Determinants of generalized anxiety and construction of a predictive model in patients with chronic obstructive pulmonary disease.

Abstract

Background: Patients with chronic obstructive pulmonary disease (COPD) frequently experience exacerbations requiring multiple hospitalizations over prolonged disease courses, which predispose them to generalized anxiety disorder (GAD). This comorbidity exacerbates breathing difficulties, activity limitations, and social isolation. While previous studies predominantly employed the GAD 7-item scale for screening, this approach is somewhat subjective. The current literature on predictive models for GAD risk in patients with COPD is limited.

Aim: To construct and validate a GAD risk prediction model to aid healthcare professionals in preventing the onset of GAD.

Methods: This retrospective analysis encompassed patients with COPD treated at our institution from July 2021 to February 2024. The patients were categorized into a modeling (MO) group and a validation (VA) group in a 7:3 ratio on the basis of the occurrence of GAD. Univariate and multivariate logistic regression analyses were utilized to construct the risk prediction model, which was visualized using forest plots. The model's performance was evaluated using Hosmer-Lemeshow (H-L) goodness-of-fit test and receiver operating characteristic (ROC) curve analysis.

Results: A total of 271 subjects were included, with 190 in the MO group and 81 in the VA group. GAD was identified in 67 patients with COPD, resulting in a prevalence rate of 24.72% (67/271), with 49 cases (18.08%) in the MO group and 18 cases (22.22%) in the VA group. Significant differences were observed between patients with and without GAD in terms of educational level, average household income, smoking history, smoking index, number of exacerbations in the past year, cardiovascular comorbidities, disease knowledge, and personality traits (P < 0.05). Multivariate logistic regression analysis revealed that lower education levels, household income < 3000 China yuan, smoking history, smoking index ≥ 400 cigarettes/year, ≥ two exacerbations in the past year, cardiovascular comorbidities, complete lack of disease information, and introverted personality were significant risk factors for GAD in the MO group (P < 0.05). ROC analysis indicated that the area under the curve for predicting GAD in the MO and VA groups was 0.978 and 0.960. The H-L test yielded χ ² values of 6.511 and 5.179, with P = 0.275 and 0.274. Calibration curves demonstrated good agreement between predicted and actual GAD occurrence risks.

Conclusion: The developed predictive model includes eight independent risk factors: Educational level, household income, smoking history, smoking index, number of exacerbations in the past year, presence of cardiovascular comorbidities, level of disease knowledge, and personality traits. This model effectively predicts the onset of GAD in patients with COPD, enabling early identification of high-risk individuals and providing a basis for early preventive interventions by nursing staff.