Predictors for mortality due to acute exacerbation of COPD in primary care: Derivation of a clinical prediction rule in a multicentre cohort study

Abstract

Background

In primary care (PC), 80% of the acute exacerbations of chronic obstructive pulmonary disease (AECOPD) are treated. However, no predictive model has been derived or validated for use in PC to help general practitioners make decisions about these patients.

Objectives

To derive a clinical prediction rule for mortality from any cause 30 days after the last PC visit.

Methods

Between December 2013 and November 2014, we performed a cohort study with people aged 40 and over who were treated for AECOPD in 148 health centres in Spain. We recorded demographic variables, past medical history, signs, and symptoms of the patients and derived a logistic regression model.

Results

In the analysis, 1,696 cases of AECOPD were included and 17 patients (1%) died during follow-up. A clinical prediction rule was derived based on the exacerbations suffered in the last 12 months, age, and heart rate, displaying an area under the receiver operating characteristic curve of 0.792 (95% confidence interval, 0.692–0.891) and good calibration.

Conclusion

This rule stratifies patients into three categories of risk and suggests to the physician a different action for each category: managing low-risk patients in PC, referring high-risk patients to hospitals and taking other criteria into account for decision-making in patients with moderate risk. These findings suggest that it is possible to accurately estimate the risk of death due to AECOPD without complex devices. Future studies on external validation and impact assessment are needed before this prediction rule may be used in clinical practice.

Keywords:

• Asthma/COPD
• prognosis/prognostic research
• general practice/family medicine
• general
• multivariate analysis
• incl. modelling

Acknowledgement

The authors thank Daniel Araúzo Palacios (Former Director of the Department of Clinical Assistance and Applications, Castile-Leon Regional Health Authority) for his critical review of the study proposal, Pilar Lorenzo Lobato (Department of Information Systems and Health Outcomes, Castile-Leon Regional Health Authority) for her assistance with data extraction, and Sara Calvo Simal (Research Unit, Burgos University Hospital) for her assistance with the statistical analysis.

Disclosure statement

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

Additional information

Funding

This work was supported by the Spanish Society of Family and Community Medicine (semFYC) under [Grant ‘Isabel Fernández 2013’]; Castile-Leon Regional Health Authority under [Grant GRS 1009/B/14].