Table 3 Risk of bias of quantitative studies based on the EPOC tool
Study/year | Selection bias | Allocation to intervention | Performance bias | Baseline differences in characteristics | Baseline differences in outcomes | Contamination | Attrition bias | Selective reporting | Other bias |
|---|---|---|---|---|---|---|---|---|---|
Randomized controlled trials (RCTs) | |||||||||
1. Hobel et al. [25] | Low: Cluster randomization with a restricted block | Low: cluster randomization. | Low: women were not aware of their intervention status nor the clinics teams | Low: participants were comparable with respect to age, marital status, gravidity, parity and preterm birth rate | Low: no difference between the groups with regards to high risk preterm problems at baseline | Low: intervention was provided on a clinic-basis rather than patients | Unclear: the number of women who left the study was not reported | Low | low |
2. Main et al. [22] | High: a random numbers table was used for the first 479 participants then the second sample of464 women was divided into groups by birthday date | High: women’s’ date of birth was used to allocate women to intervention or control | Low: In the control group neither the doctors nor the women were made aware they were at high risk of preterm birth. | Low: no differences between groups with respect to maternal age, gravidity, parity, previous preterm deliveries, and gestational age at first visit | High: More women with previous preterm birth were assigned to the intervention. | High: 8 women from the control group transferred to the clinic | High: insufficient reporting on the rate of attrition. | Unclear | Low |
3. Iams and Johnson [21] | Unclear | Unclear | Unclear | Unclear | Unclear | Unclear | Unclear | Unclear | Unclear |
4. Goldenberg et al. [24] | Low: randomization by a randomization officer | Unclear: allocation to intervention or control method was not reported | Low: nobody was aware of the intervention status | Low: No significant difference between high-risk group and high-risk control with regard to number of birth, race, age and parity | Low: no differences of previous preterm birth and multiple pregnancies between the two groups | low | Unclear: the number of missing women was not reported | Low | Low |
5. Mueller-Heubach [23] | Low: participants were selected to intervention and controlled randomly | Unclear: method of allocation to intervention or control was not reported | High: nurses were aware of the intervention status of the participants | Low | Low | High: a historical control was used in the analysis | Unclear | low | low |
Cohort studies | |||||||||
1. Herron et al. [26] | High | High | High | Low | Unclear | Low | Low | Low | Low |
2. Manuck et al. [27] | High | High | High | Low | Low | Low | Low | Low | Low |
3. Karkhanis et al. [28] (abstract) | High | High | High | Unclear | Unclear | Low | Unclear | Unclear | Unclear |
4. Burul [29] (abstract) | High | High | High | unclear | Unclear | Low | Unclear | Unclear | Unclear |
5. Cohen et al. [30] (abstract) | High | High | High | Unclear | Unclear | Low | Unclear | Unclear | Unclear |