Integrated and simplified approaches to community management of acute malnutrition in rural Kenya: a cluster randomized trial protocol

The primary research question that the study seeks to answer is: Does integrating management of acute malnutrition (including MAM and SAM) into iCCM improve coverage, quality of care, and treatment outcomes of children 6–59 months with acute malnutrition?

The study will also explore a number of secondary research questions focusing on the process variables. The secondary research questions include:

Turkana County, situated in North Western Kenya, is the largest of the 47 counties in Kenya with a total area of 77,000 Km² administratively divided into seven sub counties namely; Turkana Central, Loima, Turkana South, Turkana East, Turkana North, Kibish and Turkana West. According to the latest, 2009, Kenya Population and Housing Census (KPHS) [15] and statistical estimates, the county has a population of 855,399, and expected to grow to 1,427,797 by the year 2017. The population of children under 1 year and those below 5 years is projected to reach 25,922 and 186,243 respectively by 2017.

Turkana is drought-prone and experiences frequent, successive and prolonged droughts virtually every year. Turkana’s harsh environment and historic underdevelopment has led to the highest poverty rate in Kenya at 94%, with significantly lower indicators for human development and service coverage than the rest of the country. According to the 2009 KPHS, less than one in five (18%) of the population can read and write, only 9% roads were tarmac and less than one in five households had access to improved sanitation [15]. The average distance to a hospital, health centre or dispensary is 50 km. Serious gaps in human resources further complicate service availability. Doctor to population ratio stands at 1:70,000 while nurse to population ratio is 1:5200 [16].

According to the SMART survey conducted in June 2015 covering all the livelihood zones in Turkana county (pastoral, agro-pastoral and formal employment/business/petty trade), about one in four (25%) children in the county are acutely malnourished. The rates of acute malnutrition in Turkana Central/Loima, North and South/East indicated a very critical nutrition situation, while the nutrition situation in Turkana West was classified as critical. The weighted Global Acute Malnutrition (GAM) for Turkana County was 21.2% and SAM rate was 5%.

Isiolo County is located in the North Eastern part of Kenya and borders Marsabit County to the North, Samburu and Laikipia Counties to the West, Garissa County to the South East, Wajir County to the North East, Tana River and Kitui Counties to the South, Meru and Tharaka Nithi Counties to the South West. The county covers an area of approximately 25,700 km² and has an estimated population of 143,294 of which children aged 0–59 months are estimated to be 31,163. It consists of three sub-counties namely Isiolo, Garbatulla and Merti.

To be included in the study, the following criteria highlighted in Table 2 should be met.

The overarching our outcomes of integration are seamless service provision, maximized synergy and improved efficiency and effectiveness. The following list shows key primary and secondary outcome indicators.

At admission, the CHVs will only screen under five children (aged 6–59 months) and carryout preliminary examination for any danger signs. This will be followed by further examination for complaints including cough/cold, diarrhea, dehydration and fever, which will be followed by anthropometric measurements to classify the child in relation to nutritional status either MAM, SAM or normal. A special five color-band MUAC tape will be used to allow identification.

Children with MUAC < 9.5 cm and/or less than 4 kg on the dosage scale, with/without bilateral pitting oedema, those who fails the appetite test or show presence of iCCM danger signs or other illnesses other than iCCM diseases (diarrhoea, malaria, pneumonia), sick neonates (less than 1 month) and malnourished children less than 6 months, will be subject for referral to the health facility. However, those sick children below 2 months and caregivers who fail to consent will be excluded at the recruitment and only referred to the health facility.

If MUAC is < 11.5 cm, the child will be classified as SAM; before admitting the child directly for treatment, do the following: children presenting a MUAC of < 11.5 cm will be defined as with SAM while MAM will be defined as MUAC of between 11.5 cm and 12.5 cm. The CHVs will then conduct physical examination of MAM and SAM cases. At this stage, examination for any oedema not observed during assessment of danger signs will be detected. Appetite test will then be conducted setting the stage for either referral or exclusion. Wait for up to 30 min before deciding the child has poor appetite. If the child takes significant quantities as indicated in the Table 3, interpret as the child has good appetite. If no complications and good appetite, the child will be admitted for community-based management of malnutrition.

For SAM children, the CHVs will give first dose amoxicillin dispersible tablets (DT) on the spot and twice for 7 days before the next visit. If rapid diagnosis test (RDT) is positive for malaria, CHVs will give one dose on the spot and twice daily for 3 days (the prescriptions will clearly be explained to the mother). Then RUTF will be given to the mother/caregiver and explain the daily intakes throughout until the next visit as explained. Table 4 provides details on RUTF that should be given to children based on their weight.

With regards to MAM cases, albendazole and folic acid (for weight) will be given on the spot and further dosage explained to the mother/caregiver. Then CHVs will give RUSF for the children for 7 days until the next visit. Mother/caregivers will be advised to respect follow-up, breastfeeding and diversification of feeding in addition to RUSF. The CHVs will use the following dosing chart to administer.

At follow-up, CHVs will ask and seek for general danger signs and examine any complaints on the child. In addition, the systematic evaluation will be conducted to examine any cough, fever or diarrhea following the standard treatment and follow-up protocol. Anthropometric measurement weight and MUAC will be taken to determine any changes (gains) in the child. The CHVs will further conduct systematic evaluation to look for edema not detected at the general danger signs and appetite test. Children with either danger signs, failed appetite test, weight loss, diarrhea will be referred to the health facility, otherwise the CHVs will conduct treatment for MAMA or SAM as appropriate.

For SAM cases, during follow-up the RUTF will be provided for the next 7 days and if RDT is positive, ACT will be provided on the spot and twice daily for 3 days. Adelbanzole will be given if during the second visit the child is aged more than 12 months. In the case of MAM, the CHVs will give folic acid and continue RUSF for the next 7 days. The diversification and breastfeeding should be emphasized again at this point.

Furthermore, during follow-up all SAM children will be subject to weekly visit to CHVs for MUAC measurement and weight using the dosage scale, and collection of RUTF as per dosage scale will continue for a maximum of eight weekly visits if the child is progressing well, otherwise referral will be recommended in case of non-response to treatment. On the other hand, MAM children will be subject to bi-weekly (every 2 weeks) visits for MUAC measurement and collection of RUSF (1 sachet per day for 16 weeks for a maximum of eight visits if child is progressing well, otherwise the child needs to be referred). Table 5 outlines the follow-up procedure for both MAM and SAM.

All the activities during this intervention will be clearly documented using primary data collection tools and will be collated and for further analysis on the effectiveness of ICCM-CMAM integration.

Quantitative data will be conducted at baseline to establish coverage of SAM and MAM treatment as well as other relevant nutrition and iCCM indicators. Data on treatment outcomes will be collected during the implementation. Further, end line quantitative data collection will be done to assess the coverage of SAM and MAM treatments as well as other relevant nutrition and iCCM indicators.

These data will be collected via quantitative household surveys. Building on our extensive experience with household nutrition surveys and iCCM, We shall use questionnaires to collect detailed information on socio-demographic and economic characteristics of the household, child characteristics including health, nutrition status and standard anthropometrics in every time period. Anthropometric data, will include child’s weight, presence of oedema, height and mid-upper arm circumference will also be collected, using standard protocol [19]. Additionally, CHVs will also be interviewed to obtain data that shall inform of their understanding and practice of the laid out protocol. The same information will be compared with data collected from mothers or caregivers of children in the clusters to gauge the treatment regime received compared to the laid out practice. Altogether an understanding of the effectiveness of this service will be obtained. These questionnaires to be used will be pre-tested before fielding the final survey. Furthermore, in the Kenya context, at the level of health facility, there are registers which are used to follow-up children with MAM/SAM cases. Those registers provide valuable information on the outcomes (recovery, defaulter and death). But, for the community-based management of malnutrition, we will use similar registers to record all the management of MAM/SAM cases which will also provide information on our outcomes. We will setup a tracking system which help us to randomly select our sample size of MAM/SAM cases and follow-up them till 4 months period at each time point.

Collection on costing data for integrating nutrition into iCCM will be carried out from the health system’s perspective and focus on the costs incurred by the system (e.g. government, donors) for the delivery of this health intervention. Furthermore, we will explore the societal perspective by including costs for the caregiver e.g. the caregivers’ travel costs and lost labor time, societal/opportunity costs. Other research costs such as salaries of personnel doing research (field monitors/supervisors) will not be taken into account.

In this health intervention program, the costs incurred could be summarized into three components.

The program manager or institutional perspective is the three components listed above. The societal perspective is more global since it includes the three components of cost listed above and other household costs incurred for participating in community-based management of malnutrition (time taken to feed the child suffering from MAM/SAM, additional food purchased to feed the child and other related cost incurred).

In order to account for the inflation, all the costs will be converted in constant prices using the GDP deflator for year 2019 for Kenya. Our base year of costing exercise will be 2019. Furthermore, we will explore whether integrating management of acute malnutrition into iCCM is cost effective using the decision rule based on WHO-CHOICE (Choosing Interventions that are Cost-Effective).

We will conduct qualitative data collection in three stages: pre-intervention, midline and post intervention qualitative data collection. For each of the stages, we will conduct focus group discussions (FGDs), in-depth interviews (IDIs) and key informant interviews (KIIs) as well as observation of care provision to assure quality of services. The number of interviews per stage and sub-group to be interviewed as well as the sample size will mainly depend on the level needed to achieve saturation of concepts. However, we anticipate the following number of interviews per stage: 8 IDIs, 8 FGDs, and 14 KIIs per county. We expect that each FGDs will have 8 people. Table 6 gives an overview of the sample size for qualitative data collection.

For the qualitative data collection, purposive sampling will be used to sample women of reproductive age with children suffering from acute malnutrition, fathers of children being treated for acute malnutrition, other community members including grandmothers, health care providers, community leaders and policy/decision makers in the Ministry of Health, County health management teams, county executive officer, county director of health, implementing partners and other key organizations in the study areas.

During the implementation phase, qualitative data will be collected at mid-term whereby we will conduct in-depth interviews, key informant interviews and focus group discussions with women/ caregivers in the program, on their experiences with the integrated package and their satisfaction as well as suggested potential areas of improvement. We will also conduct focus group discussions with CHWs and key informant interviews with other stakeholders (implementing partners and the Ministry of Health Officials including the CHMT and Health Facility Workers) on their views on the progress of the integrated program including facilitating and limiting factors, and the cost and time implications of the intervention. The information gathered will be useful in engaging the stakeholders including the implementing partners, the Ministry of Health, UNICEF and WFP during the remaining section of the implementation phase in order to minimize potential risks and counter challenges experienced and to maximize on the potential benefits.

We will conduct qualitative interviews among women whose children were treated for acute malnutrition during the implementation period, and men (fathers of the children) as well as other community members to assess their experience and satisfaction with the integrated program within the period of the implementation, and potential areas of improvement, observed advantages and challenges and perceived effects of the integrated program including adherence to the treatment given, recovery rate and their relationship with the CHVs during the implementation phase. We will also conduct interviews with the CHVs to determine their experiences in implementing the integrated package including challenges and facilitators; their views on its effect on their workload; effect on the quality of care given; their views on their capability to administer treatment; and their suggestions on the possible areas of improvement.

In a non-inferiority in CRCT, we will first analyze the findings using non-parametric namely the adjusted Chi-squared test for clustered binary such as recovery rate, case fatality and defaulter rate. For continuous outcomes, we will use t-test for clustered data namely for weight gain. Furthermore, for our primary outcome we will use an econometric model namely a Probit model where the dependent variable is the recovery rate and covariates are the treatment variable (one for the intervention and zero for the control), age of the children, gender of the children, vaccination status (unvaccinated, partially vaccinated, fully vaccinated), overall length of stay, breastfeeding, co-morbidities at admission (malaria, anemia, pneumonia, measles and giardiasis), admission status (new admission, readmission, admission after default), education of the mothers, income of the mothers etc.

All digital project data will be stored in encrypted computers or laptops that are password protected, with regular back-up procedures to the server put in place to ensure that the data is not lost. All survey data will be collected via password protected tablets and uploaded to a secure cloud-based data depository after completion of the survey. Data will then be deleted from the tablets. Non-anonymized data will be stored on a secure, password projected server made accessible only to the core research team. All data will be anonymized before being made available to other project staff, ensuring the confidentiality of data following collection.

Quantitative data will be analyzed using STATA, SPSS and/or MLWin files. Prior to data sharing, it will be anonymized. The primary aim of the analyses conducted will be to answer the research questions outlined in the protocol. Quantitative data analysis will mainly be descriptive. Qualitative interviews will be audio-recorded on encrypted recorders, transcribed verbatim (and translated/transcribed as needed), anonymized and stored in digital format (MS Word compatible).

Tape-recorded qualitative data will be transcribed verbatim. Rapid analysis of the transcribed qualitative data collected during the preceding phases for the purpose of informing the succeeding phases will be done through systematic reading through the transcripts.

For further understanding and use of the data including dissemination and scientific writing, data will be coded in NVIVO (QSR International Pty Ltd), to identify primary and meta-codes and major themes. Themes will be identified, with attention to contradiction and diversity of experiences, perception and attitudes across the different interviewees. Analysis across all transcripts will be done thematically [20].

A qualitative software package (NVIVO) will be used to code the qualitative data and support analyses. Thematic analysis will be used to identify specific themes from the individual interviews using NVivo. Coding and interpretation will be undertaken by at least two members of the research team to ensure objectivity and consistency during coding. All transcripts will be anonymized to conceal the identity of the research participants. The anonymized transcripts will be stored in password protected computers, accessible only by the research team.