Epidemiological Foundations and Disease Frequency

Introduction

Epidemiology is the study of the distribution and determinants of health-related states or events in specified populations, and the application of this knowledge to control health problems. For health-care managers, epidemiology provides the language and measurement tools needed to assess need, allocate resources, evaluate services, identify inequities and anticipate future demand.

Clinical practice generally focuses on the individual patient. Epidemiology asks broader questions: How common is a condition? Who is affected? Where and when does it occur? What exposures or social conditions influence it? How effectively are prevention and treatment strategies working? These questions require precise definitions of populations, events, time periods and denominators.

This chapter develops the foundations for measuring disease frequency and population health. It distinguishes incidence from prevalence, explains mortality, survival and case-fatality, introduces person-time and standardization, and examines summary measures such as disability-adjusted life years (DALYs) and quality-adjusted life years (QALYs). It also applies the person–place–time framework to population interpretation.

Learning objectives

By the end of this chapter, the reader should be able to:

1. Define epidemiological terminology and identify appropriate populations at risk.
2. Calculate and interpret incidence, prevalence, mortality, survival and case-fatality measures.
3. Explain the role of person-time and distinguish rates from proportions and ratios.
4. Describe direct and indirect standardization and recognize confounding by age or other variables.
5. Explain DALYs and QALYs and their uses and limitations.
6. Interpret disease patterns by person, place and time.
7. Apply epidemiological measures to health-care planning and managerial decisions.

1. The epidemiological perspective

Population, health event and denominator

An epidemiological measure combines a numerator, a denominator and a time or observation frame. The numerator records events, such as new diagnoses or deaths. The denominator represents the population from which those events could arise. A denominator must therefore reflect the population at risk, not merely the population that is convenient to count.

For example, cervical cancer incidence should be calculated among people with a cervix who are free of the disease at the beginning of observation, rather than among the entire population. A measure of pregnancy complications should use pregnancies or pregnant individuals as the relevant population at risk. Individuals who cannot experience the event should generally be excluded from the denominator.

Important terms include:

• Population: the defined group to which an epidemiological statement applies.
• Source population: the population giving rise to the cases or participants observed.
• Target population: the wider population to which findings are intended to apply.
• Case definition: explicit clinical, laboratory, imaging or administrative criteria for classifying an individual as a case.
• Numerator: the number of events or persons with the characteristic of interest.
• Denominator: the population or amount of person-time in which events could occur.
• Reference period: the time interval over which events or population membership are measured.

A case definition should be sufficiently sensitive to identify most true cases when surveillance is the priority, or sufficiently specific to reduce false positives when confirmation is essential. Changing a case definition can create an apparent change in disease frequency even when the underlying disease pattern is unchanged.

Ratio, proportion and rate

A ratio compares two quantities; the numerator need not be part of the denominator. A proportion is a ratio in which the numerator is included in the denominator and is therefore bounded between 0 and 1, or 0% and 100%. A rate describes the occurrence of events over time and commonly uses person-time in the denominator.

In routine public health language, the word “rate” is sometimes used for proportions such as prevalence. Conceptually, however, a true rate includes a time dimension or an equivalent observation period. Clear reporting should state the numerator, denominator, time period, multiplier and population characteristics.

2. Incidence: measuring new disease occurrence

Incidence measures the occurrence of new cases among people initially free of the condition. It is especially useful for studying causation, evaluating prevention and estimating the risk of developing disease.

Cumulative incidence

Cumulative incidence, also called incidence proportion or risk, is the proportion of an initially disease-free population that develops the condition during a specified period.

The result is dimensionless and may be expressed as a proportion or percentage. The calculation assumes that the initial population is adequately defined, cases are counted consistently, loss to follow-up is limited or appropriately handled, and individuals are not counted more than once.

Worked example. A cohort of 2,000 employees is free of influenza-like illness on 1 January. During the following month, 160 develop the illness. The cumulative incidence is:

Thus, 8% of the cohort developed the condition during the month. If individuals leave the cohort, the interpretation becomes less direct unless their follow-up is incorporated into the analysis.

Incidence rate or incidence density

When individuals are observed for different lengths of time, incidence rate is more appropriate. It uses the total disease-free person-time contributed by the population.

Typical units are cases per person-year, cases per 1,000 person-months or cases per 100,000 person-years. The measure is not a probability and is not normally bounded by 1.

Worked example. A surveillance programme detects 45 new cases over 18,000 person-years of observation:

Expressed per 1,000 person-years, this is 2.5 cases per 1,000 person-years.

The incidence rate assumes that the event process is reasonably represented by the accumulated person-time and that the population is sufficiently stable, or that changes in risk are addressed analytically. It is particularly useful when participants enter and leave at different times.

Incidence and hazard

The incidence rate is an average event rate over an observation period. The hazard is the instantaneous event rate among those still at risk at a particular time. Survival analysis estimates hazards when censoring and varying follow-up are important. Incidence rate and hazard are related but should not be treated as identical.

3. Prevalence: measuring existing disease

Prevalence measures existing cases, whether recently diagnosed or longstanding, in a population. It is useful for planning services, estimating treatment needs and describing the burden of chronic conditions.

Point prevalence

For example, if 600 of 30,000 adults are receiving treatment for diabetes on 1 July, point prevalence is 600 ÷ 30,000 = 0.02, or 2%.

Period prevalence

The denominator and case ascertainment method must be stated because period prevalence may include cases present at the beginning and new cases arising during the period.

Relationship between prevalence, incidence and duration

In a relatively stable population, with a rare condition and approximately constant incidence, prevalence is often approximated by:

This relationship is a conceptual approximation, not a universal identity. It is most defensible when migration, mortality, recovery and changes in incidence are limited. Prevalence can rise because incidence increases, duration becomes longer through improved survival, or detection improves. Conversely, prevalence can fall because incidence declines, patients recover, mortality increases or cases leave the population.

Incidence is therefore generally the more informative measure for studying disease causation, whereas prevalence is often more useful for estimating current service demand.

4. Mortality, survival and case-fatality

Mortality rate

Mortality measures deaths in a population during a specified period. A crude mortality rate is:

Here, n is commonly 3, 4 or 5, depending on the desired scale. The denominator is often the mid-year population because it approximates the population at risk during the year. Units should be reported, such as deaths per 100,000 population per year.

Cause-specific mortality uses deaths from a specified cause in the numerator. Age-specific mortality uses a particular age group in both numerator and denominator. Infant mortality is a specialized measure involving deaths under one year of age and live births, rather than the general population denominator.

Case-fatality proportion

Case-fatality describes the severity or lethality of a disease among diagnosed cases.

If 20 of 400 diagnosed patients die, case-fatality is 5%. The measure is a proportion, although often called a case-fatality rate. Interpretation depends on complete case ascertainment, the follow-up period, disease severity, treatment and the case definition. Under-diagnosis of mild cases can artificially increase case-fatality.

Survival

Survival is the probability of remaining alive, or remaining free of another specified event, over a defined period after diagnosis or treatment. One-year survival is commonly calculated as:

For complete follow-up without censoring, this simple expression may be adequate. With incomplete follow-up, Kaplan–Meier or other survival methods are required. Relative survival compares observed survival among patients with expected survival in a comparable general population and is often used in cancer epidemiology.

Survival and case-fatality are related but not interchangeable. Survival focuses on time to death or another event; case-fatality summarizes deaths among cases over a defined period.

5. Person-time and dynamic populations

Person-time adds the periods during which individuals are observed and at risk. One person observed for five years contributes five person-years; five persons observed for one year also contribute five person-years. Person-time is particularly useful when entry, exit, loss to follow-up or study duration varies.

A person should contribute time only while eligible and at risk. After developing the outcome, that person normally no longer contributes disease-free time for a first-event incidence analysis. Person-time calculations should account for delayed entry, competing events and censoring where relevant.

A rate ratio compares incidence rates in two groups:

A rate difference is:

The rate ratio describes relative contrast; the rate difference describes excess events in the units of the rate. Health managers often need both. A large relative increase may correspond to few additional cases when baseline risk is low, whereas a modest relative increase may create substantial service demand in a large population.

6. Standardization and fair comparison

Crude measures can mislead when populations differ in age, sex or another strong determinant of disease. Standardization creates a summary measure that removes, or adjusts for, differences in the distribution of a selected variable.

Direct standardization

Direct standardization applies age-specific rates from each study population to the same standard population. The expected number of events in age group i is:

The directly standardized rate is:

Rates must use compatible units, and the same age categories and standard population must be used for all groups being compared. Direct standardization is most suitable when age-specific rates are stable and available for every group.

Indirect standardization

Indirect standardization applies standard age-specific rates to the study population to calculate expected events. The standardized mortality ratio is:

An SMR of 125 means that observed deaths are 25% higher than expected under the standard rates, assuming the comparison is appropriate. Indirect standardization is useful for small populations with unstable age-specific rates. SMRs from different populations should not automatically be compared because each may have a different age structure and expected-event calculation.

Standardization does not eliminate all confounding. It adjusts only for the variables and categories included, and residual confounding may remain.

7. Burden of disease: DALYs and QALYs

Frequency measures count cases or deaths, but health policy also needs measures combining premature mortality, disability, functioning and quality of life.

Disability-adjusted life years

The disability-adjusted life year measures health loss. It is commonly defined as:

where YLL is years of life lost due to premature mortality and YLD is years lived with disability.

A basic prevalence-based form of YLD is:

An incidence-based form uses new cases and their expected duration. Disability weights generally range from 0, representing full health, to 1, representing a health state equivalent to death. Contemporary global burden studies use specific methodological conventions for life expectancy, disability weights and uncertainty estimation.

Illustrative calculation. Suppose 100 deaths occur at an age with a remaining standard life expectancy of 30 years. YLL is 100 × 30 = 3,000 years. If 500 people live with a condition for two years and the disability weight is 0.20, YLD is 500 × 2 × 0.20 = 200 years. Total DALYs are therefore 3,200 years of healthy life lost.

DALYs support comparison across diseases and populations, but disability weights may not reflect every community’s values. They also do not inherently capture distributional fairness, caregiving burdens or social disadvantage.

Quality-adjusted life years

A QALY combines length and quality of life. In a simplified form:

Utility weights commonly range from 0, equivalent to death, to 1, equivalent to perfect health, although some instruments permit values below 0 for states considered worse than death. If a treatment provides 3 years at a utility of 0.8, it produces 2.4 QALYs.

QALYs are widely used in health technology assessment. The incremental cost-effectiveness ratio is:

The ICER is interpreted only alongside a decision threshold, uncertainty analysis, budget impact, equity considerations and ethical judgment. A cost-effective intervention may still be unaffordable at scale.

8. Person, place and time

Descriptive epidemiology begins by characterizing patterns according to person, place and time.

Person

Relevant characteristics include age, sex, gender, occupation, education, income, ethnicity, migration status, disability, behaviours, comorbidities and access to care. These variables can identify high-risk groups, reveal inequities and guide targeted interventions. They must be interpreted cautiously: group differences may reflect exposure, biology, measurement, diagnostic access, treatment or structural conditions.

Place

Place may refer to country, region, neighbourhood, workplace, school, health facility or household. Spatial differences may result from environmental exposures, climate, vector ecology, health-service access, population composition or data quality. Ecological comparisons are useful for generating hypotheses but cannot establish that every individual in a high-rate area experienced the suspected exposure.

Time

Time patterns include secular trends, seasonal variation, cyclical changes, outbreaks and birth-cohort effects. A sudden rise may indicate a true increase, improved detection, altered case definitions, reporting delays or changes in coding. Time-series interpretation should therefore examine surveillance processes as well as disease mechanisms.

A useful descriptive sequence is: define the population, verify the case definition, calculate appropriate denominators, stratify by person/place/time, compare with a meaningful reference and assess data quality before inferring causes.

9. Applied case: planning diabetes services

Illustrative case. A regional health authority is planning diabetes services. In 2024, its adult population is 100,000. The registry identifies 8,000 existing diabetes cases on 1 January. During the year, 1,200 new cases are recorded. The population contributes approximately 99,000 adult person-years because of migration and delayed registration. There are 180 deaths among people with diabetes, but only 20 are certified as directly caused by diabetes.

Analysis

Point prevalence on 1 January is 8,000 ÷ 100,000 = 8%. If the end-of-year population is similar, this is a reasonable planning estimate, though registry completeness must be checked.

Incidence rate is 1,200 ÷ 99,000 = 0.01212 cases per person-year, or approximately 12.1 per 1,000 person-years. The incidence rate is more appropriate than a simple risk because person-time is not identical for all adults.

The diabetes-specific case-fatality proportion, using all deaths among registered patients as the numerator, is 180 ÷ 8,000, or 2.25%, only if the denominator and follow-up period are explicitly defined and cases are followed adequately. The 20 certified diabetes deaths should instead be used for a cause-specific mortality analysis, not automatically for case-fatality.

The authority should stratify prevalence and incidence by age, sex, deprivation, ethnicity and locality. If one district has higher prevalence but also better screening, the apparent difference may partly reflect ascertainment. Age-standardized rates are needed before comparing districts with different age structures.

Management response

The authority should use prevalence to estimate the number of people needing continuing clinical management, incidence to plan diagnostic and prevention activity, and complication rates to forecast specialist and inpatient demand. It should audit case definitions, duplicate records, missing diagnoses and migration. Interventions should combine clinical care with population-level strategies addressing diet, physical activity, commercial determinants, poverty and access to preventive services.

Learning points

The same disease can require several measures because each answers a different question. No single frequency measure is sufficient for service planning. Denominator quality, case ascertainment and population structure are as important as arithmetic.

10. Limitations, ethics and contemporary practice

Epidemiological measures can be biased by under-reporting, differential access to diagnosis, incomplete follow-up, misclassification, changing coding systems and non-comparable denominators. Prevalence surveys may suffer from non-response. Administrative data may be large but clinically incomplete. Small numbers create unstable rates and wide uncertainty intervals.

Ethically, surveillance should use the least identifiable data needed, protect confidentiality, define legitimate data access and communicate uncertainty honestly. Results should not stigmatize communities. Equity analysis should consider who is missing from databases and whether a seemingly neutral measure hides unequal access or exposure.

Digital records, syndromic surveillance, genomic epidemiology and machine-learning systems can improve timeliness and pattern recognition. They do not remove the need for valid case definitions, representative data, denominator construction, calibration, human oversight and privacy governance. Algorithms trained on unequal health-care data may reproduce or amplify inequities. Managers should require documentation of data provenance, validation, subgroup performance and appropriate use.

Best practice includes defining the estimand before analysis, selecting a denominator that matches the population at risk, reporting numerator and denominator counts, stating the observation period, using confidence intervals where appropriate, standardizing transparently, conducting sensitivity analyses and distinguishing association from causation.

Chapter Summary

Epidemiology measures health events in defined populations using explicit numerators, denominators and time frames. Incidence measures new occurrence; prevalence measures existing disease and is influenced by incidence and duration. Mortality describes deaths in a population, case-fatality describes deaths among diagnosed cases, and survival describes remaining alive or event-free over time. Person-time permits analysis when observation varies among individuals.

Crude comparisons can be distorted by different age structures, making direct or indirect standardization valuable. DALYs combine years of life lost and years lived with disability to quantify health loss, while QALYs combine survival and health-related quality using utility weights. Person, place and time provide the central descriptive framework for identifying patterns. For health-care managers, sound epidemiological interpretation depends not only on calculation but also on case definitions, data quality, equity, ethics and the practical consequences of measurement.

Key Terms

Epidemiology: Study of the distribution and determinants of health events in populations and the application of this knowledge to control health problems.

Population at risk: People capable of experiencing the health event under study.

Case definition: Criteria used to classify a person as a case.

Cumulative incidence: Proportion of an initially disease-free population developing disease during a specified period.

Incidence rate: New cases divided by disease-free person-time at risk.

Prevalence: Proportion of a population with an existing condition.

Point prevalence: Prevalence at a specified moment.

Period prevalence: Prevalence during a specified interval.

Mortality rate: Deaths in a population during a specified time period.

Case-fatality proportion: Deaths among diagnosed cases of a disease.

Survival: Probability of remaining alive or event-free over time.

Person-time: Sum of individual times observed while at risk.

Standardization: Adjustment of rates to a common population structure.

Standard mortality ratio: Observed deaths divided by expected deaths, commonly multiplied by 100.

YLL: Years of life lost through premature death.

YLD: Years lived with disability.

DALY: Disability-adjusted life year, equal to YLL plus YLD.

QALY: Quality-adjusted life year combining time and utility-weighted health quality.

Utility weight: Value representing quality of a health state.

Person–place–time: Descriptive framework for characterizing population patterns.

Review Questions

Short Answer Questions

1. What is the difference between incidence and prevalence?
2. Why must a denominator represent the population at risk?
3. Define person-time and state one situation in which it is useful.
4. What does a case-fatality proportion measure?
5. Name the two components of a DALY.

Long Answer Questions

1. Explain cumulative incidence, incidence rate and prevalence, including their appropriate uses.
2. Describe direct and indirect standardization and explain why crude rates may be misleading.
3. Discuss mortality, survival and case-fatality as distinct measures of disease outcome.
4. Explain how DALYs and QALYs support health-care priority setting, including their limitations.
5. Describe how person, place and time analysis can guide regional health-service planning.

Analytical / Application Questions

1. A cohort of 5,000 disease-free individuals contributes 4,600 person-years and experiences 92 new cases. Calculate and interpret the incidence rate per 1,000 person-years.
2. A district has 12,000 existing cases in a population of 300,000. During the year, 900 new cases occur. Explain which measure would be used to estimate ongoing service needs and which would be used to describe new disease occurrence.
3. Two districts have different crude mortality rates but markedly different age structures. Design an analysis to determine whether the mortality difference persists after accounting for age, and explain how the result should be communicated.

Multiple Choice Questions with Answers

Question 1

Which measure is most directly appropriate for estimating the proportion of an initially disease-free cohort that develops disease during a specified period?

A. Cumulative incidence

B. Point prevalence

C. Case-fatality proportion

D. Crude mortality rate

Correct Answer: A. Cumulative incidence

Question 2

A surveillance programme observes 30 new cases during 12,000 person-years. What is the incidence rate per 1,000 person-years?

A. 0.25

B. 2.5

C. 25

D. 250

Correct Answer: B. 2.5

Question 3

Which measure is most useful for estimating the number of people currently requiring long-term services for a chronic condition?

A. Incidence rate

B. Point prevalence

C. Case-fatality proportion

D. Cause-specific mortality

Correct Answer: B. Point prevalence

Question 4

What is the principal purpose of direct standardization?

A. To remove all measurement error

B. To adjust rates to a common population structure

C. To convert prevalence into incidence

D. To calculate person-time for each participant

Correct Answer: B. To adjust rates to a common population structure

Question 5

If 15 of 300 diagnosed patients die during a defined follow-up period, the case-fatality proportion is:

A. 0.5%

B. 5%

C. 15%

D. 20%

Correct Answer: B. 5%

Question 6

Which expression correctly represents a DALY?

A. YLL − YLD

B. YLL ÷ YLD

C. YLL + YLD

D. QALY + YLD

Correct Answer: C. YLL + YLD

Question 7

A treatment provides 4 years at a utility weight of 0.75. Ignoring discounting, how many QALYs does it provide?

A. 0.75

B. 3.00

C. 4.75

D. 5.33

Correct Answer: B. 3.00

Question 8

An SMR of 125 most appropriately indicates that:

A. Observed deaths are 125% lower than expected

B. Observed deaths are 25% higher than expected

C. 25% of cases died

D. Incidence is 125 per 100,000 population

Correct Answer: B. Observed deaths are 25% higher than expected

Question 9

Which combination is the core descriptive framework for examining population disease patterns?

A. Cost, quality and access

B. Person, place and time

C. Sensitivity, specificity and validity

D. Input, process and output

Correct Answer: B. Person, place and time

Question 10

Why can prevalence increase even when incidence remains constant?

A. Disease duration may increase because survival improves

B. The denominator must always decrease

C. Case-fatality must increase

D. Standardization automatically raises prevalence

Correct Answer: A. Disease duration may increase because survival improves