Menu
Menu
Menu
Menu
Background: A model health district was initiated through a program of health system strengthening (HSS) in Ifanadiana District of southeastern Madagascar in 2014. We report population health indicators prior to initiation of the program. Methods: A representative household survey based on the Demographic Health Survey was conducted using a two-stage cluster sampling design in two strata – the initial program catchment area and the future catchment area. Chi-squared and t-tests were used to compare data by stratum, using appropriate sampling weights. Madagascar data for comparison were taken from a 2013 national study. Results: 1522 households were surveyed, representing 8310 individuals including 1635 women ages 15-49, 1685 men ages 15-59 and 1251 children under age 5. Maternal mortality rates in the district are 1044/100,000. 81% of women’s last childbirth deliveries were in the home; only 20% of deliveries were attended by a doctor or nurse/midwife (not different by stratum). 9.3% of women had their first birth by age 15, and 29.5% by age 18. Under-5 mortality rate is high: 145/1000 live births vs. 62/1000 nationally. 34.6% of children received all recommended vaccines by age 12 months (compared to 51.5% in Madagascar overall). In the 2 weeks prior to interview, approximately 28% of children under age 5 had acute respiratory infections of whom 34.7% were taken for care, and 14% of children had diarrhea of whom 56.6% were taken for care. Under-5 mortality, illness, care-seeking and vaccination rates were not significantly different between strata. Conclusions: Indicators of population health and health care-seeking reveal low use of the formal health system, which could benefit from HSS. Data from this survey and from a longitudinal follow-up study will be used to target needed interventions, to assess change in the district and the impact of HSS on individual households and the population of the district.
Madagascar, an island nation off the east coast of the African continent, is one of the poorest countries in the world, with 81.7% of the population living on less than $1.90 USD per day in 2010 [9]. Ifanadiana District, in the southeastern region of Madagascar, consists of approximately 148,000 residents, gathered into 13 communes and 195 fokontany (the smallest administrative unit). Six of the communes are only accessible by foot or motorcycle. The district contains the eastern part of Ranomafana National Park, a montane rainforest and United Nations Educational, Scientific and Cultural Organization (UNESCO) World Heritage Site. The HSS program based on the World Health Organization’s (WHO’s) six building blocks [1] supports the health system at three administrative levels: the district hospital; local health centers at the commune administrative level; and village community health workers at the fokontany level. Details on the activities of the program can be found in a forthcoming paper [10]. In brief, through partnership with the MoH, the program works at all levels of care (community health, health center and hospital) and includes the following interventions: infrastructure, staffing (quantity and capacity) and equipment (to strengthen service delivery and quality of care); strengthening procurement systems (to enable consistent drug and medical supplies); instituting an ambulance network (to improve physical access to emergency care); strengthening community health care (to assist in prevention, treatment and referral, improving access); implementing social programs (to institute needed services at village and commune level); removal of user fees (to remove financial barriers to access to care at all levels); and monitoring, evaluation and supervision (to strengthen all levels). Complementary to these interventions, specific clinical programs, including malnutrition, tuberculosis and maternal and child health, span all three levels of the health system. The first phase of work from January 2014 through April 2016 covered 4 out of 13 communes – a population of about 65,000 people – with plans to progressively roll out services to the entire district. Although initial strengthening was not randomly allocated (for logistical reasons, the first sites chosen for the HSS initiative were those near Ranomafana National Park and along the main paved road in the district), this staggered roll-out of activities allows us to assess the impact of the HSS as it unfolds over time, in which some communes receive services first (‘initial catchment areas’) and others later (Rest of District, ‘RoD’). Before any of the program’s activities had begun (other than establishment of the formal partnership between PIVOT and the MoH), we conducted a baseline survey of the health and socio-economic conditions of the district. We conducted a representative survey of Ifanadiana District using tools and methods based on the Demographic and Health Surveys (DHS) [11], a project that conducts nationally representative household surveys in low- and middle-income countries (including Madagascar) approximately every five years to assess population, health and nutrition indicators. The survey was conducted by the Madagascar National Institute of Statistics (INSTAT), which also conducts the Madagascar DHS. Our sample consisted of 1600 households selected using a two-stage cluster sampling scheme involving 80 clusters and 2 strata (PIVOT’s initial area and RoD). Eighty clusters based on maps from the 2009 census were selected at random from the total available – 40 clusters within the initial catchment area and 40 from the RoD (Figure 1). The clusters were mapped and geographical information system (GIS) coordinates were collected for each cluster. Twenty households per cluster were randomly selected for inclusion. Five teams of five (four data collectors and one field supervisor per team) conducted face-to-face surveys with participating households in April and May of 2014. Eligibility criteria for interview were based on DHS standard criteria and included individuals of reproductive age (defined as the age ranges of 15–49 for women, 15–59 for men) who were de facto residents of the household (usual members or had spent the night prior in the household). Sensitization of the community prior to field work was performed through meetings with local mayors, government and MoH representatives and village leaders. Sampling design used for the baseline survey in Ifanadiana District. Map shows the limits of the Ifanadiana District and its 13 communes. A total of 80 clusters were selected, half inside the initial PIVOT catchment area (blue) and half outside (tan), with probabilities proportional to the population size. Our indicators of interest for health and economic well-being were defined a priori and are presented in Box 1. Core indicators were chosen to represent key access- and health- related factors that a strengthened health system is expected to influence, and were based on those from the Doris Duke Charitable Foundation’s African Health Initiative Population Health Implementation and Training (PHIT) partnership [12–14]. For example, strengthening community health services and health worker training and capacitation have been shown to reduce neonatal and infant mortality [15], and a 2014 study found that elimination of user fees increased care coverage, with accompanying substantial reductions in estimated neonatal and child mortality [16]. Our indicators for care access and utilization outcomes include vaccination rates (DTP [diphtheria-tetanus-pertussis] – at least one dose and all three recommended doses, measles, and full immunization coverage by 24 months), care-seeking for children’s illnesses (defined as treatment sought from trained health professional or clinic for diarrhea, acute respiratory illness, and fever), antenatal care and delivery with a trained health professional. Our indicators for population health impact include under-five mortality, under-five children with diarrheal, febrile or acute respiratory infection (ARI – defined as cough with shallow, rapid breathing), illness in the last two weeks and maternal mortality. Health indicators, Ifanadiana District, Madagascar, 2014. We collected the following data: basic household composition and living conditions, including gender, ages and relationships among household members; structure and physical condition of the house; type and condition of latrine or toilet; electricity; and access to water, among others. Child health characteristics included date of birth, nutrition, anthropometry (height and weight), vaccination status, health insurance, diarrheal or febrile illness within two weeks prior to interview, access to selected health services, treatment for malaria, or presence of an illness requiring a visit to a health center in the last three months prior to the interview. Adult health characteristics included measures of febrile and diarrheal illness, injuries, respiratory illness, other chronic and acute conditions, anthropometry, health care utilization and reproductive history. We also collected information on adult and under-five mortality [17]. Economic data collected included 30-day and 12-month income and expenditure data, employment data, household goods, land and livestock ownership, and time lost to work or education because of illness or injury. Our questionnaires were a subset of those used in Madagascar for the DHS [11], with added questions on child development from the Malagasy version of Multiple Indicator Cluster Survey-version 4 (MICS) [18] and questions on adult health from Rwanda’s Questionnaire de Bien-Etre [19]. Economic status was assessed using tools adapted from the World Bank’s Living Standard Measurement Survey (EICV) [20]. We successfully conducted a pilot of tools and methods in February 2014 in 80 rural households in a different district. All French and Malagasy questionnaires, data collection and analysis methods were standardized and had been reviewed and approved the year prior to the study by the Madagascar National Ethics Committee and re-reviewed through internal processes at INSTAT. The study was also reviewed by the Harvard Medical School Institutional Review Board (IRB). All individual identifiers were removed prior to database delivery to the Investigators. The ongoing longitudinal component of the study was re-reviewed and approved by Harvard Medical School IRB and the Madagascar National Ethics Committee. All eligible women and men who were in the households sampled were interviewed. All residents, including children, were weighed and had their height measured (or, in the case of infants, length). No biologic samples were taken. All health indicators were defined using the standard techniques of the DHS or MICS and are presented in Box 1. Under-five mortality was estimated using the synthetic life-table method in which probabilities of death are calculated for small age segments, then combined into overall mortality rates [17]. Under-five mortality was defined as deaths per 1000 live-born children ages 0–59 months and 29 days. Maternal mortality was estimated using the sisterhood method [17]. In brief, women were asked about the deaths of their sisters who were older than 15 years at their demises. Deaths during delivery or within two months of the end of a pregnancy were considered to be deaths due to maternal causes. Vaccination rates were defined as the proportion of children aged 12–23 months who received all recommended vaccines prior to their first birthdays. Incidence of childhood illness was defined as the proportion of children under five years of age who were ill with a specific condition in the two weeks prior to the survey. Wealth indices were calculated as in the DHS method, using principal components analysis [17]. Factors that comprised the wealth index included access to electricity, water and toilets, material of roofing for houses, number of residents per bedroom and type of cooking fuel. Cutoff points for wealth quintiles were determined as the values of the wealth index closest to but less than 20%, 40%, 60%, and 80% of the cumulative distribution of the household wealth index. Data were entered into CSPro and analyzed using SPSS (SPSS Inc., Armonk, NY), Stata 13.1 (StataCorp, College Park, TX) and SAS® 9.3 (SAS Institute, Inc., Cary, NC) to calculate under-five mortality 95% confidence intervals. Sampling weights were calculated for the household, women’s and men’s surveys. To compare data between strata, we conducted chi-squared and t-tests for categorical and continuous data, respectively. All analyses, including the estimation of means, standard errors and tests, used the survey procedures available in Stata, applying appropriate sampling weights and using Taylor linearized variance estimation. To compare our data to national estimates, we used reported estimates from the 2012–2013 ENSOMD (Enquete Nationale sur le Suivi des Objectifs du Millenaire pour le Developpement a Madagascar) survey for comparison [21–23]. ENSOMD, a nationally representative health and economic survey of 75,931 individuals in 16,920 households, was conducted in Madagascar in 2012 and 2013 to assess progress toward meeting the Millennium Development Goals. A report on progress toward each of seven goals is publicly available (http://www.mg.undp.org/content/madagascar/fr/home/library/mdg/publication_1.html). Raw data from this survey were not available, so statistical analyses of differences between district- and national-level estimates are not provided.
N/A
