Thirteen observational studies were included: 12 cohort (n=22,295) and one case-control (n=132). Sample sizes ranged from 96 to 13,771.
Quality (10 cohort studies)
Four studies were prospective (three adjusted for pneumonia severity) and six were retrospective (three adjusted for pneumonia severity). Only three of the retrospective studies adhered to more than two of the eight published chart review standards.
Mortality rate
Four-hour cut-off (seven studies). Findings varied. Two of the seven studies (one prospective without severity adjustment and one retrospective) reported that antibiotics within four hours were associated with a statistically significant reduction in mortality. The reduction ceased to be statistically significant in the prospective study after adjustment for mental status and other variables. Post-hoc analysis of the only prospective study with adjustment for severity (conducted by the review authors) found a statistically significant association between early antibiotic administration and higher mortality (OR 1.99, 95% CI: 1.22 to 13.45, n=409). A possible relationship between study design and the magnitude of observed results was noted overall, with lower-quality studies more likely to indicate a survival benefit associated with early administration.
Eight-hour cut-off (four studies). One study (retrospective without severity adjustment) reported that early antibiotics were associated with a statistically significant reduction in mortality. The other three studies (including one prospective study without severity adjustment) reported no statistically significant difference between the treatment groups.
Results of three additional retrospective studies reporting length of hospital stay were inconsistent.