Addressing inequity to achieve the maternal and child health millennium development goals: Looking beyond averages

Background: Inequity in access to and use of child and maternal health interventions is impeding progress towards the maternal and child health Millennium Development Goals. This study explores the potential health gains and equity impact if a set of priority interventions for mothers and under fives were scaled up to reach national universal coverage targets for MDGs in Tanzania. Methods. We used the Lives Saved Tool (LiST) to estimate potential reductions in maternal and child mortality and the number of lives saved across wealth quintiles and between rural and urban settings. High impact maternal and child health interventions were modelled for a five-year scale up, by linking intervention coverage, effectiveness and cause of mortality using data from Tanzania. Concentration curves were drawn and the concentration index estimated to measure the equity impact of the scale up. Results: In the poorest population quintiles in Tanzania, the lives of more than twice as many mothers and under-fives were likely to be saved, compared to the richest quintile. Scaling up coverage to equal levels across quintiles would reduce inequality in maternal and child mortality from a pro rich concentration index of -0.11 (maternal) and -0.12 (children) to a more equitable concentration index of -0,03 and -0.03 respectively. In rural areas, there would likely be an eight times greater reduction in maternal deaths than in urban areas and a five times greater reduction in child deaths than in urban areas. Conclusions: Scaling up priority maternal and child health interventions to equal levels would potentially save far more lives in the poorest populations, and would accelerate equitable progress towards maternal and child health MDGs. © 2012 Ruhago et al.; licensee BioMed Central Ltd.

We use disaggregated data from Tanzania to reflect mortality and coverage in five wealth quintiles from the poorest to the richest and in rural and urban areas. Baseline coverage and mortality data for this study were extracted from the openly available, 2010 Tanzania Demographic and Health Survey (TDHS) [8]. Permission to conduct research was sought and obtained from the Tanzania National Institute of Medical Research (NIMR). We define universal coverage as 80–90% coverage, acknowledging that the ideal 100% coverage may be hard to reach. For endpoint coverage, we used targets from the 2008 Tanzania National Strategic Plan for reduction of maternal, newborn and child mortality (90% for most targets) [19]. In case national targets were lower than the current TDHS 2010 coverage levels in any of the sub-national or socioeconomic groups, TDHS data were used as endpoint coverage. Table ​Table11 below provides a summary of interventions, coverage estimates and targets. Intervention coverage (%) for maternal and child health interventions by wealth quintiles and geographical residence used as input in LiST We used the Lives Saved Tool (LiST) version 4.47 for modeling. LiST is free, downloadable software and is part of the spectrum policy modeling system developed by the John Hopkins University [20]. The tool was used to model the potential health impact of scaling up priority health interventions on maternal and child mortality for a period of five years. In this study, the baseline year is 2011 and the final year is set at the target for Millennium Development Goals, 2015. LiST is pre-loaded with country specific average data. To allow for wealth quintile and urban vs. rural analysis, we adjusted the national demographic projection to obtain population estimates for each of the five wealth quintiles as well as urban and rural areas. In other words, we partitioned the whole population into seven “sub-populations” or sub-groups. The national total fertility rate was adjusted by the five wealth quintiles and urban/rural estimates of fertility rates from Tanzanian health and demographic surveys from 1992 to 2010. The adjusted fertility rate was applied from the first year of population to the target year. The proportion of each of the quintiles, urban/rural areas to the total national population was multiplied by the first year population of the national population estimates preloaded in LiST to estimate each of the sub-group populations. Migration values were adjusted to zero. The maternal mortality ratio and under-fives mortality rates by SES quintile and urban/rural were updated for the sub-group analysis using current data from TDHS 2010. Default data for cause-specific mortality was used. However, we assumed that the higher/lower than average neonatal, infant and under-five mortality rates in each quintile reported in demographic and health survey were distributed in proportion to the original distribution of cause-specific mortality. The family planning module was updated, the total fertility rate and the unmet need for family planning was adjusted to reflect the sub-group current data. The LiST user manual provides detailed procedures for sub-group modeling [21]. The data on the effectiveness of interventions are default in LiST, updated frequently from comprehensive reviews under the Child Health Epidemiology Reference Group (CHERG) [22]. We entered the baseline coverage for each quintile, urban/rural and national level for a set of high impact priority interventions for maternal health (skilled birth attendance and health facility delivery, as proxy predictors of Basic Emergency Obstetric Care and Comprehensive Emergency Obstetric Care) into LiST. Similarly, coverage data per quintile and urban/rural for child health interventions (oral rehydration salts (ORS) for diarrhoea management, antibiotic for pneumonia treatment, Insecticide Treated Nets (ITN) and artemisinin-based combination therapy (ACTs) for the management of malaria) were entered. The TDHS 2010, does not report maternal mortality by wealth quintile, so the lowest, midpoint and high estimates were used for quintiles. To account for any possible biases the two lowest quintiles (40%) likely to have higher maternal mortality were assigned with the highest estimates of maternal mortality ratio. The modeling exercises were done by linking intervention coverage, effectiveness and cause of mortality. We observed the expected change of mortality in maternal and under-fives and lives saved over the five-year period. Details on the assumptions built into the LiST module have been well documented elsewhere [23,24]. Concentration curve and concentration index were used to measure the equity impact of the priority intervention scale up. A concentration curve is used to display the distributional impact of wealth related inequity in MMR and U5M, (Figures ​(Figures11 and ​and2).2). The baseline and endpoint mortality measured before and after intervention scale up (maternal or under five mortality) were cumulatively plotted on the y-axis, against the cumulative proportion of (mothers or under-fives) population ranked by their socioeconomic status from lowest to highest on the x axis. When the curve lies on the line of equality, all mothers or under fives, regardless of their socioeconomic status have the same mortality. If it lies above the line of equality, mortality is more prominent amongst the poorest population, indicating a pro-rich distribution. On the other hand if the curve lies below the line of equality, this indicates lower mortality in the poorest population, hence a pro poor distribution. To obtain the magnitude of inequality, we used the concentration index [25]. The measure ranges from −1 to 1, with a zero index indicating no wealth related inequity and a negative index indicating higher maternal or under five mortality among the poor. Degree of inequality in maternal mortality. Degree of inequality in under five children mortality.

N/A

• ruhago, george
• ngalesoni, frida
• norheim, ole frithjof