Observational studies have reported improved outcomes with salmeterol (SAL) and inhaled corticosteriods (ICS) relative to ipratropium (IPR) and/or no treatment in patients with chronic obstructive pulmonary disease (COPD). Some have suggested that these studies were flawed by immortal-time bias imparted by failure to account for time-dependency of drug exposure. We conducted a retrospective observational study using time-dependent analysis to assess effects of exposure to IPR, SAL, and ICS on mortality in patients with COPD.
We used data from a large US health-insurance claims database to identify patients with an initial (“index”) prescription for IPR, SAL, or an ICS from 9/97-8/00, who also had a prior medical encounter with a primary diagnosis of COPD. Linked mortality information was obtained from the US Social Security Administration and analyzed by Cox proportional hazards regression. Exposure to IPR, SAL, and ICS was characterized by time-dependent variables that were calculated based on dates of prescriptions and therapy-days supplied. Patient characteristics (age, sex, emphysema diagnosis, comorbidities, use of short-acting beta agonists, oral corticosteroids, antibiotics, or oxygen therapy, emergency visits, hospitalizations, and charges) were included in the regression as time-dependent covariates assessed at index date and 90-day intervals post-index.
We identified 12,284 patients with 1,181 deaths over 15,301 person-years of follow-up. Compared with no therapy (969 deaths, 12,428 person-years), exposure to SAL was independently associated with lower mortality (13 deaths, 425 person-years, multivariate hazard ratio [HR]=0.48 [95% CI 0.28-0.83], P=.0092), as was ICS (48 deaths, 1,114 person-years, HR=0.65 [0.48-0.87], P=.0040), but not IPR (176 deaths, 1,896 person-years, HR=0.94 [0.79-1.12], P=.4835). Results were similar in patients without history of asthma.
Therapy with SAL and ICS is associated with improved survival compared with no therapy in patients with COPD; therapy with IPR is not, however. These findings are not due to immortal-time bias.
Results of observational studies alone are insufficient to establish causality. Nevertheless, these results provides further evidence of survival benefits for SAL and ICS in patients with COPD.
T. Delea, GlaxoSmithKline; TD, MH, JE, GO Research Funding,GlaxoSmithKline; RS employment