Back

Estimating the Effects of Individual Nurses on ICU Outcomes

Brossette, S. E.; Zheng, N.; Wong, D. Y.; Hymel, P. A.

2020-11-06 nursing
10.1101/2020.11.04.20226340 medRxiv
Show abstract

A better understanding of the effects of nursing on clinical outcomes could be used to improve the safety, efficacy, and efficiency of inpatient care. However, measuring the performance of individual nurses is complicated by the non-random assignment of nurses to patients, a process that is confounded by unobserved patient, management, workforce, and institutional factors. Using the MIMIC-III ICU database, we estimate the effects of individual registered nurses (RNs) on the probability of acute kidney injury (AKI) in the ICU. We control for significant unobserved heterogeneity by exploiting panel data with 12-hour fixed effects, and use a linear probability model to estimate the near-term marginal effects of individual RN assignments. Among 270 ICU RNs, we find 15 excess high-side outliers, and 4 excess low-side outliers. We estimate that in one twelve-hour work shift, each high-side RN outlier increases the probability of AKI by about 4 percentage points, and in 25 work shifts, causes about one additional AKI. Conversely, each low-side outlier prevents about one AKI in 50 work shifts. Given the fine-grained nature of the fixed effects employed, we believe that the estimated individual nursing effects are approximately causal. We discuss our contribution to the literature and identify potential use cases for clinical deployment.

Matching journals

The top 8 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.