Health insurance is important in improving maternal health service utilization in Tanzania—analysis of the 2011/2012 Tanzania HIV/AIDS and malaria indicator survey

This cross-sectional study was designed to utilize secondary data originated from the nationally representative sample of men and women aged 15–49 years in the 2011/12 Tanzania HIV/AIDS and Malaria Indicator Survey (THMIS). Data were collected between December 2011 and May 2012. The National Bureau of Statistics (NBS) developed the sampling frame used for the survey to cover all 30 regions of Tanzania. It excluded nomadic and institutional populations such as persons in hotels, barracks, and prisons.

The sample was selected in two stages. The first stage involved the selection of 585 clusters (sample units) consisting of enumeration areas (EAs). On the mainland, 30 clusters were selected from Dar es Salaam and 20 were selected from each of the other 24 regions. In Zanzibar, 15 sample points were selected from each of the five regions.

The second stage of selection involved the systematic sampling of households. A household listing operation was undertaken in all of the selected areas prior to the fieldwork. From these lists, households to be included in the survey were selected. Approximately 18 households were selected from each sample point for a total sample size of 10,496 households. Weighting factors were added to the data file so that the results would be proportional at the national level. Eligibility criteria included women and men ages 15–49 who were either permanent residents of the selected households or visitors who stayed in the household the night before the survey were.

Data analyzed in this study came from the Individual Questionnaire of the 2011/2012 THMIS dataset. It included a total of 19,319 men and women. Out of them, a total of 8352 men were excluded to get the study population of women who had a live birth within three years prior to the survey. Out of the 10,967 women remained, only 4627 women had one or more live births in the past three years before the survey were selected. A total of 114 women were excluded for missing data on important variables. They include 14 women who were missing information on insurance coverage and 100 women who were missing information on the timing of the first antenatal visit. Finally, 4513 women who had one or more live births within three years before the survey were included for analysis. The data set for this survey is available upon request from Measures, DHS program. The URL for this specific dataset is https://​www.​dhsprogram.​com/​data/​dataset_​admin/​login_​main.​cfm;jsessionid=​4E33501BD1C8593F​A55A278F027BFD44​.​cfusion?​CFID=​12989464&​CFTOKEN=​c9b926a975283d4-05E5988A-9609-2AFA-E0772649E1FDACED​

Variables were identified from the 2011–12 THMIS individual questionnaires and selected based on previous literature. The outcome variable of interest was maternal health services utilization. Three items from the THMIS individual questionnaire were used to measure it. They include: timing of the first antenatal care visit, completing the recommended number of ANC visits, and giving birth at a health facility under skilled care. Timing of the first ANC visit is recommended during the first trimester. Four spaced visits are the minimum number of recommended focused ANC visits. Lastly, a woman is considered as having received skilled birth assistance at delivery if she gave birth in a health facility, under the care of a trained health worker with the necessary skills for assisting with birth [3].

The main independent variable in this study was health insurance coverage. Participants were asked if they have any type of health insurance eliciting a dichotomized response of “Yes” or “No.” Other covariates and confounders in this analysis included socioeconomic and demographic variables. Demographic variables included age clustered in five-year age groups, rural or urban area of residence, education level, marital status, occupation, and wealth status. Number of births in the past five years before the survey was dichotomized into “one child” or “2 or more births.” Tanzania’s regions were regrouped into nine geographical zones, including Eastern, Western, Southern, Southern High, South West Highlands, Central, Northern, Lake and the Zanzibar zone.

We used weighted wealth index to assess economic status. It was constructed using ownership of household assets ranging from a television to a bicycle or car; dwelling characteristics, such as source of drinking water, sanitation facilities, and type of flooring material; and household’s food availability or consumption. Three steps were undertaken; first, a subset of indicators common to urban and rural areas was used to create wealth scores for households in both areas. Second, separate factor scores were produced for households in urban and rural areas using area-specific indicators using principle component analysis. Third, the separate area-specific factor scores were combined to produce a nationally applicable combined wealth index by adjusting area-specific scores through a regression on the common factor scores. Wealth quintiles (from lowest to highest) were formed by dividing the generated wealth index into the five rankings—the wealth index quintiles.

A multiple logistic regression was conducted to examine independent associations between health insurance coverage with the three items used to measure maternal health services utilization. The generated adjusted odds ratios (AOR) were estimated to assess the strength of the associations and used the 95% confidence intervals (CIs) for significance testing. All the covariates were entered simultaneously into the multiple regression models. Analyses were performed using SPSS version 16. Sample weighting was applied to allow for adjustments for the cluster sampling design and sampling probabilities across clusters and strata.

Data of 4513 participants was available for analysis. Health insurance coverage was low. Of all participants, only 281 (6.2%) reported to have health insurance. Among those insured 167 (59.5%) were covered under community health insurance schemes, 95 (33.8%) covered under employment based health insurance schemes and the remaining 4.5% and 1.8% were covered under social security funds and other types of insurance schemes respectively.

Table 1 shows the results of descriptive characteristics stratified by health insurance coverage. Health insurance coverage varied with education level, occupation, marital status, place of residence, and wealth index. For example, it increased from non-educated women to secondary and above level (2.8% to 13.2%, p < 0.001). It also increased from low wealth index (3.6% to 12.7%, p < 0.001). Women residing in rural areas had lower rates compared to their counterparts in urban areas (5.8% vs. 8.5%, p < 0.01). Unemployed individuals had lower coverage compared to their employed counterparts (5.4% vs. 23.2%, p < 0.001). Health insurance coverage was high among women at the age groups 25–39 years.

Table 2 shows the characteristics of 4513 study participants stratified by their utilization of maternal health services. Utilization of maternal health services was categorized into timing of first ANC visit, completion of the recommended ANC visits, and attended by skilled health worker during labor.

In this study, timing of the first ANC visit was also low. Only 762 (16.9%) of participants reported proper timing. A higher proportion of those with health insurance had proper timing of ANC compared to their counterparts (27.0% vs. 16.2%, p < 0.001). Timing of the first ANC visit differed across various socio-demographic characteristics. For example, it was higher among women in the age group 20–24 (20.9%), residing in rural areas (20.5%), and women with at least primary education or higher education (primary incomplete 18.8%, primary complete 18.1% and secondary or higher education 17.4%) compared to those in other age groups (χ2 test p < 0.01), residing in urban areas or with no formal education (χ2 test p < 0.001). Also, timing of the first antenatal care visit increased proportionally with the wealth index. For example, 23.4% and 18.4% of those in the richest and richer quintile, respectively, had recommended timing of first visit compared with only 14.4% of the poorest group quintile (χ2 test p < 0.001). Only 14.9% of the employed mothers had proper timing of the first ANC visit compared to 22.8% among those who were self-employed (χ2 test p < 0.001). There were proportionally significant variations across the nine survey zones with the Eastern zone showing the highest proportion of women timing their first antenatal care visit as recommended (31.1%) compared to only 11.7% and 11.3% the Lake and the Central zones respectively (χ2 test p < 0.001).

A few women had completed the recommended ANC visits (Table 2). Only 322 (7.1%) completed four recommended ANC visits. There was no difference between those with or without health insurance under such context (6.8% vs. 7.2%, p = 0.460). Younger women were less likely to complete all four recommended ANC visits compared to other age groups (χ2 test p < 0.01). This was also true for the urban-rural residency (2.2% vs. 8.1%, p < 0.001). Those with low education had higher completion rates compared to the higher education cluster (χ2 test p < 0.001). Similarly, women of the richest quintile had the lowest completion rate of ANC visits compared to poorest quintile (2.4% vs. 9.9%, p < 0.001). Completion rate was also lower among single women compared to their married counterparts (χ2 test p < 0.001); and those residing in the Eastern zones (χ2 test p < 0.001).

Lastly, majority of the selected population (56.5%) reported to have delivered at health facilities under skilled attendants. A higher proportion of them had health insurance compared to their counterparts (77.6% vs. 55.1%, p < 0.001). A high number [1963 (43.5%)] of women did not have their birth attended by a skilled health worker. More young women gave birth in healthcare facilities under skilled health workers compared to older groups (χ2 test p < 0.01). Counter intuitively, a significantly high proportion of women residing in rural areas gave birth under skilled health workers than those in urban areas (86.1% vs. 50.8% p < 0.001). About 76.9% of women with secondary or higher education gave birth under skilled attendance compared with only 39.5% of those with no education (χ2 test p < 0.001). Women of the richer quintile category received skilled attendance at birth in higher proportion compared to those of the poorest wealth quintile (86.0% vs 39.1%, p < 0.001). Contrary to the results of ANC completion rates, 75.6% of single women gave birth under skilled attendance, while only 55.3% of the married women gave birth under skilled attendants (χ2 test p < 0.001). Occupational status, parity and the geographical zone also showed significance differences with skilled birth in this survey (Table 2).

Table 3 shows the results of a multivariate logistic regression analysis for the association between health insurance coverage and utilization of maternal health services. After controlling for other factors, health insurance coverage was associated with increased odds of timing the first antenatal visit as recommended (OR 1.86, p < 0.001). Women with health insurance were also more likely to deliver at a health facility under a skilled attendant compared to their counterparts (AOR 2.01 p < 0.001). There was no significant association between health insurance coverage and completing the recommended number of antenatal visits (AOR 1.33, 95% CI 0.782–2.267).

Other factors associated with timing the first antenatal visit as recommended were being in the 20–24 age group compared to the youngest age group (AOR 1.47 p < 0.01) and residing in an urban compared to rural area (AOR 1.26 p < 0.01). In the ANC completion rate model, age was the only factor associated with completing the recommended number of antenatal visits. Compared with young women in the 15–19 years, being older increased the odds of completing the recommended number of antenatal visits (p < 0.01).

Table 3 also presents results of a regression analysis in the skilled birth attendants’ model. Counter intuitively; women residing in urban areas were less likely to deliver under skilled workers (AOR 0.511 p < 0.001). Being a young mother (ages 15–19) increased the odds of delivering at a health facility compared to those in 20–24 years and above age groups. Residents from the Southern highlands were more likely to deliver under skilled health worker than those from the Eastern zone (AOR 3.67 p < 0.001).