Wu 2011.
| Methods | A parallel arm cluster‐RCT conducted in Anhui province, Eastern China, between Aug 2000 to Jul 2002. | |
| Participants |
Sample size: 20 clusters (1264 individuals). Clusters: townships were selected and paired according to: place of birth (hospital, family planning centre or other); per capital income; average number of prenatal care visits; and location. Population, proportion of farmers, infant death rate, number of midwives and number of hospital beds were also taken into account. Townships were required to have an existing health facility and the staff necessary to implement the trial. Individuals: women who had given birth in the past year were eligible for the interview. |
|
| Interventions |
Target: health system (health worker education and equipment provision) and community (IEC). Arm 1 (10 clusters, 673 women): the intervention had 3 health system components: training of community midwives, a public awareness campaign with posters and leaflets about prenatal care, and provision of equipment to health centres. Arm 2 (10 clusters, 591 women): usual health system. |
|
| Outcomes |
Trial primary outcomes: prenatal care utilisation and perinatal outcomes. Primary: ANC coverage (at least 4 visits). Secondary: ANC initiation in first trimester, health facility deliveries, stillbirths, perinatal mortality, neonatal mortality. Follow‐up: data were collected from health centre records monthly. Observation in intervention hospitals monthly. Training of midwives involved initial sessions over 2 days and meetings every 3 months. Poster and leaflets in the community throughout trial. Interviews with pregnant women conducted after delivery (mothers of dead infants were not approached for interviews). |
|
| Notes |
Funders: Academy of Finland, Finnish Ministry of Education (DPPH‐program), European Commission INCO Programme "Structural hinders to and promoters of good maternal care in rural China ‐ C HIMACA (015396). Results of a hospital‐based survey are not included in this trial report. We have excluded the perinatal mortality data reported for this trial due to multiple risk of bias concerns, including unclear denominators. |
|
| Risk of bias | ||
| Bias | Authors' judgement | Support for judgement |
| Random sequence generation (selection bias) | Low risk | Sequence generated from a coin toss. |
| Allocation concealment (selection bias) | Unclear risk | 1 township in each matched pair was assigned to intervention or control by a coin toss. Allocation concealment was not described. Matching was checked after randomisation for matching. It was not clear if allocation could be changed. |
| Blinding of participants and personnel (performance bias) All outcomes | Unclear risk | Not blinded. |
| Blinding of outcome assessment (detection bias) All outcomes | Unclear risk | Not blinded. |
| Recruitment bias (for cluster RCTs) | Unclear risk | None noted. |
| Incomplete outcome data (attrition bias) All outcomes | Unclear risk | The denominators for perinatal outcomes were not clear (results expressed at percentages) so it was not possible to assess attrition bias for these outcomes. 1306 women were eligible for the survey, and data were reported for 1264 (missing data ˜ 3%). 42 mothers were described as missing; 2 refused; 27 were out of the village; and 13 cases were missing for "other reasons". |
| Selective reporting (reporting bias) | Unclear risk | Denominators not clearly reported. Mortality stats after the intervention were compared with those before the intervention with major differences reported. |
| Analysis bias | Unclear risk | Adjustments made for clusters, but no information on ICC or what difference adjustment made. ITT not stated. |
| Other bias | High risk | The perinatal data are difficult to interpret due to differences between clusters before the intervention. Data from the community based survey showed group differences for parity, but similarities on other demographic traits. Mortality data were taken from township family planning records. The early neonatal death rate for girls' is much higher than that for boys', causing the authors to doubt the utility of mortality outcomes for the intervention. They wrote, "If the impact of the family planning policy is larger on perinatal mortality than maternal care, then it is hard for any health care intervention to have an effect on perinatal health outcomes". Authors state that the Provincial Health Board implemented a program of ANC in control and intervention townships just 8 months after the trial had begun. 2 intervention districts and 4 control districts also had a prepayment scheme for maternal care implemented during this period. These health initiatives likely contaminated the controls and diluted the effects of the intervention. Furthermore, trialists failed to distribute posters and leaflets because of poor co‐operation between family planning and health sectors, and so this component of the intervention was not completed. |
| Overall risk assessment | High risk | Due to multiple risk of bias concerns above. |
ANC: antenatal care CHW: community health worker Hb: haemoglobin IEC: information, education and communication intervention ICC: intra‐cluster correlation coefficient IPT: intermittent prophylactic treatment ITT: intention‐to‐treat LHW: lady healthcare worker NGO: non‐governmental organisations NICU: neonatal intensive care unit NMR: neonatal mortality rate OR: odds ratio P4P: payment for performance POW: pregnancy outreach worker RCT: randomised controlled trial TBA: traditional birth attendant