Going beyond the aggregated: why we have better knowledge on immunisation dropouts with individual patient data
Aggregated data means taking data from several different sources and measurements and grouping them together, and is commonly used within the public health sector. Many times this type of data can be very useful for health care practitioners and planners because it allows them to find common characteristics and identify patterns in public health. It could be useful for example if you’re trying to understand which diseases are prevalent and pose a threat to public health in a specific region, or if you would like to identify common health risks for women.
There is no arguing that aggregate data is useful, but as with most things, it has its limitations.
Although grouping numbers together has its usefulness when looking for trends, it makes it very tricky for planners who work on a level that is lower than the national level at which data is usually aggregated. When you lump big groups of people together and focus on big trends and generalisations, it can be easy to forget that there are huge variations even in small populations. What is true for town A is not nessecarily true for town B even though they belong to the same region.
The UN has said that “to fully implement and monitor progress on the SDGs, decision makers need data and statistics that are accurate, timely, sufficiently disaggregated, relevant, accessible and easy to use”. However, much of the available data is based on national-level data sources, such as censuses, Multiple Indicator Cluster Surveys (MICS) or Demographic and Health Surveys (DHS) that are costly and infrequently done. They may satisfy requirements for donors and national level stakeholders but these provide no detailed local-level data. There is a growing awareness about gaps in data from these sources, such as the lack of individually-based and up-to-date data, for instance, 25 countries or areas did not conduct a population or housing census in the years between 2007–2016. Of these countries and areas, nine were in sub-Saharan Africa.
“In the 1990s, immunisation programmes were one of the first health programmes to focus on subnational data and coverage estimates based on routine reports as part of the Reaching Every District strategy.” — Ties Boerma, WHO, 2016
One of the sources of inequity identified in SDG17 is geographic location, or place of residence. Urban and rural distinctions have typically been used to determine immunisation coverage, and it has generally been assumed that people living in urban areas have better access to services and better health outcomes. A broad urban-rural categorisation may conceal further inequalities and mask what is happening on an individual level, for instance, well-off urban areas and urban slums would both be included in the ‘urban’ category. Without disaggregation into smaller geographical units, those distinctions would remain hidden.
Data availability has improved over the years, but there remain gaps and the data collected at the local level can be either of poor quality, insufficient or unreliable. So, imagine you are a health planner located at a district health office. You have been given the job of ensuring that every child in your community has been vaccinated against Rubella. You get a report that says 8000 out of 10 000 children in a community have been vaccinated. Great! 80% of children in your community have been vaccinated, but 2000 children are still left unprotected.
Now, if the only information you have is the number of children who have and haven’t been vaccinated, it would be quite hard for you to find the 2000 children that aren’t protected against Rubella. Having data disaggregated to an individual level and tied to individual identifiers would make your job a whole lot easier. Disaggregating data into subpopulations can help communities plan appropriate programmes, decide which evidence-based interventions to select, assign limited resources where they are needed most, and see where progress is being made, or not.
This is one of the struggles faced by health planners that we have attempted to solve with MyChild Solution.
Making the hidden visible
Because all information collected with MyChild Solution is connected to children’s individual IDs, it makes it easy to get specific knowledge on the health of individuals. Instead of just getting a report that says “8000 out of 10 000 children got the vaccines they should have” you would also get a list with names and contact information for the 2000 children that did not get their vaccinations according to schedule. Why is this important? Having this individual level data helps health care planners and local decision makers identify vulnerable populations and makes it easier to make sure all children are getting the care they need.
In currently used methods, dropout rates in child vaccination programmes are calculated by taking the number of children getting the first in a vaccine series (for example DPT1) and subtracting the number of children getting the last vaccine of the same series (DPT3 in this case). Now, this sounds like a reasonable method to calculate dropout rates, you take the number of children you started off with and compare it with how many children were there at the end.
But what if these two groups of children aren’t the same? Why this method does not give an accurate picture of coverage is because there is no way to guarantee that the children who are counted as receiving DPT3 are the same ones who received DPT1. In an attempt to rectify this, we decided to look at dropout rates per individual child. We look at the children who got the first vaccine, and then look at the data for the same children to see if they got the last one as well.
Information like this will enable decision makers and health workers in the region to plan their work based on reliable and up to date information and can provide a better insight of vaccination coverage and help in the planning of relevant interventions for immunisation programmes. Indeed, in a review of the Gambian EPI Programme, it was noted that there was limited capacity of staff at facility level to calculate dropout rates. One of the strategies to reducing dropout rates was to improve dropout monitoring at all health facilities. We spoke to some of the health workers in Serrekunda Health Centre, one of the implementation sites of MyChild Solution in The Gambia who say that where once they spent a great deal of time at the end of the month having to aggregate data, now this is automatically done by MyChild Solution.
To get accurate measurement of the progress of countries towards the SDGs, it is clear that acquiring accurate and timely patient-level disaggregated data must be considered a priority. Being able to easily access data disaggregated down to an individual level will put countries on a better track to meeting the SDGs, such as SDG17 which asks for an increase in the availability of disaggregated data. But more importantly, it will help local health planners provide better care for the children in their communities.