How to prevent missing data in cost-effectiveness analyses alongside cardiovascular disease trials? A review on methods of administration of patient-reported outcomes

Le résumé fourni par la source

AIM: Missing data in patient-reported outcomes (PROs) for cost-effectiveness analyses is a widespread challenge that could bias the outcome. This study aims to determine whether methods of administering PROs are associated with missing data and to identify which factors contribute to missing data. METHODS: A systematic literature search was conducted using the terms "cardiovascular disease", "cost", and "economic evaluation". Only cost-effectiveness analyses alongside clinical trials that use PROs to assess quality of life were included. The method of administration of PROs, proportion of missing data, how missing data was handled statistically, and other study characteristics were extracted. Proportions of missing data per administration strategy were calculated. RESULTS: In total, 123 articles were included in this review, median sample size was 441 (range 50 - 29,759). Median rate of missing data was 13% (range 0-44.2%) but was not reported in 35 (28.7%) studies. How PROs were administered was not reported in 36 (29.3%) studies. A combination of methods to administer PROs resulted in the least missing data (10.4%). Postal questionnaires alone had the most missing data (24.0%), while adding telephone reminders to fill in the questionnaire cut this nearly in half (12.6%). Most studies administered PROs during visits, resulting in 18.0% missing data. How missing data were addressed statistically was not reported in 32 (26.0%) studies. CONCLUSIONS: Reporting on administration of PROs, amount of missing data, and statistical handling of missing data was often lacking. The use of a reminder or a combination of methods to administer PROs seems the most effective approach to substantially decrease the proportion of missing data. CLINICAL TRIAL NUMBER: Not applicable.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.