Background: In Ethiopia, the proportion of mothers who attend the fourth antenatal care visit is lower than the proportion who attend the first visit. Although the reasons for these dropouts were investigated, few studies introduced interventions to promote the fourth antenatal care visit. Hence, the aim of this study was to assess the effectiveness of checklist-based box system intervention on improving fourth antenatal care visit. Method: This study employed a double-blind, parallel-group, two-arm cluster randomized controlled trial to compare the effectiveness of checklist-based box system intervention with the usual standard of care as a control arm. Study clusters are assigned to intervention and control arm in 1:1 allocation ratio using simple randomization technique. Pregnant mothers below 16 weeks of gestation were enrolled. Open data kit was used to collect data from the baseline and end-line surveys, and STATA version 15.0 was used to analyse the data. A difference-in-difference estimator was used to compare fourth antenatal care visit between the intervention and control groups across time. Mixed effect multi-level logistic regression was used to examine the relationship between the dependent and independent variables. Result: Data were collected from 2224 mothers who belong to 15 intervention and 15 control clusters. The difference in difference estimation resulted in a significant difference (26.1, 95%CI: 18–34%, p < 0.0001) between the intervention and control groups. Similarly, as compared to controls, the fourth antenatal care visit was found significantly higher in the intervention clusters (432 (85.2%) Vs. 297 (53.7%), p < 0.0001)/(AOR:5.69, 95% CI:4.14–7.82). Mothers who were knowledgeable about the services given during antenatal care visits (AOR: 2.31, 95% CI:1.65–3.24) and mothers who had a high level of social support (AOR:1.47, 95% CI: 1.06–2.04) were more likely to attend the fourth antenatal care visit. Conclusion: Implementation of checklist-based box system intervention resulted in a statistically significant effect in attendance of fourth antenatal care visit. Community-level variables were found to be more important in explaining variability in the fourth antenatal care visit. It is recommended that the intervention be implemented on a larger scale. Trial registration: ClinicalTrials.gov, Retrospectively registered on 26/03/2019, with trial registration number-NCT03891030.
The trial protocol for this study was published [18], and the trial was retrospectively registered on ClinicalTrials.gov, on 26/03/2019 with trial registration number, {"type":"clinical-trial","attrs":{"text":"NCT03891030","term_id":"NCT03891030"}}NCT03891030. A two-arm, double-blind, parallel-group cluster randomized controlled trial was conducted to evaluate the effectiveness of checklist-based box system intervention on improving attendance of fourth ANC visit. This study was conducted in East Gojjam zone, one of the administrative zones of Amhara region, located in North-western Ethiopia. According to 2019 Ethiopian mini Demographic and Health Survey (M-EDHS), the region’s maternal health service utilization for the first and fourth ANC was 82.6 and 50.8%, respectively [6]. According to the 2016 EDHS, more than half of (54.1%) women aged 15–49 years in the region had no formal schooling. Similarly, the majority of (83.5%) women of reproductive age (WRA) did not use any of the three media outlets (newspaper, radio, and television) at least once a week [5]. The sample size for this study was calculated following a recommendation for cluster randomized controlled trials with an equal-sized and fixed number of clusters [19]. This study is part of a larger study that included ANC, skilled delivery, and PNC. From the three options, the proportion of PNC was chosen to provide the largest sample size. Assumptions: the third postnatal care in the control group was 16.0% [20], number of clusters available – 30 (15 per cluster on each arm), 95% confidence interval and 80% power, the intra-cluster correlation coefficient of 0.04849 [21]. The sample size was then calculated to determine the number of observations per cluster using a two-sample comparison of proportion with normal approximations. STATA version 13 was used to run the calculation. Assuming individual randomization, the sample size per arm was 194. When the cluster randomization was used, the average cluster size required was 40, for a total sample size of 1200 pregnant mothers (600 from intervention and 600 from control). As a result, each baseline and end-line survey included 1200 mothers. The minimum detectable difference was set to be 12%. Debere-markos, Gozamin, and Machakel were chosen from among the sixteen districts available in East Gojjam Zone on the bases of confirming that none of the three districts had received an intervention/project aimed at improving utilization of maternal health services. The randomization units were health posts/Kebeles in selected districts. Individual randomization was not considered because a component of this trial (demand creation) was planned to be delivered at the community-level, which would result in information contamination. Following the completion of cluster recruitment, the study team used SPSS-generated random sequence to assign clusters to the intervention and control arms in a 1:1 allocation ratio. Participants in this trial were pregnant mothers under 16 weeks of gestation who lived in the designated districts. However, mothers who required hospitalization due to severe clinical complications and mothers who required a different type of ANC follow-up than the recommended focused ANC were excluded from the study. The screening and enrolment for this study were primarily handled by HEWs. HEWs are females who have completed the 10th grade. They are recruited from the community they serve and are deployed after a one-year formal training period. They are government employees who are primarily in charge of health promotion and prevention activities [22]. A family folder: a detailed record of household-level data at health posts was used to assist HEWs in locating WRA during a community-level survey. After being identified as WRA, mothers were subjected to a two-stage screening process before being enrolled in the study. First, HEWs performed community-level screening using Stanback et al., [23] pregnancy screening checklist. This checklist was used to identify suspected pregnant women. Mothers under and on 16 weeks of gestation were targeted in this community survey. Suspected pregnant women were given a referral slip, which was then given to a nearby health center; the visiting HEWs kept a copy of the referral slip for further follow-up. Second, for laboratory confirmation of pregnancy, a facility-level screening using beta-human chorionic gonadotropin (HCG) urine test was performed. Mothers who had a confirmed pregnancy were enrolled in the study and received their first ANC on the same day. The intervention was concealed from mothers and outcome assessors. Despite receiving all of the intervention’s listed packages, mothers were unaware that they were in a different treatment group and receiving a new intervention. Because this study used cluster randomization, mothers from the same catchment received the same intervention, making them unaware that they were taking part in a new intervention because those who needed the service were receiving it. In addition, the familiar HEWs provided the community-level interventions. Data collectors were unaware of the intervention, the intervention was delivered and data was collected by a separate group of health workers. Although intervention providers were aware of the intervention, data collectors were unaware of which groups were receiving the intervention and which were not. Furthermore, they are oblivious to the reason for the intervention and the objectives it is meant to achieve. To increase ANC service utilization, this trial used community-level demand creation and facility-level dropout tracing mechanisms (Fig. 1). Community-level surveys were carried out for two reasons: first, to identify suspected pregnant mothers and refer them to a nearby health center; and second, to identify factors that prevented mothers from attending ANC follow-ups at health facilities. Summary of the intervention: at community, health post and health center level The HEWs then identified the top three issues raised by mothers, documented using reason picking cards and provided problem-based health education. To make health education sessions more evidence-based, reason picking cards were used. The health education sessions held at the mother’s home were guided by addressing the most frequently mentioned reasons for maternal health service non-utilization. This data was gathered through community surveys, and each mother will receive at least three health education sessions throughout her pregnancy, delivery, and postnatal period. Mothers who did not attend any of the recommended visits, on the other hand, would receive additional health education in proportion to the number of visits missed (Fig. 2). A more detailed explanation about the intervention could be accessed from the study’s published protocol [18]. A 12-compartment health education scheduling box. For example, if a knowledge-related factor is raised as the most common reason for missing ANC visits, the reason picking card will be placed in C-4. Health education will be focused on compartments with more cards (C-Compartment, ANC- Antenatal care, PNC- Postnatal care) Once the mother was enrolled and received the first ANC at the health center level, the service utilization dropout monitoring box was used to track consecutive visits. Whenever the mother came to health facility and receive the service, her individual folder was moved to the next compartment across the health service monitoring box (Fig. 3). The timing of ANC visits from the first to the fourth ANC followed the World Health Organization (WHO) focused ANC model (i.e. first ANC (before 16 weeks of gestation), second ANC (24–28 weeks), third ANC (30–32 weeks), and fourth ANC (36–40 weeks). Dropouts were communicated to HEWs where the mothers belong by the midwife tracing that dropout. Service utilization monitoring box. The first compartment contains a copy of a suspected pregnant mother’s referral slip. The second compartment contains mothers who received the first ANC, the third compartment contains mothers who received the second ANC, and so on. A mother’s individual folder will be transferred to the second compartment if she received her first ANC visit. Missed mothers can be easily identified and communicated with during this process. (ANC stands for antenatal care, and PNC stands for postnatal care) Mothers in the control arm received the government’s existing routine care. In the routine maternal health service delivery, there was no community level screening (to improve early attendance at antenatal care) mothers visit health facilities for the first ANC based on their convenience, when mothers were appointed for a follow-up visit the mothers were given appointment cards for their next visit, there were no service utilization monitoring boxes to monitor service utilization dropouts, there was no knowledge gap assessment to be considered in health educations, the approach to awareness creation and health education mechanisms was in mass during pregnant mothers conference, and there was no individual based/person-based health education. A detailed intervention package outlining the steps to be taken while carrying out this trial was developed. Other supporting documents, such as training manuals, person-centered manuals for HEWs and manuals for health care providers (HCP), an orientation tool for local health administrators, and specifications for both boxes, were also developed. A sensitization workshop for local health administrators, and training for HEWs and HCP was held. The boxes were then purchased for health posts and health centers, and they were distributed along with the necessary number of printed checklists, referral slips, registers, and manuals. Then a community-level survey and enrollment process was initiated. On-site supervisions and follow-up visits were carried out based on the previously developed compliance parameter. The trial was originally scheduled to last for ten months, however from the time the first mother with a confirmed pregnancy enrolled in the study and received the first ANC at the health center on January 3rd, 2019, to the end of follow-up for the last rounds of mothers on August 27th, 2020, it took around 20 months. This extended timeframe was due to the following challenges. The family folder at the health post level was not updated; additional efforts were made to locate WRA who were not on family planning. As HEWs delivered a component of this intervention, there were occasions such as annual leave, maternity leave, and transfer to other areas of the trained HEWs. This was a critical issue on health posts where there was only one HEW on staff. We used the existing weekly meeting platform between health centers and health posts for reporting suspected pregnant mothers (to send a copy of the referral slip from health post to health center). However, these meetings were not held regularly, especially for rural health facilities, and HEWs were sometimes absent from weekly meetings. Even though only a few participants experienced this, some mothers had their HCG tests rescheduled due to the test’s unavailability. A structured and pre-tested questionnaire was used to collect data. The data collection questionnaire was developed after reviewing relevant literature, national [24], and international standards [25]. This trial used the PRECEED-PROCEED model as a lens [26]. This model emphasizes fundamental concepts such as health information sources and social forces as important environmental influences on health behaviour, personal desire variation may necessitate individualized care, and health behaviours that are unacceptable to society should be approached with caution [27, 28]. The English questionnaire was translated into Amharic (local language of the study area) for the data collectors. The Amharic version of the questionnaire has been uploaded to the kobo-tool box. Then, for the baseline and end-line studies, an open data kit (ODK), a software that uses mobile devices to collect and submit data to an online server, was used to conduct face-to-face interviews. The primary goal of this trial was to determine the effectiveness of checklist-based box system intervention on improving maternal health care utilization (ANC, skilled delivery and postnatal care). The findings in this study were focusing on the first outcome: improving attendance of the fourth ANC visit. As a result, the fourth ANC visit status was compared between the intervention and control clusters. Following that, additional factors influencing the use of the fourth ANC visit were investigated. For this analysis, ‘fourth ANC visit was identified as a dependent variable. Similarly, kebele was taken as a clustering variable, and then the independent variables were categorized as level 1 (individual-level variables) and level 2 (community and facility-level variables) variables. A more detailed list of variables with their description and measurement was included in Table 1. Description of study variables, East Gojjam Zone, Northwest Ethiopia, January 2019–September 2020 Fourth antenatal care visit – Fourth ANC visits in this study refer to a pregnancy check-up achieved when a pregnant woman receives care at all four appointments. Knowledge of ANC service – This variable assessed participants’ understanding of the services required and provided during pregnancy. A composite index of eight variables (detailed in Table Table1)1) was developed and dichotomized using the mean score: those who scored the mean or higher were classified as ‘knowledgeable,’ while those who scored below the mean were classified as ‘not knowledgeable.’ Knowledge of danger signs of pregnancy – Participants’ were asked to make a list of potential danger signs that could occur during pregnancy. A composite index of eleven variables (detailed in Table Table1)1) was created and dichotomized using the mean score: those who scored the mean or higher were classified as ‘knowledgeable,’ while those who scored below the mean were classified as ‘not knowledgeable.’ Gestational age at first ANC – Participants were asked how far along they were in their pregnancy at their first ANC check-up. This was recoded as ‘less than or equal to 16 weeks’ and ‘greater than 16 weeks of gestation. Social support– Social support was measured using the following 14 item questions with ‘yes’ or ‘no’ response categories: ‘gets visits from significant others’, ‘getting useful advises’, ‘discussion on problems’, ‘having care at the time of labor and delivery, ‘feeling loved’, ‘others thankful on them’, ‘getting help on household chores’, ‘help with money at emergency’, ‘help in transportation’, ‘help when sick’, ‘attending community level discussions’, ‘member of any religious cast’, ‘attending public meetings’ and ‘help in case of conflicts’. Following that, a principal component analysis was performed, and a composite index was constructed using the principal components. This was dichotomized using the mean score: those who scored the mean or higher were classified as having ‘high social support,’ while those who scored below the mean were classified as having ‘low social support.’ (Cronbach’s alpha = 0.81). Before starting field data collection, training for data collectors and supervisors was provided to create a shared understanding of the study tools. A manual explaining question types and response categories and how to use ODK was produced and distributed to data collectors (BSC-holder midwives) and supervisors (MPH holders). The use of ODK aided in setting questions ‘required’ (to avoid unanswered questions), applying range checks for selected data values, and handling field editing before leaving the respondent. In addition to the field team, data collection was managed directly from Jimma University, where all study databases were secured with a password-protected access system. Following field data collection, the data were exported from the kobo-tool box and imported into STATA MP Version 15 for analysis. Data from the trial were analysed using intention to treat analysis. Participants were assigned to clusters based on where they lived at the start of the trial. Because the baseline outcome did not determine the cluster allocation, the difference in difference (DiD) estimator was used to compare the status of the fourth ANC visit between the four contact points (baseline vs. end-line and intervention vs. control) [29]. Similarly, the status of the fourth ANC between the intervention and control arms was compared using the chi-square test of association. Then, to identify factors influencing fourth ANC visit, a bivariate analysis was performed to test the relationship between each independent variables and the outcomes variable, then variables with p < 0.25 were included in the multivariable model. Analysing different levels of factors at a single level using ordinary logistic regression results in loss of power and introduces Type I error. The presence of clustering (individuals are nested in kebeles in this case) also violates the assumption of independence among study participants. As a result, this study employed a mixed effect, multi-level logistic regression model. The multi-level analysis allows for the simultaneous examination of the effects of group and individual level variables on individual-level outcomes while accounting for non-independence of observations within groups. Similarly, this model considers individual probability, which is statistically dependent on the participants’ place of residence. This context dependence was taken into account to obtain accurate regression estimates [30]. The presence of cluster-level variability influencing fourth ANC visit was then tested using the intercept-only model and ICC [31]. In addition, the variation between clusters was measured using the median odds ratio (MOR) [32–36] and proportional change in variance (PCV) [37]. The ICC is used to calculate the proportion of total variance in the outcome attributed to the area level, whereas MOR is used to calculate unexplained cluster heterogeneity [34]. Then, model fitness for the multi-level model was tested by using the log-likelihood ratio (LR) test. Four models were constructed during the analysis: the first model was an empty model that was used to determine how much cluster variation influenced fourth ANC visit. The second model adjusted for individual-level variables, the third model was adjusted for community-level variables and the fourth for both individual and community-level variables. A bivariate analysis was performed first, and variables with p < 0.25 were included in the second and third models. Then, variables with a p < 0.05 in the second and third models were included in the final model. The p-value of < 0.05 was used to define statistical significance, AOR together with 95% CI were used to show the strength of association and level of significance, respectively.
N/A
DIMA AI Care
