Menu
Menu
Menu
Menu
Objective. To explore the equity of utilization of inpatient health care at rural Tanzanian health centers through the use of a short wealth questionnaire. Methods. Patients admitted to four rural health centers in the Kigoma Region of Tanzania from May 2008 to May 2009 were surveyed about their illness, asset ownership and demographics. Principal component analysis was used to compare the wealth of the inpatients to the wealth of the region’s general population, using data from a previous population-based survey. Results: Among inpatients, 15.3% were characterized as the most poor, 19.6% were characterized as very poor, 16.5% were characterized as poor, 18.9% were characterized as less poor, and 29.7% were characterized as the least poor. The wealth distribution of all inpatients (p < 0.0001), obstetric inpatients (p < 0.0001), other inpatients (p < 0.0001), and fee-exempt inpatients (p < 0.001) were significantly different than the wealth distribution in the community population, with poorer patients underrepresented among inpatients. The wealth distribution of pediatric inpatients (p = 0.2242) did not significantly differ from the population at large. Conclusion: The findings indicated that while current Tanzanian health financing policies may have improved access to health care for children under five, additional policies are needed to further close the equity gap, especially for obstetric inpatients. © 2012 Ferry et al; licensee BioMed Central Ltd.
Data was collected at four rural health centers (Bitale, Nguruka, Kakonko, and Mabamba) in Tanzania's Kigoma Region, a western region bordering Burundi and separated from the Democratic Republic of Congo by Lake Tanganyika. Bitale and Nguruka are located in the Kigoma Rural district and Kakonko and Mabamba are located in Kibondo district. Nguruka, Kakonko, and Mabamba are all receiving new staff houses and operating theaters as part of health facility upgrades. Facility and patient level data were collected for four months, May 2008, September 2008, January 2009, and May 2009. The data collected is part of a larger study to prospectively assess the impact of quality upgrades in three health centers (plus one comparison health center without upgrades) on overall maternal health care utilization in the Kigoma Region of Tanzania. The health centers provide both primary and secondary care. The user fee to receive inpatient services was 2,000 TZS or 1.50 USD. In 2002, the gross national income per capita was $290 [18]. User fee exemptions are provided to the following: individuals under the age of five or over the age of sixty, pregnant mothers (e.g., deliveries, antenatal care, and postnatal care), and individuals with exempt medical conditions (e.g., HIV/AIDS, tuberculosis, diabetes, and cancer) [19]. The health centers also accept national health insurance, health benefits for government employees, and community fund insurance, a national prospective payment program that costs 5,000 TZS or 3.75 USD per year and covers services for an individual and their immediate family at dispensaries and health centers (catastrophic expenses are excluded) [19]. Project managers collected facility-level data at the beginning and end of each monthly data collection period. Facility-level data tracked included facility inputs (e.g., staffing levels, functionality of equipment, training courses offered, and progress on health center upgrades) and facility outputs (e.g., total admissions and length of stay). The patient survey and consent form were developed in English, translated into Swahili, and then back translated. The one-page questionnaire included demographic characteristics, admission diagnosis, self-reported health status, and asset ownership. The survey assessed household ownership of 10 assets: bike, radio, fowl, phone, electricity, mosquito nets, house material, type of toilet, number of rooms for sleeping, and meals eaten per day. These were selected from a previous population-based study of 1,205 women in the same region completed in July of 2007, the details of which are described elsewhere [20]. Two health workers from each health center were trained to administer the survey. Following the September 2008 data collection period, one trained interviewer from Bitale was transferred to another health center and the other trained interviewer left the post for personal reasons. The two replacement health workers were trained by the project manager and completed interviews in January 2009 and May 2009. All patients who were admitted to the four health centers were eligible to participate after providing written consent. The parents/guardians of inpatients under the age of 18 provided consent on their children's behalf. If patients were severely ill on admission, study health workers were instructed to interview them only after their conditions stabilized. Patient interviews lasted for approximately 5-10 minutes. Written consent was obtained from all participants. The study received ethical clearance from the Tanzania National Institute for Medical Research and the University of Michigan Institutional Review Board. We calculated univariate statistics for health center characteristics and demographic variables for all admissions, as well as three admission sub-types: pediatric admissions, obstetric admissions, and other admissions. Individuals under the age of 5 were classified as pediatric admissions. Individuals admitted for deliveries, post-delivery complications, or post-abortion complications were classified as obstetric admissions. Marital status was assessed for adult inpatients. Previous schooling was only assessed for inpatients at least 7 years old. Inpatients were categorized into wealth groups (quintiles) based on their asset index using population quintile cut-offs in the Kruk et al population-based survey [20]. Asset indices are frequently used to estimate permanent wealth in non-cash economies [21]. Household assets were assigned numeric values and an index was created using principal component analysis. The first component was used to determine asset weights, which were then used to calculate a continuous index of wealth [21-23]. Based on the value for the asset index, households were divided into five wealth quintiles (quintile 1 was designated as poorest and quintile 5 the richest). Individuals missing more than one asset response were not included in the wealth analysis. Assets were imputed for individuals with only one asset response missing, using logit imputation for the dichotomous assets and mean imputation for the number of mosquito nets and daily meals. A bivariate analysis comparing patients excluded from equity analysis to those classified by wealth quintile was completed, showing no meaningful differences between the two groups on demographic and illness factors. Concentration curves were constructed and concentration indices were calculated for all inpatients and the three admission sub-types. Concentration curves indicate the equity of distribution of a service graphically. Concentration curves have ascending wealth on the x-axis and a health variable on the y-axis, with a 45-degree line indicating equitable distribution and values below this line indicating disproportionate concentration of the variable among the rich. Concentration indices were also calculated for the following subgroups: patients with fee exempt status and patients required to pay a fee. The concentration index is a quantitative measure of the deviation of the concentration curve from the line of equality (45 degrees) and has been widely used in international research to quantify the degree of income inequality [24-27]. A concentration index of zero indicates perfect equity. Since admissions are a health good, concentration curves falling below the line of equity indicate a system that disproportionately benefits the wealthier individuals–i.e., where admissions are more frequent for the wealthy. A larger concentration index indicates greater inequity. A Wilcoxon rank-sum test was completed comparing the wealth distribution of all inpatients, as well as the defined subgroups, to the wealth distribution of the community population. The same test was performed to compare the wealth distribution of the subgroups to wealth distribution of all inpatients.
N/A
