The study was conducted in Malindi and Lamu districts in the Coast Province of Kenya. The region is characterized by a low level of access to health facilities for much of the population due to difficult geographical barriers, semi-arid climate and prevailing high level of poverty. Many villages are located far from main access roads with terrain that is impassable by motor vehicles, including some areas located across rivers or swamps. Lamu district includes several islands. During the rainy season, widespread flooding and wild animals from neighboring game parks further aggravate inaccessibility. At the time of the project design, Coast province was classified as a high malaria transmission zone; however, more recent Kenya Malaria Indicator surveys
[1, 12] suggest that concerted malaria prevention and control programmes are bearing fruits and transmission is declining over time.
The pilot CCM project was implemented by Kenya Red Cross Society (KRCS) that works with an extensive network of community-based volunteers in these two districts. The project was funded by the Canadian International Development Agency (CIDA) and technical support was provided by WHO and the Division of Malaria Control (DOMC, Ministry of Public Health and Sanitation - MoPHS). The pilot phase had an estimated target population of 9,600 children 3–59 months of age living in 113 communities (total population 71,850) that were generally considered “hard to reach”.
The objective of the pilot project was to increase access to artemisinin-based combination therapy (ACT) as malaria treatment for children under five in hard to reach areas. The project’s management structure was set up in such a way as to strengthen the capacity of the health system in these districts in accordance with the national Community Health Strategy, being fully integrated within the roles and responsibilities of existing health service providers. For instance, the district malaria focal person worked in close collaboration with the KRCS Project Officer to manage the project activities and Community Health Extension Workers (CHEWs), who are full-time employees of the MoPHS based in health facilities in the targeted areas, functioned as coaches to selected groups of CHWs. ACT supplies (artemether-lumefantrine (AL) packs) were supplied through normal MoPHS channels where Kenya Medical Supplies Agency provided facilities with medication based on needs (i.e. the District Pharmacy) with additional logistic support from KRCS.
In January 2009, 113 CHWs (one per village) were selected by Village Committees, based on the government’s Community Health Strategy 2006 criteria, which required them to be resident in the community, respected members of the community and literate. Preference was given to pre-existing KRCS volunteers in the village while the project recruited new CHWs and registered them as new volunteers in areas without KRCS volunteers. Project induction training was conducted by DOMC. Further refresher training for CHWs was done after six months. These trainings were conducted using approved training curriculum whose content incorporated Integrated Management of Childhood Illness (IMCI) materials that covered important and relevant components like signs and symptoms of uncomplicated malaria, danger signs for referral, correct dispensing of AL (dose regimen, importance of compliance, potential adverse reactions), the referral process, record keeping and community health promotion topics, e.g. malaria prevention through use of insecticide-treated nets (ITN) and intermittent preventive treatment in pregnancy (IPTp).
At the time of the project design, the WHO and DoMC guidelines for treatment of fever in children under five did not require diagnosis prior to treatment. Rapid diagnostic tests for malaria were not regularly available at a significant proportion of health facilities due to cost constraints. Therefore, the use of RDTs was not included in the project. In order to prescribe the AL at the community level by CHWs, the project applied for and received a formal waiver for prescription of AL across the counter from the Pharmacy and Poisons Board together with DOMC. Supplies of age-appropriate AL packs were distributed to CHWs by their respective coaches on a monthly basis, based on supplies used and monitoring reports of child fever cases treated. Compliance was measured from the volunteer recording forms and discussion with care givers. Referrals to a health facility were done when there were danger signs (e.g. complicated malaria case), adverse drug reactions or treatment failure.
Close supervision and monitoring of CHW activities was done throughout the project implementation period. CHEWs met every two weeks with CHWs at a designated health facility to verify the CHW monitoring reports and address any concerns. The health facility nurse often participated in these meetings and this contributed to enhanced coordination, particularly for referrals. The project was also discussed at monthly district level meetings and independently monitored by provincial health management team members.
Project evaluation methods
An evaluation of changes occurring during the project period was conducted by carrying out a household survey in the targeted communities before and after the intervention. Baseline and endline surveys were conducted in December 2008 and December 2009, respectively. Two-stage cluster sampling was used. In the first stage, 30 clusters (defined as villages) were randomly selected using probability proportional to size. In the second stage, 30 households were randomly selected from a list of all households in the cluster with at least one child 3–59 months of age.
Two semi-structured survey questionnaires were administered, both based in large part on standardized Malaria Indicator Survey (MIS) modules
. Data was collected on the household member list, household characteristics, mosquito net inventory, reproductive history of the woman caregiver, fever and health seeking behaviours in the youngest and next-to-youngest children, and knowledge of AL and malaria. A series of questions was added for the endline survey to assess the frequency and nature of the household’s interaction with CHWs. Two interviewers were assigned to collect data from each sampled household; one interviewed the head of the household (Form A) and the other the selected woman caregiver over 15 years of age (Form B).
Training of interviewers for field data collection procedures was conducted using KMIS 2007 materials and manuals. During data collection, field supervisors undertook daily spot checks including checking 10% of the completed questionnaires to confirm that the interviews took place appropriately. During the spot checks, at least three households were re-interviewed by the field supervisor to verify the information collected and any discrepancies were corrected before data entry in Microsoft Access.
Statistical data analysis
The main outcome variables of this study were malaria care-seeking and treatment practices. Data on child fever treatment practices was taken from the module administered to women caregivers of children 3–59 months. In households with data for two children, one child was randomly selected. From this sample of one child per household, children with fever in the last two weeks were selected for inclusion in the sub-sample to assess fever treatment practices.
Variables related to treatment seeking practices, in particular the source of advice/treatment sought and associated actions, were analysed by household characteristics including caregiver age and education level, household wealth rank (see below) and size, attendance to antenatal clinic (ANC) as a proxy for access to routine health services, malaria-related knowledge and practices (IPTp coverage, knowledge of AL as the recommended anti-malarial drug, knowledge of sleeping under bed net to prevent malaria) and village size. Source of treatment was categorized as CHW and Others, which included public and private health sector sources. Treatment was considered prompt when it was given within 24 hours of fever onset. Effective treatment was defined as receiving AL, the recommended first-line anti-malarial.
Household-level variables on asset ownership, water source, toilet facility and house flooring material were used to create a wealth index using Principal Component Analysis, a methodology that is commonly applied to DHS datasets
. Each household was given a factor score, which was used to assign a rank. In order to heighten the sensitivity of the wealth index to the lowest and highest groups, a three-level ranking was created, with the lowest rank including the poorest 20% of households, the middle rank the central 60% and the highest rank the least poor 20% of households. Internal consistency of the wealth index was assessed. In both surveys, the poorest 20% of households all shared common characteristics – all relied on surface water sources, none of them reported using toilet facilities (all used field/bush) and none of them reported ownership of any assets (with the exception at baseline of some households owning a bicycle). The least poor 20% of households were most likely to use improved sources of water, use improved toilet facilities and own the assets under question. These results suggest that the wealth rank variable generally performed as desired and may assist in understanding the differences between households associated with relative wealth, even though it is clear that, in general, all households in the project area were disadvantaged socioeconomically.
One CHW was assigned to each village regardless of the size of that village. The size of villages targeted by the project ranged from a population of 107 to 3,173 (mean 636, median 500), approximately 23 to 690 households. Village size was recoded into a four-level categorical variable: <60, 61 to 100, 101 to 200, and >200 households.
Pearson’s chi-square test for categorical variables was used to test for differences across groups; tests associated with a p-value less than 0.05 were considered as evidence for a significant difference. Data were analysed using Stata version 11.2.