HMIS and DHIS2 reporting from an EMR: what to expect

Every month, Ethiopian health facilities report their activity to the national health management information system. Since Ethiopia moved its HMIS onto DHIS2, those figures — outpatient attendances, admissions, deliveries, antenatal contacts, immunizations, laboratory tests, notifiable diseases and many more — are entered into DHIS2, where they flow to woreda, zone, region and the Ministry.

In most facilities, producing those figures is a manual job. Someone, often the HMIS officer, works through registers and tally sheets, counts, cross-checks, and types the totals into DHIS2. It takes hours or days, it competes with the officer's other work, and errors in counting are hard to detect. An electronic medical record promises to do the counting automatically. It can — but what you get depends on how the data is captured and who owns the mapping between your records and the national indicators.

Capture once, at the point of care

The principle is simple: if every visit, delivery, immunization and test is recorded as it happens, then the monthly totals are a query, not a project. Nobody re-enters anything. The OPD attendance count is the number of outpatient visits recorded; the ANC first-visit count is the number of pregnancies with a first antenatal contact in the period; the immunization counts come from the doses recorded.

This only works if the data is captured in a structured way. A free-text note saying "ANC visit, G2P1, BP normal" cannot be counted reliably. A maternity record that records the contact number, gestational age and findings in fields can. The same goes for diagnoses: a diagnosis coded in ICD-10 can be counted toward a disease indicator; a scribbled diagnosis cannot.

Indicator definitions are where the work is

DHIS2 data elements have precise definitions, and those definitions are where most EMR reporting projects go wrong. A few examples of the questions that need an answer:

• Is an outpatient attendance counted per visit or per patient? Are new and repeat visits counted separately?
• Which age bands does the indicator use, and is age calculated at the date of the visit?
• Does "ANC first visit" mean the first contact of this pregnancy at this facility, or anywhere?
• Which ICD-10 codes roll up into each disease indicator?
• Is a malaria case counted on a positive RDT, a positive microscopy, a clinical diagnosis, or all three?

These are not technical questions, and a software vendor should not answer them alone. They belong to the facility's HMIS officer, guided by the national indicator reference and the woreda or regional health office. The software's job is to make the mapping explicit and to apply it consistently.

Direct submission, or a reviewed export?

It is tempting to aim for a system that pushes figures straight into DHIS2 at month end with no human involved. Technically, DHIS2 has a well-documented API that makes this possible. In practice, we think it is the wrong place to start.

For the first months, a facility should produce its report both ways: the usual manual count and the EMR export. Comparing them shows where the mapping is wrong, where staff are not recording something consistently, and where the paper count was itself wrong. Once the two agree — or the differences are understood — the export can be trusted. Only then does automating the final step make sense, and even then many facilities prefer a person to review the figures before they are submitted.

This is the approach CareLink takes today. The reports module produces an HMIS-style monthly facility summary — OPD attendances, admissions, discharges, deliveries, ANC first visits, immunizations, lab tests and malaria RDT results — as a CSV for any period up to a year. Each facility's HMIS officer maps the indicators to its DHIS2 data elements. Direct DHIS2 submission is not built in yet, and we say so plainly.

What changes for the HMIS officer

The HMIS officer's job changes from counting to checking. Instead of tallying registers, the officer reviews the export, investigates anomalies — a sudden drop in ANC visits might be a recording problem, not a real drop — and works with departments to fix recording habits. This is a better use of a skilled person, and it tends to improve data quality across the facility, because problems are found and corrected each month instead of being buried in a total.

Common pitfalls

• Counting from the wrong date: a visit registered late at night, or entered the next morning after downtime, can fall into the wrong reporting period unless the system uses the date of care.
• Duplicate patients: if registration creates a new record instead of finding the existing one, repeat visits are counted as new attendances. Good patient search at the front desk is a reporting control.
• Partial roll-out: if one department is still on paper, its activity is missing from the export. Report what is covered, and add the paper count for the rest until it goes live.
• Silent changes to indicator definitions: when the national definitions are revised, the mapping must be revised too. Someone at the facility must own that.

Beyond the monthly report

Once activity data is captured digitally, the monthly report is only the most obvious use. Management can see visits by day and service, lab and imaging volumes, admissions and length of stay, and bed occupancy whenever they want, not once a month. Disease trends can be spotted weeks earlier. And the same data supports program reporting — maternal health, immunization, malaria — without separate registers.

Practical advice

1. Involve your HMIS officer from the first day of selection, not after go-live.
2. Ask any vendor to show you the export for a real reporting period, and how each figure is calculated.
3. Plan at least two or three months of parallel reporting before relying on the export.
4. Fix recording problems at the source — in the department — rather than adjusting the numbers afterwards.
5. Keep the mapping documented and owned by the facility, so it survives staff changes.

If you would like to see the export on our demo hospital, ask us for a walkthrough. Related reading: antenatal care tracking at health centers, where structured capture matters most.