Table 5.

Examples of possible noninferiority margins for clinical trials of treatments for community-acquired pneumonia.

End point, population	Established lower limit of antibiotic effect,a %	Proposed noninferiority margin, %
Mortality
PSI classes II–V with Streptococcus pneumoniae only	24	10
PSI classes II–V	10	10
PSI classes II–III	8	5
PSI classes III–IV	25	10
PSI classes IV–V	39	10
Defervescence by day 3 (dichotomous)
PSI class I with Mycoplasma pneumoniae only	65	20
PSI class I	35	15
PSI classes II–V	50	20
Composite clinical responseb	Varied	10–20

NOTE. PSI, pneumonia severity index.

^aBased on data reviewed in table 2 and table 3 and in the text.

^bComposite clinical responses could include either time-to-event or dichotomous end points at a specific time point. Data exist to support components, including mortality, defervescence, resolution of cough, resolution of dyspnea, resolution of chest pain, resolution of malaise, and duration of hospitalization. Patient-reported outcome instruments should be considered for clinical response end points. The appropriate patient population and selection of noninferiority margin should be appropriately justified on the basis of available data and the principles outlined in the text above and in International Conference on Harmonisation guidance documents E9 and E10.