Assessment of trial similarity and evidence consistency for indirect comparisons: preliminary results

Tags: Oral
Xiong T1, Parekh-Bhurke S2, Kong-Loke Y3, Song F3
1School of Medicine, Health Policy and Practice, University of East Anglia, Norwich, UK, 2School of Allied Health Professions, University of East Anglia, Norwich, Norfolk, UK, 3Faculty of Health, University of East Anglia, Norwich, Norfolk, UK

Background: Adjusted indirect comparisons (AIC) are increasingly used for evaluating healthcare interventions in situations where there is limited head-to-head trial evidence. However methods for assessing validity of AIC have not been systematically developed and tested. Objectives: To investigate clinical similarity of trials in AIC, and clinical consistency between direct comparisons (DC) and AIC. Methods: The Cochrane Database of Systematic Reviews was searched for Cochrane reviews (CSRs) that contained sufficient data to enable both DC and AIC analysis. Two reviewers independently extracted data using a standardized proforma. Clinical diversity and similarity were assessed by comparing characteristics of participants, interventions, and outcome measures. The results of the assessment of trial similarity and clinical consistencywere expressed through ‘similarity’ and ‘consistency’ scores, ranging from 0 (very low) to 5 (very high) for average of the three components. Similarity of trial quality was also scored from 0 (very low) to 5 (very high). Results: 95 CSRs were included in analyses, involving a total of 1109 trials (332 trials for direct and 777 trials for indirect comparisons). 75 CSRs investigated pharmaceutical interventions and 11 CSRs included surgical interventions, the remaining 9 CSRs involved different interventions including rehabilitation, psychoeducational, etc. For trial similarity assessment, 31 cases scored from 4 to 5, 57 cases from 3 to 4, 7 cases lower than 3. For clinical consistency assessment, 16 cases scored from 4 to 5, 70 cases from 3 to 4, 9 cases lower than 3. In terms of quality similarity scores, 6 cases scored 5, 54 cases were in the range of 4 to 5, 25 cases from 3 to 4, and 10 cases lower than 3. Conclusions: Evaluation of the validity and appropriateness of the AIC by assessing the trial characteristics and qualities is feasible. Validation of the similarity and consistency score is in progress.