Participants and data collection
Data for this analysis were collected as a part of the baseline situational analysis for the “Diagonal Interventions to Fast Forward Enhanced Reproductive Health” (DIFFER) project. The DIFFER interventions aim to improve access to SRH and HIV services among women in general and Female Sex Workers (FSWs), and are being tested in four countries; Mombasa (Kenya), Durban (South Africa), Tete (Mozambique), and Mysore (India). The situational analysis aimed to describe baseline service provision and access, including the demographic and socio-economic characteristics of SRH service users at sentinel facilities. Information on service provision was collected through a detailed service audit not reported in this paper. Service access and the characteristics of service users were measured using a quantitative patient exit interview. There are a total of 13 sentinel facilities for which data were collected; three in India, four in Kenya, five in Mozambique and one in South Africa. These facilities offer a range of SRH services including care for sexually transmitted infections (STIs), family planning, HIV testing and counseling, HIV care and ART, cervical cancer screening, services for victims of gender-based violence, and termination of pregnancy and/or post abortion care.
Respondents eligible to participate in the patient exit interview for the situation analysis were women older than 18 years, who had completed a visit at one of the sentinel facilities, all urban centers, and were willing to sign an informed consent form. Respondents were recruited and interviewed immediately following their consultations for SRH services. The sample size of female SRH users was powered to enable the study to detect an increase in the proportion of satisfied health service users from 60 to 80 %, at the 95 % confidence level. The sampsi command in Stata/IC 11.0 for a two-sample comparison of proportions was used. The estimation was based on the method of Fleiss, Levin and Paik [22] to estimate the sample size to achieve a given power of a two-sided test for the difference in two proportions. Across the four study sites, a total of 614 female SRH users were interviewed (India: 150, Kenya: 100, Mozambique: 99, South Africa: 265). However, data on household income and coping mechanisms were not collected within the patient exit interviews in South Africa and Mozambique. As such, only the survey data collected in India and Kenya (n = 250 SRH users) are included in this study and described further below.
In India, data were collected from women attending three facilities in Mysore city. These facilities included two government hospitals (Cheluvumba and SMT Hospitals) and one private facility (Asha Kirana Hospital). 50 participants were interviewed at each facility for a total of 150 participants. Study participants were recruited using a consecutive sampling technique. In other words, all individuals who meet the selection criteria were approached to participate in the study. While the government facilities offered a range of different SRH services, Asha Kirana’s SRH services were limited to HIV care and treatment services. Accordingly, participants from the government facilities were chosen from the full range of different SRH services provided within the government facilities, but at Asha Kirana participants were limited to HIV positive women seeking HIV care and treatment.
In Kenya, a total of 100 participants were recruited from the family planning and STI clinical services at four representative urban public health facilities in city of Mombasa; two health centers (Kisauni and Chaani) and two district hospitals (Tudor and Likoni). With the help of health care providers, and with access to the health facility registers, the study team estimated the weekly number of clients using each SRH service (estimated at between 17 and 20 clients per service). Based on this estimate, 3 to 5 clients were interviewed per available service and up to 25 clients per facility in order to gather data on the full range of SRH services. Using an interval sampling method (or nth person selection technique) every 4th client was recruited as they exited each of the SRH services of interest.
The surveys collected data on the socio-economic characteristics of respondents including age, education level, occupation, relationship/s status, place of residence, average monthly household income, number of children, religion, the services requested and those provided to them, total expenditure, including both direct medical payments (i.e., payments for any services received at the facility, medicine, tests, consultations etc.) and direct non-medical payments (including the transportation costs), travel time, waiting time at the facility, sources of financing these expenditures, degree of satisfaction with services received, perceived unmet needs, and measures of empowerment and agency.
The questionnaire for the patient exit interview was designed to collect data through face-to-face interviews. The questionnaire was translated into local languages and then back translated into English to ensure accuracy. Respondents could choose to be interviewed in English or the local language. Interviews were conducted in December 2012 in India and between November 2012 and March 2013 in Kenya. Interviews were done by trained staff, with experience in research of this nature and who are used to asking sensitive questions. The study was reviewed and approved by the local Research Ethics Committee in each site as well as the ethics committee at Ghent University as the coordinating partner. Informed consent was given by all participants before they were enrolled in the study.
Data management and analysis
Data collected from the exit-interviews were entered into a Microsoft Access database, cleaned and extracted to Stata, Version 12, for analysis. Information on costs of seeking care, sources of financing (coping mechanisms), average monthly household income and demographic variables were extracted from the database for the analysis.
Average monthly household income was used as the measure of socioeconomic status. This included income generated by all members of household from different sources (including government grants, pension etc). Household was defined in this study as “all the people who live under one roof or who eat from a common pot”. Quintile ratios and a Kakwani index were used to measure the progressivity of the costs of seeking SRH services. To calculate quintile ratios, the proportion of health spending in the lowest and the highest income quintiles were compared and tested for significant differences in means using a t-test, using a 95 % confidence interval [12]. If individuals in the lowest income quintile bore more costs than those in the highest when seeking care, spending was defined as strongly regressive. If there was no significant difference in expenditure - if the lowest quintile did not spend significantly more or less - spending was defined as weakly regressive [12].
Spending was further analysed using the Kakwani index [23, 24]. This index is a commonly used measure of equity in health care financing/payments [6, 25–27], which compares the distribution of health care spending, plotted on the concentration curve, with the distribution of income or consumption expenditure (plotted on the Lorenz curve). The Kakwani index is defined as twice the area between the concentration curve and the Lorenz curve and is calculated as;
$$ \uppi \mathrm{k} = \mathrm{C}\ \hbox{-}\ \mathrm{G} $$
Where:
C = the health spending’ concentration index
G = the Gini coefficient of the household income.
The value of the Kakwani index (πk) ranges from -2 to 1. A negative index indicates regressive spending as the concentration curve lies inside the Lorenz curve, while a positive index indicates progressive spending as the concentration curve lies outside the Lorenz curve. When the index is zero, i.e. the concentration lies on the top of the Lorenz curve, the spending is proportional [24].
Catastrophic spending was calculated as the percentage of household income spent on SRH care seeking. There is no single accepted threshold for catastrophic health care payments. The rationale behind catastrophic spending is that households must reduce spending on their basic needs, and they might go into debt or sell productive assets, risking household livelihoods, over a period of time to cope with this degree of health care spending. While the choice of the threshold is effectively arbitrary, two commonly used thresholds are 10 % of total income [1, 28, 29] or 40 % of income after spending on food (referred to as capacity to pay) [2]. The former threshold (10 % of income) is employed for this analysis and we calculated the % of household income spent on the last reported SRH care event.
Further, the strategies that were adopted by the respondents in each site to cope with the costs of seeking care were analysed considering socio-economic status of the individuals.