Evaluating a screening programme
- Measure
A screening programme is judged by how many people it reaches, what it finds and, above all, whether it cuts deaths from the disease.
What it is
Evaluating a screening programme means measuring whether it works in the population, not only whether the test is good. Operational measures (coverage, positivity rate, follow-up of positives, detection rate and positive predictive value) show whether the programme runs well. Outcome measures (disease-specific mortality, stage at diagnosis, harms) show whether it helps, and the strongest evidence is a randomised trial with disease-specific mortality as the end point.
Formula and reading
- Coverage = screened / target population × 100.
- Positivity rate = positive results / screened × 100.
- Follow-up = positives with a complete diagnostic work-up / positives × 100.
- Detection rate = confirmed cases / screened (per 1 000).
- Programme PPV = confirmed cases / positives worked up × 100.
- Interval cases: cases diagnosed between two rounds after a negative result; many suggest low sensitivity or too long an interval.
Worked example
- Data (made up): a cervical cancer screening programme; target population 20 000 women; 13 000 are screened.
- Coverage: 13 000 / 20 000 = 0.65, that is, 65 % (the made-up local target was 70 %).
- Positives: 650; positivity rate 650 / 13 000 = 0.05, that is, 5 %.
- Follow-up: 520 of the 650 complete the diagnostic work-up: 520 / 650 = 0.80 (80 %); 130 positives are left unexplored.
- High-grade lesions confirmed: 104; PPV = 104 / 520 = 0.20 (20 %).
- Detection rate: 104 / 13 000 = 0.008, that is, 8 per 1 000 screened.
- Reading: the programme misses its coverage target and loses 1 in 5 positives; improving follow-up pays more than changing the test.
Operational versus outcome measures
| Type | Measure | What it tells | Limitation |
|---|---|---|---|
| Operational | Coverage | Whether the programme reaches the population | Says nothing about whether the test works |
| Operational | Follow-up of positives | Whether positives are confirmed and treated | Needs linked records |
| Operational | Detection rate and PPV | Yield and burden of false positives | Change with prevalence and round |
| Outcome | Stage at diagnosis | Whether diagnosis comes earlier | Can improve without changing mortality |
| Outcome | Disease-specific mortality | Whether the programme saves lives | Takes years and ideally a randomised trial |
Markers
- Coverage
- Detection rate
- PPV
- Interval cases
- Disease-specific mortality
How to read it
- A programme is audited by rounds: coverage by age and territory, time from positive to diagnosis, and interval cases.
- Longer survival of detected cases does not prove benefit: mortality must be compared across the whole invited population.
- Harms are measured too: false positives, unnecessary procedures and overdiagnosis.
Common pitfalls
- Judging the programme by 5-year survival of detected cases (lead-time and length bias).
- Comparing screened and unscreened people as if they were alike: those who attend are usually healthier.
- Comparing the first round's detection rate with later rounds without noting that the first one picks up accumulated prevalence.
- Celebrating coverage without measuring follow-up of positives.
Test yourself
Think of the answer, then tap the question to see it.
How is the coverage of a screening programme calculated?
People screened divided by the target population; for example, 13 000 / 20 000 = 65 %.
Why is survival of detected cases not enough to evaluate a programme?
Because lead time, length bias and overdiagnosis lengthen apparent survival even if nobody lives longer; disease-specific mortality must be measured.
What is an interval case?
A case diagnosed between two screening rounds in a person whose result was negative; its frequency reflects the test's sensitivity and whether the interval is right.
Which design gives the strongest evidence that screening helps?
A randomised trial comparing disease-specific mortality between a population invited to screening and one not invited.
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: Cancer Screening Overview (PDQ®)–Health Professional Version (National Cancer Institute, 2023); Screening programmes: a short guide. Increase effectiveness, maximize benefits and minimize harm (World Health Organization, Regional Office for Europe, 2020).
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.