Clinical Interventions in Aging (Jun 2016)
Prediction of critical illness in elderly outpatients using elder risk assessment: a population-based study
Abstract
Michelle Biehl,1 Paul Y Takahashi,2 Stephen S Cha,3 Rajeev Chaudhry,2 Ognjen Gajic,1 Bjorg Thorsteinsdottir2 1Division of Pulmonary and Critical Care Medicine, Department of Medicine, 2Division of Primary Care Internal Medicine, 3Health Sciences Research, Mayo Clinic, Rochester, MN, USA Rationale: Identifying patients at high risk of critical illness is necessary for the development and testing of strategies to prevent critical illness. The aim of this study was to determine the relationship between high elder risk assessment (ERA) score and critical illness requiring intensive care and to see if the ERA can be used as a prediction tool to identify elderly patients at the primary care visit who are at high risk of critical illness. Methods: A population-based historical cohort study was conducted in elderly patients (age >65 years) identified at the time of primary care visit in Rochester, MN, USA. Predictors including age, previous hospital days, and comorbid health conditions were identified from routine administrative data available in the electronic medical record. The main outcome was critical illness, defined as sepsis, need for mechanical ventilation, or death within 2 years of initial visit. Patients with an ERA score of 16 were considered to be at high risk. The discrimination of the ERA score was assessed using area under the receiver operating characteristic curve. Results: Of the 13,457 eligible patients, 9,872 gave consent for medical record review and had full information on intensive care unit utilization. The mean age was 75.8 years (standard deviation ±7.6 years), and 58% were female, 94% were Caucasian, 62% were married, and 13% were living in nursing homes. In the overall group, 417 patients (4.2%) suffered from critical illness. In the 1,134 patients with ERA >16, 154 (14%) suffered from critical illness. An ERA score ≥16 predicted critical illness (odds ratio 6.35; 95% confidence interval 3.51–11.48). The area under the receiver operating characteristic curve was 0.75, which indicated good discrimination. Conclusion: A simple model based on easily obtainable administrative data predicted critical illness in the next 2 years in elderly outpatients with up to 14% of the highest risk population suffering from critical illness. This model can facilitate efficient enrollment of patients into clinical programs such as care transition programs and studies aimed at the prevention of critical illness. It also can serve as a reminder to initiate advance care planning for high-risk elderly patients. External validation of this tool in different populations may enhance its generalizability. Keywords: aged, prognostication, critical care, mortality, elder risk assessment
