Health Data Management Basics

Across Africa the systems that collect routine health data are already in place: DHIS2 is the national HMIS standard in dozens of countries, OpenMRS records run in thousands of facilities, and Ethiopia's eCHIS digitizes community health workers. The frontier is no longer installation; it is use. This Level I course builds foundational health data management competency over four weeks, earning 6 CEU in the Health Informatics and Data category.

You will learn to capture data accurately at the point of care, distinguish data elements from indicators, and trace how one patient's visit becomes a number on a national dashboard. You will run completeness, timeliness, consistency, and accuracy checks on a register or DHIS2 dataset, read an HMIS dashboard and tell a real signal from a data artefact, and apply practices that close the data-use gap. It is built for clinicians, data officers, and students across disciplines.

Course glossary

• Accuracy — whether a reported value matches reality, confirmed by verification against the source document.
• ANC4 coverage — the share of pregnant women who completed at least four antenatal-care visits; a worked example of a coverage indicator.
• Completeness — the proportion of expected reports and required data fields actually recorded and received.
• Consistency — agreement between related data elements and across time (internal and trend consistency).
• Dashboard — a single screen of charts and figures presenting indicators for quick interpretation and decision-making.
• Data artefact — an apparent pattern caused by a data problem (backlog, double-counting, bad denominator, non-reporting) rather than a real change.
• Data capture — the act of recording a health event into a register, form, or digital system.
• Data dictionary (NHDD) — the authoritative definitions of all data elements and indicators; Ethiopia's National Health Data Dictionary.
• Data element — a single measured value (a raw count or attribute), such as doses given or weight.
• Data-review meeting — a regular session where a team interrogates its own data and ends with named decisions and actions.
• Data-use gap — the failure pattern in which systems collect data successfully but fail to change decisions.
• Data-use rituals — recurring practices (review meetings, league tables, feedback loops) that turn reporting into a decision tool.
• DHIS2 (District Health Information Software 2) — open-source HMIS platform, the de facto national standard across most of Africa.
• Disaggregation — breaking an indicator down by category (age, sex, location, time) to reveal where and when a gap exists.
• eCHIS — Ethiopia's electronic Community Health Information System, digitizing the workflows of health extension workers.
• eIDSR / SORMAS — electronic disease-surveillance tools into which notifiable cases flow for outbreak detection and response.
• Health information system (HIS) — the people, processes, and technologies that collect, manage, and report health data.
• Health management information system (HMIS) — the routine system aggregating facility-level service and disease data for management and planning.
• Indicator — a defined measure of performance or status, usually a ratio of data elements, used to support decisions.
• Information Revolution — Ethiopia's transformation agenda pairing digitization with a cultural goal of routine data use, verified through facility "model" status.
• Numerator / denominator — the top and bottom of an indicator ratio; the denominator (target population) turns a count into a coverage measure.
• OpenMRS — open-source electronic medical record system widely deployed across Africa, originally hardened in HIV care.
• Outlier — a value far outside the expected range, flagging a real event or, often, a data error.
• Parallel reporting — duplicate manual reporting of the same data into multiple vertical programmes; eliminated by integration.
• Register — a structured paper or digital record logging events of one type in sequence, one event per line.
• Source document — the original record (register, card, form) on which an aggregate number rests and against which it can be verified.
• Supportive supervision — coaching-style oversight that checks and improves data quality with the worker rather than policing it.
• Timeliness — whether data arrives by the reporting deadline, in time to act on.
• Validation rule — a system check that rejects or flags an impossible or out-of-range entry at the point of capture.

Frequently asked questions

Q: I only fill in a register — why do I need to understand indicators and dashboards? A: Because the line you write becomes the number a manager acts on. Understanding where your data goes — how a register entry becomes an indicator on a national dashboard — is what makes capture meaningful rather than clerical, and it is exactly the "use of data outputs" the framework expects of every cadre (Ch3 §3.3-D2). When you can see the decision your number feeds, you capture it more carefully and you can spot when something looks wrong.

Q: Our facility collects mountains of data but nothing seems to change. Is that normal, and what can I do? A: It is the single most common pattern in African health data — the data-use gap. It is not your personal failing, but you can act on it: start or join a monthly data-review meeting, send simple feedback back to those who report, and pick one decision the data could improve. The book's view is blunt and encouraging — using the system you already own is "the cheapest available win" (Ch5 §5.2.2).

Q: What is the difference between a data element and an indicator? A: A data element is a raw measured value — "240 women had four ANC visits." An indicator is a defined measure built from elements, usually a ratio — "ANC4 coverage = 68%." The element is a fact; the indicator interprets it against a denominator so it can guide a decision.

Q: My dashboard shows coverage over 100%. Did we really over-perform? A: Almost never. A figure above 100% is usually a denominator problem (an outdated or too-small population estimate) or double-counting in the numerator. Treat it as a data-quality flag, check the indicator's definition and the source counts, and fix the denominator rather than celebrating.

Q: How do I run the basic data-quality checks if I am not a statistician? A: You do not need statistics. Check four things in order: completeness (did all facilities and fields report?), timeliness (did it arrive on time?), consistency (do related numbers agree, and does this month resemble the last few?), and accuracy (does a sample match the register?). DHIS2 flags many outliers and impossible values automatically; the rest is careful looking and going back to the source document.

Q: What is the data dictionary / NHDD and why should I care? A: It is the official definition of every data element and indicator — exactly what counts as an "ANC1 visit," which age bands to use, how each indicator is computed. You should care because without shared definitions, two facilities reporting "deliveries" may count different things and the national total is meaningless. The dictionary is what lets a number mean the same thing everywhere.

Q: We re-enter the same data into three different programme forms. Is there a better way? A: That is parallel reporting, and ending it is a recognised priority — "integration of program data streams to end parallel reporting" (Ch5 §5.2.2). The better way is integration, so data is captured once and reused: a record system that feeds the HMIS, a surveillance tool fed directly from the case. While you await integration, flag the duplication to your supervisor as a quality and time problem, not just an annoyance — every duplicate form is an extra error source.

Q: How does accurate data capture connect to data privacy and ethics? A: Closely. Accurate, minimal, well-governed capture is the foundation both of useful data and of protecting patients — the same identifiers that make a record correct also make it sensitive. This course covers the quality and use of data; the companion course "Ethics, Privacy & Security in Digital Health" covers confidentiality, consent, lawful sharing, and the data-protection laws (Ethiopia's PDPP 2024, POPIA, and others) that govern how that data must be handled.