Table 3.
Mortality among patients aged <30 years or with “good” statusa |
Mortality among patients aged 30–59 years or with “fair” statusa |
Mortality among patients aged ≥60 years or with “poor” statusa |
|||||||
---|---|---|---|---|---|---|---|---|---|
Study | No antibiotics | Antibiotic treatment |
Absolute reductionb (95% CI), % |
No antibiotics | Antibiotic treatment |
Absolute reductionb (95% CI), % |
No antibiotics | Antibiotic treatment |
Absolute reductionb (95% CI), % |
Tilghman and Finland [37] | 32/301 (11) | NA | 154/563 (27) | NA | 96/140 (69) | NA | |||
Bullowa [38] | 103/739 (14) | NA | 371/1090 (34) | NA | 80/135 (59) | NA | |||
Heinztelman et al. [28] | NA | NA | 5/7 (71) | 0/7 (0) | 3/3 (100) | 2/2 (100) | |||
Evans and Gaisford [31] | 6/34 (18) | 1/51 (2) | 18/52 (35) | 6/44 (14) | 3/4 (75) | 1/5 (20) | |||
Dowling and Lepper [26] | 23/287 (8) | 3/226 (1) | 199/613 (32) | 24/565 (4) | 109/187 (58) | 20/129 (16) | |||
Weighted average | 164/1361 (12) | 4/277 (1) | 11 (8–13) | 747/2325 (32) | 30/616 (5) | 27 (25–30) | 291/469 (62) | 23/136 (17) | 45 (39–54) |
NOTE. Data are no. of patients who died/total no. of patients (%), unless otherwise indicated. NA, not available.
Some studies scored the baseline status of the severity of illness as “good” (i.e., pneumonia severity index [PSI] classes II–III equivalent), “fair” (i.e., PSI classes III–IV equivalent), or “poor” (i.e., PSI classes IV–V equivalent); if scoring was not available, age was used.
The reduction in mortality is summarized across the studies as the difference between the group receiving antibiotics and the group not receiving antibiotics, weighted by the numbers of patients. The 95% CIs were calculated using standard linear combination variance formulas [35]. This method allows inclusion of 1-arm studies and nonrandomized 2-arm studies, which is not possible with meta-analytic techniques [36]. One small trial (by Heinzelman et al. [28]; n = 19) was not included in the summary statistics because of the 0% and 100% mortality rates among some subgroups. Also, the large, historical control study by Finland [25] was not included, because he reported only the percentage mortality according to age and did not list the numerator and denominator of patients in each age group.