Table 3 Risk of bias in included studies
Risk of bias | Assessment of bias in studies included in the lit review | General description |
|---|---|---|
Random allocation of intervention | Yes (1–4) | Four studies randomized the intervention at a facility/district level, whereas the other three studies did not randomize allocation of intervention |
No (5–7) | ||
Not clear | ||
Not applicable | ||
Concealment of allocation | Yes | No studies concealed assignment. This is expected given the nature of health systems interventions |
No (1–7) | ||
Not clear | ||
Follow-up of professionals | Yes (7) | Out of the six studies where follow-up was possible, four studies did not have the adequate 80–100% follow-up after randomization outlined by EPOC. In the other two studies, follow-up was not described |
No (1,3,4) | ||
Not clear (2,5) | ||
Not applicable (6) | ||
Blinded assessment of primary outcome(s) | Yes | None of the studies had blinded assessment of the primary outcome, as it is not possible given the nature of health systems interventions |
No (1–7) | ||
Not clear | ||
Not applicable | ||
Baseline measurement of outcomes in both groups | Yes (4,5) | Two studies included baseline measurement of outcome and differences between groups, three did not, and in two cases it was not applicable |
No (1–3) | ||
Not clear | ||
Not applicable (6,7) | ||
Baseline measurement of characteristics in both groups | Yes (4) | Most studies had unclear, or did not describe, baseline measurements of characteristics in both groups, with some reporting pooled numbers or not providing differences. One article did |
No (1,3) | ||
Not clear (2,5) | ||
Not applicable (6,7) | ||
Reliable primary outcome measure(s) | Yes | No studies presented an inter-rater reliability measure |
No (1–7) | ||
Not clear | ||
Not applicable | ||
Protection against contamination | Yes (1,3,4,6) | Almost all studies had an element of randomization, with randomization done at a facility or district level. These processes help reduce risk contamination. In the remaining two cases with comparison groups, contamination may have been compromised due to other external programs and district pairing |
No | ||
Not clear (2,5) | ||
Not applicable (7) | ||
No selective outcome reporting | Yes (1–5, 7) | Most studies reported both significant and non-significant findings. In one study, it was not clear whether there was selective reporting |
No | ||
Not clear (6) | ||
Not applicable |