Skip to content

Digital health · 5 min read

Data quality before dashboards in surveillance programs

A surveillance dashboard shows a curve. Whether that curve describes disease or describes reporting behaviour depends entirely on what happened before the aggregation.

Two curves that look identical

Cases reported can rise because incidence rose, or because four facilities that were silent last week submitted. Without a reporting rate on the same screen, those two are indistinguishable, and both trigger the same alert.

Publishing completeness and timeliness alongside the count is not a caveat. It is the unit of the measurement.

If completeness is not on the screen next to the case count, the case count is not interpretable.

Checks belong upstream

Validation applied after aggregation cannot repair a form that was never standardized. Constraints belong at capture: controlled vocabularies, required fields, and a duplicate check at the facility rather than at the district.

That moves effort to the people with the least time, which is why the capture design has to be ruthless about how few fields are truly required.

Keep a human review step

A threshold crossing is a reason to look, not a conclusion. The district review step exists so that a data artefact does not become an outbreak declaration on its way up the hierarchy.

Software should make that step fast and visible — and should never offer to skip it.

Related solutionHealth Intelligence →