What can we learn from international healthcare systems to improve early diagnosis in primary care?

Chair: Prof Jon Emery

Speakers: Prof David Weller, Prof Peter Vedsted, Prof Lyndal Trevena, Prof Surendra Shastri

Session code: CTS.4.32

www.worldcancercongress.org Overview

1. Health service initiatives in the United Kingdom David Weller, University of Edinburgh 2. Health service initiatives in Denmark Peter Vedsted, Aarhus University 3. Community symptom awareness campaigns Jon Emery, University of Melbourne 4. Symptom risk assessment tools for primary care Lyndal Trevena, University of Sydney 5. How applicable are these system initiatives to low and middle income countries? Surendra Shastri, Tata Memorial Hospital 6. Discussion

Model of Pathways to Treatment

Walter, Scott, Webster, Emery. JHSRP 2011

Healthcare system delays

Olesen BJC 2009 Prof David Weller What can we learn from international healthcare systems to improve early cancer diagnosis in primary care:

Health service initiatives in Denmark

Peter Vedsted Professor, Director

Research Centre for Cancer Diagnosis in Primary Care – CaP The Research Unit for General Practice Aarhus University Denmark

www.cap.au.dk [email protected]

Health care system in Denmark

Tax-financed with free medical care Primary care (GPs) is first line with a patient list of 1600 • Gatekeepers to hospitals and specialists Family medicine is a speciality as every other medical speciality A Dane on average contacts the GP 8 times per year (4 face- to-face) Denmark did badly: survival, stage and waiting ‘It all starts in primary care…’

More than 75-85% of are seen in primary care

90% of cancers are diagnosed based on symptoms

5-8% of cancers are found with screening (at the moment)

Allgar et al. British Journal of Cancer 2005;92:1959–70 Hansen et al. BMC Health Serv Res. 2011;11:284. Emery JD, et al. Nat Rev Clin Oncol. 2014;11:38-48. Vedsted P, et al. Scand J Prim Health Care. 2009;27:193-4. Following 6% of consultations, GPs suspected a serious disease 10% of these patients got a serious disease in 2 months

Hjertholm P, et al. Br J Gen Pract 2014 Three groups of symptoms at presentation

Symptom %

Alarm symptom 50

Serious, unspecific 20

Uncharacterisc, vague 30

Jensen H, et al. BMC Cancer 2014;14:636 Nielsen T, el al. Ugeskr Læger. 2010;172:2827-31 Need for a 3-legged diagnostic strategy

Alarm symptom (the obvious) • Urgent referral investigations for a specific cancer Unspecific, serious symptoms (the difficult) • Diagnostic centre – fast multidisciplinary assessment Vague symptoms (the common) • Quick and direct access to investigations (e.g. ultrasound)

- Vedsted, Olesen. Early diagnosis of cancer--the role of general practice. Scand J Prim Health Care. 2009;27:193-4. - Olesen, Hansen, Vedsted. Delay in diagnosis: the experience in Denmark. Br J Cancer. 2009 Dec 3;101 Suppl 2:S5-8. - Rubin, Vedsted, Emery. Improving cancer outcomes: better access to diagnostics in primary care could be critical. Br J Gen Pract. 2011;61:317-8. - Jensen H, et al. Cancer suspicion in general practice, urgent referral and time to diagnosis. BMC Cancer. 2014;14:636 The urgent referral for specific cancer

If the GP suspects cancer (a list of symptoms and signs) Has given shorter diagnostic intervals (for some)

40% are diagnosed through urgent referral in DK 60% are not diagnosed using the expedited route!

- Meechan D et al. BrJGP 2012 DOI: 10.3399/bjgp12X654551. - Elliss-Brookes et al. BrJCan 2012, 1220–1226. - Jensen H et al. BMC Cancer 2014; Urgent referral to diagnostic centres If the GP cannot allocate the patient to a specific alarm route The GP performs a filter function: • Imaging and blood samples within 2 days • If no explanation, then referral and seen within 2 days A multidisciplinary team of specialists at hospital Outpatient ‘pit-stop’ The first results from diagnostic centres

50% from the GP filter function is referred to diagnostic centre 16% of those referred get a cancer diagnosis The first results from diagnostic centres Cervix, ovarian and uterus cancer, 3% , 19% , 4% Esophagus and stomach cancer, 4%

Breast cancer, 4%

Metastasis; 5% Hematopoietic cancer; 15% Kidney cancer , 5%

Liver and biliary system cancer, 6% Colorectal cancer, Prostate cancer, 7% 12% Pancreatic cancer, 8% Ingeman ML, et al. Under preparaon Direct access to investigations

Implemented as ‘No-Yes-Clinics’ (NYC)

For the ‘low-risk-but-not-no-risk’ group

GPs have direct access to expedited investigations

• Ultrasonic investigation of abdomen, pelvis, CT, endoscopy

The GP is fully responsible

No record, history taking etc. at clinic – only a No or a Yes! Direct access to abdominal ultrasonic investigation

Suspected cancer (n=43) Cancer risk = 28% Same-day investigation (n=434) Cancer risk = 5.3% No suspicion of cancer (n=391) Cancer risk = 2.8%

Ingeman ML, et al. 2014 (under publication) The 3-legged strategy for cancer diagnosis

§ Alarm symptom (the obvious) § Urgent referral, specific cancer § Serious, unspecific symptoms (the difficult) § Diagnosc centre § Vague symptoms (the common) § Quick and direct access Thank you What can we learn from international healthcare systems to improve early cancer diagnosis in primary care:

Community symptom awareness campaigns

Jon Emery Herman Professor of Primary Care Cancer Research University of Melbourne Director of PC4

Evidence for effect of symptom awareness campaigns?

In progress Effect on early presentation (Austoker et al 2009)

Reference Cancer Country Outcome Catalano 2000 Breast USA 790 additional cases over 23 yrs Gabram 2008 Breast USA 14% increase stage 0 and 8% reduction stage IV MacKie 2003 Melanoma Scotland 34% increase thick melanomas Rossi 2000 Melanoma Italy Mean thickness reduced by 0.5mm Geczi 2001 Testicular Hungary No change in time to diagnosis Leander 2007 Retinoblastoma Honduras 38% reduction in advanced disease Lung cancer awareness campaigns

Athey et al 2012: • Campaign plus GP intervention • 20% increase CXRs • 27% increase lung cancer diagnoses Be Clear on Cancer (UK)

• 30% increase in urgent referrals for suspected lung cancer • 9.1% increase in lung cancer diagnoses during campaign months • Significant shift towards earlier stage distribution for NSCLC.

http://www.england.nhs.uk/wp-content/uploads/ 2013/12/be-clear-cancer.pdf The Improving Rural Outcomes Trial IRCO Trial Find Cancer Early Campaign

Computer-assisted telephone survey

Timing: 18 months into campaign Sample size: 725 intervention, 725 control regions

Stratified by age and gender

71% participation rate 94% response rate

Awareness of Find Cancer Early

100.0%

80.0% 43.30% Not aware 60.0% 83.40% Prompted recognition 40.0% 31.60% Recognition Recall 20.0% 7.60% 21.60% 8.80% 0.0% 0.10% 3.50% Control regions Campaign regions Symptom awareness

Campaign Control n % n % Coughing up blood 56 7.7 51 7.0 A cough or croaky voice 53 7.3 38 5.2 Becoming more short of breath 39 5.4 40 5.5 Blood in your pee 128 17.6* 60 8.3* Blood in your poo 227 31.3* 136 18.7* Problems peeing 29 4.0 22 3.0 Looser poo (diarrhoea) 44 6.1 40 5.5 Unusual pain, lump or swelling 464 63.9* 425 58.5* Unexplained weight loss 157 21.6 140 19.3

* P < 0.05 The Improving Rural Outcomes Trial

• 2x2 factorial design • Symptom awareness campaign • GP intervention • Primary outcome: time to diagnosis • 1351 cancer patients recruited • Data analysis in progress Prof Lyndal Trevena What can we learn from international healthcare systems to improve early cancer diagnosis in primary care:

Symptom risk assessment tools for primary care

A/Prof Lyndal Trevena Head Discipline of General Practice Sydney School of Public Health University of Sydney, Australia Chair Prevention and Early Diagnosis Scientific Working Group, PC4

The dilemma for primary care

Symptoms are poor predictors of cancer BUT… Most cancer patients present in primary care SO… Can we be more systematic about finding the ‘needle in the haystack’?

4th December, 2014 What are risk assessment tools? Usually an algorithm that combines a number of ‘risk factors’ (or ‘symptoms’) for the disease of interest They provide an estimate of the ‘chance’ of having (or getting) the disease of interest now (or over a period of time) Risk assessment tools based on risk factors have been around for some time (e.g. Gail model for breast cancer risk) Usually a paper-based chart or web-based tool. Some have been integrated to GP software http://canceraustralia.gov.au/affected-cancer/cancer-types/breast-cancer/ your-risk/calculate 4th December 2014 The relationship between pre-test probability (risk assessment) and the utility of a ‘test’

Positive LR chest CT for lung cancer if >30mm nodule is found on CXR = 3.7 Clinician estimates of pre-test probabilities vary widely Three clinical scenarios given to 500 physicians and 500 GPs randomly selected from NSW College registers; also sent to 205 physicians and 202 GPs randomly sampled from UK college equivalents (2001) 60% and 57% response rates Australia and UK respectively 1. Chest pain and risk of IHD (54% were within 20 percentage points) 2. Risk of DVT (only 6.7% within 20 percentage points) 3. AF and risk of stroke (57% within 20 percentage points) UK results were very similar to Australian Physicians were generally more accurate than GPs but the spread of estimates was similarly wide between cardiologists and GPs This is consistent with studies that clinical decision rules out-perform even expert clinicians

(Attia et al MJA 2001)

Clinical decision rule for diagnosing influenza

Ebell et. al. J Am Board Fam Med 2012 4th December 2014 QCancer: A tool to predict the chance (pre-test probability) of having cancer given a set of symptoms

Uses GP-based data from a representative cohort of 2.5 million men and women aged 25-89 years in UK

Predicts global cancer risk and 12 cancer types (which account for 85% all cancers)

Symptom rather than cancer- based (same symptom can occur in several cancers

www.qcancer.org

4th December 2014 Risk Factors and Symptoms included in Qcancer Algorithm Risk Factors Symptoms Age & sex Haemoptysis Haematemesis Smoking status Haematuria Deprivation score Rectal bleeding Family history of cancer Haematuria Unexplained bruising COPD Constipation, cough Endometrial hyperplasia/polyp Vaginal bleeding (women) Chronic pancreatitis Testicular lump (men) Type 2 diabetes Loss of appetite Unintentional weight loss Anaemia (HB < 11g/DL) Indigestion +/- heart burn Venous thromboembolism Dysphagia Abdominal pain or swelling Breast lump, pain, skin Night sweats Neck lump Urinary symptoms (men)

How accurate is QCancer? Threshold for investigating?? Can these tools be implemented in primary care practice?

The CAPER tools are derived from a series of case–control studies using data from a national general practice database in the UK, which estimate the positive predictive value of symptoms, signs and common investigations, singly and in combination, for a wide range of cancers.

4th December 2014

Evaluation of CAPER tools for lung and colorectal cancer in 28 UK general practices GPs were given the tools as mouse-mats or flipcharts No advice given about what risk-level to investigate Outcome was 2 –week referral or CXR (for lung cancer), number of RATs used and subsequent investigations and referrals

Hamilton et. al Br J Gen Pract Jan 2013 4th December 2014 More lung and colorectal cancers were diagnosed at what cost? New lung cancer diagnoses There was an increase in 2wk increased from 127 in the 6 months before the evaluation to referral for lung cancer from 332 to 174 during the pilot (47 extra 436 (104 extra referrals) and a 4% cases). increase in CXRs The proportion of stage 1 and There was an increase in 2 week stage 2 cancers combined (as a proxy for potential cure) increased referrals for colorectal cancer from by 19% (from 26 to 31). 1173 to 1477 (304 extra referrals) New colorectal cancers increased and increase in colonoscopies from by from 134 to 144 (10 extra 1762 to 2032 (270 extra cases). colonoscopies) No significant change in staging was seen (data not shown).

4th December 2014 Australian study with 15 GPs using QCancer

The risk tool was perceived as potentially useful for patients with complex histories. More experienced GPs were distrustful of the risk output, especially when it conflicted with their clinical judgement. Variable interpretation of symptoms meant there was a significant variation in risk assessment. When a risk output was high, GPs were confronted by numerical risk outputs creating challenges in the consultation

4th December 2014 GPs more likely to refer…implications?

‘Mid-project we looked at our numbers and we felt that under the clause, would you have referred this patient if you hadn’t been using the risk assessment tool, there was a significant minority that said, you know, the tool had pushed them to a different decision ... it was 10–15% of people that may have waited longer if they hadn’t had the tool.’ (GPL/2) ‘I think our referral thresholds for lower GI have definitely gone down.’ (GPL/3) ‘I’d say particularly it got us thinking about patients with COPD, because, um, there’s a bit in the, ah, in the lung tool which is smokers with COPD saying they automatically should have a referral for a chest X-ray, and that made us think about how frequently we should do chest X-rays in our COPD patients.’ (GPL/8)

Chiang et al. (in press BJC) 4th December 2014 These findings are similar for implementation of other risk assessment tools in general practice Five communication strategies used: 1. AR-focused strategy, used when AR assessment was considered useful for the patient; 2. AR-adjusted strategy, used to account for additional risk factors such as family history 3. Clinical judgement strategy, used when GPs considered that their judgement took multiple risk factors into account as effectively as AR; 4. Passive disregard strategy, used when GPs lacked sufficient time, access or experience to use AR; 5. Active disregard strategy, used when AR was considered to be inappropriate for the patient.

Bonner et. al. MJA 2013 4th December 2014 ..and there is an ever-increasing number of risk assessment tools being developed for primary care Strengths and limitations of RATs for cancer diagnosis in primary care Potential for more Uncertainty about which tool to use rational testing and Lack of ‘trust’ by providers The plethora of tools not linked to clinical investigation reasoning processes & guidelines Early evidence for Lack of evaluation of the impact on earlier diagnosis if ‘over-testing’, ‘over-investigation’ and ‘missed cancers’ versus ‘tests avoided’ linked to fast-track and ‘improved cancer outcomes’ referrals for some Need for improvement in user interface cancers Need for guidance on thresholds for testing

4TH December 2014 Prof Surendra Shastri Low and Middle Income Countries Perspective

Prof Surendra Shastri Tata Memorial Centre ,

Shastri, A. and S. S. Shastri (2014). "Cancer screening and prevention in low-resource settings.“ Nat Rev Cancer 14(12): 822-829. Dec 4, 2014 Disease Burden

In 2012, less-developed-region countries (LDCs)

contributed 57% and 65% to the global cancer

incidence and mortality, respectively

Ferlay, J. et al. GLOBOCAN 2012 v1.0, Cancer Incidence and Mortality Worldwide: IARC CancerBase No. 11. International Agency for Research on Cancer [Online] http://globocan.iarc.fr (2013). Disease Spectrum q Cancers that are caused by infectious agents (that is, cancers of the cervix, liver and stomach) continue to have the highest incidence and mortality in these countries. q Cancers that are related to lifestyle and environmental factors (that is cancers of the lung, breast, prostate and colon–rectum) are now appearing as the top cancers in Africa, Asia and Latin America Challenges q No health insurance q Lack of trained manpower (physicians, nurses, technicians, pathologists, radiologists) q Lack of basic infrastructure (laboratories, screening- diagnostic facilities) q Large populations coupled with poor resources q Poor health awareness and health seeking behavior So what do we see q Over 70% of the cancers report to a treatment centre at advanced stages q High treatment costs and disproportionately high mortality rates Examples q No screening for : Cytology-based screening is not feasible q No screening for breast cancer : Mammography screening is not feasible; population is young Are developed country models/systems applicable to low resource settings? q Cannot be just replicated; need to be suitably modified and adapted to circumstances q Appropriate and feasible models need to be developed q Invest in manpower and infrastructure development for low cost interventions; health awareness; legislations to protect people’s health Examples of models for low resource settings?

q Visual inspection with acetic acid; low cost HPV-DNA screening for cervical cancer q Physical breast examination by primary health care providers for breast screening q Awareness programmes and increased taxation for tobacco control

Shastri, S. S. et al. Effect of VIA screening by primary health workers: randomized controlled study in Mumbai, India. J. Natl Cancer Inst. 106, dju009 (2014). Sankaranarayanan , R. et al. HPV screening for cervical cancer in rural India. N. Engl. J. Med. 360, 1385–1394 (2009). VIA Negave VIA Chart developed by IARC Faint acetowhite Dot-like pale Acetowhite areas far Streak-like Line-like areas without a areas in the away from the TZ acetowhitening acetowhitening sharp outline endocervix

VIA Posive Thick well-defined acetowhite areas, near the Transformaon Zone (TZ) either on the endocervix or ectocervix (or both) are VIA posive

Thank You for Your Attention

65 Questions and Discussion

• Questions to speakers • Other international examples of policy initiatives • How context specific are these types of intervention? • Could any of these be applicable to your own country and how? • How well developed a primary care system is needed to make them work?