Skip to content
← All work

Statistics · Oct 2025

Hospital Infection Risk

Two regression models on 113 US hospitals, compared with 10-fold cross-validation.

Data
113 US hospitals
Method
Two multiple linear regression models in R, compared with 10-fold cross-validation
  • R
  • Regression
  • Cross-validation

The question

Hospital-acquired infections threaten patient safety and raise costs. Which hospital and patient factors explain why infection risk is higher at some hospitals than others?

What I found

Length of stay is the strongest predictor of infection risk (p < 0.001). Each extra day of average stay adds about 0.4 points of risk. Patient age adds little once stay is included (p = 0.21).

Effect on infection risk (Model 2: stay + age)

Regression coefficients with 95% confidence intervals. Age's interval crosses zero.

Both models (stay + region, and stay + age) explain a moderate share of the variation (adjusted R² 0.30 and 0.28) and predict equally well on held-out data: mean RMSE 1.14 across 10 folds. A real limitation: if hospitals report or define infections differently, the data and the model are less reliable.