Cost of malaria treatment and health seeking behaviour of children under-five years in the Upper West Region of Ghana

Background There is limited knowledge on cost of treating malaria in children under-five years in northern Ghana which poses a challenge in determining whether interventions such as the National Health Insurance Scheme (NHIS) and Community-based Health Planning and Services (CHPS) have reduced the economic burden of malaria to households or not. This study examined the malaria care seeking and cost of treatment in children under-five years in the Upper West Region of Ghana. Methods The study used a cross-sectional, quantitative design and data were collected between July and August 2016 in three districts in the Upper West Region of Ghana. A total of 574 women who had under-five children were interviewed. Socio-demographic characteristics of respondents, malaria seeking patterns for under-five children with malaria as well as direct medical and non-medical costs associated with treating under-five children with malaria were collected from the patient perspective. Analysis was performed using STATA 12. Results Out of 574 women visited, about 63% (360) had children who had malaria and sought treatment. Most treatment was done at formal health facilities such as the health centres (37%) and the CHPS (35%) while 3% had self-treatment at home. The main reason for choice of place of treatment outside home was nearness to home (53%). The average direct medical and non-medical costs associated with treating an under-five child with malaria were US $4.13 and US$3.04 respectively. The average cost on transportation alone was US$2.64. Overall, the average direct medical and non-medical cost associated with treating an underfive child with malaria was US$4.91(range: minimum = US$0.13 ±maximum = US$46.75). Children who were enrolled into the NHIS paid an average amount of US$4.76 compared with US$5.88 for those not enrolled, though the difference was not statistically significant (pvalue = 0.15). Conclusions The average cost to households in treating an under-five child with malaria was US$4.91. This amount is considerably high given the poverty level in the area. Children not insured paid a little over one US dollar for malaria treatment compared to those insured. Efforts to improve enrolment into the NHIS may be needed to reduce the cost of malaria treatment to households. Construction of more health facilities near to community members and at hard to reach areas will improve access to health care and reduce direct non-medical cost such as transportation costs.

The study was conducted in the Upper West Region (UWR). Upper West Region is one of the 10 Regions in Ghana, located in the northern part of Ghana. Its population is 680,000 (2010), and it is one of the regions with a low population density in Ghana. The Region is bordered by Burkina Faso to its North and West. The major ethnic groups in the region are the Dagaati, Sissala, and Wala [18]. Subsistence agriculture is the mainstay of the people. The region is one of the poorest regions in Ghana with the highest malaria burden in the country [19]. The prevalence of fever among under-five children in the study area is 25% [19]. There is a total of 242 health facilities providing various types of services in the Upper West Region. There are three (3) district government hospitals, one (1) Regional hospital, two (2) Christian Health Association of Ghana (CHAG) hospitals and three (3) private hospitals. The rest are five (5) Polyclinics, sixty six (66) health centres, ten (10) clinics, one hundred and forty seven (147) CHPS Compounds and four (4) maternity homes [20]. The study design was cross sectional and data were collected from July to August 2016. Quantitative methods were used in the data collection. We interviewed women of reproductive age who had under-five children. This study was part of a bigger study that aimed at determining access to maternal and child health services, cost of providing primary health care services [17], as well as costs in seeking malaria care for under-five children in the Upper West Region of Ghana. Only households with under-five children who had malaria in the last one month prior to the survey and sought treatment were included in this study. In cost of illness studies, data collection and analysis are carried out either from the societal perspective (including the patient and the health system/provider costs), health provider perspective (including costs incurred by the health provider) or patient perspective (the patient only or patient and care giver costs) [21,22]. This study adopted a patient perspective as used in similar studies [10,22–24] and estimated costs incurred by the household on under five child who had malaria as well as cost incurred by the care giver in seeking the care for the child such as transportation costs. A structured questionnaire was used to collect data and respondents were asked questions on socio-demographic characteristics, household assets, health seeking and cost of treating the most recent under-five child who had malaria within the past 1 month and sought care/treatment. A multi stage method was used in selecting respondents for the study. At the first stage, we grouped all the 11 districts in the Upper West region by the three main ethnic groups. The Dagaati districts comprised of DBI, Jirapa, Lambussie, Lawra, Nadowli, Nandom; the Wala districts comprised of Wa East, Wa Municipality, Wa West; and the Sissala districts comprised of Sissala East and Sissala West. In the second stage, we randomly selected one district from each of the ethnic groups. This method ensured a geographical representation of respondents across the region. Nadowli district was randomly selected to represent the Dagaati ethnic group, Wa West district was selected to represent the Wala ethnic group and Sissala East to represent Sissala ethnic group. In the third stage, we randomly selected a sub-district within each district and then communities in the sub-district. Interviewers were then allocated to communities to identify respondents by moving from one household to another until they meet their sample target. A team of 5 interviewers were sent to each sampled community in a day. A target of 5–8 interviews was assigned to each interviewer depending on the size of the sampled community. The interviewers were assigned to different sections of the community. In each section of the community, households were then randomly selected by the interviewer going to the centre of the section and spinning a pen. The direction in which the pen pointed was followed and from which the first house was selected. Only one household with a child below five years was interviewed in a house. In case a house had more than one households with under-five children, one household was randomly selected. If there were more than two women with under-five children in a sampled household, only one of them was randomly selected for interview. The interviewer then moves to the nearest house/household. This was continued till the maximum number of respondents assigned to the interviewer was obtained for that community. The sample size for this study was calculated by assuming that 50% of the women in the study districts had at least one child aged less than five years and with 95% confidence level [10,25], and design effect of 1.3 [26] to account for clustering effect in the districts would require a sample size of 500. Adjusting for a refusal rate of 15% yielded a sample size of 575. Graduate level data collectors were recruited for the data collection. The selection criteria included ability to understand and speak the local language, familiarity with the environment and previous experience on data collection. Data collectors were trained on the study protocol, instruments and guidelines in conducting interviews. Data collectors interviewed women of reproductive age who had under-five children. The questions included socio-demographic characteristics of mothers/caregivers, household assets as well as health seeking and cost of treating the most recent under-five child who had malaria within the past 1 month and sought care/treatment (S1 Questionnaire). The episode of malaria was based on the report given by caregivers. Treatment could either be at the health facility or at home by buying drugs from the pharmacy or drug shops. The data were double entered, cleaned and verified using EPI Data 6.1. Inconsistencies in entries between two data entry clerks who entered a questionnaire were checked from the source document and corrections made. Data were then transferred to STATA version 12 (STATA Corporation, College Station, Texas) for analysis. Additional data cleaning by way of identifying outliers, missing values and checking for consistency among variables were carried out by running frequencies and cross tabulations. The source documents were checked and in some cases phone calls were made to respondents to correct some inconsistencies. Basic frequencies, proportions and averages/means were estimated and presented in tables. The data was analyzed as survey data, and we therefore used ‘svyset’ in STATA to identify the survey design characteristics with community as the primary sampling unit (clusters). We then prefixed the estimation commands with ‘svy:’ in STATA. For the purpose of this study, a total of 15 communities were randomly sampled from the three districts. Differences in mean cost of treating malaria were tested using Student’s t-test. Direct and non-direct medical costs were calculated. The direct medical costs covered all out-of-pocket payments (OOP) for registration card, consultation, diagnosis, medicines and medical supplies on the patient; and direct non-medical costs included all out-of-pocket payments for transportation to and from health facilities (patient and caregiver) and other OOP or informal payments. Informal payments (“under the table payments”) are OOP payments made by patients/caregivers to the health staff for services and medical products that are officially free of charge at the health facility. The total direct and non-direct medical costs were summed and divided by the number of households that made expenditure to arrive at the average costs per treatment. Catastrophic payments for malaria treatment were also calculated. Catastrophic payments occur when total OOP payments for health care exceeds a certain threshold of a household’s resources (income or expenditure) [4,27,28]. Thresholds for calculating catastrophic payments vary, usually ranging from 2.5% to 40% [4,27,28]. This study used a threshold of 5% as applied in other studies [4,29] and all households that spent more than 5% of their annual income to pay for the treatment of malaria were deemed to have made catastrophic payments. The annual income of households was calculated using estimated cost of yields or proceeds from agricultural products such as crops, poultry as well as from salaries or wages from business, income from investments or gifts (for the dataset used for the analysis, See S1 Data). All costs were collected in Ghana Cedis (GH¢) and results presented in US$. The US$ conversion was based on the average exchange rate between January 2016- June 2016 (1US$ = 4GH¢). Ethical approval was obtained from the Navrongo Health Research Centre Institutional Review Board (Approval ID: NHRCIRB232) and the National Centre for Global Health and Medicine (NCGM), Japan (Approval ID: NCGM-G-0020510-00). Permission was also sought from the regional health directorate of UWR, district directorate of Wa West, Nadowli and Sissala East as well as chiefs of selected communities for the interviews. All the respondents were briefed on the study procedure and written informed consent was obtained. For the illiterate respondents, consenting was done in their preferred local language and those who agreed to participate thumb printed on the consent form.

• dalaba, maxwell ayindenaba
• welaga, paul
• oduro, abraham
• danchaka, laata l.
• matsubara, chieko