Michelle Bentivegna, MPH; Cassie Hulme, MPH; Mark H. Ebell MD, MS
Corresponding Author: Mark H. Ebell MD, MS; University of Georgia. Email: ebell@uga.edu
Section: Original Research
| FULL PDF |
Background: The aim of this systematic review is to summarize the best available evidence regarding individual risk factors, simple risk scores, and multivariate models that use patient characteristics, vital signs, comorbidities, and laboratory tests relevant to outpatient and primary care settings. Methods: Medline, WHO COVID-19, and MedRxIV databases were searched; studies meeting inclusion criteria were reviewed in parallel and variables describing study characteristics, study quality, and risk factor data were abstracted. Study quality was assessed using the Quality in Prognostic Studies tool. Random effects meta-analysis of relative risks (categorical variables) and unstandardized mean differences (continuous variables) was performed; multivariate models and clinical prediction rules were summarized qualitatively. Results: 551 studies were identified and 22 studies were included. The median or mean age ranged from 38 to 68 years. All studies included only inpatients, and mortality rates ranged from 3.2% to 50.5%. Individual risk factors most strongly associated with mortality included increased age, c-reactive protein (CRP), d-dimer, heart rate, respiratory rate, lactate dehydrogenase (LDH), and procalcitonin, as well as decreased oxygen saturation, the presence of dyspnea, and comorbid coronary heart and chronic kidney disease. Independent predictors of adverse outcomes reported most frequently by multivariate models include increasing age, increased CRP, decreased lymphocyte count, increased LDH, elevated temperature, and the presence of any comorbidity. Simple risk scores and multivariate models have been proposed, but are often complex and most have not been validated. Conclusions: Our systematic review identifies several risk factors for adverse outcomes in COVID-19 infected inpatients that are often available in the outpatient and primary care settings: increasing age, increased CRP or procalcitonin, decreased lymphocyte count, decreased oxygen saturation, dyspnea on presentation, and the presence of comorbidities. Future research to develop clinical prediction models and rules should include these predictors as part of their core dataset to develop and validate pragmatic outpatient risk scores.