OBJECTIVES: To describe and identify recruitment and data collection characteristics of noninterventional medical record review studies (NIMRRs) associated with participation rates and data collection duration.
METHODS: Recruitment and data collection monitoring data from 86 patient samples (14,249 charts) collected for health economic and outcomes NIMRRs during the last 5 years were analyzed. Patient samples were collected in nine disease areas and eight countries/regions in North America and Europe. Recruitment (e.g., disease area, country, physician specialty) and data collection (e.g., data collection form [DCF] length, type of abstractor) characteristics were summarized. Factors associated with data collection duration (i.e., number of months from start of recruitment to collection of last chart) and physician participation rates (i.e., proportion of physicians that started eligibility screening who ultimately abstracted data) were independently assessed.
RESULTS: Across all samples, the mean duration of data collection was 2.5 months and differed by country (1.6 months [UK] to 3.6 months [Belgium/Netherlands]). Samples abstracted by research coordinators took longer to collect than those abstracted by physicians (mean, 7.1 and 2.2 months, respectively). Factors independently associated with data collection duration included country and type of abstractor; disease area was not significantly associated. Across all samples, 57.2% of physicians participated in chart abstraction. Participation rates were highest in rare diseases (78.0%) and Italy (86.3%); lowest in dermatology (8.71%) and Belgium/Netherlands (23.9%). Country, disease area, and physician type were independently associated with participation rates; DCF length was not significantly associated.
CONCLUSIONS: Multiple recruitment and data collection characteristics influence timelines and participation rates. Physician abstractors, particularly those treating rare diseases, may be more motivated to participate as they have direct patient contact and want to support research in their specialties and access to new treatments. Consideration of these factors in designing NIMRRs can facilitate estimation of realistic timelines, sample sizes, and research costs.
