Assessing the Integrated Community-Based Health Systems Strengthening initiative in northern Togo: A pragmatic effectiveness-implementation study protocol

Background: Over the past decade, prevalence of maternal and child morbidity and mortality in Togo, particularly in the northern regions, has remained high despite global progress. The causes of under-five child mortality in Togo are diseases with effective and low-cost prevention and/or treatment strategies, including malaria, acute lower respiratory infections, and diarrheal diseases. While Togo has a national strategy for implementing the integrated management of childhood illness (IMCI) guidelines, including a policy on integrated community case management (iCCM), challenges in implementation and low public sector health service utilization persist. There are critical gaps to access and quality of community health systems throughout the country. An integrated facility- and community-based initiative, the Integrated Community-Based Health Systems Strengthening (ICBHSS) initiative, seeks to address these gaps while strengthening the public sector health system in northern Togo. This study aims to evaluate the effect and implementation strategy of the ICBHSS initiative over 48 months in the catchment areas of 21 public sector health facilities. Methods: The ICBHSS model comprises a bundle of evidence-based interventions targeting children under five, women of reproductive age, and people living with HIV through (1) community engagement and feedback; (2) elimination of point-of-care costs; (3) proactive community-based IMCI using community health workers (CHWs) with additional services including family planning, HIV testing, and referrals; (4) clinical mentoring and enhanced supervision; and (5) improved supply chain management and facility structures. Using a pragmatic type II hybrid effectiveness-implementation study, we will evaluate the ICBHSS initiative with two primary aims: (1) determine effectiveness through changes in under-five mortality rates and (2) assess the implementation strategy through measures of reach, adoption, implementation, and maintenance. We will conduct a mixed-methods assessment using the RE-AIM (reach, effectiveness, adoption, implementation, maintenance) framework. This assessment consists of four components: (1) a stepped-wedge cluster randomized control trial using a community-based household survey, (2) annual health facility assessments, (3) key informant interviews, and (4) costing and return-on-investment assessments for each randomized cluster. Discussion: Our research is expected to contribute to continuous quality improvement initiatives, optimize implementation factors, provide knowledge regarding health service delivery, and accelerate health systems improvements in Togo and more broadly. Trial registration: ClinicalTrials.gov, NCT03694366, registered 3 October 2018

The objective of this study is to optimize ICBHSS model implementation using the adapted RE-AIM evaluation framework. To achieve this, we have two primary objectives: (1) to determine the effectiveness of the ICBHSS model and (2) to assess the implementation strategy through measurements of reach, adoption, implementation, and maintenance. Our specific study aims include the following: This study uses a pragmatic type II hybrid effectiveness-implementation design [35] to evaluate the two primary aims of effectiveness and implementation strategy by the ICBHSS initiative using a modified RE-AIM implementation science framework [42]. See Additional file 1 for the CONSORT checklist. We will include four distinct study components: (1) a stepped-wedge cluster randomized control trial using a community-based household survey, (2) annual health facility assessments at each selected site, (3) qualitative key informant interviews conducted 1-year post-intervention reception, and (4) annual costing and return-on-investment analyses using the Community Health and Costing Tool [48] and the Lives Saved Tool (LiST) [49]. Further details about each study component are described below with Table 2 summarizing a timeline. Data collection and Integrated Community-Based Health Systems Strengthening (ICBHSS) initiation timeline based on staggered implementation* *Follows the CONSORT extension diagram for stepped-wedge cluster randomized trials [52] **CH community-based household survey, Fac health facility assessments, KIs key informant interviews, Cost survey costing and return-on-investment assessment This pragmatic design leverages the sequential or staggered rollout of the ICBHSS model and will facilitate an assessment of effectiveness as well as implementation strategy through coverage and adoption metrics by comparing geographically organized clusters [50, 51]. It includes four clusters that organize intervention health facilities by district: Bassar, Binah, Dankpen, and Kéran. The ICBHSS cluster initiation order will be randomized annually with four steps, as it was independently determined by the IH programs team that baseline health needs were similar in each cluster. Community-based household surveys adapted from Demographic and Health Survey (DHS) modules previously implemented in Togo [3] and focusing on demographic, maternal, and child health data will be conducted in each cluster at baseline, 12, 24, 36, and 48 months. These assessments will employ facility-level surveys based on the World Health Organization (WHO) Service Availability and Readiness Assessment tool (SARA) [41] and will provide effectiveness information about facility-level service quality. Surveys will be completed for each health facility annually at baseline, 12, 24, 36, and 48 months. Qualitative interviews will be completed with key informants to assess barriers and facilitators to program implementation fidelity and feasibility while also documenting contextual factors. The first key informant interviews will be conducted 12 months post-intervention start and at subsequent 12-month intervals until study end within each cluster. ICBHSS program costs and return on investment will be measured using the Community Health Planning and Costing Tool [48] and the Lives Saved Tool [49] to assess implementation strategy approaches and to inform considerations of maintenance and national planning efforts. The first assessment will be conducted 12 months post-intervention start and at annual subsequent 12-month intervals until study end within each cluster. The study will be conducted in the catchment areas of 21 public sector health facilities within the Kara region’s rural districts of Bassar, Binah, Dankpen, and Kéran. The total population is approximately 181,111 people. Study sites were selected by MoH partners and IH programmatic staff based on perceived population health needs, ongoing regional public-private program activities, and population size. All selected sites are primary healthcare facilities operated by the MoH [53] that serve rural populations. Estimated catchment population and utilization rates for these sites are listed by district in Table 3. As described in Table 2, the ICBHSS initiative will be sequentially implemented by district each year within these 21 preselected sites. List of study sites (N = 21) with estimated baseline catchment population (N = 181,111) and facility utilization rates *Data derived from 2016 Ministry of Health population estimates and 2018 Integrate Health baseline population-based sampling **Data source is 2016 Ministry of Health district-level annual reports Eligibility for inclusion is described below by study component for the community-based household surveys, facility surveys, and qualitative interviews. Females 15–49 years of age who reside in a selected household within the study catchment area are included. Participants 15–17 years of age will only be included if they have been/are currently pregnant and have a waiver of parental permission. Households will be randomly selected each year using a systematic population-weighted sampling methodology. If there is more than one eligible respondent in the household, one will be randomly selected based on a Kish selection grid [54]. All participants will have the opportunity to decline participation during the informed consent process. All 21 selected health facilities will be surveyed. See Table 3 for facility details. This component will include individuals 18 years of age or older who are either implementing partners or intervention health facility staff. Implementing partners are defined as IH programmatic staff members employed in the catchment area of interest. Health facility staff are defined as clinical or administrative MoH employees working at one of the 21 health centers. Approximately 42 key informants from the 21 health facilities (21 implementing partners and 21 clinical health facility staff) will be included. This study is powered to detect a change by cluster in under-five mortality per 1000 live births. A sample size of 7600 participants will provide 80% power to detect an estimated 30% reduction or greater of under-five mortality from the estimated baseline of 70 per 1000 live births, with an alpha of 0.05, intracluster correlation of 0.005, 20% non-response rate, and estimated 0.5 children under five per participant [9, 55]. Effect size is a conservative estimate based on past pilot experience (ClinicalTrials.gov Identifier: {“type”:”clinical-trial”,”attrs”:{“text”:”NCT03773913″,”term_id”:”NCT03773913″}}NCT03773913). The cluster order for implementation at each step will be determined randomly by an external technical advisor using a random number generator. Randomization will occur each year 8 months prior to the rollout of the intervention in the next cluster. This will enable blinding to the random order of clusters for IH and MoH staff 7involved in implementation while also allowing for an annual 8-month planning stage prior to start of the intervention. Each cluster represents a district, with a total of 21 preselected health facilities across each of the four districts. All metrics will be organized using a modified RE-AIM (reach, effectiveness, adoption, implementation, maintenance) framework [56]. Table 4 summarizes data collection and analysis plans organized by primary aim and adapted RE-AIM domains. Summary of key study measures organized by aim and domain using modified RE-AIM evaluation framework Changes in childcare-seeking behavior over time for fever, pneumonia, and diarrhea in patients presenting to health clinic, CHW, or non-clinical site. Test if these proportions increased using the same approach to mixed-effects generalized linear models as described in the primary effectiveness outcome measure. We define effectiveness metrics as those that assess the impact of the ICBHSS initiative using annual community household surveys, routine programmatic data, and the health facility assessments. The community-level primary outcome by district uses under-five mortality rates as well as the secondary outcomes of neonatal, under-one, under-two, and maternal mortality rates. We will additionally evaluate quality of care parameters focusing on timeliness of child healthcare through promptitude of treatment reception following illness diagnosis and health facility readiness scores. Lastly, we will assess equity through secondary analyses of under-five mortality by household wealth quintiles, maternal education levels, and distance from the nearest health facility. Primary outcome: under-five mortality rates. We will calculate under-five mortality rates from all births reported by respondents using a standard birth/death history table. We will calculate under-five mortality rates and compare the risk of death before age five across surveys with the Cox proportional hazards regression using intervention exposure as the explanatory variable. Children still alive and under age five at the time of survey will be right censored. Secondary outcomes: neonatal, under-one, and under-two mortality rates. We will calculate the neonatal, under-one, and under-two mortality rates from all births reported by respondents using the same methods described above adjusted for 28 days, 1 year, and 2 years. Secondary outcome: maternal mortality. We will calculate an exploratory maternal mortality analysis based on sisterhood reports [57, 58]. We define reach metrics as the proportion of target population that gained access to the ICBHSS initiative services using the annual community household surveys. We will assess the implementation strategy through individual-level ICBHSS participation with community-level (vis-à-vis CHWs) and facility-level health service utilization using health service coverage estimates and early service access for child health. We will use a mixed-effects generalized linear model to compare pre-intervention to post-intervention proportions for each reach metric while adjusting for clustering at the facility and district level and time and allowing for district-level estimates to be random effects. Our primary analysis will not include adjustment for individual-level characteristics, as each district will serve as its own control. We define adoption metrics for this study as the proportion of the community and providers changing health-seeking or providing behavior. Our evaluation of adoption will be completed using routine programmatic data and annual community household surveys. Through this domain, we will assess implementation strategy measures using community-level engagement and behavior change by the individual-level uptake of the ICBHSS intervention. We will test whether these proportions increased, applying the same approach described in the reach evaluation measures through mixed-effects generalized linear models. Metrics of implementation are expressed as fidelity and feasibility as well as documenting contextual factors [59]. Evaluation of implementation will be completed using key informant qualitative assessments (in-depth interviews with implementing partners and health facility staff) that will be conducted 1-year post-intervention at the cluster (district) level. It will complement quantitative data collected to evaluate implementation strategy and will assess emerging themes. Our study defines maintenance metrics as costs required to deliver and sustain the ICBHSS model. Evaluation of maintenance will be completed using the costing and return-on-investment analysis, which will be conducted each year of implementation at the cluster (district) level. This analysis will assess program implementation costs based on strategy design retrospectively using the Community Health Planning and Costing Tool [48] and the Lives Saved Tool [49]. These results will be used to inform planning and policy decisions and processes. Lastly, to further triangulate the validity of our findings, we will compare our baseline and 36-month estimates for the domains of reach, effectiveness, and adoption to the most recent Togo DHS or Multiple Indicator Cluster Survey (MICS) data. We will furthermore compare the domains of reach and adoption to ICBHSS through programmatic data collected at the community and facility level. We will routinely disseminate study data with key stakeholders in Togo at the national, subnational, and community level, as well as the global community of public health practitioners, researchers, and policymakers. IH staff will conduct biannual forums with local leaders, public sector health facility staff, and community members to discuss ICBHSS implementation and share findings. Based on these forums, IH and MoH partners will collaboratively decide to adapt the implementation strategy and, if deemed necessary, the intervention. IH staff will additionally participate in MoH monthly district-level meetings for district health personnel to stay current in MoH plans and to share findings from ICBHSS initiatives. Results will be published in conference abstracts and peer-reviewed journals with preference for publicly available publications in collaboration with partners at the Togolese MoH.

• lauria, molly e.
• fiori, kevin p.
• jones, heidi e.
• gbeleou, sesso
• kenkou, komlan
• agoro, sibabe
• agbèrè, a. d.
• lue, kelly d.
• hirschhorn, lisa r.