Heart rate variability, a marker of autonomic function, has shown promising prognostic results in specific populations, but has not been tested in a general medical population. We hypothesized that heart rate variability identifies high-risk medical patients early after admission to the hospital.

This was a single-center prospective cohort study of acutely admitted medical patients aged ≥18 years with a life expectancy ≥3 months, included between 2019-2023. Unstable patients needing direct admission to the intensive care unit were excluded. Heart rate variability was recorded within 24 hours of admission for 10 minutes. The standard deviation of normal-normal beats (SDNN) was the primary heart rate variability marker. Low SDNN was defined as the lowest tertile (≤22 ms). The primary outcome was 30-day all-cause mortality. The secondary outcome was 30-day readmission or mortality.

Among 721 patients included, low SDNN carried an 8-fold greater risk of 30-day mortality in univariate analysis (hazard ratio [HR] 8.3; P = .001); in multivariate analyses a 4-fold greater risk (HR 3.8; P = .037). Low SDNN was associated with the combined outcome of 30-day mortality or readmission (HR 1.5; P = .03) in multivariate analysis. In receiver operating characteristics analyses, low SDNN improved the predictive accuracy of early warning score for 30-day mortality or readmission from 0.63 to 0.71 (P = .008) but did not improve the accuracy for 30-day mortality alone.

In patients admitted due to acute medical illness, low heart rate variability predicted 30-day mortality and readmission, suggesting heart rate variability as a tool to identify patients at high and low risk of relevant endpoints.