Twelve RCTs (2,731 patients with HAP) were included.
The mean quality score was 2.42 out of 5. Five RCTs were of a high quality and seven were of a low quality.
Mortality.
Carbapenems resulted in significantly fewer deaths in comparison with fluoroquinolones or beta-lactams, alone or in combination with aminoglycosides (OR 0.72, 95% CI: 0.55, 0.95; 7 RCTs). There were no significant differences between carbapenems and fluoroquinolones or beta-lactams when only higher quality trials were included in the analysis. No significant statistical heterogeneity was reported.
Treatment success in ITT and CE patients.
There were no significant differences between carbapenems and fluoroquinolones or beta-lactams in terms of treatment success in ITT (8 RCTs) or CE patients (11 RCTs; significant statistical heterogeneity was reported). Similarly, sensitivity analyses showed no significant differences when only studies of late onset HAP were included (3 RCTs); when carbapenems were compared with beta-lactams, alone or in combination with aminoglycosides (9 RCTs; significant statistical heterogeneity reported); and when imipenem was compared with fluoroquinolones (3 RCTs) or beta-lactams (5 RCTs). However, meropenem was associated with significantly greater treatment success in comparison with beta-lactam/aminoglycoside (OR 2.30, 95% CI: 1.36, 3.91; 4 RCTs). In contrast, when only HAP due to Pseudomonas aeruginosa was considered, carbapenems were associated with significantly lower treatment success in comparison with fluoroquinolones and other beta-lactams, alone or in combination with aminoglycosides (OR 0.42, 95% CI: 0.22, 0.82; 6 RCTs).
Treatment success in microbiologically evaluable patients.
No significant differences between carbapenems and fluoroquinolones or beta-lactams, alone or in combination with aminoglycosides, were evident for treatment success in microbiologically evaluable patients overall (10 RCTs) or in the eradication of Acinetobacter baumannii (5 RCTs), Klebisella pneumoniae (4 RCTs) or Staphylococcus aureus (6 RCTs). However, carbapenems were associated with significantly fewer treatment successes in the treatment of HAP due to Pseudomonas aeruginosa (OR 0.5, 95% CI: 0.24, 0.89; 7 RCTs). No significant statistical heterogeneity was reported.
Development of resistance.
The development of resistance in Pseudomonas aeruginosa was significantly increased in patients treated with carbapenems in comparison with those treated with fluoroquinolones or beta-lactams, alone or in combination with aminoglycosides (OR 5.17, 95% CI: 1.96, 13.65; 4 RCTs). No significant statistical heterogeneity was reported.
Adverse effects.
Adverse effects were mainly mild to moderate in severity and mainly involved the gastrointestinal tract (nausea, vomiting and diarrhoea). No significant differences between carbapenems and fluoroquinolones or beta-lactams, alone or in combination with aminoglycosides, were evident in the incidence of total (5 RCTs) or drug-related (3 RCTs) adverse effects, gastrointestinal effects (4 RCTs), the number of seizures (5 RCTs), or the number of withdrawals due to drug-related adverse effects (3 RCTs). No significant statistical heterogeneity was reported.
Publication bias.
Egger’s test results suggested that smaller studies favoured carbapenems in the analysis of treatment success in ITT and CE patients, and that there was a significant risk of publication bias in these analyses.