Study design
A cluster randomized, controlled trial was conducted to assess the effect of community delivery of IPTp (cIPTp) by CHWs on the coverage of IPTp and ANC. Following a baseline survey in December 2017, cIPTp was implemented in intervention sites within two districts of Malawi over a 21-month period, from November 2018 to July 2020. In control sites, IPTp was delivered at health facilities via ANC clinics, per routine practice. The endline survey was carried out in August 2020.
Study area
The study was conducted in the districts of Ntcheu (population: 270,903), Central Region, and Nkhata Bay (population: 206,670), Northern Region, Malawi (Fig. 1). These districts were purposively selected from among the 10 districts in Malawi where the U.S. President’s Malaria Initiative (PMI) supports malaria control activities. Malaria is considered a major public health problem throughout the year in both places, with peak transmission during the rainy season from November to March. The vast majority of infections are caused by Plasmodium falciparum [9].
Both Ntcheu and Nkhata Bay are rural areas where most families are subsistence farmers. Nkhata Bay also has a small tourism economy. The median age of marriage for women in both districts is 18 years old, while the median age at first birth is 19 years old. Women in both districts are equally likely to deliver at a health facility and with a nurse or midwife present [16]. Ntcheu has almost twice as many health centers and ANC providers as Nkhata Bay (40 vs 22 and 55 vs 29, respectively), but only about 30% more pregnant women (13,544 vs 9,905 annually) (Table 1).
Table 1
Characteristics of the study districts
District
|
Public health centers providing ANC
|
ANC staff
|
Community health workers
|
Total population
|
Pregnant population
|
IPTp2+ (%)a
|
IPTp3+ (%)a
|
ANC1+ (%)a
|
ANC4+ (%)a
|
Nkata Bay
|
22
|
29
|
87
|
206,670
|
9,905
|
65.4
|
31.3
|
98.1
|
57.6
|
Ntcheu
|
40
|
55
|
81
|
270,903
|
13,544
|
60.5
|
26.6
|
94.2
|
45.9
|
ANC antenatal care, IPTp intermittent preventive treatment in pregnancy |
a National Statistical Office (NSO) and ICF Macro: Malawi demographic and health survey 2015-2016. Zomba, Malawi and Calverton, MD, USA 2017 |
CHWs in Malawi are the lowest cadre of service providers within the Ministry of Health. All CHWs receive a basic six-week training. Some receive further specialized trainings to provide specific services, such as integrated Community Case Management (iCCM), Community-Based Distribution of Contraceptives, Community Case Management of Acute Malnutrition, or Community-Based Maternal and Newborn Health (CBMNH). This study utilized CHWs trained in CBMNH.
Compared with Nkhata Bay, Ntcheu has fewer CHWs overall and per facility, and on average, each CHW in Ntcheu serves a larger population than the CHWs in Nkhata Bay. Although Ntcheu has fewer CHWs per facility, there are additional volunteers in the community called Secret Mothers. These volunteers receive minimal (one day) training in basic community maternal and newborn health concepts such as counselling of pregnant women on ANC attendance, use of mosquito nets, nutrition and birth preparedness, and their primary function is to encourage women to attend ANC in the first trimester.
Intervention
In the intervention arm, all CHWs received a three-day training on how to provide IPTp to pregnant women who had already received their initial dose at the health facility, in addition to refresher training on community-based maternal and newborn health. Training covered eligibility, administration, and record keeping for the study registers and ANC cards. CHWs were encouraged to promote utilization of IPTp and ANC attendance during routine community meetings and home visits. They were also encouraged to work with community leaders, supervisors and health facility staff to identify pregnant women and organize group sessions on the importance of ANC visits and malaria prevention, including IPTp. A total of 72 CHWs were trained (49 males and 23 females). Amongst the 60 CHWs for whom data were available, the mean age was 43.4 years (standard deviation=6.8) and had worked as CHWs for an average of 17.5 years (standard deviation=5.0).
In addition, study personnel conducted a one-day orientation for staff at the intervention health facilities and members of the Area Development Committees to explain the objectives of the study and how CHWs would carry out their duties. Staff at the intervention health facilities were also trained to identify which pregnant women received IPTp in the community, record this in the health facility register, and report this information within the District Health Information Software (DHIS2) on a monthly basis. During the second half of implementation, the study team collaborated with local non-governmental organizations to further develop and disseminate educational messages about cIPTp.
Study personnel conducted monthly supervision at all intervention health facilities. During these visits, they met with the CHWs and reviewed their registers for completeness and accuracy, answered questions, and re-stocked SP supply. Review meetings were held in each district on approximately a quarterly basis, bringing together all CHWs to highlight successes and challenges, and collectively discuss solutions.
Control
In the control arm, IPTp was administered exclusively at ANC by trained health facility workers, usually nurses. All CHWs received training on promotion of ANC attendance and prevention of malaria in general, in addition to refresher training on community-based maternal and neonatal health. They were not trained in cIPTp. Quarterly visits to control facilities were conducted to collect routine data on ANC attendance and IPTp administration.
Selection and sample size for the cross-sectional survey
The impact of the intervention was assessed by baseline and endline cross sectional surveys, including women between the ages of 16–49 years who had a pregnancy resulting in a live birth in the previous 12 months (recently pregnant women). A three-stage cluster sampling procedure was used to select survey respondents. After excluding district hospitals, non-governmental facilities, facilities that do not provide ANC, and facilities accessible only by boat, 10 health centers were randomly selected in each district (20 in total). First, the catchment area of each health center was designated as a cluster. Depending on geographical size, each cluster contained 3–18 enumeration areas (EAs). EAs are administrative data collection units, demarcated by the National Statistics Office, with an average of 250 households or 1,000 people. In the second stage, a single EA was randomly selected from each cluster using probability proportional to size. Finally, all households within the selected EA were listed, noting households with recently pregnant women. At baseline, simple random sampling was used to select 20 households with recently pregnant women. At endline, 40 households with recently pregnant women were randomly selected. If a selected household had more than one recently pregnant woman living in it, all women who met the inclusion criteria were included in the survey.
A sample size of 300 women (15 women per health facility catchment area) was required for each survey to achieve 80% power to detect a 30%-point difference between the group proportions for IPTp3+ coverage. These calculations assumed a baseline proportion of 30% IPTp3+ coverage (average proportion in the two districts) to an endline proportion of 60% coverage in the intervention arm. Sample size was estimated using PASS V14 (NCSS, LLC, Kaysville, UT) to assess sample size for cluster randomized trial, at a significance level of 0.05 and an intra-cluster correlation of 0.2. Sample size was then increased to allow for 30% non-response/ unavailability of the woman or ANC card which was used to assess the primary outcome measure.
Training for cross sectional surveys
Twenty-eight enumerators took part in each of the baseline and endline cross-sectional household surveys. At baseline, all enumerators participated in a four-day training, consisting of two days of classroom instruction on ethical research conduct, consent, use of the enumeration program, and review of the questions and sampling program. This was followed by one day of piloting the survey in one of the non-selected EAs, and a final debriefing day, during which minor updates were made to several questions to improve the readability and clarity. At endline, an effort was made to include as many of the original data collectors as possible, and training was conducted over three days.
Data collection and management
Cross-sectional survey
Surveys were conducted using Open Data Kit (ODK) forms hosted on the SurveyCTO platform (Dobility, Inc., Cambridge, MA). Files from SurveyCTO were exported into SAS V9.4 (SAS, Cary, NC) and STATA (STATACorp, College Station, TX) for analysis. The database was programmed with range checks and key fields were required. Data were checked daily for completeness by field supervisors before uploading to a central server. In addition, throughout the survey period, data checks for key variables were run on data downloaded from the SurveyCTO platform to ensure loops were working as expected.
Routine Data
Visits to study and non-study health facilities were made by study supervisors monthly. During these visits, health facility and HSAs registers were reviewed and monthly data on ANC attendance, IPTp uptake, SP stocks, and other administrative data were retrospectively abstracted.
Data analysis
A descriptive analysis was performed of the socio-demographic characteristics of recently pregnant women in the baseline and endline surveys, stratified by control and intervention areas.
The primary outcome was the change in IPTp3+ coverage over time, assessed using a difference-in-differences analysis comparing the baseline and endline cross-sectional survey data. Further difference-in-differences analyses were conducted to assess secondary outcomes, including any, two, three, or four or more ANC visits (ANC1+, ANC2, ANC3, ANC4+), IPTp1+, 2+, 4+, and 5+ coverages, number of IPTp doses, and timing of ANC initiation. Both primary and secondary outcomes were primarily based on what was recorded on the ANC card, up to a maximum of five IPTp doses. Self-reported outcomes were used only when the values from the ANC card were missing.
For binary outcomes (e.g., IPTp3+, ANC4+), difference-in-differences were calculated using a binary logistic regression model with an identity link function. Poisson regression was used for continuous outcomes (e.g., number of IPTp doses, timing of ANC initiation). In order to obtain 95% confidence intervals for the difference-in-differences, a linear model was run using the GENMOD procedure, including all the same terms as in the logistic model. All analyses accounted for clustering at the health facility level and the interaction between intervention and time. Adjusted models included gravidity, maternal age, and maternal education as additional covariates as these factors have been identified by previous studies as important determinants of ANC service use [10].
Several secondary analyses were also conducted to understand the impact of IPTp and some dynamics that could affect operationalization of the intervention. First, crude and adjusted odds of low birthweight were calculated for all women who received IPTp3+, compared with women who received fewer than three IPTp doses. The adjusted odds included gravidity, maternal age, maternal education and ANC4+ as covariates. Results were calculated for the overall sample, as well as stratified by district and survey round (baseline and endline). Second, community perceptions surrounding IPTp and ANC were summarized descriptively using several questions from the endline survey. Third, differences in IPTp3+ uptake by study arm were evaluated amongst the subset of women who lived at least five kilometers from a health facility to assess if the intervention may be more impactful in remote settings. Distance to facility was only available at endline so this analysis was based on a simple chi-square. Fourth, monthly routine service data from health facilities and visit logs from CHWs were assessed to identify potential implementation gaps.
All analyses were run in SAS Version 9.4 (SAS Institute Inc., Cary, NC) and p-values of less than 0.05 were considered statistically significant.
Ethics
The protocol was reviewed and approved by the Malawi College of Medicine Research Ethics Committee (COMREC). Representatives from the Government of Malawi Ministry of Health were involved throughout the design and implementation of the survey, and permission was obtained from the officials in each district health office prior to initiating the survey. In each EA, permission was obtained from the village leaders. Written informed consent was obtained from each respondent before data collection. Participants were told that the survey was focused on antenatal care and malaria prevention in pregnancy. The Centers for Disease Control and Prevention (CDC) Human Subjects Office determined that CDC staff were not engaged in human subjects research for this project.