IS THERE EVIDENCE FOR CONCORDANCE BETWEEN PATIENT AND PHYSICIANS FOR ASPECTS OF TREATMENTS THAT MATTER MOST? EVIDENCE FROM A REVIEW OF DISCRETE CHOICE EXPERIMENTS
Method: We conducted a systematic review in Medline, EMBASE, Econlit, PsycINFO and Web of Science for studies that elicited values from patient and healthcare providers using Discrete Choice Experiments (DCEs). The inclusion criteria were: peer-reviewed; complete empiric text in English from 1995-July 31st2015; discussing a healthcare-related topic; DCE methodology; comparing patients and healthcare providers. We conducted a synthesis based on the relative importance of coefficients to compare the difference in ranks for attributes by patients and physicians.
Result: After independent title and abstract review, there was a 96% agreement between all three reviewers on papers to include. In total, 38 papers from 15 countries were included, exploring 16 interventions in 26 diseases/indications. 14 studies (36.8%) generated attributes using input from both patients and providers and 15 (40%) piloted the survey in both groups. Framing of questions varied: most studies (90%) asked patients to choose for themselves, while providers were generally asked to choose for themselves (26%), their patients (24%), a standardized patient (21%), or predict their patients’ choice (11%). Methods varied considerably between studies to: analyze results, determine concordance between patient and physician values, and explore heterogeneity. Comparisons of concordance were made at the aggregate group level. The synthesis of coefficients highlighted that concordance between patients and physicians differed by the type of attribute.
Conclusion: DCE studies provide an excellent opportunity for determining value concordance between patients and providers, however, our findings found no consistent approach was used, and identified challenges in the interpretation of concordance from the various approaches. Concordance has also been tested at the aggregate level, which fails to account for heterogeneity in patient and physician preferences. We also show that concordance can vary according to which attributes are being considered, highlighting that concordance should not be considered a binary outcome.