1.3 Clinical Epidemiology, Screening and Health Technology Assessment

Key Takeaways

  • Use the correct at-risk denominator and interval for incidence, and the observed population for prevalence.

  • Predictive values answer patient-level questions about test results and depend on disease prevalence at a given test performance.

  • Compare absolute and relative effects over the same period. In an Irish HTA, incremental cost per QALY and budget impact answer different decisions.

  • Compare screening completion within each eligible group, ask about barriers at the relevant stage, and check whether an accommodation reduces the gap.

Last updated: November 2026

Identify the population and the denominator

Prevalence is the proportion with a condition in the specified population at a given point or, with a defined period, the proportion who had it during that interval. Cumulative incidence is the proportion of an initially at-risk group that develops the condition over a specified interval. An incidence rate divides new events by accumulated time at risk and has a time unit. For example, if 40 of 1,000 clinic patients already have a diagnosis on 1 January, point prevalence is 4%. If 24 of the remaining 960 develop it during the next year with complete follow-up, the one-year cumulative incidence is 2.5%. Computing an incidence rate instead would require the actual person-time at risk.

In Irish healthcare, population screening invites eligible people without the symptoms under investigation. A symptomatic person needs assessment and the appropriate diagnostic pathway. This distinction applies to bowel symptoms even if the patient previously had a negative BowelScreen test. The National Screening Advisory Committee advises on eligible population programmes and changes. The HSE operates them. Appraising a programme means asking about evidence that the full pathway improves important outcomes, harms of testing and follow-up, availability of a confirmatory test, and the capacity to deliver and monitor care. A rise in measured survival from diagnosis can reflect earlier diagnosis without a later death. Overdiagnosis means detecting disease that would never have caused symptoms or death during that person's lifetime. These effects help explain why test accuracy and survival from diagnosis alone cannot establish a screening benefit.

CDC field epidemiology manual · NSAC role and screening appraisal criteria · HSE BowelScreen · NCI screening and lead time.

Investigate unequal participation

Use the eligible population as the denominator when comparing screening participation across groups. Compare the same time interval and stages of the pathway, from invitation to booking and completion. Two districts may differ in age, deprivation, housing stability, transport and language access. Compare like groups where possible and check that incomplete address, invitation and attendance records do not distort the result. An observed association alone cannot identify the cause. If the apparent difference could be explained by a third factor associated with both the comparison groups and the outcome, consider confounding. An anonymous local comparison can identify where to ask questions, while accounts from patients and community partners can help explain barriers the data cannot show.

For example, suppose 300 of 500 eligible people completed screening in one group and 450 of 500 in another over the same period. The participation proportions are 60% and 90%, a 30-percentage-point difference. Check whether people received invitations they could use and whether appointment times, transport, accessibility or communication were a barrier. If patients report difficulty with a digital-only booking route, test an accessible alternative designed with them. Recheck completion among eligible people in each group and listen to their experience. Support an informed decision about participation, including the option to decline.

HSE National Screening Service equity framework, priority areas 1, 4 and 5 · CDC Field Epidemiology Manual, analysis and interpretation.

Interpret a screening result in context

This fictional group of 10,000 people has 100 cases as established by an appropriate reference standard. The figures illustrate test interpretation and do not represent any Irish programme's performance.

Test resultCondition presentCondition absent
Positive90 true positives495 false positives
Negative10 false negatives9,405 true negatives
  • Sensitivity asks how often the test is positive among people with the condition. It is 90 / (90 + 10) = 90%.
  • Specificity asks how often the test is negative among people without the condition. It is 9,405 / (495 + 9,405) = 95%.
  • Positive predictive value asks how often a positive result identifies a person who has the condition. It is 90 / (90 + 495), about 15.4%.
  • Negative predictive value asks how often a negative result identifies a person who does not have the condition. It is 9,405 / (9,405 + 10), about 99.9%.

With fixed sensitivity and specificity, a lower prevalence reduces the proportion of positive results that truly identify the condition. This affects the number of people who face further investigations for each true case identified. Check whether the performance evidence applies to the target population and whether the same credible reference standard was used. A positive screening result usually needs follow-up specified by the programme. When symptoms develop, assess them on clinical grounds.

FDA diagnostic accuracy methods and reference standards.

Test Your Knowledge

In a screening accuracy study, 90 people with the condition and 495 without it have positive tests. Ten people with the condition and 9,405 without it have negative tests. Which estimate answers what proportion of people with a positive test have the condition?

A

90 / 100 = 90%

B

9,405 / 9,900 = 95%

C

10 / 100 = 10%

D

90 / 585, approximately 15.4%

Compare benefit, cost and affordability

Suppose in a fictional trial 100 of 1,000 participants receiving usual care experience a defined complication over two years, compared with 60 of 1,000 receiving a preventive intervention. Risks are 10% and 6%. The risk ratio is 6% / 10% = 0.60 and the relative risk reduction is 40%. The absolute risk reduction is 4 percentage points. The number needed to treat is 1 / 0.04 = 25 over two years to avert one additional complication on average. Discuss potential adverse effects and uncertainty as well as benefit. Applicability to a different population depends in part on baseline risk. A large relative benefit can translate into a small absolute one where baseline risk is low.

Irish health technology assessment (HTA) uses evidence on clinical benefit and harms and can include economic evaluation. A quality-adjusted life year (QALY) combines time and health-related quality of life. For example, two years at a constant utility of 0.75 yield 1.5 undiscounted QALYs. Cost-utility analysis compares additional cost with additional QALYs versus a relevant comparator. If two hypothetical strategies yield 5.10 and 5.00 QALYs per person, at respective costs of €5,000 and €2,000 per person, the incremental cost-effectiveness ratio (ICER) is €3,000 / 0.10 = €30,000 per QALY gained. A strategy that costs more and yields worse outcomes is dominated by the comparator. Evaluate uncertainty, including changes in plausible uptake, benefit or cost assumptions where relevant to the analysis.

Budget impact asks what adopting the intervention would do to spending under a defined funding perspective, eligible population, uptake and time horizon. The same per-person ICER can be associated with different budget impacts in populations of different sizes. Account for implementation, displaced care and resulting costs where relevant. An ICER estimate by itself gives no automatic Irish reimbursement decision.

CDC measures of association · NICE NNT glossary · HIQA Irish economic evaluation and budget impact methods.

Test Your Knowledge

In a trial the two-year risk of an adverse outcome is 12% on usual care and 8% with a preventive intervention. If the study estimate applies, how many people need to be treated for two years to avert one additional adverse outcome on average?

A

25

B

About 3

C

8

D

About 13

Test Your Knowledge

A new service adds €1,200 in expected cost and 0.06 QALYs per patient compared with usual care. Which interpretation is correct?

A

The new service adds €20,000 to the national budget regardless of adoption

B

The ICER is €20,000 per QALY gained, and projected uptake and spending must be examined to estimate budget impact

C

The service automatically qualifies for reimbursement in Ireland

D

An additional €1,200 per patient means the new service is dominated

Test Your Knowledge

A screening service has 300 completions among 500 eligible people in one district and 450 among 500 in another in the same period. People in the lower-participation district describe difficulty using the only booking option, a website. Which response is most appropriate?

A

Report equal coverage because the two eligible populations have the same size

B

Conclude the website caused exactly 150 people to miss screening

C

Check comparable denominators and pathway stages, work with patients on an accessible booking option and remeasure completion in each group

D

Report 750 overall completions and use the total alone to assess the gap

Sections you finish are checked off in the contents.