How are researchers using MIMIC? A sampling-based methods review

The Medical Information Mart for Intensive Care (MIMIC) database is a widely used open data resource in critical care research. Its accessibility has led to large numbers of research studies being published; however, design and analysis decisions made by individual studies may vary, affecting comparability and reproducibility.

We conducted a sampling-based methods review of research that analysed the MIMIC database. Studies published between 2015 and 2025 were eligible for inclusion in the sampling frame. A random sample of 100 eligible studies was identified through the OpenAlex bibliographic database; data were collected using a structured extraction framework. Information was extracted on study aims, cohort definitions, exclusion criteria, outcome specification, handling of missing data, and analytical methods. The review focused on how key design and analytical decisions were reported and justified in the random sample.

Studies reported findings from general ICU and specific clinical populations; the study population could not be determined for 32 of the 100 studies. Sixty studies addressed predictive research questions, but only one predictive study reporting external validation. Core methodological decisions were frequently underreported or inconsistently justified in the full-text, including length‑of‑stay thresholds, age criteria, and mortality time horizons. Handling of missing data was not reported in more than half of studies (n = 56), despite missing data being common in MIMIC.

Among the 100 sampled publications, we observed considerable variation in study design and analysis, with key methodological decisions not always transparently reported. Improved reporting of design and analytical decisions is needed to strengthen the validity, reproducibility, and generalisability of registry‑based critical care research.