Predictors for mortality due to exacerbation of COPD in primary care: the EXAGGERATE clinical prediction rule derivation dataset
Description
Research hypothesis: Past medical history, symptoms and signs in a person who suffers an acute exacerbation of COPD (AECOPD) that is treated in primary care (PC) allows for predicting if they will die in the short term. Methods: Between December 2013 and November 2014, we studied people aged 40 and over seen for AECOPD in 148 health centres in Spain. Demographic variables, past medical history and signs and symptoms of the patients were collected. A logistic regression model for mortality from any cause was derived 30 days after the last PC visit. Findings: There were 1696 AECOPD included in the analysis. Seventeen patients (1%) died during follow-up. A clinical prediction rule was derived with the exacerbations suffered in the last 12 months, age and heart rate, with an area under the curve operator characteristic of 0.792 (95% confidence interval, 0.692 - 0.891) and good calibration. This rule stratifies patients into three categories of risk and suggests to the physician a different action for each: managing low-risk patients in PC, referring high-risk patients to hospitals and taking other criteria into account for decision-making in patients with medium risk. Interpretation: It is possible to accurately estimate the risk of death due to the exacerbation of COPD without the need to use complex instruments. This rule can help GPs optimize the diagnostic and therapeutic means used in these patients.