Sources of health data
- Concept
Health indicators come from certificates, registries, surveys and medical records; each source has its own errors.
What it is
Public health rates and indicators are built from data that someone recorded for another purpose: vital statistics, mandatory notification, disease registries, population surveys, and clinical or administrative records. Knowing how each datum is produced lets one judge its coverage, quality and biases. An indicator is never better than the source it comes from.
Worked example
- Data (made up): a department of 800 000 people registers 4 200 deaths in a year; a study estimates that 4 800 occurred.
- Registration coverage: 4 200 / 4 800 = 0.875, that is, 87.5 %; 600 deaths (12.5 %) are missing.
- Registered crude rate: 4 200 / 800 000 = 5.25 per 1 000; estimated true rate: 4 800 / 800 000 = 6.0 per 1 000.
- Quality: 378 certificates have an ill-defined cause; 378 / 4 200 = 0.09, that is, 9 %.
- Usable causes: 4 200 − 378 = 3 822 deaths with a well-defined underlying cause for analysis by cause.
- Reading: the register understates mortality by 0.75 per 1 000 and one death in 11 gives no usable cause; both must be corrected before comparing.
Data sources and their common errors
| Source | What it provides | Typical error |
|---|---|---|
| Vital statistics | Births, mortality by cause | Rural under-registration, poorly certified underlying cause |
| Mandatory notification | Cases of events of public health interest | Under-reporting, changes in case definition |
| Disease registries (cancer, birth defects) | Incidence and survival | Partial geographic coverage |
| Population surveys | Prevalence, risk factors, access | Recall and social desirability bias, non-response |
| Medical records and administrative data | Service use, diagnoses, costs | Coding for payment, only people who seek care |
Markers
- ICD
- Underlying cause
- Vital statistics
- Under-registration
- Survey
- Registry
In practice
- Filling in the death certificate well is a clinical act with public health effects: the underlying cause feeds the country's indicators.
- On the certificate the chain runs from the direct cause (top) to the underlying cause (bottom); ‘cardiorespiratory arrest’ is not an underlying cause.
- Before using a database, ask who fills it in, for what, with which definition and which part of the population is left out.
Common pitfalls
- Comparing mortality between two regions with different registration coverage as if both were complete.
- Using billing diagnoses as validated clinical diagnoses.
- Mixing years coded with different ICD versions without a bridging table.
- Reading the absence of notified cases as absence of disease.
Test yourself
Think of the answer, then tap the question to see it.
What is the underlying cause of death?
The disease or injury that started the chain of events leading directly to death; it is the cause used for mortality statistics by cause.
What does under-registration of deaths do to the mortality rate?
It understates it: the numerator is incomplete and the region looks healthier than it is.
What is one strength and one limitation of administrative health service data?
They are large, continuous and cheap, but were created for care or payment: codes can be imprecise and they include only people who seek care.
Which classification is used to code causes of death, and when did version 11 come into effect?
WHO's International Classification of Diseases (ICD); ICD-11 was adopted in 2019 and came into effect on 1 January 2022.
Connections
Educational content on epidemiological and statistical methods; the examples use made-up numbers to teach the calculation. Not a clinical or legal guide: current Colombian rules are in the official sources.
Sources: International Statistical Classification of Diseases and Related Health Problems (ICD) (World Health Organization, 2026); DANE - Estadísticas vitales nacimientos y defunciones (Departamento Administrativo Nacional de Estadística (DANE), Colombia, 2026); Principles of Epidemiology: Lesson 5, Section 4 (Centers for Disease Control and Prevention, 2012).
Spotted a mistake?We are always improving for the community. Tell us what is wrong and our team will review it.Report a mistake
Representative model, for educational purposes only.