Mortality, Morbidity, Prevalence, Incidence, and Risk Data Interpretation
Key Takeaways
- Incidence measures new cases arising over time and requires a population-at-risk denominator; prevalence measures all existing cases at a point or period and approximates incidence multiplied by average disease duration.
- Crude rates use the total population denominator; specific rates stratify by age, sex, or cause; age-adjusted rates apply a standard population distribution to enable fair comparisons across populations with different age structures.
- Case fatality rate divides deaths by diagnosed cases, while cause-specific mortality divides deaths by the total population — mixing these denominators is a common exam error.
- Relative risk compares incidence in exposed versus unexposed groups; attributable risk quantifies the excess risk due to exposure; population attributable risk extends this to the whole population.
- Person-time denominators account for varying follow-up durations in cohort studies and are essential when participants enter or leave observation at different times.
Quick Answer: Incidence counts new cases over time and needs a population-at-risk denominator; prevalence counts all existing cases at a point or period. Mortality and morbidity rates depend on choosing the correct denominator — total population for crude or cause-specific mortality, diagnosed cases for case fatality. Risk measures (relative risk, attributable risk, odds ratio) quantify how strongly an exposure associates with an outcome.
Incidence Versus Prevalence
Incidence measures the transition from disease-free to diseased state. Cumulative incidence (also called risk or incidence proportion) equals the number of new cases divided by the population at risk at the start of a defined period: CI = new cases / population at risk. It assumes a fixed population and a closed time window, and it is expressed as a proportion (e.g., 0.02 or 2%). Incidence rate (incidence density) replaces the denominator with person-time — the sum of each individual's disease-free observation time — yielding units of 1/time (e.g., 5 cases per 1,000 person-years). Person-time is the correct denominator when participants enter or leave a cohort at different times or are lost to follow-up.
Prevalence measures the burden of existing disease. Point prevalence = existing cases at one instant / total population. Period prevalence = all cases present during a defined interval / total population. The steady-state relationship prevalence ≈ incidence × average disease duration holds for chronic, stable conditions. A disease can have high prevalence but low incidence when duration is long (diabetes, arthritis), or high incidence but low prevalence when duration is short or mortality is high (influenza, untreated pancreatic cancer). Reading prevalence as if it were incidence is a classic interpretation trap.
Mortality and Morbidity Rate Denominators
| Measure | Numerator | Denominator | Interpretation |
|---|---|---|---|
| Crude mortality rate | All deaths | Total mid-year population | Overall death burden |
| Cause-specific mortality | Deaths from one cause | Total mid-year population | Deaths attributable to that cause |
| Age-specific mortality | Deaths in age band | Population in that age band | Mortality for that age stratum |
| Case fatality rate (CFR) | Deaths from disease | Diagnosed cases of disease | Severity / lethality of disease |
| Proportionate mortality | Deaths from one cause | All deaths | Relative importance of cause |
| Infant mortality rate | Deaths <1 year | Live births that year | Child health sentinel indicator |
The single most common denominator error on CPH items is computing case fatality using the total population denominator (which produces a cause-specific mortality rate instead). Case fatality rate answers the question: of people who already have the disease, what fraction die? It is not a population-based rate.
Age adjustment removes the confounding effect of differing age distributions. Direct age adjustment applies age-specific rates from the study population to a standard population (e.g., U.S. 2000 standard million) and sums expected deaths to produce a single comparable rate. Without age adjustment, a population with many older residents can appear to have higher crude mortality even if every age-specific rate is lower — a trap called the confounding by age problem. Standardized mortality ratio (SMR) is the indirect method: observed deaths in the study population divided by expected deaths based on standard rates, a ratio useful when age-specific strata are too small for stable direct adjustment.
Risk, Association, and Attributable Measures
Relative risk (RR) = incidence in exposed / incidence in unexposed. RR > 1 suggests the exposure is associated with increased disease risk; RR = 1 means no association; RR < 1 suggests a protective association. In a 2×2 table, RR = [a/(a+b)] / [c/(c+d)]. Odds ratio (OR) approximates RR when the disease is rare (rare disease assumption, <10% prevalence) and is the measure of choice in case-control studies because the researcher selects cases and controls rather than following a population. OR = ad/bc.
Attributable risk (risk difference, excess risk) = incidence in exposed − incidence in unexposed. It quantifies the absolute additional disease burden from exposure. Population attributable risk (PAR) = attributable risk × prevalence of exposure in the population; it answers how much disease in the entire population could be eliminated by removing the exposure. Population attributable fraction expresses PAR as a percentage of total incidence. These attributable measures matter for prioritizing interventions: a weak risk factor that is very common (PAR high) may warrant more public health action than a strong risk factor that is rare (PAR low).
Worked example: Suppose a cohort study of smoking and lung cancer finds incidence of 120 per 100,000 person-years among smokers and 10 per 100,000 among non-smokers. The relative risk is 120/10 = 12, indicating smokers have 12 times the lung cancer rate. The attributable risk is 120 − 10 = 110 per 100,000 person-years — the excess disease rate from smoking. If 20% of the population smokes, the population attributable risk is 110 × 0.20 = 22 per 100,000, and the population attributable fraction is 22/((120×0.20)+(10×0.80)) = 22/32 = 68.8%, meaning smoking accounts for nearly 69% of lung cancer in this population despite the exposed fraction being only one in five. This example shows why a combination of relative risk, attributable risk, and exposure prevalence is needed to understand the full public health burden.
Interpreting risk data also requires distinguishing statistical significance from public health significance. A large relative risk with a wide confidence interval and a tiny exposed population may not be actionable, whereas a modest relative risk affecting most of the population can produce substantial attributable burden. Always pair the magnitude of association (RR, OR) with the absolute measure (AR, PAR) before drawing intervention conclusions.
A cohort study reports 120 new diabetes cases among 4,000 adults followed for an average of 2.5 years. Which calculation gives the incidence rate?
A town reports 300 deaths from a hepatitis outbreak among 1,200 diagnosed cases, in a population of 50,000. Which is the case fatality rate?
Population A has a higher crude mortality rate than Population B, but every age-specific mortality rate is lower in A. Which explanation is most likely?