Section 2.1: Epidemiology & Risk Measures

Epidemiology & Risk Measures in Clinical Practice

Clinical epidemiology on the USMLE Step 3 focuses on quantitative measures used to evaluate disease frequency, assess risk, and determine the clinical efficacy or potential harms of interventions. Mastering these calculations and, more importantly, understanding their clinical interpretations and limitations is essential for clinical decision-making.

Disease Frequency: Incidence vs. Prevalence

Understanding the distinction between incidence and prevalence is fundamental to interpreting clinical research and tracking public health trends.

Incidence measures the rate at which new cases of a disease develop in a population over a specified period. It is calculated by dividing the number of new cases by the total population at risk during that time frame. The population at risk excludes individuals who already have the disease at the start of the study period. Incidence can be expressed as cumulative incidence (risk) or incidence rate (which uses person-time in the denominator to account for varying follow-up periods).

Prevalence measures the proportion of a population that has a disease at a specific point in time (point prevalence) or over a specified period (period prevalence). The numerator includes both new and pre-existing cases, while the denominator is the total population at that time.

The mathematical relationship between these two measures is represented as: $\text{Prevalence} \approx \text{Incidence} \times \text{Average Duration of Disease}$ This approximation holds true for chronic, stable diseases where incidence and duration are relatively constant. Changes in disease management directly influence this relationship:

• Interventions that prolong survival without curing (e.g., insulin for Type 1 Diabetes, disease-modifying therapies for multiple sclerosis) will increase prevalence because the duration of the disease is prolonged, even if the incidence remains unchanged.
• Curative therapies (e.g., direct-acting antivirals for Hepatitis C) will decrease prevalence by reducing the duration of active disease in the population, even if incidence remains constant.
• Primary prevention strategies reduce incidence, which subsequently decreases prevalence over time.
• Highly fatal conditions result in low prevalence relative to incidence because the duration of disease is extremely short.

Measures of Association: Relative Risk and Odds Ratios

To determine whether an exposure is associated with an outcome, researchers calculate measures of association. The choice of measure depends on the study design.

Relative Risk (RR) is the ratio of the risk of an event in an exposed group to the risk in an unexposed group. It is calculated from prospective cohort studies or clinical trials using a standard 2 by 2 table: $\text{Relative Risk (RR)} = \frac{\text{Risk in Exposed}}{\text{Risk in Unexposed}} = \frac{a / (a + b)}{c / (c + d)}$ An RR of 1.0 indicates no association between the exposure and the outcome. An RR greater than 1.0 indicates that the exposure is associated with an increased risk of the outcome (e.g., smoking and lung cancer). An RR less than 1.0 indicates a protective effect (e.g., vaccination and infection).

Odds Ratio (OR) is the ratio of the odds of exposure among cases (those with the disease) to the odds of exposure among controls (those without the disease). It is the standard measure of association in case-control studies: $\text{Odds Ratio (OR)} = \frac{\text{Odds of Exposure in Cases}}{\text{Odds of Exposure in Controls}} = \frac{a / c}{b / d} = \frac{ad}{bc}$ In a case-control study, we cannot directly calculate incidence or risk because we pre-determine the number of cases and controls. Therefore, we calculate the odds of exposure.

The Rare Disease Assumption: The OR is a reliable approximation of the RR only when the disease is rare in the population (typically affecting less than 10% of the population). Under these circumstances, the values of $a$ and $c$ are very small relative to $b$ and $d$, respectively. Consequently, $a + b \approx b$ and $c + d \approx d$, simplifying the RR formula to approximate the OR formula ($ad/bc$). If the disease is common, the OR will exaggerate the association, appearing further from 1.0 (more extreme) than the true RR.

Quantifying Intervention Effects: ARR, RRR, and Attributable Risk

In therapeutic trials, the impact of an intervention is quantified using absolute and relative metrics.

Absolute Risk Reduction (ARR) represents the absolute difference in event rates between the control group (unexposed to treatment) and the treatment group (exposed to treatment): $\text{ARR} = \text{Control Event Rate (CER)} - \text{Experimental Event Rate (EER)} = \frac{c}{c + d} - \frac{a}{a + b}$ The ARR is highly clinically relevant because it reflects the baseline risk of the population. A treatment that reduces risk from 2% to 1% has the same relative effect as a treatment that reduces risk from 50% to 25%, but the absolute benefit is far greater in the latter group.

Relative Risk Reduction (RRR) measures the proportion of baseline risk that is eliminated by the intervention: $\text{RRR} = \frac{\text{CER} - \text{EER}}{\text{CER}} = 1 - \text{RR}$ Drug advertisements frequently emphasize the RRR because it yields a larger, more impressive-sounding percentage than the ARR. However, a large RRR can mask a clinically trivial absolute benefit if the baseline risk is very low.

Attributable Risk (AR) (also known as Absolute Risk Increase, ARI, when discussing harms) measures the excess risk of disease in an exposed population that can be directly attributed to the exposure: $\text{AR} = \text{EER} - \text{CER}$ For example, if the risk of developing a side effect is 5% in patients taking a drug and 1% in those taking a placebo, the AR is 4%. This represents the absolute risk increase associated with the medication.

Attributable Risk Percent (AR%) represents the proportion of disease in the exposed group that is due to the exposure: $\text{AR\%} = \frac{\text{EER} - \text{CER}}{\text{EER}} \times 100 = \frac{\text{RR} - 1}{\text{RR}} \times 100$

Population-Attributable Risk (PAR) estimates the excess risk of disease in the entire population (both exposed and unexposed) that is due to the exposure: $\text{PAR} = \text{Risk in Total Population} - \text{Risk in Unexposed}$

The Population-Attributable Risk Fraction (PARF) represents the proportion of disease in the entire population that could be eliminated if the exposure were completely removed: $\text{PARF} = \frac{\text{PAR}}{\text{Risk in Total Population}}$ This is a crucial metric for public health, as it guides policy decisions regarding risk factor modification (e.g., smoking cessation campaigns).

Clinical Actionability: NNT and NNH

To make statistics clinically meaningful, clinicians calculate the number of patients that must be treated or exposed to see a specific outcome.

Number Needed to Treat (NNT) is the number of patients who must receive the treatment for a specified duration to prevent one additional adverse outcome: $\text{NNT} = \frac{1}{\text{ARR}}$ When calculating NNT, if the division does not result in a whole number, always round up to the next integer (e.g., an NNT of 12.1 must be reported as 13). This conservative approach ensures that clinicians do not overestimate the clinical benefit of a therapy.

Number Needed to Harm (NNH) is the number of patients who must be exposed to a risk factor or therapy for a specified duration to cause one additional adverse event: $\text{NNH} = \frac{1}{\text{ARI}} = \frac{1}{\text{Attributable Risk}}$ For NNH, if the calculation results in a decimal, always round down to the next integer (e.g., an NNH of 25.8 is reported as 25). This conservative approach prevents clinicians from underestimating the potential harms of an intervention.

On the USMLE Step 3, you must be prepared to calculate these metrics from a clinical scenario, identify how they change when baseline risks vary, and use them to counsel patients. For example, in a patient population with low baseline risk, the NNT for a drug will be higher than in a high-risk population, even if the drug's relative efficacy is identical.