Impact of a follow-up visit on two-year mortality after severe exacerbation of COPD in Finland

Background: While guidelines recommend follow-up visits after an acute exacerbation of COPD (AECOPD), the optimal healthcare facility, timing and its effect on long-term mortality are not well known. Methods: In this retrospective case–control study, Finnish nationwide healthcare and mortality registries were linked to determine the association of a follow-up visit in secondary or primary care within 3 and 6 months after severe AECOPD on 2-year mortality. A conditional fixed-effects logistic regression analysis with 23 covariates was used. A nested case–control study design was applied with matching for age, sex, and multimorbidity in a 1:4 ratio. A time-varying Cox sensitivity analysis was performed to test result robustness. Results: In 2018, the Finnish prevalent COPD population consisted of 63 442 patients, of whom 10% experienced an AECOPD. After the AECOPD, 23% of the patients had a follow-up visit within 3 months. A follow-up in primary care wasn’t associated with mortality (OR 1.06, 95% CI 0.90–1.24). However, patients with a follow-up in secondary care were at an increased mortality risk compared with controls without a follow-up (OR 1.62, 95% CI 1.37–1.92). In sensitivity analysis the effect lost statistical significance (HR 1.05, 95% CI 0.35–3.12) but a secondary care follow-up with an add-on primary care follow-up was associated with lower mortality (HR 0.56, 95% CI 0.51–0.61). The strongest predisposing factors were older age, comorbidities such as lung cancer, heart failure and atherosclerosis obliterans, home care, readmission, length of inpatient stay and prior health care use burden. By extending the follow-up time window to 6 months, the results remained largely similar. Conclusion: While we were able to quantify the effect of several patient characteristics on 2-year mortality after AECOPD, the effect of follow-up visits was unclear, likely due to residual confounding by indication. Further studies with even more robust methodology are needed.