Provision of medical supply kits to improve quality of antenatal care in Mozambique: a stepped-wedge cluster randomised trial

Background High levels of maternal and newborn mortality and morbidity remain a daunting reality in many low-income countries. Several interventions delivered during antenatal care have been shown to improve maternal and newborn outcomes, but stockouts of medical supplies at point of care can prevent implementation of these services. We aimed to evaluate whether a supply chain strategy based on the provision of kits could improve quality of care. Methods We did a pragmatic, stepped-wedge, cluster-randomised controlled trial at ten antenatal care clinics in Mozambique. Clinics were eligible if they were not already implementing the proposed antenatal care package; they served at least 200 new pregnant women per year; they had Maternal and Child Health (MCH) nurses; and they were willing to participate. All women attending antenatal care visits at the participating clinics were included in the trial. Participating clinics were randomly assigned to shift from control to intervention on prespecified start dates. The intervention involved four components (kits with medical supplies, a cupboard to store these supplies, a tracking sheet to monitor stocks, and a one-day training session). The primary outcomes were the proportion of women screened for anaemia and proteinuria, and the proportion of women who received mebendazole in the first antenatal care visit. The intervention was delivered under routine care conditions, and analyses were done according to the intention-to-treat principle. This trial is registered with the Pan African Clinical Trial Registry, number PACTR201306000550192. Findings Between March, 2014, and January, 2016, 218 277 antenatal care visits were registered, with 68 598 first and 149 679 follow-up visits. We found significant improvements in all three primary outcomes. In first visits, 5519 (14·6%) of 37 826 women were screened for anaemia in the control period, compared with 30 057 (97·7%) of 30 772 in the intervention period (adjusted odds ratio 832·40; 99% CI 666·81–1039·11; p<0·0001); 3739 (9·9%) of 37 826 women were screened for proteinuria in the control period, compared with 29 874 (97·1%) of 30 772 in the intervention period (1875·18; 1447·56–2429·11; p<0·0001); and 17 926 (51·4%) of 34 842 received mebendazole in the control period, compared with 24 960 (88·2%) of 28 294 in the intervention period (1·88; 1·70–2·09; p<0·0001). The effect was immediate and sustained over time, with negligible heterogeneity between sites. Interpretation A supply chain strategy that resolves stockouts at point of care can result in a vast improvement in quality during antenatal care visits, when compared with the routine national process for procurement and distribution of supplies. Funding Government of Flanders and the UNDP/UNFPA/UNICEF/WHO/World Bank Special Programme of Research, Development and Research Training in Human Reproduction. We did a pragmatic, facility-based, cluster randomised controlled trial with a stepped-wedge design in which participating antenatal care clinics were randomly assigned to shift from control to the intervention on prespecified dates. After 3 months of baseline data collection in all clinics, the intervention was rolled out sequentially to a new clinic every 2 months (length of the step). The trial lasted 23 months, including the 3 months of baseline data collection. The methods of the study are described in detail in the study protocol.14 Clusters were antenatal care clinics in health facilities selected purposely by the Ministry of Health according to its programmatic activities and priorities and with geographical representation of the three regions of Mozambique (north, centre, and south). Antenatal care clinics were eligible if they were not already implementing the proposed antenatal care package; they served at least 200 new pregnant women per year; they had Maternal and Child Health (MCH) nurses; and they were willing to participate. All women attending antenatal care visits at the participating clinics were included in the trial. The ten participating clinics have been described in the study protocol14 and are listed in the appendix. All clinics were Ministry of Health facilities and provided antenatal care free of charge under the public health system in which patients do not pay for maternal and child health care. All clinics provided the same standard of care in accordance with national Ministry of Health guidelines for antenatal care at health centre level. Two of the clinics (3-Songo and 5-Matola) received some additional resources from non-governmental partners. Clinics were assigned to one of ten start dates by the study statistician via a computer-generated list of random numbers. Concealment of the intervention start date was not possible for logistic reasons, because of the involvement of the Ministry of Health in the preparatory activities required to launch the intervention at each clinic. The intervention was multifaceted with four components described in the panel, which pertain to the cluster level: antenatal care kits (boxes containing supplies necessary to carry out a specific number of antenatal care visits); a cupboard to organise and store the supplies locked in the antenatal care room; a tracking sheet to monitor the kits' stock levels; and a training session for the health-care providers at the beginning of the intervention. Component 1: antenatal care kits containing the necessary medicines, laboratory supplies, materials, and equipment Four different kits were designed. Each health facility was provided with the antenatal care kits, which include the commodities required for first and follow-up antenatal care visits in accordance with Ministry of Health guidelines. Component 2: cupboard A cupboard for storage of the kits was provided in the room where the antenatal care visits took place, which allowed the nurses to have easy and quick access to all necessary materials during the antenatal care visit, and ensured secure storage of items. Component 3: tracking sheet A tracking sheet was introduced to monitor use of kits and stock levels to avoid stockouts. Component 4: training session At the start of the intervention in each site, a refresher course on the essential interventions for antenatal care was held with all the Maternal and Child Health (MCH) nurses in the health facilities involved in delivering antenatal care, as well as the pharmacist and laboratory technician. These nurses were also trained in how to use the contents of the kit, such as the proteinuria test and blood pressure measurement, and in procedures to ensure continuous supply of the medicines and materials needed. The kits were designed by two experts in procurement and supply chain management on the basis of the evidence-based antenatal care interventions in the national guidelines (see the appendix for the list of interventions). Because of the importance of the first antenatal care visit, the trial differentiated between first and follow-up antenatal care visits (visits other than the first). We designed four types of kits: Kit A contained supplies necessary to carry out 100 first antenatal care visits; Kit B contained supplies necessary to carry out 200 follow-up antenatal care visits; Kit C contained urine collection containers; and Kit D contained long-lasting impregnated bednets. For this trial, Kit A and Kit B were procured in the Netherlands and imported as ready-made boxes in accordance with national regulatory importing procedures. Kit C was purchased in Mozambique, and Kit D and antiretroviral drugs were provided through the Ministry of Health. The International Centre for Reproductive Health in Mozambique, the local implementing partner, was responsible for importation, storage, and distribution of the kits. Automatic sphygmomanometers with rechargeable batteries were part of the intervention but were not packed inside the kits because they were a one-time delivery. The full list of the products contained in Kits A and B, and photos of the kits, can be found in the appendix. Data collection started in March, 2014, in all antenatal care clinics. The first clinic entered the intervention in June, 2014, and the last in December, 2015. The dates of the launches of the intervention in each facility are shown in the appendix. Before the day of the launch, the local coordination team would be deployed to the clinic for preparation of the intervention. The kits were received, stored, and organised, the cupboard for storage was set up, and the antenatal care room or rooms were rearranged to accommodate changes in patient flow required by the introduction of the intervention. The training session was carried out with attendance of all MCH nurses involved in antenatal care services at the clinic, as well as laboratory and pharmacy technicians. The content of the training is outlined in the panel. On the day of the launch, the nurses would start by addressing the awaiting pregnant women to explain the new system of kits and their contents, the antenatal care practices they were going to receive, the importance of these practices, and the advantages expected. This process was implemented ten times—before the launch of the intervention in each facility. An official from the Ministry of Health was actively involved in every step of the trial, including the launches and the monitoring visits. Nurses were supervised by the Ministry of Health in the context of their routine monitoring activities. During the control period, each clinic provided antenatal care according to standard practice and functioned under the regular procurement system. The components of the antenatal care visits were the same in both groups, as per national guidelines (appendix). The kits contained, among other items, rapid tests for HIV, syphilis, haemoglobin, and proteinuria. Rapid tests for HIV and syphilis were already a routine part of antenatal care services before the intervention, whereas proteinuria and haemoglobin tests were only done at the laboratory. Ten practices for women attending the first antenatal care visit were initially targeted by the intervention and included in the protocol as primary outcome candidates.14 The protocol prespecified that three among those ten practices would be selected as primary outcomes, on the basis of the analysis of outcome delivery rates with data collected before the implementation of the intervention (ie, in step 1). This analysis was done during step 2 and the results are presented in the appendix. From the initial list of outcomes, iron/folic acid supplementation and tetanus toxoid administration were excluded before the analysis was done because the information required to compute these variables was not recorded in the antenatal care logbook. Screening for hypertensive disorders was split into its two elements: screening for high blood pressure and screening for proteinuria. These changes led to the final list of nine outcomes that were evaluated for the first antenatal care visits. The full list of outcomes and definitions is provided in the appendix. Among these nine practices, the three with the lowest delivery rate during step 1 were selected as primary outcomes: screening for proteinuria, screening for anaemia, and treatment of parasitic worms with mebendazole. The other six practices (screening for high blood pressure, preventive treatment for malaria, screening for HIV, treatment for HIV, screening for syphilis, and treatment for syphilis) were selected as secondary outcomes. Secondary outcomes also included delivery of antenatal care practices in follow-up visits. These practices are described in the appendix. A composite outcome was also prespecified in the protocol. All outcomes were measured with routine data extracted from the antenatal care logbook. Antenatal care nurses are required to register all antenatal care visits and practices delivered in each visit in standardised logbooks provided by the Ministry of Health (appendix). Although the Ministry of Health compiles data from monthly summaries generated by the nurses on the basis of the data in the logbooks, no system is in place for systematic digital storage of this information. Women do not have clinical records, and the logbook was the only source document for the trial. Data management procedures were developed and implemented in all ten participating antenatal care clinics with the purpose of transferring data in the logbook to the data management centre. Nurses were trained in how to complete the logbooks correctly by use of the coding system. Research assistants hired for the trial regularly reviewed the logbooks in each antenatal care clinic, took digital photos of each page of the logbook, and sent them to the data management centre in Maputo. We originally intended to link first and follow-up antenatal care visits for each woman by assigning a unique study participant ID at enrolment; however, implementation of such a system was challenging in practice and it is only available for a subset of the data. Data management and quality control procedures have been described in the protocol14 and are summarised in the appendix. These procedures included several data quality audits and monitoring activities. The Ministry of Health provided the number of antenatal care visits in each selected clinic for 2011, which was used to calculate the sample size, as reported in the study protocol.14 We assumed, conservatively, a baseline frequency of 30% for each selected health practice, and an increase to 60% with implementation of the intervention. For a 0·05 alpha level, 80% power, and an intra-cluster correlation coefficient of 0·05, six clusters were needed. To protect against pre-randomisation exclusions and dropouts, we decided to include ten clusters (antenatal care clinics). All sites completed the study and it is therefore overpowered for these initial assumptions. Analyses were done according to the intention-to-treat principle.15 Because we prespecified three primary outcomes, the level of significance was adjusted for multiple comparisons and set to 0·016 rather than 0·05. For dichotomous outcomes, a multilevel logistic regression was used to estimate odds ratios [OR] and 99% CI for exposure to the intervention across datapoints during the intervention period compared with the pre-intervention period. For the composite outcome score, mean differences and 99% CI were computed by use of multilevel linear regression.15 The clinic was entered in these models as a random effect. Effect sizes were adjusted for time trends by including time in the model as a fixed effect. This trial is registered with the Pan African Clinical Trial Registry, number PACTR201306000550192. The funders of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The corresponding author, APB, EB, SG, MHN, and SM had full access to all the data in the study. All authors had final responsibility for the decision to submit for publication.