Effectiveness and safety of misoprostol distributed to antenatal women to prevent postpartum haemorrhage after child-births: A stepped-wedge cluster-randomized trial

N/A

Background: Oral misoprostol, administered by trained health-workers is effective and safe for preventing postpartum haemorrhage (PPH). There is interest in expanding administration of misoprostol by non-health workers, including task-shifting to pregnant women themselves. However, the use of misoprostol for preventing PPH in home-births remains controversial, due to the limited evidence to support self-administration or leaving it in the hands of non-health workers. This study aimed to determine if antenatally distributing misoprostol to pregnant women to self-administer at home birth reduces PPH. Methods: Between February 2013 and March 2014, we conducted a stepped-wedge cluster-randomized trial in six health facilities in Central Uganda. Women at 28+ weeks of gestation attending antenatal care were eligible. Women in the control-arm received the standard-of-care; while the intervention-arm were offered 600mcg of misoprostol to swallow immediately after birth of baby, when oxytocin was not available. The primary outcome (PPH) was a drop in postpartum maternal haemoglobin (Hb) by ≥2g/dl, lower than the prenatal Hb. Analysis was by intention-to-treat at the cluster level and we used a paired t-tests to assess whether the mean difference between the control and intervention groups was statistically significant. Results: 97% (2466/2545) of eligible women consented to participate; 1430 and 1036 in the control and intervention arms respectively. Two thousand fifty-seven of the participants were successfully followed up and 271(13.2%) delivered outside a health facility. There was no significant difference between the study group in number of women who received a uterotonic at birth (control 80.4% vs intervention 91.4%, mean difference=-11.0%, 95% confidence interval [CI] -25.7% to 3.6%, p=0.11). No woman took misoprostol before their baby’s birth. Shivering and fever were 14.9% in the control arm compared to 22.2% in the intervention arm (mean difference=-7.2%, 95%CI -11.1% to -3.7%), p=0.005). There was a slight, but non-significant, reduction in the percentage of women with Hb drop ≥2g/dl from 18.5% in the control arm to 11.4% in the intervention arm (mean difference=7.1%, 95%CI -3.1% to 17.3%, p=0.14). Similarly, there was no significant difference between the groups in the primary outcome in the women who delivered at home (control 9.6% vs intervention 14.5%, mean difference -4.9; 95% CI -12.7 to 2.9), p=0.17). Conclusion: This study was unable to detect a significant reduction in PPH following the antenatal distribution of misoprostol. The study was registered with Pan-African Clinical Trials Network ( PACTR201303000459148 , on 19/11/2012).

Ethical approval was obtained from the School of Medicine Research and Ethics Committee at Makerere University, Kampala, Uganda, and the Uganda National Council for Science and Technology. Permission to carry out the study was obtained from the District Health Office (DHO) and respective in-charges of the health facilities. After information and counselling, eligible women provided written informed consent and received an information sheet in either English or Luganda. The study was registered with the Pan African Clinical Trials Network (PACTR201303000459148) on 19/11/2012. We employed a stepped-wedge cluster-randomized trial design [22] because current evidence on misoprostol use and postpartum haemorrhage would render a placebo-controlled trial unethical [19, 22–25] and all facilities ultimately get the intervention. A cluster was defined as a health facility catchment area. All health facilities started as control-arm facilities. Then in a prior-determined random order, two facilities “crossed over” to become intervention facilities during each of the subsequent three steps for a total of four steps (Fig. 1). Stepped-wedge schema for the trial. Six clusters were enrolled at baseline. The white (non-shaded) cells marked “C” represent the control period. The gray (shaded) cells marked “I” represent the intervention period The random sequence for starting the intervention was determined before the start of the study by using computer generated number sequence. The principal investigator implemented the randomization. Each step lasted for two months, and women were followed up on 3rd to 5th day post delivered. Because of the nature of the intervention, it was not possible to blind the intervention to the care-givers, research team or study participants. The study participants were recruited from six health facilities in Mpigi district, Uganda, between February 2013 and March 2014. The majority of people in the district are of low socioeconomic status, with peasant farming and fishing as their main economic activities. The district health infrastructure consisted of 31 health units (25 government and 6 non-government). These included one private hospital, one Health Centre IV, 13 Health Centre IIIs, and 16 Health Centre IIs. The district recorded a skilled birth attendant rate of 30 % (2010–11 District Annual Report), although the national average was 58 % [21]. We enrolled study participants at the antenatal clinic of the Health Centre IV and the five Health Centre IIIs. Two health facilities held their antenatal clinics from Monday to Friday, while the other four had two dedicated antenatal-care service days per week. Maternity services at the Health Centre IIIs were staffed by midwives, while the Health Centre IV had three medical officers in addition to the 7 midwives, and provided comprehensive emergency obstetric care. The staff in the antenatal clinic and delivery wards were involved in recruiting and following up the women, which allowed the intervention to be delivered as part of ongoing maternity care. Clinics: All the 31 health facilities were screened for eligibility. The eligibility criteria were a) that a minimum of 50 pregnant women attended antenatal clinic (for the first time) in the month prior to the start of the study, and b) that the person in-charge of health facility agreed for the facility to participate. Of the nine facilities that registered a minimum of 50 antenatal first-time attendees per month, we excluded three; the hospital because health-care services are paid for (private hospital); one health facility because its in-charge declined to participate; and another because it was difficult to access the women after home birth during the wet season due to seasonal rivers and swamps. Women: Within the antenatal clinics of participating health facilities, we included all pregnant women who were 28 weeks or more of gestation, and who had no plans to leave the district during pregnancy delivery or in the immediate postpartum period. We excluded women who had a planned elective caesarean-section delivery or previous caesarean section scars. Study staff briefed pregnant women attending the antenatal clinic about the study objectives and design of the study in a group. The key messages to the pregnant women included: 1. The benefits of delivering at the health facility, 2. Excessive bleeding after child-birth was dangerous to a woman’s life, 3. The availability of an effective drug (oxytocin) to stop excessive bleeding that can be given by trained health worker at the time of birth in the health facility. 4. For those willing to participate, the need to alert the research assistant by telephone when and where the delivery occurred. We repeated the sessions about the study in every antenatal clinic throughout the study period (in both control and intervention phases). After the discussion, we invited those eligible to participate. Each participant gave a written informed consent. At the time of the trial, the standard-of-care was that a women who delivered at a health facility should receive oxytocin to prevent PPH, while women who delivered at home received no uterotonic. Women in the intervention period were given 600 micrograms (mcg) of oral misoprostol at enrolment to the study to self-administer after childbirth if delivery happened outside a health facility, or when there was no oxytocin at the health facility. The three tablets of misoprostol (200 mcg each for a total of 600mcg) in aluminum foil were packaged in a plastic envelope. Women were given the following instructions; “1. Not to take the misoprostol tablets when the baby is still inside the womb, 2. To swallow all the three tablets immediately after the birth of the baby, if delivery occurred at home, or if no oxytocin was given by the health provider. If she had twins, she was to swallow the tablets after the birth of second twin. 3. To keep the packaging of tablets (foil) after swallowing them and to give it to research assistant when she visits her. 4. To carry along the study tablets (misoprostol) when going to deliver at a health facility. Hand the misoprostol tablets to the attending midwife or research assistant if delivery occurred at health facility.” The research assistants and health facility staff in the antenatal clinics and delivery wards from the six study facilities were trained on the protocol for 5 h. This comprised of study material and key messages to women attending antenatal clinic, and was delivered by the principal investigator. Weekly supervisory visits by the principal investigator followed the initial training and further training was given as requested or as assessed by the principal investigator. The study participants were interviewed face-to-face by a trained research assistant. We used a pre-tested questionnaire to collect socio-demographic characteristics including maternal age, education, marital status, maternal occupation and religious affiliation. We also inquired about parity, gestation at first antenatal visit, the use of prophylactic anti-malarials, transport costs to the health facility for antenatal care, and delivery plans. We established gestational age from the woman’s last normal menstrual period (LNMP) or ultrasound scan estimation. In a few cases where we did not have LNMP or an ultrasound scan, we used fundal height to approximate the gestational age [26]. Trained research assistants measured haemoglobin (Hb) levels at enrolment (during their third trimester antenatal care visit) and three to five days after delivery using a portable HemoCueR Hb 301 system. as described in another part of the study that looked at haemoglobin status of pregnant women [27]. All pregnant women enrolled in the study continued receiving standard antenatal care at the local health facility. A sticker identifying them as enrolled study participants was placed on their hand-held antenatal cards to make it easier to identify them at repeat antenatal visits or when they reported in labour. The sticker had three telephone numbers that the women could call to contact the study team once they had delivered or if they had any problems or questions about study. At enrolment, women were advised to deliver in a health facility as per national guidelines. They were also advised to seek care in case they had excessive bleeding after child birth, the placenta had not delivered within one hour, or the baby did not cry immediately after birth or developed a fever. The study kept a log of participants’ names, contact telephone numbers and the name of the village health worker where they lived. The study team contacted any woman who had passed her estimated delivery date to identify if she had given birth and from where. Midway through the study, we observed that the names of participants in the log book were often not what the women were called by the community members in the village, so we subsequently modified our procedures to ask participants for the names (petty names) the community members usually called them. Research assistants visited the woman either at home or at the health facility after birth to measure the haemoglobin and complete postnatal questionnaire. Participants were defined as lost-to-follow-up when we were unable to physically contact them eight weeks after the expected date of delivery. The primary outcome was PPH, defined as a drop in maternal haemoglobin by 2g/dl or more, lower than the prenatal Hb [10]. Secondary outcomes were: postpartum anaemia defined as Hb < 11 g/dl when assessed within 7 days and Hb < 12 g/dl if assessed after the 7th day after childbirth [28], place of child-birth, use of any uterotonics for prevention of PPH, referral to a health facility after delivery, blood transfusion and maternal death. We asked the women about side effects related to misoprostol use, such as fever (self-report of body feeling hot), chills, shivering and how they coped with them. Safety was defined as swallowing of the medicine after delivery of the baby or babies. Specific to the intervention group, we also assessed the timing of swallowing misoprostol, and its acceptability to women. We asked the women in the intervention group to keep the blister package of the misoprostol (used or unused) and hand it to research assistant at home during the follow up visit or to the nurse at the health facility where the woman delivered. Sample size was calculated taking into account the clustering effect. We assumed a between cluster correlation coefficient km = 0.2, a minimum of 200 pregnant women per health facility in each phase (m), and proportion experiencing PPH of 12.0 % [9]. Assuming 80 % power to detect a difference of 50 % in PPH proportions between the two groups with a type I error of 5 %, using formula for matched cluster trial [29], the study needed six health facilities in each arm, Analysis was conducted on an intention-to-treat principle, based on the period (intervention or control) at which women were enrolled into the study. We compared the characteristics of women enrolled in the control and the intervention periods at individual level and these were summarized as percentages for categorical outcomes, and means (and standard deviations) for continuous outcomes. Cluster-level summaries of some women’s characteristics in the control and intervention periods were computed and presented as means and standard deviations. The primary outcome (postpartum Hb ≥ 2g/dl lower than prenatal Hb) in each arm was expressed as the mean of the six cluster-level proportions of women who had PPH measured in each cluster in intervention and control periods respectively. The effect of the intervention was measured as the difference in the means of primary outcome of the two groups (with 95 % confidence interval [CI]). We used paired t-tests to assess whether the mean difference between the two groups was statistically different from zero. We also applied a paired t-test to assess the difference in the means of the secondary outcomes between the control and intervention groups. Uterotonic use included a summary statistic of uterotonic received at birth and was cross-tabulated with place of birth. The acceptability of misoprostol as a number of home births who ingested misoprostol and were willing to use it next pregnancy or recommend it to relative. Results were summarized as frequency distribution. Because the study did not have a lag phase, some of women recruited during the control period delivered in the intervention period.

• Dima Health