Table 4—

Sensitivity analyses based on exposure to formoterol, and exclusion of subjects taking maintenance salmeterol treatment

All deathsAsthma deaths
Simple contingency two-by-two table method
 Formoterol deaths#24/27,363 (0.09)6/27,363 (0.02)
 Comparator deaths#12/19,045 (0.06)2/19,045 (0.01)
 OR (95% CI)1.4 (0.7–3.1)2.1 (0.4–21.2)
Peto’s method
 Studies with data n125
 OR (95% CI)
  Fixed effects1.4 (0.7–2.7)2.2 (0.5–8.8)
  Random effects1.4 (0.7–2.7)2.2 (0.5–8.8)
Bayesian method
 Relative risk (95% credible interval)
  Vague prior1.2 (0.6–2.9)1.8 (0.3–12.3)
  Informative prior1.3 (0.9–1.9)3.1 (1.2–8.8)
  • #: presented as deaths/total subjects (%). : the prior used for total deaths was the confidence interval for the odds ratio for all deaths of 0.8–2.1 from SMART (Salmeterol Multicentre Asthma Research Trial); for asthma deaths the confidence interval 1.2–15.3 was used.