Introduction to Multi-Source Health Data Analysis

Imagine you are working with a public health team in Nigeria. The local health authority wants to know how well a maternal and child health programme is performing. Facilities report that 800 children received a particular service. An NGO reports that its outreach teams served 300 children. A community survey estimates that 950 children received the service. You now have three different numbers.

Which one is correct? Should you add 800 and 300 to get 1,100 children? Should you accept the survey estimate of 950? Or should you investigate further?

A good health data analyst does not automatically choose one figure or add all the numbers together. The analyst first investigates what each source measures and how the information was collected. That is the foundation of multi-source health data analysis and triangulation.

In Module 3, you learned about digital data collection, routine health reporting, KoboToolbox and DHIS2. Now we learn how information from different sources can be compared responsibly. These skills are relevant to Health Information Management, public health, Monitoring and Evaluation, NGO programmes, research and health system planning.

4.1.1 What is a data source?

A data source is where information originates. In healthcare, data sources can include patient registers, hospital information systems, routine facility reports, laboratory records, community surveys, NGO programme reports, population surveys and health information platforms.

Different sources may describe the same health problem from different perspectives. A facility register may record services provided at a clinic, while a community survey may collect information about people's experiences or use of services. Both sources may be useful, but they do not necessarily measure exactly the same thing.

4.1.2 What is multi-source health data analysis?

Multi-source health data analysis involves examining information from two or more sources to develop a more complete understanding of a health question or problem.

Suppose a district health manager wants to understand childhood immunisation coverage. The manager may consider facility immunisation records, outreach programme records, population estimates and community survey findings.

Facility records may provide detailed information about recorded services. Population estimates may help establish the number of people eligible for a service. Surveys may provide information about vaccination reported by sampled households, depending on the survey design and questionnaire. Comparing these sources can help the manager understand what the data suggests and what needs further investigation.

4.1.3 Why use multiple data sources?

1. To obtain a broader picture. One dataset may not answer every question.
2. To identify inconsistencies. Two reports may provide different figures for the same period.
3. To investigate data quality. Differences may reveal missing records, duplicate entries or inconsistent definitions.
4. To understand context. A survey may identify barriers that are not visible in routine facility totals.
5. To support better decisions. Decision-makers can use the combined evidence to identify questions, investigate problems and determine appropriate next steps.

However, more datasets do not automatically produce better evidence. The sources must be suitable for the question and interpreted carefully.

4.1.4 A simple Nigerian health-sector example

Consider a fictional local government area in Nigeria. The health authority wants to understand why reported immunisation coverage appears lower than expected. The analyst has three sources.

Before comparing the figures, the analyst must establish whether the sources cover the same geographical area, refer to the same reporting period, measure the same vaccine or service, refer to the same age group, count doses, visits or unique children, and include overlapping service activity. Without this information, a numerical comparison could be misleading.