Results
Characteristics of Patients With Undiagnosed COPD
The entire PAC-COPD cohort included 342 patients (93% men) with a mean (SD) age of 67 (9) years and a mean (SD) post-bronchodilator FEV1 of 52% (16%) predicted during clinical stability (Table 1). A total of 117 patients (34%) fulfilled the criteria of "undiagnosed COPD". Table 1 shows the comparisons of sociodemographic and clinical characteristics for these two groups. Undiagnosed patients were younger and more physically active, had fewer symptoms and better health status, and had milder airflow limitation and fewer comorbidities; in addition a higher proportion of these patients reported that they currently smoked (Table 1). A total of 33 (28%) patients with severe COPD and 5 (4%) patients with very severe COPD reported that they had never been diagnosed as having a respiratory disease prior to their first hospitalisation. The Charlson comorbidities are shown in Additional file 1: Table S2 http://www.biomedcentral.com/1471-2466/15/4/additional.
Undiagnosed patients reported a significantly lower use of health care resources due to respiratory symptoms in the 12 months prior to their first hospitalisation for a COPD exacerbation. The number of unscheduled visits to the primary care surgery was similar in both groups (Table 2).
Short-term Effects Associated With a COPD Diagnosis
Figure 2 shows the short-term effects associated with a COPD diagnosis on smoking cessation. The proportion of current smokers after hospital discharge decreased significantly more in newly diagnosed COPD patients than in those with a previous COPD diagnosis (16% vs. 5%). Despite significantly different baseline values at hospitalisation (Figure 2), the interaction between diagnosis group and time was significant (p = 0.019).
(Enlarge Image)
Figure 2.
Short-term effects of a new COPD diagnosis on smoking cessation. P-values were obtained from a logistic regression model with active smoking as the outcome and the interaction between diagnosis status and time (period) included as explanatory variables. For further explanations, see the main manuscript text.
Long-term Prognosis of Newly Diagnosed COPD Patients
During a mean (SD) of 1.87 (0.98) years of follow-up, 44% of previously diagnosed patients and 28% of newly diagnosed required re-hospitalisation. This corresponds to 0.25 and 0.14 annual hospitalisation rates (p < 0.01), respectively (Figure 3, panel A). However, this risk of re-hospitalisation was similar in both groups after adjusting for other covariates in a Cox regression multivariate model (Table 3). The proportion of patients who required admission was higher in previously diagnosed patients when compared with newly diagnosed patients for the mild, moderate and severe spirometric COPD groups (20% vs. 7%, 36% vs. 23% and 49% vs. 36%, respectively). The proportion of patients within the very severe COPD group who required admission was 63% in previously diagnosed patients and 100% for newly diagnosed patients; however, the very small sample size prevented any statistical comparisons.
(Enlarge Image)
Figure 3.
Kaplan-Meier curves show the cumulative hospitalisation-free rate (panel A) and survival rate (panel B) according to previous COPD diagnosis.
During a mean (SD) of 3.28 (0.85) years, overall survival rates (Figure 3, panel B) of previously diagnosed and newly diagnosed patients were similar (87% and 84%, respectively; p = 0.51) at all severity stages (80% and 93% in mild, 92% and 85% in moderate, 87% and 81% in severe, and 64% and 60% in very severe patients).