Every year, hundreds of papers are published using DHS Program data. Datasets for individual DHS surveys are available for download from The DHS Program website by completing a simple registration form. Researchers and students can also access the IPUMS DHS data collection to facilitate comparative analysis of DHS Program data. Harmonized across time and space, IPUMS DHS features consistently named and coded variables for 32 African countries and 9 Asian countries, with more datasets constantly being added.
The IPUMS DHS team recently announced the winners of this year’s IPUMS Global Health paper awards, which highlight some of the most exciting research to take advantage of the IPUMS DHS integrated datasets. These papers were selected from a highly competitive field that included nearly 50 studies released in 2020.
Fan and Loria resolve a puzzle in prior research on intimate partner violence (IPV): Why is the relationship between IPV and contraceptive use negative in some countries and positive in others? Using 30 IPUMS DHS samples from 17 countries, the authors demonstrate that the relationship between IPV and family planning is modified by macro contextual factors, including legal prohibitions and national levels of female empowerment. This study stands out not just for answering an important social science question but also in its creative use of the broad range of information collected in DHS Program surveys, including variables on contraceptive use and type, family size preferences, husband-wife disagreement on fertility goals, various indicators of women’s status (e.g., education, employment, decision-making), and domestic violence. In addition, the authors draw on IPUMS DHS variables to determine the direction of causality: from the experience of IPV to increased contraceptive use, rather than from contraceptive use to increased incidence of IPV.
This study leverages geographic heterogeneity to determine the effect of reduced malaria burden on low-birth-weight rates across communities in 19 sub-Saharan African countries. Low birth weight is a serious health risk associated with cognitive and physical difficulties among children. This careful and cleverly designed study analyzes IPUMS DHS data from countries with at least two surveys and GPS data on survey cluster locations. After using optimal matching to pair DHS Program survey clusters separated in time, the authors use a difference-in-difference approach to compare the incidence of low birth weight in areas that did and did not experience malaria decline. Results reveal a substantial decline in low birth weight resulting from declines in malaria prevalence, especially for first-born children.
Congratulations to these scholars on this impressive accomplishment!
IPUMS DHS is a system that makes it easy to find and review thousands of DHS survey variables and to download a single fully harmonized data file with only the variables and samples that interest you. IPUMS DHS currently includes variables from DHS survey samples from 32 African and 9 Asian countries; more samples are constantly being added.
Special thanks to our guest blog contributors, Elizabeth Boyle and Miriam King!
Elizabeth Heger Boyle is Professor of Sociology & Law at the University of Minnesota. She studies the role of international laws and policies on women and children’s health around the world. She has written extensively on the impetus for and impact of laws related to female genital cutting, including the book Female Genital Cutting: Cultural Conflict in the Global Community. Her current research focuses on abortion policies globally and their effects; this includes a 2015 article in the American Journal of Sociology. Professor Boyle is currently co-Principal Investigator (with Dr. Miriam King) on IPUMS DHS, a National Institute of Child Health and Human Development grant that integrates Demographic and Health Surveys over time and across countries to make them more user-friendly for researchers. Professor Boyle has a Ph.D. in Sociology from Stanford University and a J.D. from the University of Iowa.
Miriam L. King is a Senior Research Scientist at the Institute for Social Research and Data Innovation at the University of Minnesota. She has managed data integration projects on the U.S. Current Population Survey, the U.S. National Health Interview Survey, and, most recently, the Demographic and Health Surveys. Her research has focused on the history of the U.S. census, data integration methods, U.S. historical fertility differences, living arrangements, and disparities in access to insurance for same-sex couples. Dr. King has a Ph.D. jointly in Demography and History from the University of Pennsylvania.
Don’t read French? You can use the translate feature at the top of the page!
Si vous êtes intéressé à être présenté dans la série de blogs ‘DHS Data Users’, faites-le nous savoir ici en nous soumettant votre exemple d’utilisation des données du programme DHS.
Comment êtes-vous impliqué
dans les enquêtes de The DHS Program ?
J’ai travaillé pendant trois années comme superviseur national du volet ménage des Enquêtes Démographiques et de Santé (EDS)-Continue au Sénégal. Durant cette expérience, en plus du suivi de la collecte, j’ai participé activement dans le traitement des données (éditions secondaires) et dans l’analyse des données. Lors de l’EDS-Continue, j’ai bénéficié de formations en matière de :
L’échantillonnage de l’EDS ;
Les procédures de traitement des données de
Les méthodes dynamiques de formation pour adultes.
Cette dernière a changé ma façon d’animer les ateliers. En toute modestie, si aujourd’hui la qualité de mon enseignement ou d’animation est appréciée, c’est en grande partie grâce à la formation sur les méthodes dynamiques de formation pour adultes que j’ai suivi.
Et les ateliers de l’utilisation et de l’analyse de données des enquêtes de
The DHS Program ?
2019, The DHS Program m’a engagé pour animer un atelier, Tendances des
Indicateurs du Paludisme au Bénin.
L’atelier portait sur l’analyse des tendances temporelles des indicateurs du paludisme au Bénin. La finalité était de contribuer à l’amélioration des capacités des 18 acteurs opérationnels de la mise en œuvre du programme paludisme par :
une formation sur la méthodologie de la collecte, le calcul des indicateurs clés du paludisme, notamment l’identification de numérateur et du dénominateur de chaque indicateur;
un examen des intervalles de confiances et leur interprétation
Comment utilisez-vous les données des enquêtes de The DHS Program lors de votre travail actuel ?
Depuis 2018, je
suis Data Manager de l’Evaluation prospective des programmes du Fonds Mondial
(Tuberculose, VIH, Paludisme) où je suis chargé de la gestion et de l’analyse
des données de l’évaluation.
Les données de l’EDS nous permettent de vérifier l’exactitude des données des programmes de santé mais aussi d’estimer les connaissances, attitudes et pratiques de la population sur les maladies telles que la tuberculose, le VIH ou encore le paludisme.
Quel est le problème de
population ou de santé qui vous intéresse le plus, et pourquoi ?
particulièrement intéressé par la planification familiale. La plupart des décès
maternels restent liés aux grossesses rapprochées, trop nombreuses, précoces ou
tardives. C’est pourquoi la planification familiale constitue une méthode des
plus efficaces et peu couteuse pour améliorer la santé des femmes et des
enfants et lutter contre la mortalité maternelle et infantile.
En effet, en dépit des progrès réalisés dans le secteur de la santé au Sénégal, les taux de mortalité maternelle et infantile n’ont pas connu l’évolution espérée ces dix dernières années. Le pourcentage de décès de femmes liés à la grossesse est l’un des plus élevés de la sous-région (29%). Par ailleurs, la mortalité infantile (42 décès pour 1 000 naissances vivantes, EDS-C 2017) reste au même niveau depuis quelques années et ce à cause, notamment, du nombre important de décès néonataux (28 décès pour 1 000 naissances vivantes, EDS-C 2017) qui représentent la moitié des décès infantiles.
Ingénieur statisticien de formation avec
une spécialisation en informatique décisionnelle, Ibrahima GAYE est aussi
titulaire d’un Master en management de projets et d’un Master en santé publique
spécialité Méthode Quantitatives et Économétriques pour la Recherche en santé,
pour lequel il a utilisé les données de l’EDS dans le cadre de son mémoire de
Master en santé publique sur : « Analyse multiniveau de l’utilisation
de la contraception au Sénégal ». Il est maintenant en train d’écrire sa
thèse de Doctorat en santé publique : « Contribution du modèle
Age-Période-Cohorte (APC) à l’étude de la prévalence contraceptive au
Sénégal », pour lequel il utilisera également les données de l’EDS.
I then transitioned from being a workshop participant to a workshop co-facilitator, facilitating the 2017 Regional Malaria Indicator Trends Workshop in Uganda. This workshop brought together NMCP monitoring and evaluation (M&E) program managers from Liberia, Malawi, Nigeria, Sierra Leone, and Uganda to examine trends in malaria indicators.
More recently, I co-facilitated the 2018 Ghana Malaria Trends Workshop. This workshop brought together district malaria health officers to analyze trends in household survey indicators in Ghana. This was a great workshop because I was able to work with the data I am most familiar with! The output from this workshop is published on The DHS Program website.
How has NMCP
used DHS data for programmatic decision making?
After the release of the 2016 GMIS, NMCP noticed a low uptake of artemisinin-based combination therapy (ACTs) in the Northern region, but the use of SP/Fansidar was high, which is not a recommended treatment for malaria in children. This triggered us to do additional research to figure out what was going on in this region and investigate which outlets were distributing SP. We realized that people were not receiving SP from public health facilities but from private clinical shops and other drug peddlers. The 2016 GMIS results provided a snapshot of the malaria case management situation in the Northern region and provided us justification to explore further. To solve this problem, NMCP implemented a sensitization activity to ensure people in the region know the recommended treatment and sources to get the correct treatment.
Another example of evidence-based decision making was the implementation of a malaria sensitization campaign using data the 2016 GMIS. Malaria prevalence by microscopy in the Eastern region increased between the 2014 GDHS and 2016 GMIS. This was a worrying trend because in Ghana we normally only see high malaria prevalence in the Northern and Upper West regions. NMCP looked more critically at the 2016 GMIS results and saw that while insecticide-treated net (ITN) ownership was high, the proportion of people who recognized the cause and symptoms of malaria was very low. As a result, NMCP implemented a community level sensitization activity in four districts of the Eastern region.
How do you use MIS
survey data during your daily job?
I recently collaborated on a research paper using DHS data. The paper, published in The Malaria Journal, used survey data from the 2014 GDHS and the 2016 GMIS to examine ITN use behavior by exploring how several household and environmental variables related to use among Ghanaians with access to an ITN. This further analysis paper has been extremely helpful for programmatic decision making here at NMCP.
What data are
you looking forward to in the upcoming 2019 GMIS?
I am interested in further examining the information about the type of nets in households. NMCP finished a mass long-lasting insecticidal net (LLIN) distribution campaign in 2018 and implemented a school-based piperonyl butoxide (PBO) net distribution campaign in 2019. The 2019 GMIS results will provide information on the reach and use of these nets across Ghana as well as where people obtained their nets.
Written by: Samuel Oppong
Samuel Oppong is a Monitoring and Evaluation Specialist with the Ghana National Malaria Control Programme. He coordinators M&E activities in vector control interventions, routine data quality audits, and SMC. He is involved in capacity building of national, regional, district and health facility staff on capturing, reporting, and analyzing malaria-related data from routine health information systems as well as other malaria data sources. He also leads capacity building programs of national, regional, and district staff on conducting data quality audits as well as onsite training, supportive supervision (OTSS) on malaria data management.
If you are interested in being featured in the ‘DHS Data Users’ blog series, let us know here by submitting your example of DHS Program data use.
Over the past four years, the IPUMS-DHS program has grown substantially, in both the magnitude of available data and in use. As of June 2019, more than 2,000 users have accessed the IPUMS-DHS database, and multiple papers have been published using DHS data through IPUMS-DHS.
One of the advantages of accessing DHS data through IPUMS-DHS is that variables are harmonized across surveys, facilitating comparative research. Recent research using IPUMS-DHS data highlight innovative methods and fascinating results:
Becker conducted a comparative study looking at control of female sexuality and male absenteeism in 34 countries and found that women in historically pastoralist societies face more restrictive norms.
Boyle and Svec recently published a paper on the international transmission of gender norms and female genital cutting (FGC) in six DHS countries. Results indicate that women’s decision making status is an important factor in FGC abandonment.
Di Brisco and Farina explored the methodological challenge of measuring gender disparities through individual perceptions and enlightening the pertinence of the poset methodology for the analysis of ordinal variables and response profiles. They used IPUMS-DHS data from 16 African countries.
IPUMS-DHS is also being used to train the next generation of analysts and data users. The Quantitative Global Health Analysis course taught at the University of Minnesota this spring relied on IPUMS-DHS as a primary data source for its students. Final products were research posters using the data. Research questions explored by students analyzing IPUMS-DHS data included:
How Violence against Women Affects Fertility and Family Planning in Uganda
Changes in and Predictors of Antenatal Care for Women in Mali
Effects of Family Size and Food Insecurity on Child Mortality in Ethiopia
Understanding Variation in Vaccination Status in Ethiopia
Vitamin A Vaccination and Deficiency in Uganda
Perceptions of HIV/AIDS in India in the Context of Education
IPUMS-DHS Data Update: As of June 2019, the IPUMS-DHS database includes 156 samples from 38 countries and nearly 15,000 consistently coded variables, including all standard DHS variables from DHS Phases 1 through 7 and many country-specific variables. Learn more on our website and read our previous blogs on the IPUMS-DHS collaboration here.
This new blog series, DHS Data Users, captures examples of how you, the data user, have incorporated data from DHS, MIS, and/or SPA surveys into your analyses, at your institution, or to influence policies or programs. If you are interested in being featured in the ‘DHS Data Users’ blog series, let us know here by submitting your example of DHS Program data use.
The year 2018 saw an upswell of interest in health system quality with the publication of three global reports highlighting critical deficits in quality in health systems in low- and middle-income countries [1,2,3]. Much of the empirical basis for these reports was drawn from the Service Provision Assessments (SPA), the lesser-known surveys conducted by The Demographic and Health Surveys (DHS) Program, which provide comprehensive assessments of health systems in low-resource settings from Haiti to Nepal.
These surveys include a detailed audit of facility resources, provider interviews, direct observations of primary care services, and exit interviews with patients or caretakers. Each assessment is a sample of the complete health system (public and private) or in some cases a complete census. The resulting wealth of data enables assessment of structural inputs to quality of care, the care process – both competent care and user experience – and some outcomes from care, primarily user confidence in the health system. A small but increasing number of researchers is delving into all the SPA data have to offer. Among the insights the SPA surveys have yielded just from my own research are:
Most health systems assessed are not fully prepared for basic health care.
A comparative study of 8,443 facilities in 9 countries based on SPA surveys between 2007 and 2015 found that hospitals averaged between 69% (Senegal 2012-2014) and 82% (Tanzania 2015, Namibia 2009) on the service readiness index defined by the World Health Organization for primary health facilities. Non-hospitals achieved at best 68% readiness (Namibia 2009) and at worst only 41% (Uganda 2007, Bangladesh 2014) . Within primary care services – antenatal care, family planning, and sick child care – service-specific service readiness is not highly predictive of competent care being delivered.
In Kenya, where the 2010 SPA did include direct observation of labor and delivery, both structural quality of maternity care and observed clinical quality was higher in facilities in wealthier areas than facilities in poorer areas, with women in the poorest areas receiving care that complied with only half of recommended clinical guidelines on average .
Across 8 countries, adherence to clinical guidelines was lower in sick child care, where providers completed only 38% of the standard Integrated Management of Childhood Illness (IMCI) items, than in family planning (46%) and antenatal care (57%) . The median sick child consultation lasted only 8 minutes . Focusing specifically on Malawi, where the survey team conducted a limited re-examination of sick children, providers diagnosed pneumonia in only 1 in 5 children who showed symptoms of pneumonia per the IMCI guidelines .
Analysis of the 2013-2014 Malawi SPA survey with a simultaneous household survey suggested that poor quality care may contribute to avertable neonatal mortality, with a predicted prevalence of neonatal mortality of 28.3 deaths per 1,000 in lower quality facilities and 5.2 deaths per 1,000 in higher quality facilities, among women who would choose higher quality if it were more accessible to them .
As attention shifts from describing health system quality to improving it at scale, robust and ongoing measurement will be an essential tool for governments and researchers alike, particularly the direct observation of care delivery and perspective from patients themselves that makes the SPA such a unique and valuable resource.
Dr. Hannah Leslie is a Research Associate at the Harvard Chan School of Public Health; she served as the Measurement Research Lead for the Lancet Global Health Commission on High-Quality Health Systems in the SDG Era. She received her MPH and Ph.D. in Epidemiology from the University of California, Berkeley. Her research has made extensive use of the Service Provision Assessment surveys to 1) develop metrics of structure and process quality in LMICs, 2) describe current quality of care, and 3) assess predictors and effects of poor quality. Her recent work focuses on effective coverage calculations, patient experience measurement, and quality of care as a driver of HIV testing and treatment retention.