USMLE Step 3 - Biostatistics and Epidemiology Exam
Practice Guide 2025-2026
1. Study design looks at two groups *Cohort Study*
and looks risk factors and fol-
lows for development of disease *relative risk* assesses results of Cohort Studies
observe PROspectively over time 1. Define Exposure/Risk Factor
what happens to groups with 2. Assess Development of an Outcome/Risk (Disease)
certain exposures or illnesses
Help ID future Health Consequences of a Particular Disor-
der
- Study unusual risk factors because captures subjects with
and without the unusual risk factor
2. The outcome is known from the *Case control studies*
start of the study & go back to
find exposures Odds ratio used to assess results of Case Controls
Retrospective study. Looks at *Odds Ratio = AD/BC*
odds of exposure on develop-
1. Identified Disease (Outcome Status)
ment of illness
- Control is No Disease
2. Asses Exposure/Risk Factors in the PAST
, USMLE Step 3 - Biostatistics and Epidemiology Exam
Practice Guide 2025-2026
3. Simultaneous measurement of *Cross-sectional study*
an exposure and outcome within
a specific population at a Certain *(aka "prevalence study")*
point in time
Examine association between risk factors and outcomes in
*"snapshot" study* - usually a specific population at one particular time "snapshot"
uses *surveys* - inexpensive & 1. Define the population
easy 2. Pick 1 time: Estimate the prevalence of risk factors and
disease at once
4. Types of Different Study Designs
5. Pragmatic Trials (PCT) High External Validity (Generalizability)
more reflective of practice situations;
Diverse pt pops w/fewer or minimal exclusion criteria
(Subjects similar to population of interest with disorder)
Active controls (standard/other common care)
Treatment reflects practice & is less standardized
,6. Explanatory Trials High Internal Validity (Causality)
7. Use of a Control Group Provide better estimate of the ettect of the intervention
Allow researchers to discard potential ettects of extraneous
factors that can distort the real ettect of intervention
Reduce Bias: Confounding, Placebo Ettect, Natural History
Increase the INTERNAL Validity of the study
8. Absolute Risk Reduction Attributable Risk Reduction=
Disease rate in Placebo group - Disease rate in Treatment
group
Attributable Risk Reduction
[C/C+D] - [A/A+B]
+ is the Treatment group (A & B)
-- is the placebo group (C & D)
, 9. Number needed to treat (NNT) Inverse of Absolute Risk Reduction
NNT = 1/ARR
ARR = Control group event rate - Experimental group event
rate
10. Absolute Risk Reduction (ARR) % risk in control group - % risk in treatment group
11. Attributable Risk Excess risk assumed by Exposed group
(risk in exposed - risk in unexposed) so outcome is attrib-
uted to exposure
[A / A+B] - [C/ C+D]
+ is Exposused group (A & B)
-- is non-exposed group (C &D)
12. Number Needed to Harm Inverse of Absolute Risk Increase (ARI)
Practice Guide 2025-2026
1. Study design looks at two groups *Cohort Study*
and looks risk factors and fol-
lows for development of disease *relative risk* assesses results of Cohort Studies
observe PROspectively over time 1. Define Exposure/Risk Factor
what happens to groups with 2. Assess Development of an Outcome/Risk (Disease)
certain exposures or illnesses
Help ID future Health Consequences of a Particular Disor-
der
- Study unusual risk factors because captures subjects with
and without the unusual risk factor
2. The outcome is known from the *Case control studies*
start of the study & go back to
find exposures Odds ratio used to assess results of Case Controls
Retrospective study. Looks at *Odds Ratio = AD/BC*
odds of exposure on develop-
1. Identified Disease (Outcome Status)
ment of illness
- Control is No Disease
2. Asses Exposure/Risk Factors in the PAST
, USMLE Step 3 - Biostatistics and Epidemiology Exam
Practice Guide 2025-2026
3. Simultaneous measurement of *Cross-sectional study*
an exposure and outcome within
a specific population at a Certain *(aka "prevalence study")*
point in time
Examine association between risk factors and outcomes in
*"snapshot" study* - usually a specific population at one particular time "snapshot"
uses *surveys* - inexpensive & 1. Define the population
easy 2. Pick 1 time: Estimate the prevalence of risk factors and
disease at once
4. Types of Different Study Designs
5. Pragmatic Trials (PCT) High External Validity (Generalizability)
more reflective of practice situations;
Diverse pt pops w/fewer or minimal exclusion criteria
(Subjects similar to population of interest with disorder)
Active controls (standard/other common care)
Treatment reflects practice & is less standardized
,6. Explanatory Trials High Internal Validity (Causality)
7. Use of a Control Group Provide better estimate of the ettect of the intervention
Allow researchers to discard potential ettects of extraneous
factors that can distort the real ettect of intervention
Reduce Bias: Confounding, Placebo Ettect, Natural History
Increase the INTERNAL Validity of the study
8. Absolute Risk Reduction Attributable Risk Reduction=
Disease rate in Placebo group - Disease rate in Treatment
group
Attributable Risk Reduction
[C/C+D] - [A/A+B]
+ is the Treatment group (A & B)
-- is the placebo group (C & D)
, 9. Number needed to treat (NNT) Inverse of Absolute Risk Reduction
NNT = 1/ARR
ARR = Control group event rate - Experimental group event
rate
10. Absolute Risk Reduction (ARR) % risk in control group - % risk in treatment group
11. Attributable Risk Excess risk assumed by Exposed group
(risk in exposed - risk in unexposed) so outcome is attrib-
uted to exposure
[A / A+B] - [C/ C+D]
+ is Exposused group (A & B)
-- is non-exposed group (C &D)
12. Number Needed to Harm Inverse of Absolute Risk Increase (ARI)
