Abstract
Objectives. To develop and validate a screening tool designed to identify detained people at increased risk for COVID-19 mortality, the COVID-19 Inmate Risk Appraisal (CIRA). Design. Cross-sectional study with a representative sample (development) and a case-control sample (validation).Setting. The two largest Swiss prisons.Participants. (1) Development sample: all male persons detained in Pöschwies, Zurich (n=365); (2) Validation sample: case-control sample of male persons detained in Champ-Dollon, Geneva (n=192, matching 1:3 for participants at risk for severe course of COVID-19 and participants without risk factors).Main outcome measures. The CIRA combined seven risk factors identified by the World Health Organization and the Swiss Federal Office of Public Health as prognosis of severe COVID-19 to derive an absolute risk increase in mortality rate: Age ≥60, cardiovascular disease, diabetes, hypertension, chronic respiratory disease, immunodeficiency, and cancer. Results. Based on the development sample, we proposed a three-level classification: average (<3.7), elevated (3.7-5.7), and high (>5.7) risk. In the validation sample, the CIRA identified all individuals considered vulnerable by national recommendations (having at least one risk factor). The category “elevated risk” maximized sensitivity (1) and specificity (.97). The CIRA had even higher capacity in discriminating vulnerable individuals according to clinical evaluation (a four-level risk categorization based on a consensus of medical staff). The category “elevated risk” maximized sensitivity and specificity (both 1). When considering the individuals classified as extremely high risk by medical staff, the category “high risk” had a high discriminatory capacity (sensitivity=.89, specificity=.97). Conclusions. The CIRA scores have a high discriminative ability and will be important in custodial settings to support decisions and prioritize actions using a standardized valid assessment method. However, as knowledge on risk factors for COVID-19 mortality is still limited, the CIRA should be considered preliminary. Underlying data will be updated regularly on the website www.prison-research.com, where the CIRA algorithm is freely available.