Menu
Menu
Menu
Menu
Objective To determine if improved geographical accessibility led to increased uptake of maternity care in the south of the United Republic of Tanzania. Methods In a household census in 2007 and another large household survey in 2013, we investigated 22 243 and 13 820 women who had had a recent live birth, respectively. The proportions calculated from the 2013 data were weighted to account for the sampling strategy. We examined the association between the straight-line distances to the nearest primary health facility or hospital and uptake of maternity care. Findings The percentages of live births occurring in primary facilities and hospitals rose from 12% (2571/22 243) and 29% (6477/22 243), respectively, in 2007 to weighted values of 39% and 40%, respectively, in 2013. Between the two surveys, women living far from hospitals showed a marked gain in their use of primary facilities, but the proportion giving birth in hospitals remained low (20%). Use of four or more antenatal visits appeared largely unaffected by survey year or the distance to the nearest antenatal clinic. Although the overall percentage of live births delivered by caesarean section increased from 4.1% (913/22 145) in the first survey to a weighted value of 6.5% in the second, the corresponding percentages for women living far from hospital were very low in 2007 (2.8%; 35/1254) and 2013 (3.3%). Conclusion For women living in our study districts who sought maternity care, access to primary facilities appeared to improve between 2007 and 2013, however access to hospital care and caesarean sections remained low.
We used information from two geo-referenced household surveys covering the same five districts in the south of the United Republic of Tanzania: (i) a census of all 243 612 households in 2007 – primarily designed to evaluate the impact of intermittent preventive treatment with antimalarials on infant survival;24 and (ii) a sample survey in 2013 that assessed the impact of a home-based counselling strategy on neonatal care and survival.25 In both surveys, the study population comprised women who had had a live birth in the 12 months before the survey and reported on uptake of pregnancy and intrapartum care. The study area covers three districts of the Lindi region and two districts of the Mtwara region.26 Most of the residents of these districts are poor and live in mud-walled houses in rural villages. Between 2009 and 2013, two dispensaries in the study area were upgraded to become health centres and 14 new dispensaries were inaugurated. By 2013, the study population was served by 156 dispensaries, 15 health centres and six hospitals within the study area and by another two hospitals just outside the district boundaries. All except four of the 179 health facilities serving the study area in 2013 – i.e. two mission hospitals, one mission dispensary and one private health centre – were public facilities that provided maternal health services free of charge.27 In both 2007 and 2013, all eight hospitals serving the study area provided caesarean sections on a daily 24-hour basis, three of the hospitals had maternity waiting homes and all of the hospitals and seven of the health centres were equipped with ambulances. Ambulance use – e.g. for hospital referral – was, however, severely constrained by shortages of fuel, human resources and funds for repair. Although all except one of the 179 facilities offered delivery care, basic emergency obstetric care was not consistently available in the study area.27–29 The survey methods are described in detail elsewhere.24,25 In brief, we used a modular questionnaire, administered in Swahili, to assess coverage of essential interventions during pregnancy and childbirth. Use of personal digital assistants to collect data facilitated the checking of standard ranges, consistency and completeness at the time of data entry.30 Household wealth was assessed by asking each household head about household assets and housing type. We mapped the study households using a global positioning system. The positions of the relevant health facilities had been recorded in previous surveys. In 2007, we surveyed all 243 612 households in the five study districts. In 2013, however, we sampled 169 324 households, which were selected by following a two-stage sampling survey.25 Using the results of the national 2012 census, in which 247 350 households were recorded in the study area, we first sampled so-called subvillages. This sampling was proportional to the number of households in each subvillage – typically about 80–100. We included all households in the subvillages with fewer than 130 households, but used segmentation for subvillages with more than 131 households. Our main outcomes of interest were uptake of at least four visits for antenatal care, delivery in a health facility and delivery by caesarean section. Using a combination of coordinates and the nearstat command in Stata version 13 (StataCorp. LP, College Station, United States of America), we calculated straight-line distances between each surveyed household and: (i) the nearest antenatal clinic, which could have been in a primary facility or a hospital; (ii) the nearest primary facility offering delivery care; and (iii) the nearest hospital. We did this separately for 2007 and 2013. In the 2007 survey, we attempted to impute the coordinates of households for which no such coordinates were recorded, from the coordinates for neighbouring households. Household wealth quintiles were constructed separately for 2007 and 2013, using principal component analysis.31 All analyses were conducted in Stata version 13. For the 2013 data, we accounted for the different sampling structures of the 2007 and 2013 surveys by weighting subvillages by the inverse chance of being included. The percentages reported for 2013 – but not those reported for 2007 – are therefore weighted values. For both 2007 and 2013, we assessed the effect of: (i) distance to nearest antenatal clinic on uptake of at least four visits for antenatal care; (ii) distance to nearest primary facility on delivery in a primary facility; (iii) distance to nearest hospital on hospital delivery; and (iv) distance to nearest hospital on birth by caesarean section. For the analysis of the effect of distance on delivery in a primary facility, we excluded births where a hospital was the nearest facility. We first used generalized linear models to calculate crude prevalence ratios (cPR) with 95% confidence intervals (CI). We compared the prevalence of each indicator by increasing distance to a primary health facility or hospital and then compared the prevalence of each indicator between 2007 and 2013 within each distance group.32 We adjusted the crude prevalence ratios for potential confounding by the mother’s age, parity, district of residence, education, ethnic group and occupation and her household’s wealth quintile. Using multilevel logistic regression without weighting, we fitted an interaction term between distance to facility and survey year and used the likelihood ratio test to calculate a corresponding P-value. We also used ArcGIS version 9.2 (ESRI, Redlands, USA) to map the absolute increases in facility delivery and caesarean section by administrative ward – as percentages of the live births – between 2007 and 2013. Ethical clearance was obtained from the institutional review boards of Ifakara Health Institute, and the Tanzanian National Institute of Medical Research and the ethics committees of the London School of Hygiene and Tropical Medicine and the Swiss cantons of Basel-Stadt and Basel-Land. The study population was informed about the surveys by the local government authorities and again, one day prior interview, by a sensitizer who used information sheets in the local language. Written consent to participate was obtained from household heads and the women who answered questions about pregnancy and childbirth.
N/A
