Menu
Menu
Menu
Menu
Background The Maternal Mortality Ratio in Mozambique has stagnated at 405 deaths per 100,000 live births with virtually no progress over the last 15 years. Low Institutional Birth Rates (IBRs) levelling around 50% in many rural areas constitute one of the contributing reasons. Demand-side financing has successfully increased usage of maternal health services in other countries, but little information exists on in-kind incentives in rural Africa. The objective was to test the impact on Institutional Birth Rates of giving a USD 5.50 baby package incentive to every woman who came to give birth in a health centre in a rural, poor district of Cabo Delgado, Mozambique. Methods and findings The intervention was implemented in one district in 2010 with the remaining 15 districts serving as controls. The total population in the 16 districts in 2006 was just under 1.5 million people. IBRs were observed from 2006 to 2013 (53 months before and 55 months after the intervention began). The non-intervention districts showed a slight increase, from a mean IBR of 0.39 (SD = 0.10) in 2006 to 0.67 (SD = 0.13) in 2014. The intervention district had a dramatic increase in IBRs within six months of the start of the intervention in 2010, which was sustained until the end of the study. Adjusting for the background increase and for confounders, including health facilities and health personnel per district, and taking clustering in districts into account, the estimated rate ratio of institutional births in the intervention district was 1.80 (95% CI 1.72, 1.89 p<0.001). Conclusion Women were almost twice as likely to have an institutional birth following the introduction of the baby package.
Connected by one tarmac road to the rest of the country, Cabo Delgado is the most northern province of Mozambique. When compared to other provinces in the country, it has some of the poorest indicators as demonstrated in Table 1. In 2009, when the intervention was planned, Ancuabe had a population density of 22.6 people per square kilometre, the vast majority of whom lived off subsistence farming. Fifty-three percent of the population between the ages of 15 and 40 years of age were women. The population in the 16 districts were attended to by 110 health centres and 996 clinical staff. Data from DHS 2011 [16]. *per 1000 live births. The intervention was performed in the district of Ancuabe where 112,610 (2009) people live mostly by subsistence farming. In the district, 99% of the population have no running water, 98% live in mud houses with straw roofs and only 56% have access to latrines (Table 2). Data according to the National Health Information System (NHIS). CEmONC: Comprehensive Emergency Obstetric and Neonatal Care. There was no facility providing Comprehensive Emergency Obstetric and Neonatal Care (CEmONC) in the district. On average, women travelled 10 kilometres to reach the nearest health centre. The remaining districts of the province served as control districts. The study was carried out in all 16 rural districts of Cabo Delgado Province, comprising one intervention district (Ancuabe) and 15 comparison districts. The distribution of baby packages was implemented in Ancuabe beginning on June 1, 2010. For all 16 Districts, the number of institutional births was observed from January 2006 to December 2013, comprising 53 months before and 55 months after the start of the intervention. At the start (2006), the 16 Districts had a total population of 1,433,344, 7% of whom (104,694) lived in the intervention district of Ancuabe, and 83% (1,328,650) in the 15 non-intervention rural districts. The population of the province reached a total of 1,671,344 by 2013. Due to its different social, geographic and service coverage pattern, the only urban district of the Province, the capital Pemba (which had the provincial referral hospital) was not included in the analysis. Therefore, its population of 135,438 was excluded, making the control and intervention districts more comparable. The intervention consisted of giving a ‘baby package’ to every woman who had a birth (live or stillbirth) in one of the 6 health centres in Ancuabe district. This is in line with national policy guidance that identifies incentives as one of the ratified strategies to increase institutional birth rates. Those who had twins received two baby packages. The baby package consisted of a traditional cloth (capulana), a plastic basin, a bar of soap and 3 cloth nappies. The choice of items was decided following focus group discussions with women’s groups before the start of the intervention. The package was the same for all women, independent of income or distance from the health centre. The mean cost per package was USD 5.50, including transport of the materials. No promotional campaign was organised. Packages were bought from a bulk supplier in the city of Nampula on a six-month basis. The supplier organised the 250km transport to the district store room. The Chief District Health Nurse then distributed the packages on a monthly basis from the central district store to each health centre, with occasional transport support from the Swiss non-profit organization, SolidarMed. The package was given to the mother by the nurse at each health centre upon discharge and her name and fingerprint were recorded as confirmation of receipt. Every three months, stocks were monitored by a nurse who reconciled the number of distributed packages with the number of institutional births occurring in each health centre, in order to avoid fraud. Health centre stock-out of baby packages occurred on only one occasion (February 2012), due to delays in purchase and delivery. No supply side interventions were implemented concurrently with baby package intervention over the time period monitored. For the purpose of stock control, a paper-based register was introduced at all participating health facilities in the intervention district. For each baby package given, the nurse from the health centre was asked to record the name and age of the mother, the date of delivery, mother’s village of origin and distance from the health centre. Once a month, the study nurse entered those data into a spreadsheet by Microsoft Excel (version 2010) and gave stock monitoring feedback to the nurse at each health centre. Monitoring visits were performed every 4 to 6 weeks by NGO staff to ensure that the baby package register was correctly filled and that patients were receiving the baby package as registered. Institutional birth numbers used for this study were not, however, extracted from the baby package register, but from the Mozambican National Health Information System (NHIS), which was in place prior to the baby package intervention and was not modified for the study purposes. To assess the accuracy of the NHIS data, we compared the number of institutional births in the intervention district as registered by NHIS with the numbers of institutional births registered by the baby package registers. The discrepancies in the year 2010, 2011, 2012 and 2013 were 2.2%, -1.1%, 0%, and 0.003%, respectively. This indicates an overall good quality of NHIS data. The NHIS is routinely checked by the government HIS department. No other additional data checks were implemented for the purposes of this project. The number of institutional deliveries which occurred in each rural district was extracted from the NHIS, which defines institutional delivery as “giving birth in a public health facility with assistance from a trained provider”[14]. To calculate institutional delivery rates, we divided the numbers of institutional deliveries by the number of expected deliveries in each district. The Ministry of Health calculates this denominator by using the same yearly crude birth rate of 45 per 1000 which is multiplied by the estimated population for that year. These population estimates in turn are based on yearly projections based on the 2007 census published by the Instituto Nacional de Estatistica which are used by the National Health Information System. Given inaccuracies in these estimates, the number of recorded deliveries in the intervention district turned out to be greater than the number of expected deliveries, resulting in coverage of more than 100%. The denominator of expected deliveries does, however, provide a basis on which to measure the rates of institutional births, as there is no obvious reason why the error would differ between districts and therefore is not expected to introduce bias in the estimate for the effect of the intervention. In our analysis, we used the rate of institutional deliveries, rather than the proportion, since a proportion cannot be greater than 100%. The baby package intervention commenced on 1st June 2010. In order to divide the year into before and after the start of the intervention, the number of institutional births in 2010 was divided into January to May and June to December. The expected number of births for the two parts of the year was calculated assuming that there was no seasonality in the number of expected deliveries. Since the outcome of institutional delivery is a rate, we used Poisson regression to estimate the effect of the baby package on the number of institutional deliveries per expected birth. We checked the assumptions and the fit using residual plots. We allow for the background increase in institutional delivery rates in the 15 rural control districts. After graphical checks, we assumed a linear increase over time. To account for the clustering within districts, we include random effects for the districts for the intercept and slope which allows both the absolute starting value of institutional delivery rate and the background increase to vary by district. We adjusted for the potential confounders, the number of health facilities per district and the number of health personnel per district. We then estimated the additional increase associated with the intervention. Following graphical checks, we find that there is a sudden effect of the intervention and that this increase is maintained over time. The study is a retrospective analysis of data collected by the routine health registration system which consists of aggregated data over time. The implementation of the baby package incentive followed national recommendations and was distributed by the Ministry of Health. As such, it was part of a routine health service strategy implementation. No personal data were collected and/or included for the analysis. Ethical clearance was sought from the head of the ethics committee at the time of implementation. It was confirmed that no ethical approval nor consent was necessary for the analysis given the routine nature of the proceedings. Baby packages continue to be offered routinely to all women who give birth in a health centre in the district of Ancuabe as of the date of submission of this manuscript. The findings reported in this paper have been disseminated at provincial and national level on multiple occasions, in both scientific forums and Ministry of Health meetings. The study has been designed and implemented by the Swiss Non-Profit Organization SolidarMed in conjunction with the Mozambican Ministry of Health. SolidarMed has been working in Mozambique for 20 years collaborating with the Ministry of Health in order to improve health care in the province of Cabo Delgado. The baby package intervention was designed and implemented by SolidarMed and the Ministry of Health in partnership. There were no other partners. Data were not collected specifically for the intervention. All data analyzed were routinely collected across all districts through the Ministry’s routine health management information system and therefore could not be tampered with or altered in any way. There were no incentives or benefits to any partner, implementing health worker or individual health centre, which was linked to the success or failure of the intervention in any way. There is no known conflict of interest to be declared.
