and mailing a questionnaire about home safety to strategy was used to ensure that financial reimbursement was potential participants to an injury prevention trial are not considered an inducement to participate. To minimize factors that improved recruitment in high-quality studies.5,7 errors, eligibility criteria were double checked by a study A systematic review reported on a variety of strategies to researcher at the time of the interview. improve recruitment, the most common being the use of We advised at the initial training and re-enforced letters, e-mails, and telephone calls to clinicians3,4;however, throughout the whole study period that all study clinicians similar to our findings, these factors did not significantly should invite all consecutive patients presenting with LBP to increase recruitment. the study. If clinicians did not enroll consecutive patients, it would have the potential to include sampling bias in the parent TRIGGERS study. However, we do not believe that this would Limitations and Strengths introduce bias into this study of factors influencing recruitment. Some of the strengths of this study are the large number In this study, clinicians were not randomly allocated to of clinicians that participated in this LBP study and the either training by telephone or by face-to-face visit. The large number of patients recruited in a short period. These training method was determined by the clinician, and this large numbers have enabled us to robustly assess clinician choice may reflect other confounders in the practice.30 and operational features that, in combination, could lead Clinicians that opted to be trained by telephone may have to successful recruitment of patients to LBP studies in chosen this due to their busier clinic schedule suggesting primary care. contact with a larger number of patients per day than other One weakness of this study was that the factors investigated clinicians and, hence, an increased opportunity to recruit. may apply predominantly to simple observational studies. The Regardless, the finding that, in a simple observational study, simplified design, minimal role required by the study training of clinicians by telephone appears to be at least as clinicians, and the reimbursement for the time and inconve- effective as face-to-face training for recruitment has impor- nience may have contributed to the rapid patient recruitment. tant implications. The training administered by telephone was The factors associated with recruitment were relatively delivered one to one, as opposed to face-to-face training unexpected. Other clinician characteristics not investigated in where 1 or more clinicians (up to 5) were trained at a given this study may be important in influencing recruitment. time. The individualized training and feedback are effective Although previous studies have described financial in improving recruitment.31 The results that better recruit- reimbursement as important for recruiting clinicians and ment is associated with referral of more ineligible participants patients,9,29 one of the few systematic reviews does not could be due to a higher overall number of invitations. The support this. Clinicians who identify reimbursement as a key relation with professional membership is a complete mystery. reason for participating in an RCT are no more likely to recruit The recruitment of primary care clinicians in this study patients than those who do not.20 In the current observational was based on an invitation letter sent by mail, and despite this study, financial reimbursement for both clinicians and relatively passive method of recruitment, we could interest a participants may have influenced recruitment; however, we suitable number of clinicians in participating in our study in a could not assess this using our current methods. In addition, relatively short while. Had we used more active methods of we could not assess if practice-level characteristics of the encouraging clinicians to participate, such as providing providers affected recruitment. Factors such as full-time vs educational seminars and distributing advertisements and part-time employment status, ownership of multiple practices, newsletters to association databases, we may have had more employment of other therapists, specialty practices, number of rapid recruitment of clinicians. Chiropractors were identified patients treated per week, average duration of consultation by a Google search using the words “Chiropractor Sydney.” session, referral patterns, university affiliation, or other The order of appearance in the search may be affected by the organizational factors were not measured in the current search engine optimization, hence, favoring chiropractors study, and therefore, their effect on patient recruitment who have greater knowledge. However, it remains unclear remains unclear. Future research might explore the influence whether these clinicians would have recruited more subjects of practice characteristics on research recruitment rates. In to the study. addition, we did not attend the clinics to observe if the In this study, we contacted the Australian Physiotherapy providers were completing participant recruitment as per Association as our primary means of identifying physiothera- protocol, and this is a limitation of the study. pists but used a Google search to identify chiropractors. It In this study, the money reimbursed for eligible and/or would have been better to use similar methods to identify ineligible patients was to cover clinicians' time spent in both professions. Future studies should use similar methods 191 Journal of Manipulative and Physiological Therapeutics Steffens et al 7 Volume xx, Number x Factors Predicting Recruitment of Participants

to recruit primary care clinicians. Either, physiotherapy and Supervision (provided oversight, responsible for orga- chiropractic associations should be contacted and using the nization and implementation, writing of the manu- Google engine to identify nonmember clinicians. script): D.S., C.M., M.F, M.H, L.P, C.W, J.L. Data collection/processing (responsible for experi- ments, patient management, organization, or reporting Future Research data): D.S., C.M., M.F., M.H., L.P., C.W., J.L. There is a small but emerging body of literature on factors Analysis/interpretation (responsible for statistical anal- influencing recruitment. The operational factor (trained by ysis, evaluation, and presentation of the results): D.S., telephone) identified as influencing recruitment in this study C.M., M.F., M.H., L.P., C.W., J.L. may only be appropriate in simple study designs but should be Literature search (performed the literature search): N.A. investigated further due to the potential to make clinician Writing (responsible for writing a substantive part of the recruitment easier and cheaper. To date, there are no studies manuscript): D.S., C.M., M.F., M.H., L.P., C.W., J.L. investigating if primary care clinicians that are members of Critical review (revised manuscript for intellectual content, their respective associations are more or less likely to this does not relate to spelling and grammar checking): D.S., participate and recruit patients for research. This information C.M., M.F., M.H., L.P., C.W., J.L. would be of value, as clinicians that are members may be easier to contact through their professional association. Future studies should also investigate if practice-level characteristics could influence patient recruitment. In addition, other potentially important factors that could influence recruitment and/or clinicians' behavior, such as Practical Applications number of contacts made with the clinician and reimbursement • This study provides evidence that, in relatively for time involved for the clinician and administrative staff, need simple observational studies for LBP, it should be to be considered in future studies. A better understanding of the relatively easy to recruit large numbers of patient characteristics associated with successful recruitment is participants from primary care. also urgently needed. Further research on factors that could • However, even in studies that recruit the required maximize recruitment rate must be conducted. Factors that sample size in a reasonable time frame, identifying influence patient recruitment in primary care are complex and factors that increase recruitment seems challeng- remain unclear. ing, providing a strong case for the urgent need for more studies in this area.

CONCLUSIONS Although patient recruitment is a challenge, this study of recruiting participants from primary care clinicians for a large observational study of LBP has been positive. Factors identified as influencing recruitment were professional REFERENCES association (negative association), training by telephone, and referring a higher number of ineligible participants. 1. Blanton S, Morris D, Prettyman M, et al. Lessons learned in participant recruitment and retention: the EXCITE trial. Phys This study has revealed factors associated with recruit- Ther 2006;86:1520-33. ment rate, although the ability to predict which clinician will 2. Bowen J, Hirsch S. Recruitment rates and factors affecting recruit based on operational and clinicians characteristics recruitment for a clinical trial of a putative anti-psychotic agent seems restricted. in the treatment of acute schizophrenia. Hum Psychopharmacol 1992;7:337-41. 3. McDonald A, Knight R, Campbell M, et al. What influences recruitment to randomised controlled trials? A review of trials FUNDING SOURCES AND POTENTIAL CONFLICTS OF INTEREST funded by two UK funding agencies. Trials 2006;7:1-8. 4. Sully BG, Julious SA, Nicholl J. A reinvestigation of recruitment No funding sources or conflicts of interest were reported to randomised, controlled, multicenter trials: a review of trials for this study. funded by two UK funding agencies. Trials 2013;14:166. 5. Treweek S, Pitkethly M, Cook J, et al. Strategies to improve recruitment to randomised controlled trials. Cochrane Database Syst Rev 2010;(4):MR000013. CONTRIBUTORSHIP INFORMATION 6. Mapstone J, Elbourne D, Roberts I. Strategies to improve Concept development (provided idea for the research): recruitment to research studies. Cochrane Database Syst Rev 2007;(2):MR000013. D.S.,CM.,M.F.,M.H.,L.P.,C.W.,J.L. 7. Treweek S, Lockhart P, Pitkethly M, et al. Methods to improve Design (planned the methods to generate the results): DS., recruitment to randomised controlled trials: Cochrane systematic C.M., M.F., M.H., L.P., C.W., J.L. review and meta-analysis. BMJ Open 2013;3:e002360. 192 8 Steffens et al Journal of Manipulative and Physiological Therapeutics Factors Predicting Recruitment of Participants Month 2015

8. Thoma A, Farrokhyar F, McKnight L, Bhandari M. How to of a randomised controlled trial. BMC Musculoskelet Disord optimize patient recruitment. Can J Surg 2010;53:205-10. 2010;11:1-6. 9. Rosemann T, Szecsenyi J. General practitioners' attitudes 20. Rendell JM, Merritt RD, Geddes JR. Incentives and disincen- towards research in primary care: qualitative results of a cross tives to participation by clinicians in randomised controlled sectional study. BMC Family Pract 2004;5:1-5. trials. Cochrane Database Syst Rev 2007:MR000021. 10. Foy R, Parry J, Duggan A, et al. How evidence based are 21. Prescott R, Counsell C, Gillespie W, et al. Factors that limit the recruitment strategies to randomized controlled trials in primary quality, number and progress of randomised controlled trials. care? Experience from seven studies. Fam Pract 2003;20:83-92. Health Technol Assess 1999;3:1-143. 11. Lannin N, Cusick A. Factors affecting patient recruitment in an 22. Steffens D, Ferreira ML, Maher CG, et al. Triggers for an acute rehabilitation randomized controlled trial. Am J Occup episode of sudden onset low back pain: study protocol. BMC Ther 2006;60:177-81. Musculoskelet Disord 2012;13:7. 12. Page MJ, French SD, McKenzie JE, O'Connor DA, Green SE. 23. Williams CM, Maher CG, Latimer J, et al. Efficacy of Recruitment difficulties in a primary care cluster randomised paracetamol for acute low-back pain: a double-blind, rando- trial: investigating factors contributing to general practitioners' mised controlled trial. Lancet 2014;384:1586-96. recruitment of patients. BMC Med Res Methodol 2011;11:35. 24. Improving primary health care for all Australians. Ageing 13. Chalmers I, Rounding C, Lock K. Descriptive survey of non- DoHa. Canberra: Commonwealth of Australia; 2011. commercial randomised controlled trials in the United 25. Stata Statistical Software: Release 12 [computer program]. Kingdom, 1980-2002. BMJ 2003;327:1-4. Version 12. College Station, TX: StataCorp LP; 2011. 14. Bower P, Wilson S, Mathers N. Short report: how often do UK 26. Ewing G, Rogers M, Barclay S, et al. Recruiting patients into a primary care trials face recruitment delays? Family Pract 2007; primary care based study of palliative care: why is it so 24:601-3. difficult? Palliat Med 2004;18:452-9. 15. Abraham N, Young J, Solomon M. A systematic review of 27. Rollman B, Fischer G, Zhu F, Belnap B. Comparison of reasons for nonentry of eligible patients into surgical electronic physician prompts versus waitroom case-finding on randomized controlled trials. Surgery 2006;139:469-83. clinical trial enrollment. J Gen Intern Med 2008;23:447-50. 16. Jenkinson CE, Winder RE, Sugg HV, et al. Why do GPs 28. Hayward R, Porcheret M, Mallen C, Thomas E. Recruiting exclude patients from participating in research? An exploration patients and collecting data for an observational study using of adherence to and divergence from trial criteria. Fam Pract computerised record pop-up prompts: the PROG-RES study. 2014;31:364-70. Prim Health Care Res Dev 2013;14:21-8. 17. Spaar A, Frey M, Turk A, Karrer W, Puhan MA. Recruitment 29. Langley C, Gray S, Selley S, Bowie C, Price C. Clinicians' barriers in a randomized controlled trial from the physicians' attitudes to recruitment to randomised trials in cancer care: a perspective: a postal survey. BMC Med Res Methodol 2009;9:14. qualitative study. J Health Serv Res Policy 2000;5:164-9. 18. Murray CJ, Vos T, Lozano R, et al. Disability-adjusted life 30. Mann B, Wood E. Confounding in observational studies years (DALYs) for 291 diseases and injuries in 21 regions, explained. Open Epidemiol J 2012;5:18-20. 1990-2010: a systematic analysis for the Global Burden of 31. Donovan JL, Lane JA, Peters TJ, et al. Development of a Disease Study 2010. Lancet 2012;380:2197-223. complex intervention improved randomization and informed 19. Williams C, Latimer J, Maher C, et al. PACE—the first placebo consent in a randomized controlled trial. J Clin Epidemiol 2009; controlled trial of paracetamol for acute low back pain: design 62:29-36. 193

Chapter Ten

Conclusions

194

10.1. Aim

The primary aim of this thesis was to contribute to a better understanding of the mechanisms for low back pain. New knowledge was acquired in a number of ways that included interviewing primary care clinicians, conducting relevant systematic reviews, measuring exposure to back pain risk factors, and exploring prognosis for patients with chronic low back pain.

The first study contributed knowledge on the mechanisms of onset of back pain, identifying the short and long-term risk factors that primary care clinicians consider important in triggering an episode of low back pain (Chapter Two). The studies described in Chapter Four and Chapter Five continued this theme investigating a range of physical, psychosocial (Chapter Four) and environmental factors (Chapter Five) that increase risk for an episode of sudden onset, acute low back pain. The study presented in Chapter Six aimed to systematically review whether magnetic resonance imaging findings of the lumbar spine predict future low back pain. In terms of back pain management, the study presented in Chapter Seven aimed to examine the prognosis and prognostic factors for patients with chronic low back pain. The systematic review presented in Chapter Eight aimed to investigate if the presence of magnetic resonance imaging findings identifies patients with low back pain who respond better to particular interventions. Finally, the study presented in Chapter Nine aimed to identify factors that influence recruitment of participants to a large observational study.

10.2. Overview of principal findings

The observational study described in Chapter Two revealed that Australian primary care clinicians believe that biomechanical risk factors (89.3%), such as lifting (17.5%), prolonged sitting (9.1%) and physical trauma (8.9%), are the most likely short-term risk factors for low back pain. Biomechanical risk factors (54.2%), such as prolonged sitting (13.4%) and lifting (10.9%), and individual risk factors (39%), such as physical inactivity (9.1%) and other individual risk factors (5.8%) are the most endorsed long-term risk factors. Surprisingly, commonly reported risk factors, such as psychological or psychosocial factors (0.6% and 3.1% for short and long-term respectively), and genetic risk factors (0.0% and 0.2% for short and long-term respectively) were considered unimportant by clinicians. 195

Prior to this thesis there had been no high quality study that used a case-crossover design to determine the effects of physical, psychosocial and meteorological factors on the risk of an episode of sudden onset, acute back pain. The case-crossover study described in Chapter Four demonstrated for the first time that brief exposure to a range of physical factors (e.g. manual tasks involving awkward postures, or manual tasks involving an object that could not be positioned close to the body) and psychosocial factors (being distracted during a task or being fatigued) can considerably increase the risk of an episode of low back pain. However, these associations were not moderated by habitual physical activity, BMI, previous episodes of low back pain, depression or anxiety. Age moderated the risk associated with exposure to heavy loads and sexual activity. Chapter Five presents a case-crossover study investigating the influence of weather conditions on risk of low back pain. The findings demonstrated that there was no association between temperature, relative humidity, air pressure, wind direction and precipitation and risk of back pain. Higher wind speeds slightly increased the odds of back pain onset, although the effect was not considered clinically important.

The systematic review investigating magnetic resonance imaging (MRI) findings as a predictor of future low back pain (Chapter Six) identified twelve longitudinal studies. Of these, most enrolled small samples, investigated different MRI findings and presented varied clinical outcomes. Across the 46 MRI findings investigated, no consistent associations with clinical outcomes were identified. Three different studies reported associations for Modic changes with pain, disc degeneration with disability in samples with current low back pain and disc degeneration with pain in a mixed sample of patients with and without current low back pain.

The prognosis study described in Chapter Seven revealed that patients with chronic low back pain presenting to a private, community-based group exercise program improved clinical outcomes significantly, with greater improvements in disability compared to pain at 12 months. The predictors investigated accounted for only 10% and 15% of pain and disability outcomes, respectively, suggesting that there is much to learn about the factors influencing recovery in this group of patients.

Chapter Eight reports the findings from a systematic review investigating if the presence of magnetic resonance imaging findings identifies patients with low back pain who respond better to particular interventions. Although this review identified eight clinical trials, 196

investigating 38 interactions for low back pain and sciatica, only two individual trials suggested some magnetic resonance imaging findings that might be effect modifiers for specific interventions. It is unknown if these subgroup interactions accurately represent the association, given the limited number of suitable trials and the heterogeneity across them.

The observational study described in Chapter Nine found three variables (clinicians not members of their professional association, clinicians trained by phone and clinicians who referred a larger number of ineligible patients) associated with the rate at which primary care clinicians recruited patients to the study. However, the applicability and understanding of some of these factors seems counterintuitive; indicating that identifying primary care clinicians likely to recruit at faster rates is complex.

The findings of these studies have advanced the understanding of mechanism of low back pain in relation to risk, prognosis and response to treatment. There are several important implications and directions for future research that arise from these studies.

10.3. Implications and suggestions for future research

10.3.1. Mechanism: Risk factors for low back pain

Currently, there are a number of recognised risk factors for low back pain (HOOGENDOORN et al., 2000; LINTON, 2001; HAMBERG-VAN REENEN et al., 2007; HENEWEER et al., 2011; LANG et al., 2012), however, most of these risk factors are based on long term exposure (e.g. smoking), and many are not modifiable (e.g. age). The identification of risk factors in primary care is crucial to help the development of new research, which may lead to future prevention programs (RUBIN, 2007). The study presented in Chapter Two, based on primary care clinicians views, identified important short (biomechanical) and long-term (biomechanical and individual) risk factors for low back pain. While primary care clinicians beliefs on biomechanical and individual risk factors are aligned with past research (TAYLOR et al., 2014), the lack of consideration towards psychosocial and genetic risk factors is quiet surprising. Thus, the reasons why primary care clinicians consider psychosocial and genetic risk factors unimportant need to be further investigated to aid management and prevention of this condition. Previous studies suggest that psychosocial risk factors (e.g. low job control) are often correlated with biomechanical risk factors (e.g. intensive load) (MACDONALD et al., 2001). One reason why experienced clinicians think 197

these factors are not important, may be that they have limited expertise in assessing these factors as triggers for low back pain. Another reason may be that this population represents a general sample of the population, while previous studies have focused on occupational settings (HOOGENDOORN, VAN POPPEL et al., 2000; KERR et al., 2001; HOY et al., 2010). Although low back pain is commonly reported as multifactorial (TAYLOR, GOODE et al., 2014), future studies are needed to investigate if two or more factors present higher risk of back pain development.

The large case-crossover study presented in Chapter Four provides clear evidence that brief exposure to a range of physical and psychosocial triggers substantially increased risk for a new episode of back pain. One of the advantages of this study is the novel design used and that the risk factors found to increase the risk of a low back pain episode are readily modifiable. In this robust design, participants act as their own controls, and therefore the perfect matching of cases and controls eliminates potential effects of unmeasured confounders, such as genetic and lifestyle influences, on back pain development (HOOGENDOORN, VAN POPPEL et al., 2000; HENEWEER, STAES et al., 2011). The fact that the risk factors found in this study are modifiable (e.g. lifting) will support the development of new prevention approaches for back pain. Previous research has focused on factors that are not modifiable (e.g. age and height) or involve long-term exposure (e.g. smoking) (KOPEC, SAYRE and ESDAILE, 2004; SHIRI et al., 2010). While we found a strong association with most of factors investigated, future studies are needed to validate these factors in other populations. Additionally, there may be other potential risk factors we failed to measure that may be modifiable.

Well-designed studies controlling exposure to these risk factors, either at home or in the workplace, should be a priority as secondary prevention of low back pain could reduce individuals’ suffering and reduce heath expenditure. Also, future studies could evaluate the effect of educating patients with previous history of back pain, providing accurate information about the natural history of the condition to help reduce patient’s concerns and to promote compliance with prevention strategies. The results of this study will also have significant implications for clinicians and policy makers for the control of low back pain episode.

The results of the case-crossover study presented in Chapter Five showed no association between temperature, relative humidity, air pressure, wind direction, precipitation 198

and sudden onset, acute low back pain. Higher wind speed and wind gust speed, only slightly increase the risk of back pain and, while this reached statistical significance, the magnitude of the increase was not considered clinically important. Interestingly, this is the first study to use a robust case-crossover design to investigate the influence of the weather on back pain. Future research could focus on a wide range of patients’ characteristics (e.g. beliefs, mood, memory) to explain individual differences in weather sensitivity. Future research should determine if this insignificant association of weather parameters with back pain found in Sydney, Australia, holds in more extreme climate conditions and in different clinical settings. At the present, it seems that other risk factors are more important for the development of acute back pain, such the factors investigated in Chapter Four (physical and psychosocial).

Lumbar imaging is routinely prescribed for the diagnosis of patients with low back pain (JENSEN et al., 2008), however, the importance of the findings remains controversial (MODIC and ROSS, 2007). Previous studies revealed high rates of abnormalities on MRI in people without low back pain (JARVIK et al., 2001), though, this may represent markers of early pre-symptomatic disease that is later characterized by episodes of back pain. Chapter Six reported the results from a systematic review on the association of magnetic resonance imaging and future low back pain. Although this review found that three single studies presented significant associations (Modic changes with pain and disc degeneration with disability, in samples with current low back pain; and disc degeneration with pain in a mixed sample), it is uncertain if these estimates accurately represent the association given the quality, sample size and heterogeneity among the included studies. Thus, further large, high- quality studies that address the aforementioned problems are clearly needed to help determine the clinical meaningfulness of lumbar imaging in relation to low back pain.

Investigations into the association between lumbar MRI findings and low back pain are complicated as multiple findings are present at the same time. Findings such as lumbar intervertebral disc protrusions or endplate changes, almost always co-exist with other degenerative disc findings, such as disc height reduction and signal intensity (WANG, VIDEMAN and BATTIE, 2012). An initial strategy to advance this area of investigation would be to recognise which MRI findings typically occur together and whether clusters of findings are more predictive of outcome than single findings.

199

10.3.2. Management: Prognosis and subgroups for low back pain

Understanding prognostic factors that are associated with better or worse disease outcome can help identify possible determinants and causal pathways for low back pain, which may lead to more effective management strategies (HAYDEN et al., 2010). The findings of the prognosis study reported in Chapter Seven showed that patients with chronic low back pain who presented to an exercise program incorporating cognitive behaviour therapy improved considerably over the course of one year. However, the predictors investigated accounted for only 10% and 15% of pain and disability outcomes, respectively. This information on prognostic outcome is important for patients and clinicians as it helps to set realistic expectations and can be used to guide decision making regarding the need for additional interventions. It is likely that prognosis depends on multiple factors (HAYDEN et al., 2009). Further investigation of important and novel predictors are needed (e.g. stress, job satisfaction, beliefs). Advanced phases of investigation are needed to progress the low back pain prognosis field, including confirmation studies for prognostic factors with more frequent (e.g. monthly) and longer follow-up (e.g. longer than 12 months).

Selecting a representative cohort is a key consideration in designing studies on the prognosis of back pain. For studies investigating the prognosis of low back pain, the ideal is to assemble a sample that is at risk of developing chronic low back pain and then identifying an inception cohort from incident cases (COSTA LDA et al., 2007). These inception cohorts generally provide stronger evidence on prognosis than cohorts assembled from available cases (e.g. survival cohorts) and are therefore recommended for future research. Lastly, intervention strategies for low back pain should make greater use of prognostic data to support the theoretical rationale for interventions and to identify the group of patients most likely to benefit from it.

The effects of most clinical interventions for the management of low back pain reported in trials are usually classified as small or moderate at best (HAYDEN et al., 2005). There is always the argument that different patients will not respond similarly to the same intervention and, therefore, subgroups should be considered (COSTA LDA et al., 2013). The systematic review in Chapter Eight found eight trials that investigated 38 subgroup interactions; one presented a significant effect modifier for low back pain and one for sciatica populations. Although some statistically significant subgroup interactions were noted, it is questionable if the estimates accurately represent the effect modification. This is largely due 200

to the limited number, heterogeneity and overall quality of studies found. Interestingly, subgroup studies are a research priority in the low back pain field since 2007 (HENSCHKE et al., 2007; COSTA LDA, KOES et al., 2013), however, there are only a few clinical trials and most are underpowered. Therefore, well-designed, adequately powered trials are required. Moreover, the general content and reporting of subgroup analyses is rather poor. It is thus recommended, that authors use available guidelines when performing subgroup analyses to ensure that they are reliable and of a good standard (ROTHWELL, 2005).

Another problem with subgroup analysis is that the sample size required is around four times larger than if the only interest was the main effect of treatment (CUZICK, 1999). Therefore, acquiring a reasonable sample size is always a challenge for subgroup studies. The combination of data across multiple studies may be a practical option to gain power and overcome some of the concerns reported in our review. However, data should be combined only when studies are homogeneous. This is not always obvious and requires great caution.

10.3.3. Factors influencing recruitment rate

Identifying important factors that influence recruitment to observational studies is required to improve the efficiency, impact and success of research studies (WILLIAMS et al., 2014). Many of the factors investigated in Chapter Nine did not appear to influence participant recruitment rate. The identification of factors that increase recruitment remains challenging, providing a strong case for the urgent need for more studies in this area. For this reason, future research should appropriately focus on identifying other variables that could be incorporated with factors already known to improve participant recruitment rate.

Recruiting participants to research studies in primary care setting presents some unique challenges as primary care studies often rely on clinicians to screen and enrol patients. One possible way to optimise recruitment of patients to observational studies is the use of a computerised pop-up prompts. Previous studies have reported that one of the main problems in recruiting patients was that clinicians are time poor and normally forget about the study due to the large amount of patients seen per day (SPAAR et al., 2009). This new method reminds clinicians of potential eligible patients using computer prompts when patients’ information is entered into their electronic medical records. This approach could potentially enhance recruitment to future research studies. 201

Finally, the series of studies described in this thesis provide new, important information that lead to a better understanding of the mechanisms of low back pain. It is hoped that the findings and recommendations that arise from this thesis are widely adopted in future low back pain research. 202

10.4. References

COSTA LDA, C., N. HENSCHKE, C. G. MAHER, K. M. REFSHAUGE, R. D. HERBERT, J. H. MCAULEY, A. DAS and L. O. COSTA, Prognosis of chronic low back pain: design of an inception cohort study. BMC Musculoskelet Disord, 8, p. 11, 2007. COSTA LDA, C., B. W. KOES, G. PRANSKY, J. BORKAN, C. G. MAHER and R. J. SMEETS, Primary care research priorities in low back pain: an update. Spine (Phila Pa 1976), 38, 2, p. 148-156, 2013. CUZICK, J., Interaction, subgroup analysis and sample size. IARC Sci Publ, 148, p. 109- 121, 1999. HAMBERG-VAN REENEN, H. H., G. A. ARIENS, B. M. BLATTER, W. VAN MECHELEN and P. M. BONGERS, A systematic review of the relation between physical capacity and future low back and neck/shoulder pain. Pain, 130, 1-2, p. 93-107, 2007. HAYDEN, J. A., R. CHOU, S. HOGG-JOHNSON and C. BOMBARDIER, Systematic reviews of low back pain prognosis had variable methods and results: guidance for future prognosis reviews. J Clin Epidemiol, 62, 8, p. 781-796 e781, 2009. HAYDEN, J. A., K. M. DUNN, D. A. VAN DER WINDT and W. S. SHAW, What is the prognosis of back pain? Best Pract Res Clin Rheumatol, 24, 2, p. 167-179, 2010. HAYDEN, J. A., M. W. VAN TULDER, A. MALMIVAARA and B. W. KOES, Exercise therapy for treatment of non-specific low back pain. Cochrane Database Syst Rev, 3, p. CD000335, 2005. HENEWEER, H., F. STAES, G. AUFDEMKAMPE, M. VAN RIJN and L. VANHEES, Physical activity and low back pain: a systematic review of recent literature. Eur Spine J, 20, 6, p. 826-845, 2011. HENSCHKE, N., C. G. MAHER, K. M. REFSHAUGE, A. DAS and J. H. MCAULEY, Low back pain research priorities: a survey of primary care practitioners. BMC Fam Pract, 8, p. 40, 2007. HOOGENDOORN, W. E., M. N. VAN POPPEL, P. M. BONGERS, B. W. KOES and L. M. BOUTER, Systematic review of psychosocial factors at work and private life as risk factors for back pain. Spine (Phila Pa 1976), 25, 16, p. 2114-2125, 2000. HOY, D., P. BROOKS, F. BLYTH and R. BUCHBINDER, The Epidemiology of low back pain. Best Pract Res Clin Rheumatol, 24, 6, p. 769-781, 2010. 203

JARVIK, J. J., W. HOLLINGWORTH, P. HEAGERTY, D. R. HAYNOR and R. A. DEYO, The Longitudinal Assessment of Imaging and Disability of the Back (LAIDBack) Study: baseline data. Spine (Phila Pa 1976), 26, 10, p. 1158-1166, 2001. JENSEN, T. S., J. KARPPINEN, J. S. SORENSEN, J. NIINIMAKI and C. LEBOEUF-YDE, Vertebral endplate signal changes (Modic change): a systematic literature review of prevalence and association with non-specific low back pain. Eur Spine J, 17, 11, p. 1407- 1422, 2008. KERR, M. S., J. W. FRANK, H. S. SHANNON, R. W. NORMAN, R. P. WELLS, W. P. NEUMANN and C. BOMBARDIER, Biomechanical and psychosocial risk factors for low back pain at work. Am J Public Health, 91, 7, p. 1069-1075, 2001. KOPEC, J. A., E. C. SAYRE and J. M. ESDAILE, Predictors of back pain in a general population cohort. Spine (Phila Pa 1976), 29, 1, p. 70-77; discussion 77-78, 2004. LANG, J., E. OCHSMANN, T. KRAUS and J. W. LANG, Psychosocial work stressors as antecedents of musculoskeletal problems: a systematic review and meta-analysis of stability- adjusted longitudinal studies. Soc Sci Med, 75, 7, p. 1163-1174, 2012. LINTON, S. J., Occupational psychological factors increase the risk for back pain: a systematic review. J Occup Rehabil, 11, 1, p. 53-66, 2001. MACDONALD, L. A., R. A. KARASEK, L. PUNNETT and T. SCHARF, Covariation between workplace physical and psychosocial stressors: evidence and implications for occupational health research and prevention. Ergonomics, 44, 7, p. 696-718, 2001. MODIC, M. T. and J. S. ROSS, Lumbar degenerative disk disease. Radiology, 245, 1, p. 43- 61, 2007. ROTHWELL, P. M., Treating individuals 2. Subgroup analysis in randomised controlled trials: importance, indications, and interpretation. Lancet, 365, 9454, p. 176-186, 2005. RUBIN, D. I., Epidemiology and risk factors for spine pain. Neurol Clin, 25, 2, p. 353-371, 2007. SHIRI, R., J. KARPPINEN, P. LEINO-ARJAS, S. SOLOVIEVA and E. VIIKARI- JUNTURA, The association between smoking and low back pain: a meta-analysis. Am J Med, 123, 1, p. 87 e87-35, 2010. SPAAR, A., M. FREY, A. TURK, W. KARRER and M. A. PUHAN, Recruitment barriers in a randomized controlled trial from the physicians' perspective: a postal survey. BMC Med Res Methodol, 9, p. 14, 2009. 204

TAYLOR, J. B., A. P. GOODE, S. Z. GEORGE and C. E. COOK, Incidence and risk factors for first-time incident low back pain: a systematic review and meta-analysis. Spine J, p. 2014. WANG, Y., T. VIDEMAN and M. C. BATTIE, ISSLS prize winner: Lumbar vertebral endplate lesions: associations with disc degeneration and back pain history. Spine (Phila Pa 1976), 37, 17, p. 1490-1496, 2012. WILLIAMS, C. M., C. G. MAHER, M. J. HANCOCK, J. H. MCAULEY, C. W. LIN and J. LATIMER, Recruitment rate for a clinical trial was associated with particular operational procedures and clinician characteristics. J Clin Epidemiol, 67, 2, p. 169-175, 2014.

205

APPENDIX

206

Appendix A: Media coverage of Chapter Five

Television:

1. Channel 9 News: http://buff.ly/U5R25m

2. ABC News:

Broadcasts:

1. 2UE

2. 2UE 954 News Talk

3. 6PR Perth

4. 4BC Brisbane

Online:

1. Wiley: http://au.wiley.com/WileyCDA/PressRelease/pressReleaseId-111044.html

2. News.com.au: http://www.news.com.au/national/breaking-news/back-pain-dont-blame-it- on-the-rain/story-e6frfku9-1226984395748

3. Life hacker Australia: http://www.lifehacker.com.au/2014/07/why-your-back-pain-has-

nothing-to-do-with-the-weather/

4. Business insider Australia: http://www.businessinsider.com.au/low-back-pain-dont-blame- the-weather-2014-7

5. The Washington Post: http://www.washingtonpost.com/news/to-your- health/wp/2014/07/10/no-uncle-fred-the-weather-has-nothing-to-do-with-your-back-pain/ 207

6. Channel 9 News: http://news.ninemsn.com.au/health/2014/07/10/14/08/back-pain-don-t-

blame-it-on-the-rain

7. Live Science: http://www.livescience.com/46740-back-pain-not-linked-weather.html

8. Vancouver Desi: http://www.vancouverdesi.com/lifestyle/dont-curse-weather-for-low-

back-pain/768630/

9. News Medical: http://www.news-medical.net/news/20140710/Acute-episodes-of-low-

back-pain-not-linked-to-weather-conditions.aspx

10. Northern Voices Online: http://nvonews.com/dont-curse-weather-for-low-back-pain/

11.TVNZ One News: http://tvnz.co.nz/world-news/back-pain-don-t-blame-rain-6024295

12. Science Codex: http://www.sciencecodex.com/low_back_pain_dont_blame_the_weather-

137261

13. Medical Daily: http://www.medicaldaily.com/lower-back-pain-not-made-worse-

inclement-weather-such-humidity-or-cold-292130

14. E Newspaper of India: http://www.eni.network24.co/lifestyle/dont-curse-weather-for-

low-back-pain-12574_13

15. EurekAletr: http://www.eurekalert.org/pub_releases/2014-07/w-lbp070814.php

16.MedicalXpress: http://medicalxpress.com/news/2014-07-pain-dont-blame-weather.html

17. Daily Mail: http://www.dailymail.co.uk/wires/aap/article-2687005/Back-pain-dont-

blame-rain.html

18. Yahoo News: http://news.yahoo.com/feel-bones-back-pain-not-linked-weather-

071628526.html 208

19. The Australian: http://www.theaustralian.com.au/news/latest-news/back-pain-dont- blame-it-on-the-rain/story-fn3dxiwe-

1226984395748?nk=179f05082c5bbe67589cada8dbfdbe1d

20. TechieTonics: http://www.techietonics.com/health-tonics/back-pain-does-not-link-to-the- weather-conditions-posture-is-to-be-blamed.html

21. Apple Balla: http://www.appleballa.com/2014/07/154959/dont-curse-weather-low-back- pain

22. Nature World News: http://www.natureworldnews.com/articles/7988/20140710/lower- back-pain-related-weather-study.htm

23. Web MD: http://www.webmd.boots.com/back-pain/news/20140710/weather-lower-back- pain

24. Daijiworld.com: http://www.daijiworld.com/news/news_disp.asp?n_id=247590

25. Science World Report: http://www.scienceworldreport.com/articles/15929/20140710/low-back-pain-not-linked-to- weather-conditions.htm

26. Health Medicine Network: http://healthmedicinet.com/i/low-back-pain-dont-blame-the- weather/

27. Business Standard: http://www.business-standard.com/article/news-ians/don-t-curse- weather-for-low-back-pain-114071000586_1.html

28. Best Health: http://besthealth.bmj.com/x/news/758164/news- item.html?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+bestheal th%2Fnews+(Best+Health%3A+Latest+news) 209

29. Topix: http://www.topix.com/forum/health/back-pain/TTCM5SSIBICJR3OTE

30. The times of India: http://timesofindia.indiatimes.com/life-style/health-

fitness/health/Dont-curse-weather-for-low-back-pain/articleshow/38136126.cms

31. University Herald: http://www.universityherald.com/articles/10332/20140710/lower-

back-pain-weather-conditions-sydney-australia.htm

32. Huffington Post US: http://www.huffingtonpost.com/2014/07/10/weather-low-back-

pain_n_5573968.html

33. National Pain Report: http://americannewsreport.com/nationalpainreport/researchers-say-

weather-not-linked-to-back-pain-8824219.html

34. Daily RX: http://www.dailyrx.com/low-back-pain-onset-not-tied-weather-factors-

humidity-and-temperature

35. Upstart Magazine: http://www.upstartmagazine.com/weather-doesnt-impact-lower-back- pain-says-study/295012/

36. NVO News: http://nvonews.com/weather-doesnt-cause-lower-back-pain/

37. Daily Mail Australia: http://www.dailymail.co.uk/health/article-2687631/Dont-listen- granny-weather-NO-impact-state-bad-back.html

38. KSBY 6: http://www.ksby.com/news/study-finds-weather-does-not-affect-back-pain/

39. Bayou Buzz: http://www.bayoubuzz.com/healthcare/healthcare-news/item/702916-

medical-news-today-back-pain-not-brought-on-by-weather-except-for-trivial-wind-effect

40. KYTX 19: http://www.cbs19.tv/story/25984211/health-alert-back-and-neck-pain-not- linked-to-weather 210

41. My Foxny: http://www.myfoxny.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

42. 13 ABC: http://www.13abc.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

43. Daily Digest News: http://dailydigestnews.com/2014/07/researchers-stop-blaming-the-

weather-for-your-lower-back-pain/

44. The Daily Express: http://www.express.co.uk/life-style/health/487969/Research-shows-

no-link-between-cold-weather-and-back-pain

45. Philly.com:

http://www.philly.com/philly/health/HealthDay689584_20140710_Don_t_Blame_Bad_Weat

he for_Your_Aching_Back.html

46. Fox News: http://www.foxnews.com/health/2014/07/10/feel-it-in-your-bones-back-pain-

not-linked-with-weather/

47. Design & Trend: http://www.designntrend.com/articles/16542/20140710/weather-does-

not-affect-lower-back-pain-latest-study-suggests.htm

48. Headlines & Global News: http://www.hngn.com/articles/35733/20140710/weather-

conditions-increase-low-back-pain-study.htm

49. Counsel & Heal: http://www.counselheal.com/articles/10419/20140710/weather-patterns- not-tied-to-back-pain.htm

50. CBS Atlanta: http://atlanta.cbslocal.com/2014/07/10/study-low-back-pain-not-linked-to-

weather/

51. TIME: http://time.com/2970789/achy-back-dont-blame-the-weather/ 211

52. Science Recorder: http://www.sciencerecorder.com/news/lower-back-pain-dont-blame-

the-weather/

53. Medical News Today: http://www.medicalnewstoday.com/articles/279435.php

54. Los Angeles Times: http://www.latimes.com/science/sciencenow/la-sci-sn-back-pain-

bad-weather-20140710-story.html

55. Red Orbit: http://www.redorbit.com/news/health/1113188840/low-back-pain-not-caused-

by-the-weather-071014/

56. Tech Times: http://www.techtimes.com/articles/10146/20140710/achy-back-due-bad-

weather.htm

57. IndiLeak: http://www.indileak.com/dont-curse-weather-for-low-back-pain/

58. Nelms Pharmacy: https://nelmspharmacy.com/article.php?id=689584

59. Health Magazine: http://news.health.com/2014/07/10/dont-blame-bad-weather-for-your-

aching-back/

60. Science Daily: http://www.sciencedaily.com/releases/2014/07/140710081200.htm

61. Arthritis Research UK: http://www.arthritisresearchuk.org/news/general- news/2014/july/weather-conditions-do-not-affect-low-back-pain.aspx

62. Mental Help:

http://www.mentalhelp.net/poc/view_doc.php?type=news&id=165586&cn=72

63. Smile-on News: http://www.smile-onnews.com/news/view/weather-not-to-blame-for-

low-back-pain

64. India Today: http://indiatoday.intoday.in/story/low-back-pain-monsoon/1/370700.html 212

65. CVBT: http://www.centralvalleybusinesstimes.com/stories/001/?ID=26265

66. Kuam News: http://www.kuam.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

67. People Say About: http://peoplesayabout.com/dont-blame-bad-weather-for-your-aching-

back/

68. News Hub:

http://au.newshub.org/feel_it_in_your_bones_back_pain_not_linked_with_weather_1926696.

html

69. News 10 ABC: http://www.news10.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

70. NEWS 724 : http://news724.com/feel-it-in-your-bones-back-pain-not-linked-with-

weather/

71. WTVM 9: http://www.wtvm.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

72. Immunology: http://immunologynews.blogspot.com.au/2014/07/lower-back-pain-not-

triggered-by.html

73. Economy Lead: http://www.economylead.com/lifestyle/back-pain-linked-weather-22667

74. Isupon: http://isupon.com/dont-curse-weather-for-low-back-pain/

75. Celebrities Snitch: http://celebritiessnitch.com/bad-back-dont-blame-the-rain-research-

shows-no-link-between-cold-weather-and-back-pain/

76. VEOOZ: http://www.veooz.com/news/wHJyc4h.html 213

77. 6 Minutes: http://www.6minutes.com.au/news/latest-news/weather-link-to-back-pain-a- lot-of-hot-air

78. RTT NEWS: http://www.6minutes.com.au/news/latest-news/weather-link-to-back-pain-a- lot-of-hot-air

79. News Ledge: http://www.newsledge.com/back-pain-blues-study-refutes-weather-link-

7672

80. Health Central: http://www.healthcentral.com/dailydose/cf/2014/07/10/study_finds_lower_back_pain_not_tie d_to_weather

81. Betty Hardwick Center: http://www.bhcmhmr.org/poc/view_doc.php?type=news&id=165586&cn=72

82. Regular News Update: http://regularnewsupdate.com/?p=48887

83. 6 WLNS: http://www.wlns.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

84. Delhi Daily News: http://www.delhidailynews.com/news/Weather-does-not-cause-back- pain-1405011439/

85. Biocompare: http://www.biocompare.com/Life-Science-News/165005-Low-Back-Pain-

Don-t-Blame-The-Weather/

86. Breaking News: http://palashbd.com/tag/says-study/

87. Fox 23: http://www.myfoxmaine.com/story/25985267/dont-blame-bad-weather-for-your- aching-back 214

88. Chicago Tribune: http://www.chicagotribune.com/health/la-sci-sn-back-pain-bad- weather-20140710,0,4484041.story

89. Silo Breaker: http://news.silobreaker.com/bad-weather-is-not-affecting-your-back-pain--

but-high-winds-might-5_2268078554707132465

90. Khaleej Times: http://www.khaleejtimes.com/kt-article-display-

1.asp?section=health&xfile=/data/health/2014/July/health_July10.xml

91. Rheumatology: http://www.rheumatologyupdate.com.au/latest-news/weather-link-to-

back-pain-a-lot-of-hot-air

92. Z News: http://zeenews.india.com/news/health/fitness/don-t-curse-weather-for-low-back-

pain_28709.html

93. Ethiopia News Hub:

http://et.newshub.org/don_t_curse_weather_for_low_back_pain_1938312.html

94. Science News Line: http://www.sciencenewsline.com/summary/2014071011150015.html

95. Health Living: http://healthylng.blogspot.com.au/2014/07/don-blame-bad-weather-for-

your-aching.html

96. Health Day – News for Healthier living: http://consumer.healthday.com/bone-and-joint- information-4/backache-news-53/don-t-blame-bad-weather-for-your-aching-back-

689584.html

97. News Max – Health: http://www.newsmaxhealth.com/Health-News/backache-pain-

symptoms-weather/2014/07/10/id/581893/

98. NBC News: http://www.nbcnews.com/id/55615058 215

99. The Advertiser – Adelaide: http://www.adelaidenow.com.au/news/breaking-news/back-

pain-dont-blame-it-on-the-rain/story-fni6ul2m-

1226984395748?nk=f1339b0422d39b4ed34178ee98d4198a

100. Deccan Herald: http://www.deccanherald.com/content/418918/dont-curse-weather-low- back.html

101. Fox 5 – Las Vegas: http://www.fox5vegas.com/story/25985267/dont-blame-bad-

weather-for-your-aching-back

102. Yotta Fire: http://yottafire.com/2014/07/does-weather-affect-back-pain-no-new-study-

finds/

103. ANI News: http://www.aninews.in/newsdetail9/story175283/weather-doesn-039-t-

cause-low-back-pain-say-scientists.html

104. The Morung Express: http://www.morungexpress.com/health/118511.html

105. The Free Press Journal: http://freepressjournal.in/dont-curse-weather-for-low-back-pain/

106. The University of Sydney:

http://sydney.edu.au/news/84.html?newscategoryid=1&newsstoryid=13764

107. 12 WBoy: http://www.wboy.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

108. Perth Now: http://www.perthnow.com.au/news/breaking-news/back-pain-dont-blame-it-

on-the-rain/story-fnhrvfuw-1226984395748?nk=f1339b0422d39b4ed34178ee98d4198a

109. Herald Sun Melbourne: http://www.heraldsun.com.au/news/breaking-news/back-pain-

dont-blame-it-on-the-rain/story-fni0xqi4-

1226984395748?nk=0f0749c3e80365b2b3cc4dc0e5f26705 216

110. Medicine Net: http://www.medicinenet.com/script/main/art.asp?articlekey=179416

111. Bright Surf: http://www.brightsurf.com/news/headlines/98424/Low_back_pain_Dont_blame_the_weather

.html

112. New Kerala: http://www.24dunia.com/english-news/shownews/8/Don-t-curse-weather- for-low-back-pain/19150408.html

113. Armenian Medical Network: http://www.health.am/ab/more/low-back-pain-dont-blame/

114. Eye Witness News: http://www.wfsb.com/story/25985267/dont-blame-bad-weather-for- your-aching-back

115. Womens Health: http://www.womenshealth.gov/news/healthday/en/2014/jul/10/689584.html

116. The Siasat Daily: http://www.siasat.com/english/news/dont-curse-weather-low-back- pain

117. ABC News 4: http://www.abcnews4.com/story/25985267/dont-blame-bad-weather-for- your-aching-back

118. KHQ: http://www.khq.com/story/25985267/dont-blame-bad-weather-for-your-aching- back

119. InteliHealth: http://www.intelihealth.com/news/dont-blame-bad-weather-for-your- aching-back?level=0

120. KMPH Fox 26: http://www.kmph.com/story/25985267/dont-blame-bad-weather-for- your-aching-back 217

121. MedBroadcast: http://www.medbroadcast.com/health_news_details.asp?news_id=31673&news_src=1&news

_channel_id=1007#.U7-TsvmSx8E

122. My Fox Houston: http://www.myfoxhouston.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

123. Mangalorean: http://www.mangalorean.com/news.php?newstype=local&newsid=494792

124. The Daily Telegraph: http://www.dailytelegraph.com.au/news/breaking-news/back-pain- dont-blame-it-on-the-rain/story-fni0xqi3-

1226984395748?nk=179f05082c5bbe67589cada8dbfdbe1d

125. Geelong Advertiser: http://www.geelongadvertiser.com.au/news/national/back-pain- dont-blame-it-on-the-rain/story-fnjbnvyf-1226984395748

126. Spire Healthcare: http://www.spirehealthcare.com/patient-information/health- news/orthopaedic-surgery/801734511-low-back-pain-cannot-be-linked-to-weather- conditions-/

127. Health Canal: http://www.healthcanal.com/disorders-conditions/pain/52954-low-back- pain-don-t-blame-the-weather.html

128. Wonder woman: http://wonderwoman.intoday.in/story/low-back-pain- monsoon/1/111692.html

129. Terra Daily: http://www.terradaily.com/reports/Low_back_pain_Dont_blame_the_weather_999.html 218

130. Fox 42: http://www.fox42kptm.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

131. Fox 2127: http://www.fox2127.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

132. People’s Media: http://peoplesmedia24.com/dont-blame-bad-weather-for-your-aching-

back/

133. The George Institute: http://www.georgeinstitute.org.au/media-releases/rain-and-back-

pain-not-

linked?utm_content=buffer4669e&utm_medium=social&utm_source=facebook.com&utm_c

ampaign=buffer

134. Harvard Medical School: http://www.health.harvard.edu/blog/bad-weather-isnt-blame-

aching-back-201407117262

135. Guelph Mercury: http://www.guelphmercury.com/living-story/4627655-study-finds-no- link-between-weather-and-lower-back-pain/

136. TV 3: http://www.tv3.ie/entertainment_article.php?locID=1.803.1098&article=139873

137. Montlhly Prescribing Reference: http://www.empr.com/back-pain-dont-blame-the-

weather/article/360458/

138. Health Professionals Network: http://www.hcplive.com/articles/Weather-Conditions-

Not-Associated-With-Lower-Back-Pain

139. Doctors Lounge: http://www.doctorslounge.com/index.php/news/pb/48085

140. The Record: http://www.therecord.com/living-story/4627655-study-finds-no-link-

between-weather-and-lower-back-pain/ 219

141. The Baltimore Sun: http://www.baltimoresun.com/health/la-sci-sn-back-pain-bad-

weather-20140710,0,2527158.story

142. 8 News Now: http://www.8newsnow.com/story/25985267/dont-blame-bad-weather-for-

your-aching-

back?utm_source=feedburner&utm_medium=feed&utm_campaign=Feed%3A+lasvegasnow

%2Fhealth+(8NewsNOW.com+-+Health+News)

143. Start at 60: http://www.startsatsixty.com.au/health/stop-blaming-it-on-the-weather

144. Your News Now: http://www.hometownstations.com/story/25985267/dont-blame-bad-

weather-for-your-aching-back

145. WMBB Northwest Florida: http://www.wmbb.com/story/25985267/dont-blame-bad-

weather-for-your-aching-back

146. GTN News: http://www.mygtn.tv/story/25985267/dont-blame-bad-weather-for-your-

aching-back

147. Central Coast News: http://health.keyt.com/story/25985267/dont-blame-bad-weather-

for-your-aching-back

148. Health e Galaxy: http://www.hegalaxy.com/weather-conditions-not-associated-with-

low-back-pain/

149. The Hamilton Spectator: http://www.thespec.com/living-story/4627655-study-finds-no- link-between-weather-and-lower-back-pain/

150. Australian News: http://www.australiannews.net/index.php/sid/223696183/scat/88f7d0d02bea1b33/ht/Weather-

doesnt-cause-low-back-pain-say-scientists 220

151. 7 News Denver: http://www.thedenverchannel.com/news/study-finds-no-link-between-

certain-weather-conditions-lower-back-pain

152. Record Search Light: http://www.redding.com/news/national/study-finds-no-link-

between-certain-weather-conditions-lower-back-pain

153. Health News Digest: http://www.healthnewsdigest.com/news/weather0/Low-Back-Pain-

Don-t-Blame-the-Weather_printer.shtml

154. WBKO: http://wn.wbko.com/story/25985267/dont-blame-bad-weather-for-your-aching-

back

155. NBC 40: http://www.nbc40.net/story/25985267/dont-blame-bad-weather-for-your-

aching-back

156. 760 KFMB: http://www.760kfmb.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

157. Action 3 News: http://health.kmtv.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

158. Bio-Medicine: http://www.bio-medicine.org/medicine-news-1/Low-back-pain-3F-Dont-

blame-the-weather-127490-1/

159. Israel Foreign Affairs News: http://israelforeignaffairs.com/dont-attribute-inclemency-

for-the-aching-back/

160. Live 5 News: http://www.live5news.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

161. Tulsa’s Channel: http://www.ktul.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back 221

162. Walb News: http://www.walb.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

163. Kusi News: http://www.kusi.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

164. Drugs.com: http://www.drugs.com/news/don-t-blame-bad-weather-your-aching-back-

52308.html

165. Medline Plus: http://www.nlm.nih.gov/medlineplus/news/fullstory_147239.html

166. Wrex: http://www.wrex.com/story/25985267/dont-blame-bad-weather-for-your-aching-

back

167. Yahoo Health: http://health.yahoo.net/news/s/hsn/don-t-blame-bad-weather-for-your-

aching-back

168. Healthy Living: http://healthyliving.msn.com/diseases/back-pain/dont-blame-bad-

weather-for-your-aching-back

169. Weekly Times Now: http://www.weeklytimesnow.com.au/news/national/back-pain-

dont-blame-it-on-the-rain/story-fnjbnvyg-1226984395748

170. NBC Right Now: http://www.nbcrightnow.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

171. News Net 5: http://www.newsnet5.com/news/national/study-finds-no-link-between-

certain-weather-conditions-lower-back-pain

172. WAFB: http://www.wafb.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back 222

173. KSFY: http://www.ksfy.com/story/25985267/dont-blame-bad-weather-for-your-aching-

back

174. WAFF: http://www.waff.com/story/25985267/dont-blame-bad-weather-for-your-aching-

back

175. The Courier Mail Brisbane: http://www.couriermail.com.au/news/breaking-news/back-

pain-dont-blame-it-on-the-rain/story-fnihsfrf-

1226984395748?nk=179f05082c5bbe67589cada8dbfdbe1d

176. Black Christian News: http://www.blackchristiannews.com/2014/07/if-you-have-pain-

in-your-lower-back-dont-blame-the-weather/

177. KSLA News: http://www.ksla.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

178. News West 9: http://www.newswest9.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

179. Healing Well: http://news.healingwell.com/index.php?p=news1&id=689584

180. KTRE: http://www.ktre.com/story/25985267/dont-blame-bad-weather-for-your-aching-

back

181. My Fox Orlando: http://www.myfoxorlando.com/story/25985267/dont-blame-bad-

weather-for-your-aching-back

182. WCAX: http://www.wcax.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

183. Sify News: http://www.sify.com/news/weather-doesn-t-cause-low-back-pain-say-

scientists-news-news-ohlmaBahdgccf.html 223

184. Fox Oregon: http://www.kptv.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

185. KXXV: http://www.kxxv.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

186. WPSD Local: http://www.wpsdlocal6.com/story/25985267/dont-blame-bad-weather- for-your-aching-back

187. My Fox Los Angeles: http://www.myfoxla.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

188. Wnem: http://www.wnem.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

189. KEYC TV Mankato: http://www.keyc.com/story/25985267/dont-blame-bad-weather- for-your-aching-back

190. WHLT 22: http://www.whlt.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

191. KATV: http://www.katv.com/story/25985267/dont-blame-bad-weather-for-your-aching- back

192. CBS 5 AZ: http://www.kpho.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

193. KTIV: http://www.ktiv.com/story/25985267/dont-blame-bad-weather-for-your-aching- back

194. WAOW: http://www.waow.com/story/25985267/dont-blame-bad-weather-for-your- aching-back 224

195. Oklahoma’s Own: http://www.newson6.com/story/25985267/dont-blame-bad-weather- for-your-aching-back

196. Kotatv: http://www.kotatv.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

197. Valley News Live: http://www.valleynewslive.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

198. My Fox Memphis: http://www.myfoxmemphis.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

199. WXYZ Detroit: http://www.wxyz.com/news/national/study-finds-no-link-between- certain-weather-conditions-lower-back-pain

200. Independent Mail: http://www.independentmail.com/news/national/study-finds-no-link- between-certain-weather-conditions-lower-back-pain

201. Irish Health: http://www.irishhealth.com/article.html?id=23849

202. The Indy Channel: http://www.theindychannel.com/news/study-finds-no-link-between- certain-weather-conditions-lower-back-pain

203. WKRG News: http://ww2.wkrg.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

204. ABC 6: http://www.abc6.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

205. News Plex: http://wn.newsplex.com/story/25985267/dont-blame-bad-weather-for-your- aching-back 225

206. My 13 LA: http://www.my13la.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

207. Be Live: http://phucanpc.com/8120/dont-blame-bad-weather-for-your-aching-back/

208. Lycos News: http://news.lycos.com/entertainment/weather-not-related-to-back-ache-

9ab8817dbfd88f499f11b5e88f498903/

209. WIFR Rockford: http://wn.wifr.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

210. KKTV Sothern Colorado: http://wn.kktv.com/story/25985267/dont-blame-bad-weather- for-your-aching-back

211. Capital Gazette: http://www.capitalgazette.com/parade/health/does-weather-cause- aches/article_3ced89f1-7c0c-5259-817b-a13c0e94cdb8.html

212. Cinema Blend: http://www.cinemablend.com/pop/Scientists-Claim-Link-Between-Cold-

Weather-Back-Pain-Nonexistent-65185.html

213. Fox 16: http://www.fox16.com/story/d/story/back-pain-not-caused-by-the- weather/30094/RvNSiTKnf06Hlp3Tkkr6MA

214. FM News Talk: http://www.971talk.com/news/health/weather-doesnt-cause-aggravate- back-pain

215. Value Based Care in Rheumatology: http://www.valuebasedrheumatology.com/vbcr- news/news-feed/1172-back-pain-not-brought-on-by-weather-except-for-trivial-wind-effect

216. Joint Pain 101: http://jointpain101.com/achy-back-pain-isnt-due-to-bad-weather-latest- study-finds-tech-times/ 226

217. India News Hub: http://indianewshub.com/weather-cause-back-pain/

218. Faraday’s Natural Foods and Supplements: http://www.faradaysnaturalfoods.com/common/news/news_results.asp?task=Headline&id=1

5368&StoreID=EADB2AA619684A9AB9AF9A7D0A20FF81

219. Wild by Nature: http://www.wildbynature.com/common/news/news_results.asp?task=Headline&id=15368&S toreID=D272A3B93180420D908E136E9D7E775D

220. Lovelock Pharmacy: https://lovelockpharmacy.com/article.php?id=689584

221. My Center Pharmacy: https://mycenterpharmacy.com/article.php?id=689584

222. US News: http://health.usnews.com/health-news/articles/2014/07/10/dont-blame-bad- weather-for-your-aching-back

223. Wate: http://www.wate.com/story/25985267/dont-blame-bad-weather-for-your-aching- back

224. Clinical Research: http://www.clinicalresearch.com/NewsDetail.aspx?id=689584

225. WMBF News: http://www.wmbfnews.com/story/25985267/dont-blame-bad-weather- for-your-aching-back

226. Jose Marcos – Doencas Reumaticas: http://doencasreumat.blogspot.com.au/2014_07_11_archive.html

227. 1 Click News: http://1clicknews.com/is-the-weather-to-blame-for-lower-back-pain/

228. The Malay Mail Online: http://m.themalaymailonline.com/features/article/dont-blame- the-weather-for-lower-back-pain-study-says 227

229. Gulf Bend Center:

http://www.gulfbend.org/poc/view_doc.php?type=news&id=165586&cn=72

230. Priyo News: http://news.priyo.com/2014/07/13/dont-curse-weather-low-back-pain-

113952.html

231. Bakersfield Now: http://wn.bakersfieldnow.com/story/25985267/dont-blame-bad-

weather-for-your-aching-back

232. 0 Hag: http://www.0hag.com/dont-listen-granny-weather-impact-state-bad-back/

233. WLTZ First News: http://www.wltz.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

234. Daily Health Headlines: http://www.dailyhealthheadlines.com/article/health-

headlines/feel-it-in-your-bones-back-pain-not-linked-with-weather

235. The Health Site: http://www.thehealthsite.com/news/weather-to-blame-for-lower-back-

pain/

236. CTV News: http://www.ctvnews.ca/health/is-the-weather-to-blame-for-lower-back-pain-

1.1911889

237. Astro Awani: http://english.astroawani.com/news/show/is-the-weather-to-be-blamed-

for-lower-back-pain-39770

238. Free Malaysia Today:

http://www.freemalaysiatoday.com/category/leisure/2014/07/14/is-the-weather-to-blame-for-

lower-back-pain/

239. Malaysian Digest: http://malaysiandigest.com/features/509026-don-t-blame-the- weather-for-lower-back-pain-study-says.html 228

240. International Business Times: http://au.ibtimes.com/articles/559112/20140714/back-

pain-backache-weather-world-health-association.htm#.U8OTEPmSx8E

241. Iafrica: http://lifestyle.iafrica.com/wellness/949058.html

242. The New Age: http://thenewage.co.za/131503-12-53-

Is_the_weather_to_blame_for_lower_back_pain

243. The Roger Hedgecock Show: http://www.rogerhedgecock.com/story/25985267/dont- blame-bad-weather-for-your-aching-back

244. Reuters UK: http://uk.reuters.com/article/2014/07/14/us-weather-back-pain-

idUKKBN0FJ1SI20140714

245. Huffington Post Canada: http://www.huffingtonpost.ca/2014/07/14/back-pain-causes-

weather_n_5584538.html

246. Ciencias Medicas News: http://elbiruniblogspotcom.blogspot.com.au/2014/07/dont-

blame-bad-weather-for-your-aching.html

247. ABC 40 KRHD: http://www.abc40.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

248. WITN: http://wn.witn.com/story/25985267/dont-blame-bad-weather-for-your-aching-

back

249. WBTV: http://www.wbtv.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

250. Hon News: http://www.hon.ch/News/HSN/689584.html 229

251. WOWKRV: http://www.wowktv.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

252. WDAM: http://www.wdam.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

253. Newsday: http://www.newsday.com/news/health/don-t-blame-bad-weather-for-your-

aching-back-1.8755826

254. 14 News: http://www.14news.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

255. WSAV: http://www.wsav.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

256. KPLCTV: http://www.kplctv.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

257. WLOX 13: http://www.wlox.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

258. WGEM: http://www.wgem.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

259. My Fox Nepa: http://www.myfoxnepa.com/story/25985267/dont-blame-bad-weather-

for-your-aching-back

260. WDRB: http://www.wdrb.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

261. WSPA 7: http://www.wspa.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back 230

262. Winnipeg Free Press: http://www.winnipegfreepress.com/arts-and-life/life/health/dont- blame-bad-weather-for-your-aching-back-266581831.html

263. My Fox Wausau: http://www.myfoxwausau.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

264. KCBD: http://www.kcbd.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

265. News Channel 12: http://www.wjtv.com/story/25985267/dont-blame-bad-weather-for- your-aching-back

266. News R India: http://newsr.in/n/Health/750jc4162/Weather-to-blame-for-lower-back- pain.htm

267. WKOW Madison: http://www.wkow.com/story/25985267/dont-blame-bad-weather-for- your-aching-back

268. Herald Whig: http://www.whig.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

269. West Virginia Illustrated: http://www.wvillustrated.com/story/25985267/dont-blame- bad-weather-for-your-aching-back

270. The Mercury: http://www.themercury.com.au/news/breaking-news/back-pain-dont- blame-it-on-the-rain/story-fnj6ehgr-1226984395748

271. WRCB Chattanooga: http://www.wrcbtv.com/story/25985267/dont-blame-bad-weather- for-your-aching-back

272. Hawaii News Now: http://www.hawaiinewsnow.com/story/25985267/dont-blame-bad- weather-for-your-aching-back 231

273. KWWL: http://www.kwwl.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

274. ABC Kait 8: http://www.kait8.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

275. My Fox Phoenix: http://www.myfoxphoenix.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

276. My Fox DC: http://www.myfoxdc.com/story/25985267/dont-blame-bad-weather-for- your-aching-back

277. WTHR Indiana’s News Leader: http://www.wthr.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

278. My Fox Boston: http://www.myfoxboston.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

279. CBS 8: http://www.cbs8.com/story/25985267/dont-blame-bad-weather-for-your-aching- back

280. My Fox Dallas-Fort Worth: http://www.myfoxdfw.com/story/25985267/dont-blame- bad-weather-for-your-aching-back

281. WSFA 12: http://www.wsfa.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

282. My Fox Tampa Bay: http://www.myfoxtampabay.com/story/25985267/dont-blame-bad- weather-for-your-aching-back

283. WSET-TV Lynchburg Danville Roanoke: http://www.wset.com/story/25985267/dont- blame-bad-weather-for-your-aching-back 232

284. WKRN-TV Nashville: http://www.wkrn.com/story/25985267/dont-blame-bad-weather-

for-your-aching-back

285. Fox 51 WOGX: http://www.wogx.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

286. News Channel 5: http://www.newschannel5.com/story/25985267/dont-blame-bad-

weather-for-your-aching-back

287. Fox 29 WFLX: http://www.wflx.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

288. Yahoo! News India: https://in.news.yahoo.com/dont-curse-weather-low-back-pain-

090004253.html

289. 9&10 Northern Michigan’s News Leader:

http://www.9and10news.com/story/25985267/dont-blame-bad-weather-for-your-aching-back

290. The Courant: http://www.courant.com/health/sns-rt-us-weather-back-pain-

20140714,0,296648.story

291. KTBS 3: http://www.ktbs.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

292. Townsville Bulletin: http://www.townsvillebulletin.com.au/news/breaking-news/back-

pain-dont-blame-it-on-the-rain/story-fnjbnvyi-1226984395748

293. KLTV 7: http://www.kltv.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

294. Rocket News: http://www.rocketnews.com/2014/07/feel-it-in-your-bones-back-pain-not-

linked-with-weather/ 233

295. Yahoo! News UK and Ireland: https://uk.news.yahoo.com/feel-bones-back-pain-not-

linked-weather-071628526.html?.tsrc=lgwn#iELc3Jm

296. WAVE 3 News: http://www.wave3.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

297. Gold Coast Bulletin: http://www.goldcoastbulletin.com.au/news/breaking-news/back-

pain-dont-blame-it-on-the-rain/story-fnjbnvyk-1226984395748

298. KVVU-TV Fox 5 Las Vegas: http://www.fox5vegas.com/story/25985267/dont-blame-

bad-weather-for-your-aching-back

299. Web India 123:

http://news.webindia123.com/news/Articles/India/20140711/2422516.html

300. WVVA – The two Virginias: http://www.wvva.com/story/25985267/dont-blame-bad-

weather-for-your-aching-back

301. WFMJ: http://www.wfmj.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

302. Truth Dive: http://truthdive.com/2014/07/11/Weather-doesn-t-cause-low-back-pain-say- scientists.html

303. KTVN 2 News: http://www.ktvn.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back

304. NBC 2: http://www.nbc-2.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

305. KTEN 10 Texoma: http://www.kten.com/story/25985267/dont-blame-bad-weather-for-

your-aching-back 234

306. WRBL: http://www.wrbl.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

307. WTOC 11: http://www.wtoc.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

308. KFVS 12: http://www.kfvs12.com/story/25985267/dont-blame-bad-weather-for-your- aching-back

309. Prokerala News: http://www.prokerala.com/news/articles/a472570.html

310. KSWO 7 News Lawton/ Wichita Falls: http://www.kswo.com/story/25985267/dont- blame-bad-weather-for-your-aching-back

311. ABC 2 WBAY: http://www.wbay.com/story/25985267/dont-blame-bad-weather-for- your-aching-back

312. Mizo News: http://www.mizonews.net/world/dont-curse-weather-for-low-back-pain/

313. ABC 10 News: http://www.10news.com/news/national/study-finds-no-link-between- certain-weather-conditions-lower-back-pain

314. Neuro Talk: http://neurotalk.psychcentral.com/showthread.php?s=77a5572aa77f20b8e90925adf927a025& p=1082100#post1082100

315. Latest News Link: http://latestnewslink.com/2014/07/weather-not-tied-to-back-pain- study/

316. Big News Network: http://www.bignewsnetwork.com/index.php/cat/eb75a2fd5e16e873/ 235

317. Personal Liberty Digest: http://personalliberty.com/study-finds-link-certain-weather-

conditions-lower-back-pain/

318. Cliprender: http://www.cliprender.com/clip/Is-The-Weather-To-Be-Blamed-For-Lower-

Back-Pain%3F

319. Turn to 10: http://ww2.turnto10.com/story/25985267/dont-blame-bad-weather-for-your-

aching-back

320. Focus Italy: http://www.focus.it/ADNKronos/salute-sfatato-legame-meteo-dolore-clima-

non-influenza-mal-di-schiena_C65.aspx

321. Meteo e scienze del cielo e della terra: http://www.meteoweb.eu/2014/07/salute-sfatato-

legame-meteo-dolori-clima-non-influenza-mal-schiena/298772/

322. Australian Doctor: http://www.australiandoctor.com.au/news/latest-news/weather-not-a-

barometer-for-back-pain

323. ABLX Boston: http://www.manewsday.com/national/65203-don-t-listen-to-granny-the-

weather-has-no-impact-on-the-state-of-your-bad-back.html

324. Sina Women Eladies – China: http://eladies.sina.com.tw/getnews.php?newsid=97932

325. Viet Times: http://www.viet-times.com.au/gia-dinh/suc-khoe/1300534-thoi-tiet-khong-

anh-huong-den-chung-dau-lung?device=desktop

326. Tien Phong Online: http://khoe360.tienphong.vn/gia-dinh-suc-khoe/khong-co-bang-

chung-dau-lung-do-thoi-tiet-734237.tpo

327. Path Finder:

http://www.pathfinder.gr/stories/3752169/%CF%80%CE%BF%CE%BD%CE%BF%CF%82

-%CF%83%CF%84%CE%B7-%CE%BC%CE%B5%CF%83%CE%B7- 236

%CE%BC%CE%B7%CE%BD-

%CE%B1%CE%BA%CE%BF%CF%85%CF%84%CE%B5-%CF%84%CE%B7-

%CE%B3%CE%B9%CE%B1%CE%B3%CE%B9%CE%B1-%CF%83%CE%B1%CF%82/

328. AArgauer Zeitung: http://www.aargauerzeitung.ch/leben/gesundheit/ist-das-schlechte-

wetter-schuld-am-hexenschuss-128166931

329. Anteka: http://www.apteka.ua/article/299054

330. RMF 24: http://www.rmf24.pl/nauka/news-bole-w-krzyzu-nie-przez-

pogode,nId,1467274

331. Tiscali: http://lifestyle.tiscali.it/salute/feeds/14/07/11/t_16_ADN20140711130058.html

332. TVN Meteo: http://tvnmeteo.tvn24.pl/informacje-pogoda/ciekawostki,49/bole-w-

krzyzu-to-nie-przez-pogode,127818,1,0.html

333. Sante Log: http://www.santelog.com/news/rhumatologie/arthrite-et-lombalgie-une-

question-de-meteo-_12595_lirelasuite.htm

334. Time Turk: http://www.timeturk.com/tr/2014/07/10/buyukanneler-yaniliyor-

olabilir.html#.U8TruPmSx8E

335.Wissenschaft: http://www.wissenschaft-

aktuell.de/artikel/Hexenschuss__Das_Wetter_ist_nicht_schuld1771015589598.html

336. Gazete A24: http://www.gazetea24.com/yerel-basin-haber/buyukanneler-yaniliyor-

olabilir_12387770.html

337. Aponet: http://www.aponet.de/aktuelles/kurioses/20140710-kein-wettereinfluss-auf-

rueckenschmerzen.html 237 Dr Daniel Steffens CURRICULUM Address: 4/18 Market Street | Rockdale, Sydney NSW 2216 VITAE Mobile: + 61 2 0423 786 695 | E-mail: [email protected]

EDUCATION DOCTOR OF PHILOSOPHY (2011 - 2015), The George Institute for Global Health, Sydney Medical School, The University of Sydney, Australia. Thesis title: Mechanisms of low back pain. Supervisor: Professor Chris G Maher Main projects: . Clinicians’ views on factors that trigger a sudden onset of low back pain. . Factors predicting recruitment of participants to an observational study of low back pain conducted in primary care. . Prognosis of chronic low back pain in patients presenting to a private community-based group exercise program. . Does magnetic resonance imaging predict future low back pain? A systematic review. . Triggers of an acute onset of low back pain – results of a case-crossover study. . Does the weather affect back pain? A case-crossover study. . Do MRI findings identify patients with low back pain who respond best to particular interventions? A systematic review. . Triggers for an episode of sudden onset low back pain: Study protocol.

BACHELOR OF PHYSIOTHERAPY - HONOURS (2001 - 2005), University of Santa Cruz do Sul, Brazil.

EXPERIENCE LECTURER, Macquarie University, Department of Chiropractic, Faculty of Science, Australia (ACADEMIC) (2014-present). Subject: Chiropractic 2 – CHIR 311 (Research Methods). . Designing, preparing and developing teaching materials. . Delivering lectures. . Setting and marking assignments.

EXPERIENCE CLINICAL RESEARCH ASSISTANT, Neuroscience Research Australia, University of New South (RESEARCH) Wales (2015 - present). Project: Standing Tall - A home-based balance exercise program. NHMRC-funded randomised controlled trial involving 500 participants. . Administering participants comprehensive physical assessment. . Maintaining study database. . Delivering study intervention to participants. . Developing study materials.

RESEARCH ASSISTANT, The George Institute for Global Health, The University of Sydney, Australia (2010 – 2011). Project: Paracetamol for low back pain (PACE). NHMRC-funded randomised controlled trial involving 1650 participants. . Recruited and enrolled study clinicians and participants. . Provided additional administrative support and carried out general clerical duties. . Administered participant’s interviews and prepared assessments. . Wrote research manuscript, presentations, reports and ethical submissions. . Developed and maintained research databases (Excel, FileMaker Pro). Achievement: Received a discretionary bonus for recognition and reward of exceptional performance.

Dr Daniel Steffens - Curriculum Vitae Page 1 of 4

238

EXPERIENCE RESEARCH ASSISTANT, Faculty of Health Science, The University of Sydney, Australia (2009 – (RESEARCH) 2010). Project: Prognosis of chronic back pain. One year follow-up study that involved 118 participants presenting to private care. . Performed face-to-face and telephone data collection. . Performed systematic literature searching. . Liaised with data collection sites. . Performed data entry and statistical analysis (SPSS). . Wrote protocols, ethical documents and scientific papers. . Achievement: Completion of the study on time with minimal loss to follow-up. Published study in relevant scientific journal.

SUPPORT OF Keira BEILKEN and Vicky DUONG - Does weather affect daily pain levels in patients with acute JUNIOR low back pain? A longitudinal cohort study. Honours Students , Macquarie University - 2015. RESEARCH STUDENTS Matthew STEVENS - Patients’ and clinicians’ views on triggers for low back pain. Doctor of Philosophy, The University of Sydney – 2014.

Patricia PARREIRA – Nominating triggers for low back pain patients. Doctor of Philosophy, The University of Sydney – 2014.

PUBLICATIONS DANIEL STEFFENS, Manuela L Ferreira, Jane Latimer, Bart W Koes, Fiona M Blyth, Paulo H (FULL PAPER) Ferreira, Christopher G Maher. What triggers an episode of acute low back pain? A case- crossover study. Arthritis Care & Research, 2015; 67(3):403-10.

DANIEL STEFFENS, Chris G Maher, Manuela L Ferreira, Mark J Hancock, Leani SM Pereira, Christopher M Williams, Jane Latimer. Clinician characteristics and operational factors have limited influence on participant recruitment in primary care: Results from an observational study. Journal of Manipulative and Physiological Therapeutics, 2015; xx:1-8.

DANIEL STEFFENS, Chris Maher, Qiang Li, Manuela Ferreira, Leani Pereira, Bart Koes, Jane Latimer. Weather does not affect back pain: results from a case-crossover. Arthritis Care & Research, 2014; 66(12):1867-72.

DANIEL STEFFENS, Chris Maher. Effectiveness of extracorporeal shock wave therapy in chronic plantar fasciitis. American Journal of Physical Medicine and Rehabilitation, 2014; 2:1-2.

DANIEL STEFFENS, Chris Maher, Manuela Ferreira, Mark Hancock, Timothy Glass, Jane Latimer. Clinicians’ views on factors that trigger a sudden onset of low back pain. European Spine Journal, 2014; 23(3):512-9.

DANIEL STEFFENS, Mark Hancock, Chris Maher, Jane Latimer, Rob Satchill, Manuela Ferreira, Paulo Ferreira, Melissa Partington, Anna-Louise Bouvier. Prognosis of chronic low back pain in patients presenting to a private community-based group exercise program. European Spine Journal, 2014; 23(1):113-9.

DANIEL STEFFENS, Mark Hancock, Chris G Maher, Ciaran Williams, Tue Secher Jensen, Jane Latimer. Does magnetic resonance imaging predict future low back pain? A systematic review. European Journal of Pain, 2013; 18(6):755-65.

Dr Daniel Steffens - Curriculum Vitae Page 2 of 4

239

PUBLICATIONS DANIEL STEFFENS, Paula R Beckenkamp, Mark Hancock, Dulciane Nunes Paiva, Jennifer A Alison, (FULL PAPER) Sergio S Menna-Barreto. Activity level predicts 6-minute walk distance in healthy older females: an observational study. Physiotherapy, 2013; 99(1):21-6.

DANIEL STEFFENS, Manuela L Ferreira, Christopher G Maher, Jane Latimer, Bart W Koes, Fiona M Blyth and Paulo H Ferreira. Triggers for an episode of sudden onset low back pain: study protocol. BMC Musculoskeletal Disorders, 2012; 13:7.

DANIEL STEFFENS, Chris Maher. Conflicting findings on effectiveness of low level laser therapy for tendinopathty. British Journal of Sports Medicine, 2011; 45:459.

Antonio MV Silva, Luis U Signori, Guilherme C Torres, DANIEL STEFFENS, Rodrigo DM Plentz. Neuromuscular electrical stimulation versus strength training in elderly women. Geriatria & Gerontologia, 2008; 2(1):12-16 [Portuguese].

DANIEL STEFFENS, Paula R Beckenkamp, Isabella M Albuquerque, Dulciane N Paiva, Serigio SM Barreto. Occupational exposure to tobacco dust – effects on the respiratory system. Pulmão RJ, 2007; 16:86-90 [Portuguese].

PUBLICATIONS DANIEL STEFFENS, Manuela Ferreira, Jane Latimer, Paulo H Ferreira, Bart W Koes, Fiona Blyth, (PROCEEDINGS) Qiang Li, Chris Maher. What triggers an episode of low back pain? Results of a case-crossover study. Proceedings XIII International Back Pain Forum. Campos do Jordao, Brazil, 2014. p52.

DANIEL STEFFENS, Chris Maher, Qiang Li, Manuela Ferreira, Leani Pereira, Bart W Koes, Jane Latimer. Could the weather triggers an episode of low back pain? A case-crossover study. Proceedings XIII International Back Pain Forum. Campos do Jordao, Brazil, 2014. p81.

DANIEL STEFFENS, Mark Hancock, Chris G Maher, Ciaran Williams, Tue S Jensen, Jane Latimer. Does magnetic resonance imaging predict future low back pain? A systematic review. Proceedings Pain in Europe VIII. Florence, Italy, 2013. p1117.

DANIEL STEFFENS, Manuela Ferreira, Chris Maher, Jane Latimer, Bart Koes, Fiona Blyth, Paulo Ferreira. Does the method of training of recruiting clinicians influence recruitment to a low back pain case-crossover study. Proceedings Odense International Forum XII. Odense, Denmark, 2012. p.178. DANIEL STEFFENS, Manuela Ferreira, Chris Maher, Jane Latimer, Bart Koes, Fiona Blyth, Paulo Ferreira. Clinician’s views on triggers for sudden onset low back pain. Proceedings Odense International Forum XII. Odense, Denmark, 2012. p.195.

DANIEL STEFFENS, Paula R Beckenkamp, Mark Hancock, Dulciane N Paiva, Sergio S Menna- Barreto, Jennifer A Alison. Six minute walk distance in healthy elderly active and sedentary female. Proceedings Australian Physiotherapy Association Biennial Conference 2011. Brisbane, Australia, 2011. p.126.

DANIEL STEFFENS, Mark J Hancock, Rob Satchill, Manuela Ferreira, Paulo Ferreira, Chris G Maher, Melissa Partington, Ana-Louise Bouvier. Prognosis of patients with chronic low back pain presenting to a private functional group exercise program. Proceedings Australian Physiotherapy Association Biennial Conference. Brisbane, Australia, 2011. p.146.

Dr Daniel Steffens - Curriculum Vitae Page 3 of 4

240

RESEARCH COTUTELLE AWARD (2011 – 2015), jointly awarded degree between The University of Sydney GRANTS AND (Australia) and the Federal University of Minas Gerais (Brazil). AWARDS TRAVEL GRANT (2014), Research Student Grant Scheme (RSGS), School of Public Health, Sydney Medical School, The University of Sydney (AUD$ 2602.00).

TOP BEST ABSTRACT (2013), EFIC-Pain in Europe VII Congress. Florence, Italy.

TRAVEL GRANT (2013), Research Student Grant Scheme (RSGS), School of Public Health, Sydney Medical School, The University of Sydney (AUD$ 1200.00).

TRAVEL GRANT (2012), Postgraduate Research Support Scheme (PRSS), School of Public Health, Sydney Medical School, The University of Sydney (AUD$ 1020.74).

REVIEWER JOURNAL OF PHYSIOTHERAPY (SCIENTIFIC ARTICLES) MANUAL THERAPY

WORLD CONFEDERATION FOR PHYSICAL THERAPY

BMC MUSCULOSKELETAL DISORDERS

JOURNAL OF MEDICAL INTERNET RESEARCH PROTOCOLS

REFEREES PROF. CHRIS G MAHER (PhD Supervisor) Director, Musculoskeletal Division. The George Institute for Global Health, The University of Sydney Telephone: +61 2 9657 0382 E-mail: [email protected]

A/PROF. MARK HANCOCK (Previous employer) Senior Lecturer, Musculoskeletal physiotherapy. Macquarie University Telephone: +61 2 9850 6622 E-mail: [email protected]

Dr Daniel Steffens - Curriculum Vitae Page 4 of 4