Twenty-nine studies were included (n=1,656 episodes of PCP). These included eight RCTs, four phase 2 trials, two intensive care unit selections, eight observational cohort studies and seven case series. Only 82 (6.9 per cent) of the patients from the three-centre study were included in analyses.
Treatment response rates were 73 per cent for TMP-SMX and 68 per cent for clindamycin-primaquine, resulting in positive outcomes for second-line therapy (OR 2.1, 95% CI: 1.1, 3.2 for TMP-SMX and OR 2.7, 95% CI: 1.3, 4.0 for clindamycin-primaquine). The response rate for pentamidine was 44 per cent, with an increased risk of second line therapy (OR 0.8, 95% CI: 0.6, 1.0). Atovaquone showed no significant benefit as a second line treatment.
Analysis using data from the 82 patients in the three-centre study showed intravenous pentamidine was associated with a significantly increased risk of death at three months compared to TMP-SMZ (adjusted relative risk 12.4, 95% CI: 4.0, 38.3) and that second-line treatment with pentamidine and TMP-SMZ improved over time.