Measuring patient-reported outcome measures (PROMs) is essential for improving intensive care medicine, but it is challenging and prone to bias. This study investigates the representativeness of a multicenter, PROM-based cohort of Intensive Care Unit (ICU) survivors. Demographic, clinical and ICU admission characteristics of the multicenter, PROM-based cohort (n = 6 ICUs) were compared to those of all ICU survivors in Dutch hospitals (n = 73 ICUs), in the years 2019 and 2022, based on data from the National Intensive Care Evaluation registry. Comparison between the PROM-based cohort (n = 2,454) and the national registry cohort (n = 89,154) revealed predominantly similarities in demographic, clinical, and ICU admission characteristics. Nevertheless, ICU survivors in the PROM-based cohort had a higher severity of illness (59 vs. 56 points) and mortality probability (19% vs. 16%), were more often mechanically ventilated during the first 24 hours after ICU admission (49% vs. 34%), had higher ICU and hospital lengths of stay (respectively, 4.7 vs. 3.6 days, and 16 vs. 14 days), and lower mortality rates in-hospital and at 3, 6, and 12 months after ICU admission (respectively, 1.4% vs. 4.6%, 4.2% vs. 10%, 6.2% vs. 13%, and 9.2% vs. 17%). ICU survivors in the PROM-based cohort share similar demographic, clinical, and ICU admission characteristics with the national ICU population. However, severity of illness, lengths of stay, and mortality rates deviate from the national registry cohort. These findings highlight external validity concerns, urging researchers and policymakers to consider this when using outcome data from a PROM-based cohort.
Gemert et al. (Tue,) studied this question.