hospital readmission, mortality, older patients
Background: A real-time electronic health record (EHR) predictive model that identifies older patients at risk for readmission and mortality may assist the health care team in improved patient care.
Purpose: This study was performed to generate an automated 30-day readmission and 30-day mortality risk model using data from the EHR in hospitalized older adults.
Methods: This was a retrospective cohort study. Included were patients age 65 years and older admitted to the hospital from July 2012 to December 2013. An automated predictive model was derived from variables collected from the EHR including socioeconomic factors, medical diagnoses and health care utilization. The study sample was randomly divided into derivation (70%) and validation (30%) cohorts. Multiple logistic regression analysis was performed to derive a prediction model. A scoring system was developed for estimating risk of 30-day readmission.
Results: The study included 11,223 patients in one hospital, of which 46% were male, 20% were age > 85 years, 6.2% were black, 60% required emergency admission, 2.8% required an ICU stay and 62.7% were discharged home. Overall 30-day readmission and mortality rates were 13.7% and 1.5%. The risk model predicted 30-day readmission, with c-statistics of 0.62 (95% confidence interval [CI]: 0.61–0.64) and 0.62 (95% CI: 0.60–0.65) in the derivation and validation cohorts, respectively. A readmit risk score was developed that ranged from 0 to 20. The readmission rate increased as the score increased: score 0–4, readmission rate = 8.38%; score 5–9, readmission rate = 13%; and score > 10, readmission rate 20% (P < 0.0001) in the derivation cohort. Results were similar for validation cohort. The risk model predicted 30-day all-cause mortality with c-statistics of 0.81 (95% CI: 0.77–0.86) and 0.73 (95% CI: 0.66–0.81) in the derivation and validation cohorts. The variables associated with mortality included discharge to nursing home, urgent admission status, social worker consultation and diagnoses of respiratory issues and dementia.
Conclusion: A promising automated model generated by EHR data to predict 30-day readmissions and mortality among hospitalized older adults, these findings will be used by the health care system to incorporate a real-time alert into physician workflow. Efforts to improve care will include interventions targeted at the highest-risk group.
Khan A, Hook ML, Singh M, Vollbrecht M, Malsch A, Malone ML. Using an automated model to identify older patients at risk for 30-day hospital readmission and 30-day mortality. J Patient Cent Res Rev. 2015;2:203-204. doi: 10.17294/2330-0698.1221