Prevalence

Prevalence in Health Economic Evaluation and Planning

Prevalence measures the existing caseload of a condition, counting everyone in a population who has it at the stated time, however long ago each case began. Health economists use it to set the number of patients a new treatment would reach when it is introduced and the annual cost of a disease. The stock of cases is filled by new cases and emptied by recovery and death, so prevalence moves with survival as well as with incidence, and a treatment that keeps people alive without curing them raises prevalence and the spending that follows it. Prevalence also shapes the chance that a positive test result is correct, and regulators use it to decide which medicines count as orphan medicines.

Counting the existing caseload as a proportion

The CDC teaching text Principles of Epidemiology in Public Health Practice defines prevalence as the proportion of persons in a population who have a particular disease or attribute at a specified point in time or over a specified period. Its numerator includes all cases, new and pre-existing, whereas incidence is limited to new cases.

$$P = \frac{C}{N}$$

Here $P$ is the prevalence, $C$ is the number of people with the condition at the stated time or during the stated period, and $N$ is the population at that time or over that period.

Prevalence is a proportion between 0 and 1 with no time unit. The CDC text acknowledges that it is sometimes called the prevalence rate, but it is not a rate per unit of person-time.

The time frame defines the variant. Point prevalence counts people with the condition on one date, and period prevalence counts anyone who had it at some time during an interval such as a year. Spronk and colleagues describe a further form used with general practice records, contact prevalence, which counts people with at least one encounter with a health care professional for the disease during a year. Comparing prevalence between two groups gives the prevalence ratio, although Pearce observes that when the concern is the public health burden of disease, the absolute prevalence and the prevalence difference are often of greatest interest.

Incidence, duration and the size of the stock

Prevalence is based on both incidence and the duration of illness, a link the incidence page develops for prevention. Pearce gives the formal relation for a population in a steady state.

$$\frac{P}{1-P} = I \times D$$

Here $P$ is the prevalence, $P/(1-P)$ is the prevalence odds, $I$ is the incidence rate among people without the condition and $D$ is the average duration of the condition. Pearce states that the relation usually requires incidence and disease status to be unrelated to migration and population size, and average duration to stay constant over time. Such a population will rarely exist, but it is approximated where incidence and the relevant exposures are not changing markedly. When prevalence is small, the term $(1-P)$ is close to 1 and prevalence is approximately $I \times D$.

Pearce also notes that prevalence will often depend on age even when incidence does not, so comparisons between populations need age to be controlled for.

Choices that change a prevalence estimate

Each data source counts a slightly different group of people. Spronk and colleagues computed three prevalence proportions from the same Dutch general practice records for 2012. For long-lasting diseases, one-year period prevalence was between 58.3% and 206.6% higher than point prevalence on 31 December, and between 26.2% and 79.7% higher than contact prevalence. The choice of prevalence definition moved the proportions far more than the choice of denominator moved incidence rates.

Contact prevalence is close to what an analyst sees in activity or claims data, and Spronk and colleagues warn that it depends on the condition and on the amount of care a patient needs. People with a chronic condition may not visit their general practitioner in a particular year, or may be seen only in secondary care, so a count of service users can sit well below the number of people living with the condition. Records also miss undiagnosed disease, while a survey that tests people misclassifies some of them, a problem the point prevalence page treats.

A survey estimate also carries sampling error. When a prevalence from $r$ cases among $n$ people sampled enters a probabilistic sensitivity analysis, it is usually given a beta distribution with parameters $r$ and $n-r$, which the ISPOR-SMDM task force on parameter estimation calls a natural match for binomial data.

Prevalence in budgets, burden estimates and test decisions

Prevalence enters health economic work wherever the question concerns people already living with a condition. Each use needs the estimate to match the population and time frame of the decision.

Sizing the eligible population for budget impact

The ISPOR good practice report on budget impact analysis by Sullivan and colleagues lists the size of the eligible population among the key elements of the analysis and asks for data specific to the decision maker's own population. Prevalence gives the stock of people eligible when a treatment is introduced, and incidence gives those who become eligible later. Where a simple cost calculator cannot capture changes in the eligible population, the report allows a cohort or patient-level model. The worked example below follows a treatment that enlarges its own eligible population.

Prevalence-based cost of illness

Jo describes the prevalence-based design, which estimates the burden of a condition over a period, usually a year, as the most commonly used approach in a cost-of-illness study. Jo adds that for health problems declining in magnitude, prevalence-based estimates will generally be lower than incidence-based ones. The years lived with disability page sets out how burden of disease studies combine prevalence with disability weights.

Prevalence as the starting probability for a test

In a diagnostic or screening decision, the prevalence in the group tested is the pre-test probability. The positive predictive value page shows how the share of positive results that are true falls with prevalence for a test of fixed sensitivity and specificity. For the economics of case finding, the testing cost of each true case found rises in inverse proportion to prevalence.

$$C_{case} = \frac{c_t}{P \times Se}$$

Here $C_{case}$ is the testing cost per true case detected, $c_t$ is the cost of testing one person, $P$ is the prevalence in the group tested and $Se$ is the sensitivity of the test. With an illustrative test costing GBP 20 and a sensitivity of 0.9, a prevalence of 0.01 gives 20 / (0.01 × 0.9) = GBP 2,222 per case found, and a prevalence of 0.001 gives 20 / (0.001 × 0.9) = GBP 22,222.

Leeflang and colleagues found that test accuracy can shift with prevalence, probably through patient spectrum, so accuracy estimates from a high-prevalence setting may not transfer to a low-prevalence screening population.

Rare disease rules written in terms of prevalence

Orphan medicine rules turn on prevalence, and the EU and US rules define it differently. The European Medicines Agency states that, to qualify for orphan designation, the prevalence of the condition in the EU must not be more than 5 in 10,000, unless marketing the medicine would be unlikely to generate sufficient returns to justify the investment. Article 3 of Regulation (EC) No 141/2000 applies the ceiling to the persons affected when the application is made. In the United States, the FDA rule at 21 CFR 316.21 asks for documentation that fewer than 200,000 persons are affected, and for that purpose defines prevalence as the number of persons in the United States diagnosed with the condition at the time the designation request is submitted. The EU test is a proportion of the population, while the US test is a count of diagnosed people that leaves undiagnosed cases out.

Worked example of a life-extending treatment in a rare condition

All figures in this example are illustrative. A health system covers 2,000,000 people. A chronic condition has an incidence rate of 0.0001 per person-year among people without it and an average duration of 4 years under current care, and the population is close to a steady state.

1. Prevalence under current care. The prevalence odds is 0.0001 × 4 = 0.0004, so the prevalence is 0.0004 / 1.0004 = 0.00039984, or about 4.0 per 10,000. The number of people living with the condition is 0.00039984 × 2,000,000 = 799.7, about 800.

2. Prevalence once a new treatment is established. The new treatment extends average duration to 6 years without curing anyone or changing incidence. The prevalence odds becomes 0.0001 × 6 = 0.0006, the prevalence is 0.0006 / 1.0006 = 0.00059964, or about 6.0 per 10,000, and the caseload is 0.00059964 × 2,000,000 = 1,199.3, about 1,199.

3. The orphan threshold. Before the treatment the condition sits below the EU ceiling of 5 in 10,000. A prevalence estimate made at the new steady state would put it above the ceiling, although the EU rule counts the persons affected when the application is made.

4. Annual spending. Current care costs GBP 6,000 per patient a year and the new treatment GBP 24,000, and every patient is assumed to be treated. Spending before is 800 × 6,000 = GBP 4.8 million. At the new steady state it is 1,199 × 24,000 = GBP 28.776 million, an increase of GBP 23.976 million.

5. Splitting the increase. Moving the original 800 patients to the new treatment adds 800 × 18,000 = GBP 14.4 million. The extra 399 patients who are alive because of the treatment add 399 × 24,000 = GBP 9.576 million. The two parts sum to 14.4 + 9.576 = GBP 23.976 million, matching step 4.

6. Interpretation. A budget impact estimate that held the caseload at its current 800 would report an increase of GBP 14.4 million a year and miss GBP 9.576 million that arrives as prevalence builds over the following years. The rise in prevalence here signals that the treatment works, and a cost-effectiveness analysis would weigh the extra years of life against the extra spending.

Prevalence beside neighbouring measures

Prevalence is a stock measure, and most confusion comes from substituting it for a flow measure or for one of its own variants. Point and period prevalence differ only in the time frame of the count, and the prevalence ratio compares two groups in one survey.

The flow counterpart of prevalence is cumulative incidence when a model needs the risk of onset over a period, and the incidence rate when it needs a rate per person-time.

Common misreadings

Most errors with prevalence come from reading it as something it does not measure. Each one alters a patient count or cost that a model carries forward.

1. Reading falling prevalence as success. Prevalence also falls when patients die sooner, and rises when treatment keeps them alive.
2. Sizing a budget impact model from prevalence alone. This misses the stream of newly eligible patients and, for a life-extending treatment, the growth of the pool.
3. Mixing definitions. Period, point and contact prevalence from the same records can differ by more than a factor of two for long-lasting diseases.
4. Treating service counts as total prevalence. Activity and claims data count people in contact with care, not everyone with the condition.
5. Carrying test accuracy across settings. Sensitivity and specificity can shift with the spectrum of patients that also changes prevalence.

Sources

• Briggs AH, Weinstein MC, Fenwick EAL, Karnon J, Sculpher MJ, Paltiel AD. Model parameter estimation and uncertainty analysis: a report of the ISPOR-SMDM Modeling Good Research Practices Task Force Working Group-6. Medical Decision Making. 2012;32(5):722-732. https://doi.org/10.1177/0272989X12458348
• Centers for Disease Control and Prevention. Principles of Epidemiology in Public Health Practice, 3rd ed. Lesson 3: Measures of Risk, Section 2 (Morbidity frequency measures). Atlanta: CDC; 2006, page last reviewed 18 May 2012. https://archive.cdc.gov/www_cdc_gov/csels/dsepd/ss1978/lesson3/section2.html
• Electronic Code of Federal Regulations. Title 21, section 316.21, Verification of orphan-drug status. Text as at 1 October 2026. https://www.ecfr.gov/current/title-21/section-316.21
• European Medicines Agency. Orphan designation overview. Page last updated 6 March 2026. https://www.ema.europa.eu/en/human-regulatory-overview/orphan-designation-overview
• Jo C. Cost-of-illness studies: concepts, scopes, and methods. Clinical and Molecular Hepatology. 2014;20(4):327-337. https://doi.org/10.3350/cmh.2014.20.4.327
• Leeflang MMG, Rutjes AWS, Reitsma JB, Hooft L, Bossuyt PMM. Variation of a test's sensitivity and specificity with disease prevalence. CMAJ. 2013;185(11):E537-E544. https://doi.org/10.1503/cmaj.121286
• Pearce N. Effect measures in prevalence studies. Environmental Health Perspectives. 2004;112(10):1047-1050. https://doi.org/10.1289/ehp.6927
• Regulation (EC) No 141/2000 of the European Parliament and of the Council of 16 December 1999 on orphan medicinal products, Article 3, legislation.gov.uk copy, text as at 31 December 2020. https://www.legislation.gov.uk/eur/2000/141/article/3
• Spronk I, Korevaar JC, Poos R, et al. Calculating incidence rates and prevalence proportions: not as simple as it seems. BMC Public Health. 2019;19:512. https://doi.org/10.1186/s12889-019-6820-3
• Sullivan SD, Mauskopf JA, Augustovski F, et al. Budget impact analysis: principles of good practice. Report of the ISPOR 2012 Budget Impact Analysis Good Practice II Task Force. Value in Health. 2014;17(1):5-14. https://doi.org/10.1016/j.jval.2013.08.2291