The study is a pragmatic, single-centre, three-arm, parallel, open-label, stratified cluster-randomized controlled trial with 1:1:1 allocation. A cluster is defined as a geographical community or village(s)/settlement(s) serving as the proximate catchment area for at least one public primary health facility and one patent medicine vendor (PMV) within the study sites/strata in Ebonyi state.
A three-arm parallel design, with equal number of clusters in each arm and equal sample size in each cluster, will be used to assess the effect of two interventions compared to control:
- Control Arm: No intervention. This arm involves the usual practice of provision of MRDT services by individual health care providers (in public health facilities and PMVs) with basic training in MRDT. Patent medicine vendors that have offered the MRDT service previously but are not currently doing so will be re-supplied with MRDT kits to resume provision of MRDT services for the study period.
- Social Group Arm: Social group intervention. This arm involves control treatment plus the sensitization and education of social groups about MRDT.
- Social Group/Provider Arm: Social group/provider intervention. Control treatment, the sensitization and education of social groups about MRDT plus the training of health care providers in health communication with patients/caregivers (clients) about MRDT.
The trial protocol development was guided by the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) 2013 checklist (see Additional file 1).
Study area
The study is being conducted in three sites/strata in Ebonyi state, south-eastern Nigeria. Ebonyi state is located in the south-east geopolitical zone of Nigeria between latitude 50 40’ and 60 54’N and longitude 70 30’ and 80 30’E with land area of 5,953 sq. km.35 It shares borders with Benue state to the north, Enugu state to the west, Cross River state to the east, Imo and Abia state to the south.35,36 The State lies in the Cross River plains with rainy/wet season from April to October and dry/harmattan season from October to March.35 Floods often occur during the rainy season due to poor drainage systems, stagnant rivers and ponds which expose the State to mosquito infestations and high malaria burden.35 The vegetation of the state is mostly savannah in the drier northern part and forests in the wetter southern part.35
The state’s projected population for 2016 was 2,897,401 (based on the 2006 national census figure and a growth rate of 2.8 percent) with male making up 48.62 percent, female is 51.38 percent and under-5 years age group is 20 percent.35 The people of Ebonyi are primarily of the Igbo language and ethnic extraction with ten dialects/minor languages spoken across the state.36 English, especially its local variant, the pidgin, is a widely spoken language in the state. People of other languages and ethnic groups in Nigeria also live in the state especially in the capital and urban areas. Most inhabitants practice Christianity.36 The State is divided into 3 senatorial zones (Ebonyi north, Ebonyi central and Ebonyi south), with 13 Local Government Areas (LGSs) (with Abakaliki LGA as the administrative and political capital), 64 development centres, 138 autonomous communities and 215 political wards.35 Each autonomous community has a traditional ruler and consists of autonomous villages each having a village head called chairman. Each autonomous village is made up of smaller villages or settlements each having a village/settlement head.
The public sector is the main driver of the State economy and agriculture is the major occupation.36 The state has several solid mineral resources (including lead), crude oil and natural gas36 but there are few large-scale commercial mines and industries.35,36 The state is called "the salt of the nation" because of huge salt deposit at the Okposi and Uburu Salt Lakes.36 Traditional industries and works of art include blacksmithing, pottery works and wood works (carved doors, stools, walking sticks, traditional flutes, wooden mortars and pestles).36
Health care in Ebonyi state (like other states in Nigeria) is provided by the public and private sectors under the overall guidance of the federal government through the federal ministry of health (and its agencies) and the national council on health. The federal government (through the federal ministry of health – FMoH) provides health services in the state through tertiary health centres. The state government (through state ministry of health – SMoH) provides health care through secondary health centres (general hospitals). The state government also supports the local governments in providing primary health care (PHC) through PHC centres. The Ebonyi state Malaria Elimination Program (SMEP), within the department of public health in the state ministry of health, coordinates the efforts to combat malaria in the state.
Other health care service providers recognized by the National Health Act 2014 include the private healthcare providers, traditional healthcare providers and alternative health care providers.37 These can be grouped as the private health sector and subdivided as: formal private sector (private hospitals and clinics) and informal private sector (pharmacies, patent medicine vendors (PMVs), traditional healers and alternative health care providers).4 Majority of febrile patients, especially from the lower socio-economic groups seek care primarily in the informal private sector.14,16,18
The three study sites/strata consist of the three senatorial zones: Ebonyi north, Ebonyi central and Ebonyi south. Ebonyi north consists of four local government areas (LGAs) with a 2016 population of 998,473.38 It includes Ebonyi urban area/metropolis made up of the state capital city, Abakaliki and the adjacent city, Ebonyi. However, it is a predominantly rural area. The vegetation is mostly savannah and it is drier than the other sites. Ebonyi central consists of four local government areas (LGAs) with a 2016 population of 861,55438 and is a predominantly rural area. The vegetation changes from savannah towards forests and the weather become wetter as one move through this central zone southward towards the southern zone. Ebonyi south consists of five local government areas (LGAs) with a 2016 population of 1,037,37638 and is also predominantly rural. The vegetation consists of forests and wetter weather compared to the other sites/zones. The three sites are all endemic for malaria with year-round transmission and a high malaria burden.
Participants and participant timeline
Participants
Clusters
Clusters participating in the study include villages/settlements (a community) that have at least one eligible public primary health care facility and one eligible patent medicine vendor (PMV) (see below) and are easily accessible (close to a motorable road) even in the rainy season. Clusters either participating in or that participated in any similar interventions within the preceding one year will be excluded from the study. Clusters that are too close (less than 15kms apart), and not separated by a buffer area or natural barrier, and urban clusters (in cities/towns) will be excluded from the study to minimize contamination between clusters within strata.39,40 Non-consenting clusters will also be excluded from the study.
Social groups
Social groups are any set of persons within society with particular demographic, economic or social characteristics.41 Examples of social groups include women associations/meetings, village meetings, men associations, youth associations, market/trade unions, elders’ fora, parent-teacher association etc. Social groups that are recognized by cluster heads/authorities will participate in the study and non-consenting social groups will be excluded.
Primary health facilities and individual health care providers
Eligible public primary health facilities are those providing MRDT services, maternal and child health care services including immunisation; attending to at least an average of four fever cases (or suspected cases of malaria) per day and having at least two staff that are at least junior community health extension workers (JCHEWs). Individual public health care providers involved in the diagnosis (and treatment) of malaria in these facilities are eligible. Eligible patent medicine vendors (PMVs) are those with basic training in MRDT services and either currently offering or previously offered MRDT services. Non-consenting health facilities and individual providers will be excluded from the study.
Households and community members
Households will be eligible to participate if there is a report of any case of fever or suspected malaria among under-5 children, 5 years and above children, and adults (excluding pregnant women) in the household in the preceding two weeks to a population-based household survey. For each household, consent will be obtained from the mother of the house or the female primary care giver who will be the respondent to the survey. Community members that have resided in the community for at least two years will be eligible to participate in a focus group discussion (FGD). Non-consenting households and community members will be excluded.
Participant timeline
Eligible participants will be enrolled and randomly allocated into the three study arms.
The main study outcomes will then be measures through a baseline household survey and an end-line household survey three months after the end of intervention. The participant timeline is depicted below in accordance with the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) guideline (see figure 1).
Interventions
The interventions in this study include: (1) the sensitization and education of social groups about MRDT (social group intervention) and (2) the sensitization and education of social groups plus the training of health care providers in health communication with client’s about MRDT (social group/provider intervention). Participants in the control arm will not receive any of these interventions.
The sensitization and education of social groups
The social group intervention will involve the sensitization and education of social groups within each cluster in the social group arm. There will be three episodes of group discussion/interaction, one per month, with each social group during the intervention period of three months. Health education messages designed to change wrong beliefs and perceptions, promote right beliefs and perceptions, increase knowledge and promote key actions will be communicated to social group members in an interactive session/group discussion. The health education message is adopted from a national framework42 and modified for the study.
Each discussion will be a one-hour event. One investigator will moderate each discussion with a group discussion guide. In order to make the intervention more pragmatic and to optimize adherence to trial intervention, the discussions will take place at the usual meeting time of each social group, or at an agreed time, and at their usual meeting point/venue or at a central location as is most convenient for the participants. The intervention will be discontinued for a social group or cluster whenever such request is made by the respective social group head/leader or cluster head/authority. The first episode will be divided into seven sub-headings viz: (1) introduction (2) individual beliefs and opinions (3) key facts about malaria (4) key facts about MRDT (5) demonstration of testing with MRDT kit (6) key actions to practice and promote (7) closing remark. The moderator will explain the research problem and objectives to the participants and the broad role the participants (social group members) are expected to play in addressing the problem. Each participant will then share his/her beliefs and opinion about malaria with respect to cause, symptoms, complications/fatalities, diagnosis, treatment, where to seek care and why, and prevention. The discussion will explore the reasons for and against these views and there will be questions and answers from participants. The moderator will then highlight the key facts about malaria. Key facts about MRDTs will also be highlighted and discussed. Demonstration of MRDT will be done on volunteer participants and the procedure will be explained and the result interpreted to them.
The moderator will then highlight and discuss the key actions that the participants should exercise and promote as individuals, social group members, opinion leaders, parents, heads of households, family members, friends and relatives, neighbours etc. These actions include recognizing the symptoms of malaria especially in under-5 children; promptly seeking care with providers that carry out MRDT (preferably in health centres) and requesting for the “simple test” for malaria (MRDT), receiving the test, asking for the result, and receiving treatment for malaria (with ACTs) only when the test is positive; making sure under-5 age children receives treatment within 24 hours of on-set of fever; acquiring and giving the right ACTs, in the right dose, for the right number of days. The moderator will then give a closing remark and thank the participants for their time.
The second and third episodes will take similar format as the first but will focus more on re-emphasizing the key facts and actions to practice and promote so as to reinforce participants’ knowledge and attitude. The participants will be encouraged to share their experiences in the preceding four weeks for discussion and clarifications. Visits to the social groups will end with the third episode of group discussion. However, the cell phones of the investigators will be open to group members who might still sought clarifications about malaria and MRDT during the subsequent months.
Other supportive interventions will include twice weekly reminder text messages to social group members that have cell phones; regular phone calls to social group heads (to urge them to remind and encourage the group members during meeting sessions) and regular visits by a provider (PMV or public) to the scheduled meetings of each social group for provision of MRDT services and ACTs for positive results.
The training of health care providers
The clusters in the social group/provider arm will also receive the social group treatment as described above and in addition, will receive provider training in health communication with clients. Health communication involves the use of communication strategies to inform and influence individual and community decisions that enhance health and it is a vital aspect of provider-patient relations.43 The provider training will involve a one-day sensitisation and training workshop for the participating health care providers at the Abakaliki (the capital of Ebonyi state). The research team will administer the workshop with the aid of a power point projector, a provider training guide and a health communication guide. There will be a pre-post-test to assess the knowledge of participants with respect to malaria, MRDT and communication. The training will be divided into two parts with four sessions in each. The first part will focus on sensitising the participants. The four sessions include (1) Pre-test (2) Background information: on the research problem and aim (objectives) and malaria (3) Demonstration of testing with MRDT kit (4) Key actions to practice and promote. The facilitators of the training will use a power point presentation to highlight the objectives of this part of the workshop and then to present an explanation of the problem he research wants to address and how, the objectives it hopes to achieve and the broad role the participants/providers are expected to play in this respect. Volunteer participants will then perform and receive MRDT, explain the procedure and interpret the results. This will be followed by a question and answer session on the sensitivity and specificity of MRDT, adherence to test results and what to do if result is negative, the benefits of universal testing and complete adherence to test results. The facilitators will then highlight the key actions that need to be practiced and promoted by providers. Questions and comments will be respectively answered and discussed.
The second Part of the training will focus on improving the health communication skills of participants with the aid of a health communication guide. The four sessions in this part include (1) Background information on health communication (2) Practical session (including simulation) on health communication with clients (3) Closing session and filling of assessment forms (4) Post-test. The facilitators will use power point presentation to highlight the objectives of this part of the workshop and to talk about communication and health communication with regards to definitions, communication components, process, channels/media, types, aids etc. This will be followed by a practical session and simulation of real life situation. The facilitators will then divide the participants into two groups – public provider group and private provider group. Each group will be sub-divided into clients and providers. Each client/patient sub-group will visit their respective providers with malaria-like illness. Their providers will then engage them in health communication about malaria and MRDT (with or without the use of a self-written/improvised guide). The strengths and weaknesses of each group’s communication techniques during the simulation will be noted by the other group. These will be read out by a facilitator for discussion by all parties.
The facilitators will then introduce the participants to a health communication guide on how to effectively educate their clients during the discharge of their duties. They will be educated on how to use the guide in communicating with all categories of clients including those that requested for MRDT or freely accepted providers’ suggestion of a test or initially declined from doing the test. The participants will then apply the guide in another round of simulation exercise. Participants will then fill an assessment form on which they will rate the training process and logistics, state what they believe or know previously that has now been corroborated, what they believe or know previously that has now been contradicted, what they just learnt for the first time, any additional key actions that can be taken or promoted to increase the use of MRDT and adherence to test results. The participants will be encouraged to use the health communication guide to effectively communicate with their clients (suspected malaria cases) about MRDT at the course of their regular duties. A facilitator will then give the closing remark and implore the participants to develop their communication skills by putting it to practice immediately.
The training workshop will be followed by a twice weekly reminder text messages and a monthly visit to each of the participating provider for supportive supervision (on-the-job training) and monitoring throughout the rest of the three month intervention period. Each support visit will aim at assessing providers’ performance, exploring their experiences in the previous weeks and addressing their challenges in health communication with their clients. The on-the-job training will make the intervention more pragmatic and will optimise adherence to the trial intervention. The intervention will be discontinued for a health care provider whenever such request is made by the respective provider.
The trained providers will also be subjected to mystery client monitoring. The mystery client assessment of providers' performance will guide the researchers in properly identifying those that require more attention and in what specific areas during the support visits. Mystery clients will be trained on how to interact with providers and how to fill the assessment form thereafter. A mystery client will visit a participating provider/health facility and claim he has malaria based on familiar symptoms. If proposed by provider, he will freely accept MRDT otherwise, he will request for MRDT. He will all the while note whether the provider engaged him in health communication with basic standard techniques. He will then fill the assessment form after leaving the provider. Each provider/health facility will be visited by a mystery client at least twice.
Objectives
The primary objectives include:
- To evaluate the effectiveness of the sensitization/education of social groups about MRDT (social group intervention) in increasing the demand (use and/or request) for MRDT compared to usual practice (control)
- To evaluate the effectiveness of the sensitization/education of social groups about MRDT and training of providers in health communication with patients/caregivers about MRDT (social group/provider intervention) in increasing the demand for MRDT compared to usual practice (control)
- To evaluate the effectiveness of the social group/provider intervention in increasing the demand for MRDT compared to the social group intervention
The secondary objectives include:
- To evaluate the effect of the interventions on the care seeking practices and anti-malarial drug use pattern among community members
- To assess the effect of the interventions on the knowledge and opinion of respondent female heads of households (female primary care givers) about malaria and malaria diagnosis
- To evaluate the effect of the interventions on the number of suspected malaria cases visiting the public primary health centres
- To assess the effect of the interventions on the knowledge and opinion of health care providers and their practice of health communication about malaria and malaria diagnosis
- To evaluate the cost-effectiveness of the interventions
- To assess the level of demand (use and/or request) for MRDT, care seeking practices and anti-malarial drug use pattern among community members with fever or malaria-like illness
- To assess the knowledge and opinion level of female head of households (female primary care givers) and of health care providers about malaria and malaria diagnosis
- To ascertain the factors that influence the demand for MRDT among community members
Hypotheses
- The social group intervention is more effective (and more cost-effective) in increasing the demand (use and/or request) for MRDT compared to control.
- The social group/provider intervention is more effective (and more cost-effective) in increasing the demand for MRDT compared to control and compared to social group intervention alone.
The social group intervention is expected to enhance the knowledge and opinion of community members about malaria and malaria diagnosis as well as their preference for MRDT. This will lead to a moderate increase in their demand for MRDT compared to community members in the control arm. The social groups/provider intervention is expected to (to a greater extent) enhance the knowledge and opinion of the community members and their preference for MRDT. This intervention will also enhance the knowledge and opinion of health care providers about malaria and malaria diagnosis as well as their preference for MRDT. These will lead to a large increase in the demand for MRDT among community members compared to community members in the control arm. The summary of the study’s logical framework is shown in Figure 2.
Outcome measures
Primary outcome
The primary outcome has two components:
- The first primary outcome is the proportion of under-5 children with fever or malaria-like illness in the preceding two weeks to a population-based household survey that received MRDT. A co-outcome is the proportion of these children for whom their caregivers requested for MRDT
- The second primary outcome is the proportion of 5 years and above children and adults (excluding pregnant women) with fever or malaria-like illness in the preceding two weeks to a population-based household survey that received MRDT. A co-outcome is the proportion of these subjects that requested for or whose caregivers requested for MRDT
Secondary outcomes:
The main secondary outcomes at the community level include:
- The proportion of under-5 children with fever or malaria-like illness in the preceding two weeks to a population-based household survey whose caregivers sought care with a provider, those that sought care the same or next day and among those for whom care was sought, type of provider with whom care was sought.
- The proportion of these under-5 children that took any anti-malarial drug and those that took ACTs.
- The proportion of 5 years and above children and adults (excluding pregnant women) with fever or malaria-like illness in the preceding two weeks to a population-based household survey that sought care with a provider, those that sought care the same or next day and among those that sought care, the type of provider with whom care was sough.
- The proportion of these children and adults that took any anti-malarial drug and those that took ACTs.
- The proportion of respondent female heads of households (female primary care givers) that have good knowledge and opinion about malaria and malaria diagnosis
The secondary outcomes at the provider level include:
- The number of suspected malaria cases visiting the public primary health centres (monthly) measured using patients’ register
- The proportion of providers that have good knowledge and opinion and practice of health communication about malaria and malaria diagnosis
Cost outcomes include:
Total cost of the social group and social group/provider interventions, average cost per provider and average cost per social group member
Measurement of study outcomes
Population-based household survey
Survey questionnaire adopted from the 2015 Nigeria Malaria Indicator Survey (NMIS)15 woman’s questionnaire, and modified to collect data on both the primary and community level secondary outcomes will be pre-tested in non-participating clusters before the survey. Interviewers will be recruited and trained over a one-week period to administer the questionnaire. The training of the interviewers will include a detailed review and explanation of the questionnaire items, interview techniques, how to provide information to household respondents about the survey, how to obtain consent, the translation of key words in the questionnaire to local language and how to administer the questionnaire. Baseline survey will be carried out before intervention (see figure 1) to collect data for the assessment of the level of demand (use and/or request) for MRDT among community members with fever or malaria-like illness in Ebonyi state. A section of the questionnaire will collect data on the determinants of the demand for MRDT. End-line survey will be carried out three months after the end of intervention (see figure 1) to collect data for the evaluation of the effects of the interventions on the demand for MRDT among community members.
The interviewers will be accompanied by research supervisors at the start of the baseline survey for monitoring and supportive supervision. This will be followed by at least once weekly visit for monitoring and supportive supervision. The household survey questionnaire is designed to collect data about the following items from the respondent female head of household (the mother of the house or female primary care giver):
- Basic socio-demographic characteristics (of the respondent and of eligible household members reported to have had fever/malaria in preceding two weeks)
- Fever/malaria-like illness management including care seeking and demand for MRDT
- Respondent’s knowledge and opinion about malaria and malaria diagnosis
- The determinants of the demand for MRDT
Focus group discussion (FGD)
Focus group discussions (FGDs) will be conducted at baseline to collect data on the factors that influence the demand (use and or request) for MRDT in the communities. Nine (9) focus group discussions (FGDs) will be carried out across the three study sites/strata. Participants will be purposively selected from among providers, male and female community members (who have resided in the community for at least two years). There will be three focus group discussions (FGDs) with providers, one per stratum; three FGDs with male community members, one per stratum; three FGDs with female community members, one per stratum.
Investigators will administer the focus group discussions (FGDs) using a focus group discussion question guide prepared in English and pre-tested in non-participating clusters. The focus group discussion question guide will consist of both very open-ended and more targeted questions designed to explore the providers’ and community members’ knowledge and practice concerning MRDT and their perceptions on the determinants of the demand for MRDT in their communities/villages. The more targeted focus group questions are based on the proximal determinants in the study’s logical framework. The more targeted questions are combined with the very open-ended questions to identify additional determinants. Before commencement of each focus group discussions (FGD), the investigators will collect background data of participants such as age, sex, level of education, occupation, years of experience with use of MRDT (for providers). Each FGD will consist of 8-10 participants and will last for about one hour. The FGD will be audio-recorded and later transcribed (and translated) verbatim into English before analysis.
Provider survey
Provider survey will be conducted at baseline and end-line (three months after the end of intervention) to collect data for the assessment of the effect of the interventions on the knowledge and opinion of health care providers and their practice of health communication about malaria and malaria diagnosis. The provider survey questionnaire is similar to the household questionnaire and is modified to collect data about the following items from the health care providers:
- Basic socio-demographic characteristics
- Knowledge and opinion about malaria and malaria diagnosis
- The practice of health communication about malaria and malaria diagnosis
- The determinants of the demand for MRDT
Review of patient register
Patient registers are routinely kept by public primary health facilities. The following relevant data are contained in the registers: patient’s biodata (name, age, sex, address etc), date, symptoms, type of test done, test result, type of drug given. The participating patent medicine vendors (PMVs) will be asked to keep similar records. The registers will be used by the research team for monitoring, supportive supervision and evaluation. The investigators will review the patient registers for the nine months pre-intervention and nine months post-end-of-intervention records. The document review will provide data for evaluating the effect of the interventions on the number of suspected malaria cases visiting the public primary health centres.
Documentations of the implementation/intervention process
The process of implementation of the study will be documented including challenges encountered and how these were addressed. The social group and provider intervention processes will be documented and/or recorded with respect to the name of the participants, participants’ phone number/contact, individual contributions, participants’ observations and evaluation of the intervention logistics and suggestions etc. The study personnel and the organisers of the interventions will also record their experience on challenges and possible solutions. These data will be used for monitoring and to guide the implementation of similar interventions in the future.
Costing of the intervention process
Direct and indirect costs of the intervention process will be assessed from the perspective of the implementer of intervention and that of the recipient of intervention (the societal perspective) using standard economic evaluation methods. Cost data will be estimated primarily from the documentation of the implementation/intervention process and the project financial accounts. The costing will be guided by the research budget.
Data management plan and quality assurance
During the household surveys, the research supervisors will revisit an average of thirty households in each participating cluster with a specially designed questionnaire to double check on responses and coverage. The supervisors will collect completed questionnaires from the interviewers weekly and crosscheck/review for internal consistency and completeness. Questionnaires with internal inconsistencies and or missing data will be returned to the respective field staff for correction with the respective respondent. The questionnaires will be serially (and uniquely) numbered and data will be double-entered using Microsoft Excel 2007 (Microsoft Inc., Redmond, Washington) and will be verified using Stata version 15 (Stata Corp, College Station, TX, USA). The datasets compare utility in Stata will be used to verify the datasets and any discrepancies identified will be crosschecked against the corresponding original questionnaire and corrected before analysis. The variables and data in the data set will also be examined in details and range checks will be done to ensure the data entry was done correctly and appropriately.
The audio recordings of the focus group discussions (FGDs) will be transcribed (verbatim) within 24 hours of recording and translated into English before analysis. The questionnaires, the audio recordings and the verbatim transcript of the focus group discussions (FGDs) will be stored in secure area while the electronic data file will have a back-up file. Access to study materials and data files by unauthorized persons will be prevented.
Sample size
After comparing the required sample size for each of the primary objectives, the largest sample size will be used. This is based on the primary objective to evaluate the effectiveness of the social group/provider intervention compared to the social group intervention in increasing the demand for MRDT in the communities. The sample size is estimated using the methods recommended for stratified cluster randomized trial39,40 and will be based on the first primary outcome of the proportion of under-5 children with fever or malaria-like illness in the preceding two weeks to a population-based household survey that received MRDT. A plausible estimate of the primary outcome of 11% in the control arm, based on the reported value for the South-East zone (that includes Ebonyi state) in the Nigeria Malaria Indicator Survey (NMIS) 201515, will be used.
Assuming a coefficient of variation between clusters within strata of 0.16, a sample size per cluster of 40 eligible under-5 children, 80% power at 5% probability of type I error, the trial will require 6 clusters per arm to detect a difference of 20% in the primary outcome between the control arm and the social group arm (an increase from 11% to 31%) and a difference of 20% between the social group arm and the socio group/provider arm (an increase from 31% to 51%). A sample size per cluster (cluster size) of 40 subjects with 6 clusters per arm gives a total cluster number of 18, with 240 subjects per arm and a total sample size of 720 subjects. The sample size per cluster (of 40) will be inflated by 20% to compensate for any probable invalid records and other field eventualities. This gives a final cluster size of 50 subjects with 300 subjects per arm and a total sample size of 900 subjects.
The same sample size will be used to assess the second primary outcome of the proportion of 5 years and above children and adults (excluding pregnant women) with fever or malaria-like illness in the preceding two weeks to a population-based household survey that received MRDT. Equal number of clusters (6) will be allocated to each arm and to each stratum and, despite variations in population size of clusters, equal number of fixed sample size (of 50 for each target population) per cluster will be used to optimize statistical efficiency.39,40
Sample size is estimated for the provider survey to give the expected level of precision (at 95% confidence level) for determining the proportion of providers that have good knowledge and opinion about malaria and malaria diagnosis. Based on the number of clusters (6) allocated to each arm and about 7-8 individual providers per cluster who consented to participate in the study, we expect to survey 42 providers per arm (7 per cluster). Assuming the outcome measure of 50% in each of the control and social group arms and 70% in the social group/provider arm, with an intracluster correlation coefficient (ICC) of 0.01, we can estimate the true outcome measure with ± 15.6% precision in each of the control and social group arms and ± 14.3% precision in the social group/provider arm.
Sampling technique (recruitment)
Stratified multistage (cluster) sampling will be employed. Eighteen clusters will be randomly selected from a list of eligible clusters across three strata (6 in each stratum) using the “sample” command in Stata. To enhance reproducibility, random-number seed will be set using the “set seed” command before running the “sample” command. If written consent is not provided by any of the selected cluster(s) before randomization, replacement cluster(s) will be randomly selected from the remaining list of eligible clusters using same technique. The summary of the trial profile is shown in figure 3.
Within the selected clusters, an average of four eligible social groups per cluster will be purposively selected and non-consenting social group(s) prior to randomization will be replaced. All eligible and consenting health facilities will be selected across clusters and the individual providers involved in the diagnosis and treatment of suspected malaria cases in these facilities will participate in the study. A number eligible individual providers and community members, who have resided in the community for at least two years, will be purposively selected for the focus group discussion. All the households in the selected clusters will be enumerated and screened for inclusion during the pre- and post-intervention surveys. If the required sample size is not reached after the interviewers have reached the end of the enumerated households, non-eligible households will be revisited and re-assessed to see whether they have become eligible.
Randomization
Randomization will be done using Stata version 12 (Stata Corp, College Station, TX, USA). Stratified randomisation technique will be used to assign the 18 clusters across the three strata to the three treatment arms, after written consent is obtained from all recruited participants, by a statistician that will not otherwise take part in the study. The 6 clusters within each stratum will be randomly allocated to the three study arms in the ratio of 2:2:2 using a programme written in Stata. To ensure the desired balance in cluster number in the study arms, restricted randomization technique will be employed for the within strata allocation. The summary of the trial profile is shown in figure 3.
It will not be possible to blind the interviewers that will administer questionnaires in the household survey and the respondent female heads of households because of the pragmatic nature of the trial as they could acquire treatment knowledge from informal sources or grapevine. Some of the respondents could also be social group members or providers. It will also not be possible to blind the investigators that will administer intervention or participants that will receive intervention (social groups and providers).
Data analysis
Data will be double-entered using Microsoft Excel 2007 (Microsoft Inc., Redmond, Washington) and analysed using Stata version 15 (Stata Corp, College Station, TX, USA). Since the number of subjects/household members about whom data will be collected is expected to vary across clusters (due to the method employed for household survey), the “sample” command will be used to randomly select equal number of subjects across clusters for the analysis. All analyses will be by intention-to-treat.
The effect of the interventions on the primary outcomes will be analysed using cluster-level methods for cluster randomized trials with small number of clusters per treatment arm.39,40 Point estimates of the intervention effects (risk difference) in each intervention arm compared to control and in both intervention arms compared to each other, will be computed from the unweighted mean of cluster-level summaries (proportions) of the outcome measures in each study arm. But since the cluster size (sample size across clusters) is constant, this point estimate will be identical to that obtained from the weighted average of individual values.39,40 Also, since the number of clusters across strata is constant, the estimated risk difference will be identical to the weighted average of stratum-specific risk difference.
An overall test of the null hypothesis of no difference between any of the study arms will be conducted in a two-way analysis of variance (2-way ANOVA) of the cluster-level proportion on stratum and treatment arm. This overall test is to guide the interpretation of any subsequent significant findings in pair-wise comparisons. If the distribution of the cluster summaries in each study arm is markedly skewed, logarithmic transformation may be considered before analysis. Before conducting pair-wise significance testing and computing a confidence interval, an estimate of the within-stratum between-cluster variance will be obtained as the residual mean square from a two-way analysis of variance of the cluster-level proportions on stratum and treatment arm, including interaction terms. The within-stratum between-cluster variance will then be used in a stratified t-test to tests the null hypotheses of no difference in the primary outcomes between each intervention arm and the control and between both intervention arms. It will also be used in computing the corresponding 95% confidence interval of each risk difference.
Adjusted analysis based on cluster summaries will be done in a two-staged procedure. In the first stage, covariate-adjusted residual will be obtained for each cluster using standard multiple linear regression analysis, incorporating the stratum (as a fixed effect) and all baseline cluster-level covariates of interest but excluding the intervention effect. The potential baseline covariates of interest will include cluster-level summaries of the primary outcome measures, mean knowledge and opinion score of respondents (about malaria and malaria diagnosis) and other baseline variables found to differ between the study arms and which can be determinants of a particular outcome. Only cluster-level summaries of baseline covariates will be used in the adjusted analysis because baseline and follow-up data will be on different individuals since the study employed a repeated cross-sectional design.39,40 The difference-residual for each cluster will be obtained as the difference between the observed outcome in each cluster and the predicted outcome in the absence of intervention effect. The covariate-adjusted residuals will replace the cluster-level proportions in the second stage in estimating the intervention effects which are thus adjusted for the covariates in the first stage.
The same cluster-level methods will be used to evaluate the effect of the interventions on the secondary outcomes. Comparative analysis of baseline data will be used to assess the balance between the treatment arms and potential baseline variables that will be reported include the cluster-level and individual-level summaries of the primary and secondary outcome measures, the age and sex of individual subjects, the respondent female head of household’s age, educational level, occupation, knowledge and opinion level about malaria and malaria diagnosis, and average knowledge and opinion score. Summary statistics of baseline data will be used to assess the level of demand for MRDT, care seeking practices and anti-malarial drug use pattern among community members with fever or malaria-like illness; the knowledge and opinion level of female head of households about malaria and malaria diagnosis; the knowledge and opinion level of health care providers about malaria and malaria diagnosis.
To ascertain the factors that influence the demand for MRDT in the communities, a thematic analysis of the focus group discussions (FGDs) will be done using the method recommended by Braun and Clarke.44 The audio recordings of the focus group discussions (FGDs) will be transcribed (and translated) verbatim into English and the transcript will be compared with the original recording to check for ‘accuracy’ before conducting the analysis. Exact and meaning-based translation will be used. QDA Miner Lite 2 (by Provalis Research) will be used to manage the coding and analysis process.