Journal of the New Zealand Medical Association Vol 132 | No 1491 | 8 March 2019 Where are we going with cancer treatment in New Zealand?

New Zealand Society of Gastroenterology Guidelines on Therapeutic Drug Monitoring in Inflammatory Bowel Disease

Developing a regional Lifespan of New Zealand Advancing transgender cancer service; lessons Second World War veterans healthcare teaching in from the Waitemata Cancer from one large cemetery: the Aotearoa/New Zealand Patient Experience Survey case for a national-level study Publication Information published by the New Zealand Medical Association

NZMJ Editor NZMA Chair Professor Frank Frizelle Dr Kate Baddock

NZMJ Production Editor NZMA Communications Manager Rory Stewart Diana Wolken

Other enquiries to: NZMA To contribute to the NZMJ, fi rst read: PO Box 156 www.nzma.org.nz/journal/contribute The Terrace Wellington 6140 © NZMA 2019 Phone: (04) 472 4741

To subscribe to the NZMJ, email [email protected]

Subscription to the New Zealand Medical Journal is free and automatic to NZMA members. Private subscription is available to institutions, to people who are not medical practitioners, and to medical practitioners who live outside New Zealand. Subscription rates are below. All access to the NZMJ is by login and password, but IP access is available to some subscribers. Read our Conditions of access for subscribers for further information www.nzma.org.nz/journal/subscribe/conditions-of-access If you are a member or a subscriber and have not yet received your login and password, or wish to receive email alerts, please email: [email protected] The NZMA also publishes the NZMJ Digest. This online magazine is sent out to members and subscribers 10 times a year and contains selected material from the NZMJ, along with all obituaries, summaries of all articles, and other NZMA and health sector news and information.

Subscription rates for 2019 New Zealand subscription rates Overseas subscription rates Individuals* $317 Individual $442 Institutions $549 Institutions $591 Individual article $30 Individual article $30

*NZ individual subscribers must not be doctors (access is via NZMA Membership) New Zealand rates include GST. No GST is included in international rates. Note, subscription for part of a year is available at pro rata rates. Please email [email protected] for more information. Individual articles are available for purchase by emailing [email protected]

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 2 www.nzma.org.nz/journal CONTENTS

EDITORIAL 63 7 Vision screening in New Zealand: Where are we going with cancer an audit of the B4 School Check treatment in New Zealand? William Muller, Logan Mitchell, Tamara Mullaney, Frank Frizelle Graham Wilson ARTICLES 71 Considering evidence for ethnicity 11 bias using assessment case Relationship between estimated scenarios and medical student glomerular ltration rate and correctness and certainty incident cardiovascular disease Mike Tweed, Gordon Purdie, in an ethnically diverse primary Cameron Lacey care cohort Emma Church, Katrina Poppe, 78 Matire Harwood, Suneela Mehta, District health board of residence, Corina Grey, Vanessa Selak, ethnicity and socioeconomic Mark R Marshall, Susan Wells status all impact publicly funded insulin pump uptake in 27 New Zealand patients with The factors that lead to a delay type 1 diabetes between general practitioner Benjamin J Wheeler, Rhiannon Braund, referral of symptomatic patients Barbara Galland, Anastasia Mikuscheva, and specialist diagnosis of Esko Wiltshire, Craig Je eries, colorectal cancer: an audit in the Michel de Lange Bay of Plenty District Health Board Ada Yee, Tara Linton, Mei-Sze Lee, CLINICAL CORRESPONDENCE Simon Raimes 90 38 Beau lines and telogen ef uvium Developing a regional cancer following aluminium phosphide service; lessons from the poisoning Waitemata Cancer Patient Samaneh Nakhaee, Nasim Zamani, Omid Mehrpour Experience Survey Jonathan Koea, Sue French, 93 Karen Hellesoe, Peter Sandiford Successful intravascular 46 lithotripsy for severely calci ed New Zealand Society of le anterior descending coronary Gastroenterology Guidelines on artery stenosis Therapeutic Drug Monitoring in Aleksandra Pineda, Aniket Puri, Bijan Jahangiri In ammatory Bowel Disease Afrasyab Khan, Arvenia B Berahmana, RESEARCH LETTER Andrew S Day, Murray L Barclay, Michael Schultz 96 Lifespan of New Zealand Second World War veterans from one large cemetery: the case for a national-level study Nick Wilson, Glyn Harper

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 3 www.nzma.org.nz/journal CONTENTS

LETTERS 107 99 Nothing short of universal Comment on: “Quality of electronic healthcare coverage, fully funded records documenting adverse drug and without charge both in primary reactions…” and hospital care is good enough John S Fountain, Susan H Kenyon for New Zealand Stephen Main 101 Switching from gabapentin to METHUSELAH pregabalin Pauline McQuoid 109 Pre-eclampsia and risk of dementia 104 later in life Advancing transgender healthcare 100 YEARS AGO teaching in Aotearoa/New Zealand Althea Gamble Blakey, Gareth Treharne 110 Subclavian Aneurysm Cured by Proximal and Distal Ligature By STANLEY BROWN, F.R.C.S.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 4 www.nzma.org.nz/journal SUMMARIES

Relationship between estimated glomerular ltration rate and incident cardiovascular disease in an ethnically diverse primary care cohort Emma Church, Katrina Poppe, Matire Harwood, Suneela Mehta, Corina Grey, Vanessa Selak, Mark R Marshall, Susan Wells This is a large study examining whether there is a relationship between kidney function (eGFR) and the risk of having a cardiovascular event such as a heart attack or stroke in New Zealanders. The esti-mated Glomerular Filtration Rate (eGFR) result indicates how well the kidneys fi lter waste from the blood and it can identify the presence of kidney damage. The study found that participants with worse kidney function were more likely to go on to have a cardiovascular event. Patients included in the study attended their family doctor for their heart and diabetes check. At this check, a doctor or nurse collects information on any risk factors and calculates a patient’s risk of having a heart attack or stroke over the next fi ve years. The CVD risk is used to guide treatment decisions. After taking into account other CVD risk factors such as smoking, blood pressure and cholesterol levels, Māori and Pa-cifi c patients without diabetes or CVD were at higher risk of developing heart disease with worsen-ing eGFR compared to Europeans. Indian people had similar risk and Other Asians had lower risk compared to Europeans. Further studies are needed to understand whether including kidney func-tion as part of the heart and diabetes check could improve the accuracy of risk prediction and lead to more tailored care of patients seeing their doctor. The factors that lead to a delay between general practitioner referral of symptomatic patients and specialist diagnosis of colorectal cancer: an audit in the Bay of Plenty District Health Board Ada Yee, Tara Linton, Mei-Sze Lee, Simon Raimes Most patients diagnosed with colon or rectal cancer (CRC) have been referred to a specialist by their GP. The Ministry of Health provides guidelines for symptoms that raise a High Index of Suspicion (HIS) for CRC and if GPs refer patients with HIS then they should have a fast track diagnosis within two weeks. This audit examined the referral information of 181 patients who were referred by their GP and subsequently diagnosed with CRC and found that only 36% actually satisfi ed the HIS criteria. Patients who did not satisfy the criteria generally waited a lot longer for a diagnosis. This suggests that the HIS criteria are too strict, and we believe it is now necessary to review these and also to in-troduce a national referral form that prompts the GP to include all the essential information that might then facilitate a more rapid diagnosis. Developing a regional cancer service; lessons from the Waitemata Cancer Patient Experience Survey Jonathan Koea, Sue French, Karen Hellesoe, Peter Sandiford A cancer patient experience survey used in the National Health System (NHS) in the UK was administered to patients with a diagnosis of cancer at Waitemata District Health Board in 2013 and 2015. Overall patient experience benchmarked with or exceeded that of the NHS in clinical metric’s such as wait for clinic appointment, provision for family support and confi - dence in medical care. The survey did indicate further need for detailed information to be available to patients on fi nancial support such as the pharmaceutical subsidy card.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 5 www.nzma.org.nz/journal SUMMARIES

New Zealand Society of Gastroenterology Guidelines on Therapeutic Drug Monitoring in In ammatory Bowel Disease Afrasyab Khan, Arvenia B Berahmana, Andrew S Day, Murray L Barclay, Michael Schultz Various medications are available for the treatment of infl ammatory bowel disease. The effective use of these medications is important and leads to better outcomes. Monitoring the level of medications and/or their breakdown products (therapeutic drug monitoring) in a patient’s blood can lead to more effective and individualised dosage recommendations. These blood tests are available in New Zealand. Therapeutic drug monitoring can decrease the risk of drug toxicity and help in decision making for further treatment if one therapy seems to be failing. This document represents the New Zealand Society of Gastroenterology Guidelines on Therapeutic Drug Monitoring in Infl ammatory Bowel Disease. Vision screening in New Zealand: an audit of the B4 School Check William Muller, Logan Mitchell, Graham Wilson This study reviewed the results of the four-year-old vision screening programme (B4 School Check) in Southern and Tairawhiti DHBs over a six-month period in 2016. Of the 2,109 children screened, 176 (8.3%) ‘failed’ the vision test and were referred on to either optometrist or DHB ophthalmology clinic. Only 91 (51.7%) were determined to have been seen as a result of that referral, and of them 39 (42.8%) were found to have normal vision. Of those referred and seen with abnormal vision, the vast majority (>92%) needed glasses for simple refractive error, while only 10% required any treatment other than glasses. Concerningly, 48.3% of children referred from the B4SC vision screening pro-gramme were not able to be confi rmed as having received optometric or ophthalmic review as a con-sequence of that referral. Considering evidence for ethnicity bias using assessment case scenarios and medical student correctness and certainty Mike Tweed, Gordon Purdie, Cameron Lacey Analysis of student responses allowing for the ethnicity of the patient as described in a test can be used to investigate for evidence of variations in response patterns. Variations in response patterns may indicate bias based on ethnicity. In this study, no difference in correctness of responses, certainty in responses or correctness for levels of certainty was detected. District health board of residence, ethnicity and socioeconomic status all impact publicly funded insulin pump uptake in New Zealand patients with type 1 diabetes Benjamin J Wheeler, Rhiannon Braund, Barbara Galland, Anastasia Mikuscheva, Esko Wiltshire, Craig Je eries, Michel de Lange Insulin pump therapy is becoming increasingly common for those living with type 1 diabetes, and has been publicly funded in New Zealand since 2012. This study aimed to examine national uptake of publicly funded insulin pumps from 2012 to 2016 using data provided by the Ministry of Health. The main fi ndings show that from 2012 to 2016 there has been a steadily increasing uptake of pumps. Despite publicly funded access, disparities in use appear to exist, including by district health board of residence as well as traditionally described socio-de- mographic barriers to healthcare (ethnicity, socioeconomic status, age, gender). Efforts to understand and reduce these disparities are required.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 6 www.nzma.org.nz/journal EDITORIAL

Where are we going with cancer treatment in New Zealand? Tamara Mullaney, Frank Frizelle

n 2018, the Government directed the “Cancer care at a Crossroads“ conference Ministry of Health to develop a new set in Wellington a month ago.3 The aim of of performance measures to improve this conference was to develop a vision for I 1 health outcomes for New Zealanders. This long-term cancer control, specifi c to the is intended to prioritise population health needs of Aotearoa/New Zealand. This week’s outcomes with a view to health resource edition of the Journal documents some optimisation. The Ministry aims to have issues with cancer treatment. the new measures in place in early 2019. In The paper by Koea et al4 focuses on the developing these new measures, the Minis- development of the cancer service at Wait- try has been asked to consider the following emata District Health Board (DHB), New criteria: Zealand’s largest single health provider with • a mix of health system and population a catchment of about 600,000. In this study health improvement measures they used a survey to assess patients’ views • alignment with government prior- on currently available services. The ques- ities, for example, child wellbeing and tionnaire that was used was developed for mental health the National Health Service (NHS) in Britain. In the reported study they assessed patients • be quantifi ed and timed in 2013 and again in 2015 across all tumour • availability of data to monitor streams. Not surprisingly they showed progress an improvement in most aspects of care, • sector engagement and support however the proportion of patients invited • focus on health issues with alignment to participate in cancer research decreased to socio-economic determinants from 35.9% to 21.4% (p<0.001). What is interesting is no similar measuring tool One of the important aspects of these designed specifi cally for New Zealand exists, changes will be the application of these new which would permit limited comparison and measures to cancer treatment. Our survival benchmarking across other DHBs. rates from cancer lag behind those of Australia, our most vulnerable populations Yee et al5 report signifi cant variation in continue to experience poorer outcomes time to diagnosis of non-emergency referrals from cancer than other New Zealanders and of symptomatic patients subsequently the costs of cancer care keep on increasing shown to have colorectal cancer in Bay of exponentially. The previous government Plenty DHB. The High Index of Suspicion had the ministry focusing on faster cancer target for fast-track diagnosis is 14 days from treatment times and national tumour receipt of the referral letter. In fact, only a standards stream pathways.2 The national quarter of their study population patients tumour standards had both cancer stream followed this pathway, with a median time specifi c and generic factors in the pathway. to their diagnosis of 21 days from general All pathways also had a focus on improved practitioner (GP) referral date. This was in data collection. These developments appear, contrast to a median of 67 days for all other however, to have run out of steam under the patients who did not enter this pathway. present government. When the national criteria for assessment were applied, the patients allocated a Grade Concern about the lack of a coordinated 1 urgency by the specialist waited a median plan in cancer treatment led to a recent of 26 days for diagnosis, compared with over

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 7 www.nzma.org.nz/journal EDITORIAL

100 days for those who were graded less The majority of people (71%) were diag- than Grade 1. Patients referred by their GP nosed with bowel cancer following referral to with new lower gastrointestinal symptoms a clinic, the proportion of patients diagnosed who went direct to colonoscopy had a following a referral from screening services median time to diagnosis of 32 days, while was 3%, and 26% were diagnosed following those who were seen in clinic had median presentation at an emergency department. time to diagnosis 81 days. They found that The 90-day post-operative mortality GPs were failing to identify and correctly in patients undergoing major resection refer nearly half of those who did satisfy the reported in the NBOCA audits for the criteria. A standardised referral form which four-year period 2013–16 was 3.8%. In the prompts the inclusion of all required infor- 2017 report, the 90-day mortality in patients mation and fi ndings would also potentially undergoing emergency surgery was signifi - improve the referral pathway, a step many cantly higher than for those having elective DHBs have already undertaken. Thus, how surgery (10.3% compared to 1.9%). There you enter the colorectal cancer diagnostic was wide variation in 90-day mortality pathway matters. between trusts. The issue of variation in treatment of In people with colon cancer, the 90-day patients in colorectal cancer is also the mortality following resection was 4.8%. focus of a paper released on 4 March on the People who have emergency surgery do 6 Ministry of Health website. The opaquely worse. Of people with colorectal cancer, titled and misreported Bowel Cancer Quality the proportion who undergo major surgical Improvement Report 2019, builds on the resection performed as an emergency in 7 out-of-date Piper study. This report shows New Zealand was 19.6%. In people with clearly the variation that is occurring in colon cancer, the rate of emergency surgery treatment and outcomes between district was 24.7%, and in people with rectal cancer health boards across all indicators. This the rate was 4.4%. The highest rates of is despite the establishment of a national emergency surgery were for Māori (23.8%), bowel cancer working group in 2011, with females (21.1%), people younger than 50 the purpose being to improve the care of years (27.1%) old and those over 75 years patients with colorectal cancer. old (21.2%), again refl ecting that our most Some of the more interesting statistics vulnerable populations continue to expe- from this report are: rience poorer outcomes.

Figure 1: Proportion of people diagnosed with bowel cancer following presentation at an emergency department, by district health board of domicile, 2013–16.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 8 www.nzma.org.nz/journal EDITORIAL

Figure 2: Observed 90-day post-operative mortality (elective admissions only) for patients diagnosed with bowel cancer, by district health board of service, 2013–16.

There are large variations also docu- Hopefully then with the ministry’s new mented in how people are treated, for focus on population health outcomes, and instance who gets an operation, who gets their stated goal of attempting to ensure that radiotherapy, what sort of radiotherapy is health resources are used optimally, we will given (short course versus long course). see less of this variation and faster, more While some of these variations may have evidence-based treatment. While we wait very good reasons such as local resourcing, for the Ministry’s new set of performance given that the decisions should be built on measures, we hope that it enables a more the same evidence base, then the level of coordinated national plan and direction of variation is a surprise. cancer treatment.

Competing interests: Nil. Author information: Tamara Mullaney, Senior Lecturer, University of Otago, Christchurch; Frank Frizelle, Department of Surgery, Christchurch Hospital, Christchurch. Corresponding author: Professor Frank Frizelle, Department of Surgery, Christchurch Hospital, Christchurch. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7822

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 9 www.nzma.org.nz/journal EDITORIAL

REFERENCES: 1. http://www.health.govt.nz/ service; lessons from 6. http://www.health. new-zealand-health-system/ the Waitemata Cancer govt.nz/publication/ health-targets Patient Experience Survey. bowel-cancer-quality-im- 2. http://www.health. 2019; 132(1491):38–45. provement-report-2019 govt.nz/our-work/ 5. Yee A, Linton T, Lee M-S, 7. http://www.nzma.org.nz/ diseases-and-conditions/ Raimes S. The factors that journal/read-the-journal/ national-cancer-pro- lead to a delay between all-issues/2010-2019/2018/ gramme/cancer-initiatives/ general practitioner refer- vol-131-no-1476-8- faster-cancer-treatment ral of symptomatic patients june-2018/7582 3. http://www.otago.ac.nz/ and specialist diagnosis cancer-care/index.html of colorectal cancer: an audit in the Bay of Plenty 4. Koea J, French S, Hellesoe District Health Board. K, Sandiford P. Devel- 2019; 132(1491):27–37. oping a regional cancer

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 10 www.nzma.org.nz/journal ARTICLE

Relationship between estimated glomerular ltration rate and incident cardiovascular disease in an ethnically diverse primary care cohort Emma Church, Katrina Poppe, Matire Harwood, Suneela Mehta, Corina Grey, Vanessa Selak, Mark R Marshall, Susan Wells

ABSTRACT AIM: To investigate eGFR as an independent risk factor for CVD in a New Zealand primary care cohort, stratified by disease status (prior CVD, diabetes or no CVD or diabetes). METHOD: The PREDICT-CVD open cohort study is a large, ethnically diverse, New Zealand primary care cohort, generated by using a web-based CVD risk assessment tool. Using encrypted identifiers, participant profiles were linked anonymously to a regional laboratory database (to determine renal function) and to national hospitalisation and mortality datasets. Analyses using a single baseline eGFR measurement were undertaken in three clinical sub-cohorts of participants: those with prior CVD (n=29,742), with diabetes (n=44,416) and with neither CVD nor diabetes (n=192,696). The association between baseline eGFR (by category ≥90, 60–89.9, 30–59.9, and <30ml/min/1.73m2) and incident CVD was analysed with Kaplan Meier plots and Cox regression models. RESULTS: A er adjustment for traditional CVD risk factors, there was an inverse relationship between CVD risk and eGFR, up to an eGFR of 60ml/min/1.73m2 in all three clinical sub-cohorts, and up to an eGFR of 90ml/min/1.73m2 in the sub-cohort with CVD or diabetes. Compared to eGFR ≥90ml/min/1.73m 2, the adjusted hazard ratios of a new CVD event for eGFR <30ml/min/1.73m2 in the CVD, diabetes and no CVD/ no diabetes sub-cohorts were 2.29 (95% CI 2.00–2.61), 4.71 (3.92–5.67) and 2.78 (2.05–3.77), respectively. Compared to European/Other ethnic groups, Māori participants remained at greater adjusted risk of a new CVD event in all clinical sub-cohorts and Pacific people only in the no CVD/no diabetes sub-cohort, whereas Indian participants had a similar adjusted risk to European/Other, and Other Asian patients were consistently at lower adjusted risk. Sensitivity analyses for individuals with consecutive eGFR results (>90 days apart) yielded similar results. CONCLUSION: This study has confirmed that, in a large ethnically diverse primary care cohort, eGFR is a significant independent predictor of CVD risk, and the risk varies by ethnic group.

stimated glomerular fi ltration rate min/1.73m2 (moderately decreased), 15–29 (eGFR) is used in the detection and ml/min/1.73m2 (severely decreased) and Eclassifi cation of chronic kidney disease <15ml/min/1.73m2 (end stage renal disease).1 (CKD) as a marker of chronic renal impair- International population-based cohorts ment. CKD classifi cation is based on two con- indicate that the lower the level of eGFR, the secutive eGFR results at least 90 days apart. greater the risk of incident cardiovascular It is usually categorised into fi ve levels from disease (CVD) events, hospitalisation and eGFR ≥90ml/min/1.73m2 (normal), 60–89ml/ deaths.2,3 A Chronic Kidney Disease-Progno- min/1.73m2 (mildly decreased), 30–59ml/ sis Consortium (CKD-PC) meta-analysis of 1.4

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 11 www.nzma.org.nz/journal ARTICLE

million participants from 30 cohort studies recommendations incorporating national in the US showed that, after adjustment for guidelines are displayed after the online traditional cardiovascular risk factors and form has been completed. The anonymised albuminuria, the risk gradient for cardio- patient profi les are stored on a secure vascular mortality increased linearly with off-site server (Enigma Solutions Ltd) and decreasing eGFR below the threshold of form the research cohort. The cohort can be 75ml/min/1.73m2.4 linked to regional and national databases Existing studies differ in terms of popu- through a patient’s encrypted National lations studied, defi nitions of CVD and Health Index (a unique identifi er in New CKD, estimating equations for GFR, and Zealand health services [NHI]) providing thresholds below which eGFR increases CVD data on laboratory tests, deaths, hospital- risk. Currently there are no epidemiological isations and medication dispensing history. 10 studies investigating the impact of renal Over 95% of New Zealanders have an NHI. impairment on CVD risk in New Zealanders Good agreement has been found between and ethnic subgroups who have higher risk the data stored in PREDICT and patient data 11 of both CKD and CVD.5,6The relationship within primary care electronic records. between renal function and new CVD events Study population may also differ by disease status (prior CVD Between 2002 and 2015, over 450,000 or diabetes). Both patients with diabetes people between the ages of 35 and 74 years and/or prior CVD have higher risk of new had a CVD risk assessment undertaken using CVD events and increased risk of renal the PREDICT online forms. The PREDICT impairment due to underlying vascular software is available in approximately disease, shared risk factors or diabetic 35–40% of primary care practices in New nephropathy.7 Kerr et al found that people Zealand (mainly Auckland and Northland) with a history of CVD had an additional 20% serving around 35% of the resident popu- fi ve-year CVD risk on top of the observed lation.9 After excluding patients who had risk of participants without a history of an International Classifi cation of Disease-10 CVD.8 Given these differences, stratifying Australian modifi cation (ICD-10-AM) coded analyses into three distinct risk groups (prior history of dialysis or kidney transplan- CVD, diabetes and no CVD/no diabetes) when tation, participants were divided into the examining the relationship between renal three sub-cohorts: CVD, diabetes (without function and CVD, increases the homoge- CVD) or no CVD/no diabetes. The CVD neity of the subgroups assessed and provides cohort was derived in three ways. Firstly, more clinically meaningful results. if a GP or nurse reported a patient’s prior Using a large ethnically diverse primary history of angina, myocardial infarction, care cohort generated in the course of coronary artery bypass grafting (CABG), routine CVD risk assessment in New Zealand percutaneous coronary intervention (PCI), primary care, we aimed to determine the transient ischaemic attack (TIA), ischaemic impact of eGFR on fatal and non-fatal CVD stroke, peripheral vascular disease (PVD) or events independent of other known CVD atherosclerotic vascular procedure during risk factors and whether the risk varied by the PREDICT risk assessment. Secondly, disease status (prior CVD or diabetes) and from ICD-10-AM coded CVD-related hospital ethnic group. admissions in the state-funded system (over 95% of acute CVD events are managed in Methods publicly funded hospitals).12 Thirdly, linkage to the national pharmaceutical database Study design (PHARMS) identifi ed patients with three or The PREDICT-CVD cohort study has been more prescriptions of anti-anginal medica- 9 previously described. Briefl y, patients have tions in the previous fi ve years from records their CVD risk assessed by general practi- of subsidised pharmaceutical dispensing. tioners (GPs) using PREDICT online forms The diabetes sub-cohort does not include that are integrated with their enrolled any patients with a history of CVD. This patients’ electronic health records. This sub-cohort was also derived in three web-based tool is used in primary care for ways: reported by their practitioner as the assessment and management of CVD having diabetes, prior hospitalisation with risk. The risk score and evidence-based

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 12 www.nzma.org.nz/journal ARTICLE

ICD-10-AM diabetes codes, and linkage to Indian and Fijian and who were therefore the PHARMS database detecting at least one assumed to be Fijian Indian), Other Asian prescription for insulin or oral hypogly- and fi nally European and Other ethnic caemic agents in the six months prior to the groups (European/Other). 14 patient’s CVD assessment (see Appendix for Covariates full defi nitions and ICD-10-AM codes). Variables from PREDICT include age Additional CVD history (derived from date of birth), sex, ethnicity, History of heart failure was defi ned as self-reported family history of premature any prior hospitalisation for heart failure, CVD, smoking status, systolic and diastolic dispensing of metolazone in the last six blood pressure (BP), and total cholester- months, or dispensing of loop-diuretics at ol-to-high density lipoprotein ratio (TC/HDL). least three times in the previous fi ve years. Body mass index (BMI) was included in the History of atrial fi brillation was defi ned as PREDICT template but was not a mandatory any prior hospitalisation or if recorded in fi eld for CVD risk assessment. the PREDICT online form. Family history of premature CVD was Ascertainment of renal function defi ned as a self-reported familial history Serum creatinine results were provided of ischaemic heart disease or ischaemic where available through linkage to stroke occurring in a father or brother ‘TestSafe’, the repository for results in before 55 years of age, or a mother or sister Northland and Auckland. We identifi ed before 65 years of age. Smoking status was the serum creatinine result nearest to the defi ned as never smoked, previous smoker fi rst PREDICT assessment and within either (giving up over 12 months ago or within the the two years prior or 14 days after that previous 12 months) or current smoker (of assessment. From the serum creatinine, any amount). Linkage to the primary health a baseline eGFR was calculated using the organisation enrolment database provided Chronic Kidney Disease Epidemiology information on the New Zealand Depri- Collaboration (CKD-EPI) equation.13 The vation Index (NZDep). The NZDep score analysis was stratifi ed by eGFR categories assigns a measure of socioeconomic status to ≥90 (reference group), 60–89.9, 30–59.9, and a patient’s geographical area of residence.15 <30ml/min/1.73m2. It is derived from variables in the Census that signify eight dimensions of relative The Jaffe method for serum creatinine deprivation and has been used as quintiles measurement is used by all laboratories in this analysis from the least deprived (1) to providing results for this cohort (S. Musaad, the most deprived (5). personal communication, 2017). Creatinine testing standardised to the isotope-dilution The PHARMS database also provided mass spectrometry (IDMS) methodology was information on dispensed cardiovascular gradually introduced into New Zealand labo- and diabetes-related medications. ratories after 2008. Follow-up and outcomes The procedure for identifying patients Entry to the study was prospective from 1 with albuminuria in the diabetes sub-cohort August 2002 and participants were electroni- was using the protein/creatinine ratio (mg/ cally followed up until they either died, were mmol or g/mol), urine protein/24 hours lost to follow-up or reached the date of data (mg/24 hours) or albumin/creatinine ratio extraction (31 December 2015). The index (mg/mmol [ACR]), with thresholds as defi ned date for this study was taken as the date in the Kidney Disease: Improving Global of the fi rst completed PREDICT assessment Outcomes (KDIGO) guidelines.1 The albu- (baseline assessment). minuria result closest to baseline was used New CVD events were identifi ed using within two years prior and 14 days post ICD-10-AM codes from mortality records PREDICT assessment. and hospital discharges. New CVD events Ethnicity include fatal and non-fatal acute coronary Ethnic groups were based on national syndromes, unstable angina, myocardial ethnicity data prioritisation protocols in the infarction, coronary heart disease, heart following order: NZ Māori, Pacifi c, Indian failure, PVD, TIA, ischaemic stroke and (including individuals who identifi ed as both haemorrhagic stroke (See Appendix for ICD-10-AM codes).

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 13 www.nzma.org.nz/journal ARTICLE

Statistical analysis measures of renal function less than 60ml/ 2 Statistical analyses centred on devel- min/1.73m , at least 90 days apart) for the 1 oping survival models for incidence of diagnosis of CKD. Sensitivity analyses were any CVD event, taking serum eGFR as a performed for the diabetes sub-cohort potential predictor. Kaplan-Meier plots were by also controlling for albuminuria as a produced to show disease-free survival over binary variable (positive or negative result). time in each of the three sub-cohorts, strat- Albuminuria was not included in the main ifi ed by eGFR category. Time from baseline analyses as only 68% of participants had a assessment was used as the time scale in baseline result from any source. multivariable Cox regression models strat- All analyses were done using R software ifi ed by the three sub-cohorts. version 3.2.3. All models controlled for age group, Ethical approval ethnicity, sex, smoking status, NZDep The PREDICT study was approved by the quintile, systolic BP, diastolic BP, TC/HDL, Northern Region Ethics Committee Y in 2003 family history of premature CVD, any (AKY/03/12/314) with subsequent annual lipid-lowering, BP-lowering, diabetes-related, approval by the National Multi Region Ethics antiplatelet or anticoagulant medications Committee since 2007 (MEC07/19/EXP). at baseline. The CVD sub-cohort was also adjusted for any history of atrial fi brillation Results or heart failure, and the diabetes sub-cohort Between 1 August 2002 and 31 December was adjusted for BMI (as it was the only 2015, 455,822 patients aged 35–74 years had sub-cohort with suffi cient data). For all a CVD risk assessment using the PREDICT models, the proportional hazards assumption online form. Those with an ICD-10-AM coded was assessed using Schoenfeld residual plots history of hospitalisation for renal dialysis and the linearity of the association between or transplantation (n=14,256) or missing continuous predictors and the hazard eGFR (n= 174,712) were excluded. Partici- function via Martingale residual plots. pants were then allocated to disease status Due to missing data, a single eGFR result sub-cohorts; 29,742 with prior CVD, 44,416 was used in the main analyses, with sensi- with diabetes but no prior CVD and 192,696 tivity analyses conducted using confi rmed in the no CVD/no diabetes sub-cohort (Figure eGFR results for patients with eGFR 1). The baseline characteristics are displayed <60ml/min/1.73m2 meeting the KDIGO in Table 1, stratifi ed by sub-cohort. guideline requirements (two consecutive

Figure 1: Flowchart of PREDICT-CVD cohort with sub-cohorts.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 14 www.nzma.org.nz/journal ARTICLE

Table 1: Baseline characteristics of each sub-cohort.

Characteristic CVD Diabetes No CVD/no diabetes n=29,742 n=44,416 n=192,696 Age group 35–44 1,117 (4) 7,529 (17) 24,229 (13)

45–54 5,259 (18) 14,482 (33) 68,098 (35)

55–64 10,599 (36) 13,743 (31) 66,202 (34)

65–74 12,767 (43) 8,662 (20) 34,167 (18)

Female 10,954 (37) 21,344 (48) 88,515 (46)

Ethnicity NZ Māori 4,001 (14) 5,791 (13) 20,215 (10)

Pacific 3,492 (12) 12,225 (28) 21,646 (11)

Indian 2,527 (9) 7,063 (16) 14,957 (8)

Other Asian 1,779 (6) 5,135 (12) 19,294 (10)

European/Other 17,943 (60) 14,202 (32) 116,584 (61)

NZDep quintile 1 (least deprived) 5,681 (19) 5,892 (13) 49,303 (26)

quintile 2 5,349 (18) 6,454 (15) 40,334 (21)

quintile 3 5,238 (18) 6,953 (16) 34,193 (18)

quintile 4 5,879 (20) 9,502 (21) 33,142 (17)

quintile 5 (most deprived) 7,594 (26) 15,613 (35) 35,712 (19)

Diabetes 10,487 (35) 44,416 (100) ~

Heart failure 4,588 (15) ~ ~

Atrial fibrillation 4,138 (14) ~ ~

Current smoker 4,391 (15) 6,450 (15) 25,656 (13)

Previous smoker 7,944 (27) 7,460 (17) 30,930 (16)

Never smoked 17,406 (59) 30,505 (69) 136,110 (71)

Systolic BP (mmHg) 131 (17.5) 132 (17.2) 129 (16.7)

TC/HDL ratio 3.8 (1.20) 4.2 (1.31) 4.0 (1.18)

BP-lowering medication 22,971 (77) 26,430 (60) 45,474 (24)

Lipid-lowering medication 21,500 (72) 23,863 (54) 28,293 (15)

Antiplatelet/anticoagulant medication 21,649 (73) 16,947 (38) 16,890 (9)

Diabetes-related medication 6,920 (23) 27,840 (63) ~

BMI (kg/m2) <25 ~ 5,815 (13) ~

25–29.9 ~ 25,097 (57) ~

≥30 ~ 12,013 (27) ~

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 15 www.nzma.org.nz/journal ARTICLE

Table 1: Baseline characteristics of each sub-cohort (continued).

eGFR (ml/min/1.73m2) ≥90 9,139 (31) 21,863 (49) 96,371 (50)

60–89.9 16,046 (54) 18,909 (43) 89,814 (47)

30–59.9 4,096 (14) 3,302 (7) 6,269 (3)

<30 461 (2) 342 (1) 242 (0.1)

Mean follow-up (years) 5.0 5.4 4.1

Median follow-up (years) 4.6 5.3 3.6

Person-years 149,793 238,794 782,542

Number of fatal and non-fatal first events 7,560 4,328 6,527

Crude event rate per 1,000 person-years with 50 (49–52) 18 (18–19) 8 (8–9) 95% confidence intervals

Categorical values given as n (%) and continuous variables as mean (standard deviation). Three missing values for NZDep quintile; two missing values for smoking status; two missing values for diastolic BP; 43 missing values for TC/HDL.

In comparison to the other sub-cohorts, per 1,000-person years (95% CI 18–19), and the group with prior CVD were older and eight per 1,000-person years (95% CI 8–9), included fewer women. People with CVD or respectively (Table 1). diabetes were more likely to live in the most The Kaplan-Meier plots in Figure 2 show deprived areas of residence (NZDep quintile incident hospitalisations or deaths occurring 5). The proportions of Māori patients during follow-up in each of the three were similar across all three sub-cohorts. sub-groups by eGFR category. With each However, there were higher proportions of category of decreasing eGFR, the plots show Pacifi c, Indian and Other Asian patients in reduced disease-free survival probability the diabetes sub-cohort compared to those over time, and a sharp decrease within the with prior CVD. In the CVD sub-cohort, fi rst 500 days under 60ml/min/1.73m2 in the over one-third had diabetes, over 15% had CVD sub-cohort and under 30ml/min/1.73m2 heart failure and nearly 14% had atrial in the diabetes sub-cohort, compared to a fi brillation. In the diabetes sub-cohort, more gradual event rate in the sub-cohort nearly 63% were prescribed diabetes-re- with no CVD or diabetes. lated medications, and 84% were either Adjusted hazard ratios (adjHR) for new overweight or obese. The median times CVD events by sub-cohort are shown in between the eGFR result and the baseline Table 2. Compared to an eGFR ≥90ml/ PREDICT visit were 56, 29 and 88 days for min/1.73m2, there was an increase in the the CVD, diabetes and no CVD/no diabetes adjHR by each category of worsening sub-cohorts, respectively. Using a defi nition renal impairment, with a nearly fi ve-fold of CKD as eGFR under 60ml/min/1.73m2, the higher risk of a fatal or non-fatal CVD event CVD sub-cohort had the highest proportion among people with diabetes and an eGFR of individuals with CKD (16%) and the no under 30ml/min/1.73m2 (adjHR 4.71[95% CVD/no diabetes sub-cohort the lowest (3%). CI 3.92–5.67]). There were signifi cant Of the 30,350 patients who were tested in increases in the risk of new CVD events for the diabetes sub-cohort, 40% were found to eGFR results under 90ml/min/1.73m2 in the have albuminuria. CVD and diabetes sub-cohorts, and under The crude event rates for ischaemic CVD 60ml/min/1.73m2 in the no CVD/no diabetes were highest in the CVD sub-cohort at 50 per sub-cohort. The 60–89.9ml/min/1.73m2 1,000-person years (95% confi dence interval category of eGFR remained signifi cant in the [CI] 49–52). The diabetes and no CVD/no diabetes sub-cohort when albuminuria was diabetes sub-cohorts had event rates of 18 included in the model.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 16 www.nzma.org.nz/journal ARTICLE

Table 2: Adjusted hazard ratios (with 95% confi dence intervals) for fatal and non-fatal cardiovascular events by sub-cohort.

CVD Diabetes No CVD/no diabetes eGFR(ml/min/1.73m2) ≥90 1 1 1

60–89 1.10 (1.03–1.17) 1.12 (1.04–1.20) 1.00 (0.94–1.05)

30–59 1.50 (1.39–1.61) 1.95 (1.76–2.17) 1.54 (1.40–1.70)

<30 2.29 (2.00–2.61) 4.71 (3.92–5.67) 2.78 (2.05–3.77) Ethnicity Euro/Other 1 1 1

NZ Māori 1.09 (1.02–1.17) 1.38 (1.25–1.52) 1.57 (1.45–1.69)

Pacific 0.90 (0.84–0.98) 1.02 (0.93–1.12) 1.17 (1.08–1.27) Indian 1.05 (0.96–1.14) 1.09 (0.98–1.21) 0.98 (0.88–1.10)

Other Asian 0.68 (0.60–0.78) 0.62 (0.53–0.72) 0.65 (0.58–0.73) Age group (years) 35–44 1 1 1

45–54 1.13 (0.98–1.30) 1.46 (1.29–1.64) 1.87 (1.67–2.10)

55–64 1.21 (1.05–1.39) 1.95 (1.72–2.20) 3.24 (2.88–3.64)

65–74 1.54 (1.33–1.77) 2.61 (2.28–2.99) 6.23 (5.51–7.04) Sex Male 1 1 1

Female 0.81 (0.77–0.85) 0.67 (0.63–0.71) 0.65 (0.61–0.68) NZDep quintile 1 1 1 1

quintile 2 1.01 (0.93–1.11) 1.06 (0.92–1.21) 1.06 (0.98–1.15)

quintile 3 1.24 (1.14–1.35) 1.14 (1.00–1.29) 1.25 (1.15–1.35)

quintile 4 1.30 (1.20–1.41) 1.36 (1.21–1.54) 1.31 (1.21–1.42)

quintile 5 1.48 (1.37–1.60) 1.45 (1.29–1.63) 1.59 (1.46–1.72)

Diabetes 1.29 (1.21–1.38) ~ ~

Heart failure 1.39 (1.31–1.47) ~ ~

Atrial fibrillation 1.29 (1.21–1.38) ~ ~ Never smoked 1 1 1

Previous smoker 1.13 (1.07–1.19) 1.12 (1.03–1.22) 1.15 (1.07–1.22)

Current smoker 1.49 (1.40–1.59) 1.50 (1.38–1.63) 1.76 (1.64–1.87)

Systolic BP (mmHg) 1.00 (1.00–1.01) 1.01 (1.01–1.01) 1.01 (1.01–1.01)

Diastolic BP (mmHg) 0.99 (0.99–1.00) 1.00 (0.99–1.00) 1.00 (1.00–1.00)

TC/HDL ratio 1.07 (1.05–1.09) 1.10 (1.08–1.12) 1.15 (1.13–1.17)

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 17 www.nzma.org.nz/journal ARTICLE

Table 2: Adjusted hazard ratios (with 95% confi dence intervals) for fatal and non-fatal cardiovascular events by sub-cohort (continued).

BP-lowering medication 1.29 (1.19–1.39) 1.41 (1.30–1.53) 1.51 (1.42–1.60)

Lipid-lowering medication 0.93 (0.87–0.99) 0.91 (0.84–0.97) 0.90 (0.84–0.96)

Antiplatelet/anticoagulant medication 1.15 (1.07–1.22) 1.14 (1.06–1.22) 1.38 (1.29–1.48)

Diabetes-related medication 1.16 (1.09–1.25) 1.06 (0.98–1.14) ~ BMI (kg/m2) <25 ~ 1 ~

25–29.9 ~ 0.93 (0.83–1.03) ~

≥30 ~ 0.93 (0.83–1.04) ~

Statistically significant results in bold.

Figure 2: Kaplan-Meier curves showing disease-free survival by eGFR category in each sub-cohort.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 18 www.nzma.org.nz/journal ARTICLE

Figure 2: Kaplan-Meier curves showing disease-free survival by eGFR category in each sub-cohort (continued).

Compared to people of European/Other in all three sub-cohorts, showed similar descent, Māori were at increased risk in results to the main analyses (see Appendix 2). all three sub-cohorts, particularly in the sub-cohort with no CVD or diabetes (adjHR Discussion 1.57 [95% CI 1.45–1.69]). In the diabetes This study is the fi rst to examine the sub-cohort, Indian participants were at relationship between reduced eGFR and higher risk than their European/Other incident CVD in New Zealanders, including counterparts, but this difference was not ethnic subpopulations. CVD risk increased signifi cant. However, those of Other Asian with each category of worsening eGFR in ethnicity were at signifi cantly lower risk all three sub-cohorts. The main insight than European/Other in all three sub-co- from our study is that, controlling for horts. Pacifi c participants had reduced CVD known risk factors, there were signifi cant risk in comparison to European/Other in the increases in incident CVD risk for people CVD sub-cohort after controlling for renal with eGFR under 90ml/min/1.73m2 in the function and CVD risk factors but higher risk CVD and diabetes sub-cohorts, and under in the no CVD/ no diabetes sub-cohort. 60ml/min/1.73m2 in the no CVD/no diabetes Men, increasing age group, increasing sub-cohort. The signifi cance of fi ndings in quintile of deprivation, current smokers, the 60–89.9ml/min/1.73m2 eGFR category higher systolic BP and TC/HDL ratio and in the CVD and the diabetes sub-cohorts is receiving BP-lowering medications were important because this is considered to be in associated with higher risk. the non-CKD range. Of note, full CKD staging The defi nition of CKD is limited as it was according to KDIGO guidelines1 could not based on a single measure of creatinine and be used in our study because of limited may include participants with acute kidney data regarding albuminuria. It is therefore injury (AKI). For the participants with an possible that some of those with eGFR in the eGFR <60ml/min/1.73m2, confi rmatory eGFR non-CKD range did in fact have albuminuria results were available for 3,163 participants and therefore CKD at baseline. However, the in the CVD sub-cohort (69.4%), 2,473 partici- insight from our study remains: people with pants in the diabetes sub-cohort (67.9%) and mildly reduced eGFR but CVD or diabetes 3,223 participants in the sub-cohort with no have an increased risk of incident CVD CVD or diabetes (49.5%). Sensitivity analyses compared with those with a normal eGFR. using confi rmatory eGFR <60ml/min/1.73m2

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 19 www.nzma.org.nz/journal ARTICLE

Our study also provided insights into The results of this study were similar to effect modifi cation of eGFR on incident CVD the CKD-PC meta-analyses. However, those risk by ethnicity. Māori had an elevated risk studies did not examine the patient group of a new CVD event compared to European/ with prior CVD as a separate cohort. CKD-PC Other in all three sub-cohorts, over and meta-analyses that have combined high- and above their eGFR category. Importantly, low-risk cohorts show signifi cance for CVD these fi ndings persisted after controlling for risk at eGFRs either below 75ml/min/1.73m2 multiple covariates including socioeconomic independent of albuminuria,20,21 or between deprivation, smoking and medication use. 75–89ml/min/1.73m2 with albuminuria Of note, once eGFR is considered, the risk only.22,23 Many studies have shown a link of CVD among Indian people is not signifi - between renal insuffi ciency and cardiovas- cantly elevated in any sub-cohort. However, cular outcomes in cohorts with diabetes and as 75% of the current Indian population in some studies have additionally linked mild New Zealand are immigrants, it is unclear renal insuffi ciency (60–90ml/min/1.73m2) whether this will change over time as the with these outcomes in diabetic cohorts.24,25 healthy migrant effect abates. This situation In the sub-cohort with no CVD and no is similar for Pacifi c people, and in fact the diabetes, increased CVD risk below eGFR adjusted hazard ratio is lower for them in 60ml/min/1.73m2 is consistent with similar the CVD sub-cohort. studies.2,26 Most studies use single baseline Limitations of this study arise from the eGFR results in their analyses, however, this measurement and modelling of renal study was also able to provide confi rmed function. Some of the creatinine results consecutive eGFR results (>90 days apart) for pre-2008 will not have been standardised sensitivity analyses which yielded similar to the IDMS methodology and more precise results to the main analyses. fi ltration markers (cystatin C) were not Another critical unknown is the asso- available in this study.16 In New Zealand, ciation of rate of decline in eGFR with the CKD-EPI equation has not been vali- subsequent incident CVD risk (adjusted for dated, and it is unclear whether the term confounders and fi rst eGFR), with impli- for the African-American race should be cations for greater declines in eGFR as applied to non-Asian non-white people.17 potential markers of even greater increases Moreover, renal function testing was at the in incident CVD risk.27 discretion of physicians, and 39.6% of people As is usual with real-world evidence, there had missing values for eGFR. The largest is potential for residual confounding from limitation of this study was the use of a unmeasured but prognostically important single baseline creatinine-based eGFR value. variables, and information bias from misclas- Acute and chronic kidney disease have sifi cation. BMI was not included in the CVD different aetiologies and are differentiated and no CVD/no diabetes sub-cohort, and in the KDIGO guidelines by duration (>90 important variables such as physical activity 1 days for CKD). This has implications for and diet were also not included in our data using creatinine-based eGFR at an individual source and may all have affected outcomes level to predict the risk of a new CVD event differentially between comparison groups. using a single creatinine measurement, Linking and merging electronic health data since there can be signifi cant variability in is known to carry several limitations since serum creatinine levels in the same indi- the data has not been obtained to answer vidual on repeat testing and abnormal a specifi c research question, but for other results may be due to acute fl uctuation in clinical, administrative or fi nancial purposes. renal function as opposed to CKD, or the The utility of the information is dependent presence of both CKD and AKI in the same on the quality of data from different sources, individual (pre-existing CKD is a signifi cant data completeness, data curation and the 18 risk factor for AKI). However, the new CVD classifi cation algorithms used for the disease consensus statement published in February states, risks factors and outcomes. For 2018 recommends routine collection of eGFR instance, medication data were derived from for risk stratifi cation and therefore it is dispensed prescriptions from the national likely that missingness of confi rmatory eGFR pharmaceutical claims warehouse but there values for CKD classifi cation will be miti- is no way of actually knowing if patients took 19 gated in the future. them. Finally, the NZDep score is a proxy

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 20 www.nzma.org.nz/journal ARTICLE

measure of deprivation based on meshblock of this study, it would appear that eGFR area, which cannot indicate individual socio- is an independent predictor of new CVD economic position. events, and this relationship varies by The ethnicity data used in this study, ethnic group and by disease status. This collected by the Ministry of Health, may have several clinical implications combined South Asian ethnicities other than for CVD risk assessment in New Zealand. the Indian group (eg, Sri Lankan, Pakistani, Firstly, as an independent predictor of Bangladeshi), with Other Asian ethnicities new CVD events, further New Zealand- (eg, Chinese, Japanese and other South East based research is required to understand Asian ethnicities). Therefore, as South Asian whether a single-creatinine-based eGFR people will have a similar risk factor profi le result could improve risk prediction in to Indian people, it is likely that Chinese and multivariable CVD risk equations, particu- people of South East Asian ethnicity will larly for subgroups at increased risk of CKD have even lower risk of CKD than repre- and CVD. Secondly, it is clear that ethnicity sented here. Recent changes to the ethnicity modifi es the risk of a new CVD event, even protocols for the Health and Disability after accounting for renal function, and Sector in New Zealand in 2017, include the there may be scope for improved clinical requirement to record up to six ethnicities management of ethnic subgroups according at the most detailed level of classifi cation to renal impairment and disease status. (Level 4).28 This will allow the different risk It is certainly plausible that ethnic-spe- profi les between South Asian groups and cifi c equations would result in improved Other Asian ethnicities to be distinguished, prediction and therefore earlier inter- as currently intended in both the CVD risk vention in high-risk ethnic groups. Finally, assessment and CKD detection guidelines in the results of our study may suggest that New Zealand.19,29 incident CVD risk is signifi cant for people with ostensibly normal eGFRs between This large primary care cohort of 60 and 90ml/min/1.73m2 in the CVD and participants with representation of the diabetes sub-cohorts, and there may be main ethnic groups in New Zealand has opportunity for heightened attention to provided evidence of the relationship modifi cation of risk factors and preventive between renal function and new CVD care in these patients. events at the population level. On the basis

Appendix 1 The defi nitions used for this study are outlined below. The dispensing history is taken from the time of patient’s CVD assessment from the Pharmaceutical claims collection database (PHARMS) between 1 January 2005 to 31 December 2015. ICD-10-AM codes used to identify sub-cohorts and CVD events are from hospital records (and mortality records for CVD outcomes) from 1 January 2005 to 21 December 2015. Sub-cohorts CVD sub-cohort Patients with prior CVD were identifi ed based on meeting one of the following criteria:

PREDICT online form Patients with a history of angina, myocardial infarction, coronary artery bypass gra ing (CABG), percutaneous coronary intervention (PCI), transient ischaemic attack (TIA), ischaemic stroke, peripheral vascular disease (PVD) or atherosclerotic vascular procedure were identified by their practitioner.

Any hospital admissions with ICD-10-AM codes listed under “New CVD events.” CVD-related codes

Three or more prescriptions of Subsidised pharmaceutical dispensing of glyceryl trinitrate, anti-anginal medications in the isosorbide dinitrate, isosorbide mononitrate, nicorandil, previous five years pentaerythritol tetranitrate or perhexiline maleate.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 21 www.nzma.org.nz/journal ARTICLE

Within the CVD sub-cohort, history of heart failure and atrial fi brillation were also iden- tifi ed through:

Heart Any prior hospitalisation with heart failure-related code (ICD-10-AM codes listed under failure “New CVD events”) OR Any dispensing of metolazone in the previous six months or dispensing of loop diuretics (bumetanide and frusemide) on at least three occasions in the previous five years.

Atrial History of atrial fibrillation identified from PREDICT fibrillation OR Any hospital admission with atrial fibrillation-related code (ICD-10-AM I48). Diabetes sub-cohort The diabetes sub-cohort does not include any patients with a history of CVD. Patients with diabetes were identifi ed based on meeting one of the following criteria:

PREDICT online form Patients with type 1 or type 2 diabetes were identified by their practitioner.

Any hospital admission with ICD-9 or 10-AM codes: ICD-9 250 and ICD-10-AM E10-E14. diabetes-related codes

At least one prescription for insulin Subsidised pharmaceutical dispensing of insulin, acarbose, or oral hypoglycaemic agents in chlorpropramide, glibenclamide, gliclazide, glipizide, the previous six months metformin, pioglitazone, rosiglitazone, tolazamide, tolbutamide. Medications at baseline Multivariable analyses used any lipid-lowering, BP-lowering, diabetes-related, antiplatelet of anticoagulant medications in the six months prior to PREDICT risk assessment. This was identifi ed either on the PREDICT online form or from medication dispensing history (PHARMS). Drug classes include:

Antiplatelet or Antiplatelet Anticoagulant anticoagulant Aspirin Dabigatran Clopidogrel Phenindione Dipyridamole Rivaroxiban Prasugrel Warfarin Ticagrelor Ticlopidine

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 22 www.nzma.org.nz/journal ARTICLE

BP-lowering ACE Inhibitor Beta-blocker Benazepril Acebutolol Captopril Alprenolol Cilazapril Atenolol Enalapril Bisoprolol Lisinopril Carvedilol Perindopril Celiprolol Quinepril Labetolol Trandolapril Metoprolol Nadolol ARB Oxprenolol Candesartan Pindolol Losartan Propanolol Sotalol Other Timolol Amiloride Clonidine CCB Clopamide Amlodipine Hydralazine Dilitiazem Methyldopa Felodipine Triamterene Isradipine Nifedipine Thiazide Verapamil Bendrofluazide Chlorthiazide Cycylopenthiazide Hydrochlorothiazide Indapamide Methyclothiazide

Lipid-lowering Statin Other Atorvastatin Acipimox Ezetimibe with Simvastatin Bezafibrate Fluvastatin Cholestyramine Pravastatin Clofibrate Simvastatin Colestipol hydrochloride Ezetimibe Gemfibrozil Nicotinic acid

Diabetes-related Insulin Metformin Acarbose Pioglitazone Chlorpropramide Rosiglitazone Glibenclamide Tolazamide Gliclazide Tolbutamide Glipizide

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 23 www.nzma.org.nz/journal ARTICLE

CVD events CVD events were identifi ed using ICD-10-AM codes from mortality records and hospital discharges from 1 January 1988 to 31 December 2015. Prior and new CVD events include: • fatal and non-fatal acute coronary syndromes (I200, I210-I214, I219-I221, I228, I229), • unstable angina (I200), • myocardial infarction (I210-I214, I219-I221, I228, I229), • coronary heart disease (I200, I201, I218-I214, I219-I221, I228-I236 I238, I240, I248, I249, I253-I256, I460-1469), • heart failure (I50, I110, I130, I132, I500, I501, I 509), • peripheral vascular disease (I711, I713, I715, I718, I7100-I7103, I7021-I7024, I739-I745, I748, I749, E1050-E1052, E1150-E1151, E1451, E1452), • transient ischaemic attack (G450-G453, G458-G468), • ischaemic stroke (I630-I636, I638, I639, I64) and • haemorrhagic stroke (I600-I616, I618, I619). Appendix 2 Confi rmed CKD results, with two measurements greater than 90 days apart recommended by the KDIGO guidelines, have been used in separate sensitivity analyses, different to the fully adjusted models which used a single baseline eGFR in the main analyses. For the comparison group, results were counted as normal if they had a confi rmed result ≥60ml/ min/1.73m2 or a patient had an index result alone in this same range, as normal results do not need to be repeated. The diabetes sub-cohort in this analysis was also adjusted for confi rmed albuminuria result. Compared to a reference group of greater than or equal to 60ml/min/1.73m2, the adjusted hazard ratios for the three sub-cohorts are all similar to the main analyses. Sensitivity analyses using confi rmed CKD (sequential measurements >90 days apart), displaying adjusted hazard ratios and 95% confi dence intervals by eGFR category in each sub-cohort.

eGFR (ml/min/1.73m2) CVD sub-cohort Diabetes sub-cohort No CVD/No diabetes sub-cohort ≥60 1 1 1

30–60 1.47 (1.37–1.57) 1.74 (1.53–1.97) 1.70 (1.53–1.89)

<30 2.34 (2.03–2.68) 3.38 (2.69–4.26) 3.25 (2.33–4.55)

Adjusted for age group, ethnicity, sex, smoking status, NZDep quintile, systolic BP, diastolic BP, TC/HDL, family history of premature CVD, and any lipid-lowering, BP-lowering, diabetes-related, antiplatelet or anticoagulant drugs in the six months prior to PREDICT risk assessment. CVD sub-cohort also adjusted for diabetes status, any history of heart failure and any history of atrial fibrillation. Diabetes sub-cohort also adjusted for confirmed albuminuria and BMI.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 24 www.nzma.org.nz/journal ARTICLE

Competing interests: SW had a Stevenson Fellowship in Health Innovation and Quality Improvement during the conduct of the study. KP holds the Heart Foundation Hynds Senior Fellowship. MM is a full-time employee of Baxter Healthcare (Asia) Pte Ltd. CG holds a fellowship from the National Heart Foundation. SM, VS, CG and KP report grants from Health Research Council of New Zealand during the conduct of the study. EC reports grants from NZ College of Public Medicine during the conduct of the study. Acknowledgements: The authors would like to thank the primary health care organisations, affi liated general practitioners, practice nurses and patients for their contributions to this study. The development of the PREDICT cohort is the result of a collaboration between epidemiologists and other clinical researchers at the University of Auckland, IT specialists at Enigma Solutions Ltd (a private IT company who developed and maintain the PREDICT software and webserver), primary health care organisations (and their member GPs), non-governmental organisations (New Zealand Guidelines Group, National Heart Foundation, Diabetes New Zealand, Diabetes Auckland), several district health boards and the Ministry of Health. The PREDICT software platform is owned by Enigma Publishing (PREDICT is a trademark of Enigma Solutions Ltd). Author information: Emma Church, Public Health Medicine Registrar, National Screening Unit, Ministry of Health, New Zealand; Katrina Poppe, Section of Epidemiology and Biostatistics, School of Population Health, University of Auckland, Auckland; Matire Harwood, School of Population Health, University of Auckland, Auckland, New Zea- land; Suneela Mehta, Section of Epidemiology and Biostatistics, School of Population Health, University of Auckland, Auckland; Corina Grey, Section of Epidemiology and Biostatistics, School of Population Health, University of Auckland, Auckland; Vanessa Selak, Section of Epidemiology and Biostatistics, School of Population Health, University of Auckland, Auckland; Mark R Marshall, Department of Renal Medicine, Middle- more Hospital, Counties Manukau Health, Auckland; Sue Wells, Section of Epidemiology and Biostatistics, School of Population Health, University of Auckland, Auckland. Corresponding author: Emma Church, c/o National Screening Unit, Ministry of Health, 133 Molesworth Street, Thorndon, Wellington 6011. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7823

REFERENCES: 1. KDIGO. KDIGO 2012 to estimate future risk of 5. Best Practice Advocacy clinical practice guideline cardiovascular disease: Centre New Zealand (BPAC for the evaluation and Prospective cohort study. NZ). The detection and management of chronic BMJ. 2017; 357:j2099. management of patients kidney disease. Kidney Int doi: 10.1136/bmj.j2099. with chronic kidney Suppl. 2013; 3(1):1–150. 4. Matsushita K, van der disease in primary care. 2. Thomas B, Matsushita K, Velde M, Astor BC, et al. BPJ. 2015; 66:36–45. Abate KH, et al. Global Association of estimated 6. Chan WC, Wright C, cardiovascular and renal glomerular fi ltration rate Riddell T, et al. Ethnic and outcomes of reduced and albuminuria with socioeconomic disparities GFR. J Am Soc Nephrol. all-cause and cardio- in the prevalence of 2017; 28(7):2167–2179. vascular mortality: a cardiovascular disease in 3. Hippisley-Cox J, Coupland collaborative meta-analysis New Zealand. N Z Med J. C, Brindle P. Development of general population 2008; 121(1285):11–20. and validation of QRISK3 cohorts. Lancet. 2010; 7. Webster AC, Nagler EV, risk prediction algorithms 375(9731):2073–2081. Morton RL, Masson

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 25 www.nzma.org.nz/journal ARTICLE

P. Chronic kidney 15. Atkinson J, Salmond 23. van der Velde M, Matsushi- disease. Lancet. 2017; C, Crampton P. 2014. ta K, Coresh J, et al. Lower 389(10075):1238–1252. NZDep2013 Index of estimated glomerular 8. Kerr AJ, Broad J, Wells Deprivation. Wellington, fi ltration rate and higher S, et al. Should the fi rst Department of Public albuminuria are associ- priority in cardiovascular Health, Wellington ated with all-cause and risk management be School of Medicine cardiovascular mortality. A those with prior cardio- and Health Sciences. collaborative meta-analysis vascular disease? Heart. 16. Shlipak MG, Matsushita of high-risk population 2009; 95(2):125–129. K, Ärnlöv J, et al. Cystatin cohorts. Kidney Int. 2011; 79(12):1341–1352. 9. Wells S, Riddell T, Kerr C versus creatinine in A, et al. Cohort profi le: determining risk based on 24. Papademetriou V, Lovato The PREDICT cardiovas- kidney function. N Engl J L, Doumas M, et al. cular disease cohort in Med. 2013; 369(10):932–943. Chronic kidney disease and New Zealand primary 17. Hossain F, Kendrick-Jones intensive glycaemic control care (PREDICT-CVD J, Ma TM, Marshall MR. The increase cardiovascular 19). Int J Epidemiol. estimation of glomerular risk in patients with type 2017; 46(1):22–22j. fi ltration rate in an Austra- 2 diabetes. Kidney Int. 2015; 87(3):649–659. 10. Ministry of Health lian and New Zealand New Zealand. National cohort. Nephrology (Carl- 25. Wang Y, Katzmarzyk PT, health index questions ton). 2012; 17(3):285–293. Horswell R, et al. Kidney and answers. http:// 18. Chawla LS, Bellomo R, function and the risk of www.health.govt.nz/ Bihorac A, et al. Acute cardiovascular disease our-work/health-identity/ kidney disease and renal in patients with type 2 national-health-index/nhi-in- recovery: Consensus report diabetes. Kidney Int. formation-health-consumers/ of the acute disease quality 2014; 85(5):1192–1199. national-health-index- initiative (ADQI) 16 work- 26. Garcia-Gil M, Parramon D, questions-and-answers group. Nat Rev Nephrol. Comas-Cufi M, et al. Role of Updated 2018. 2017; 13(4):241–257. renal function in cardio- 11. Riddell T, Kenealy T, Wells 19. Ministry of Health. vascular risk assessment: S, et al. Audit of health 2018. Cardiovascular A retrospective cohort data captured routinely Disease Risk Assessment study in a population with in primary healthcare and Management for low incidence of coronary for the clinical decision Primary Care. Wellington: heart disease. Prev Med. support system PREDICT Ministry of Health. 2016; 89:200–206. (PREDICT CVD-4). HCIRO. 20. Nitsch D, Grams M, Sang Y, 27. Van Pottelbergh G, 2008; 12(1):15–21. et al. Associations of esti- Mamouris P, Opdeweegh N, 12. Kerr A, Williams MJ, mated glomerular fi ltration et al. Is there a correlation White H, et al. The all rate and albuminuria with between an eGFR slope New Zealand acute mortality and renal failure measured over a 5-year coronary syndrome quality by sex: A meta-analysis. period and incident improvement programme: BMJ. 2013; 346:f324. cardiovascular events in the following 5 years Implementation, meth- 21. Wen CP, Matsushita K, among a Flemish general odology and cohorts Coresh J, et al. Relative practice population: A (ANZACS-QI 9). N Z Med risks of chronic kidney retrospective cohort J. 2016; 129(1439):23–36. disease for mortality study. BMJ Open. 2018; 13. Johnson DW, Jones and end-stage renal 8(11):e023594. doi: 10.1136/ GR, Mathew TH, et al. disease across races bmjopen-2018-023594. Chronic kidney disease are similar. Kidney Int. and automatic reporting 2014; 86(4):819–827. 28. Ministry of Health. HISO 10001:2017 Ethnicity Data of estimated glomerular 22. Fox CS, Matsushita K, Protocols. Wellington: fi ltration rate: New Woodward M, et al. Ministry of Health. 2017. developments and revised Associations of kidney recommendations. Med J disease measures with 29. Ministry of Health. 2015. Aust. 2012; 197(4):224–225. mortality and end-stage Managing Chronic Kidney 14. Ministry of Health. 2004. renal disease in individuals Disease in Primary Care: Ethnicity Data Protocols with and without diabetes: National Consensus for the Health and Disabil- A meta-analysis. Lancet. Statement. Wellington: ity Sector. Wellington: 2012; 380(9854):1662–1673. Ministry of Health. Ministry of Health.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 26 www.nzma.org.nz/journal ARTICLE

The factors that lead to a delay between general practitioner referral of symptomatic patients and specialist diagnosis of colorectal cancer: an audit in the Bay of Plenty District Health Board Ada Yee, Tara Linton, Mei-Sze Lee, Simon Raimes

ABSTRACT AIM: To determine factors leading to delay between referral by the GP of symptomatic patients and subsequent specialist diagnosis of colon or rectum cancer (CRC). METHOD: A retrospective audit of patients with new CRC referred by their GP over a 30-month period to the specialist services. Analysis of referral letters, specialist grading and subsequent results of investigations. We focused on the High Index of Suspicion (HIS) criteria for suspected CRC. RESULTS: Only 65 out of 181 patients fulfilled the HIS criteria and of these only half were correctly identified in the referral letter. Only 48 who fulfilled HIS criteria were graded as urgent by the specialist and had their fast-track diagnostic test within a median of 21 days (5–114). The remaining 133 waited a median of 67 days (10–387) (p<0.001). The diagnosis was reached faster if the patient went straight to colonoscopy rather than initial outpatient assessment: median 32 versus 81 days (p=0.008). CONCLUSION: The HIS Urgent pathway only identified a third of patients and so the criteria should be reviewed. GPs frequently failed to recognise and refer those who met the criteria. A standardised referral form prompting the inclusion of all required information would improve this.

olorectal cancer (CRC) is the second are not as effective in New Zealand.5 Delay most common cause of cancer death in presentation and more advanced disease in both males and females in New at diagnosis are even more of a problem in C 1 4,6 Zealand. New Zealand has one of the Māori and Pacifi ca patients. highest incidences of CRC in the world, Even as the Bowel Cancer Screening 2,3 similar to Australia and Ireland. Despite Programme is being introduced in New this, a higher proportion of patients in Zealand, it is important to remember that New Zealand fi rst present as emergencies most new CRC diagnoses will still present and this country also has a higher rate of through the symptomatic route, as they disseminated Stage IV disease compared do now.7 While up to a third of patients 4 to Australia and the UK. This indirectly continue to present through the emer- suggests that diagnostic pathways for CRC gency route, most will report symptoms

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 27 www.nzma.org.nz/journal ARTICLE

that might indicate large bowel pathology checked to the Cancer Registry. Each patient to their GP and be referred for specialist was identifi ed by their National Health assessment. There have been a number of Index (NHI) number. initiatives to help guide the referral of these Only patients diagnosed with a primary 8 patients. However, previous data suggest CRC who had been referred by their GP to that there is still a concerning delay in the General Surgery or Gastroenterology 9 reaching a diagnosis. services were included in the audit. The Most countries with a high incidence of exclusion criteria were: CRC have adapted rapid diagnosis pathways • Acute/emergency admission for patients with signifi cant symptoms.10,11 • Recurrence of previously treated CRC The New Zealand Ministry of Health has provided guides for High Index of Suspicion • Surveillance programme diagnosis (HIS) symptoms and signs for most of the • Internal referrals from another more common cancers, including CRC.12 specialty (incidental fi nding of CRC) Patients who fulfi l any of the HIS criteria • Patients diagnosed elsewhere and should undergo their fi rst diagnostic test or referred to BOPDHB for treatment. FSA within 14 days of receipt of an appro- Patient characteristics were collected priate referral. and recorded onto an electronic database. The primary aim of this study was to Information included patient demographics, analyse the factors that lead to a delay ethnicity, type of and urgency of GP referral between GP referral and specialist diagnosis letters together with symptoms, signs and of CRC and thus to identify how we might results of investigations that had been improve the diagnostic pathway for symp- included in the referral letter. The specialist tomatic patients. This was achieved by an grading of urgency and the time to grading audit of all patients referred from their GP from referral were recorded. The grading who were subsequently diagnosed with a system used in BOPDHB at that time was: new CRC in the Bay of Plenty DHB (BOPDHB). Grade 1 Urgent, Grade 2a Semi-Urgent, We particularly wanted to examine the HIS Grade 2b Routine and Grades 3 and 4 referral route to determine how effective referral fails to meet criteria. The days from this is at providing a rapid diagnosis. referral to fi rst specialist assessment (FSA) and to diagnostic test confi rmation of diag- Methods nosis (same as FSA if ‘direct to test’), together Patients with the specialist’s record of patient history, clinical signs and fi ndings were recorded. This retrospective audit was carried out Site, histopathology and stage of the tumour at the Bay of Plenty DHB of patients with were also recorded. a new diagnosis of adenocarcinoma of the colon or rectum between 1 January 2014 Table 1 shows the High Index of Suspicion to 30 June 2016. Patients diagnosed during for Cancer Criteria (HIS Urgent) as defi ned that period were identifi ed from the DHB per Ministry of Health (MoH) Faster Cancer database using ICD codes and then cross- Treatment defi nitions (2016).13

Table 1: HIS Urgent criteria.

If the patient presents with one or more of the following red flags, then the referral should be triaged as ‘High Suspicion of Cancer’ Known or suspected bowel cancer (on imaging, or palpable or visible on rectal examination)

Unexplained rectal bleeding (benign anal causes treated or excluded) WITH iron deficiency anaemia (haemoglobin and ferritin below the local reference range)

Altered bowel habit (looser and/or more frequent) >6 weeks duration PLUS unexplained rectal bleeding (benign anal causes treated or excluded) AND aged ≥50 years

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 28 www.nzma.org.nz/journal ARTICLE

Table 2: Demographics.

Demographics Population (n=181) Number %/Range

Sex Female 78 43.1

Male 103 56.9

Mean age, years Total 71.2 24–96

European/Pakeha 71.9 24–90

Māori 63.8 38–96

Ethnicity European/Pakeha 163 90.0

Māori 15 8.3

Asian 2 1.1

Pacifica 1 0.6

Statistical analysis Results Data was analysed using IBM SPSS A total of 181 patients met the criteria for Statistics for Windows, Version 21.0 inclusion in this audit. The demographic (Armonk, NY, US). information for the patients referred by Comparisons were analysed using their GP is shown in Table 2. The site of the Kruskal-Wallis or Mann-Whitney U test primary tumour was similar to PIPER as depending on the number of groups was the age distribution of the BOP patients involved. A two-sided p-value <0.05 was (see Table 3 and Figure 1). Although there considered to be statistically signifi cant. is a higher than the national average Māori In addition to analysis of the pathways population in the Bay of Plenty, only 8.3% of taken by patients in BOP DHB, our patient new CRC cases were Māori. It is notable that group was also compared to the PIPER study the Māori patients were on average eight to demonstrate whether the CRC population years younger than the European/Pakeha audited was representative of New Zealand patients as also seen in the PIPER study. as a whole.4

Table 3: Site of tumour BoP vs PIPER.

Population n=181 Population n=5667 Bay of Plenty (p=0.491) PIPER Female (78) Male (103) Female Male

Colon 56 (44.8%) 69 (55.2%) 2,155 (51.4%) 2,038 (48.6%)

Rectum 22 (39.3%) 34 (60.7%) 525 (37.5%) 876 (62.5%)

Unknown n/a n/a 35 (47.9%) 38 (52.1%)

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 29 www.nzma.org.nz/journal ARTICLE

Figure 1: Site of tumour in age groups.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 30 www.nzma.org.nz/journal ARTICLE

Figure 2: Time between GP referral and diagnosis.

The time between GP referral and diag- receipt of the referral letter. In fact only nosis is shown in Figure 2. Only 41% of 48 (26.5%) patients followed this ‘ideal’ patients with subsequently proven CRC had pathway, with a median time to diagnosis of their diagnostic test within 30 days. The 21 days (mean 24.9 days) from GP referral remaining patients waited varying lengths date. This was in contrast to a median of 67 of time, with 32% waiting more than 90 days days (mean 87.6 days) for all other patients on a ‘routine’ waiting list. (p<0.001) who did not for one reason or The ‘ideal’ rapid diagnostic pathway another follow the rapid route. We then would be identifi cation and recognition of explored the factors that might have led to HIS symptoms and signs prompting Urgent patients not following the ideal pathway. HIS Referral by the GP and then a grading We found a worrying inconsistency in the recommendation by the Specialist of Grade type of referral letter used; in total there 1 for urgency of investigation or clinic were 10 different letter headings used by appointment. The median time from writing the GPs (see Table 4). We recognised that the referral letter to specialist grading was the ‘Suspected CRC’ referral forms preceded six days. The High Index of Suspicion target the use of the HIS forms and that these were for fast-track diagnosis is 14 days from still in use on the electronic system. We

Table 4: GP referral letter types.

GP actual letters Grouped referrals Total • HIS urgent HIS Urgent 73 • Urgent CRC

• CRC semi-urgent HIS Semi-urgent 21 • HIS semi-urgent

• HIS HIS Routine 26 • HIS CRC • Routine CRC

• Urgent Non HIS urgent/semi-urgent 25 • Semi-urgent

• Routine Routine 36

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 31 www.nzma.org.nz/journal ARTICLE

Figure 3: HIS urgent vs HIS other vs Non HIS with days to diagnosis.

included ‘Suspected CRC’ as being the same In Table 5 the effect of the specialist level of concern as ‘High Index of Suspicion’ grading on the delay between referral and and grouped these into those marked diagnosis is shown. The 90 patients allo- Urgent, Semi-Urgent or Routine (see Table cated a Grade 1 urgency by the specialist 4). In total, 120 patients were classed as HIS/ waited a median of 26 days for diagnosis Suspected CRC referrals of varying urgency compared with over 100 days for those who and 61 as non-HIS. were graded less than Grade 1 (p<0.001). The time in days between GP referral and This emphasises the importance of the infor- the diagnosis of cancer is shown in Figure mation provided in the referral for alerting 3, with the patients grouped into three the specialist to Grade appropriately. The main groups: HIS—Urgent, HIS—Other and correlation between the urgency of the GP non-HIS. referral and the subsequent specialist grade is shown in Table 6.

Table 5: Days to diagnosis (specialist grade urgency).

Population (N=181) SPECIALIST GRADINGS No. of patients Mean (days) Median (days) Range (days)

1 90 31.5 26.0 5–155

2a 55 107.8 100.6 14–387

2b 13 172.2 166.0 35–297

3 4 56.0 48.0 34–100

4 1 - - 1

Missing data 16 76.3 42.0 7–288

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 32 www.nzma.org.nz/journal ARTICLE

Table 6: Urgency of GP grading and subsequent grading by specialists.

Specialist grade urgency 1 2a 2b 3 4 Missing Total

GP grading

HIS Urgent 48 15 0 4 0 6 73 HIS Other 25 15 3 0 0 4 47

Non-HIS 17 27 10 0 1 6 61

Total 90 57 13 4 1 16 181

In Table 7 we have analysed the time to days. The fact that some patients still had diagnosis for patients with different HIS long waits until diagnosis is of concern and criteria on the basis of all the information the lower part of the table shows the type of collected. It should be emphasised that some GP referral letter. GP HIS urgent referrals were incorrect as It is notable that only 36 of the 65 patients they were not supported by the information (55%) were correctly identifi ed as HIS and test results provided (‘false-positives’) Urgent and referred appropriately by their and some patients who fulfi lled the criteria GP. While some of the patients who were but were referred with a lesser degree of missed as HIS Urgent by their GP were still urgency (‘false-negatives’). A total of 65 identifi ed as Grade 1 Urgent by the grading patients fulfi lled at least one of the criteria specialist, there was still a statistically signif- (eight fulfi lled more than one)—this is only icant delay in time to diagnosis in the other 36% of all new diagnoses. The time to diag- 29 missed or ‘false-negative’ HIS Urgent nosis for these patients was a mean of 42.4 criteria patients (median 38 vs 21 days (median of 22) days with a range of 5 to 176 p=0.002).

Table 7: HIS Urgent Criteria Match.

HIS URGENT Criteria Match Population (n=181) HIS 1 HIS 2 HIS 3 Total matching criteria (%) 23 (12.7) 27 (14.9) 23 (12.7)

Time to diagnosis Mean 35.8 38.0 43.8

Median 21.0 26.0 23.0

Range 7–166 10–132 5–176

GP gradings HIS urgent 15 15 12

HIS semi-urgent 3 2 2

HIS routine 3 1 1

Non-HIS urgent/semi-urgent 1 6 4

Routine 1 3 4

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 33 www.nzma.org.nz/journal ARTICLE

Table 8: Warning symptoms.

Symptom HIS criteria eligible Not HIS criteria eligible Total Rectal bleeding 41 (22.7%) 45 (24.9%) 86 (47.5%)

Iron deficiency anaemia 21 (11.6%) 37 (30.4%) 58 (32.0%)

Change in bowel habit >6 weeks 46 (25.4%) 30 (16.6%) 76 (42.0%)

Patients referred by their GP with new It has confi rmed our concerns that there lower gastrointestinal symptoms largely are long delays for a signifi cant number of went direct to colonoscopy (140/181) with patients between GP referral and specialist the remainder being seen in an outpatient diagnosis in BOPDHB.14 While other DHBs clinic fi rst. Nearly all the clinic patients then may have their own referral process to went on to have an endoscopy apart from ensure the timely diagnosis of CRC, it is a few frail elderly patients for whom only likely that the picture in the Bay of Plenty is imaging was utilised. The straight to scope not greatly different elsewhere.5,15 route led to a signifi cantly faster diagnosis There has to be concern about the HIS of than the clinic route: median 32 days versus CRC route of referral as this is a national 81 days (p=0.008). standard for timely access to diagnostic We also examined three of the recognised services.15 The fact that only 36% of the ‘warning’ symptoms for signifi cant large patients with proven CRC in BOP actually bowel pathology that when combined met one or more of the HIS criteria is of the may reach the criteria level for HIS urgent greatest concern. It could be argued that referral but do not do so alone. In Table 8 only a prospective study will provide an the fi gures are shown for the frequency of accurate fi gure for HIS criteria. However, these symptoms either combined to reach the point of this audit was to assess the the HIS criteria or alone when the HIS level available evidence provided in the GP was not met. Recognisable PR blood loss was referral letter, together with the results of the most frequent symptom but over half any tests undertaken prior to referral. This the patients did not reach the HIS criteria. It was the information available to the grading was of concern that 42% of patients with PR specialist to make an informed decision on bleeding did not have a documented digital the grading of the urgency of FSA or diag- rectal examination in the referral letter. nostic testing. This audit has demonstrated It was notable that 37 of the 58 patients that there are too many different referral with iron defi ciency anaemia did not have forms available to GPs in BOPDHB. We have recognisable PR bleeding thus reducing the also found that the information provided impact of this fi nding on the rapid diagnosis is often incomplete. We believe there is a of CRC. strong case to re-design the form to prompt provision of all the necessary information for referral of a patient with suspected Discussion CRC. This has also been the conclusion of This audit has examined the non-emer- a recent study in Hawke’s Bay.16 It would gency referral by their GP of symptomatic seem sensible that this is a nationally agreed patients who are subsequently proven to format for a single referral form. have CRC. The results provide detailed Too few of our patients with CRC went information on the process from sending through the fast track 14-day pathway. This a referral letter to defi nitive diagnosis, audit has shown that if patients ‘miss’ being which in most patients is at colonoscopy. categorised for urgent 14-day fast track, they The demographics of the patients diagnosed may wait a long time for a defi nitive diag- with CRC in the Bay of Plenty are similar to nostic test. It has shown that in BOP patients the picture nationally.4 The results demon- fell broadly into either a rapid diagnosis strate that there are both national and local route or routine one and could then wait issues that need to be addressed in order to four months or more. An additional more improve this diagnostic pathway. inclusive semi-urgent (42 day) pathway

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 34 www.nzma.org.nz/journal ARTICLE

might be a reasonable solution. This type A retrospective review of the grading of pathway is detailed in the MoH bowel tool in the Auckland region also showed cancer working group ‘National Referral that patients with CRC can present with Criteria for Direct Access Colonoscopy or CT a number of symptoms and a signifi cant Colonography’ guidelines, though does not proportion of these are ‘atypical’ symptoms.9 17 seem to be a widely recognised standard. The most common single symptom in our This could pick up a considerable number audit was the passage of recognisable blood of patients who fail to reach the HIS Urgent PR (48% of all patients). This symptom alone criteria but nevertheless have worrying does not trigger the HIS Urgent pathway, ‘red fl ag’ symptoms or new iron defi ciency but if the patient has a palpable lesion on anaemia. In this audit 24.9% of CRC patients digital rectal examination (DRE) or has had PR bleeding only, 16.6% had a change in iron defi ciency anaemia or has a prolonged bowel habit only and 20.4% had IDA without change to looser bowel motions, then it does. recognisable PR blood loss and failed to More than half of the patients presenting reach the HIS level for referral. with PR blood loss had none of the above The other solution would be to broaden symptoms or fi ndings. While PR bleeding is the HIS criteria as in some other countries. a common symptom in the population, it has However, there has been criticism of this to be considered signifi cant in those over because large numbers of patients undergo 50 years of age if it is a new symptom and diagnostic tests when they do not have CRC. other benign causes have been excluded. It 18–22 This audit suggests the New Zealand was of note that 42% of our patients with HIS criteria have been set too stringently PR bleeding did not have a recorded DRE and not enough patients are undergoing in their referral form and there were a few fast track diagnosis. While it would not be cases were a palpable lesion was almost possible to make the HIS criteria inclusive certainly missed. We feel this reinforces for all patients, we suggest that they do need the need for a structured referral form that to be reconsidered. We make this statement prompts the inclusion of this important part with the proviso that we have not been of the work up. able to audit the workload implications of The next most common symptom was a relaxing the HIS criteria within this audit. change in bowel habit to a looser stool or As such an action would produce further more frequent motions (42% of all patients). capacity issues for colonoscopy services, a A number of previous studies have shown prospective study is probably the best way that a change in bowel habit is associated of establishing the risk to benefi t ratio of with a diagnosis of CRC, but few have estab- changing the HIS criteria. lished a high enough positive predictive As the considerable majority of asymp- value to warrant investigation as it is such a tomatic patients do not have CRC, other common symptom.24 The positive predictive diagnostic options for improving triage value for patients of all ages has been found may need to be considered. This might to be less than 5%,25 though may be higher also include the use of immunochemical for older patients. Change in bowel habit based faecal occult blood testing (iFOBT).23 to increasing bowel motions is of greater A number of other studies have looked to predictive strength than passing harder or improve the sensitivity of triage for referral less frequent motions.26 It seems unlikely on the basis of presenting symptoms and that this symptom alone could be used as fi ndings.24–26 part of the HIS Urgent pathway. In 2008, Canterbury DHB created a novel We found that 32% of our CRC patients pathway (CCrSP) to identify those symp- had evidence of iron defi ciency anaemia at tomatic individuals at greater risk of CRC, the time of referral. This is similar to the providing them with priority access to fi ndings in a recent study from the Nether- colonoscopy or targeted CT colonography lands which found that 33.9% of patients directly from primary care. Rectal bleeding, with newly diagnosed CRC were anaemic.28 change in bowel habit to looser and unex- IDA is recognised to have a high predictive plained iron defi ciency anaemia were given value for gastrointestinal luminal cancer. greater weight in this tool.27 The UK NICE Guidelines (2018) specifi es

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 35 www.nzma.org.nz/journal ARTICLE

IDA as a criterion for Fast Track Two Week function well. We believe that such audits Wait Pathway because of the association should be incorporated into all cancer diag- with occult gastrointestinal bleeding.29 A nosis pathways to allow continued quality recent study from the UK found patients improvement. presenting with other symptoms who were anaemic were 3.2 times as likely to have Conclusions CRC as those without anaemia; this was This audit of the CRC diagnostic pathway even higher in patients with change in in the Bay of Plenty has identifi ed both local bowel habit (8.3 versus 1.8).30 We noted that and national factors that contribute to a 57 of 181 patients (31.5%) did not have a delay in time to specialist diagnosis after GP FBC included by the GP. This could poten- referral of symptomatic patients. The present tially be improved by providing a structured High Index of Suspicion Urgent 14-day referral form. pathway only identifi ed just over a third We feel that the results of our audit should of new diagnoses. We found that GPs were prompt further work into the predictive failing to identify and correctly refer nearly value of both single indicators and those a half of those who did satisfy the criteria. with combinations of symptoms and We conclude that there needs to be further fi ndings. This work should ideally be at a work on either broadening the criteria for national level. HIS Urgent referral or introducing a semi- At a DHB level, there has been consid- urgent pathway for symptoms and fi ndings erable work to improve the sensitivity with lower, but still positive, predictive and the timeliness of the CRC diagnostic value. A standardised referral form which pathway. The results of this audit have prompts the inclusion of all required infor- allowed a clearer understanding of the mation and fi ndings would also potentially parts of the diagnostic process that do not improve the referral pathway.

Competing interests: Nil. Author information: Ada Yee, Surgical Registrar, Department of General Surgery, Tauranga Hospital, Tauranga; Tara Linton, Surgical Registrar, Department of General Surgery, Tauranga Hospital, Tauranga; Mei-Sze Lee, Surgical Registrar, Department of General Surgery, Tauranga Hospital, Tauranga; Simon Raimes, Consultant Surgeon, Department of General Surgery, Whakatane Hospital, Whakatane. Corresponding author: Mr Simon Raimes, Consultant Surgeon, Department of General Surgery, Whakatane Hospital, Stewart St, Whakatane 3120. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7824

REFERENCES: 1. Ministry of Health NZ. able from: http://www.wcrf. 4. Firth MJ, Sharples KJ, New Cancer Registrations org/int/cancer-facts-fi gures/ Hinder VA, et al. Methods 2015. 2015 [21st Mar data-specifi c-cancers/ of a national colorectal 2018]; Available from: colorectal-cancer-statistics cancer cohort study: the http://www.health.govt. 3. Arnold M, Sierra MS, PIPER Project. N Z Med nz/publication/new-can- Laversanne M, et al. J. 2016; 129:25–36. cer-registrations-2015 Global patterns and 5. Tiong J, Gray A, Jackson C, 2. International WCRF. trends in colorectal cancer et al. Audit of the associ- Colorectal Cancer Statistics. incidence and mortality. ation between length of 2012 [21st Mar 2018]; Avail- Gut. 2017; 66:683–91. time spent on diagnostic

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 36 www.nzma.org.nz/journal ARTICLE

work-up and tumour stage 14. Singh H, Khan R, Giardina identifi cation of colorectal in patients with symptom- TD, et al. Post-referral cancer. Eur J Gastroenterol atic colon cancer. ANZ J colonoscopy delays in diag- Hepatol. 2014; 26:1408–14. Surg. 2017; 87:138–42. nosis of colorectal cancer: 23. Kaul A, Shah A, Magill FH, 6. Hill S, Sarfati D, Blakely a mixed-methods analysis. et al. Immunological faecal T, et al. Survival dispar- Qual Manag Health occult blood testing: a ities in Indigenous and Care. 2012; 21:252–61. discriminatory test to iden- non-Indigenous New 15. Murray M, Brown J, Hinder tify colorectal cancer in Zealanders with colon V, et al. The colorectal symptomatic patients. Int J cancer: the role of patient cancer patients’ journey: Surg. 2013; 11:(4):329–31. comorbidity, treatment the Auckland region. N Z 24. Astin M, Griffi n T, Neal and health service factors. Med J. 2011; 124:18–28. R, et al. The diagnostic J Epidemiol Community 16. Ling E. Improving colonos- value of symptoms for Health. 2010; 64:117–23. copy referral information colorectal cancer in 7. New Zealand Guidelines from primary health care primary care: a systematic Group. Suspected Cancer providers in a peripheral review. Br J Gen Pract. in Primary Care; guidelines hospital. Bay of Islands, 2011; 61(586):e231–e243. for investigation, referral New Zealand: New 25. Shapley M, Mansell and reducing ethnic Zealand Association of G, Jordan J, Jordan K. disparities. In: NZ MoH, General Surgeons; 2018. Positive predictive values editor. Wellington2009. 17. National Bowel Cancer ≥5% in primary care 8. Hamilton W, Sharp D. Diag- Working Group Documents. for cancer: systematic nosis of colorectal cancer in National referral criteria review. Br J Gen Pract. primary care: the evidence for direct access outpatient 2010; 60(578):e366–e377. base for guidelines. Fam colonoscopy. Ministry 26. Thompson MR, O’Leary DP, Pract. 2004; 21:99–106. of Health; Nov 2015. Flashman K, et al. Clinical 9. Hsiang JC, Bai W, Lal D. 18. Couch DG, Murphy JH, assessment to determine Symptom presentations Boyle KM, Hemingway the risk of bowel cancer and other characteristics of DM. Straight to fl exible using Symptoms, Age, colorectal cancer patients sigmoidoscopy: rational- Mass and Iron defi ciency and the diagnostic perfor- ization of 2-week wait anaemia (SAMI). BJS. mance of the Auckland referrals in suspected 2017; 104:1393–1404. Regional Grading Criteria colorectal cancer. Colorec- 27. Sanders AD, Stevenson C, for Suspected Colorectal tal Dis. 2015; 17:980–3. Pearson J, et al. A novel Cancer in the South 19. Aslam MI, Chaudhri S, pathway for investigations Auckland population. N Z Singh B, Jameson JS. of colorectal symptoms Med J. 2013; 126:95–107. The “two-week wait” with colonoscopy or 10. National Institute for referral pathway is not computer tomography Health and Care Excel- associated with improved colonography. NZMJ. lence. Referral Guidelines survival for patients with 2013; 126:45–57. for Suspected Cancer colorectal cancer. Int J 28. Wilson, MJ, Dekker JWT, in Adults and Children. Surg. 2017; 43:181-5. Harlaar JJ, et al. The role NICE; 2015. p. 16–7. 20. Banerjea A, Voll J, of preoperative iron 11. Benton S, Steele R, Logan R, Chowdhury A, et al. defi ciency in colorectal et al. NICE referral guide- Straight-to-test colonoscopy patients: prevalence and lines for suspected cancer: for 2-week-wait referrals treatment. Int J Colorec Dis. colorectal cancer and faecal improves time to diagnosis 2017; 32(11):1617–1624. occult blood testing. Ann of colorectal cancer and 29. National Institute for Clin Biochem. 2016; 53:7–9. is feasible in a high-vol- Healthcare Excellence. 12. New Zealand Guide- ume unit. Colorectal Suspected Cancer: Recogni- lines Group. Suspected Dis. 2017; 19:819–26. tion and referral. Primary cancer in primary care: 21. Currie AC, Evans J, Smith care investigation fi ndings. guidelines for inves- NJ, et al. The impact of the London. NICE; 2018 tigation, referral and two-week wait referral 30. Mashlab S, Large P, Laing reducing ethnic disparities. pathway on rectal cancer W, et al. Anaemia is a Wellington: New Zealand survival. Colorectal risk stratifi cation tool for Guidelines Group; 2009. Dis. 2012; 14:848–53. symptomatic patients 13. Ministry of Health. Faster 22. Patel RK, Sayers AE, Seedat referred via the two week Cancer Treatment: high S, et al. The 2-week wait wait pathway for colorectal suspicion of cancer defi ni- service: a UK tertiary cancer. Annals of RCSE. tions. Wellington: Ministry colorectal centre’s 2018; 100(5):350– 356. of Health; April 2016. experience in the early

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 37 www.nzma.org.nz/journal ARTICLE

Developing a regional cancer service; lessons from the Waitemata Cancer Patient Experience Survey Jonathan Koea, Sue French, Karen Hellesoe, Peter Sandiford

ABSTRACT AIM: As part of a project developing cancer service capability, the National Health Service (NHS) Cancer patient experience survey was used to assess the currently available services at Waitemata District Health Board (WDHB). METHODS: Patients presenting with cancer to WDHB in the previous 12 months were mailed a copy of the survey, to the initial cohort in 2013 and to the second in 2015. Results were compared between survey periods and with the 2015 NHS Cancer Patient Experience Survey. RESULTS: In 2013, 329 patients completed the survey while 319 responded in 2015. Over 90% of patients classed their experience as good or excellent in both survey periods and comparison showed significant overall improvement (p=0.001) in patient experience between 2013 and 2015. Overall, WDHB benchmarked with the NHS experience but the NHS performed better than WDHB in cancer nurse specialist contact for ongoing support and information related to eligibility for financial assistance. CONCLUSION: The results of these confirm that the NHS Cancer Patient Experience Survey provides meaningful data within New Zealand and that WDHB cancer patient experience has improved over the survey period. This data has assisted WDHB in prioritising patient information resources and treatment planning in developing services.

aitemata District Health Board and the initiative to establish local delivery (WDHB) provides care for a pop- of infusional chemotherapy in the Auckland Wulation of 597,500, of whom one region to help minimise patient travel and third are over 50 years of age in the north- expense, a small working group was estab- west of Auckland city. It is New Zealand’s lished to develop service capability to ensure largest single health provider catchment.1 that WDHB was compliant with national FCT While surgery for a number of cancers (gas- targets and to meet service needs for future trointestinal, breast, skin, head and neck and on-site infusional chemotherapy. urological) have been performed at WDHB Early in this process it was appreciated that there has not been a discrete cancer service, the development of any new cancer service and patients requiring chemotherapy and presented a unique opportunity to consult radiation therapy were treated at nearby directly with patients and to determine Auckland City Hospital. This often involves areas of patient satisfaction and dissatis- prolonged travel times and stress for pa- faction with the current service delivery. As tients and whanau, and is associated with in- an initial enquiry into the needs of WDHB creased patient expenditure related to trans- cancer patients it was elected to administer port costs and parking as well as increased the National Health Service Patient Expe- time away from income earning activities. rience Survey3 to a large cohort of cancer However, with the release of the national patients undergoing or recently fi nished with 2 faster cancer treatment (FCT) targets in 2013 treatment. This survey was chosen since it

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 38 www.nzma.org.nz/journal ARTICLE

has been validated for use in cancer patients. boards (Counties, Auckland City and Wait- Potential comparison of results obtained emata). Following consultation, the survey by ourselves with those in the National was reviewed and approved by the Māori, Health Service (NHS) was feasible and it has Pacifi c and Asian Health Services at WDHB. also been used successfully in a number Survey data was analysed descriptively of sites outside of the UK, notably Victoria, with medians and Wilson’s 95% confi dence 4 Australia. The survey was initially adminis- intervals determined using Excel (Microsoft, tered in March 2013 when the working group California, US). Comparisons were made was fi rst established and at the beginning of for individual questions between the NHS the introduction of oncology clinical nurse survey results for 20155 and the WDHB specialists (CNS) for each tumour stream. It results for each of the two survey periods. A was subsequently readministered in May comparison was also made between the two 2015. This investigation summarises the WDHB survey periods. Given the number results of the survey and the impact CNS had of individual comparisons being made, on cancer patient experience. the false discovery was set at 5% using the Benjamini-Hochberg method.6 No attempt Methods was made to adjust for differences in case Surviving patients from all tumour mix between the surveys. In addition to the streams and resident in the WDHB comparisons on individual questions, a sign 7 catchment, who presented for the fi rst time test was performed to compare the overall with a diagnosis of cancer in the previous performance of Waitemata with the NHS in 12 months, were sent a paper copy of the each time period, and to determine whether National Health Service Cancer Patient there was a signifi cant change between Experience Survey3 with a prepaid return the two surveys. This was done by simply envelope. Two postal reminders were sent at counting the number of questions in which four and eight weeks and all patients were one survey performed better than the other given 12 weeks to complete and return the (positive signs) and vice versa (negative survey. All responses were anonymous. The signs), independently of the magnitude of results were then collated and entered onto this difference. an electronic spreadsheet for analysis. The 2013 NHS survey was used in 2013 and the Results 2015 version in 2015. Both surveys were the Survey administration same and administered in the same way. The survey was administered in 2013 and Patients were identifi ed from WDHB again in 2015. In 2013, surveys were sent to records and these were correlated with 439 patients of who 329 (75%) responded. In National Cancer Registry data to ensure that 2015, survey forms were sent to 520 patients only patients with a defi nitive diagnosis of of whom 319 (61%) patients responded. cancer were included. Similarly, this tech- Ethnicity of the respondents was recorded nique was used to ensure that only patients in both surveys with the majority being New presenting to WDHB for the fi rst time with a Zealand European (74.4% 2013, 78% 2015), diagnosis of cancer (either localised or meta- Māori (3.9% 2013, 2.7% 2015), Pacifi c (6.2% static) were included. Patients in the WDHB 2013, 3.3% 2015), Chinese (3.6% 2013, 2.7% catchment who had cancers diagnosed and 2015), Indian (1.1% 2013, 1.7% 2015). treated exclusively in the private system were not included and those patients with Demographics Demographically, 60% of respondents a cancer diagnosis who were managed with were female in 2013 and 55% in 2015. The best supportive care only were not included. median age of responders was 66 years This project was registered with the WDHB (range 30–87 years) in 2013 and 69 years Awhina Research and Knowledge Centre (range 23–79 years) in 2015. Eighty-seven and approved by the Northern A Regional percent of patients were treated in a public Ethics Committee (ref 13/NTA/127 number hospital in both 2013 and 2015, and 70% 16776). Written approval was obtained and 60% respectively did not have any long- from the Chief Executive Offi cers of the standing disability. three Auckland metropolitan district health

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 39 www.nzma.org.nz/journal ARTICLE

Figure 1: Domains where the performance of the National Health Service was signifi cantly worse than Waitemata District Health Board in 2013 and 2015 (proportion of positive responses with 95% confi - dence intervals).

Reported outcomes by question Summary comparison of surveys The survey outcomes are summarised In 2014/15, of the 34 comparable ques- in Figures 1–4 and Table 1, comparing tions, there was a higher proportion of WDHB outcomes across the two surveys positive responses in the WDHB survey with those of the NHS. In comparison to the than the NHS in 19 questions and a lower NHS 2015 survey, WDHB had signifi cantly proportion in 15. This difference was more positive responses in fi ve questions not statistically signifi cant. In 2013/14, in 2013/14 and four questions in 2014/15. 19 questions had a higher proportion of There were more positive responses in the positive responses in the NHS survey and NHS survey in eight and four questions 15 in the WDHB survey, again not a signif- respectively. icant difference. However, comparing

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 40 www.nzma.org.nz/journal ARTICLE

Figure 2: Domains where the performance of the National Health Service in 2015 was signifi cantly better than Waitemata District Health Board in 2013 and 2015 (proportion of positive responses with 95% confi dence intervals). CNS: Cancer nurse specialist.

results from 2013/14 survey with those deliberately global in scope and there was from 2014/15 there were 47 questions no intention of analysing the performance where the proportion of positive responses of separate tumour streams although that increased in second period versus 21 where work will be undertaken in future surveys. the proportion declined. The sign test for Overall, all the measures of patient expe- this difference was statistically signifi cant rience at WDHB improved between the (p=0.001) suggesting that the quality of 2013 and 2015 survey. The only signifi cant cancer care in Waitemata improved over decline between the two surveys was in time, despite the fact that few of these the proportion of patients invited to partic- differences in positive response rate were ipate in cancer research which decreased statistically signifi cant at the level of indi- from 35.9% to 21.4% (p<0.001: Figure 4). vidual questions. Increasing patient access to clinical trials is a future regional priority for Auckland. A Discussion regional tissue bank has been established as The NHS Cancer Patient Experience having Phase 1 clinical trials unit at Auckland survey was utilised to assess the status of the City Hospital and it is hoped that these, and services available to WDHB cancer patients other initiatives, will increase the enrolment and to help defi ne and prioritise service of patients in clinical research protocols. improvements. The NHS Cancer Patient Measures showed that service provision Survey was utilised because it has been metrics (eg, satisfactory wait for outpatient validated in patients with a wide range of assessment, surgery date not cancelled) cancers and widely used in the UK3 as well were at high levels and communication as in other health constituencies.4 Unfortu- of investigation results, diagnosis and the nately no similar measuring tool designed results of surgery were well performed. specifi cally for New Zealand exists, although However, there was a lack of written this is planned.8 Use of this survey also material pertaining to both diagnostic permitted limited comparison and bench- investigations and surgical procedures. marking of WDHB services against the Similarly, few patients received a written National Health Service. The survey was summary of treatment or treatment plan

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 41 www.nzma.org.nz/journal ARTICLE

Table 1: Domains where there was no signifi cant difference between the National Health Service and Waitemata District Health Board (proportion of positive responses with 95% confi dence intervals). CNS: Cancer nurse specialist.

Question NHS 2015 WDHB 2013 WDHB 2015 % 95% C.I. % 95% C.I. % 95% C.I. I saw my GP no more than two times about a health 81.9 (81.6–82.2) 79.5 (74.4–83.8) 80.0 (75.1–84.2) problem related to my cancer

I completely understood the explanation of what was 73.1 (72.7–73.4) 74.1 (68.8–78.9) 71.4 (66.1–76.2) wrong with me

I was given written information about the type of 71.6 (71.3–72.0) 71.2 (65.4–76.3) 72.2 (66.7–77.1) cancer and it was easy to understand

The possible side e ects of treatment were explained 72.7 (72.4–73.1) 70.5 (64.7–75.6) 76.4 (71.0–81.0) to me in a way that I could definitely understand

From my CNS I have always or mostly had answers 88.5 (88.2–88.8) 87.2 (81.6–91.2) 91.4 (86.9–94.5) that I could understand to my important questions

Hospital sta gave me information about how to get 54.7 (54.2–55.2) 50.6 (43.1–58.1) 60.6 (53.1–67.6) financial help or benefits that I might be entitled to

A er the operation a sta member explained how it 77.6 (77.2–78.0) 72.9 (66.5–78.5) 75.3 (68.7–80.9) had gone in a way that I could completely understand

I had confidence in all the doctors treating me 84.1 (83.7–84.4) 85.4 (80.1–89.5) 89.4 (84.1–93.0)

Doctors did not talk in front of me as if I weren’t there 81.3 (81.0–81.7) 78.5 (72.6–83.5) 75.5 (68.9–81.1)

My family and those close to me definitely had enough 72.1 (71.6–72.5) 66.0 (58.9–72.4) 68.8 (61.2–75.4) opportunity to speak to a doctor

I had confidence and trust in all the ward nurses 72.4 (72.0–72.8) 69.5 (63.2–75.2) 74.9 (68.2–80.5) treating me

There were always or nearly always enough nurses on 66.1 (65.7–66.6) 68.2 (61.8–74.0) 66.8 (59.8–73.2) duty to care for me in hospital

I was always given enough privacy when discussing 84.8 (84.4–85.1) 82.2 (76.6–86.7) 85.6 (79.8–89.9) my condition or treatment

The hospital sta always did everything they could to 83.9 (83.5–84.2) 86.1 (80.5–90.3) 84.6 (78.4–89.3) help control my pain

Hospital sta told me who to contact if I was worried 93.7 (93.4–93.9) 95.3 (91.6–97.4) 93.4 (88.8–96.2) about my condition a er I le hospital

As far as I know my GP was given enough information 95.4 (95.2–95.5) 93.1 (89.2–95.6) 96.4 (93.5–98.0) about my condition and the treatment I had at the hospital

The di erent people treating and caring for me defi- 60.7 (60.3–61.0) 64.3 (58.5–69.8) 65.8 (60.2–70.9) nitely/always work well together to give me the best possible care

I have been given a care plan 33.0 (32.6-33.4) 33.8 (28.0–40.0) 34.5 (28.7–40.8)

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 42 www.nzma.org.nz/journal ARTICLE

Figure 3: Domains where the performance of the National Health Service was signifi cantly better than Waitemata District Health Board in 2013 but not in 2015 (proportion of positive responses with 95% confi dence intervals).

and development of a written or digital focus. Since the 2015 survey, cancer-spe- resource that could be used by patients to cifi c social workers and psychologists have review issues pertaining to specifi c investi- been appointed to WDHB to work in this gational procedures as well as summarising area. Similarly, a number of WDHB patients treatment plans and steps, is being indicated that transport issues and living considered. This fi nding was previously situation were not taken into account when highlighted in the only national survey of clinical staff planned treatment. Since travel New Zealand cancer patients9 and also by by road in Auckland is often time-con- our own surgical department in relation to suming and expensive, these factors should the specifi c needs of patients with upper be central to any decisions made around gastrointestinal cancers.10 As yet, New treatment location and timing. Zealand has no national information policy When compared to the NHS, WDHB for cancer patients and the provision of recorded improved performance in areas information resources is undertaken by related to patient communication and individual district health boards and is care (Figure 1). The two WDHB survey largely paper based. periods bridged the introduction of tumour In comparison to the NHS, WDHB under- stream-specifi c CNS at WDHB. The WDHB performed in specifi cs around cancer nurse surveys indicate that, at times, patients were specialist (CNS) contact for ongoing support not always made aware of their specifi c CNS and imparting of information related to or their contact details. CNS have now been eligibility for fi nancial assistance from a working at WDHB for over four years and high user pharmaceutical card (Figure 2). are now an indispensable part of patient This emphasises the importance of prac- treatment pathways. A further survey of tical issues for patients with cancer and WDHB patients is being undertaken in how these are often overlooked by hospital 2018 to assess whether there has been staff with either a clinical or a service improvement in this measure.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 43 www.nzma.org.nz/journal ARTICLE

Figure 4: Domains where the performance of Waitemata District Health Board was signifi cantly better than the National Health Service in 2013 but not 2015.

Overall the response rate for both WDHB quantify the proportion of responses by surveys was good. However, the surveys can tumour stream or clinical stage. Future be criticised for not specifi cally addressing surveys will look to obtain this data. the experience and health needs of Māori Finally, over 90% of WDHB patients indi- and Pacifi c people. The nominated ethnic cated that their experience was good or response rates from these groups were excellent and is comparable to the reported signifi cantly lower than the expected popu- experience from the UK. The survey lations in the WDHB catchment area. This is responses have assisted WDHB in priori- concerning since Māori have a higher inci- tising the following initiatives: dence of a number of cancers (eg, lung and • Development of more comprehensive upper gastrointestinal) and notably poorer disease-specifi c patient information outcomes than non-Māori. Recent research resources and a formal written care has shown that Māori patients value plan for each patient to facilitate Māori-specifi c clinical support and assis- understanding of diagnosis and tance with fi nancial and transport issues treatment. is more likely to be required.11 WDHB has recently employed a dedicated with Māori • Development of treatment site-spe- CNS and aim to strengthen relationships cifi c information particularly for with primary health care providers to assist those patients requiring treatment Māori cancer patients. at multiple sites within the Auckland metropolitan area. An important limitation of this analysis is that it has not made any adjustment for • Increasing patient access to clinical differences in the case mix between the trials and research protocols. two survey periods, or between the WDHB • Ensuring there is adequate resource and NHS surveys. Patient demographics available (CNS, psychology and social including age, gender and ethnicity were work full-time equivalents) to guide similar in both the 2013 and 2015 surveys. and assist patients through their However, the survey responses were anony- treatment journey. mised and it was therefore not possible to

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 44 www.nzma.org.nz/journal ARTICLE

Competing interests: Nil. Acknowledgements: The work of Sandra Scheene is acknowledged in administering the questionnaire to patients and ensuring the responses were collated. Author information: Jonathan Koea, Faster Cancer Treatment Service, Waitemata District Health Board, Takapuna, Auckland; Sue French, The Centre for Innovation and the Faster Cancer Treatment Service, Waitemata District Health Board, Takapuna, Auckland; Karen Hellesoe, Faster Cancer Treatment Service, Waitemata District Health Board, Takapuna, Auckland; Peter Sandiford, The Centre for Innovation and the Faster Cancer Treatment Service, Waitemata District Health Board, Takapuna, Auckland. Corresponding author: Jonathan Koea, Upper Gastrointestinal Unit, Department of Surgery, North Shore Hospital, Private Bag 93503, Takapuna, Auckland 0620. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7825

REFERENCES: 1. http://www.waitemata- multiple testing. Journal of in upper gastrointestinal dhb.govt.nz/ Accessed the Royal Statistical Society cancer: What patients June 26, 2018 Series B (Methodological) and their families want 2. http://www.health.govt.nz/ 1995:289–300. http://www. to know. ANZ J Surg 2018; our-work/diseases-and-con- jstor.org/stable/2346101?se- Early view publication 12 ditions/cancer-programme/ q=1#page_scan_ June 2018. DOI: 10.1111/ faster-cancer-treat- tab_contents ans.14565. http://www. ment-programme 7. Armitage P, Berry G. ncbi.nlm.nih.gov/ Accessed June 26, 2018 Statistical methods in pubmed/?term=9.%09O- tutaha+B%3B+Srini- 3. http://www.quali- clinical research. Third vasa+S%3B+Koea+JB.+In- ty-health.co.uk/surveys/ Edition. Blackwell formation+needs+in+up- national-cancer-pa- Scientifi c Publications per+gastrointestinal+- tient-experience-survey Oxford 1994 pp 449–450. cancer%3A+What+pa- Accessed June 26, 2018. 8. http://www.hqsc.govt. tients+and+their+fami- nz/our-programmes/ 4. http://www.health.vic. lies+want+to+know.+ANZ+- health-quality-evaluation/ gov.au/about/health-strat- J+Surg+2018%3B+Ear- projects/health-quali- egies/cancer-care/ ly+view+publica- ty-and-safety-indicators/ cancer-projects/victori- tion+12+June+2018.+- patient-experience/ an-cancer-patient-experi- DOI%3A+10.1111%- Accessed June 26, 2018. ence-survey-tool-project 2Fans.14565 Accessed June 26, 2018 9. O’Brien I, Britton E, Sarfati 11. Slater T, Matheson A, D, Naylor W, et al. The 5. http://www.england.nhs. Davies C, Tavite H, et voice of experience: results uk/statistics/wp-content/ al. It’s whanaungatanga from Cancer Control New uploads/sites/2/2016/0 and all that kind of stuff: Zealand’s fi rst national 6/20160713-CPES-2015- Māori cancer patients cancer care survey. N Z Presentation_revised.pdf experiences of health Med J 2010; 123(1325); Accessed June 26, 2018 services. J Primary Health http://www.nzma.org.nz/ 6. Benjamini Y, Hochberg Care 2013; 5:208–314. journal/123-1325/4419/ Y. Controlling the false h ttp://www.ncbi.nlm.nih. Accessed June 26, 2018. discovery rate: a practical gov/pubmed/24294619 and powerful approach to 10. Otutaha B, Srinivasa S, Koea JB. Information needs

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 45 www.nzma.org.nz/journal ARTICLE

New Zealand Society of Gastroenterology Guidelines on Therapeutic Drug Monitoring in In ammatory Bowel Disease Afrasyab Khan, Arvenia B Berahmana, Andrew S Day, Murray L Barclay, Michael Schultz

ABSTRACT The incidence of inflammatory bowel disease (IBD) in New Zealand has increased over the last several decades. The management of IBD has been transformed since the introduction of monoclonal antibody drugs. Other medications used in the treatment of IBD include amino-salicylates, steroids, thiopurines and methotrexate. Therapeutic drug monitoring (TDM) involves the measurement of serum drug levels or active metabolites and anti-drug antibodies. TDM is essential for a personalised approach to the management of patients with IBD and is used to optimise drug e icacy and reduce the risk of toxicity. In IBD, TDM can be used for checking adherence, evaluating drug toxicity, identifying hypermethylators, assessing loss of response and in decisions regarding treatment escalation or de-escalation. Management decisions in patients on a thiopurine are facilitated by checking TPMT enzyme activity and thiopurine metabolite levels. Measurement of drug trough levels and anti-drug antibodies can result in individualised treatment decisions in patients on biologics. In addition to using TDM in patients who fail therapy, proactive TDM can potentially facilitate early treatment decisions, albeit more work is needed in this area. The clinical benefits of reactive TDM are well documented and this has been shown to be cost e ective. Studies have shown that combination therapy in patients on a biologic leads to better clinical outcomes. E ective use of drugs in the treatment of IBD is even more imperative in the New Zealand setting due to relatively fewer options of funded tr eatment, and the limitations on the use of available drugs. This document represents the current guidelines of the New Zealand Society of Gastroenterology on TDM in IBD.

he incidence of infl ammatory bowel anti-infl ammatory and systemic immu- disease (IBD) in New Zealand has risen nomodulating medications are typically Tsubstantially in the last fi ve decades.1–6 used in the treatment of IBD. These include IBD comprises Crohn’s disease (CD) and amino-salicylate preparations (5-ASA), cor- ulcerative colitis (UC) with a small group of ticosteroids, thiopurines, methotrexate and patients remaining unclassifi ed (IBDU). The monoclonal antibodies. aetiology of IBD is unknown, however it is For many years the thiopurines, azathi- recognised that environmental factors play a oprine (AZA) and 6-mercaptopurine (6MP) role resulting in an abnormal host immune have been the mainstay of IBD treatment response to luminal antigens in a genetically in those with moderate to severe disease, susceptible host leading to sustained gastro- to maintain remission and minimise corti- intestinal and systemic infl ammation with costeroid use. However, many patients are resultant bowel wall damage and increased either intolerant or fail to respond to these 7,8 risk of colorectal cancer. Locally effective drugs.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 46 www.nzma.org.nz/journal ARTICLE

The management of IBD has been trans- paucity of data. The authors were invited formed since the introduction of monoclonal based on their expertise and include adult antibodies or so-called biologic drugs. The and paediatric gastroenterologists and a two agents publicly funded in New Zealand clinical pharmacologist; all with clinical and are infl iximab (Remicade®, Janssen, research interests in infl ammatory bowel Auckland, New Zealand) and adalimumab disease. Feedback was sought from the NZSG (Humira®, AbbVie, Wellington, New executive committee members and inte- Zealand). Both are anti-tumour necrosis grated. All authors as well as NZSG executive factor inhibitors (anti-TNF). However, a committee members were in agreement number of monoclonal antibodies against regarding these recommendations. These other components of the infl ammatory local guidelines have been developed to help cascade are either in development or the effective use of available drugs keeping available for use elsewhere in the world.9 in view the availability and funding setting Therapeutic drug monitoring (TDM) in New Zealand. involves the measurement of levels of drugs or their active metabolites in blood and can Treatment with also include detection of anti-drug anti- bodies. TDM is fully funded in New Zealand thiopurines and can guide clinicians to optimise the Thiopurines are thought to specifi cally drug dosage regimen to increase effi cacy target Rac1 activation, which is key for and reduce risk of drug toxicity. TDM can activating CD4-positive T-lymphocytes. also explain treatment failure at times and The prodrug AZA is metabolised to 6-MP, result in appropriate switching of drug another prodrug, which is then metabolized class. Measurement of drug levels in New by three different pathways as shown in Zealand is routinely available for AZA, 6MP the abridged Figure 1. The main products and anti-TNF drugs. Antibodies to anti-TNF of each of these three pathways are drugs (anti-drug antibodies: ADA) can also 6-thioguanine nucleotide (6TGN), 6-methyl- be measured. TDM can be undertaken at mercaptopurine (6MMP) and thiouric acid. regular intervals or in specifi c clinical situ- Thiouric acid is excreted in the urine. 6TGN ations, such as when the drug appears to be are the main active metabolites that are failing, as it is based on the understanding thought to be responsible for therapeutic that there is a relationship between disease response. Because 6TGN takes time to accu- outcomes and adequate drug exposure. mulate in cells, this contributes towards the slow therapeutic onset of thiopurines Due to the limited number of biologics (10–12 weeks or even longer). However, available in New Zealand, optimum and adequate treatment with thiopurines has effective use of available drugs is of been shown to induce remission in 49% the utmost importance. The executive of CD and 42% of UC patients with a low committee of the New Zealand Society of rate of loss of response.16 In addition to Gastroenterology (NZSG) identifi ed the need the commonly used AZA/6MP, thioguanine for such a guideline to support clinicians in (TG) is a non-conventional thiopurine that their management decisions by examining evades several intermediates in the thio- published literature regarding TDM.10–15 The purine enzymatic pathway and is converted guideline took the Australian statement as directly to 6TGN nucleotide.17 the framework and adapted this to the New Zealand setting. Literature search focused Thiopurine side effects can be dose-de- on studies published regarding treatment pendent or dose-independent (idiosyncratic) options in IBD (specifi cally infl iximab and (Table 1). Up to 50% of individuals treated adalimumab among the biologics as these with thiopurines are reported to have are available in New Zealand) and the effect side effects, but this is variable depending of TDM on clinical outcomes. Confl icts of on the study. Cessation of thiopurines in interest—if any—were disclosed by the 25.6% of patients due to adverse effects was 18 authors, and disagreements regarding reported in a New Zealand study. There are specifi c guidelines were resolved in light concerns regarding nodular regenerative of published literature pertaining to that hyperplasia (NRH) in patients on TG and specifi c guideline. Clarifi cation has been the reported rates are 0 to 62%; however, given in specifi c sections where there is a the higher rates reported in earlier studies

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 47 www.nzma.org.nz/journal ARTICLE

Figure 1: Altered metabolic pathway after co-administration of allopurinol with preferential metabo- lism to TGNs.

Allopurinol inhibits XO and TPMT. Thiopurinemethyltransferase : TPMT. Xanthine oxidase : XO. 6MMP: 6-methyl mercaptopurine. HPRT: Hypoxanthine phosphoribosyl transferase. IMPDH: Inosine monophosphate dehydrogenase. GMPS: guanosine monophosphate synthetase. TGN: thioguanine nucleotides. may be due to high (>40mg/day) dosing and establish TPMT enzyme activity before signifi cant previous patient exposure to initiating thiopurines, however starting other thiopurines. Recent research shows at a low dose (eg, 50mg once a day of AZA that NRH is rare with thiopurine treatment, or 25mg once a day of 6MP) is safe while including TG, and that the clinical course is monitoring for myelosuppression on a often benign and pathology is reversible.19 weekly basis until TPMT activity results are Testing for thiopurine methyltransferase available. It is noteworthy that blood trans- (TPMT) enzyme activity is recommended fusion within three months prior to TPMT 29 in all patients commencing a thiopurine. testing can affect TPMT activity testing. In One in every 200–300 patients have TPMT these patients, careful dose escalation with defi ciency and these patients universally weekly monitoring for myelosuppression have severe leucopenia with standard and repeat TPMT enzyme testing three dosing. TMPT testing and adjustment of months after the last transfusion is recom- medication dosage markedly reduces this mended. TPMT genotype testing can be done risk. Knowledge of TPMT activity testing is in patients with a recent blood transfusion 29 also useful to establish the correct dosage in to check their TPMT status. However, the patients with intermediate or low enzyme genotype test detects a limited number of activity.27,28 Good practise would be to mutations affecting enzyme activity and is not a substitute for close monitoring.30

Table 1: Side effects of thiopurines.18,20–26

Side-e ects of thiopurines Approximate incidence Gastrointestinal symptoms 5–24% Headache 1% Fatigue or myalgias 7.1% Allergic reactions 3–6.5% Myelotoxicity 1.8–3.7%

Infections 1.1%

Hepatitis 3.4–8.8% Pancreatitis 1.8–6.2%

Nodular regenerative hyperplasia 1.28–1.5% at 10 years

Lymphoproliferative disorder Standardised incidence ratio of 5.7*

Non-melanoma skin cancer Standardised incidence ratio of 7.0†

*The absolute risk varies between 1 in 5,000 for patients between 20–30 years old and 1 in 500 for those more than 70 years old. †Risk increases with age, duration and ongoing use of thiopurines.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 48 www.nzma.org.nz/journal ARTICLE

Table 2: Dosage of thiopurine depending on TPMT enzyme activity.106,107

TPMT enzyme AZA dose 6MP dose TG dose activity Normal Start with normal dose Start with normal dose Start at 0.3mg/kg/day - not (2–3mg/kg/day). (1–1.5mg/kg/day). exceeding 25 mg/day in adults.

Intermediate Start with 33–50% of Start with 33 - 50% of Start with 33–50% of target target dose target dose dose. (eg, 1–1.5mg/kg/day). (eg, 0.5–0.75mg/kg/ Consider thrice a week dosing. day).

Low or absent Start with 10% of Start with 10% of target Start with 10% of target dose. target dose. dose. Alternate days or Alternate days or thrice a week Alternate days or thrice a week dosing is dosing is recommended. thrice a week dosing is recommended. recommended.

Escalate the dose of AZA and 6MP as required allowing time to reach steady state before dose change (two weeks in normal and intermediate TPMT activity; and four weeks in low TPMT activity). Conversion factor of 0.5 can be used for converting AZA dosage to 6MP. TPMT: thiopurine methyltransferase. AZA: Azathioprine. 6MP: 6-mercaptopurine. TG: thioguanine.

Adequate thiopurine dosing target range of 235–450 pmol/8x108 RBC In the 90% of patients with normal TPMT with AZA or 6MP. TG should be discontinued activity, target doses will usually be in the if liver enzymes, especially alkaline phos- order of 2–3mg/kg for AZA and 1–1.5mg/kg phatase, increase two-fold, if the leukocyte 9 for 6MP, with dose adjustment based on thio- count decreases below 1x10 /L or platelet purine metabolite concentrations as below. count falls signifi cantly. In the past, routine Some clinicians prefer to start patients on liver biopsies were recommended (due to a lower dose and increase over time in the fear of nodular regenerative hyperplasia). hope of reducing the risk of toxicity and However, liver biopsies are now performed intolerance but evidence for this is poor.31 only in cases of suspected liver involvement Treatment initiation with 33–50% of the (increased liver enzymes, decreased platelet target dose is recommended in patients with count) and/or signs of (non-cirrhotic) portal intermediate TPMT enzyme activity and hypertension (NCPH). As observed in the at 10% of the target dose in patients with largest case series (N=111), nodular regen- low or absent enzyme activity (Table 2).31 erative hyperplasia was not increased in TG After commencing thiopurines, full blood users (6%) and was not associated with clini- 33 count and liver enzymes should be checked cally relevant liver pathology. weekly for one month, then monthly for two Although one early study showed a high months and then every three months with rate of nodular regenerative hyperplasia ongoing monitoring for adverse effects. (NRH) and hepatic fi brosis with thioguanine Using thioguanine therapy in IBD, subsequent extensive Dutch experience with TG at lower doses of around Extensive Dutch experience has enabled 0.3mg/kg/day has shown a low incidence of the following guidelines for use of TG.32 The this adverse effect and no clinically signif- dose in IBD is 0.3mg/kg/day, not exceeding icant morbidity.33,34 The prevalence of NCPH 25mg/day in adults, adjusted according due to TG and conventional thiopurines is to thiopurine metabolite concentrations. very low in the Netherlands (with over 4,000 During the fi rst year, regular blood count IBD patients on TG in 2017).19 However, and liver function tests are mandatory and treatment with TG should be discontinued in 6-TGN levels of 800–1200 pmol/8x108 RBC cases of histological proven hepatotoxicity. should be targeted, as compared to the usual

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 49 www.nzma.org.nz/journal ARTICLE

Table 3: Interpretation of thiopurine metabolite levels and recommendations for management.36–39

Thiopurine metabolite levels Interpretation Management Indiscernible or negligible 6TGN Medication non-adherence Check adherence and educate. & 6MMP

Low 6TGN Underdosing or medication Increase dose and monitor a er Low 6MMP non-adherence checking adherence.

Low 6TGN Thiopurine hypermethylator Add allopurinol and reduce thio- High 6MMP or “shunter” purine dose to 25–33%. (6TGN:6MMP>20)

Normal 6TGN* Adequate dosing Continue current treatment if qui- 6MMP<5700 pmol/8×108 RBC. escent disease or escalate therapy if active IBD.

High 6TGN Increased risk of adverse Decrease dose. Add another drug High 6MMP events if active disease.

*Normal range for 6TGN is 235 to 450 pmol/8x108 RBCs. 6TGN: 6-thioguanine. 6MMP: 6-methyl mercaptopurine. RBC: red blood cells. IBD: inflammatory bowel disease.

Monitoring of thiopurine helps guide subsequent treatment deci- metabolites sions (Figure 2). Frequency of response to thiopurines has been shown to be around Ongoing disease activity while on a thio- 78% in patients with 6TGN levels higher purine can be due to primary non-response, than 235 pmol/8x108 RBCs and can result in loss of response, medication non-adherence, steroid-free remission in approximately 65% underdosing or altered metabolism with of patients on thiopurines compared with preferential production of the hepatotoxic 41% response in those with concentrations 6MMP at the expense of low concentrations <235 pmol/8x108 RBCs.44 of the therapeutic 6TGN (Table 3). Under- dosing and medication non-adherence are Low TGN levels indicate medication important causes of ongoing active disease in non-adherence, underdosing or thiopurine patients on thiopurines.35 Active disease can hypermethylation (Table 3). In contrast, TGN be present in symptomatic as well as asymp- levels greater than 450 pmol/8x108 RBCs are tomatic patients and should be confi rmed associated with a higher risk of leukopenia with endoscopy, radiology or biochemical and myelotoxicity.40 6MMP is a potentially indices or a combination of these. toxic metabolite and is associated with hepa- totoxicity, especially at levels of greater than The correlation between weight-based 5,700 pmol/8×108 RBC.45 High 6MMP has also dosing and 6TGN levels is not strong.40 In been associated with myelotoxicity.46 addition there is ethnic variability in 6TGN levels likely due to differences in thiopurine Patients who are intolerant to AZA can metabolism, which is an area of ongoing often be successfully switched to 6-MP or research.41,42 Hence routine monitoring of thioguanine.47–49 Patients can tolerate and thiopurine metabolites is now often recom- respond to TG even if they have previously mended in all IBD patients whether they experienced pancreatitis secondary to AZA have active disease or are in remission, with or 6-MP.50 the latter to help reduce the risk of disease Thiopurine hypermethylators or 43 fl are. Monitoring of thiopurine metabolites ‘shunters’ is also helpful in identifying underdosage, About 18% of patients with normal TPMT non-adherence and those with unconven- activity can have altered drug metabolism tional thiopurine metabolism.43,44 However with thiopurine hypermethylation. These there are no randomised controlled studies patients (also known as ‘shunters’) metab- supporting routine thiopurine TDM in IBD. olise AZA or 6-MP preferentially to the toxic In patients with active IBD or those with metabolite 6MMP instead of 6TGN (Figure adverse effects suspected to be secondary 1). Hence the level of 6TGN is low and that of to thiopurine toxicity, metabolite testing

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 50 www.nzma.org.nz/journal ARTICLE

Figure 2: Algorithm for initiation and monitoring of treatment with thiopurines in patients with infl ammatory bowel disease (IBD).

TPMT: thiopurine methyltransferase. 6MMP: 6-methyl mercaptopurine. 6TGN: 6-thioguanine. TG: thioguanine.

6MMP in these patients is high. These pref- erential 6MMP producing patients typically Treatment with have a ratio of 6MMP to 6TGN of >20.51 This biologics ratio can be corrected by the addition of allo- Biologic or monoclonal antibody drugs are purinol 50–100mg daily and reduction of the protein-based treatment modalities, usually 52,53 thiopurine dose to 25–33% of baseline. IgG molecules, that bind to their antigen For example, a 75kg patient may receive with high specifi city. Anti-TNFα antibodies 100mg allopurinol and 50–100mg AZA. have been used in the treatment of IBD After commencement of allopurinol, it is since the late 1990s and were introduced in recommended that thiopurine metabolites New Zealand in the early 2000s. TNF-α is a should be rechecked in approximately four cytokine that plays a key role in mediating weeks and the dose of AZA or 6-MP should infl ammation in a number of autoimmune be adjusted accordingly. At least fortnightly diseases. It is produced as a transmem- monitoring of full blood count is recom- brane molecule and after proteolytic mended in this interim period. There is some cleavage of this molecule, the soluble TNF-α evidence to suggest that the mechanism for is released. Anti-TNFα antibodies prevent allopurinol correcting the 6TGN/6MMP ratio the interaction of TNFα with its receptor 54 is via indirect inhibition of TPMT. and thus inhibit its effector functions in infl ammation. Infl iximab is a chimeric

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 51 www.nzma.org.nz/journal ARTICLE

antibody while adalimumab is a fully However, the results of this single study need humanised IgG1 kappa antibody. Infl iximab to be validated and further data is needed and adalimumab act against transmem- before repeat antibody testing is routinely brane, soluble and receptor-bound TNF-α. recommended. Transient antibodies are not Anti-TNF antibodies bind to TNF and form associated with loss of response to anti-TNFα complexes. The rate and mode of clearance agents or lack of response to dose escalation. of these complexes is not entirely clear. Depending on the assay used (and hence the At least some clearance is achieved by different units of measurement), infl iximab uptake via antigen presenting cells and ADA levels greater than 9.0U/ml and 9.1ug/ml this is likely to play a major role in immu- have been found to be closely related (90% nogenicity and the formation of anti-drug specifi c) with an unsuccessful result from antibodies(ADA).55 Approximately 10–30% dose escalation.60,61 In New Zealand, the ADA of patients treated with anti-TNF agents for assay has recently been changed from an IBD have primary non-response (14 and 12 ELISA to a homogeneous mobility shift assay weeks after induction with infl iximab and (HMSA), which is not drug concentration-de- adalimumab respectively) while secondary pendent and the result will likely be reported non-response occurs in approximately as absent, low or high antibody titre.62 56–58 23–46% patients by 12 months. Drug levels of biologics and Anti-drug antibodies recommendations for management ADAs can be directed against the antigen While there are no prospective binding or Fab portion of the monoclonal randomised controlled trials showing antibody or against the Fc portion. Those benefi t of TDM-based biologic dosing directed against the Fab portion are likely compared with clinically-based dose opti- to be neutralising antibodies as they will misation, TDM of biologics can be very inhibit binding of the monoclonal antibody helpful in guiding treatment especially in with its target and at the same time increase patients with secondary loss of response. its clearance via the immune system. Anti- There is a wide range of measured concen- bodies directed against other portions of trations of these drugs and clearance is the monoclonal antibodies are generally related to the severity of disease, disease not neutralising as the drug can still bind phenotype, patient’s body mass index to TNF-α but the ADAs can cause increased and gender as well as differences in phar- immune-mediated drug clearance. macokinetics and drug clearance due The results of different quantitative assays to multiple mechanisms, which could of ADAs are often not comparable and if be immune mediated or non-immune repeated testing is required this should be mediated.63 These drugs are unique in that undertaken using the same assay. In addition, the severity of disease is strongly associated the units of measurement used in assays are with drug clearance, hence larger doses not uniform. Available reports indicate a are required in severe disease.64 Patients possible different management approach to with active disease on anti-TNFs can have patients with low level and transient ADAs mechanistic failure (blocking TNF is not compared to those with high and sustained working) or pharmacokinetic failure due ADAs.59,60 In one cohort study, it was reported to increased drug clearance, which can be that among patients developing ADAs to immune or non-immune mediated. Conse- infl iximab, 72% had sustained ADAs while quently, patients can be divided into three 28% had transient ADAs. Dose escalation of groups (Table 4). Patients with mecha- infl iximab resulted in a response in approx- nistic failure are not likely to benefi t from imately 70% patients with transient ADAs other treatment options which have the compared with just 16% with sustained ADAs same mechanism of action or are within the (p=0.0028).60 If detected on testing, consider same class. This drug failure is likely due repeating antibody titres eight weeks later, to the disease being driven by pro-infl am- as these (especially against infl iximab) can matory mediators that are not targeted by be transient and disappear on repeat testing. that specifi c drug.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 52 www.nzma.org.nz/journal ARTICLE

Table 4: Interpretation of drug trough levels and anti-drug antibodies in patients with active IBD on anti-TNF agents.

Drug trough levels Anti-drug antibodies

Mechanistic failure Acceptable Variable

Immune-mediated pharmacokinetic failure Low High

Nonimmune-mediated pharmacokinetic failure Low or absent Undetectable (check drug adherence)

There is ongoing debate regarding are needed, proactive TDM and individu- proactive versus reactive TDM in patients alised management of patients is logical, in clinical remission. Retrospective studies especially in the funding environment in have shown that proactive TDM in patients New Zealand, and can help clinicians use the on anti-TNF agents in clinical remission available drugs effectively. It has to be noted result in lower hospitalisations, operative that the benefi ts of proactive TDM mainly interventions, lower rates of immunogenicity come from retrospective studies and the two and lower rates of drug discontinuation.69,70–72 prospective studies regarding this had some The TAILORIX study compared proactive limitations as discussed above. In patients TDM and clinical dose escalation in patients with disease remission on an anti-TNF agent, receiving infl iximab for Crohn’s disease periodic TDM is recommended, especially if and found no difference in outcomes at it is likely to change management (Figure 3). one year.69 However, a large proportion TDM is also recommended in patients after of patients in the clinical arm had dose induction as well as before drug cessation or escalation, which could have had an effect a drug holiday.65,72 Currently dose escalation on the study results. The TAXIT study did in certain situations for both CD and UC is not fi nd a signifi cant difference in clinical funded and these guidelines should be imple- remission measured at a single point in time mented in that context. In cases of presumed (at one year) between clinically-based dose disease remission and TDM indicating the changes compared to proactive TDM-based need for dose escalation, a careful reas- dose changes.70 However the dosage of sessment should be done to ensure disease both groups in the TAXIT study was opti- remission. Undetectable or low trough levels mised to a target trough level based on TDM are associated with low risk of relapse after before randomisation, which could affect cessation of therapy in selected patients.72 the primary outcome. And notably a signifi - A steady state trough level of 3–8ug/ml and cantly higher proportion of Crohn’s disease 5–12ug/ml is recommended for infl iximab patients attained remission after initial dose and adalimumab respectively. However, optimisation via TDM to a target trough it is also recognised that patients with a level. In addition, the risk of disease fl ares in severe disease phenotype, such as perianal the proactive TDM group was less than those or fi stulising disease, may require higher in the clinically-based dosing group. While concentrations. Reduction in anti-TNFα the primary outcome of disease remission dose is recommended in patients in clinical measured at a single point in time was not remission who have supra-therapeutic met in this trial, signifi cant information was trough levels, without any deterioration gained regarding the benefi ts of proactive in outcomes but resulting in signifi cantly TDM.71 Proactive TDM in patients with IBD in increased cost-effectiveness.10,73,74 remission can result in no change in dosage, In patients with active IBD (supported by dose escalation or dose de-escalation. Dose endoscopy, radiology or biochemical indices) de-escalation will result in cost-savings, and and suspected nonimmune-mediated patients needing dose de-escalation is not pharmacokinetic failure (subtherapeutic a negligible proportion as the TAXIT study anti-TNF drug level and negative ADA), showed that 26% of patients had dose de-es- the initial recommended step is to review calation. Although currently the evidence medication adherence and then escalate is limited and further prospective studies the dose with consideration of adding an

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 53 www.nzma.org.nz/journal ARTICLE

immunomodulator (IMM)—if not already patients is primarily mediated by cyto- utilised.61,75 Repeat TDM within 4–8 weeks is kine(s) not inhibited by the initial biologic. recommended. Failing this, a switch within In this group of patients, addition of an the same class (anti-TNF) or to a different immunomodulator is recommended, espe- class may be required. It is important to cially in those with ADA when switching confi rm active disease in symptomatic to a different biologic.82 The management patients with relevant investigations as a of these patients can be challenging due signifi cant proportion of IBD patients have to the limitations of funded biologics in symptoms due to irritable bowel syndrome New Zealand currently. Detailed discussion (IBS). Studies show a range of prevalence of regarding options including involvement IBS in IBD patients with a pooled prevalence in clinical trials of a different class of agent of 30.9%.76–80 should be undertaken if available. In patients with active IBD and Ideally, samples for drug trough levels immune-mediated pharmacokinetic failure should be taken within 24 hours prior to the (subtherapeutic anti-TNF drug level, positive scheduled dose, especially for infl iximab ADA), consider rechecking ADA in eight where there are large differences in concen- weeks to detect transient ADA.60 Dose esca- tration over the eight-week dosing interval. lation of anti-TNFα and IMM can be added This is a little less critical for adalimumab in these patients, especially if the ADA where there is only about a 50% difference titre is available and low.61 Transient ADA between peak and trough concentrations can develop randomly at any point during over the two-week dose interval. treatment, while persistent and clinically It is important to note that dose escalation signifi cant ADA usually appear during the of infl iximab and adalimumab should ideally 81 fi rst year of therapy. Changing or therapy be continued based on TDM, however, there escalation in patients with possible tran- are limitations on funded dose escalations. sient ADA should be individualised and Dose escalation of infl iximab to 10mg/ weighed against close monitoring and kg every eight weeks for three doses or repeated antibody testing depending on equivalent is currently funded.83 A second disease severity. In primary loss of response re-induction of infl iximab can be done 16 with immune-mediated pharmacokinetic weeks after the fi rst one. For adalimumab, a failure, switching within the same class maintenance dose of 40mg every two weeks and addition of an IMM is recommended, is currently funded.84 Increasing dosage or 14 especially if ADA titre is available and high. shortening dosage interval is currently not A drug of a different class can be used in funded and additional doses can be nego- primary non-response but currently there tiated with the pharmaceutical company on are no funded options available in New a case by case basis. Zealand.61 In secondary loss of response with immune-mediated pharmacokinetic Testing assays failure, switching within the same class Multiple assays are available currently is recommended initially along with the for testing the concentration of drugs as addition of an IMM.14 A drug of a different well as the titres of anti-drug antibodies. class can be used if there is no response to There is some consensus and agreement the initial switch.61 among different assays available for testing infl iximab trough levels.85 However, there Patients have a mechanistic failure is more variability among different assays (regardless of ADA status) if they have active for measurement of the trough levels of disease while on maintenance anti-TNF adalimumab.86 therapy with adequate drug concentrations (infl iximab more than 5ug/ml and adali- ADA detection can be affected by mumab more than 7.5ug/ml). This type the presence of the drug when using of secondary loss of response has been drug-sensitive assays like enzyme-linked observed in 30% of patients with IBD.60,61,75 immunosorbent assays (ELISA) or radio-im- 87 Switching to a drug of a different class is munoassays (RIA). Drug-tolerant assays recommended in these patients, as it is like the electrochemiluminescence and assumed that the infl ammation in these HMSA assays are more resistant to this effect

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 54 www.nzma.org.nz/journal ARTICLE

Figure 3: Algorithm for therapeutic drug monitoring and management of patients with infl ammatory bowel disease (IBD) on anti-TNF agents.

ADA: anti-drug antibodies. IMM: immunomodulator. * Dose escalation of infliximab to 10mg/kg every eight weeks for three doses or equivalent is currently funded. Dose escalation of adalimumab is not funded but can be negotiated with the pharmaceutical company on a case-by-case basis. compared with an ELISA. The HMSA assay Combination therapy will detect antibodies in more patients and Combination therapy with an IMM (thio- will ultimately provide better information purine or methotrexate) and a biologic for clinicians. results in a signifi cantly higher rate of One of the important aspects of TDM is steroid free remission and mucosal healing; the timing of results once testing is done. In and lower rates of secondary loss of the current setting results are not available response.90–92 Combination therapy results immediately and any dose optimisation in higher biologic concentrations, at least based on TDM can be delayed. Point-of-care in part due to a signifi cantly lower rate of testing provides an opportunity to have TDM formation of ADA as well as a lower titre of results immediately available and acted ADA if formed.13,90 In patients on biologic upon.88,89 No data from large prospective monotherapy, the addition of an IMM can studies regarding point-of-care testing is result in improving response to the biologic available at present. by reducing or eliminating ADAs.90 There is

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 55 www.nzma.org.nz/journal ARTICLE

no increased risk of infection with combi- However, aspects such as age and gender, nation therapy.93 Patients being treated with and the consideration of EBV testing in adalimumab have a lower rate of developing adolescence require an individualised ADA compared to those on infl iximab.94,95 approach in the paediatric population. Combination treatment with adalimumab Combination therapy could be initiated and an IMM leads to increased biologic drug early in patients with risk factors for concentrations and lowers immunogenicity, aggressive disease in the paediatric popu- but increased clinical effi cacy has not yet lation including older age (6–17 years old been determined.95 compared with younger patients), female In patients being started on infl iximab, sex, ileocolonic or ileal disease, perianal combination therapy with an IMM is disease, fi stulising disease, steroid depen- recommended. There is data showing dence or refractoriness, previous surgery 100,101 that low-dose AZA is suffi cient to achieve and growth failure. the immunologic benefi ts of preventing Cost e ectiveness of TDM antibody formation in combination TDM has been shown to result in cost therapy.96 Data regarding a lower dose savings without any adverse effect on of IMM in combination therapy is not clinical outcomes in a systemic review of conclusive, and monitoring of thiopurine multiple studies involving patients with metabolites is still useful in cases of IBD and rheumatoid arthritis.102 A more treatment failure, monitoring of treatment recent meta-analyses supports the cost-effec- adherence, detecting hypermethylators tiveness of TDM in IBD with the effect more (‘shunters’) and investigating drug toxicities. evident for reactive TDM.103 Better clinical It has been shown that adding a thio- outcome due to reactive TDM can result in purine to a biologic is benefi cial in fewer disease-related complications and preventing the formation of ADA even in need for surgery. Furthermore, TDM results patients with no previous response to a thio- in dose de-escalation in a number of patients purine.97 The optimum treatment duration resulting in cost savings. The benefi ts of of an IMM in combination with a biologic reactive TDM are well recognised and it is for the prevention of ADA is yet to be deter- currently recommended in international mined but continuing the IMM or biologic guidelines.10,12,104 There are no prospective (in combination therapy) for one year is controlled studies showing cost-effectiveness recommended, keeping in view the possible of proactive TDM, however retrospective risks of combination therapy and the clinical data has suggested some benefi t.105 The benefi ts.13 The small increased risk of malig- limited options of biologics available in New nancy with combination therapy (while Zealand for IBD can make disease treatment using a biologic and/or a thiopurine) should diffi cult, especially in cases of mechanistic be considered when continuing treatment failure of an anti-TNF agent. Additionally, beyond one year.13,98 consideration of surgical intervention is Methotrexate can be added to a biologic potentially done earlier due to limited avail- instead of a thiopurine in cases of intol- ability of biologic drugs. Currently there is erance to thiopurines. There is currently no no local data regarding cost-effectiveness consensus about the optimal dose of meth- of TDM, however international data can be otrexate in this setting. Further studies are extrapolated to the scenario in New Zealand. ongoing in this area, including the COMBINE The judicious and effective use of funded study (Clinical Outcomes of Methotrexate medications for IBD in New Zealand is even Binary treatment with INfl iximab or adali- more important due to relatively fewer mumab in practicE).99 options available and the limitations on their funded use. TDM in the New Zealand The use of combination therapy is also setting can help clinicians get the most out important in children and adolescents, of the available drugs and provide individu- as discussed in a recent clinical report.100 alised care to their patients.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 56 www.nzma.org.nz/journal ARTICLE

drug toxicity, identifying hypermethylators, Conclusion assessing loss of response and in decisions TDM is now an essential component in the regarding treatment escalation or de-esca- management of IBD patients and the corner- lation. In addition to using TDM in patients stone of individualised treatment. TDM is who fail therapy, proactive TDM can facil- used to optimise drug effi cacy and reduce itate early treatment decisions and has been the risk of toxicity. In IBD, TDM is also indi- shown to lead to better outcomes. cated for checking adherence, evaluating

Competing interests: Dr Day reports personal fees from Abbvie and Janssen outside the submitted work. Author information: Afrasyab Khan, Gastroenterology Advanced Trainee, Gastroenterology Unit, Southern District Health Board, Dunedin; Clinical Lecturer, Department of Medicine, University of Otago, Dunedin; Arvenia B Berahmana, Trainee Intern, Gastroenterology Unit, Southern District Health Board, Dunedin; Andrew S Day, Professor, Department of Paediatrics, University of Otago, Christchurch; Murray L Barclay, Professor, Department of Medicine, University of Otago, Christchurch; Department of Gastroenterology, Christchurch Hospital, Christchurch; Michael Schultz, Associate Professor, Gastroenterology Unit, Southern District Health Board, Dunedin; Department of Medicine, University of Otago, Dunedin. Corresponding author: Michael Schultz, Associate Professor of Medicine (Gastroenterology), Department of Medicine, University of Otago, PO Box 56, Dunedin 9054. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7826

REFERENCES: 1. Eason RJ, Lee SP, Tasman- Infl amm Bowel Dis. 2016 cancer in patients with Jones C. Infl ammatory Sep; 22(9):2238–44. infl ammatory bowel bowel disease in Auckland, 5. Gearry RB, Day AS. Infl am- disease. Minerva Chir. New Zealand. Aust N Z Med matory bowel disease in 2017 Dec; 72(6):520–9. J. 1982 Apr; 12(2):125–31. New Zealand children--a 9. Chan HC-H, Ng SC. 2. Schlup M, Maclaurin BP, growing problem. N Z Med Emerging biologics in Barbezat GO, de Lambert J. 2008 Oct 3; 121(1283):5–8. infl ammatory bowel BM. Crohn’s disease: a 6. Coppell K, Galts C, disease. J Gastroenterol. ten year retrospective Huizing F, et al. Annual 2017 Feb; 52(2):141–50. review at Dunedin Incidence and Phenotypic 10. Mitrev N, Vande Casteele hospitals. N Z Med J. 1986 Presentation of IBD in N, Seow CH, et al. Review Mar 12; 99(797):141–4. Southern New Zealand: article: consensus state- 3. Wigley R, Maclaurin BP. An 18-Year Epidemiolog- ments on therapeutic A study of ulcerative ical Analysis. Infl amm drug monitoring of colitis in New Zealand, Intest Dis. 2018 Sep 17. anti-tumour necrosis factor showing a low incidence 7. Stidham RW, Higgins therapy in infl ammatory in Maoris. Br Med J. 1962 PDR. Colorectal Cancer bowel diseases. Aliment Jul 28; 2(5299):228–31. in Infl ammatory Bowel Pharmacol Ther. 2017 4. Su HY, Gupta V, Day Disease. Clin Colon Dec; 46(11–12):1037–53. AS, Gearry RB. Rising Rectal Surg. 2018 11. Vande Casteele N, Herfarth Incidence of Infl amma- May; 31(3):168–78. H, Katz J, et al. American tory Bowel Disease in 8. Baek S-J, Kim S-H. Coli- Gastroenterological Asso- Canterbury, New Zealand. tis-associated colorectal ciation Institute Technical

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 57 www.nzma.org.nz/journal ARTICLE

Review on the Role of Ther- patients with infl ammatory 25. Moran GW, Dubeau M-F, apeutic Drug Monitoring in bowel disease. Pharma- Kaplan GG, et al. Clinical the Management of Infl am- coepidemiol Drug Saf. predictors of thiopurine-re- matory Bowel Diseases. 2004 Aug; 13(8):563–7. lated adverse events in Gastroenterology. 2017 19. Simsek M, Meijer B, Crohn’s disease. World J Sep; 153(3):835–857.e6. Ramsoekh D, et al. Gastroenterol. 2015 Jul 12. Feuerstein JD, Nguyen GC, Clinical Course of 7; 21(25):7795–804. Kupfer SS, et al. American Nodular Regenerative 26. Bahi M, Walmsley RS, Gastroenterological Associ- Hyperplasia in Thiopurine Gray AR, et al. The risk ation Institute Guideline on Treated Infl ammatory of non-melanoma skin Therapeutic Drug Monitor- Bowel Disease Patients. Clin cancer in New Zealand ing in Infl ammatory Bowel Gastroenterol Hepatol Off in infl ammatory bowel Disease. Gastroenterology. Clin Pract J Am Gastroen- disease patients treated 2017 Sep; 153(3):827–34. terol Assoc. 2018 Jun 1. with thiopurines. J 13. Bots S, Gecse K, Barclay M, 20. Subramaniam K, D’Rozario Gastroenterol Hepatol. D’Haens G. Combination J, Pavli P. Lymphoma and 2018 May; 33(5):1047–52. Immunosuppression in other lymphoproliferative 27. Sayani FA, Prosser C, IBD. Infl amm Bowel Dis. disorders in infl ammatory Bailey RJ, et al. Thiopurine 2018 Feb 15; 24(3):539–45. bowel disease: a review. methyltransferase enzyme 14. Ding NS, Hart A, De J Gastroenterol Hepatol. activity determination Cruz P. Systematic 2013 Jan; 28(1):24–30. before treatment of review: predicting and 21. Kotlyar DS, Lewis JD, infl ammatory bowel optimising response to Beaugerie L, et al. Risk disease with azathioprine: anti-TNF therapy in Crohn’s of lymphoma in patients effect on cost and adverse disease - algorithm for with infl ammatory bowel events. Can J Gastroenterol practical management. disease treated with J Can Gastroenterol. 2005 Aliment Pharmacol Ther. azathioprine and 6-mercap- Mar; 19(3):147–51. 2016 Jan; 43(1):30–51. topurine: a meta-analysis. 28. Newman WG, Payne K, 15. Maaser C, Sturm A, Vavric- Clin Gastroenterol Hepatol Tricker K, et al. A pragmat- ka SR, et al. ECCO-ESGAR Off Clin Pract J Am ic randomized controlled Guideline for Diagnostic Gastroenterol Assoc. 2015 trial of thiopurine meth- Assessment in Infl ammato- May; 13(5):847–858. yltransferase genotyping ry Bowel Disease. J Crohns e4; quiz e48–50. prior to azathioprine Colitis. 2018 Aug 23. 22. Peyrin-Biroulet L, Khos- treatment: the TARGET study. Pharmacogenomics. 16. Gisbert JP, Nino P, Cara rotehrani K, Carrat F, et al. 2011 Jun; 12(6):815–26. C, Rodrigo L. Compar- Increased risk for nonmel- ative effectiveness of anoma skin cancers in 29. Cheung S-T, Allan RN. azathioprine in Crohn’s patients who receive thio- Mistaken identity: disease and ulcerative purines for infl ammatory misclassifi cation of TPMT colitis: prospective, bowel disease. Gastro- phenotype following long-term, follow-up enterology. 2011 Nov; blood transfusion. Eur J study of 394 patients. 141(5):1621–1628.e1–5. Gastroenterol Hepatol. Aliment Pharmacol Ther. 23. Seksik P, Mary J-Y, 2003 Nov;15(11):1245–7. 2008 Jul; 28(2):228–38. Beaugerie L, et al. 30. Donnan JR, Ungar WJ, 17. van Asseldonk DP, Sander- Incidence of nodular Mathews M, Rahman son J, de Boer NKH, et al. regenerative hyperplasia P. Systematic review of Diffi culties and possibilities in infl ammatory bowel thiopurine methyltrans- with thiopurine therapy disease patients treated ferase genotype and in infl ammatory bowel with azathioprine. enzymatic testing strat- disease--proceedings of Infl amm Bowel Dis. 2011 egies. Ther Drug Monit. the fi rst Thiopurine Task Feb; 17(2):565–72. 2011 Apr; 33(2):192–9. Force meeting. Dig Liver 24. Qiu Y, Mao R, Zhang S, 31. Gardiner SJ, Gearry RB, Dis Off J Ital Soc Gastroen- et al. Safety Profi le of Begg EJ, et al. Thiopurine terol Ital Assoc Study Liver. Thiopurines in Crohn dose in intermediate and 2011 Apr; 43(4):270–6. Disease: Analysis of 893 normal metabolizers of 18. Gearry RB, Barclay ML, Patient-Years Follow-Up in thiopurine methyltransfer- Burt MJ, et al. Thiopurine a Southern China Cohort. ase may differ three-fold. drug adverse effects in a Medicine (Baltimore). Clin Gastroenterol Hepatol population of New Zealand 2015 Oct; 94(41):e1513. Off Clin Pract J Am

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 58 www.nzma.org.nz/journal ARTICLE

Gastroenterol Assoc. 2008 treat my infl ammatory ylmercaptopurine-induced Jun; 6(6):654–60; quiz 604. bowel disease-patients leukocytopenia during 32. Mulder C. Personal with thiopurines? thiopurine therapy in communication with Prof World J Gastrointest infl ammatory bowel Chris Mulder, Department Pharmacol Ther. 2016 disease patients. J Gastro- of Gastroenterology and Nov 6; 7(4):524–30. enterol Hepatol. 2017 Hepatology, Vrije Univer- 40. Osterman MT, Kundu Jun; 32(6):1183–90. siteit Medical Center, R, Lichtenstein GR, 47. Amin J, Huang B, Yoon 1081 HV Amsterdam, Lewis JD. Association of J, Shih DQ. Update 2014: The Netherlands. 2018. 6-thioguanine nucleotide advances to optimize 33. van Asseldonk DP, Jharap levels and infl ammatory 6-mercaptopurine and B, Verheij J, et al. The bowel disease activity: azathioprine to reduce Prevalence of Nodular a meta-analysis. Gastro- toxicity and improve Regenerative Hyperplasia enterology. 2006 Apr; effi cacy in the management in Infl ammatory Bowel 130(4):1047–53. of IBD. Infl amm Bowel Dis. Disease Patients Treated 41. Andoh A, Tsujikawa T, 2015 Feb; 21(2):445–52. with Thioguanine Is Not Ban H, et al. Monitoring 48. Beswick L, Hair CS, Dowl- Associated with Clinically 6-thioguanine nucleotide ing D. Letter: successful Signifi cant Liver Disease. concentrations in Japanese mercaptopurine therapy Infl amm Bowel Dis. 2016 patients with infl am- after azathioprine-related Sep; 22(9):2112–20. matory bowel disease. J pancreatitis in patients 34. Dubinsky MC, Vasili- Gastroenterol Hepatol. with IBD. Aliment auskas EA, Singh H, et 2008 Sep; 23(9):1373–7. Pharmacol Ther. 2013 al. 6-thioguanine can 42. Lim SZ, Chua EW. Revisit- Jan; 37(1):162. cause serious liver ing the Role of Thiopurines 49. Gallego-Gutierrez S, injury in infl ammatory in Infl ammatory Bowel Navas-Lopez VM, Kolorz M, bowel disease patients. Disease Through Phar- Bartosova L, et al. Success- Gastroenterology. 2003 macogenomics and ful Mercaptopurine Usage Aug; 125(2):298–303. Use of Novel Methods despite Azathioprine-In- 35. Seidman EG. Clinical use for Therapeutic Drug duced Pancreatitis in and practical application Monitoring. Front Phar- Paediatric Crohn’s Disease. of TPMT enzyme and macol. 2018; 9:1107. J Crohns Colitis. 2015 6-mercaptopurine metab- 43. Gilissen LPL, Wong Aug; 9(8):676–9. olite monitoring in IBD. DR, Engels LGJB, et 50. Bonaz B, Boitard J, Marteau Rev Gastroenterol Disord. al. Therapeutic drug P, et al. Tioguanine in 2003; 3 Suppl 1:S30-38. monitoring of thiopurine patients with Crohn’s 36. Spencer EA, Dubinsky MC. metabolites in adult disease intolerant or Therapeutic Drug Moni- thiopurine tolerant IBD resistant to azathioprine/ toring in Infl ammatory patients on maintenance mercaptopurine. Aliment Bowel Disease: History therapy. J Crohns Colitis. Pharmacol Ther. 2003 and Future Directions. 2012 Jul; 6(6):698–707. Aug 15; 18(4):401–8. Pediatr Clin North Am. 44. Stocco G, Londero M, 51. van Egmond R, Chin P, 2017 Dec; 64(6):1309–26. Campanozzi A, et al. Useful- Zhang M, et al. High TPMT 37. Beswick L, Friedman AB, ness of the measurement of enzyme activity does not Sparrow MP. The role of azathioprine metabolites in explain drug resistance due thiopurine metabolite the assessment of non-ad- to preferential. Aliment monitoring in infl amma- herence. J Crohns Colitis. Pharmacol Ther. 2012 tory bowel disease. Expert 2010 Nov; 4(5):599–602. May; 35(10):1181–9. Rev Gastroenterol Hepatol. 45. Dubinsky M, Lamothe 52. Ansari A, Patel N, 2014 May; 8(4):383–92. S, Yang T, et al. Phar- Sanderson J, et al. 38. Cappello M, Morreale GC. macogenomics and Low-dose azathioprine The Role of Laboratory metabolite measurement or mercaptopurine in Tests in Crohn’s Disease. for 6-mercaptopurine combination with allo- Clin Med Insights Gastro- therapy in infl ammatory purinol can bypass many enterol. 2016; 9:51–62. bowel disease. Gastroenter- adverse drug reactions ology. 2000; 118(4):705–13. in patients with infl am- 39. Meijer B, Mulder CJ, matory bowel disease. van Bodegraven AA, 46. Meijer B, Kreijne JE, van Aliment Pharmacol Ther. de Boer NKH. How I Moorsel SAW, et al. 6-meth- 2010 Mar; 31(6):640–7.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 59 www.nzma.org.nz/journal ARTICLE

53. Ansari A, Elliott T, 61. Yanai H, Lichtenstein L, 67. Armuzzi A, Felice C. IBD: Baburajan B, et al. Long- Assa A, et al. Levels of Infl iximab dose optimi- term outcome of using drug and antidrug anti- zation in IBD-proactive allopurinol co-therapy as bodies are associated with or reactive? Nat Rev a strategy for overcoming outcome of interventions Gastroenterol Hepatol. thiopurine hepatotoxicity after loss of response to 2014 Dec; 11(12):706–8. in treating infl ammatory infl iximab or adalimum- 68. Vaughn BP, Marti- bowel disease. Aliment ab. Clin Gastroenterol nez-Vazquez M, Pharmacol Ther. 2008 Hepatol Off Clin Pract J Am Patwardhan VR, et al. Sep 15; 28(6):734–41. Gastroenterol Assoc. 2015 Proactive therapeutic 54. Blaker PA, Arenas-Her- Mar; 13(3):522-530.e2. concentration monitoring nandez M, Smith MA, et al. 62. Personal Communication of infl iximab may improve Mechanism of allopurinol with Canterbury Health outcomes for patients induced TPMT inhibition. Laboratories, Christchurch, with infl ammatory bowel Biochem Pharmacol. 2013 New Zealand. 2018. disease: results from a Aug 15; 86(4):539–47. 63. Lopetuso LR, Gerardi pilot observational study. 55. van Schie KA, Ooijevaar-de V, Papa V, et al. Can We Infl amm Bowel Dis. 2014 Heer P, et al. Therapeu- Predict the Effi cacy of Anti- Nov; 20(11):1996–2003. tic TNF Inhibitors can TNF-alpha Agents? Int J 69. OP029. Drug-concentration Differentially Stabilize Mol Sci. 2017 Sep 14; 18(9). versus symptom-driven Trimeric TNF by Inhibiting 64. Brandse JF, Mathot RA, dose adaptation of Monomer Exchange. Sci van der Kleij D, et al. Infl iximab in patients with Rep. 2016 Sep 8; 6:32747. Pharmacokinetic Features active Crohn’s disease: a 56. Ben-Horin S, Chowers and Presence of Antidrug prospective, randomised, Y. Review article: loss of Antibodies Associate With multicentre trial (Tailorix). response to anti-TNF treat- Response to Infl iximab J Crohns Colitis. 2016 Mar ments in Crohn’s disease. Induction Therapy in 1; 10(suppl_1):S24–S24. Aliment Pharmacol Ther. Patients With Moderate 70. Vande Casteele N, Ferrante 2011 May; 33(9):987–95. to Severe Ulcerative M, Van Assche G, et al. 57. Roda G, Jharap B, Neeraj Colitis. Clin Gastroenterol Trough concentrations N, Colombel J-F. Loss of Hepatol Off Clin Pract J Am of infl iximab guide Response to Anti-TNFs: Gastroenterol Assoc. 2016 dosing for patients with Defi nition, Epidemiology, Feb; 14(2):251–258.e1–2. infl ammatory bowel and Management. Clin 65. Papamichael K, Chachu disease. 2015 Jun; 148. Transl Gastroenterol. KA, Vajravelu RK, et al. 71. Ben-Horin S. Drug 2016 Jan 7; 7:e135. Improved Long-term Level-based Anti-Tumor 58. Ford AC, Sandborn WJ, Outcomes of Patients Necrosis Factor Therapy: Khan KJ, et al. Effi cacy With Infl ammatory Bowel Ready for Prime Time? of biological therapies Disease Receiving Proactive Gastroenterology. 2015 in infl ammatory bowel Compared With Reactive Jun; 148(7):1268–71. disease: systematic review Monitoring of Serum 72. Ben-Horin S, Chowers Y, and meta-analysis. Am J Concentrations of Infl ix- Ungar B, et al. Undetectable Gastroenterol. 2011 Apr; imab. Clin Gastroenterol anti-TNF drug levels in 106(4):644–59, quiz 660. Hepatol Off Clin Pract J Am patients with long-term Gastroenterol Assoc. 2017 59. Ungar B, Anafy A, Yanai remission predict success- Oct;15(10):1580–1588.e3. H, et al. Signifi cance of ful drug withdrawal. low level infl iximab in 66. Mitchell RA, Shuster Aliment Pharmacol Ther. the absence of anti-infl ix- C, Shahidi N, et al. The 2015 Aug; 42(3):356–64. imab antibodies. World Utility of Infl iximab 73. Steenholdt C, Brynskov J, J Gastroenterol. 2015 Therapeutic Drug Moni- Thomsen OO, et al. Indi- Feb 14; 21(6):1907–14. toring among Patients vidualised therapy is more with Infl ammatory Bowel 60. Vande Casteele N, Gils A, cost-effective than dose Disease and Concerns Singh S, et al. Antibody intensifi cation in patients for Loss of Response: A response to infl iximab and with Crohn’s disease who Retrospective Analysis of a its impact on pharmaco- lose response to anti-TNF Real-World Experience. Can kinetics can be transient. treatment: a randomised, J Gastroenterol Hepatol. Am J Gastroenterol. 2013 controlled trial. Gut. 2016; 2016:5203898. Jun; 108(6):962–71. 2014 Jun; 63(6):919–27.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 60 www.nzma.org.nz/journal ARTICLE

74. Ungar B, Levy I, Yavne Y, et colitis patients with clinical care testing for infl iximab al. Optimizing Anti-TNF-al- and endoscopic evidence and faecal calprotectin in pha Therapy: Serum Levels of remission: prospective IBD patients with a second- of Infl iximab and Adalim- multicenter study. Scand ary loss of response. J umab Are Associated with J Gastroenterol. 2014 Crohns Colitis. 2018 Jan 16; Mucosal Healing in Patients Jun; 49(6):674–80. 12(supplement_1):S470–1. With Infl ammatory Bowel 81. Ungar B, Chowers Y, 89. Ametzazurra A, Rivera N, Diseases. Clin Gastroenterol Yavzori M, et al. The Hernández AM, et al. Rapid Hepatol Off Clin Pract J Am temporal evolution of Point-of-Care Monitoring Gastroenterol Assoc. 2016 antidrug antibodies in of Anti-Infl iximab Anti- Apr;1 4(4):550-557.e2. patients with infl ammatory bodies in Patients with 75. Roblin X, Rinaudo M, Del bowel disease treated Infl ammatory Bowel Tedesco E, et al. Devel- with infl iximab. Gut. 2014 Disease Treated with the opment of an algorithm Aug; 63(8):1258–64. Reference Infl iximab or incorporating pharmaco- 82. Strik AS, Bots SJA, D’Haens CT-P13 in Routine Clinical kinetics of adalimumab G, Lowenberg M. Optimi- Practice. Gastroenterology. in infl ammatory bowel zation of anti-TNF therapy 2017 Apr 1; 152(5):S391. diseases. Am J Gastroenter- in patients with Infl am- 90. Ben-Horin S, Waterman M, ol. 2014 Aug; 109(8):1250–6. matory Bowel Disease. Kopylov U, et al. Addition 76. Teruel C, Garrido E, Expert Rev Clin Pharmacol. of an immunomodulator Mesonero F. Diagnosis and 2016; 9(3):429–39. to infl iximab therapy elim- management of functional 83. Pharmaceutical Manage- inates antidrug antibodies symptoms in infl ammatory ment Agency (PHARMAC). in serum and restores bowel disease in remis- Form RS1581. [Internet]. clinical response of patients sion. World J Gastrointest 2019 [cited 2019 Jan 4]. with infl ammatory bowel Pharmacol Ther. 2016 Available from: http:// disease. Clin Gastroenterol Feb 6; 7(1):78–90. www.pharmac.govt. Hepatol Off Clin Pract J 77. Piche T, Ducrotte P, nz/2019/01/01/RS1581.pdf Am Gastroenterol Assoc. 2013 Apr; 11(4):444–7. Sabate JM, et al. Impact of 84. Pharmaceutical Manage- functional bowel symp- ment Agency (PHARMAC). 91. Colombel JF, Sandborn toms on quality of life Form SA1621. [Internet]. WJ, Reinisch W, et al. and fatigue in quiescent 2018 [cited 2019 Jan 4]. Infl iximab, azathioprine, Crohn disease and irritable Available from: http:// or combination therapy bowel syndrome. Neuro- www.pharmac.govt. for Crohn’s disease. N gastroenterol Motil Off J nz/2018/08/01/SA1621.pdf Engl J Med. 2010 Apr Eur Gastrointest Motil Soc. 15; 362(15):1383–95. 85. European Medicines 2010 Jun; 22(6):626–e174. Agency. Guideline on 92. Panaccione R, Ghosh 78. Jonefjall B, Strid H, Ohman bioanalytical method S, Middleton S, et al. L, et al. Characterization validation. Eur Med Combination therapy of IBS-like symptoms in Agency Lond. 2009; with infl iximab and patients with ulcerative azathioprine is superior to 86. Bodini G, Giannini EG, colitis in clinical remis- monotherapy with either Furnari M, et al. Compar- sion. Neurogastroenterol agent in ulcerative colitis. ison of Two Different Motil Off J Eur Gastro- Gastroenterology. 2014 Techniques to Assess intest Motil Soc. 2013 Feb; 146(2):392-400.e3. Adalimumab Trough Sep; 25(9):756–e578. Levels in Patients with 93. Lichtenstein GR, Rutgeerts 79. Vivinus-Nebot M, Crohn’s Disease. J Gastro- P, Sandborn WJ, et al. A Frin-Mathy G, Bzioueche intest Liver Dis JGLD. pooled analysis of infec- H, et al. Functional bowel 2015 Dec; 24(4):451–6. tions, malignancy, and symptoms in quiescent mortality in infl iximab- and 87. Vaughn BP, Sandborn infl ammatory bowel diseas- immunomodulator-treated WJ, Cheifetz AS. Biologic es: role of epithelial barrier adult patients with infl am- concentration testing disruption and low-grade matory bowel disease. in infl ammatory bowel infl ammation. Gut. 2014 Am J Gastroenterol. 2012 disease. Infl amm May; 63(5):744–52. Jul; 107(7):1051–63. Bowel Dis. 2015 Jun; 80. Fukuba N, Ishihara S, 21(6):1435–42. 94. Sandborn WJ, Hanauer Tada Y, et al. Prevalence of SB, Rutgeerts P, et al. 88. Restellini S, Bessissow T, irritable bowel syndrome- Adalimumab for main- Lemieux C, et al. P711 A like symptoms in ulcerative tenance treatment of pilot study using point of

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 61 www.nzma.org.nz/journal ARTICLE

Crohn’s disease: results of cade(R)] or Conventional in Infl ammatory Bowel the CLASSIC II trial. Gut. Therapy. J Crohns Colitis. Disease: A Systematic 2007 Sep; 56(9):1232–9. 2017 Jun 1; 11(6):680–9. Review With Meta-Analy- 95. Chalhoub JM, Rimmani 99. Steiner S. Clinical sis. J Crohns Colitis. 2018 HH, Gumaste VV, Sharara Outcomes of Methotrexate Nov 15; 12(11):1302–15. AI. Systematic Review and Binary treatment with 104. Bressler B, Marshall JK, Meta-analysis: Adalimum- INfl iximab or adalimumab Bernstein CN, et al. Clinical ab Monotherapy Versus in practicE [Internet]. 2015. Practice Guidelines for Combination Therapy Available from: http:// the Medical Management with Immunomodulators www.improvecarenow. of Nonhospitalized for Induction and Main- org/the_combine_study Ulcerative Colitis: The tenance of Remission 100. Day AS, Gulati AS, Patel Toronto Consensus. and Response in Patients N, et al. The Role of Gastroenterology. 2015 May with Crohn’s Disease. Combination Therapy in 1; 148(5):1035–1058.e3. Infl amm Bowel Dis. 2017 Pediatric Infl ammatory 105. Giráldez J, Ferreiro R, Aug; 23(8):1316–27. Bowel Disease: A Clinical López JG, et al. GM-032 96. Zator ZA, Regueiro M, Report from the North Economic impact of Goldby-Reffner K, et al. American Society for adalimumab monitoring Sa1961 Low-dose Concom- Pediatric Gastroenterology, through algorithm use in itant Azathioprine is as Hepatology and Nutrition. patients with infl amma- Effective as High-dose J Pediatr Gastroenterol tory bowel disease. Eur Azathioprine in Preventing Nutr. 2018 Feb;66(2):361–8. J Hosp Pharm. 2017 Mar Immunogenicity to Infl ix- 101. Gupta N, Bostrom AG, 1; 24(Suppl 1):A171. imab. Gastroenterology. Kirschner BS, et al. 106. Dean L. Thioguanine Thera- 2018 Feb 15; 150(4):S417. Incidence of stricturing py and TPMT Genotype. In: 97. Bar-Yoseph H, Waterman and penetrating compli- Pratt V, McLeod H, Rubin- M, Almog R, et al. Preven- cations of Crohn’s disease stein W, Dean L, Kattman B, tion of Antidrug Antibody diagnosed in pediatric Malheiro A, editors. Medi- Formation to Infl iximab patients. Infl amm Bowel cal Genetics Summaries. in Crohn’s Patients with Dis. 2010 Apr; 16(4):638–44. Bethesda (MD): National Prior Failure of Thiopu- 102. Martelli L, Olivera P, Center for Biotechnology rines. Clin Gastroenterol Roblin X, et al. Cost-ef- Information (US); 2012. Hepatol Off Clin Pract J fectiveness of drug 107. Relling MV, Schwab M, Am Gastroenterol Assoc. monitoring of anti-TNF Whirl-Carrillo M, Suarez- 2017 Jan; 15(1):69–75. therapy in infl ammatory Kurtz G, Pui C-H, Stein CM, 98. D’Haens G, Reinisch bowel disease and rheuma- et al. Clinical Pharmaco- W, Colombel J-F, et al. toid arthritis: a systematic genetics Implementation Five-year Safety Data review. J Gastroenterol. Consortium Guideline for From ENCORE, a European 2017 Jan 1; 52(1):19–25. Thiopurine Dosing Based Observational Safety 103. Ricciuto A, Dhaliwal J, on TPMT and NUDT15 Registry for Adults with Walters TD, et al. Clinical Genotypes: 2018 Update. Crohn’s Disease Treated Outcomes with Thera- Clin Pharmacol Ther. 2018. with Infl iximab [Remi- peutic Drug Monitoring

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 62 www.nzma.org.nz/journal ARTICLE

Vision screening in New Zealand: an audit of the B4 School Check William Muller, Logan Mitchell, Graham Wilson

ABSTRACT AIMS: To assess the accuracy of the B4 School Check (B4SC) vision screening programme in two distinct regions of New Zealand. METHODS: A retrospective audit of all children who were screened for vision in the Southern and Tairawhiti District Health Boards, between 1 April and 30 September 2016. Results from the B4SC screening programme (n=2,109) were compared to records for all children who were screened and subsequently presented to an optometrist at a DHB eye clinic (n=116). RESULTS: The B4SC produced a sensitivity in the range of 54.7% to 94.7% and a specificity of 93.8% to 95.7%. There was a low positive predictive value (PPV), between 29.5% and 51.1%, with a relatively high number of false positive referrals. The negative predictive value (NPV) was higher, however, between 97.8% and 99.7%, meaning nearly all children who passed screening had no visual impairment. CONCLUSIONS: The high NPV is reassuring that very few children with visual impairment are missed by screening. The low PPV was consistent with the international literature and is related to a tendency for over-referral from the B4CS programme. Further work could evaluate increasing the threshold for referral to reduce the number of false positive cases.

mblyopia, or ‘lazy eye’, is the most In 2008, a nationwide screening common ophthalmological condition programme was introduced in New Zealand, affecting children, with a preva- the B4 School Check (B4SC), which includes A 1–3 lence of approximately 1–5%. It arises vision screening. Vision screening is from abnormal visual development during performed by vision-hearing technicians childhood, a process occurring competi- (VHT) who aim to test all four-year-old tively between fellow eyes, and can be due children in New Zealand. The current B4SC to strabismus, refractive error (especially protocol assesses the VA of each eye using anisometropia, a difference in the refrac- letter matching with the Parr Vision chart tive state between fellow eyes), or ‘pattern with confusion bars at four metres (this is deprivation’ (visual impairment due to the same as the Sheridan-Gardener chart structural eye abnormalities).3 Amblyopia which is used at six metres).11 Testing is typi- typically manifests as reduced visual acuity cally performed at the child’s educational or (VA) in one eye and can lead to issues later day-care facility. Failure of the child to pass in life, such as limitation of occupational the B4SC vision screening, defi ned as VA of choices and visual impairment if something 6/12 or worse in one or both eyes, results in happens to the better-seeing eye.4,5 Amblyo- referral to either a community optometrist, pia treatment is partly effective so long as it private ophthalmologist or district health occurs before the critical period of child vi- board (DHB) eye clinic at the discretion of sual development (about seven to nine years the child’s parent/caregiver. A VA of 6/9 of age),6,7 and so many countries have some in one eye and 6/6 in the other results in form of screening to detect amblyopia while rescreening, and a VA of 6/9 or in both eyes it is still amenable to treatment.8–10 (ie, no difference between the eyes) results

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 63 www.nzma.org.nz/journal ARTICLE

in a pass. If on rescreening, the VA remains was collected: identifying information (ie, at 6/9 in one eye and 6/6 in the other, or name, date of birth and NHI number), date becomes worse in either eye, the child is of screening, ethnicity, sex, unaided visual referred.12 acuity of right and left eyes obtained through Currently, there is little published data screening, and referral outcome from regarding the performance of this screening screening (ie, pass, referral or rescreen). programme. Therefore, the primary aim of Southern and Tairawhiti DHB Eye this study was to assess the accuracy of the Clinic Data B4SC vision screening programme in terms Identifying details for all children aged of sensitivity, specifi city, positive predictive four to seven years (a window maximising value (PPV) and negative predictive identifi cation of children who could have value (NPV), in two distinct regions in theoretically been screened within the study New Zealand, as well as to assess how VA period) who had been seen at a Southern measurement by VHTs compare to optome- DHB eye clinic (Dunedin Hospital or Inver- trists or DHB eye-clinic assessment. cargill Hospital) or a Tairawhiti DHB eye clinic (Gisborne Hospital) between 1 April Methods and 30 November 2016 were obtained retro- This study was a retrospective audit of spectively. This longer data collection period all children who were screened for vision allowed for delay between VHT referral and between 1 April 2016 and 30 September fi rst specialist assessment at an eye clinic. 2016, in the B4SC programme, in two distinct Identifying details were used to crossmatch New Zealand DHBs: Southern (SDHB) and children to the MoH Database. Tairawhiti (TDHB), potentially allowing Every child who presented to a Southern identifi cation of differences in screening or Tairawhiti DHB eye clinic, and who throughout New Zealand. was also screened between 1 April and The SDHB catchment area is comprised 30 September 2016 (regardless of their of the southern half of the South Island screening outcome) was de-identifi ed of New Zealand, and has a population and the following additional clinical of about 320,000, 10% of whom are New information from the patient notes was Zealand Māori. The TDHB catchment area recorded: unaided and best-corrected (as is comprised of Gisborne and the East Cape applicable) visual acuities of right and left of the North Island of New Zealand, and eyes, the visual acuity test used, cause of has a population of about 48,000, 50% of visual impairment (if any), ophthalmic whom are New Zealand Māori. The SDHB is examination fi ndings including cycloplegic relatively less deprived compared with the refraction (if performed) or ocular pathology, national average, whereas the TDHB is rela- and management (glasses, penalisation tively more deprived.13 therapy, surgery, observation or discharge). Data for this study were obtained from Community optometrist clinical three sources: New Zealand Ministry of data Health (MoH) B4SC Database, Southern We approached all local community and Tairawhiti DHB eye clinic records, and optometrists within the Southern DHB and community optometrists located within Tairawhiti DHB catchment regions for their the Southern and Tairawhiti DHB regions. voluntary provision of clinical records. Iden- Identifying details of the children from each tifying details of all children aged four to source was used to cross-match screening seven years seeing an optometrist between 1 results of children to any results from April and 31 October 2016 was recorded by follow-up eye-care. Crossmatched cases each optometry practice. Again, this longer were then de-identifi ed. data collection period allowed for delay The Ministry of Health Database between referral and optometry assessment. Screening information for every child who Identifying details of every child who was was screened for vision in Southern and reported as having attended a community Tairawhiti DHBs within the study period was optometry practice was cross-matched to the obtained from the MoH, and the following MoH Database.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 64 www.nzma.org.nz/journal ARTICLE

The same clinical data was collected for Ethical approval every child who presented to a community Ethical approval for this study was optometrist and was also on the MoH obtained from the University of Otago Database (regardless of screening outcome), (Health) Research Ethics Committee. Specifi c as for children seen at DHB clinics. informed consent for participation from Data analysis the children’s parents/caregivers was not Sensitivity, specifi city, PPV and NPV required for this retrospective audit. were estimated for pooled data. The study Māori consultation cohort consisted of children screened who Specifi c consultation occurred with the voluntarily presented to an eye healthcare relevant iwi (Ngāi Tahu and Ngāti Porou) professional, meaning it was not a sample through the Ngāi Tahu Research Consul- representative of all children undergoing tation Committee and Ngāti Porou Hauora. screening. Therefore, the calculation was performed using a ‘best-case, worse-case’ model. A contingency table was constructed Results for the study population, comparing the Data collection screening outcome (pass or fail) to what DHB data was obtained from three the screening outcome would have been if hospitals: Dunedin and Invercargill performed by an eye-care provider (pass Hospitals (SDHB), and Gisborne Hospital or fail), to fi nd totals for true positive, true (TDHB). There were 21 optometry prac- negative, false positive and false negative tices identifi ed in the SDHB region. Three results. These values, along with the practices in the SDHB did not have records screening referral rate and estimations of available for retrospective identifi cation and the total prevalence of amblyopia in the review, and another was unable to supply population, could then be used to construct data for children seen throughout half of another table estimating screening perfor- the data collection period. There were three mance characteristics for the population of optometry practices in the TDHB region, all children who were screened. The estimated of which supplied clinical records. prevalence of visual impairment in New A summary of screening and referral Zealand four-year-olds used was 4.5%.14 results are shown in Table 1. The best-case scenario for screening is that 8.3% of children were referred from the all cases of unaccounted amblyopia belong B4 School Check, about half were identifi ed to children with a positive screen but not as having attended a DHB eye clinic or presenting for follow-up, and worst-case optometry practice; follow-up information that all cases of unaccounted amblyopia was unable to be obtained for referred belong to children who passed screening. children not seen. This allowed the calculation of best-and- worst case ranges for sensitivity, specifi c, Screening outcomes NPV and PPV, in which the true values lie. True positive, false positive, false negative and true negative outcomes from screening The relative performance of each DHB was for the study cohort for both SDHB, TDHB, estimated by calculating the apparent PPV and combined, were determined by for both DHBs from the absolute numbers B4SC referral status and VA measured at of true positives and false positives found in secondary assessment and are shown in the study, as this calculation is independent Table 2. A child was found to have visual of any differences between disease preva- impairment if VA at optometry or DHB clinic lence in the two DHBs. was 6/9 or worse in one eye, or 6/12 or worse Totals of ocular abnormalities and the in both eyes—the B4SC referral criteria. management of children who presented for Table 3 demonstrates the screening perfor- further assessment were also collected, as mance measures calculated from best-case, well as a subgroup of children with severe worst-case analysis. visual impairment, defi ned as VA of 6/24 or The potential range of sensitivity was very worse in either eye. wide (best-case; worst-case range of 94.7%

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 65 www.nzma.org.nz/journal ARTICLE

Table 1: Number of children screened and outcomes of screening of the B4 School Check vision screening in the Southern and Tairawhiti DHBs in New Zealand from 1 April to 30 September 2016.

Number of children (%)

Southern District Tairawhiti District Health Combined Health Board Board Eligible for screening 1,942 (100) 402 (100) 2,344 (100)

Under care of eye health professional at 54 (2.8) 26 (6.5) 80 (3.4) time of screening (thus not screened)

Received screening 1,739 (92.1) 370 (98.4) 2,109 (93.1)

Referral from screening 134 (7.7) 42 (11.4) 176 (8.3)

Referral from screening and seen at 60 (44.8) 31 (77.1) 91 (51.7) optometrist or DHB [29 optometrist; [31 optometrist; 1 DHB] [60 optometrist; 31 DHB] (One case was seen by both 32 DHB] optometrist and DHB)

Referral from screening and not seen at 74 (55.2) 11 (26.2) 85 (48.3) optometrist nor DHB eye clinic

Passed screening and seen at optometrist 10 (0.6) 15 (4.6) 25 (1.3) or DHB eye clinic

Table 2: Outcomes of testing in the B4 School Check vision screening in children subsequently presenting to an optometrist or DHB eye clinic in the Southern and Tairawhiti DHBs in New Zealand from 1 April to 30 September 2016.

Screening outcome Total (%) [SDHB, TDHB]

True positives 52 (44.8) [34, 18]

False positives 39 (33.6) [26,13]

False negatives 5 (4.3) [1, 4]

True negatives 20 (17.2) [9, 11]

Total 116 (100) [70, 46]

Table 3: Performance characteristics of the B4 School Check for the SDHB and TDHB vision screening in the Southern and Tairawhiti DHBs in New Zealand from 1 April to 30 September 2016 in best-case and worst-case scenarios.

Screening outcome Best-case scenario Worst-case scenario True positives [A] 90 52

False positives [B] 86 124

False negatives [C] 5 43

True negatives [D] 1,928 1,890

Total (number) 2,109 2,109

Sensitivity [A/A+C] 94.7% 54.7%

Specificity [D/D+B] 95.7% 93.8%

Positive Predictive Value [A/A+B] 51.1% 29.5%

Negative Predictive Value [D/D+C] 99.7% 97.8%

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 66 www.nzma.org.nz/journal ARTICLE

Table 4: Ocular abnormalities in children screened by the B4SC with visual impairment, and the management by DHB and optometry practices in SDHB and TDHB.

Ocular abnormality Number (%) [optometry, Management Number (%) [optometry, DHB] DHB] Refractive error 53 (93.0) [35, 18] Glasses 47 (82.4) [32, 15]

Review later 6 (10.5) [3, 3]

Strabismus 3 (5.3) [2, 1] Ophthalmology referral 2 (3.5) [2, 0]

Review later 1 (1.8) [0, 1]

Amblyopia 8 (14.0) [5, 3] Glasses 7 (12.2) [4, 3]

Penalisation 3 (5.3) [1, 2]

Ophthalmology referral 1 (1.8) [1, 0]

Other pathology 0 (0.0) [0, 0]

to 54.7%), which means this study is unable One child with severe visual impairment was to provide a meaningful estimate of vision seen at a DHB eye clinic and was diagnosed screening sensitivity. PPV of screening was with anisometropic amblyopia. Two children low (29.5% to 51.1%), however, the estimated with severe visual impairment were seen range of NPV was high (97.8% to 99.7%). at optometry practices. One of these was VA comparison between screening diagnosed with myopia and treated with spectacles, and the other was diagnosed with and optometry and DHB clinic anisometropic amblyopia, and referred to a assessment DHB eye clinic for further management. Screening VA were compared to each child’s corresponding VA from optometrist or DHB eye clinic (the ‘true’ VA), using a Discussion two-sided, paired t-test. The mean VA from This study assesses the accuracy of vision screening was 6/9.80 while the mean ‘true’ screening in two distinct regions of New VA was 6/9.27. A two-sided, paired t-test had Zealand and indicates that there are very a p value of less than 0.05, meaning there is few children with reduced vision who are a statistically signifi cant (but not clinically missed by screening. However, the relatively signifi cant) difference between VA measured low PPV suggests that there is a tendency to by VHT from optometrists and DHB clinics. over-refer and an increased VA threshold for referral would reduce the number of false SDHB and TDHB vision screening positive referrals. comparison The main fi nding of this study is a low The apparent PPV for SDHB was 56.7%, PPV, in the range of 29.5% to 51.1%, for whereas the apparent PPV for TDHB was the vision screening component of the B4 58.1%, suggesting there is likely very little School Check; at least half of referrals have difference in screening performance normal vision. This is consistent with the between the SDHB and TDHB. only other study evaluating the B4SC since Treatment outcomes its inception, which found a PPV of 31% in Any ocular abnormality and management South Auckland,14 as well as with fi ndings of children who were screened and found from other childhood vision screening to have truly reduced vision (less than 6/9 in programmes around the world. Reported on eye) by DHB clinics and optometrists was PPVs range from 35% to 85%, including recorded and is displayed in Table 4. the Rotterdam AMblyopia Screening Effec- Severe visual impairment tiveness Study, a seven-year follow-up study, which found a PPV of 42%.15–17 Reassuringly, Of all children screened for vision in the there was no difference in the apparent PPV SDHB and TDHB, and who received further calculated from the raw outcomes between assessment at a DHB eye clinic or optometry the SDHB and TDHB, suggesting that the practice, there were three children (2.5%) performance of the B4SC in these two areas who met the threshold for severe visual is comparable. impairment (6/24 or worse in either eye).

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 67 www.nzma.org.nz/journal ARTICLE

PPV correlates with prevalence, and over time.19 Again, this is not necessarily amblyopia has a relatively low prevalence inappropriate for a screening programme (1–5%). Also, the function of a screening functioning to identify cases of reduced programme is not to decide which cases do vision, rather than precisely quantify it. or do not have a condition, rather, it is to The main issue identifi ed by this study identify cases at risk and requiring further is the large proportion of children who assessment, and a lower PPV refl ects this. failed screening and were not identifi ed as The implication for a relatively low PPV, presenting to a DHB clinic or community however, is a signifi cant number of unnec- optometrist in the study period, particularly essary referrals are made to DHB eye clinics, in the SDHB; 55% of children referred in the as well as causing unnecessary worry and SDHB were unaccounted for. Explanations inconvenience for parents and caregivers. include children presenting beyond the data The high NPV found is reassuring, collection window, missing records from suggesting children who pass vision optometrists, and children not presenting screening are very unlikely to have a at all after referral. Data collected from visual defi cit. This is ideal for a screening community optometrists was voluntary, programme, particularly one like the B4SC limiting the number of practices supplying where screening is a one-off event, as it data. This potentially concerning fi nding, minimises the number of cases of reduced particularly if many children are indeed vision not detected by screening. The low not presenting for further follow up after PPV and high NPV may mean the B4SC screening referral, warrants further inves- criteria for referral could be raised without tigation as it may be an area that can be negatively impacting on the number of improved. After a referral is generated by children that may be missed, which would the B4SC, the parents are given respon- result in fewer false positive referrals. sibility for organising follow-up at an Further study could investigate the optimal optometry practice, DHB eye clinic or threshold for vision screening referral. Most private ophthalmologist. Potential barriers visual defi cits found in children referred for parents initiating this include the from the B4SC were due to refractive error, perceived costs of optometry appoint- and most of these cases were managed with ments and glasses prescriptions. There are a prescription of glasses. The small number optometry practices which do offer free of children with severe visual impairment services for children and the Children’s seen by optometrists either had refractive Spectacle Subsidy does subsidise prescrip- error amenable to glasses or were appro- tions, however, the parents’ knowledge of priately referred. The implication for this is these services is uncertain.20 DHB appoint- that optometrists are very well placed to be ments, while free, are not equally accessible managing children referred from the B4SC throughout New Zealand—in the SHDB, for and referral directly to them rather than instance, a child can be referred directly to DHB clinics would reduce reliance on DHBs Dunedin Hospital Ophthalmology Service for the management of these children, and from the BS4C, but not to the Invercargill this is the case in the TDHB. Hospital Ophthalmology Service. Therefore, This study also found VA measured at it is important to further investigate what VHT screening is statistically different from is happening to children who are not found VA measured at optometry and DHB eye to present to an eye healthcare professional clinics, assumed to be a child’s true VA. after a vision screening referral, and what However, it is reassuring that this difference barriers, if any, parents have.Another possi- in not clinically signifi cant, despite inherent bility for further study is to sample children limitations of vision screening: suboptimal in the community with gold-standard visual testing environments for VHTs (often in acuity testing and compare to screening preschools with associated distractions) outcomes, allowing determination of true compared to an optometry practice; a values of sensitivity, specifi city, PPV and relative lack of training for VHTs (the Vision NPV. Future vision screening research in in Pre-schoolers Group has shown that the New Zealand may also look at other aspects accuracy of lay-screeners is less than that of screening, such as the role of photo- of trained eye-care professionals);18 and screeners, which are emerging in various that a child’s VA is not necessarily constant societies as an augmentation to current screening programmes.21,22

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 68 www.nzma.org.nz/journal ARTICLE

Competing interests: Nil. Acknowledgements: The authors would like to thank The Maurice and Phyllis Paykel Trust for providing scholarship to William Muller to help fund this study. The authors would also like to thank Pat Tuohy and the Ministry of Health for providing help with the study design and data collection. Prof Nick Wilson provided helpful comments in the manuscript. Author information: William Muller, Department of Medicine, University of Otago, Dunedin School of Medicine, Dunedin; Logan Mitchell, Consultant Ophthalmologist, Dunedin Hospital, Clinical Senior Lecturer, University of Otago, Dunedin School of Medicine, Dunedin; Graham Wilson, Ophthalmologist, Gisborne Hospital, Clinical Senior Lecturer, University of Otago, Gisborne. Corresponding author: Dr Logan Mitchell, Consultant Ophthalmologist, Dunedin Hospital, Clinical Senior Lecturer, University of Otago, Dunedin School of Medicine, Dunedin. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7827

REFERENCES: 1. Attebo K, Mitchell P, 7. Epelbaum M, Milleret 12. Ministry of Health. Cumming R, et al. Prev- C, Buisseret P, Dufi er JL. The B4 School Check: alence and causes of The sensitive period for A handbook for prac- amblyopia in an adult strabismic amblyopia in titioners. Wellington: population. Ophthalmolo- humans. Ophthalmology. Ministry of Health; 2008. gy. 1998 Jan; 105(1):154–9. 1993 Mar; 100(3):323–7. 13. New Zealand Health 2. Holmes JM, Clarke MP. 8. Hendler K, Mehravaran System. 2016. http://www. Amblyopia. Lancet S, Lu X, et al. Refractive health.govt.nz/new-zea- Lond Engl. 2006 Apr 22; Errors and Amblyopia land-health-system/ 367(9519):1343–51. in the UCLA Preschool my-dhb 3. Robaei D, Rose KA, Ojaimi Vision Program; First Year 14. Langeslag-Smith MA, E, et al. Causes and Results. Am J Ophthalmol. Vandal AC, Briane V, et associations of amblyopia 2016 Dec; 172:80–6. al. Preschool children’s in a population-based 9. Matsuo T, Matsuo C, vision screening in New sample of 6-year-old Matsuoka H, Kio K. Zealand: a retrospective Australian children. Arch Detection of strabismus evaluation of referral Ophthalmol Chic Ill 1960. and amblyopia in 1.5- and accuracy. BMJ Open. 2015 2006 Jun; 124(6):878–84. 3-year-old children by a Nov 27; 5(11):e009207. 4. Chua B, Mitchell P. preschool vision-screen- 15. Juttmann R, Rotterdam Consequences of amblyopia ing program in Japan. AMblyopia Screening Effec- on education, occupation, Acta Med Okayama. tiveness Study (RAMSES) and long term vision loss. 2007 Feb; 61(1):9–16. steering committee. The Br J Ophthalmol. 2004 10. Kvarnström G, Jakobsson Rotterdam AMblyopia Sep; 88(9):1119–21. P, Lennerstrand G. Visual Screening Effectiveness 5. Engel-Yeger B. Evaluation screening of Swedish Study (RAMSES): compli- of gross motor abilities and children: an ophthalmo- ance and predictive self perception in children logical evaluation. Acta value in the fi rst 2 years. with amblyopia. Disabil Ophthalmol Scand. 2001 Br J Ophthalmol. 2001 Rehabil. 2008; 30(4):243–8. Jun; 79(3):240–4. Nov; 85(11):1332–5. 6. Daw NW. Critical periods 11. Sheridan MD, Gardiner 16. Peterseim MMW, Papa and amblyopia. Arch PA. Sheridan-Gardiner test CE, Wilson ME, et al. Ophthalmol Chic Ill 1960. for visual acuity. Br Med J. The effectiveness of the 1998 Apr; 116(4):502–5. 1970 Apr; 2(5701):108-9. Spot Vision Screener

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 69 www.nzma.org.nz/journal ARTICLE

in detecting amblyopia lay screeners in the vision Wellington, New Zealand: risk factors. J AAPOS Off in preschoolers study. Ministry of Health, 2013. Publ Am Assoc Pediatr Invest Ophthalmol Vis Sci. 21. Sanchez I, Ortiz-Toquero Ophthalmol Strabismus. 2005 Aug; 46(8):2639–48. S, Martin R, de Juan V. 2014 Dec; 18(6):539–42. 19. Tarczy-Hornoch K, Cotter Advantages, limitations, 17. Chang C-H, Tsai R-K, Sheu SA, Borchert M, et al. and diagnostic accuracy M-M. Screening amblyopia Prevalence and causes of photoscreeners in early of preschool children with of visual impairment detection of amblyopia: a uncorrected vision and in Asian and non-His- review. Clin Ophthalmol stereopsis tests in Eastern panic white preschool Auckl NZ. 2016; 10:1365–73. Taiwan. Eye Lond Engl. children: Multi-ethnic 22. Williams T, Morgan LA, 2007 Dec; 21(12):1482–8. Pediatric Eye Disease High R, Suh DW. Critical 18. Vision in Preschoolers Study. Ophthalmology. Assessment of an Ocular Study Group. Preschool 2013 Jun; 120(6):1220–6. Photoscreener. J Pediatr vision screening tests 20. Hauora M. Review of Ophthalmol Strabismus. administered by nurse Children’s Spectacle 2018 May 1; 55(3):194–9. screeners compared with Subsidy – Final report

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 70 www.nzma.org.nz/journal ARTICLE

Considering evidence for ethnicity bias using assessment case scenarios and medical student correctness and certainty Mike Tweed, Gordon Purdie, Cameron Lacey

ABSTRACT AIMS: There is inequitable distribution of health risks, exposures, resources and outcomes by ethnicity. This may be contributed to by health professional bias. The aim of this study was to investigate the relationship between ethnicity of patients, within written assessment case scenarios, and medical students’ response correctness and certainty. METHODS: Otago Medical School students sit a 150 MCQ progress test with item level response certainty. Patient ethnicity for 60 MCQ case scenarios was varied between two ethnic groups (New Zealand European, Māori) and none specified. Analysis of responses by patient ethnicity was undertaken to compare: odds of correctness; level of certainty; correctness for level of certainty and also by year groups and ability. RESULTS: One thousand one hundred and three students sat the test. There was no significant di erence in odds of correctness or level of certainty by the ethnicity of the patient case scenario. These did not di er significantly by year groups or ability groups, or for correctness by level of certainty. CONCLUSIONS: No systematic di erences in correctness or certainty of student responses to case scenarios by patient ethnicity were detected. Further exploration is warranted, including incorporating more ethnicity descriptors, analysis of incorrect answers, analyses for patterns responses over time by individual students, and selecting questions where varying patient ethnicity is expected to alter the correct response or di iculty.

hen assessing medical students, In a research context, New Zealand assessment content should maxi- medical students have demonstrated mise authenticity,1,2 with realistic implicit and explicit ethnic bias favouring W 2,3 10 clinical problems. To provide better vali- New Zealand Europeans. Medical students dation evidence for results, written assess- may demonstrate preferences for some ments of healthcare students are often based patient demographic groups,11 and simply on clinical scenarios.4,5 Threats to validity including ethnicity in a case description include construct under-representation.6 can infl uence how fi nal-year medical Therefore, patients’ case scenarios/vignettes students talked about patients.12 This vari- should include demographics,7 with assess- ation in response, related to case scenario ment patient demographics matching the ethnicity, might indicate a bias related to the healthcare population.8 When ethnicity, or assessment or a bias related to the candi- other demographics, are only included in dates and warrants further exploration. the case scenario when the correct response Doctors’ practice may be affected by a or item diffi culty is dependent on the de- variety of biases,13–15 with patient demo- mographic, this may serve to cue test-wise graphics infl uencing certainty of diagnosis.16 students, so reducing validation evidence.9

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 71 www.nzma.org.nz/journal ARTICLE

Therefore, there is the possibility that the The current standard policy is that all case correctness of a clinical decision, and also scenarios that relate to an individual patient the certainty in that decision, might be should include their ethnicity and that the infl uenced by perception of the patient ethnicities across the test as a whole should demographics. Self-monitoring, as refl ection- approximate to the New Zealand census in-action of decision certainty, is authentic to population. 17,18 daily practice. Therefore, when assessing To investigate the relationship between medical students there is potential value in the patient ethnicity of the scenario and considering their demonstration of certainty correctness, certainty and the relationship in responses, stratifi ed by self-monitoring between them, six versions of the RKT were descriptors, to questions related to deci- created and students were randomised to sion-making situations, and this has proved which version they received (Table 1). possible in a variety of assessments.19–22 The questions chosen to have variable Building on this, it is possible to explore patient ethnicity were reviewed by clini- how the ethnicity of a patient, as part of an cians (CL, MT) to ensure that the change in assessment case scenario, will affect the ethnicity did not affect the correct answer odds of medical students making correct and/or the perceived diffi culty of the responses, their certainty in those responses question. and the odds of correctness for certainty. The aim of this study is to investigate the The standard questions were questions relationship between ethnicity of patients, where case scenario ethnicity was not within written assessment scenarios, and included as the question was not about an the responses of medical students. individual patient, was vital to the question and could not be varied or was not vital but Method included as per current policy. The Retained Knowledge Test (RKT) is Analysis a progress test sat by all year 2–5 medical Analysis explored whether correctness students at Otago Medical School. The RKT and certainty differed by case scenario is computer-delivered twice per year and ethnicity, student year or student total consists of 150 MCQ questions in random number responses correct. order. These tests are not answered under To estimate the odds ratios of a correct examination conditions, with the students answer between levels of certainty, year having a two-week window to respond but cohort and meeting the standard, among are encouraged to answer in a single sitting all 150 questions a mixed model logistic if possible. Also the RKT includes student regression was used with terms for level of certainty in their responses, with certainty certainty, student year, meeting the standard being categorised as no, low, moderate or for the year cohort, question group, question high, based on the students’ perception of and a random term for student. Level of the need for additional resources, should certainty interaction terms with student this question be an actual clinical situation.21 year and standard were added to the model

Table 1: Patient ethnicity across six versions of the RKT.

Patient ethnicity for questions RKT version 1–90 91–110 111–130 131–150 1 Standard Māori NZ Euro None

2 Standard Māori None NZ Euro

3 Standard NZ Euro None Māori

4 Standard NZ Euro Māori None

5 Standard None Māori NZ Euro

6 Standard None NZ Euro Māori

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 72 www.nzma.org.nz/journal ARTICLE

to test for differences in levels of certainty Ethics approval odds ratios by student year and standard. To University of Otago Ethics committee estimate the odds ratios of a correct answer approval was granted (reference number between ethnicity of case scenarios, among D17/126). non-standard questions, a mixed model logistic regression was used with terms for ethnicity of case scenario, student year, Results question group, question and random term One thousand one hundred and three for student. Level of certainty, meeting the students took the RKT, with number of standard, and interaction terms of ethnicity students, correct responses and levels of of case scenario with level of certainty, certainty described in Table 2. student year, standard and question were For all 150 questions, the odds of being added to the model to test for differences correct were signifi cantly related to level in case scenario ethnicity odds ratios by of certainty, year and meeting the standard level of certainty, student year, standard (all p<0.0001) (Table 3). The odds ratios for and question. For the odds ratios of levels of level of certainty were signifi cantly different certainty between ethnicity of case scenarios between the year cohorts (p<0.0001) and a model was used with the same terms and whether the student met the standard multinomial distribution and cumulative (p<0.0001). logit link was used. Meeting the standard There were no signifi cant differences and interaction terms of ethnicity of case in the odds of a correct answer between scenario with student year, standard and the ethnicity scenarios (p=0.84) (Table 4). question were added to the model to test These odds ratios did not differ signifi cantly for differences in case scenario ethnicity between levels of certainty (p=0.23), the year odds ratios by student year, standard and cohorts (p=0.29), whether the student met question. When there was a signifi cant the standard (p=0.92) or question (p=0.57). interaction, tests for differences between There were no signifi cant differences in ethnicity of case scenarios for each level the odds of increasing certainty between of the other term in the interaction were the ethnicity of case scenarios (p=0.33). The adjusted for multiple comparisons with odds ratio for increasing certainty between the Holm-Bonferroni method. The glimmix Māori and New Zealand European was 0.98 procedure of SAS 9.4 (SAS Institute Inc., (95%CI 0.94, 1.02; p=0.23) and between no Cary, North Carolina, US) was used, with ethnicity and New Zealand European 1.00 between-within degrees of freedom. (95%CI 0.97, 1.04; p=0.87). These odds ratios

Table 2: Students and responses.

Year 2 Year 3 Year 4 Year 5 Total Took RKT 288 276 262 277 1,103

Above standard 68.8% (198) 78.3% (216) 84.4% (221) 91.7% (254) 80.6% (889)

Correct (%)* 31 (25–36) 39 (34–44) 50 (43–55) 57 (51–64) 43 (33–54)

Certainty† None (%) 10 (0–74) 47 (1–68) 25 (1–50) 8 (0–36) 24 (0–61)

Low (%) 52 (16–85) 24 (13–53) 31 (18–48) 31 (19–48) 30 (16–62)

Moderate (%) 9 (5–16) 14 (7–21) 19 (11–27) 23 (15–32) 15 (8–25)

High (%) 2 (1–5) 7 (2–13) 11 (4–20) 18 (8–29) 7 (2–17)

%percentage, median and interquartile range. *Percent of 150 questions, †Percent of questions answered.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 73 www.nzma.org.nz/journal ARTICLE

Table 3: Odds ratio of a correct answer for level of certainty, student year and meeting the standard.

Percent correct Odds ratio (95% confidence interval)* P Certainty None 31.1% (16,557/53,310) 1.00

Low 35.5% (22,523/63,416) 1.71 (1.65, 1.77) <0.0001

Moderate 63.5% (17,931/28,230) 4.58 (4.40, 4.78) <0.0001

High 83.2% (15,644/18,797) 11.94 (11.31, 12.62) <0.0001

Year 2 31.1% (13,363/42,937) 1.00

3 39.4% (16,193/41,072) 1.30 (1.20, 1.41) <0.0001

4 49.8% (19,255/38,680) 1.77 (1.63, 1.91) <0.0001

5 58.1% (23,844/41,064) 2.16 (2.00, 2.34) <0.0001

Standard Below 28.5% (8,738/30,692) 1.00

Above 48.0% (63,917/ 133,061) 1.93 (1.80, 2.08) <0.0001

*from multivariable model. did not differ signifi cantly between the year contributing to this.14,15,23,24 This is high- cohorts (p=0.74) or whether the student met lighted for outcomes for Māori compared the standard (p=0.56). The odds were signifi - with New Zealand European for many cantly differed between questions (p=0.048). healthcare indicators.25,26 Robust processes There were no individual questions with are required in the development of investi- signifi cant differences between ethnicity of gations of racial/ethnic bias in assessment case scenarios after adjusting for multiple of medical student and healthcare profes- comparisons. sionals.27 This study included three versions of each question with the ethnicity of the Discussion patient in the scenario varied, and reviewed by clinicians to ensure that no difference No systematic differences in correctness in responses was to be expected. The or certainty of student responses between randomised design of question allocation case scenario patient ethnicities were and randomised order of question delivery, detected. Given that there were increased with the responses to 60 questions by 1,103 odds of being correct with increased students add to the robustness of the results, certainty, year group and ability which are which have narrow confi dence intervals. similar to results described previously,21 it is likely that the students approached this RKT There are limitations to this study. Several similarly to other assessments. factors could have an effect on responses. All questions that related to individual There is considerable literature on the patients had ethnicity described, apart from relationships between patient ethnicity and where removed for the study, with three healthcare outcomes, with many factors

Table 4: Odds ratio of a correct answer for ethnicity of case scenarios.

Ethnicity of case scenario Percent correct Odds ratio (95% confidence interval) P NZ European 45.3% (9,889/21,852) 1.00

Māori 45.1% (9,867/21,873) 0.99 (0.95, 1.03) 0.56

None 45.2% (9,886/21,871) 1.00 (0.95, 1.04) 0.85

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 74 www.nzma.org.nz/journal ARTICLE

versions of each of these study questions, possible that paper-based case scenarios will which should minimise any effect due to not reveal bias that is present.24 Variation in cueing. All questions were reviewed by certainty could refl ect cognitive load, and practising clinicians to ensure no expected there was no difference detected at different difference in correct answer or diffi culty, levels of certainty or ability in this study. which should minimise any effect due to This study set out to investigate for the complexity. Students are already used to presence of bias in clinical decision-making. ethnicity descriptors, which may or may not Bias related to the perceived ethnicity of be vital to the correct response and/or diffi - the patient may be primarily due to other culty of a question, which should minimise factors such as personal interactions, any effect due to unfamiliarity. rather than clinical decision-making.10,11,14,15 Another limitation is that patient ethnicity In addition, the biases may affect health bias by students may have led to a different outcomes related to some diagnostic condi- 24 incorrect response being given. The analysis tions more than others. However, there here was correct/incorrect only, where all were no differences between case scenario incorrect responses were treated the same. ethnicities detected in this study between An analysis the safety of different incorrect questions for certainty or for individual responses allowing for certainty would add question level analysis for certainty. to the analysis,22 but was not possible as the This study does raise future areas for incorrect responses had not been stratifi ed investigation. One question worth exploring for safety. would relate to differences in responses The case scenario ethnicities used as when the patient ethnicity does make a demographic descriptors in this analysis difference to correct response or diffi culty, 28 were limited to Māori and New Zealand such as sore throat management. Given European. There is greater ethnic diversity the important association between patient in current day New Zealand. Bias across ethnicity and health both in New Zealand 23,26,27 many ethnicities could be investigated in a and internationally, there is some study with a suffi cient number of questions. risk associated with unintended learning consequences of assessments using only It may be that patient ethnicity bias that scenarios in which patient ethnicity has no would be expressed in clinical practice were impact on correct response or diffi culty. not necessarily detectable in this research. This requires exploration as it may promote This was a single low-stakes test with cohort- students to ignore inclusion of patient level analysis, so inferring an absence of ethnicity in decision making, which is not bias among all individual students, espe- desirable for practice. cially related to high-stakes assessments and subsequent practice, was not possible. It is In conclusion, analysis of student responses conceivable that a small number of students allowing for the ethnicity of the patient in the may have had response patterns that may case scenario can be used to investigate for indicate a degree of bias, but that this infor- evidence of variations in response patterns mation was lost in the cohort level analysis. that may indicate bias based on ethnicity. Extrapolating to the high cognitive load and In this study, no difference in correctness or stress of medical practice is limited and it is certainty was detected.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 75 www.nzma.org.nz/journal ARTICLE

Competing interests: Nil. Author information: Mike Tweed, Senior Lecturer, Department of Medicine, University of Otago, Wellington; Gordon Purdie, Biostatistician, University of Otago, Wellington; Cameron Lacey, Senior Lecturer, Māori Indigenous Health Institute (MIHI), University of Otago, Christchurch. Corresponding author: Dr Mike Tweed, Senior Lecturer, Department of Medicine, University of Otago, PO Box 7343, Wellington 6242. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7828

REFERENCES: 1. van der Vleuten C. Validity 8. Aeder L, Altshuler L, General Internal Medicine. of fi nal examinations in Kachur E, et al. The” 2013; 28(11):1504–10. undergraduate medical Culture OSCE”-Introducing 14. Hall WJ, Chapman MV, Lee training. BMJ: British a formative assessment KM, et al. Implicit racial/ Medical Journal. 2000; into a postgraduate ethnic bias among health 321(7270):1217–9. program. Education for care professionals and 2. Round J, Conradi E, Poulton Health. 2007; 20(1):11. its infl uence on health T. Improving assessment 9. Sarnacki RE. The effects of care outcomes: a system- with virtual patients. Med test-wiseness in medical atic review. American Teach. 2009; 31(8):759–63. education. Evaluation & Journal of Public Health. 3. Epstein RM. Assessment in the Health Professions. 2015; 105(12):e60–e76. medical education. New 1981; 4(2):207–21. 15. Maina IW, Belton TD, England Journal of Medi- 10. Harris R, Cormack D, Ginzberg S, et al. A decade cine. 2007; 356(4):387–96. Stanley J, et al. Ethnic of studying implicit racial/ 4. Boland RJ, Lester NA, bias and clinical deci- ethnic bias in healthcare Williams E. Writing sion-making among New providers using the implicit multiple-choice questions. Zealand medical students: association test. Social Academic Psychiatry. an observational study. Science & Medicine. 2017. 2010; 34(4):310. BMC Medical Educa- 16. Lutfey KE, Link CL, Grant tion. 2018; 18(1):18. 5. Beullens J, Van Damme RW, et al. Is certainty more B, Jaspaert H, Janssen PJ. 11. Haider AH, Sexton J, Sriram important than diagnosis Are extended-matching N, et al. Association of for understanding race multiple-choice items unconscious race and social and gender disparities?: An appropriate for a fi nal test class bias with vignette- experiment using coronary in medical education? Med based clinical assessments heart disease and depres- Teach. 2002; 24(4):390–5. by medical students. JAMA. sion case vignettes. Health 2011; 306(9):942–51. Policy. 2009; 89(3):279–87. 6. Downing SM, Haladyna TM. Validity threats: 12. Ewen S, Barrett J, Paul 17. Eva KW, Regehr G. overcoming interference D, et al. When a patient’s Self-assessment in the with proposed interpre- ethnicity is declared, health professions: a tations of assessment medical students’ deci- reformulation and research data. Medical Education. sion-making processes are agenda. Academic Medi- 2004; 38(3):327–33. affected. Internal Medicine cine. 2005; 80(10):46–54. Journal. 2015; 45(8):805–12. 7. Fors UGH, Muntean V, 18. Eva KW, Regehr G. Botezatu M, Zary N. 13. Chapman EN, Kaatz A, Knowing when to look it Cross-cultural use and Carnes M. Physicians up: a new conception of development of virtual and implicit bias: how self-assessment ability. patients. Med Teach. doctors may unwittingly Academic Medicine. 2009; 31(8):732–8. perpetuate health care 2007; 82(10):S81–4. disparities. Journal of

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 76 www.nzma.org.nz/journal ARTICLE

19. Tweed MJ, Thomp- tional information from ities in New Zealand: son-Fawcett M, Schwartz P, multiple choice question cross-sectional study. Wilkinson TJ. A confi dence assessments. BMC Medical Lancet. 2006; 367:2005–09. and safety approach to Education. 2017; 17(1):106. 26. Robson B, Harris R. Hauora: MCQ scoring. Focus on 23. Smedley BD, Stith AY, Màori Standards of Health Health Professional Educa- Nelson AR. Institute of IV. A study of the years tion: A Multi-disciplinary Medicine, Committee 2000–2005. Wellington: Te Journal. 2012; 13(3):84–92. on Understanding and Ropu Rangahau Hauora 20. Tweed M, Schwartz P, Eliminating Racial and a Eru Pomare. 2007. Thompson-Fawcett M, Ethnic Disparities in 27. Harris R, Cormack D, Curtis Wilkinson TJ. Determining Health Care. Unequal E, et al. Development measures of insight and treatment: confronting and testing of study tools foresight from respons- racial and ethnic dispar- and methods to examine es to multiple choice ities in health care. ethnic bias and clinical questions. Med Teach. Washington, DC: National decision-making among 2013; 35(2):127–33. Academies Press; 2003. medical students in 21. Tweed M, Purdie G, 24. Paradies Y, Ben J, Denson New Zealand: The Bias Wilkinson T. Low N, et al. Racism as a and Decision-Making in performing students determinant of health: Medicine (BDMM) study. have insightfulness when a systematic review and BMC Medical Education. they refl ect-in-action. meta-analysis. PloS one. 2016; 16(1):173. Medical Education. 2015; 10(9):e0138511. 28. Heart Foundation NZ. 2017; 51(3):316–23. 25. Harris R, Tobias M, Group a streptococcal 22. Tweed M, Stein S, Wilkin- Jeffreys M, et al. Effects of sore throat management son T, et al. Certainty self-reported racial discrim- guideline. 2014. and safe consequence ination and deprivation on responses provide addi- Māori health and inequal-

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 77 www.nzma.org.nz/journal ARTICLE

District health board of residence, ethnicity and socioeconomic status all impact publicly funded insulin pump uptake in New Zealand patients with type 1 diabetes Benjamin J Wheeler, Rhiannon Braund, Barbara Galland, Anastasia Mikuscheva, Esko Wiltshire, Craig Je eries, Michel de Lange

ABSTRACT AIMS: Insulin pump therapy (CSII) is becoming increasingly common for those living with type 1 diabetes (T1D), and has been publicly funded in New Zealand since 2012. The aim of the current study was to examine national uptake of publicly funded pumps from 2012 to 2016, with a focus on the proportion of patients using pumps analysed according to district health board (DHB) as well as demographic characteristics. METHODS: Data from nationally held data collections including the New Zealand Virtual Diabetes Register were used to calculate the overall and subgroup proportions using pumps. Logistic regression analysis was then used to estimate the independent contributions of DHB of residence and sociodemographic characteristics to variations in pump use. RESULTS: Between 2012 and 2016, CSII for those living with T1D (n=17,338) increased from 1.6 to 11.3% overall. However, speed of uptake di ered by DHB of residence, ethnicity, degree of deprivation, age and gender. A four-fold di erence in uptake between highest and lowest using DHBs was seen a er adjusting for known confounders. CONCLUSIONS: From 2012 to 2016 there has been a steadily increasing uptake of CSII. Despite publicly funded access, disparities in use appear to exist, including by DHB of residence as well as traditionally described socio-demographic barriers to healthcare. E orts to understand and reduce these disparities are required.

ntensive diabetes management has be- traditional injection regimens, including im- come the standard of care in type 1 dia- provements in glycaemic control, reductions betes (T1D) since landmark trials such as in severe hypoglycaemia and improvements I 5–10 the Diabetic Control and Complications Trial in patient-reported quality of life. (DCCT) and the Epidemiology of Diabetes Despite ongoing technological advances, Intervention and Complication Trial (EDIC) many are still not achieving target blood confi rmed that tighter glycaemic control glucose control.3,11 In particular, those of signifi cantly reduces rates of long-term com- lower socioeconomic position and non-Eu- 1,2 plications. With this has come an increase ropean ethnicity appear particularly in uptake of insulin pump therapy (CSII) and disadvantaged.12,13 There are many factors other forms of associated diabetes tech- that are likely to infl uence these observa- 3,4 nology, particularly among children. CSII tions, including inequalities in healthcare offers a range of potential advantages over due to deprivation as well other factors.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 78 www.nzma.org.nz/journal ARTICLE

Inequalities in access to diabetes technology information attached to a unique alpha- have now been confi rmed in single-centre numeric patient identifi er, the National cohorts,14,15 Canadian,16 German17 and New Health Index [NHI]), the National Minimum Zealand national data.4 Dataset (hospital discharges from 1988), The New Zealand Pharmaceutical the Mortality Collection (all hospital and Management Agency (PHARMAC) intro- community deaths), and the Pharmaceutical duced publicly funded access to CSII in Collection (PHARMS, all claims by commu- September 2012. Access to funding can be nity-based pharmacists for the dispensing complex, and has undergone some changes of publicly funded prescription pharmaceu- since 2012. Currently there are two main ticals and associated equipment, including categories: 1) those experiencing severe all dispensing of diabetes-related drugs/ unexplained and recurrent hypoglycaemia, devices for New Zealand). The data in each and 2) those with suboptimal glycaemic of these collections are indexed to indi- control (defi ned as typical HbA1c results of vidualised patient National Health Index between 65 and 90mmol/mol. Up to 2014, numbers or NHIs (from 2005 onwards in uptake was continuing to rise for all groups, PHARMS), allowing linkage of patient-level but was considerably slower in those of demographic, hospital discharge, mortality non-European ethnicity, and in those who and pharmaceutical dispensing data. were more socioeconomically deprived.4 Diagnoses in the National Minimum Both these factors, while related, were Dataset and the Mortality Collection are independent predictors of CSII4 uptake. coded to successive revisions of the Inter- What remains unclear is why this should be national Statistical Classifi cation of Diseases the case given the overall improved public and Related Health Problems, Australian access provided by PHARMAC. In addition, Modifi cation (ICDAM). The Ministry of the role that New Zealand’s district health Health also holds the Virtual Diabetes board (DHB) structure may play in access to Register (VDR), an annually updated CSII has not been examined. DHBs in New national register of all patients with any Zealand are diverse in not only geography, form of diabetes mellitus (except gesta- but also in population sociodemographic tional diabetes) identifi ed using records of make up. Funding priorities and challenges hospital admissions with diabetes, diabetes also differ, as well as access to sub-speciality outpatient clinic visits, retinal screening, services. Diabetes specialists may also differ repeated HbA1c laboratory tests, and insulin in their comfort or support for new tech- or oral hypoglycemic use.21,22 At the time the nologies, as seen in recent paediatric data study was initiated (November 2017), linked from Australasia, where access to intensive data from the latest VDR iteration22 and the therapy from new diagnosis of T1D differed above collections were available up to 31 based on geographical location.18 December 2016. The aim of the current study was Study population therefore to examine the New Zealand-wide The Ministry of Health identifi ed all national uptake of publicly funded CSII patients who appeared on the VDR at any from 2012 to 2016 (being the most recent time between 1 September 2012, and 31 complete year data at the time of data December 2016 (the study period). This request), with a focus on the proportion was then linked to demographic infor- of patients using pumps with analysis mation, diabetes-related dispensing records, according to DHB and demographic hospital discharge, and mortality data characteristics. from the relevant National Collections. All data were indexed to an encrypted NHI to Methods preserve anonymity, prior to providing to Data sources the study team. The New Zealand Ministry of Health holds As the VDR does not distinguish between several national data collections covering types of diabetes, we utilised a previously 4 the estimated 4.87 million people living in developed and published algorithm to New Zealand.19 These data sets provided all identify and exclude diabetes patients data for this study,20 including the National affected by non-T1D subtypes (eg, type 2 Health Index Collection (demographic diabetes, cystic fi brosis, monogenic and

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 79 www.nzma.org.nz/journal ARTICLE

Figure 1: Study fl ow diagram.

neonatal forms of diabetes). The algorithm a numerical indication of deprivation for was further adapted to capture T1D patients small geographical units can be calculated, who may have taken metformin (tradi- using a 1–10 scale (1 representing the least tionally used for type 2 diabetes) at some deprived and 10 representing the most stage currently or in the past (Figure 1). Due deprived). These were converted to quintiles to coding issues, some patients determined for all data presented in this study. by our algorithm to have T1D had had at Patient characteristics at study entry least one hospital discharge with a coded were taken from the fi rst insulin dispensing diagnosis of type 2 diabetes. These patients record during the study period. For the were retained in the study population, analysis of insulin pump use by demo- but they were fl agged so that a sensitivity graphic characteristics and region over analysis could be undertaken. time, we used the values of the prioritised Population characteristics ethnicity, NZDep2006 score, and funding For each member of the study population, DHB associated with the fi rst insulin basic demographic and socioeconomic dispensing in the relevant calendar year. data was available, including prioritised Statistical analysis self-identifi ed ethnicity, the funding district The methods shown in Figure 1 were used health board (DHB), as recorded at the to ascertain the number of patients with T1D time of each dispensing claim, and the during the study period, and in each year New Zealand Index of Deprivation 2006 (annual denominators). Similarly, we ascer- 23 (NZDep2006), a validated measure of tained the number of patients who received socioeconomic position (using an area-based at least one dispensing of an insulin pump measure of social deprivation). In brief, the and/or pump consumables per calendar NZDep2006 Index enables assignment of year (annual numerators). We then deter- levels of social and economic deprivation mined the annual numbers of patients with to a given residential address. From this, T1D according to their DHB of residence and

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 80 www.nzma.org.nz/journal ARTICLE

Figure 2: Insulin pump uptake over time as a proportion of the total T1D population by individual patient characteristics.

(A) Gender; (B) ethnicity, NZ/EUR–NZ European/European; (C) deprivation quintile (Q1 least deprived, Q5 most deprived—derived from NZdeprivation2006 score); (D) age in years. sociodemographic characteristics, identifi ed pump users within these subgroups and Results calculated the corresponding proportions. Patients with T1D were identifi ed from national data collections (see Figure 1). The Pearson’s Chi-square tests were used to basic demographics of these 17,338 patients compare differences between proportions. identifi ed with T1D are seen in Table 1. CSII We repeated these analyses in a sensitivity uptake, between September 2012 (when CSII analysis confi ned to the subgroup of patients was publicly funded) and the end of 2016, who had at least one hospital discharge with as a proportion of total the T1D population a diagnosis of type 1 diabetes (T1D) and and by individual patient characteristics are no discharge diagnoses of type 2 diabetes, shown in Table 2, as well as graphically in as well as to those included who had been Figure 2. prescribed metformin. Finally, to explore the independent contributions of DHB region These highlight that during the fi ve years and sociodemographic characteristics to vari- of publicly funded CSII, from 2012 to end of ations in pump use, we undertook a logistic 2016 the proportion using CSII has increased regression analysis to estimate odds ratios from 1.6% in 2012 to 11.3% in 2016. Over (ORs) and 95% confi dence intervals (95% this period increases are seen in the CIs), adjusted for the demographic character- proportions using CSII among all available istics and region. All statistical analyses were individual patient characteristics (eg, age, performed using R version 3.4.24 gender, ethnicity, level of deprivation and DHB of residence), however the rate of Ethical approval increase was not equally distributed (Table The study received ethical approval 2, Figure 2, Table 4). from the University of Otago Human Ethics Committee (Health), reference number To further explore this, logistic regression HD17/022. analyses were conducted. Table 3 reveals

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 81 www.nzma.org.nz/journal ARTICLE

Table 1: Characteristics of the study population to be using a pump than those in decile 1 at entry. (decile 5 OR 0.6 [95% CI 0.5–0.7]). To test assumptions made in our T1D study algo- Characteristic Number (%) rithm (Figure 1), further sensitivity analyses (n= 17,338) were done. The results and odds ratios were Gender very similar in sensitivity analyses confined to the subgroup of the study population Female 8,657 (49.9) with at least one discharge diagnosis of type Male 8,681 (50.1) 1 diabetes and no type 2 diagnoses (data not shown), and with additional analyses Age group (years) subtracting those who had been previously 0–9 931 (5.4) been exposed to metformin. Table 4 high- lights regional disparities based on DHB of 10–19 2,454 (14.2) residence adjusting for age, gender, ethnicity 20–29 2,929 (16.9) and deprivation. These reveal a greater than four-fold difference in CSII between the 30–39 3,140 (18.1) DHBs with the highest and lowest uptake, 40–49 2,532 (14.6) despite identical PHARMAC access. In addition, for three DHBs (not for the other 50–59 2,295 (13.2) 17), there is a signifi cant interaction between 60–69 1,747 (10.1) childhood age and DHB. As seen in Figure 2 children are more likely to get pumps 70 + 1,310 (7.6) than adults overall, and this relationship is Ethnicity (prioritised) stronger for Counties Manakau, Midcentral and Nelson Marlborough. In Counties NZ European 13,142 (75.8) Manakau, the OR for child is between 4.3 and Māori 1,753 (10.1) 10.5, and for an adult it is between 0.7 and 1.4; In Midcentral, these ORs are [4.4–11.2] Asian 1,134 (6.5) for a child and [0.7–1.6] for an adult, and in Pacific 728 (4.2) Nelson they are [5.4–13.1] and [0.7–1.6]. Other 236 (1.4) Discussion Unknown 345 (2.0) This study examines national patterns of NZ Deprivation (06) quintile insulin pump uptake, in New Zealanders living with T1D, from the introduction of 1 (least deprived) 3,061 (17.8) PHARMAC publicly funded access criteria. 2 3,223 (18.7) The main fi ndings are that since funding was introduced, access has considerably 3 3,459 (20.1) expanded. Overall New Zealand uptake of 4 4,080 (23.7) CSII was at 11.3% by end of 2016, compared to at least 40% in the US,25 38% in Canada16 5 (most deprived) 3,420 (19.8) and 16% in Italy.26 Despite public access criteria, several sociodemographic disparities that certain patient characteristic were in access appear to exist. When controlled for signifi cantly less likely to access CSII, when all available variables, these include differ- all other variables are controlled for. Specif- ences in utilisation by DHB, age, gender, ically, females were more likely than males ethnicity and socioeconomic position. to be using a pump (adjusted OR 2 [(1.8–2.2)] (this was particularly so from age 20 years Insulin pump uptake has not previ- to age 60 years—data not shown); Maori, ously been examined by DHB including Pacific and Asian patients were significantly controlling for other potential confounders. less likely to be using a pump than New The key fi nding of this study is that DHB of Zealand Europeans (adjusted ORs 0.4 [95% residence is independently associated with CI 0.3–0.4], 0.2 [95% CI 0.1–0.3] and 0.2 [95% insulin pump uptake despite theoretically CI 0.1–0.3], respectively); and those in more equal access to PHARMAC access criteria deprived deciles were significantly less likely (up to four-fold difference between highest

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 82 www.nzma.org.nz/journal ARTICLE

Table 2: Annual proportions (%) of patients with type 1 diabetes using an insulin pump, overall and by demographic characteristics.

Characteristic Number of pump users/number with type 1 diabetes (%) 2012 2013 2014 2015 2016

Total 213/12,985 (1.6) 916/13,629 (6.7) 1,290/14,131 (9.1) 1,496/14,479 (10.3) 1,685/14,931 (11.3)

Gender Female 126/6,028 (2.1) 570/6,410 (8.9) 799/6,674 (12.0) 920/6,802 (13.5) 1,025/7,073 (14.5)

Male 87/6,957 (1.3) 346/7,219 (4.8) 491/7,457 (6.6) 576/7,677 (7.5) 660/7858 (8.4)

Age group (yr) 0–9 45/360 (12.5) 89/369 (24.1) 113/410 (27.6) 133/457 (29.1) 137/465 (29.5)

10–19 85/1,759 (4.8) 291/1,808 (16.1) 376/1,788 (21) 395/1,797 (22) 432/1,817 (23.8)

20–29 16/2,017 (0.8) 107/2,161 (5) 185/2,345 (7.9) 231/2,424 (9.5) 272/2,530 (10.8)

30–39 23/1,928 (1.2) 138/2,067 (6.7) 190/2,185 (8.7) 236/2,252 (10.5) 247/2,383 (10.4)

40–49 17/2,121 (0.8) 113/2,172 (5.2) 180/2,197 (8.2) 205/2,161 (9.5) 249/2,215 (11.2)

50–59 12/2,027 (0.6) 91/2,074 (4.4) 116/2,117 (5.5) 144/2,159 (6.7) 171/2,164 (7.9)

60–69 9/1,562 (0.6) 66/1,649 (4) 96/1,695 (5.7) 115/1,767 (6.5) 132/1,827 (7.2)

70+ 6/1,211 (0.5) 21/1,329 (1.6) 34/1,394 (2.4) 37/1,462 (2.5) 45/1,530 (2.9)

Ethnicity (prioritised) NZ European 189/10,523 (1.8) 826/10,927 (7.6) 1,161/11,221 (10.3) 1,344/11,455 (11.7) 1,505/11,715 (12.8)

Māori 12/1,174 (1) 45/1,272 (3.5) 64/1,361 (4.7) 78/1,410 (5.5) 99/1,477 (6.7)

Asian 5/435 (1.1) 14/522 (2.7) 20/594 (3.4) 27/641 (4.2) 27/736 (3.7)

Pacific 2/439 (0.5) 11/479 (2.3) 15/501 (3) 17/523 (3.3) 18/539 (3.3)

Other 3/126 (2.4) 6/147 (4.1) 6/160 (3.8) 9/160 (5.6) 13/172 (7.6)

NZDep20061 1 (least deprived) 54/2,330 (2.3) 224/2,436 (9.2) 303/2,527 (12) 354/2,593 (13.7) 395/2,711 (14.6)

2 53/2,420 (2.2) 189/2,546 (7.4) 280/2,637 (10.6) 329/2,728 (12.1) 361/2,806 (12.9)

3 38/2,580 (1.5) 186/2,724 (6.8) 264/2,815 (9.4) 316/2,895 (10.9) 345/2,972 (11.6)

4 43/102 (1.4) 208/3,236 (6.4) 276/3,316 (8.3) 303/3,395 (8.9) 352/3,484 (10.1)

5 (most deprived) 25/2,502 (1) 108/2,630 (4.1) 164/2,772 (5.9) 190/2,799 (6.8) 227/2,882 (7.9)

1NZDep2006—New Zealand Deprivation index 2006.

and lowest using DHBs). Differing access access criteria,16 which may be impacting to diabetes regimens has previously been access to CSII for individuals living in some described in children at new diagnosis in DHBs. These factors appear more likely Australasia, with quite different approaches than inadequate size of DHB or inadequate taken by diabetes teams depending on access to specialised services in some DHBs, centre of treatment.27 This suggests that as differences were also observed between availability of diabetes team members with DHBs with tertiary specialist services and experience or enthusiasm for new diabetes large urban populations, eg, Canterbury technologies may in part be an infl uencing DHB had one of the lower proportions of factor in CSII uptake. Staff may also be using CSII use. Other factors that need further personal judgements28 or centre-specifi c exploration are differences in funding and

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 83 www.nzma.org.nz/journal ARTICLE

Table 3: Proportions of patients with type 1 diabetes using an insulin pump in 2016 according to demo- graphic characteristics, with crude and adjusted odds ratios.

Characteristics Proportion using an Crude odds ratio Adjusted odds insulin pump in 2016 (95% CI) ratioa (95% CI) Gender Female (14.5) 1.8 (1.7–2.1) 2 (1.8–2.2)

Male (ref) (8.4) 1.0 1.0

Age group (years) 0–9 (29.5) 4.3 (3.6–5.2) 4.6 (3.8–5.7)

10–19 (23.8) 1.8 (1.5–2.1) 1.8 (1.5–2.2)

20–29 (ref) (10.8) 1 1

30–39 (10.4) 1.1 (0.9–1.3) 1.1 (0.9–1.3)

40–49 (11.2) 0.9 (0.8–1.1) 0.9 (0.7–1.1)

50–59 (7.9) 0.7 (0.6–0.9) 0.7 (0.5–0.8)

60–69 (7.2) 0.5 (0.4–0.7) 0.5 (0.4–0.6)

70+ (2.9) 0.2 (0.1–0.3) 0.2 (0.1–0.3)

Ethnicity (prioritised) NZ European (ref) (12.8) 1 1

Māori (6.7) 0.5 (0.4–0.6) 0.4 (0.3–0.4)

Asian (3.7) 0.3 (0.2–0.4) 0.2 (0.1–0.3)

Pacific (3.3) 0.2 (0.1–0.4) 0.2 (0.1–0.3)

Other (7.6) 0.6 (0.3–0.9) 0.4 (0.2–0.8)

Unknown (7.9) 0.6 (0.4–0.9) 0.7 (0.5–1.1)

NZDep2006 quintileb 1 (least deprived) (ref) (14.6) 1 1

2 (12.9) 0.9 (0.7–1) 0.9 (0.7–1.0)

3 (11.6) 0.8 (0.7–0.9) 0.8 (0.6–0.9)

4 (10.1) 0.7 (0.6–0.8) 0.6 (0.5–0.8)

5 (most deprived) (7.9) 0.5 (0.4–0.6) 0.6 (0.5–0.7)

aOdds ratios are adjusted for DHB and all other variables in the table. bNZDep2006—New Zealand Deprivation Index 2006. staffi ng levels between various diabetes Zealand and international research showing DHB diabetes teams available to support disparities in access to CSII due to sociode- technological advances. These have been mographic factors.4,17,30 Importantly, in this previously shown to be quite disparate 2016 data, disparities in access for those of between different Australasian centres, Māori and Pacifi c ethnicity were present, including those within New Zealand.29 and appear to be independent of socio- The other variables that appear to economic deprivation, which was also an infl uence CSII uptake were age, gender, independent predictor of pump uptake. This ethnicity and socioeconomic position. These independent relationship to inequality by fi ndings confi rm the results of previous New ethnicity and deprivation has been seen in

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 84 www.nzma.org.nz/journal ARTICLE

Table 4: Proportion of patients in 2016 using pump therapy by district health board funding region.1

District health board Proportion using pump (%) Crude odds ratio Adjusted odds ratio2 Canterbury 161/1,947 (8.3) 1 (ref) 1 (ref)

Auckland 193/1,603 (12) 1.5 (1.2–1.9) 1.7 (1.3–2.1)

Waitemata 135/1,598 (8.4) 1 (0.8–1.3) 1.3(1–1.6)

Waikato 238/1,464 (16.3) 2.2 (1.7–2.7) 2.8 (2.3–3.5)

Southern 192/1,264 (15.2) 2 (1.6–2.5) 2.2 (1.7–2.7)

Capital and Coast 126/1,078 (11.7) 1.5(1.1–1.9) 1.6 (1.3–2.1)

Counties Manukau 79/1,026 (7.7) 0.9 (0.7–1.2) 1.4 (1–1.8)

Bay of Plenty 96/668 (14.4) 1.9 (1.4–2.4) 2.4 (1.8–3.2)

Midcentral 63/589 (10.7) 1.3(1–1.8) 1.5(1.1–2.1)

Hawke’s Bay 65/578 (11.2) 1.4(1–1.9) 1.8 (1.3–2.4)

Hutt 66/564 (11.7) 1.5 (1.1–2) 1.9 (1.3–2.5)

Nelson Marlborough 68/541 (12.6) 1.6(1.2–2.1) 1.7 (1.2–2.3)

Northland 44/462 (8.3) 1.2 (0.8–1.6) 1.5 (1.1–2.2)

Taranaki 49/397 (12.3) 1.6 (1.1–2.2) 1.8 (1.2–2.5)

Lakes 28/338 (8.3) 1 (0.6–1.5) 1.2 (0.8–1.9)

South Canterbury 30/236 (12.7) 1.6 (1–2.4) 1.7 (1.1–2.6)

Whanganui 13/218 (6) 0.7 (0.4–1.2) 0.9 (0.5–1.5)

Wairarapa 26/123 (21.1) 3 (1.8–4.7) 4 (2.4–6.5)

West Coast 7/142 (4.9) 0.6 (0.2–1.2) 0.7 (0.3–1.5)

Tairawhiti 6/95 (6.3) 0. 7 (0.3–1.6) 1.2 (0.5–2. 7)

1In descending order of type 1 diabetes population size; 2adjusting for all variables in Table 2. other New Zealand Health data.4,31 Combined very poor glycaemic control have previously with the DHB data, these fi ndings have been shown to have potentially very good important implications for diabetes care and outcomes when using CSII.34 However, it also funding. Those of non-European ethnicity remains important to remember that other and with greater degrees of socioeconomic aspects of diabetes care have been shown deprivation have previously been shown to be equally important for successful to have poorer diabetes outcomes.32,33 In outcomes, including education and close Australia and the US the vast majority of CSII diabetes team support.2,35 Unfortunately, is funded via private health insurance and our funding model via PHARMAC is unable these access disparities therefore are likely to allocate equal money between devices/ to be even greater. CSII is an important tool, pharmaceuticals and diabetes support and New Zealand has world-leading public and education. Funders should pay close access, when specifi c criteria are met. Our attention to this issue when funding is allo- criteria currently disadvantage those with cated to new diabetes technologies, ideally if less healthy control and this is likely contrib- access criteria are used to limit access based uting to disparity and unequal of access. on glycaemic control, equal funding would In the UK for instance, having less healthy also be put in place for supporting those glycaemic control is actually a potential who may have barriers to meeting or under- indication for a trial of CSII, and those with standing access criteria.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 85 www.nzma.org.nz/journal ARTICLE

Like all studies, this one has strengths and CSII, and public access criteria are rela- weaknesses. Key strengths are the complete tively broad (based on hypoglycaemia or national data drawn from routinely elevated glycaemic control) it is expected collected linked data sources, as well as the that numbers accessing ongoing private uniformity of access to the national access CSII are negligible. In addition, by early criteria between DHBs. In addition this data 2013 it is anticipated that nearly all eligible spans the period of time between access pre-existing patients using private- or hospi- criteria being monitored by an “insulin tal-funded CSII would have crossed over to pump panel” (2012–2014) and transition fully funded access and thus been included to independent online specialist access in our data. This is likely the explanation for criteria (2014–ongoing), and highlights that the more rapid rise in access seen between speed of uptake does not appear to have 2012 and 2013. Finally, this study addresses been specifi cally altered by this change in uptake of CSII only. Due to weaknesses in administration. Potential weaknesses are current data linking it remains impossible the methods used to determine T1D from to accurately describe glycaemic outcomes these linked data sources. We have used using national data collections. Efforts to an algorithm technique to deal with this. address this defi cit are required to allow for This algorithm ensures likely contami- adequate clinical benchmarking. nation of the sample with type 2 diabetes In conclusion, since approval for publicly is negligible but may lead to small loss of funded insulin pumps from September T1D from the total sample. As an example, 2012 there has been a steadily increasing 100 individuals using CSII were excluded uptake. Overall use still appears lower with our algorithm, a number considered than many similar high-income countries. to be of negligible impact to the conclusions Despite public access criteria, disparities in drawn. In addition, given the use of national uptake appear to exist, and in addition to de-identifi ed routinely collected data, we did traditionally described socio-demographic not have access to the full electronic records barriers to healthcare, DHB of residence is of each participant. This means that we are also an independent predictor of uptake. left with an incomplete understanding of Efforts to reduce these disparities are why sociodemographic variables such as required. In particular, changes to aspects DHB of residence and deprivation adversely of the access criteria that appear to limit impact access to CSII. access based on age, socioeconomic position As this data comes from public pharma- and ethnicity should be considered, and for ceutical dispensing data, we also have no those who do not meet access criteria partic- access to data on privately funded CSII, ularly due to unhealthy glycaemic control, and those who moved between private- or more funding for education and support to hospital-funded access following availability ensure more equal overall outcomes should of access criteria. However, as no insurance also be incorporated. companies in New Zealand reimburse for

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 86 www.nzma.org.nz/journal ARTICLE

Competing interests: Dr Wheeler reports grants from University of Otago during the conduct of the study. Acknowledgements: We thank Analytical Services at the Ministry of Health for providing the data for this study. This study was funded by the Dunedin School of Medicine, University of Otago. Author information: Benjamin J Wheeler, Paediatric Endocrinology, Southern District Health Board, Dunedin; Department of Women’s and Children’s Health, University of Otago, Dunedin; Rhiannon Braund, NZ Pharmacovigilance Centre, Preventive and Social Medicine, University of Otago, Dunedin; Barbara Galland, Department of Women’s and Children’s Health, University of Otago, Dunedin; Anastasia Mikuscheva, Department of Women’s and Children’s Health, University of Otago, Dunedin; Esko Wiltshire, Paediatrics and Child Health, University of Otago, Wellington; Craig Jefferies, Starship Children’s Health, Auckland; Michel de Lange, Biostatistics Unit, Dunedin School of Medicine, University of Otago, Dunedin. Corresponding author: Associate Professor Benjamin J Wheeler, Paediatric Endocrinologist, Department of Women’s and Children’s Health, University of Otago, PO Box 913, Dunedin 9054. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7829

REFERENCES: 1. Diabetes Control and 4. McKergow E, Parkin L, a multiple daily insulin Complications Trial Barson DJ, et al. Demo- injection regimen (basal Research Group. Effect graphic and regional once-daily glargine plus of intensive diabetes disparities in insulin pump mealtime lispro) and treatment on the develop- utilization in a setting continuous subcutaneous ment and progression of of universal funding: a insulin infusion (lispro) long-term complications New Zealand nationwide in type 1 diabetes: a in adolescents with study. Acta Diabetol randomized open parallel insulin-dependent diabetes 2017; 54(1):63–71. multicenter study. Diabetes mellitus: Diabetes Control 5. Misso Marie L, Egberts Care 2009; 32(7):1170–6. and Complications Trial. J Kristine J, Page M, et al. 8. Cooper MN, O’Connell SM, Pediatr 1994; 125:177–88. Continuous subcutaneous Davis EA, et al. A popula- 2. The Diabetes Control insulin infusion (CSII) tion-based study of risk and Complications Trial versus multiple insulin factors for severe hypogly- Research Group. The Effect injections for type 1 caemia in a contemporary of Intensive Treatment of diabetes mellitus. Cochrane cohort of childhood-onset Diabetes on the Develop- Database of Systematic type 1 diabetes. Diabetolo- ment and Progression of Reviews 2010; (1). gia 2013; 56(10):2164–70. Long-Term Complications 6. Pankowska E, Blazik M, 9. Yeh H, Brown TT, Maruthur in Insulin-Dependent Dziechciarz P, et al. Contin- N, et al. Comparative Diabetes Mellitus. New uous subcutaneous insulin effectiveness and safety of England Journal of Medi- infusion vs. multiple daily methods of insulin delivery cine 1993; 329(14):977–86. injections in children and glucose monitoring 3. Phelan H, Clapin H, Bruns with type 1 diabetes: a for diabetes mellitus: A L, et al. The Australasian systematic review and systematic review and Diabetes Data Network: meta-analysis of random- meta-analysis. Annals fi rst national audit of ized control trials. Pediatr of Internal Medicine children and adolescents Diabetes 2009; 10(1):52–8. 2012; 157(5):336–47. with type 1 diabetes. Med J 7. Bolli GB, Kerr D, Thomas 10. de Bock M, Gunn AJ, Holt Aust 2017; 206(3):121–25. R, et al. Comparison of JA, et al. Impact of insulin

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 87 www.nzma.org.nz/journal ARTICLE

pumps on glycaemic in Germany. Diabetes 25. Pickup J. Insulin pumps. control in a pump-naive technology & therapeutics Int J Clin Pract Suppl paediatric regional 2012; 14(12):1105–9. 2010(166):16–9. population. Journal of 18. Dharrshinee S, S ASH, 26. Bonfanti R, Lepore G, paediatrics and child health Martin dB, et al. Insulin Bozzetto L, et al. Survey on 2012; 48(3):247–52. regimens for newly the use of insulin pumps in 11. Miller KM, Foster NC, Beck diagnosed children with Italy: comparison between RW, et al. Current state of type 1 diabetes mellitus in pediatric and adult age type 1 diabetes treatment Australia and New Zealand: groups (IMITA study). Acta in the U.S.: updated data A survey of current Diabetol 2016; 53(3):403–12. from the T1D Exchange practice. Journal of Paedi- 27. Selvakumar D, Al-Sallami clinic registry. Diabetes atrics and Child Health HS, de Bock M, et al. Care 2015; 38(6):971–8. 2017; 53(12):1208–14. Insulin regimens for newly 12. Zuijdwijk CS, Cuerden 19. Statistics New Zealand. diagnosed children with M, Mahmud FH. Social National population type 1 diabetes mellitus determinants of health estimates: at 31 July 2018 in Australia and New on glycemic control in 2018 [Available from: Zealand: A survey of pediatric type 1 diabetes. http://archive.stats.govt. current practice. Journal of J Pediatr 2013; 162:730–5. nz/browse_for_stats/ paediatrics and child health 13. Carter PJ, Cutfi eld WS, population/estimates_ 2017; 53(12):1208–14. Hofman PL, et al. Ethnic- and_projections.aspx 28. Lawton J, Kirkham J, ity and social deprivation accessed 21 August 2018. Rankin D, et al. Who independently infl uence 20. Ministry of Health. gains clinical benefi t metabolic control in National Collections 2018 from using insulin pump children with type 1 [Available from: http:// therapy? A qualitative diabetes. Diabetologia www.health.govt.nz/ study of the perceptions 2008; 51:1835–42. nz-health-statistics/nation- and views of health 14. Thompson RJ, Agostini K, al-collections-and-surveys/ professionals involved in Potts L, et al. Deprivation collections accessed the Relative Effectiveness and ethnicity impact on 21 August 2018. of Pumps over MDI and diabetes control and use of 21. Jo EC, Drury PL. Develop- Structured Education treatment regimen. Diabet ment of a Virtual Diabetes (REPOSE) trial. Diabet Med 2013; 30(4):491–4. Register using Informa- Med 2016; 33(2):243–51. 15. Powell PW, Chen R, tion Technology in New 29. de Bock M, Jones TW, Kumar A, et al. Socio- Zealand. Healthc Inform Fairchild J, et al. Chil- demographic effects on Res 2015; 21(1):49–55. dren and adolescents biological, disease care, 22. Chan WC, Papaconstan- with type 1 diabetes in and diabetes knowledge tinou D, Lee M, et al. Can Australasia: An online factors in youth with type administrative health survey of model of care, 1 diabetes. J Child Health utilisation data provide workforce and outcomes. Care 2013; 17(2):174–85. an accurate diabetes Journal of paediatrics and child health 2018. 16. Shulman R, Stukel TA, prevalence estimate for a Miller FA, et al. Insulin geographical region? Diabe- 30. Hine P, Senniappan S, pump use and discontin- tes Research and Clinical Sankar V, et al. Deprivation uation in children and Practice 2018; 139:59–71. impedes success of insulin teens: a population-based 23. Salmond C, Crampton P, intensifi cation in children cohort study in Ontario, Atkinson J. NZDep2006 and adolescents with Type Canada. Pediatr Diabetes Index of Deprivation. 1 diabetes; longitudinal 2017; 18(1):33–44. Wellington: Universi- linear mixed modelling of a retrospective obser- 17. Icks A, Razum O, Rosenbau- ty of Otago 2007 vational cohort. Diabet er J, et al. Lower frequency 24. R: A language and envi- Med 2011; 28(3):338–44. of insulin pump treatment ronment for statistical in children and adolescents computing [program]. 31. Harris R, Tobias M, of Turkish background Vienna, Austria: R Foun- Jeffreys M, et al. Effects with type 1 diabetes: dation for Statistical of self-reported racial analysis of 21,497 patients Computing, 2017. discrimination and

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 88 www.nzma.org.nz/journal ARTICLE

deprivation on Maori 33. Senniappan S, Hine P, Poor Glycemic Control health and inequalities in Tang W, et al. The effect of During a 12-Month Cohort New Zealand: cross-sec- socioeconomic deprivation Trial. Journal of diabetes tional study. Lancet 2006; on effi cacy of continuous science and technology 367(9527):2005–9. subcutaneous insulin 2018; 12(5):1080–81. 32. Carter PJ, Cutfi eld WS, infusion: a retrospective 35. Skinner TC, Cameron FJ. Hofman PL, et al. Ethnic- paediatric case-controlled Improving glycaemic ity and social deprivation survey. Eur J Pediatr control in children and independently infl uence 2012; 171(1):59– 65. adolescents: which aspects metabolic control in 34. de Bock M, Rossborough of therapy really children with type 1 J, Siafarikas A, et al. matter? Diabet Med diabetes. Diabetologia Insulin Pump Therapy in 2010; 27(4):369–75. 2008; 51(10):1835–42. Adolescents With Very

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 89 www.nzma.org.nz/journal CLINICAL CORRESPONDENCE

Beau lines and telogen ef uvium following aluminium phosphide poisoning Samaneh Nakhaee, Nasim Zamani, Omid Mehrpour

luminium phosphide (AlP) is a type reported in the State of California.2 There of fumigant that is applied to protect are very few publications describing fol- Astored grains from bacterial and low-up for survivors of AlP poisoning, and in fungal contamination. Contact of AIP with addition skin and nail complication induced water releases phosphine gas (PH3), which by AlP are rarely reported. is a very dangerous poison. Even a small This report summarises the clinical case amount of exposure to this chemical causes of a 24-year-old female who attended our signifi cant toxicity. AlP poisoning is an im- emergency department after intentional portant clinical problem in Asian countries ingestion of one, three-gram, undissolved such as Iran and India, and there are several AlP tablet, with four glasses of water. The reports of its poisoning from both develop- diagnosis of AlP poisoning was confi rmed by 1 ing and developed countries. AlP poisoning patient history, collateral history from her is not common in Western countries because closest relatives , and a positive silver nitrate there is a trend of much less use of AlP, but paper test to phosphine. Her initial clinical is reported. For instance, between 2005 and manifestations were: hypotension with 2009, there were 10 cases of AlP poisoning blood pressure of 90/50mmHg tachycardia

Figure 1: Severe diffuse hair loss, posterior view.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 90 www.nzma.org.nz/journal CLINICAL CORRESPONDENCE

Figure 2: Beau lines in fi nger nail.

with a pulse rate of 120bpm; tachypnea, months later, she referred to us again with respiratory rate of 20 per minute; and hair loss and nail deformity (Figures 1 and reduced oxygen saturation of 96% with 2). She had not consumed any medication, hypoxemia, PaO2 59mmHg, associated with and her hormonal profi le, thyroid tests, a pH of 7.39, and bicarbonate of HCO3 of serum iron, ferritin and total iron binding 12.9mmol/L. The patient was admitted to the capacity were normal. By history there was intensive care unit and received intravenous no evidence of anorexia, crash dieting or crystalloid fl uid, vasopressors (dopamine low protein intake. A smear and biopsy of and norepinephrine), digoxin, glucagon, the scalp were negative for alopecia areata, magnesium sulfate, sodium bicarbonate, scalp infections, and other skin disorders, vitamin C, vitamin E, heparin and N-acetyl- including lichen planus. Trichotillomania cysteine. Twenty-four hours after admission, was very unlikely because of the diffuse she became shocked, with more profound hair loss. The patient was followed for fi ve hypotenion; blood pressure of 75/50mmHg, months by the end of which her nail changes with severe metabolic acidosis with bicar- and hair loss recovered completely without bonate of 7mmol/L; and hypoxemia with any treatment. oxygen saturation of 70%, and associated We believe this represents telogen respiratory distress. She was then intu- effl uvium (TE), which is non-scarring acute bated and put on mechanical ventilation. onset alopecia of diffuse hair loss caused by Thirty-six hours after admission, echocardi- metabolic, hormonal or physiologic stress. It ography showed a left ventricular ejection occurs when a signifi cant number of anagen fraction of less than 20%, global hypoki- hairs are triggered to stop growing prema- nesia of the left ventricle, and mitral valve turely, by any stimulus, and subsequently regurgitation. An intra-aortic balloon pump enter catagen phase, followed by the telogen (IABP) was inserted. Eight days later her phase with spontaneous recovery in six blood pressure was 110/80 and IABP was months .3 Beau lines are transverse depres- removed. Echocardiography on day 20 of sions in the nail plate due to a slowing/ admission showed an increase of LVEF to disruption in cell growth from the more 50%. Twenty-seven days after admission, proximal matrix which results in a thinner she was discharged in good condition. Four

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 91 www.nzma.org.nz/journal CLINICAL CORRESPONDENCE

nail plate.4 It can be caused by trauma, a mercury, thallium and arsenic poisoning local disease involving the nail fold, meta- can cause TE and Beau lines .5 These diag- bolic, infl ammatory or traumatic causes; noses should also be considered in any and in this case shock and cardiac failure patient who has passed the acute phase of related to AlP poisoning. Various drugs and the severe systemic disease, possibly related poisonings can directly induce hair and to toxins or poisoning, and subsequently nails disorders. It has been proposed that presenting with hair loss and nail deformity.

Competing interests: Nil. Author information: Samaneh Nakhaee, Medical Toxicology and Drug Abuse Research Center, Birjand University of Medical Sciences (BUMS), Iran; Nasim Zamani, Department of Clinical Toxicology, Loghman Hakim Hospital, Shahid Beheshti University of Medical Sciences, Tehran, Iran; Omid Mehrpour, Rocky Mountain Poison and Drug Center, Denver, CO, USA. Corresponding author: Dr Omid Mehrpour, Rocky Mountain Poison and Drug Center, Denver, CO, USA. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7830

REFERENCES: 1. Mehrpour O, Jafarz adeh M, 3. Malkud S. Telogen emy of Dermatology. Abdollahi M. A systematic effl uvium: a review. 2015; 73:849–55. review of aluminium phos- Journal of clinical and 5. Patel S, Tosti A. An phide poisoning. Archives diagnostic research: overview of management of Industrial Hygiene and JCDR. 2015; 9:WE01. of drug-induced hair and Toxicology. 2012; 63:61–73. 4. Braswell MA, Daniel CR, nail disorders. Clinical 2. Rubin A. Phosphine: Risk Brodell RT. Beau lines, Practice. 2014; 11:327. characterization docu- onychomadesis, and ment. Available at: https:// retronychia: A unifying www.cdpr.ca.gov/docs/ hypothesis. Journal of risk/rcd/phosphine.pdf the American Acad-

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 92 www.nzma.org.nz/journal CLINICAL CORRESPONDENCE

Successful intravascular lithotripsy for severely calci ed le anterior descending coronary artery stenosis Aleksandra Pineda, Aniket Puri, Bijan Jahangiri

ABSTRACT Percutaneous coronary intervention (PCI) of heavily calcified stenoses remains a significant challenge to interventional cardiologists. Over the last years, high-pressure balloons, cutting balloons as well as atherectomy devices have been used to tackle such lesions. Lithoplasty is a method of lesion modification using intravascular lithotripsy (IVL, shockwave) to treat particularly calcified coronary lesions. A 55-year-old male without previous cardiac history reported chest pain and was found to have ECG changes anterolaterally and significantly elevated troponins. Coronary angiogram showed heavily calcified severe proximal le anterior descending (LAD) stenosis. The attempts to predilate the lesion with semi-compliant and non-compliant balloons at high pressures were unsuccessful so the decision was made to use a shockwave balloon. Subsequently a drug-eluting stent was successfully implanted and post-dilated at high pressure. Final angiography showed an excellent result, also confirmed in optical coherence tomography (OCT).

55-year-old male reported chest pain a 3.5/28mm drug-eluting stent was implant- and was found to have ECG changes ed and post-dilated at high pressure. Final A anterolaterally. No previous cardi- angiography showed an excellent result. ac history was reported. His troponin was OCT showed an optimal stent expansion and signifi cantly elevated. Coronary angiogram apposition (Figure 1C). showed heavily calcifi ed severe (95%) prox- imal LAD stenosis with associated coronary Discussion aneurysm (Figure 1A). Multiple attempts Percutaneous coronary intervention to predilate the proximal LAD lesion with (PCI) of heavily calcifi ed, resistant stenoses semi-compliant and non-compliant balloons remains a challenge to interventional cardi- at high pressures were unsuccessful. The ologists despite developments in procedural OCT confi rmed extensive circular calcifi ca- techniques and technology. Inadequate tion in proximal LAD (Figure 1A). The distal lesion preparation may lead to stent under- RVD (reference vessel diameter) was 3.5mm deployment, which consequently predisposes and minimum lumen diameter approxi- to in-stent restenosis and stent thrombosis.1 mately 2mm, making the vessel not suitable Over the years, high-pressure balloons, for rotational atherectomy, so coronary lith- cutting balloons and atherectomy devices otripsy angioplasty was performed. We used have been used to tackle such lesions.1 3.0/12mm Shockwave balloon (Shockwave Medical, Fremont, California) infl ated to 4 Treatment of heavily calcifi ed lesions is atm and performed fi ve cycles of lithotrip- associated with higher procedural risk and sy (10 pulses each). OCT confi rmed cracks overall increased risk of adverse cardiovas- of calcifi cation and signifi cant increase in cular events at two-year follow-up related lumen diameter (Figure 1B). Subsequently mainly to higher TVR (target vessel revascu- larisation) rate.2

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 93 www.nzma.org.nz/journal CLINICAL CORRESPONDENCE

Figure 1: Angiographic view and OCT (cross section, longitudinal view and 3D view of the vessel) showing a heavily calcifi ed proximal LAD stenosis before treatment (A), the result after lithoplasty with calcium cracks (B) and after stent implantation (C).

Lithoplasty is a novel approach of lesion has been also confi rmed in OCT showing a modifi cation using intravascular lithotripsy signifi cant increase in acute area gain and (IVL, shockwave) in a balloon developed to favourable stent expansion.4 Together with treat calcifi ed coronary lesions. The tech- other advantages such as the ease of use nology utilises high-speed sonic pressure and less procedural complications, intravas- waves which selectively disrupts superfi cial cular lithotripsy has become an attractive and deep calcium within the vessel and alternative for treatment of heavily has minimum impact on healthy tissue.3 calcifi ed complex lesions, especially in the The data from the Disrupt CAD study is presence of increased calcifi ed burden in very promising showing that the lithop- the ageing population. The intravascular lasty balloon-based therapy resulted in 98% lithotripsy is now also available in Christ- device success and facilitated 100% stent church Hospital, which will allow for this delivery.3 It also demonstrated a low major treatment method to be used routinely in adverse cardiac event rate with minimal patients with severe calcifi ed coronary vascular complication.3 The effi cacy of IVL artery disease from our region.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 94 www.nzma.org.nz/journal CLINICAL CORRESPONDENCE

Competing interests: Nil. Acknowledgements: We thank the patient for allowing us to share his details. Author information: Aleksandra Pineda, Interventional Fellow, Christchurch Hospital, Christchurch; Aniket Puri, Consultant Cardiologist, Christchurch Hospital, Christchurch; Bijan Jahangiri, Senior Registrar, Christchurch Hospital, Christchurch. Corresponding author: Aleksandra Pineda, Christchurch Hospital, 2 Riccarton Ave, Christchurch 8011. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7831

REFERENCES: 1. Farag M, Costopoulos C, lesions with drug-eluting 69, Issue 11 Supple- Gorog DA, et al. Treatment stents. Int J Cardiol. 2017 ment, March 2017. of calcifi ed coronary Mar 15; 231:61–67. 4. Ali Z, Brinton TJ, artery lesions. Expert 3. Brinton TJ, Ali Z, Di Mario Hill JM, et al. Optical Rev Cardiovasc Ther. C, et al. Performance of Coherence Tomography 2016 Jun; 14(6):683–90. the lithoplasty system in Characterization of Coro- 2. Généreux P, Redfors B, treating calcifi ed coronary nary Lithoplasty for Witzenbichler B, et al. lesions prior to stenting: Treatment of Calcifi ed Two-year outcomes after results from the Disrupt Lesions: First Description. percutaneous coronary CAD OCT sub-study. J JACC Cardiovasc Imaging. intervention of calcifi ed Am Coll Cardiol. Volume 2017 Aug; 10(8):897–906.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 95 www.nzma.org.nz/journal RESEARCH LETTER

Lifespan of New Zealand Second World War veterans from one large cemetery: the case for a national-level study Nick Wilson, Glyn Harper

or veterans of the Second World War with the online Cenotaph database held (WW2), most studies suggest long- by Auckland Museum.8 This allowed us to Fterm adverse impacts on their health, exclude veterans who were also in other especially if they had been prisoners of war wars and who were in the military forces of (POWs). But such studies do not always fi nd other countries. that these veterans had an increase in all- To further minimise the chance of the cause mortality, possibly due to the ‘healthy veteran also being a WW1 veteran, we soldier effect’ (references available on also only included the age cohorts born request). For New Zealand, being a veteran between 1905 and 1928 (ie, they would at of the First World War (WW1) has been the most only be aged 13 years in the last 1 associated with a reduced lifespan, but such year of WW1 [1918] and those born in 1928 a study has never been performed for WW2 would only turn 17 in the last year of WW2 veterans. This is despite adverse morbidity [1945], a year below the legal admission impacts being described in this group from age). The Cenotaph data also supplied status 2 survey data, psychiatric studies (eg, by Ma- on having been a POW or not, and if the 3 4–6 cleod ) and interview data. Indeed, in 1946 veteran had served in the Māori Battalion. there were 22,846 WW2 personnel receiving We created a synthetic cohort matched war pensions for service-related disabilities, to each real veteran in the dataset with an and in the 1945–1946 year the New Zea- assigned lifespan value based on that of land Government spent £58,302 on medical the average New Zealand man who was treatment of war pensioners.7 The detailed born in the same year and for the age they study by Parr4 reported that many veter- were in 1946 (the year after the one when ans “came home with invisible wounds”. It the war ended). These values have all been recorded that in 1985 more than 10,000 of estimated for fi ve-year intervals by a large these veterans were offi cially recognised as Statistics NZ (SNZ) study9 and we interpo- suffering from “nervous system disabilities” lated the values for birth years in between (Parr, p14), with cases of post-traumatic the fi ve-year values provided by SNZ. stress disorder and alcohol abuse. So to commence some initial exploration of the lifespan issue, we examined an administra- Results tive dataset which already had the age of As detailed in Table 1, this cohort of deaths for WW2 veterans buried at a large veterans had a signifi cantly lower lifespan New Zealand cemetery. relative to the SNZ estimates for the matched synthetic cohort of New Zealand males born Methods in the same year and being alive in 1946 (ie, a 4.9-year defi cit; 68.5 vs 73.4 years). Within We obtained an administrative dataset the veteran cohort there was also a signifi - (in Excel) of the names, date of burial and cantly lower lifespan for those in the Māori age of death of war veterans buried in the Battalion and for those who were not buried veterans section of a large urban cemetery next to their spouse/partner. But there was (Taita Lawn Cemetery, Naenae, Lower Hutt no signifi cant difference by former POW City). We then cross-checked the names (and status (albeit based on small numbers). military service numbers where available)

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 96 www.nzma.org.nz/journal RESEARCH LETTER

Table 1: Lifespan results for the veteran cohort for male New Zealand military personnel involved in WW2 from one large New Zealand cemetery.

Specific population Mean age (years) Median age P-value Comment at death (SD) (years) at death Full veteran cohort (n=702) (New Zealand mil- 68.5 68 <0.0001 (Kruskal- For birth years 1905 to 1928, itary who participated in WW2 but no other (11.0) Wallis test for median birth year = 1917 (ie, wars, though we included those who were two groups) age 22 years at the start of also in J Force, the New Zealand military WW2 in 1939) force occupying Japan in 1946 to 1948).

Synthetic comparison cohort with nationally 73.4 73.5 representative lifespans calculated from the (SD not meaningful) SNZ lifetables (matched to the birth cohort of the 702 veterans, for life expectancy in 1946).

Prisoner of war (n=53) 68.8 68 0.869 Many of these men were tak- (10.1) (ANOVA) en prisoner in Crete in 1941

Not a prisoner of war (n=649) 68.5 68 (11.1)

Participation in the Māori Battalion (ie, likely 60.6 57 0.0034 (ANOVA) of Māori ethnicity) (n=16) (9.3)

Not in the Māori Battalion (n=686) (nearly all 68.7 68 European, but includes some Māori in other (11.0) military units)

Buried next to spouse/partner (n=287)* 72.3 73 <0.00001 (ANO- (10.0) VA)

Not buried next to spouse/partner (n=415) 65.9 65 (11.0)

*The dataset included details about if a female spouse/partner was buried in linked plots labelled “1” and “2” beside each other (for simplicity we excluded those cases of a man buried next to a man in linked plots, though in some cases the relationship was detailed, eg, being brothers).

movement of men to the Hutt Valley after Discussion WW2 for work may also mean that those The apparent lifespan defi cit for these buried in this cemetery refl ect a cohort born WW2 veterans (relative to estimates of in quite different rural and urban settings lifespan from national lifetables) is plau- from around New Zealand. sible, given the international and New Given these limitations, these results Zealand literature concerning the long-term should be seen as very tentative and more health impacts of being a WW2 veteran. as a demonstration of what could be studied Although a small percentage of WW2 if a more appropriate sample was used, eg, veterans are still currently alive, their a random selection of all the New Zealand deaths in the future are unlikely to make military personnel. Such work would need much impact on such results (eg, there could to be well resourced, as extensive archival be men who joined in the last year of the and genealogical database work is required war [1945] at age 18 years [born in 1927] to establish the dates of birth and death of who were aged 92 years in 2019). each veteran. Even more work would be Instead, the much larger methods required to collect data on occupational limitation with this analysis is that this class, military rank, if wounded, and marital is just one cemetery which may not be status (the latter given the interesting results nationally representative. Even though the above that having a spouse may be asso- deprivation maps of the Lower Hutt City and ciated with a lifespan benefi t). Such further Upper Hutt City10 areas suggest a broad mix analysis with a nationally representative of socioeconomic groups, it could still be that sample seems important if New Zealand the veteran population tended to reside in society is to obtain a better understanding of the more deprived areas in these cities. The the full impact of this major war.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 97 www.nzma.org.nz/journal RESEARCH LETTER

Competing interests: Nil. Acknowledgements: Chris Gousmett, Corporate Information Manager, Hutt City Council, who kindly provided the cemetery dataset on veterans. Author information: Nick Wilson, Department of Public Health, University of Otago, Wellington; Glyn Harper, Professor of War Studies, Massey University, Palmerston North. Corresponding author: Professor Nick Wilson, Department of Public Health, University of Otago, Mein St, Newtown, Wellington. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7832

REFERENCES: 1. Wilson N, Clement C, 5. Hopner V. Home from war 9. Statistics New Zealand. A Summers JA, Bannister [PhD thesis]. Auckland: History of Survival in New J, Harper G. Mortality of Massey University, 2014. Zealand: Cohort life tables fi rst world war military 6. Smith BMS. Ordinary men: 1876–2004. Wellington: personnel: comparison extraordinary times [PhD Statistics New Zealand, of two military cohorts. thesis]. Auckland: Univer- (with online tables at: BMJ 2014; 349:g7168. sity of Auckland, 2011. http://www.stats.govt.nz/ browse_for_stats/health/ 2. Salmond GL, Salmond 7. Statistics New Zealand. The life_expectancy/cohort- CE. The Health of Former New Zealand Offi cial Year- life-tables.aspx), 2006. Servicemen. New Zealand, Book, 1946. http://www3. War Pensions Medical stats.govt.nz/New_Zealand_ 10. White P, Gunston J, Research Trust Board, 1977. Offi cial_Yearbooks/1946/ Salmond C, Atkinson J, 3. Macleod AD. Psychiatric NZOYB_1946.html?_ga=2. Crampton P. Atlas of socio- casualties of World War II. 1094705.1255873467.1547 economic deprivation in N Z Med J 2000; 113:248–50. 827122-1031987571.15399 New Zealand NZDep2006. Wellington: Ministry of 4. Parr A. Silent Casualties: 70965#idsect1_1_310990 Health, 2008. [Map for Hutt New Zealand’s Unspoken 8. Auckland War Memo- Valley DHB: http://www. Legacy of the Second World rial Museum. Cenotaph health.govt.nz/system/ War. Auckland: Tandem database. http://muse. fi les/documents/publica- Press, 1995. aucklandmuseum.com/ tions/huttvalley.pdf]. databases/Cenotaph/ locations.aspx

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 98 www.nzma.org.nz/journal LETTER

Comment on: “Quality of electronic records documenting adverse drug reactions…” John S Fountain, Susan H Kenyon

dverse drug reactions (ADRs) are an being recorded to ensure there is appropriate important contributor to mortality, robustness and maximal clinical utility of Amorbidity and healthcare costs. A information through sharing of information”. 1998 meta-analysis in the US concluded The necessity to collect robust and that ADRs were the fourth to sixth leading sharable ADR data has been recognised cause of death in hospitalised patients, with and a solution is being developed through 1 serious ADRs occurring in 6.7% of patients. the auspices of the Health Information Since that report, ADR-related deaths in the Standards Organisation (HISO) with input US have increased from 0.08 to 0.12/100,000 from Medsafe and clinical experts. HISO, population over the eight-year period 1999 a committee supported by the Ministry to 2006, with death signifi cantly more likely of Health’s Data and Digital directorate, in those aged over 55 years and greatest in promotes the adoption of fi t-for-purpose 2 those aged 75 years or more. Importantly health information standards for the New it is recognised that signifi cant numbers of Zealand health system. An Adverse Reaction ADRs are preventable—72% in one review Reporting Standard (ARRS) is currently 3 of occurrence in developed nations. While under development for evaluation, and is defi nitive New Zealand research in this expected to be released during 2019. This area is lacking we are not immune to this standard is based on the SNOMED CT termi- phenomenon, particularly given an ageing nology to support effi cient entry, retrieval 4 population. and reuse of clinical information. Once A primary step for recognising and approved, ARRS can underpin a nationally therefore preventing ADRs is the collection integrated ADR reporting system to enable of accurate and shareable data describing both individual patient alerting and national these events. However, the recent article by pharmacovigilance. Braund et al has highlighted shortcomings This will be an important step in in this area due to data inaccuracies, lack of improving patient safety and wellbeing. precision and diffi culties in accessing event But for these benefi ts to be realised there 5 information. The authors conclude that: must be uptake from vendors to incorporate “…information transfer between elec- this initiative into existing and future IT tronic systems needs high-quality data to products, and the cooperation of clinicians be entered at the time of a reaction initially to effectively utilise this solution.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 99 www.nzma.org.nz/journal LETTER

Competing interests: Nil. Author information: John S Fountain, Clinical Advisor, Best Practice Advocacy Centre (BPACnz), Dunedin, Clinical Senior Lecturer, Department of General Practice and Rural Health, University of Otago, Dunedin; Susan H Kenyon, Manager, Clinical Risk Management, Medsafe, Ministry of Health, Wellington. Corresponding author: John S Fountain, Best Practice Advocacy Centre (BPACnz), Level 8, 10 George Street, PO Box 6032, Dunedin 9056. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7833

REFERENCES: 1. Lazarou J, Pomeranz BH, 3. Braund R, Lawerence CK, Zealand hospitals. N Z Med Corey PN. Incidence of Baum L, et al. Quality of J. 2017; 130(1460):21–32. adverse drug reactions electronic records docu- 5. Angamo TM, Chalmers L, in hospitalized patients. menting adverse drug Curtain CM, Bereznicki JAMA. 1998; 279(15):1200–5. reactions within a hospital LRE. Adverse-drug-reac- 2. Shepherd G, Mohorn setting: identifi cation of tion-related hospitalisations P, Yacoub K, May DW. discrepancies and informa- in developed and devel- Adverse drug reaction tion completeness. N Z Med oping countries: a review deaths reported in United J. 2019; 132(1488):28–37. of prevalence and States vital statistics. 4. Robb G, Loe E, Maharaj A, contributing factors. Drug Ann Pharmacother. et al. Medication-related Saf. 2016; 39(9):847–57. 2012; 46(2):169–175. patient harm in New

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 100 www.nzma.org.nz/journal LETTER

Switching from gabapentin to pregabalin Pauline McQuoid

abapentin and pregabalin are fully the same conversion ratio across gabapen- subsidised on prescription in New tin’s dose range.8,9 This risks giving too GZealand. Originally developed as anti- much pregabalin when converting from the convulsants, they are more widely used for higher end of gabapentin’s dose range. The neuropathic pain. The proposed mechanism conversion ratio used in one study was “of for analgesic activity is binding to the α2δ the author’s creation”3 and in another study subunit of voltage-gated calcium channels in the authors assumed that pregabalin had six the central nervous system, which reduces times greater pharmacological effect against the release of excitatory neurotransmitters neuropathic pain than gabapentin based on such as glutamate.1 Numbers needed to the maximum dose of each medicine.8 Bock- treat for 50% reduction in neuropathic pain brader et al5 derived a potency ratio from are reported as 6.3 for gabapentin (5·0–8·3) EC50 data in post-herpetic neuralgia, which and 7.7 for pregabalin (6.5–9.4), and are refl ects the pharmacological activity of the dose-related.2 Switching from gabapentin to two medicines. pregabalin may be considered for effi cacy Actual and modelled bioavailability data6,7 or tolerability reasons. Although there is were used to calculate the approximate no clear evidence that either gabapentin or amount of gabapentin absorbed, then Bock- pregabalin is more effective than the other brader’s potency ratio5 and a correction 1 for neuropathic pain, patients may bene- factor for pregabalin’s bioavailability (90%)1 fi t from switching. In an open-label study, were applied to develop a conversion analgesia improved after switching from algorithm: gabapentin to pregabalin.3 Table 1: Proposed switching doses. There is no established guidance on converting between gabapentin and Gabapentin Suggested dose of pregabalin pregabalin.4 The manufacturers of both dose pregabalin and gabapentin advise that 100mg tds 50mg bd if they are to be stopped or changed to another medication, the dose should be 200mg tds 75mg bd tapered gradually over at least one week.4 300mg tds 100mg bd This gradual withdrawal is to minimise the risk of seizures where they are being 400mg tds 125mg bd used for patients with seizure disorders.4 500mg tds 150mg bd The importance of a slow withdrawal in patients with neuropathic pain remains 600mg tds 150mg m and 175mg n unknown,4 however discontinuation 700mg tds 175mg bd OR 175mg m and symptoms have been reported with abrupt 200mg n cessation of both gabapentin and prega- balin. Discontinuation symptoms reported 800mg tds 200mg bd include insomnia, nausea, anxiety, pain, 900mg tds 200mg m and 225mg n and sweating.4 Determining dose equiva- lence between gabapentin and pregabalin 1,000mg tds 225mg bd is complicated by gabapentin’s nonlinear bioavailability, in contrast to pregaba- 1,100mg tds 225mg m and 250mg n lin’s linear bioavailability.6 Gabapentin’s 1,200mg tds 250mg bd bioavailability ranges from 80% with 100mg tds6 to 35% with 1,200mg tds.7 Despite this, 1,600mg tds 300mg bd most published guidance on switching uses Pregabalin is available as capsules of 25mg, 75mg and 150mg in New Zealand.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 101 www.nzma.org.nz/journal LETTER

If analgesic effect is suboptimal 1–2 weeks analgesic effect during the tapering after completing the switch, pregabalin can down and titrating up phases. be titrated up at seven-day intervals to a The adverse consequences of giving 10 maximum daily dose of 300mg bd. too much pregabalin when switching As expected, pregabalin doses at the from gabapentin are an increase in the higher end of the dose range are more intensity of side effects. Many patients are conservative than other published reluctant to continue a medicine if they algorithms.3,9 have unpleasant side effects when starting Options for switching treatment. Treatment choices in neuropathic pain are limited, so losing an option due There are three approaches described in to potentially avoidable side effects can be the literature: frustrating. Ifuku et al8 reported a signifi cant 1. Stop/start: Take the last dose of increase in the number of patients with gabapentin at night and start peripheral oedema after switching from the target dose of pregabalin the gabapentin to pregabalin. following day. Two papers utilising The starting dose of pregabalin 75mg this approach reported that it was bd recommended by the manufacturer effective and well-tolerated.3,8 seems to have been too high for some of 2. Cross-taper: A pharmacokinetic simu- our patients started on pregabalin de novo lation model compared a stop/start (not switching) and they have described approach with a four-day cross-taper unpleasant cognitive side effects including whereby 50% of the gabapentin dose dysphoria, feeling “spaced out” or “out of and 50% of the target pregabalin dose it”, excessive sedation and dizziness. We are is given for four days, followed by now recommending a lower starting dose discontinuation of gabapentin and of 75mg at night, or 25–50mg in people who 5 use of target dose of pregabalin. Both may be susceptible to adverse effects, eg, approaches were pharmacokinetically adults over 75 years old.11 This is consistent comparable. with recommendations for both gabapentin 3. Taper down and stop gabapentin then and pregabalin to consider lower starting gradually titrate pregabalin up: This doses and/or slower titration in patients is the approach recommended by the who may be more susceptible to adverse manufacturers. A gradual reduction effects (such as frail elderly).10 The dose may be more likely to avoid with- of pregabalin should be reduced in renal drawal symptoms. The main problem impairment (eGFR ≤60ml/min).10 with this approach is possible loss of

Competing interests: Nil. Author information: Pauline McQuoid, Clinical Pharmacist, Medwise, Tauranga. Corresponding author: Pauline McQuoid, Clinical Pharmacist, Medwise, PO Box 6164, Tauranga 3146. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7834

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 102 www.nzma.org.nz/journal LETTER

REFERENCES: 1. Ben-Menachem E. uploads/2014/09/What-con- gabapentin with pregabalin Pregabalin pharmacol- version-ratio-is-recom- in postherpetic neuralgia ogy and its relevance mended-between-prega- therapy. Pain Medicine. to clinical practice. balin-and-gabapentin. 2011; 12:1112–1116. Epilepsia. 2004; 45:13–8. pdf Accessed 5/7/18. 9. Schug S, Zekry O. Conver- 2. Finnerup NB, Attal N, 5. Bockbrader HN, Budhwani sion of gabapentin to Haroutounian S, et MN, Wesche DL. Gabapen- pregabalin: simple and al. Pharmacotherapy tin to pregabalin therapy easy! The Australian for neuropathic pain transition: a pharmaco- Pain Society Newsletter in adults: systematic kinetic simulation. Am J Vol 33, Issue 5 July 2013. review, meta-analysis and Ther. 2013 Jan; 20(1):32–6. Available from http:// updated NeuPSIG recom- 6. Bockbrader HN, www.apsoc.org.au/PDF/ mendations. The Wesche D, Miller R, et Newsletters/2013_eNews/ Lancet Neurology. al. A comparison of the APS_Newsletter_JUL13. 2015; 14(2):162–173. pharmacokinetics and pdf Accessed 5/7/18. 3. Toth C. Substitution of pharmacodynamics of 10. bpacnz. Prescribing gabapentin therapy with pregabalin and gabapen- gabapentin and pregabalin: pregabalin therapy in tin. Clin Pharmacokinet. upcoming subsidy changes. neuropathic pain due 2010; 49(10):661–669. bpacnz, March 2018. Avail- to peripheral neurop- 7. Gidal BE, DeCerce J, able from: http://bpac.org. athy. Pain Medicine. Bockbrader HN, et al. nz/2018/gabapentinoids. 2010; 11:456–465. Gabapentin bioavailabil- aspx Accessed 27/1/19. 4. Palliative Meds Info: ity: effect of dose and 11. Haslam C, Nurmikko T. “What conversion frequency of adminis- Pharmacological treatment ratio is recommended tration in adult patients of neuropathic pain in between pregabalin and with epilepsy. Epilepsy older persons. Clin Interv gabapentin?” February Res. 1998 Jul; 31(2):91–9. Aging. 2008; 3(1):111–20. 2017. Available from 8. Ifuku M, Iseki M, Hidaka http://olh.ie/wp-content/ I, et al. Replacement of

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 103 www.nzma.org.nz/journal LETTER

Advancing transgender healthcare teaching in Aotearoa/New Zealand Althea Gamble Blakey, Gareth J Treharne

his letter provides a commentary that education about transgender healthcare and expands on the recent work from c) failure to set aside transphobic personal Oliphant et al1 who outline the new values.2 Our own preliminary research T 3 healthcare guidelines and pathways in confi rms these issues. transgender healthcare in Aotearoa/New We also found a notable additional issue Zealand. Here, we identify challenges in in- around what it is to ‘be’ transgender: staff stigating more education about such health- who teach trainee healthcare practitioners care in the training of doctors and other understand the topic of ‘being’ transgender healthcare practitioners. as eminently caught up with issues of Increasing recognition of transgender ‘self’ and ‘selfhood’ and that as such, these and non-binary gender identities, and issues can be inordinately sensitive. Thus, associated health disparities, indicate an we understand that the corresponding unaddressed and growing need for teaching level of skill, and values, required to effec- about transgender healthcare. Globally, and tively teach transgender healthcare might in Aotearoa/New Zealand there is a corre- be beyond levels currently understood sponding need for high-quality research on to be effective; ‘messing up’ transgender how to best deliver this education. Exam- education can have similar dire and ining the international literature and our personal consequences to ‘messing up’ a own research with transgender community consultation in which a transgender person members, we fi nd insight into why this comes out. educational ‘gap’ may have arisen and how Challenge 2 it might best be addressed. Evidence from the healthcare sector In some ways, it seems easy to under- indicates that challenges in teaching trans- stand how such an educational gap might gender healthcare might also be magnifi ed have come about. Those who are trans- by a general backdrop of persistent discrim- gender or of non-binary gender identity are ination of the transgender/non-binary often treated ‘differently’ to those who are populous, 2 specifi cally against those who cisgender, whose gender identity aligns with undertake healthcare training. A survey their sex assigned at birth.1 Thus, teaching administered by the New Zealand Medical about transgender healthcare issues can also Students Association4 reveals a marked stig- be understood to be a practice that might matisation of students who are transgender be ‘different’ to customary teaching about or of non-binary gender identity. This issue healthcare, and perhaps ‘too diffi cult’ for is signifi cant because it can perpetuate some teaching staff—namely those who lack pre-existing gaps in provision of healthcare experience with transgender people or foun- and a general tolerance of discrimination, dational expertise on transgender healthcare. scaffolded by power hierarchies across Challenge 1 educational and healthcare organisations. Past research has established three main This discrimination of trainee healthcare barriers to effective transgender healthcare practitioners is likely to have a powerful teaching: a) healthcare practitioners’ lack negative effect on both the provision of of understanding about transgender iden- a service and the quality of graduating 5 tities, b) gaps in healthcare practitioners’ practitioners.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 104 www.nzma.org.nz/journal LETTER

Challenge 3 a shortfall apparently still unaddressed. Our recent research suggests a further Similarly, medical course convenors in personal challenge in engaging healthcare Aotearoa/New Zealand note limited time teachers in teaching about transgender to cover LGBT content due to “clashes healthcare. The topic of transgender with ever growing range of other essential 11 healthcare can lead a teacher to expe- content.” Convenors also note that the rience intense feelings of inadequacy and growing requirement to generally cultivate ‘lacking’ the required expertise, both about trainee healthcare practitioners’ refl ective the healthcare needs of transgender people thinking will also compete with any ‘diffi cult’ but also about their own pedagogical skills. issue requiring concentrated resources, eg, This can be despite extensive general physical space/small group work. experience in educating trainee healthcare Hope practitioners.3 Those who teach healthcare Oliphant et al’s work offers some practitioners must themselves be sensitively important progress in in Aotearoa/New helped to address several important issues Zealand. Their recently released guidelines around teaching and learning. and care pathways for gender affi rming and Challenge 4 gender diverse healthcare1 set standards Considering the broader climate in which for all healthcare practitioners to provide, transgender education takes place, we also uphold, develop and incorporate into all fi nd negative messages about ‘being trans- future relevant contexts. gender’, even when diagnostic categories Other recent research in Aotearoa/New are revised in attempts to improve the Zealand offers evidence about a further care of transgender people. Changes to the important starting point for our quest to World Health Organization (WHO) Inter- develop transgender healthcare education national Classifi cation of Diseases (ICD) research. Recent empirical research reveals (mid-2018, into effect 2022) see ‘gender the challenging nature of the central need identity disorder’ reclassifi ed as ‘gender to cultivate teachers’ and trainee healthcare incongruence’.6 This reclassifi cation moves practitioners’ values, as part of challenges away from the category of ‘disorder of described above.12 Values development adult personality’ (mental, behavioural can be an intricate and potentially chal- and neurodevelopmental disorders) to lenging process with a high likelihood of one relating to sexual health7 noted by the failure.12 Cultivating the values of either WHO’s department of reproductive health teacher or student demands skill and time, and research coordinator as a move aimed and discourse about personal values can to reduce discrimination—“better social easily cause offence and fail to ensure acceptance for these individuals”.7 However, ethical duties are met. In other words, a the revised classifi cation can still be inter- signifi cant portion of learning about trans- preted as pathological, which reinforces gender healthcare issues may not simply the idea that being cisgender is the ‘norm’ result from instigating relevant content or and perpetuates the ‘legitimate’ questioning pedagogic process but necessarily require a of transgender people’s identities.8 As one sensitive and specifi c pedagogic discourse transgender person put it: around values.12 “I wish that people, especially doctors, Concerns and starting points acknowl- would understand and accept me.”9 edged here offer a chance to bring Aotearoa/ Challenge 5 New Zealand to the international fore in trainee healthcare practitioner education Further challenges lie in the signif- and research on transgender healthcare, icant resource issues faced by the trainee and in doing so foster future generations of healthcare education system. One in-depth healthcare practitioners to competently and examination of US medical education confi dently serve all community members reveals that “the reported time dedicated well. Developing teacher skills and values to to LGBT-related topics... was small... the negotiate a system of persistent disadvantage quantity, content covered and perceived experienced by transgender people needs to quality of instruction varied substantially”,10 be a dedicated focus of future research.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 105 www.nzma.org.nz/journal LETTER

Competing interests: Nil. Author information: Althea Gamble Blakey, Research Fellow, Otago Medical School & Professional Practice Fellow, Early Learning in Medicine Programme, University of Otago, Dunedin; Gareth J Treharne, Associate Professor, Department of Psychology, University of Otago, Dunedin. Corresponding author: Althea Gamble Blakey, Research Fellow, Otago Medical School & Professional Practice Fellow, Early Learning in Medicine Programme, University of Otago, Dunedin. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7835

REFERENCES: 1. Oliphant J, Veale J, 5. Cruess S, Cruess R. Chapter 9. Human Rights Commission. Macdonald J, Carroll 3: Transforming a medical To be who I am: Report of R, et al. Guidelines curriculum to support the inquiry into discrim- for gender affi rming professional identity ination experienced by healthcare for gender formation. In: Byyny R, transgender people; 2007 diverse and transgender Paauw D, Papadakis M, [cited 2018 Dec 5]. Available children, young people Pfeil S, [Eds]. Medical from: http://www.hrc.co.nz/ and adults in Aotearoa, professionalism best fi les/5714/2378/7661/15- New Zealand. NZ Med J. practices: Professionalism Jan-2008_14-56-48_HRC_ 2018; 131(1487):86–96. in the modern era. Chicago: Transgender_FINAL.pdf 2. Winter S, Diamond M, Alpha Omega Alpha Medi- 10. Obedin-Maliver J, Green J, Karasic D, et al. cal Society; 2017, p15–31. Goldsmith E, Stewart L, Transgender people: Health 6. WHO (World Health Orga- White W, et al. Lesbi- at the margins of society. nization). Classifi cations: an, gay, bisexual, and Lancet. 2016;388:390–400. ICD -11 is here! 2018 [cited transgender–related 3. Treharne G, Gamble 2018 Dec 5]. Available content in undergradu- Blakey A. I’m not an from: http://www.who. ate medical education. expert: Medical teaching int/classifi cations/icd/en/ JAMA. 2011; 306:971–7. staff and transgender 7. Scott E. Gender dysphoria 11. Taylor O, Rapsey C, people’s perspectives is no longer listed as a Treharne G. Sexuality on why transgender mental disorder; 2018 and gender identity healthcare is overlooked Jun 20 [cited 2018 Dec teaching within preclin- in medical education. 5]. Available from: http:// ical medical training in Presentation to the Critical metro.co.uk/2018/06/20/ New Zealand: content, Health Education Studies gender-dysphoria-no-lon- attitudes and barriers. N Conference. 29 May 2018, ger-listed-mental-dis- Z Med J. 2018; 131:35–44. Queenstown, New Zealand. order-7645770/ 12. Gamble Blakey A, Pick- 4. New Zealand Medical 8. Riggs D, Ansara Y, Treharne ering N. Putting it on the Students’ Association: Press G. An evidence-based table: Towards better release to TV ONE: Final model for understanding cultivating medical Results of NZMSA Medical the mental health expe- student values. Med Sci Student Bullying Survey, riences of transgender Educ. 2018; 28:533–42. Wellington, NZ. 2015. Australians. Austral Psych. 2015; 50:32–9.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 106 www.nzma.org.nz/journal LETTER

Nothing short of universal healthcare coverage, fully funded and without charge both in primary and hospital care is good enough for New Zealand Stephen Main

would like to applaud Gauld et al’s recent suggestion otherwise produces dismissive editorial on the ‘elephants in the room’ arguments to the contrary.2 Even the College I for our health system.1 of GPs pays only lip service to supporting As they say, it is 80 years since the socialist equitable access for all. I am not aware of government of Michael Savage enabled at any signifi cant voice within that organ- least what we have of a universal public isation that campaigns for abolition of health service. The political vision was patient charges. that all services, from community doctors Community healthcare in New Zealand is (GPs) through to highly specialised hospital characterised by fragmentation, duplication treatments would be funded from general and complex, clumsy payment systems taxation and provided free of charge at the including highly variable patient charges. point of delivery. So-called free healthcare is of course not To the undying shame of the profession, free at all. Funded by public money and the biggest resistance to implementing these raised by taxation, it is paid for by all of the laudable principles in full came from the people. Everyone pays tax—with the possible doctors themselves. Co-payments for GP exception of the very poor, the super rich consultations in New Zealand remained and the international corporations who and persist. They vary enormously and are manage to dodge their full responsibilities high enough to discourage many, especially by using offshore tax havens! those with the least means, from seeking New Zealand spends less public money healthcare when they or their families need per capita on healthcare than any other it. This elephant in the room means that OECD country. Next up is the UK with its New Zealand general practice is effectively still-free GP service. part of the private sector for healthcare Come on New Zealand, surely you can with all the divisive potential to further afford to do better than this. By all means increase health inequities between well off keep a private sector for those who wish to and poorly off people. How primary care in work that way for the wealthy worried well, the shape of general practice can function as but let’s at least consider a fully funded, the lynchpin coordinator of care of the sick salaried GP service available without out-of- under these circumstances escapes me. pocket charges to patients. Money gets in the Given the current private business model, way and is inimical to the trust and under- GP fees remain almost universal and any standing needed for healthcare.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 107 www.nzma.org.nz/journal LETTER

Competing interests: Nil. Author information: Stephen Main, General Practitioner, Hokianga Health, Rawene Hospital, Hokianga. Corresponding author: Dr Stephen Main, General Practitioner, Hokianga Health, Rawene Hospital, Hokianga. [email protected] URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7836

REFERENCES: 1. Gauld R, et al. The versary year: general 2. Marshall T. Co-payments ‘elephants in the room’ practice charges and must go, the No Case. for New Zealand’s health ownership models. N Z J Prim Health Care system in its 80th anni- Med J 2019; 132(1489):8–14. 2011; 3(3):230–1.

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 108 www.nzma.org.nz/journal METHUSELAH

Pre-eclampsia and risk of dementia later in life Pre-eclampsia is characterised by new onset of hypertension and other signs of organ dysfunction arising during pregnancy. Endothelial and vascular dysfunction play a notable role in the pathophysiology of this disorder. This nationwide cohort study was carried out in Denmark. The cohort consisted of more than a million women with more than 20 million person-years of follow up. The results of the study showed that those with a history of pre-eclampsia had more than three times the risk of vascular dementia later in life compared with those who had not had pre-eclampsia. BMJ 2018; 363:k4109 Awareness of fetal movements and care package to reduce fetal mortality The aim of this study was to test the hypothesis that introduction of a reduced fetal movement (RFM), care package for pregnant women and clinicians that increased women’s awareness of the need for prompt reporting of RFM and that standardised management, including timely delivery, would alter the incidence of stillbirth. The study was conducted in 33 hospitals in the UK and Ireland. Over a two-year period, data were collected from 409,175 pregnancies. The RFM care package did not reduce the risk of stillbirth. The researchers conclude that the benefi ts of a policy that promotes awareness of RFM remains unproven. Lancet 2018; 392:1629–38 Pantoprazole in patients at risk for gastrointestinal bleeing in the ICU Prophylaxis for gastrointestinal stress ulceration is frequently given to patients in the intensive care unit (ICU), but its risks and benefi ts are unclear. This report is about a trial involving patients admitted to the ICU who were at risk for gastrointestinal bleeding. Three thousand two hundred and ninety-eight patients were randomly assigned to receive either 40mg of IV pantoprazole (a proton-pump inhibitor) or placebo daily during their ICU stay. The primary outcome was death by 90 days after randomisation. The conclusion reached in this study were that among adult patients in the ICU who were at risk for gastrointestinal bleeding, mortality at 90 days and the number of clinically important events were similar in those assigned to pantoprazole and those assigned to placebo. N Engl J Med 2018; 379:2199–208

URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7837

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 109 www.nzma.org.nz/journal 100 YEARS AGO

Subclavian Aneurysm Cured by Proximal and Distal Ligature By STANLEY BROWN, F.R.C.S.

he patient, a male aged 47, was ad- A double ligature of No. 4 chromic mitted to Southland Hospital on 2nd catgut was placed on the vessel and tied, TJuly, 1917, complaining of a swelling controlling pulsation in the aneurysm. at the root of the neck on the right side, A further double ligature was now placed which swelling he had noticed for about ten on the third part of the vessel distal to the months. Occasionally he had numbness and sac. tingling sensations in the fi ngers supplied The clavicle was replaced, fi xed at its outer by the median nerve. No history of specifi c end by a wire suture, and the periosteum disease nor of injury could be found. brought together as well as possible, any Examination revealed a pulsating defect being made good by fi bres of pecto- swelling about the size of a hen’s egg, ralis major. The sternomastoid was sutured evidently an aneurysm of the subclavian and the wound closed in the usual manner. artery, apparently chiefl y affecting the Progress.—Four hours later pressure on second part of the vessel. There was no the fi nger nails evidenced good circulation. difference in the radial pulses. There was no pain in the arm. Treatment.—Three months’ rest in 30th October, 1917.—Evidence of mild bed and medical treatment caused no infection in the wound, which was opened improvement, so operation was advised. with a probe, and some seropurulent fl uid th On 24 October, 1917, under heated evacuated—no radial pulse, but circulation ether anaesthesia the inner two-thirds of excellent, though there was considerable the clavicle was resected subperiosteally, oedema of the limb. and the clavicular head of the sterno- 14th December, 1917.—Owing to infection, mastoid divided. The internal jugular, the resected portion of the clavicle had external jugular, and subclavian veins were necrosed. The sequestrum was removed, retracted respectively inward, outward, and considerable new bone formation was and downward, the scalenus anticus muscle found posteriorly. There was still no percep- displayed, and on it the phrenic nerve. The tible radial pulse. fi rst part of the subclavian artery was now th palpated at great depth, and gently cleared 9 March, 1918.—Wound healed for with the fi nger. An aneurysm needle was some weeks. Radiogram showed good new now passed beneath the vessel, which could clavicle. Patient has been having massage then be pulled well up into view, revealing since seven days after original operation. He the fact that the dilatation was confi ned had now almost full use and strength in the chiefl y to the third part, extending slightly right arm, and all swelling had disappeared. into the second part. The four branches all There was still a palpable tumor at the root originated from the fi rst part, making it of the neck, but no pulsation. There was no possible to apply a ligature distal to them. radial pulse. The patient was discharged on this date, and unfortunately has not reported since.

URL: http://www.nzma.org.nz/journal/read-the-journal/all-issues/2010-2019/2019/vol-132-no-1491- 8-march-2019/7838

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 110 www.nzma.org.nz/journal published by the New Zealand Medical Association

NZMJDigest http://www.nzma.org.nz/publications/nzmjdigest

The NZMA publishes the e-magazine NZMJDigest 10 times a year. It contains news and views from the profession and the NZMA, including the NZMA Chair's editorial, along with highlights from and links to the New Zealand Medical Journal. Click on the image above to view the latest issue. We welcome contributions from members and readers. To contribute to the NZMJDigest, please email [email protected]

NZMJ 8 March 2019, Vol 132 No 1491 ISSN 1175-8716 © NZMA 111 www.nzma.org.nz/journal