To ascertain the relationship between the length of time ICSs were used and mortality, we carried out nested case-control analyses examining a series of time frames. Subjects (n = 503) > 65 years of age who died within 90 to 365 days of hospital discharge were compared to age-matched and gender-matched control subjects who survived to the index date with respect to the timing of exposure to ICSs before the death of the case patient. Those who died (data not shown) had more comorbidities and physician visits, and received more prescriptions for respiratory medications other than ICSs than did control subjects.
In comparison of mutually exclusive time windows of 0 to 30 days, 30 to 60 days, 60 to 90 days, and > 90 days prior to death, the receipt of ICSs < 30 days prior to death was significantly associated with reduced mortality from all causes (Fig 3). Further, ICS receipt within 30 days prior to death was associated with reduced numbers of deaths due to both COPD (adjusted OR, was 0.61; 95% CI, 0.41 to 0.91) and cardiovascular causes (adjusted OR, 0.54; 95% CI, 0.34 to 0.86) [Fig 3]. As Figure 3 shows, mortality associations weakened with longer intervals after the receipt of ICSs.
In order to determine whether the long-term use of ICSs was associated with a reduction in mortality, we repeated the case-control analysis by including only case patients (n = 322) and their control subjects who had survived for 6 months after discharge from the hospital. Those who died between 5 months and 1 year were compared to those who survived with respect to the use of ICSs and other drugs over 6 months preceding death. Again, only the receipt of ICSs within 30 days was associated with a reduced mortality from all causes (adjusted OR, 0.54; 95% CI, 0.38 to 0.75). The main findings of our study were as follows: (1) in COPD patients > 65 years of age, ICS use after hospital discharge was associated with a 25% reduction in all-cause mortality that was not affected by excluding people with a previous diagnosis of asthma or previous ICS use; (2) a substantial reduction in mortality (approximately 50%) was also observed with ICS use in patients aged 34 to 65 years; (3) the reduction in mortality appeared to be largely ascrib-able to reduced cardiovascular mortality and to some extent mortality from COPD; (4) patients treated with ICSs had mortality that was comparable to those treated with neither bronchodilators nor ICSs and was lower than for those treated with broncho-dilators, but not with ICSs; and (5) the effect of ICSs was most evident in the short term, most notably with drug receipt within the preceding 30 days.
Figure 3. The ORs for death in subjects > 65 years of age depending on the timing of ICS use prior to death. ORs with 95% CIs are shown for overall, COPD, and cardiovascular mortality. They were adjusted for the number of prescriptions for P-agonists, ipratropium, oral theophyllines, antimicrobials, and oral corticosteroids that had been dispensed since hospital discharge as well as for age, sex, the number of physician visits in the year prior to hospitalization, and Charlson comorbidity score.