Expanding uterotonic protection following childbirth through community-based distribution of misoprostol: Operations research study in Nepal

Article Outline

• Abstract
• 1. Introduction
• 2. Materials and methods
  • 2.1. The study setting
  • 2.2. The intervention
  • 2.3. The data
    • 2.3.1. Household surveys
    • 2.3.2. Project management information system
  • 2.4. Outcome variables
  • 2.5. Statistical analysis
• 3. Results
  • 3.1. Feasibility
  • 3.2. Institutional delivery
  • 3.3. Acceptability
  • 3.4. Safety
• 4. Discussion
• 5. Misoprostol Study Group members
• Acknowledgements
• References
• Copyright

Abstract 

Objective

To determine feasibility of community-based distribution of misoprostol for preventing postpartum hemorrhage (PPH) to pregnant woman through community volunteers working under government health services.

Methods

Implemented in one district in Nepal. The primary measure of performance was uterotonic protection after childbirth, measured using pre- and postintervention surveys (28 clusters, each with 30 households). Maternal deaths were ascertained through systematic health facility and community-based surveillance; causes of death were assigned based on verbal autopsy.

Results

Of 840 postintervention survey respondents, 73.2% received misoprostol. The standardized proportion of vaginal deliveries protected by a uterotonic rose from 11.6% to 74.2%. Those experiencing the largest gains were the poor, the illiterate, and those living in remote areas.

Conclusion

Community-based distribution of misoprostol for PPH prevention can be successfully implemented under government health services in a low-resource, geographically challenging setting, resulting in much increased population-level protection against PPH, with particularly large gains among the disadvantaged.

Keywords: Community-based distribution, Misoprostol, Nepal, Operations research, Postpartum hemorrhage, Self-administration

Back to Article Outline

1. Introduction 

Postpartum hemorrhage (PPH) remains a major killer worldwide and it is estimated to cause 30% of maternal deaths [1]. In Nepal, maternal mortality is high (281 per 100000; 95% CI, 178–384) [2]. According to a 1998 study in Nepal [3], almost half of these deaths (46%) can be attributed to PPH. There is a consensus that active management of the third stage of labor (AMTSL) can prevent most PPH deaths due to atony [4], but despite serious efforts in Nepal to increase institutional deliveries, the proportion remains low (17.7%) [2]. Only 18.7% of deliveries are attended by skilled providers [2], many of whom do not practice AMTSL. Given geographic and resource constraints, high levels of AMTSL will only be achieved over the long term. To complement such efforts over the short to medium term, it is worth investigating simpler, less resource-intensive interventions with the potential to reach women who otherwise would not receive skilled care.

The study reported in this paper does not address the mortality or morbidity impact of misoprostol use; such research has been done elsewhere (see below). In studies of misoprostol efficacy for PPH prevention, misoprostol has been compared with other uterotonics, usually oxytocin, and with placebo. The usual criterion for severe PPH has been measured blood loss greater than 1000mL. Comparison with oxytocin has been useful in establishing relative efficacy. A recent Cochrane review [5] documented that oxytocin is 25% more effective than misoprostol in preventing PPH due to atony (although need for blood transfusion was less likely with misoprostol). The same paper also reviewed studies comparing misoprostol with placebo. Given that the main programmatic interest in misoprostol is for use where women would otherwise get no uterotonic, this is the more relevant comparison. The Cochrane review considered several studies looking at blood loss, comparing misoprostol with placebo, generally using 600μg of oral misoprostol and most of which were conducted in a hospital setting. Because of heterogeneity in methods, a summary measure could not be calculated. Benchimol et al. [6] showed no difference in the proportion of women with bleeding greater than 1000mL comparing misoprostol, oxytocin, and placebo. An underpowered study in South Africa in which a smaller 400-μg dose was used [7] showed a nonsignificant reduction in favor of misoprostol (RR 0.65; 95% CI, 0.35–1.22). A health center-based study in Guinea-Bissau [8] obtained similar results in reduction in severe PPH in favor of misoprostol (RR 0.66; 95% CI, 0.45–0.98). Most relevant for our setting, an Indian study [9] used peripheral health center-based auxiliary-nurse midwives who practiced expectant management rather than active management of the third stage. This trial, then, gives a purer measure of the PPH reduction effect of misoprostol compared with nothing, which is the relevant comparison for the more than 80% of women in Nepal who deliver at home with no skilled attendant. The Indian study showed a larger reduction (RR 0.20; 95% CI, 0.04–0.91) than the other studies cited, in which the controls received other elements of AMTSL.

Of the studies covered in the Cochrane review [5] none showed major adverse events; however, this issue received more focused attention in another recent review [10]. The studies that have addressed either treatment or prophylactic use of misoprostol have generally used blood loss as the endpoint. Given the rarity of death or serious morbidity in such studies, even aggregating results from multiple trials provides poor statistical power. Hofmeyr et al. [10] reviewed 46 randomized controlled trials with over 40000 study subjects, among which there were 11 deaths. The number of deaths was greater among those receiving misoprostol than among those receiving placebo but the difference was not statistically significant.

Misoprostol is administered orally and does not require special temperature control provisions, making it suitable in the absence of skilled care and/or when oxytocin is not available [11]. A recent cost-effectiveness modeling study [12] of community-based distribution of misoprostol for home deliveries in a south Asian setting found it to be a cost-effective intervention, with a cost per death averted of $1401. However, there has not yet been a documented community-based distribution at scale of misoprostol through public sector cadres. This operations research study was designed to test the feasibility of community-based delivery of misoprostol by existing public-sector community health volunteers, with self-administration of three 200-μg tablets by women delivering at home. This PPH-related work was one of a broader set of activities, intended to improve maternal and neonatal outcomes, implemented by the district public health system with support from the Nepal Family Health Program (NFHP) funded by the United States Agency for International Development (USAID). The primary purpose was to investigate: (1) the feasibility (including service coverage and uptake of promoted self-care practices); (2) the acceptability (to providers and beneficiaries); and (3) the safety of community-based distribution of misoprostol to pregnant women. A key secondary purpose was to show that making misoprostol available for self-administration at home delivery would not undermine efforts to increase skilled care at birth. The study protocol was reviewed and approved by the ethics committee of the Nepal Health Research Council.

Back to Article Outline

2. Materials and methods 

2.1. The study setting 

The study was set in one district (Banke) in the plains (terai) along the border with India. The target of the study was the 88% of the district that is rural (437000 population). Banke has a public sector hospital and a private medical college with a hospital. Both of these provide 24-hour delivery service with the capacity to perform cesarean deliveries. There are also 3 primary health centers (PHCs), 9 health posts, and 35 subhealth posts staffed by trained health workers some of whom have limited maternity-related skills but few could be called “skilled birth attendants.” In addition, there are 665 rural Female Community Health Volunteers (FCHVs) who work as health promoters and also serve as community-based distributors. On average, each FCHV serves about 120 households, works 3–8hours per week, has received some training, is reimbursed for some travel expenses, and has worked as an FCHV for an average of 11years. Most (62%) are illiterate and their median age is 40years.

2.2. The intervention 

Under a broader community-based maternal-neonatal effort, NFHP provided limited support to selected health facilities, including: training to key staff (refreshers on AMTSL, use of partogram, management of eclampsia, and other aspects of emergency obstetric care); equipment (delivery-related instruments, blood pressure cuffs, and scales); and minor repair and renovations of maternity facilities.

The main intervention was support and training to peripheral health workers and FCHVs to enable them to: identify pregnant women in their area, provide prenatal health education, dispense misoprostol (three 200-μg tablets) late in pregnancy (8th month), and make early postnatal home visits. We aimed to reach all pregnant women with misoprostol. The FCHVs were also required to document their work. Primarily because of difficulties they had with this task, 254 of the 665 rural FCHVs were not asked to dispense misoprostol, but instead to refer women to dispensing FCHVs. Prenatal health education by FCHVs was done in the home, over 3-4 sessions through the second and third trimesters, and involved other family members, notably mothers-in-law. It included advice on: seeking prenatal care and planning for institutional delivery, recognition of and timely response to danger symptoms, self-care during pregnancy, and essential newborn care. Information on misoprostol was also given. It included the importance of taking the 3 pills only after delivery and the dangers of taking them earlier, and the expected side effects and how to manage them.

The early postnatal home visit was used to reinforce messages on danger signs and essential newborn care, to document use of misoprostol and any symptoms experienced, and to recover and account for all unused misoprostol.

The FCHV training time was 7days, of which 3 were focused on misoprostol. Most trainers were government staff working at the district health office or in peripheral health facilities. At least one NFHP project staff was present at all training sessions. Most supervision was done by government staff but NFHP project staff were also involved in review meetings and field supervision. Over the 2.5years of implementation reported here, 5 fulltime-equivalent NFHP field project staff provided oversight for misoprostol distribution and other community-based maternal-neonatal work. Their primary role was monitoring and documentation. District health office staff provided management oversight, with support from an NFHP project officer.

2.3. The data 

Two methods were used to measure program performance.

Household surveys (which included 30 clusters of 30 households each) were conducted at baseline and at endline. Required sample size was determined assuming that, for key outcome variables measured, prevalence was 50% at baseline and shifted by at least 10% by endline, with 5% type I error and 20% type II error. This yielded sample sizes (for a simple random sample) of 408 for each survey. To account for design effect due to the cluster sampling, the number was doubled and further increased by 10% to account for possible nonresponse to yield a final estimated sample size of 898.

Primary study respondents were women delivering a live or dead infant of 28weeks or greater of gestation over the preceding 12months. Sample clusters were selected from the 414 rural wards in the district using probability-proportionate-to-estimated size (PPES), based on census data [13]. The same clusters selected at baseline were surveyed at endline to minimize intersample sociodemographic variability. Surveyors mapped each selected ward then randomly chose an index household from among the numbered households. Index households were independently selected at baseline and at endline. If an eligible respondent was present, surveyors conducted an interview and then proceeded to the closest house where they again interviewed eligible respondents. This procedure was repeated until 30 respondents were interviewed. To form the sample of 900 recently delivered women, 5935 households were visited at baseline, 4964 at endline.

A local research organization (Valley Research Group) conducted the surveys, supervised by the investigators. Baseline field work was done in May to June 2005, about 6months before full implementation began. Of the 30 clusters selected, 2 were dropped for security reasons. Endline field work was in June to July 2007. Pre- and postcomparisons were restricted to the 28 clusters for which we had both pre- and postdata.

FCHVs and health-facility-based staff recorded information on all program beneficiaries including from FCHV home visits to all women who received misoprostol; 97% were visited within 7days of delivery. Information collected at home visit included compliance, complications, and any side effects of misoprostol.

2.4. Outcome variables 

The principal performance measure used was use of a uterotonic drug immediately after birth. In our context, this meant taking either misoprostol or oxytocin. To measure receipt of oxytocin through household survey, we used responses to 2 questions: “did you receive an injection in the thigh or buttock immediately after delivery” and “who was in attendance at the delivery?” When an injection was given and a qualified health worker was present, we assumed that the injection was oxytocin. However, based only on responses to this question, our measure may over or underestimate use of oxytocin for PPH prevention. In this setting, unqualified practitioners are known to give postpartum injections of other substances, notably tetanus toxoid.

Misoprostol use was ascertained by asking if, after birth, the woman had used “Matri Suraksha Chakki” (the name used for misoprostol distributed under the program). Our analysis assumes that women received uterotonic protection if they took 2 or 3 misoprostol tablets. In addition, the survey instrument (available from the authors) collected data on respondent demographics and a variety of other household practices and care-seeking behaviors targeted by the broader intervention. Neonatal mortality was also measured, although it is not reported in this paper.

To determine the safety of community-based distribution of misoprostol, from the start, we also systematically collected information on maternal deaths, identified through quarterly visits by project staff to the 3 heath facilities providing delivery services and through FCHV reports for deaths in the community. FCHVs reported to their local health facility the names of the deceased woman and her husband and where they lived. The reports were collected at district level. Subsequently, guided by the FCHV reporting the case (who received a financial incentive of about US $5 per case identified), project staff visited the homes of women who had died and conducted a detailed and standardized verbal autopsy covering the circumstances of the death, and whether the woman had received and used misoprostol or delivered in a health facility and received an injection to prevent bleeding. The verbal autopsy tool and approach were adapted from JHPIEGO's earlier work in Indonesia [14]. Findings were reviewed independently by 2 obstetricians to assign cause of death. In cases of disagreement, they met to discuss the cases and revisited the family if needed.

Thirteen of the 46 Village Development Committees (the most peripheral geographic-administrative unit in Nepal, hereafter referred to as Village) either had a high population-per-FCHV ratio or lower program performance (reaching a smaller proportion of expected pregnancies). To identify missing maternal deaths in these Villages, more intensive follow-up was done by project staff (May–July 2008). This involved meetings with community groups and local persons of influence. Two additional deaths were thus identified. Maternal deaths in the program area were tracked and classified by misoprostol use. Actual denominators for misoprostol receipt and use are known from our project management information system. The number of nonrecipients was estimated based on survey [2] and census data [13]. The Central Bureau of Statistics provides population projections based on the 2001 census. This estimate for rural Banke was used, together with the crude birth rate for rural Nepal from the DHS 2006 [2] to estimate the annual number of births multiplied by the 2.5years of program implementation (26000).

2.5. Statistical analysis 

Given the self-weighting design of the PPES sampling approach, indicators derived from binary variables were estimated as unweighted proportions unless otherwise noted; in some cases direct standardization was used when presenting baseline and endline estimates of a variable to account for intersample differences in literacy and wealth status. Although the same clusters and the same sampling approach were used for baseline and endline surveys, the same households were not; thus, respondent characteristics differed somewhat between the 2 samples (Table 1).

Table 1. Characteristics of survey respondents at baseline and endline.

Characteristics	Baseline (n=840)	Endline (n=840)	Comparison
1. Mean age of respondent (in completed years)	25.1	24.9	P=0.45
2. Mean number of births	3.10	2.74	P<0.001
3. Percentage of literate respondents	29.4%	37.0%	OR 1.41 (95% CI, 1.14–1.75)
4. Wealth status (% of respondents):
Quintile 1 (least wealthy)	25.0	15.0	χ2=36.9, P<0.001
Quintile 2	19.5	20.5
Quintile 3	20.4	19.7
Quintile 4	19.4	20.5
Quintile 5 (most wealthy)	15.7	24.3

Note: Baseline and endline values of characteristics 1 and 2 were compared using the t test; baseline and endline values of characteristic 3 were compared using logistic regression that adjusted for cluster sample design; baseline and endline values of characteristic 4 were compared using the Pearson χ² test.

In the presentation of results, odds ratios and confidence intervals were adjusted for household wealth, literacy, and cluster design effect using logistic regression to address differences found between baseline and endline samples. Analyses were performed using Stata version 9 (Stata Corp, College Station, TX, USA).

Households were ranked into wealth quintiles by household assets, using the same procedures as in the DHS.

Back to Article Outline

3. Results 

3.1. Feasibility 

We were interested in demonstrating whether it is feasible to implement community-based distribution of misoprostol for PPH prevention at district scale within the government health system and achieve high coverage, particularly among the disadvantaged. Key elements of relative disadvantage assessed include wealth, literacy, and remoteness.

Program monitoring data show that it is feasible to achieve high coverage of misoprostol for PPH prevention through community-based distribution at district scale within the government health system. From January 2006 through June 2008, 18761 women received misoprostol and related counseling; 13969 (74.5% of those receiving misoprostol) took it. Endline survey data show that 73.2% of recently delivered women (95% CI, 66.3–80.1) reported having received misoprostol from an FCHV during pregnancy. The endline survey also shows actual rates of use for misoprostol and oxytocin (Fig. 1).

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 1 

  Uterotonic coverage at endline among women with live vaginal births (n=816). Abbreviation: miso, misoprostol. Note: In some cases, because of rounding, percentages do not sum to 100%.

Overall uterotonic coverage among women with vaginal births increased from 11.6% at baseline to 74.2% at endline (standardized) (OR 25.0; 95% CI, 15.6–40.1). At endline, 53.9% reported taking 2 or 3 tablets of misoprostol and 25.2% were given oxytocin; 4.5% had both. The increase in uterotonic coverage was a function of misoprostol used at home deliveries, an increase in health facility deliveries, and an increase in the proportion of health facility deliveries in which care included postpartum oxytocin (Table 2).

Table 2. Oxytocin and misoprostol use by location of delivery, baseline, and endline (among live vaginal births).

Note: The figures presented in this table are not standardized by wealth quintiles and literacy, as they are in the other tables. Where reference is made to “oxytocin given,”’ this is based on positive responses to 2 questions: trained health worker present at delivery and injection given after delivery. For ethical reasons we administered an abbreviated version of our survey instrument to women who delivered stillborns. This did not provide us information on uterotonic use or location of delivery.

Fig. 2 shows a dramatic increase in uterotonic coverage across all wealth strata, with the greatest proportionate gain in the two lowest quintiles.

• 
  • View Large Image
  • Download to PowerPoint

• Fig. 2 

  Uterotonic coverage among women with vaginal births, by household assets wealth quintiles, % (baseline n=813, endline n=816). Note: At baseline the only stratum with uterotonic coverage higher than the lowest stratum at a level that is statistically significant was the richest stratum (P=0.011). At endline, the third (P=0.029) and the richest (P=0.049) strata had significantly higher coverage than the lowest stratum.

We also determined how effectively the program reached illiterate mothers. At baseline, 6.9% of illiterate compared with 19.8% of literate mothers reported having uterotonic protection at delivery (i.e. oxytocin), (OR 3.3; 95% CI, 2.1–5.1). At endline there was no significant difference in proportions protected: 73.4% among illiterates, 76.9% among literates (OR 1.2; 95% CI, 0.8–1.8).

The third dimension of social disadvantage assessed was remoteness of place of residence. Based on the Village they resided in, time required to reach emergency obstetric care by usual means of transport was determined by project staff based on local inquiries and imputed to each survey respondent. Among the 840 respondents included in the analysis, 35.7% lived less than 1hour from the closest such site; 42.9% lived 1–3hours away; and 21.4% lived more than 3hours away (the percentages were the same in both surveys because sampling was done in the same clusters and with the same number of respondents per cluster).

3.2. Institutional delivery 

As well as providing misoprostol for home deliveries, the intervention promoted and supported institutional delivery. The institutional delivery rate among live births increased from 10.9% at baseline to 14.8% at endline (standardized) (OR 1.47; 95% CI, 1.13–1.93). Among women resident within 3hours of a suitable facility, a significantly higher proportion of deliveries occurred in health facilities at endline than at baseline. For those living further away, institutional deliveries remained uncommon. Nevertheless, once the intervention was established, uterotonic protection coverage (oxytocin and misoprostol), was uniformly high across remoteness strata (Table 3).

Table 3. Health facility delivery and uterotonic coverage by remoteness.

	Hours to closest emergency obstetrical care facility	Baseline	Endline	OR (95% CI) Baseline is the reference category
	Denominator is live births including cesareans	n=827	n=831
Proportion delivering in a health facility, by remoteness	≤1hour	14.5%	21.8%	1.36 (0.95–1.93)
	1–3hours	9.3%	17.5%	1.72 (1.14–2.58)
	>3hours	4.0%	3.9%	0.92 (0.17–4.92)
	Denominator is live vaginal births	n=813	n=816
Proportion of deliveries protected by a uterotonic, by remoteness	≤1hour	14.0%	73.8%	16.8 (7.7–37.0)
	1–3hours	9.1%	74.8%	32.7 (14.8–72.3)
	>3hours	8.5%	75.7%	39.0 (17.6–86.3)

Note: Percentages in columns B and C are unadjusted. Odds ratios and associated confidence intervals for baseline-endline comparisons were calculated using logistic regression, adjusting for wealth, literacy status, and cluster design. Note that for both baseline and endline surveys, those living less than 1hour from the health facility accounted for 35.7% of the sample; those living 1–3hours away accounted for 42.9%; and those more than 3hours away accounted for 21.4%.

3.3. Acceptability 

To achieve sustained high coverage the service must be acceptable to both beneficiaries and providers. Seventy-four percent of women who had live vaginal births at endline (604/816) received misoprostol. Among the misoprostol recipients, 26.5% (160/604) consumed no tablets. Women who received but did not take misoprostol gave several reasons for not using it. Three quarters (121/160) stated that they did not take misoprostol either because they delivered at a health facility, they were delivered by a health worker, or they received an injection after delivery from a health worker; our data show that 90% (109/121) of such women received oxytocin. The remaining quarter (39/160) of non-using recipients delivered without uterotonic protection. The most common reasons reported were that they forgot to take it (9%) or they chose not to take it because the placenta delivered quickly and they believed (mistakenly) that once the placenta delivered it was too late to take the misoprostol (6% of non-using recipients). Fear of side effects accounted for 5% of recipients who did not use misoprostol.

Misoprostol use is associated with prostaglandin-related side effects, notably shivering. Anticipating such effects was part of the FCHV counseling for pregnant women. Among endline survey respondents having live vaginal deliveries (n=816) misoprostol users were more likely to report shivering than non-users (18.3% vs 14.2%) but this difference was not statistically significant (Table 4). The interval between delivery and administration of the survey was as long as 12months, so recall of minor side effects may have been affected. From project monitoring data, incidence of shivering was slightly higher than the survey result, 22% vs 18.3%.

Table 4. Problems experienced in the first 6hours following delivery by use status (from endline survey).

Type of problems experienced within 6hours of delivery	Used misoprostol, % (n=444)	Non-users, % (n=372)	OR (95% CI)
Felt faint or dizzy	21.8	27.4	0.70 (0.49–0.99)
Fainted or lost consciousness	3.8	7.5	0.48 (0.25–0.90)
Shivering	18.5	15.0	1.37 (0.82–2.28)
Nausea	3.4	6.2	0.51 (0.29–0.89)
Fever	9.0	9.4	0.87 (0.49–1.55)
Watery stool	2.0	1.3	1.64 (0.61–4.37)
Headache	6.3	6.7	0.99 (0.43–2.27)

Note: Logistic regression was used to generate odds ratios and associated confidence intervals for comparisons between users and non-user values, adjusting for wealth, literacy status, and cluster design. Analysis was limited to women with live vaginal births. “Used misoprostol” was defined as “took one or more tablets of misoprostol.”

3.4. Safety 

Among the 447 survey respondents taking misoprostol, none reported taking it before delivery and only 1.8% of users (8/447) took fewer than 3 tablets. The project database has follow-up data on all 13969 women who took misoprostol; only one reported having taken it before delivery. Although the woman was aware that she should not take it before delivery, she seemed to have taken all 3 tablets to “act out” in the context of domestic conflict. She was at 38weeks of gestation and was taken to hospital for observation where she had a normal labor and a healthy baby within 24hours.

Subtracting the number of women known to have received misoprostol (18761) from the estimated number of births over this period (26000) gives 7239 non-recipients.

Among 13969 misoprostol users, there were 10 deaths, giving an observed mortality ratio of 72 per 100000 (mortality ratios for Banke were calculated based on complete or very nearly complete mortality reporting for the rural parts of the district, not on survey data, so no confidence intervals are used). The observed mortality ratio was significantly lower than the expected mortality based on national data (281 per 100000; 95% CI, 178–384 [2]), and lower than the ratio among non-users (292 per 100000; i.e. 35 deaths among 12031 births). Note that this denominator, 12031, is the sum of the 4792 women in our project management information system who received but did not take misoprostol and the estimated 7239 women who did not receive misoprostol, based on survey and census data. For the district as a whole over this period (excluding the municipality) the observed maternal mortality ratio was 173 per 100000, with 29% of deaths attributed to PPH (vs 46% expected from the Nepal Maternal and Mortality Study [3]).

The numbers of deaths by specific cause are too small to evaluate differences in cause-specific mortality by misoprostol use status. Detailed review of all 10 deaths among those taking misoprostol gave no indication that misoprostol played a role. One of the 10 deaths was attributed to possible rupture of a scarred uterus. The woman continued to bleed after her home delivery and after taking misoprostol. As a result she was subsequently given injections by a worker at a local medical shop.

Back to Article Outline

4. Discussion 

This study has shown that it is feasible to achieve high population coverage of misoprostol through distribution by trained community volunteers under the government primary healthcare system in Nepal. The approach was especially effective in extending such protection to the disadvantaged—women who were poorer, illiterate, or living in more remote areas.

Furthermore, we showed that expanding access to uterotonic protection through community-based distribution of misoprostol and self-administration can complement efforts to increase institutional deliveries. According to the government's Health Management Information System, of all the districts in the country, Banke had one of the biggest increases in institutional deliveries (from 11% to 20.4%) from the year pre-intervention (2004/5) to the most recent year for which a report is available (2006/7). Nationally, the increase was from 11.3% to 15.3% [15]. We also documented, through our survey data, an increase in the proportion of health facility deliveries in which oxytocin was used for PPH prevention, from 64% to 86%.

The work described here was implemented largely with the resources of the district public health system and with only modest external support, much of which was for initial training, monitoring, and documentation. While the misoprostol was donated, the expense (about US $0.30 per woman for all 3 tablets) can be managed even with the government's resource constraints.

An important determinant of treatment acceptability is toleration of side effects. Several investigators have expressed concern that frequency of side effects, notably shivering, may limit the utility of misoprostol [16]. Although we documented shivering in 1 in every 5–6 women using misoprostol, this did not undermine willingness to use it; this is consistent with findings by Caliskan et al. [17].

To be successful at scale, an intervention also needs to be acceptable to those providing the service. With the new maternal-neonatal duties FCHVs assumed in Banke, their average number of hours per week may have increased, although we did not document this. One could imagine that for some this may limit their willingness to take on further such duties. Nepal has periodically conducted surveys of FCHVs, drawing a random sample of 100 FCHVs in participating districts. In these surveys [18], FCHVs are routinely asked about how much time they would like to work in the future compared with the present. In 2005, the year before implementation, 62% of those surveyed in Banke (n=100) responded that in future they would like to work longer hours; 35% said about the same as now, and 4% said they would like to work less. In 2008, following FCHVs’ assumption of misoprostol-related and other new duties, the same question was asked. Of 100 respondents, 69% said more, 25% said the same, and 6% said fewer hours. This suggests that FCHVs have generally been happy to take on new functions and are interested in taking on more.

Misoprostol is an abortifacient and is also used in labor induction; if used inappropriately intrapartum or while a woman is pregnant, it can cause complications for the woman and fetus. The community volunteers dispensing misoprostol were trained to advise women to take the medication only after delivery. We documented that this information was universally given and understood and almost universally complied with. We believe it is important that misoprostol be dispensed with clear instruction on not taking it in pregnancy or during labor; with this information given, based on our experience in Banke we believe it can be used safely at scale in community programs.

In this study, for safety monitoring, we also tracked maternal mortality. We feel that although this was not done through household level registration and demographic surveillance, measures taken to ensure completeness were robust (including special follow-up efforts in low-performance areas). Cause of death was determined by verbal autopsy. The numbers of deaths by specific causes are too small to draw any conclusions about any specific protective or risk-enhancing effect of misoprostol. Nevertheless, the total number of deaths documented, particularly among those using misoprostol, was lower than expected given Nepal's national maternal mortality ratio. Although this suggests there may have been impact, in the absence of pre-intervention mortality documentation, we cannot determine the impact on maternal mortality. Furthermore, any reduction in mortality that occurred could not be attributed solely to increased uterotonic coverage. The intervention included elements intended to improve utilization of other maternal health services and encourage appropriate household practices.

There are several limitations to this study. Our objective was to explore the feasibility of a large-scale approach in a difficult setting where few women deliver in health facilities with trained workers. The evaluation design was a simple pre- and post- intervention with no comparison area. This is an inherently weak design for making causal inference although for the main measure used in this study—uterotonic protection—the lack of comparison area does not weaken our conclusions. For other observed changes it is important to note that the misoprostol intervention was not implemented in isolation. Other policy and program initiatives occurred over this period and undoubtedly affected Banke. Notably, the government introduced a new program in which service providers were given an incentive to attend deliveries and women delivering in health facilities were given a payment to partially offset transport and other costs. This probably contributed to increasing the proportion of deliveries occurring in health facilities.

Side effects of misoprostol have been identified by some authors as a problem that could restrict its programmatic utility and most studies have documented higher rates of shivering than we found in our study. This no doubt reflects differences in how shivering was measured. In hospital-based studies, this was captured by direct observation by clinical staff regardless of whether the patient considered it significant. Since our measure of shivering and other side effects was by self-report and, in the case of our survey data, reported as much as a year after the experience, it is to be expected that study subjects will only recall the side effect if they considered it significant at the time. Nevertheless, with over 13000 women having used misoprostol in our pilot, there has been little if any indication of resistance to future use because of unacceptable side effects. In a recent review [10], a 400-μg dose of misoprostol was found to be as effective as 600μg in reducing PPH risk, but gave fewer side effects. Investigators or program managers contemplating future similar operations research work should consider using the lower dose.

The main measure of program performance we used was uterotonic protection, which included use of misoprostol or receipt of an injection, presumed to be oxytocin, after delivery. It is possible that we have overestimated the coverage estimate for oxytocin provided for PPH prevention because the information was collected through household surveys and it was therefore not possible to confirm that reported injections were oxytocin. Furthermore, some respondents may have misreported oxytocin given for labor augmentation as immediate postpartum oxytocin for PPH prevention.

Nevertheless, lessons from this study can help direct public health practice in Nepal and elsewhere. First, we suggest that uterotonic coverage—the proportion of all deliveries protected through use of oxytocin, misoprostol or other uterotonics—would be a useful indicator for monitoring progress in extending safe-motherhood services.

We believe the next logical step in Nepal, before making decisions on country-wide implementation, would be a limited scale-up of community-based distribution of misoprostol using FCHVs, validating that adequate program performance (including safe use) can be maintained using a more streamlined approach within the district public health system, minimizing the need for external resources.

There has been a sometimes unhelpful tension in safe-motherhood efforts between advocates of institutional delivery services and those promoting community-based approaches. Unlike child health, which benefits from several highly effective community-based interventions, maternal health is inherently more challenging. To achieve substantial reductions in maternal risk, there is no substitute for wide use of skilled care at delivery, including a full range of clinical support services and ready access to full emergency obstetric care. But the excellent need not be the enemy of the good. Substantially increasing skilled care at delivery will be difficult in Nepal and elsewhere. As we continue efforts to improve coverage and quality of delivery services, interim measures can contribute to driving down the burden of maternal deaths. This study has demonstrated that high coverage with a preventive dose of uterotonic can be achieved with modest means. Much wider use of misoprostol, focusing on those not yet reachable with more definitive care, targets the principal cause of maternal death in low-resource countries and has the potential to significantly reduce PPH deaths due to atony.

Back to Article Outline

5. Misoprostol Study Group members 

Asha Pun, Shailes Neupane, Bharat Ban, Jaya Bahadur Karki, Kehar Singh Godar, Angad Bahadur Shahi, Ram Chandra Silwal, Ram Bahadur Shrestha, and Shilu Aryal.

Back to Article Outline

Acknowledgements 

The program was supported the United States Agency for International Development (USAID), principally through a Nepal-based bilateral project (the Nepal Family Health Program II, implemented by JSI R&T; CA#: 367-00-02-00017-00) as well as through two centrally-funded projects (HARP-GRA, implemented by the Johns Hopkins Bloomberg School of Public Health and the ACCESS project, implemented by JHPIEGO). The contents do not necessarily reflect the views of USAID or the United States Government. Misoprostol was provided by PLAN and Venture Strategies for Health and Development.

Conflict of interest

We declare that we have no conflicts of interest.

Back to Article Outline

References 

1. World Health Organization. Maternal mortality in 2005. Geneva: World Health Organization; 2007;
   • View In Article
2. Nepal Demographic and Health Survey 2006. Kathmandu: Ministry of Health and Population, NewERA, and MACRO Inc.; 2007.
   • View In Article
3. Pathak LR, Maternal Mortality and Morbidity Study, Family Health Division, DoHS, Ministry of Health and Population, Kathmandu: Government of Nepal; 1998.
   • View In Article
4. International Joint Policy Statement: FIGO/ICM global initiative to prevent post-partum hemorrhage. J Obstet Gynaecol Can. 2004;26(12):1100-2–1108-11
   • View In Article
5. Gulmezoglu AM, Forna F, Villar J, Hofmeyr GJ. Prostaglandins for preventing postpartum hemorrhage. Cochrane Database of Syst Rev. 2007;3:CD000494
   • View In Article
6. Benchimol M, Gondry J, Mention JE, Gagneur O, Boulanger JC. Role of misoprostol in the delivery outcome. [in French] J Gynecol Obstet Biol Reprod (Paris). 2001;30(6):576–583
   • View In Article
   • MEDLINE
7. Hofmeyr GJ, Nikodem VC, de Jager M, Gelbart BR. A randomized placebo control trial of oral misoprostol in the third stage of labour. Br J Obstet Gynaecol. 1998;105(9):971–975
   • View In Article
   • MEDLINE
8. Hoj L, Cardoso P, Nielsen BB, Hvidman L, Nielsen J, Aaby P. Effect of sublingual misoprostol on severe postpartum haemorrhage in a primary health center in Guinea-Bissau: randomized double blind clinical trial. BMJ. 2005;331(7519):723
   • View In Article
   • CrossRef
9. Derman RJ, Kodkany BS, Goudar SS, Geller SE, Naike VA, Bellad MB, et al. Oral misoprostol in preventing haemorrhage in resource-poor communities: a randomized controlled trial. Lancet. 2006;368(9543):1248–1253
   • View In Article
   • Abstract
   • Full Text
   • Full-Text PDF (95 KB)
   • CrossRef
10. Hofmeyr GJ, Gulmezoglu AM, Novikova N, Linder V, Ferreira A, Piaggio G. Misoprostol to prevent and treat postpartum haemorrhage: a systematic review and meta-analysis of maternal deaths and dose-related effects. Bull World Health Organ. 2009;87(9):666–677
    • View In Article
    • CrossRef
11. World Health Organization. WHO Recommendations on the Prevention of Postpartum Hemorrhage. A summary of the results from a WHO Technical Consultation. Geneva: World Health Organization; 2006;
    • View In Article
12. Sutherland T, Bishai DM. Cost-effectiveness of misoprostol and prenatal iron supplementation as maternal mortality interventions in home births in rural India. Int J Gynecol Obstet. 2009;104(3):189–193
    • View In Article
13. Census 2001. Kathmandu: Central Bureau of Statistics. Government of Nepal; 2002.
    • View In Article
14. Sanghvi H, Wiknjosastro G, Chanpong G, Fishel J, Ahmed S, Zulkarnain M. Prevention of postpartum hemorrhage study. Baltimore: MNH Program; 2004;
    • View In Article
15. Annual Reports 2004/5, 2006/7. Kathmandu: Department of Health Services, Ministry of Health and Population, Government of Nepal; 2005, 2007.
    • View In Article
16. Chong YS, Su LL. Misoprostol for preventing PPH: some lessons learned. Lancet. 2006;368(9543):1216–1218
    • View In Article
    • Full Text
    • Full-Text PDF (47 KB)
    • CrossRef
17. Caliskan E, Dilbaz B, Meydanli MM, Ozturk N, Narin MA, Haberal A. Oral misoprostol for the third stage of labor: a randomized controlled trial. Obstet Gynecol. 2003;101(5 pt 1):921–928
    • View In Article
    • MEDLINE
    • CrossRef
18. Analytic Report[s] on Female Community Health Volunteers, 2005, 2008. Kathmandu: NewEra; 2006, 2009.
    • View In Article