Validation of Chronic Obstructive Pulmonary Disease (COPD) Algorithm

Overview

Under-diagnosis of Chronic Obstructive Pulmonary Disease (COPD) is prevalent, with an estimated 75% of adults suffering from COPD in the US, without clinical recognition. Often, the first diagnosis of COPD comes with a flare or exacerbation. In one study, 34% of patients were first diagnosed during hospitalization for an exacerbation, an event associated with a 1-year mortality rate of 26%. When COPD is finally diagnosed, it is often in the late stages, with an average lung function of 50% of normal. Conversely, COPD can be over-diagnosed, defined as symptoms in an individual without airflow obstruction. Over-diagnosed people have significantly higher rates of hospitalization, ER visits, and ambulatory care visits because individuals are treated for a disorder they don't have and are not being treated for the disorder they do have. Lack of diagnostic clarity places patients at risk of medication complications without potential benefit. Conversely, failure to diagnose preempts the benefits of therapy. To date, no reliable solution has been found to address this problem.

Guideline based therapy for COPD includes bronchodilators in conjunction with smoking cessation, vaccination and pulmonary rehabilitation. Bronchodilators have repeatedly been shown to improve symptoms, lung function and quality of life and to reduce exacerbations and possibly mortality in individuals with documented COPD. Despite these findings, the US Preventive Services Task Force recommends against screening for COPD because its value has not been demonstrated in "asymptomatic" individuals. Symptoms such as cough, dyspnea, and wheezing are nonspecific and are common in other disorders such as asthma, kidney failure, and heart disease. Furthermore, COPD is an indolent disease, allowing accommodation to the slowly progressive symptoms. To address this significant under-diagnosis problem, the Global Obstructive Lung Disease (GOLD) Committee recommends "targeted case finding" for patients at risk. Smoking is a risk factor for COPD; however, a quarter of COPD patients are never smokers. The GOLD Guidelines suggest two assessment tools, these tools are designed to detect early symptomatic disease with physician administered patient questionnaires, limiting their use for automated, widespread screening. Risk stratification tools relying entirely on existing, structured Electronic Health Record (EHR) data would provide an opportunity for screening interventions. The most efficient EHR tool would not include symptoms as a predictor variable since the documentation of symptoms are inconsistent and are typically recorded in unstructured text. Automated screening facilitates targeted screening of high-risk patients, including those individuals with limited contact with the health system. This study has created an accurate tool for this purpose that was cross-validated on a potentially biased set of patients who already underwent spirometric testing.