SMR: Standardized Mortality Ratio
Standardized Mortality Ratio is the observed number of deaths divided by the expected number given case mix. SMR < 1.0 means lower-than-expected mortality; > 1.0 means higher. Used widely in ICU benchmarking (APACHE, MPM) and inpatient quality surveillance.
How SMR is calculated
SMR = (observed deaths) / (expected deaths). Expected deaths come from a risk-adjustment model that predicts mortality probability for each patient given diagnoses, severity, age, and other clinical inputs, then sums those probabilities across the cohort. Common SMR models: APACHE IV (ICU), MPM-III (ICU), HCUP risk-adjustment models (inpatient general).
Interpreting SMR
- SMR < 1.0 — fewer deaths than the model predicted; outperforming case-mix-adjusted expectation.
- SMR ≈ 1.0 — performing as expected.
- SMR > 1.0 — more deaths than expected; investigate process-of-care.
Confidence intervals matter. SMR 1.05 with a 95% CI of 0.92-1.21 is not statistically distinguishable from 1.0; SMR 1.05 with CI 1.02-1.08 is.
Where Vizier fits
Vizier computes SMR for ICU and inpatient general populations from EHR-pulled clinical and administrative data using published risk-adjustment models. Continuous calculation (vs. quarterly reports) lets quality and ICU teams identify negative trajectory months before annual benchmarks publish.