Socioeconomic Inequalities in Care-Seeking for Children Under Five Before and After the Free Healthcare Initiative in Sierra Leone: An Analysis of Population-Based Survey Data


 Background: There are real socioeconomic inequalities between and within countries, leading to disparities in the use of health services. In children under five years of age, this could be responsible for child mortality by depriving children of healthcare. By removing healthcare fees, the Free Healthcare Initiative (FHCI) adopted in Sierra Leone can contribute to reducing these inequities in healthcare-seeking for children. This study aimed to assess the socioeconomic inequalities in healthcare-seeking for children under five years of age before and after the implementation of the FHCI.Methods: We included 1207, 2815, and 1633 children under five years of age with fever from the 2008, 2013, and 2016 nationwide surveys database, respectively. Concentration curves were drawn for the period before (2008) and after (2013 and 2016) the implementation of the FHCI to assess socioeconomic inequalities in healthcare-seeking. Finally, concentration indices were calculated to understand the magnitude of these socioeconomic inequalities.Results: We found that before the implementation of the FHCI, there were inequalities in care-seeking for children under five (concentration index (CIn) = 0.168, standard error (SE) = 0.049) in favor of the wealthy households. These inequalities decreased after the implementation of the FHCI but remained in favor of the rich households (CIn = 0.039, SE = 0.040). There were more disparities in the healthcare-seeking between districts before the implementation of the FHCI. After the initiation of the FHCI, these disparities were reduced, and 12 of the 14 districts had a CIn around the value of equality. Conclusion: Our study observed pro-rich inequalities in care-seeking for children under five years of age before the implementation of the FHCI and a decrease in these inequalities after its implementation, even if it remained pro-rich. To continue to reduce wealth-related inequalities, policy actions can focus on the increase of availability of health services in the districts where the care-seeking was pro-rich.

This study used data from the 2008 and 2013 Sierra Leone Demographic and Health Surveys (DHS) and the 2016 Malaria Indicator Survey (MIS). These are national representative household surveys in which women aged 15-49 years were interviewed. Details of the complete description of the interview method are available elsewhere [29,30]. Our study focused on febrile children under ve years of age whose caregivers sought healthcare in the two weeks preceding each of the surveys.
Settings: Sierra Leone is a 71,740 square kilometers West African country located on the southwest coast with a population estimated at 7,396,000 in 2016 [31]. The population is young in general, with 63% aged less than 25 years and children under ve representing approximately 17% of the total population [32].
The climate is tropical with two seasons and vegetation ranging from the savannah in the North to the rainforest in the South. For these surveys, the country was divided into four administrative regions: The northern, eastern, southern, and western regions. Each region was subdivided into districts with a total of 14 districts in the country. The health system in Sierra Leone is organized into three levels. The primary level is made up of peripheral health units (PHUs), of which 229 are community health centers (CHCs), 386 community health posts (CHPs), and 559 maternal and child health posts (MCHPs). Meanwhile, the secondary level is made up of 21 district hospitals [33], and the tertiary level includes regional and specialized hospitals. There are six teaching hospitals and several private clinics and hospitals spread across the 14 districts of the country. The distribution of the wealth index is not equal between the rural and urban areas. Approximately 61% of the population in the urban area are richest and in the rural area, 28% are poorest [30], while the inequality-adjusted human development index was low at 0.266 in 2017 [34].

Variables
The outcome variable was the proportion of children under ve years of age whose caregivers sought care in the two weeks immediately preceding the surveys. We de ned the place in which care was rst sought for fever (public and private) to construct the care-seeking variable.

Sampling method
A two-stage cluster sampling method was used in the three Sierra Leone population-based surveys. The whole country was subdivided into enumeration areas (EAs). Each EA included several households. In the rst stage, EAs were selected with strati ed probability proportional to sample size. The place of residence (urban or rural) was used to stratify the EAs. In the second stage, from the EAs, households were selected using systematic random sampling. The sampling frames were developed based on the 2004 census for the 2008 and 2013 surveys and on the 2015 census for the 2016 survey [29,30,32].

Statistical methods
Statistical analyses were performed using Stata version 15.0. The characteristics of the children's caregivers and those of the febrile children during the two weeks preceding the survey were described rst. The differences between participants' characteristics and between the surveys were assessed using chisquare tests to determine the trend. Analyses were carried out separately for each of the three surveys to evaluate the proportion of caregivers seeking care for their febrile children. The descriptive analyses were weighted for probability sampling and adjusted for strati cation and clustering.
For the analysis of wealth related-inequalities, participants were grouped-according to their socioeconomic status-into wealth quintiles, as follows: Poorest (1st quintile), poorer (2nd quintile), middle (3rd quintile), richer (4th quintile), and richest (5th quintile). Concentration curves and the concentration index were used to assess inequalities in the use of healthcare. The concentration curves were used to examine the trend in the pattern of the socioeconomic inequalities in healthcare-seeking, while the concentration index was used to assess the magnitude of the inequalities in healthcare-seeking.
Concentration curves: The concentration curves were built using two keys variables: The independent wealth index and the healthcare-seeking for febrile children under ve. The concentration curves represent a plot of the cumulative percentage of caregivers seeking care (y-axis) against the cumulative percentage of the households, ranked by the wealth index, beginning with the poorest, and ending with the richest (x-axis) [36].
We compared the concentration curves to the line of equality which is the 45-degree line running from the bottom left-hand corner to the upper right-hand corner, indicating the absence of inequalities in healthcare-seeking between the households ranked in the wealth quintiles. When the percentage of caregivers seeking care takes higher values among poorer households, the concentration curves lie above the line of equality. On the contrary, when it takes lower values among poorer households, the concentration curves lie below the line of equality. The farther the curves are above the line of equality, the more concentrated the health variable is among the poor [36].

Concentration index
The concentration index estimates the magnitude of wealth-related inequalities in healthcare-seeking. The formula of the concentration index is as follows where represents healthcare-seeking, μ represents its mean, r is the fractional rank of an individual in the wealth index distribution, and cov is the covariance between healthcare-seeking and the fractional rank of the wealth index [36]. The concentration index is calculated as twice the area between the curve and the line of equality. The concentration index is bounded between -1 and 1. In absence of wealth-related inequalities, the concentration index is zero. The concentration index takes a negative value when the curve lies above the line of equality, indicating a disproportionate concentration of healthcare-seeking among the poor. It takes a positive value when it lies below the line of equality, indicating a concentration of healthcare-seeking among the richer [36]. QGIS 3.12 software was used to map the concentration indices for healthcare-seeking for children under ve at the health district level [37]. The characteristics of the study participants were described using sociodemographic and healthcare variables, which are summarized in Table 1. In addition, more than 45% of the mothers in the three surveys were aged between 25 and 34 years and were living in the northern region (over 37%). In terms of educational level, the proportion of women without education decreased from 72.5% in 2008 to 68% and 60% in 2013 and 2016, respectively. According to the surveys, most of the women were from rural areas (over 64%) and were Muslims. More than 45% of the children were aged between 12 and 35 months and belonged to families with 6-10 members (over 51%).  In 2008, the Bonthe, Kailahun, and Tonkolili districts were the places where care-seeking for febrile children favored the poor households. In the Kono, Kenema, Kambia, and Bombali districts, care-seeking was concentrated in the wealthy households (Fig. 3).

Results
In 2013, care-seeking for febrile children was concentrated in the Bo and Western Area Rural districts among the poor households. In the other districts it was in favor of the rich households (Fig. 4). Figure 5 depicts the wealth-related inequalities in the healthcare-seeking for children under ve, strati ed by districts in Sierra Leone after the implementation of the FHCI (2016). These results suggest that the inequalities in care-seeking were more concentrated in the Port Loko, Western Area Rural, and Koinadugu districts in favor of the rich households and, on the contrary, were in favor of the poor households in the Western Area Urban district. In the Kono, Bombali, and Kambia districts, equality in healthcare-seeking between the rich and poor households was nearly achieved.

Concentration index of healthcare-seeking for children under ve with fever
To better appreciate the magnitude of the inequalities, we computed the concentration index as reported in Table 2.
The positive concentration index suggests that healthcare was more accessible to the rich households. A decrease in the magnitude of the inequalities was noted after the implementation of the FHCI (2013-2016), as indicated by a 2016 concentration index, close to zero (0.039); however, it remained pro-rich, though not statistically signi cantly.