. Author manuscript; available in PMC: 2017 Mar 1.

Published in final edited form as: Contemp Clin Trials. 2016 Jan 16;47:185–195. doi: 10.1016/j.cct.2016.01.003

Table 2.

Statistical models and power for respiratory outcomes (Section 2.10 and 2.11)

Outcome	Model	Assumptions	Detectable Effect (80% Power)
Pneumonia	Poisson regression	Incidence 2/100 person-years; 10,000 per arm	1.3/100 person-yrs
Respiratory exacerbation episodes	Generalized linear model with negative binomial response	Exacerbation rate ~ 1/year; 1200 per arm with COPD	0.062 (rate ratio of 95%, treated to untreated)
Change in percent-predicted pulmonary function (e.g., FEV₁) between baseline and follow-up	Linear regression	Standard deviation of 230 ml for a 3 year change; 962 per arm	29 ml
Asthma control score^a	Generalized linear model with negative binomial response	Standard deviation of change over 6 months of 0.87; 340 per arm	0.19 in score

Secondary outcome.