Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments?
Download as PPTX, PDF1 like352 views
The document discusses the sensitivity of disease-specific instruments compared to generic measures in assessing cancer outcomes, highlighting the need for more reliable assessment methods. It analyzes various tools like EQ-5D-3L, EORTC-8D, and QLU-C10D in their effectiveness to capture the quality of life in cancer patients and emphasizes the differences in their utility and responsiveness. Further research is necessary to assess the implications of using these measures in health technology assessments and patient care.
Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments?
- 1. Paula Lorgelly Deputy Director, OHE Visiting Professor, Division of Cancer Studies, King’s College London C2E2 Rounds 5th July 2017 Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments?
- 2. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Acknowledgements • Cancer 2015 is funded by the Victorian Cancer Agency Translational Research Program, Australia • The cancer patients who agreed to participate and hospitals from which they were recruited • Cancer 2015 investigators, specifically Stephen Fox, David Thomas and John Parisot • Brett Doble, Monash PhD student (now Oxford) • Richard Norman, Curtin University
- 3. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Background – Generic MAUIs • Cost utility analyses require preference-based measures of outcome to generate QALYs • Norm has been to employ generic preference based measures (multi attribute utility instruments, MAUIs) • EQ-5D or SF-6D • Criticisms (both health economists and clinicians) that these are not sensitive to certain disease specific improvements
- 4. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Background – Condition Specific MAUIs • How do we introduce sensitivity to disease? • Could map from a condition specific non-preference- based measure to a generic preference-based measure • Could use a bolt-on to a generic measure • Could elicit patient preferences using the generic measure • Growing trend to develop condition-specific preference- based measures (CSPBM)
- 5. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Background - CSPBM • Two approaches • developing CS measures from scratch or • modifying an existing non-preference based measures • CSPBMs use is limited within HTA, but instruments/ algorithms are increasing • Research is required to aid adoption/value considerations • evidence of validity, responsiveness and sensitivity
- 6. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? CSPBM in oncology • EORTC-8D (Rowen et al, 2011) • QLU-C10D (King et al, 2016) • Both map responses from the EORTC QLQ-C30 a generic measure of quality of life for cancer patients • EORTC = European Organisation for Research and Treatment of Cancer
- 7. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Which to choose?
- 8. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Objective • To compare and contrast the EQ-5D-3L, the EORTC-8D and the QLU-C10D • To understand the sensitivity of each to different sites and stages of cancer and the effect of variations in patient functioning on each outcome measure over time • What do we already know? • Lorgelly et al., 2017, Quality of Life Research • Evidence of convergent validity and content validity but limited agreement between EORTC-8D and EQ-5D-3L • Variability in sensitivity between the baseline values and the QALY estimations of each • Researchers and decision makers advised to be cautious
- 9. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? EORTC-QLQ-C30 (version 3) • 30 questions • physical functioning • role functioning • emotional functioning • social functioning • cognitive functioning • pain • fatigue • nausea and vomiting • dyspnoea • sleep • appetite • constipation • diarrhoea • financial impact • global health • 3 summary measures • Physical functioning (0-100) • Symptoms (0-100) • Global QoL (0-100)
- 10. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? EORTC-8D • Developed from EORTC-C30 using Rasch analysis and factor analysis on dataset of multiple myeloma patients • TTO valuation using ‘healthy’ individuals from the North of England • An algorithm takes responses to EORTC-C30 and gives a utility score (10 questions in 8 domains) • 81,920 unique health states • Range 0.292-1.000 (UK specific algorithm) • Physical functioning • Role functioning • Pain • Emotional functioning • Social functioning • Fatigue • Nausea • Constipation & diarrhoea
- 11. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? QLU-C10D • Developed from EORTC-C30 using apriori inclusion/ exclusion with Confirmatory Factor Analysis and Rasch analysis on a meta-dataset of 18 studies from 8 countries • DCE with 16 choice sets of the general Australian public • An algorithm takes responses to EORTC-C30 and gives a utility score (13 questions in 10 domains) • 410=1,048,576 unique health states • Range -0.095-1.000 (Australian-specific algorithm) • Physical functioning • Role functioning • Social functioning • Emotional functioning • Pain • Fatigue • Sleep • Appetite • Nausea • Bowel problems
- 12. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? EQ-5D • Originally referred to as the EuroQol-5D • Five domains • Three level and five level versions • Dataset includes EQ-5D-3L (5L version subsequently used) • 3L, 35 = 243 unique health states • Range -0.594-1.000 (UK MVH tariff) • Mobility • Self-care • Usual activities • Pain/Discomfort • Anxiety/Depression
- 13. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? EQ-5D-3L EORTC-8D QLU-C10D Maximal theoretical contributions for each dimension
- 14. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Data • Cancer 2015 is a large-scale longitudinal prospective population-based cancer genomic cohort study • Enrols treatment naïve cancer patients (all tumour sites except leukaemia) • Phase 1 targeted the enrolment of 1000 patients from 5 hospitals in Victoria • Follow-up at 3, 6 and 12 months depending on severity • Collection of tumour samples (blood), baseline questionnaire of demographics, site and staging, treatment (pharma, chemotherapy, radiotherapy, surgery), EQ-5D-3L, and EORTC-QLQ-C30 • Also linked to administrative reimbursement data • www.cancer2015.org
- 15. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Summary of Patient Accrual 316 182 505 259 328 Patients Recruited by Hospital Cabrini Institute Warrnambool Hospital Peter MacCallum Cancer Centre Geelong Hospital Royal Melbourne Hospital Cancer 2015 Dashboard as at 13-Aug-2014 Total Patients = 1590 GRAMPIANS 2% LODDON MALLEE 5% BARWON SOUTH WEST 26% GIPPSLAND 6% HUME 4% NORTHERN & WESTERN METRO 25% EASTERN METRO 11% SOUTHERN METRO 21% OTHER 0% VICTORIA METRO
- 16. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? 0 50 100 150 200 250 300 350 400 450 500 Breast Prostate Colorectal Lung Melanoma HeadandNeck Renal Oesophagogastric BoneandSoTissue Cervical CancerofUnknownPrimary Other CentralNervousSystem Pancreac Bladder Ovarian Thyroid Endometrial Tescular Anal Hepac Biliary Lymphoma Not Specified IV III II I 0 Cancer histo-site & stage
- 17. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Representativeness Anal 0.17% Biliary -0.89% Bladder -0.05% Bone and Soft Tissue -0.78% Breast 6.13% Cancer of Unknown Primary -0.14% Central Nervous System 0.12% Cervical 1.56% Colorectal -2.23% Endometrial -1.42% Head and Neck 9.80% Hepatic -0.59% Lung 0.58% Lymphoma -3.73% Melanoma 1.57% Oesophagogastric -0.45% Other -5.57% Ovarian -0.34% Pancreatic -0.67% Prostate -3.18% Renal 1.04% Testicular -0.07% Thyroid -0.87%
- 18. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Case history – Patient X • Cohort baseline questionnaire: male, aged 60-64, lung cancer, metro Victoria, presented asymptomatic/ incidental in July 2012, no previous history of Ca, no familial history of Ca, no other chronic conditions, smoker • Genomics panel: RB1 (retinoblastoma 1); FGFR3 (fibroblast growth factor receptor 3) • Follow-up: Tx – surgery in July 2012, chemotherapy in Aug 2012, radiotherapy Jan 2013 • QoL in July 2012: Functioning 91, Symptoms 10, Global Health 100, VAS 95, EQ-5D 0.848 • QoL in Oct 2013: Functioning 49, Symptoms 54, Global Health 33, VAS 60, EQ-5D 0.587
- 19. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Case history cont. • PBS activity: Atorvastatin and Ramipril since Nov 2011 (continued through anticancer treatment) • Etoposide and Carboplatin in Sept 2012 • Mirtazapine in Oct 2012 and Diazepam in Nov 2012 • MBS activity: GP appointment in April 2012 (note no activity prior to this) • Pathology and Radiology in May 2012 • More path and diag-radio July to Nov 2012 • Psychologist in Nov 2012 and Feb 2013 • Radiology in Jan 2013 • VAED activity: July 2012 1 day stay for H61B (ICD- AM C787), Aug-Nov 2012 18 separate day admissions for R63Z (Z511)
- 20. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Deceased Cancer Dx Cancer 2015 enrolment GP visit QoL QoL Chemotherapy Pathology /radiology Psychologist Psychologist Radiotherapy Statins and BP lowering meds
- 21. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Methods • Estimated generic and cancer-specific baseline utility values for each respondent • Similarly estimated utility values at each follow-up point and then using area under curve estimated QALYs • Simple comparison of values • Regression analyses to test sensitivity to site and stage
- 22. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Sample characteristics • N = 1,663 patients with baseline health state • Mean age 62 years • 56% male • 14% current smoker, 47% ex-smoker • 40% have health insurance (indicator of high income) • 74% had curative treatment intention, 19% had palliative • 15% treated (recruited) in private hospital
- 23. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Site of cancer breast prostate headneck colo lung bone cervic renal CUP oesoph othersite
- 24. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Stage of cancer stage 0 stage 1 stage 2 stage 3 stage 4 stage 5 unable to stage
- 25. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Health Status @ baseline • EQ-5D-3L mean 0.749 (range: -0.594, 1) • EORTC-8D mean 0.830 (range: 0.292, 1) • QLU-C10D mean 0.736 (range: -0.022, 1) • Correlation coefficient: • EORTC-8D vs EQ-5D-3L = 0.746 • QLU-C10D vs EQ-5D-3L = 0.745 • EORTC-8D vs QLU-C10D = 0.947 • Interclass correlation coefficient (ICC) – measure of agreement • EORTC-8D vs EQ-5D-3L = 0.595 (moderate) • QLU-C10D vs EQ-5D-3L = 0.739 (strong) • EORTC-8D vs QLU-C10D = 0.770 (strong)
- 26. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Histogram of baseline utility values
- 27. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Coefficient values on site – regression of baseline utility (breast is ref)
- 28. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? QALYs • N = 1,142 • Mean follow-up = 435 days • Follow-up max days = 1,146 (3.14 years) • EQ-5D-3L meanQ = 0.862 (range: -0.108, 3.138) • EORTC-8D meanQ = 0.911 (range: 0.001, 3.078) • QLU-C10D meanQ = 0.821 (range: -0.000, 3.054) • Correlation > 0.95 • Difference +ve for 8D while –ve for 10D (-0.049 & 0.042, p<0.001)
- 29. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Correlation in cancer QALY estimates
- 30. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Regression for difference in QALYs
- 31. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Discussion • The three instruments produce different baseline values but notably different QALY estimates • Variation between EORTC-8D and QLU-C10D of interest as use the same initial instrument, and overlapping questions • There would appear to be differences in the sensitivity of each instrument to different sites and stages of cancer, and the responsiveness of QALYs to identify severity of disease and change in patient functioning • Differences in baseline values are of interest as these values would populate HTA decision models • It is acknowledged that while the values are on the same (theoretical) scale they accommodate different ranges on that scale
- 32. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Conclusion • It is well known that generic preference-based measures often produce different conclusions. • This analysis suggests that the performance of the cancer-specific measures, the EORTC-8D and the QLU- C10D, differ considerably to that of the EQ-5D-3L
- 33. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Further research • Further research is required to understand the reasons for the variability, particularly if recommendations for reimbursement/adoption change in light of using one instrument over another • Further research is required to understand if the new five-level version, the EQ-5D-5L, experiences similar issues or is as hoped an improvement on the three-level
- 34. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? References • King MT et al. “QLU-C10D: a health state classification system for a multi-attribute utility measure based on the EORTC QLQ-C30.” Quality of Life Research 25 (2016): 625-636. • Rowen D et al. “Deriving a preference-based measure for cancer using the EORTC QLQ-C30.” Value in Health 14 (2011): 721-731. • Thomas DM et al. “Cancer 2015: a longitudinal whole-of-system study of genomic cancer medicine.” Drug Discovery Today 20 (2015): 1429- 1432. • Lorgelly et al. “Condition-specific or generic preference-based measures in oncology? A comparison of the EORTC-8D and the EQ-5D-3L.” Quality of Life Research, 26 (2017): 1163–1176. • Lorgelly et al. “Realising the value of linked data to health economic analyses of cancer care: a case study of cancer.” Pharmacoeconomics, 34 (2016): 139-154.
- 35. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? Questions? Suggestions? Comments?
- 36. Outcome Measures in Cancer: Do disease specific instruments offer greater sensitivity than generic instruments? To enquire about additional information and analyses, please contact Paula Lorgelly at plorgelly@ohe.org To keep up with the latest news and research, subscribe to our blog, OHE News Follow us on Twitter @OHENews, LinkedIn and SlideShare Office of Health Economics (OHE) Southside, 7th Floor 105 Victoria Street London SW1E 6QT United Kingdom +44 20 7747 8850 www.ohe.org OHE’s publications may be downloaded free of charge from our website.