During the course of our trial (August 2012-August 2013) we received word that Zambia would be undertaking its 6th Standard Demographic Health Survey (DHS). The Demographic Health Surveys are nationally representative household surveys that provide data for a wide range of monitoring and impact evaluation indicators in the areas of population, health, and nutrition. They typically involve large sample sizes (usually between 5,000 and 30,000 households) and are usually conducted about every 5 years, to allow comparisons over time. Since 1984, more than 130 nationally representative household-based surveys have been completed under the DHS programme in about 70 countries.
In Zambia, the DHS is implemented by the Central Statistics Office (CSO), with whom we worked closely during the planning and design phase of our work. For the 2013/14 DHS, the CSO worked in partnership with the Ministry of Health as well as the University Teaching Hospital (UTH) – Virology Laboratory, the Tropical Disease Research Center (TDRC) and the Department of Population Studies at the University of Zambia (UNZA). The fieldwork took place from August 2013 – April 2014. With a sample size of 15,920 households, it involved interviews with 16,411 women and 14,773 men aged 15-49.
The survey results for each participating country are typically published in a brief preliminary report, a more detailed final report and a summary type report on key findings. The latter two reports are widely distributed and constitute the primary output of the project. We recently had the opportunity to see the preliminary report and thought it would be interesting to compare some of the key diarrhoea related indicators to our findings in Eastern and Southern province.
It is important to note some of the potential reasons for the differences. First and foremost, the DHS figures cited are national averages, while the cited COTZ findings are specific to our project districts – which were selected based on being rural and underserved, and therefore more likely to face a greater diarrhoea burden. Here we present the individual COTZ district-level figures at endline (for both the intervention and comparator districts) and compare them to the mean figure presented in the preliminary 2014 DHS presentation.
The second difference worth noting is that the data for the DHS was collected over the course of 9 months, and again the figure presented is an average. This would inevitably drown out seasonal variations in diarrhoea incidence over the course of the year. Our endline data was collected in August (2013), a month that typically has higher diarrhea rates on average. For example, in the graph below (based on data from our Health Centre Impact Assessment from Monze) we can see the average diarrhoea caseloads seen at Monze health centres per month from 2009-2012. We’ve included an overall average trend line as well (aqua blue), which shows August as the month with the third highest average caseload. This may also account for some of the difference between COTZ and DHS figures. While the DHS is a vital tool for public health professionals (planning, baselines, benchmarks, policy, etc.) and provides us with crucial, nationally representative, open source data, they are not ideal for measuring the effects of programs/interventions that have not been in place for a long time, and are typically not seen as replacements for well-conducted specialized surveys.
For those who are interested, here is the full slide deck of the preliminary DHS results. These were presented at a press conference in Lusaka in Sep-14:
A note on the data in this post
The data contained in this blog post are unpublished and based on preliminary analysis of data from the ColaLife Operational Trial in Zambia (COTZ). Final calculations may vary and will be published in peer reviewed literature in due course. In the interim, the following citation may be used: Ramchandani, R. et al. (forthcoming). ColaLife Operational Trial Zambia (COTZ) Evaluation. Johns Hopkins Bloomberg School of Public Health, Baltimore. Related correspondence should be sent to Rohit Ramchandani (email@example.com) and copied to Simon Berry (firstname.lastname@example.org).