Integrated Sustainable childhood Pneumonia and Infectious disease Reduction in Nigeria (INSPIRING) through whole system strengthening in Jigawa, Nigeria: study protocol for a cluster randomised controlled trial

BACKGROUND: Child mortality remains unacceptably high, with Northern Nigeria reporting some of the highest rates globally (e.g. 192/1000 live births in Jigawa State). Coverage of key protect and prevent interventions, such as vaccination and clean cooking fuel use, is low. Additionally, knowledge, care-seeking and health system factors are poor. Therefore, a whole systems approach is needed for sustainable reductions in child mortality. METHODS: This is a cluster randomised controlled trial, with integrated process and economic evaluations, conducted from January 2021 to September 2022. The trial will be conducted in Kiyawa Local Government Area, Jigawa State, Nigeria, with an estimated population of 230,000. Clusters are defined as primary government health facility catchment areas (n = 33). The 33 clusters will be randomly allocated (1:1) in a public ceremony, and 32 clusters included in the impact evaluation. The trial will evaluate a locally adapted ‘whole systems strengthening’ package of three evidence-based methods: community men’s and women’s groups, Partnership Defined Quality Scorecard and healthcare worker training, mentorship and provision of basic essential equipment and commodities. The primary outcome is mortality of children aged 7 days to 59 months. Mortality will be recorded prospectively using a cohort design, and secondary outcomes measured through baseline and endline cross-sectional surveys. Assuming the following, we will have a minimum detectable effect size of 30%: (a) baseline mortality of 100 per 1000 livebirths, (b) 4480 compounds with 3 eligible children per compound, (c) 80% power, (d) 5% significance, (e) intra-cluster correlation of 0.007 and (f) coefficient of variance of cluster size of 0.74. Analysis will be by intention-to-treat, comparing intervention and control clusters, adjusting for compound and trial clustering. DISCUSSION: This study will provide robust evidence of the effectiveness and cost-effectiveness of community-based participatory learning and action, with integrated health system strengthening and accountability mechanisms, to reduce child mortality. The ethnographic process evaluation will allow for a rich understanding of how the intervention works in this context. However, we encountered a key challenge in calculating the sample size, given the lack of timely and reliable mortality data and the uncertain impacts of the COVID-19 pandemic. TRIAL REGISTRATION: ISRCTN 39213655 . Registered on 11 December 2019. The INSPIRING Jigawa Trial is a cluster randomised controlled trial (cRCT), evaluating a three-part whole system strengthening intervention package, in Kiyawa Local Government Area (LGA), Jigawa State, Nigeria. This will be complemented with an embedded ethnographic process evaluation and economic evaluation to determine cost-effectiveness. The trial will run from January 2021 to September 2022, following a formative phase and COVID-19 adaptation from December 2019 to December 2020 (Fig. ​(Fig.1).1). The primary research question and secondary process and economic evaluation research questions are detailed in Table ​Table11. SPIRIT trial schematic Impact, process and economic evaluation research questions The trial will be conducted in Kiyawa LGA, Jigawa State. The population is approximately 230,000 people, with 25% estimated to be children under-five years. The area is predominantly Islamic with two main ethnic groups—Hausa and Fulani, living in mostly rural communities [12]. At study inception, Kiyawa LGA had 33 government health facilities, made up of primary health centres (PHCs), basic health centres and health posts, and a further two private primary care facilities and private pharmacy shops. Children who meet WHO criteria for referral should be sent to PHCs or hospital care at Dutse General or Rasheed Shekoni Teaching hospital, both located in the state capital, Dutse, in the neighbouring LGA (Fig. ​(Fig.2).2). Maternal and child health services are provided free at selected facilities, but out-of-pocket costs still pose a financial barrier to access [4]. Kiyawa LGA was selected through discussion between Save the Children Nigeria, the State Ministry of Health (SMOH) and the Jigawa State Government, using pre-defined criteria (e.g. mortality burden, road access). Map of Kiyawa LGA with study clusters This is a pragmatic community-based cRCT, following a formative phase of intervention development, concept testing and community engagement. Impact against the primary outcome will be evaluated using an open prospective cohort, recording vital events (pregnancies, births and under-five child deaths) amongst a random sample of women on a rolling 4-monthly basis. Secondary outcomes will be recorded during cross-sectional surveys conducted at baseline (January–March 2021) and end line (July–September 2022) (Fig. ​(Fig.1).1). The same random sample of compounds, and all eligible women and children residing in these compounds, will be included in both surveys and prospective cohort. Clusters are defined as the geographical catchment area of the 33 government primary care facilities. Clusters range in population size from approximately 1680 to 24,200 (median 5840). All villages in Kiyawa LGA will be included in the study (n = 425 enumerated villages); however, the Kiyawa PHC cluster will be excluded from the impact evaluation as this facility is in a peri-urban area and provides a level of inpatient care—unlike the other primary facilities. Villages are made up of compounds, which are defined as the group of structures (including ‘households’) where extended families live and have shared resources. The target population for the impact evaluation is women of childbearing age, defined as 16–49 years inclusive, and children aged 0–59 months inclusive, who reside permanently in Kiyawa LGA. Women and children will be considered lost to follow-up if they migrate, are temporarily unavailable, or consent is withdrawn. Children of women who are lost to follow-up, or die during the study period, will still be included if they continue to reside in the same compound and an eligible caregiver can provide consent. The intervention is open to all community members, with no age or gender restrictions. The intervention is a locally adapted ‘whole systems strengthening’ package of three evidence-based activities: PLA men’s and women’s groups [8], Partnership Defined Quality Scorecard (PDQS) [13–15] and training, mentorship and provision of basic essential equipment and commodities for child health (Table ​(Table1).1). The details of the community groups and PDQS have been refined through a co-design process, using a modified version of the community conversation methodology. In its original form, the approach has been used across a range of settings in Africa to shift and influence behaviour change in relation to health, including HIV/AIDS [16] and female genital mutilation [17]. Though typically applied as an intervention itself, our reformulation was designed to deepen engagement with communities at scale as part of the concept testing process. Conversations were organised in a similar manner to the PLA groups that will form one component of the intervention, providing an additional layer for exploring the acceptability of this approach prior to implementation. Groups of stakeholders, including men, women and healthcare workers, were brought together through a series of iterative and interactive discussions to explore perceptions of key concepts underpinning the interventions (e.g. challenges related to child health), community relationships with healthcare workers and how key intervention components would work best in their communities, including location and timing for delivery of groups and incentives. Full details of this methodology and results will be published separately. Groups of approximately 20–30 men and women in each community will be convened. Group meetings will be held separately for men and women to allow gender norms to be respected, and tailor meetings to different locations and times. Groups are open to all community members, and sharing of group messages will be encouraged; men and women with young children and those currently pregnant will be particularly encouraged to join. The groups will work through a four-phase PLA cycle, following an initial introductory session. The phases include the analysis of the situation and prioritisation of child health issues in their communities (phase 1), planning actions to tackle these issues (phase 2), implementing the selected strategies (phase 3) and collective self-evaluation (phase 4) (Fig. ​(Fig.3).3). The PDQS process focussed specifically on healthcare service quality issues, with group members and healthcare workers joining together to mutually agree on quality indicators and joint actions. PDQS will be incorporated into all 4 phases of the PLA cycle, where healthcare workers will attend men’s and women’s groups. Each phase can involve multiple physical meetings, decided on by group members (see the “Ethical considerations” section for details on COVID-19 mitigations). The completion of one PLA cycle is anticipated to take 4–5 months, and groups will repeat the cycle multiple times over the 18-month intervention period. Women’s and men’s group PLA cycle We plan 340 groups to ensure the recommended coverage of one group per 500 population [6]. Groups will be led by 170 trained community health volunteers (CHVs), provided with a stipend. CHVs will facilitate an average of two groups, with 12 to 28 groups per cluster, depending on the population. CHVs will be recruited from the local communities, meeting the following criteria: previous experience in community mobilisation, basic knowledge of child health, fluent in Hausa, minimum basic primary education and resident in Kiyawa LGA. Where possible, facilitators will be recruited from those working as Volunteer Community Mobilizers (VCM), Team Supervisors or other community health workers. They will receive a 5-day training, covering the PLA approach, their roles and responsibilities, participatory skills development using observation and feedback, use of the flipbook job aid and collaboration with key community stakeholders. Following training, a questionnaire assessment will be administered, and only those showing adequate competencies in networking, negotiation and mediation, and basic child health knowledge will be retained. CHVs will be supervised by Save the Children study staff based in Dutse, in collaboration with the SMOH, Kiyawa LGA staff and health facility staff. The staff at health posts, basic health centres and PHCs within the intervention clusters will receive a package of trainings to support child health, including iCCM for community health workers, IMCI for facility staff, nutrition management and vaccination supply chain management (Table ​(Table2).2). Training will be delivered at the start of the intervention period and then followed up with mentorship and supervision. Training and supervision will be delivered by Save the Children Nigeria in collaboration with the SMOH. A post-training ‘start-up kit’ will be provided, including basic essential equipment (e.g. respiratory rate timers, thermometers, MUAC tape) and guideline booklets and charts. Description of health system strengthening components Training modules and M&E tools Job aids and chart booklets Start-up kit, including RR timer, ORS + zinc, paediatric ambu bag, MUAC tapes, thermometer Vaccine carrier bags 3 electric and 2 solar refrigerators M&E tools, charts and guidelines Training modules and M&E tools Job aids and chart booklets MUAC tapes Plumpy nut 25 staff at 2 referral hospitals 6 staff at 3 PHCs including CHEWs, CHOs, nurses and doctors Annual biomedical engineer equipment audit Bi-monthly* mentorship visits by Save the Children staff 12 Lifebox pulse oximeters with universal and paediatric clip probes 8 oxygen concentrators 40 oxygen cylinders *This schedule applies to the first 6 months post-training, after which the supervision and mentorship plan will be reviewed CHEW community health extension worker, CHO community health officer, CHIPS Community Health Influencers Promoters and Services, RR respiratory rate, ORS oral rehydration salt, MUAC mid-upper arm circumference, PHC primary health centre All clusters, including control, will have access to a child health radio programme throughout the project period (Fig. ​(Fig.1).1). The radio programme consists of weekly dramas, covering prevention, recognition and care-seeking for common childhood diseases, including pneumonia. There will be a monthly call-in session, where members of the public can ask questions and share experiences. The radio station is accessible across the whole study area. The two referral hospitals located in Dutse LGA (Fig. ​(Fig.2),2), and three PHCs in Kiyawa LGA will be supported with oxygen and pulse oximetry, to ensure minimum standard of care is available for children meeting WHO IMCI referral criteria in intervention and control clusters. The primary outcome is under-five mortality. This is defined as the number of children aged 7 days to 59 months reported to have died over a 12-month period who resided in sampled compounds (numerator), divided by the total number of children aged 7 days to 59 months who reside in sampled compounds (denominator). Deaths in the first week of life are excluded due to the challenges of accurate reporting of stillbirths and early neonatal deaths [20], and we do not expect the intervention to impact perinatal causes. Secondary outcomes are detailed in Table ​Table33 and are based on the INSPIRING Programme Theory of Change. Primary and secondary outcomes *Deaths which are verified by a field supervisor during a verbal autopsy visit **Includes 1 dose BCG, 4 doses oral polio vaccine, 2 doses rotavirus vaccine, 3 doses DPT-Penta, 3 doses PCV-10 and 2 doses measles The 33 clusters will be randomised using a 1:1 ratio between intervention and control. Randomisation will be done publicly, with representatives from the LGA and all the wards, following community consent for the study and the randomisation process. The names of the health facility clusters will be written on pieces of paper, folded and placed in a container by study staff. A community representative will then pull the folded pieces of paper out of the container one by one, in front of community witnesses. The first 17 will be assigned to the intervention, and the final 16 will be assigned to control. Due to the nature of the intervention, communities will not be blinded to their status; however, we will not inform data collectors of the intervention status of villages or that they work across intervention and control clusters, and investigators will be blinded during primary analysis. During the formative phase, we conducted a community mapping exercise to create a sampling frame, enumerating all villages and the total number of compounds. We will sample proportionately to cluster size, with a minimum of 50 compounds per cluster. Simple random sampling will be used to generate a list of compound numbers within each cluster. During the baseline survey, data collectors will use an Expanded Programme of Immunization approach [25] to number all the compounds in a village, following these steps: (1) locating the centre of the village with support from a community gate keeper, (2) spinning a pen and walking in the direction it points, (3) numbering the first compound encountered as 1 and (4) walking in a clockwise direction until all compounds are numbered. They will then recruit the compounds according to the generated random sample. The same compounds will be included in all subsequent surveys, and GPS coordinates will be taken to facilitate follow-ups. All eligible women, and children under-five years of age who are under their direct care, residing within sampled compounds will be recruited. The under-five mortality rate in Jigawa was reported as 192/1000 livebirths in 2017 [26], and deaths in the first week of life may reasonably account for 15% of the total under-five deaths. More recently, a participatory radio programme on child health and COVID-19 was delivered across the LGA (June–December 2020) and vaccine coverage has increased (unpublished data), which we assume to have reduced mortality. We therefore used a conservative baseline mortality of 100 deaths amongst children aged 7 days to 59 months per 1000 livebirths. With the following parameters, we will have a minimum detectable effect size of 30% [6]: (a) 380 children per cluster, (b) 80% power, (c) 5% significance, (d) intra-cluster correlation of 0.007 [27] and (e) coefficient of variance of cluster size of 0.74. We assume on average there are three children under-five per compound (i.e. 127 compounds per cluster), and therefore, we will sample 4480 compounds overall to allow for ~ 10% loss to follow-up. The sample size was calculated using Stata’s -clustersampsi- command [28]. We will review these assumptions following the baseline survey. Data collection will be done by 20 female clinical data collectors recruited from Jigawa, with two field supervisors. They will work in teams of 3–4, accompanied by a lay data collector to support community mapping. A sub-set of data collectors will be retained between surveys to conduct prospective vital event reporting. Data will be collected using a custom-built CommCare application on Android tablets. The forms will have in-built skip patterns and cleaning rules to minimise data entry errors. Data will be uploaded daily to a central secure server, and participant ID and data cleaning checks will be conducted continually during the data collection period. Data collectors will be trained using a hybrid approach, with a mix of practical sessions, at-home assignments and online webinars over a 2-week period. This will be followed by a field pilot in a neighbouring LGA and mock interviews with a group of women in the community who took part in questionnaire co-design discussions. The baseline and endline surveys are planned to take 3 months, with approximately 70 compounds completed each day. Three respondent types will be interviewed in each compound: (1) head of compound, (2) head of households and (3) eligible women. The head, or most senior member of the compound present at the time of the survey, interview will include compound membership, structure and construction; shared compound and individual household assets and income; and community cohesion. Household heads will be asked about household membership and asset ownership. Women will be asked about themselves, and then about each of the children under-five who are in their direct care. Questions include recent birth history, demographics and social status, knowledge of pneumonia, recent illness episodes and care-seeking, vaccination status, feeding practices, smoke exposure, community cohesion and wellbeing (using the Short Warwick-Edinburgh Mental Wellbeing Scale [23]). Questions on pneumonia knowledge and recent illness will use multi-media (i.e. videos and sounds) to provide examples of key signs and symptoms. In the endline survey, additional questions will be asked about intervention exposure and engagement. A clinical screening for malnutrition and pneumonia will be conducted on all children aged 0–59 months who are present in the compound at the time of the survey. The assessment will follow the WHO’s 2014 IMCI guidelines [19], enhanced with pulse oximetry using the Lifebox pulse oximeter (Acare Technology, New Taipei City, Taiwan). If a child is found with pneumonia, the data collector will inform the caregiver and advise them to seek care and facilitate referral in cases of severe illness. Following recruitment, the same compounds will be visited on a four-monthly rolling basis. All previously registered women will be asked about their current pregnancy status, the outcome of previously reported and completed pregnancies and the vital status of all registered children. Any new children under-five will be registered and included in the cohort. Verbal autopsies (VAs) will be conducted for all deaths in children under-five, including those reported in the prior 12 months from birth histories. This time period was chosen to minimise recall bias, but ensure a culturally appropriate period of time has passed since the death. We will use the COVID-19 adapted WHO 2016 VA tool and include locally adapted social autopsy questions which explore care-seeking pathways [29, 30]. The primary analysis will be intention-to-treat, comparing under-five mortality between intervention and control clusters over a 12-month period (July 2021–June 2022), using mixed-effects logistic regression, with fixed effects for the trial arm (intervention or control), and random effects to adjust for compound-level clustering and trial cluster. Data from the first 6 months of intervention implementation (January–June 2021) will be considered as a baseline period, before any intervention effect would be expected. Any imbalances in baseline characteristics between arms will be presented to the Trial Steering Committee (TSC), who will advise on whether the primary analysis should be adjusted for these. We will conduct a complete case analysis, where only children with confirmed survival outcomes will be included; a sensitivity analysis using multiple imputation will be done, if outcome data is found to be missing at random. Secondary analyses, using the same mixed-effects model approach, will explore differences in pneumonia prevalence, coverage of prevent and protect indicators, knowledge and care-seeking between intervention and control clusters (Table ​(Table3).3). Changes in pneumonia prevalence, knowledge and care-seeking between baseline and endline surveys will be described and compared using chi-squared tests.