Master Biostatistics & Literature Appraisal
for USMLE Step 3

Core Concepts

Biostatistics and literature appraisal are fundamental for evidence-based practice and critical evaluation of medical research.

• Study Designs Hierarchy (Strongest to Weakest Evidence for Causation):
  • Meta-analysis/Systematic Review: Combines results from multiple studies to obtain a pooled estimate; mitigates individual study bias if well-conducted.
  • Randomized Controlled Trial (RCT): Gold standard for intervention efficacy; randomly assigns participants to intervention or control, minimizing confounding.
  • Cohort Study: Observational; follows a group exposed to a factor and a group not exposed over time to see who develops an outcome. Can determine incidence and relative risk.
  • Case-Control Study: Observational; compares exposure history between individuals with a disease (cases) and individuals without (controls). Determines odds ratio. Prone to recall bias.
  • Cross-sectional Study: Observational; measures exposure and outcome at a single point in time (prevalence). Cannot establish temporality or causality.
  • Case Series/Report: Descriptive; describes characteristics of a few patients with a particular disease or unusual presentation. Hypothesis generating.
• Bias: Systematic error leading to a deviation from the truth.
  • Selection Bias: Differences between study groups (e.g., non-random assignment, healthy user bias, loss to follow-up).
  • Information Bias: Errors in data collection or measurement (e.g., recall bias, observer bias, interviewer bias).
  • Confounding: An extraneous variable distorts the observed association between exposure and outcome. Can be controlled for in design (randomization) or analysis (stratification, regression).
• Validity:
  • Internal Validity: Extent to which the observed effects are due to the intervention/exposure and not other factors (bias, confounding).
  • External Validity (Generalizability): Extent to which findings can be applied to other populations or settings.
• Hypothesis Testing:
  • Null Hypothesis (H0): No difference or association.
  • Alternative Hypothesis (HA): A difference or association exists.
• P-value: Probability of observing the data (or more extreme) if the null hypothesis were true. Typically, p < 0.05 is statistically significant.
  • Type I Error (alpha, α): Rejecting H0 when it is true (false positive). Set by significance level (e.g., 0.05).
  • Type II Error (beta, β): Failing to reject H0 when it is false (false negative).
  • Power (1-β): Probability of correctly rejecting H0 when it is false. Increases with sample size, effect size, and alpha.
• Confidence Intervals (CI): Range of values likely to contain the true population parameter.
  • A 95% CI means if the study were repeated many times, 95% of the CIs would contain the true value.
  • If CI for RR or OR includes 1.0, or CI for mean difference includes 0, then the result is NOT statistically significant.
  • Narrower CI = more precise estimate.
• Measures of Association/Effect:
  • Relative Risk (RR): (Risk in exposed)/(Risk in unexposed). Used in cohort studies and RCTs.
  • Odds Ratio (OR): (Odds of exposure in cases)/(Odds of exposure in controls). Used in case-control studies.
  • Absolute Risk Reduction (ARR): Risk(control) - Risk(intervention).
  • Relative Risk Reduction (RRR): (ARR)/(Risk in control) or 1 - RR.
  • Number Needed to Treat (NNT): 1/ARR. Number of patients to treat for one additional beneficial outcome. Round UP.
  • Number Needed to Harm (NNH): 1/Absolute Risk Increase. Number of patients to expose for one additional harmful outcome. Round DOWN.
  • Hazard Ratio (HR): Ratio of event rates in two groups over time, used in survival analysis (e.g., Kaplan-Meier curves). HR < 1 indicates lower event rate in intervention group.
• Diagnostic Test Characteristics:
  • Sensitivity: (True Positives)/(All with disease). Rule OUT with a high SN-NOUT.
  • Specificit: (True Negatives)/(All without disease). Rule IN with a high SP-PIN.
  • Positive Predictive Value (PPV): (True Positives)/(All Positives). Probability of disease given a positive test. Highly affected by prevalence.
  • Negative Predictive Value (NPV): (True Negatives)/(All Negatives). Probability of no disease given a negative test. Highly affected by prevalence.
  • Likelihood Ratios (LR):
    • LR+: Sensitivity / (1-Specificity). How much a positive test increases the probability of disease.
    • LR-: (1-Sensitivity) / Specificity. How much a negative test decreases the probability of disease.
• Blinding: Concealing treatment assignment to prevent bias.
  • Single: Patient unaware.
  • Double: Patient and investigator unaware.
  • Triple: Patient, investigator, and outcome assessor unaware.

Clinical Presentation

• On USMLE Step 3, biostatistics and literature appraisal concepts appear as clinical vignettes describing research studies, journal articles, pharmaceutical advertisements, or public health scenarios.
• Questions require critical evaluation of study design, identification of biases, interpretation of statistical results (p-values, CIs, NNT, diagnostic test metrics), and application of evidence to patient care decisions.
• You may be asked to choose the most appropriate study design for a given research question or to identify flaws in a presented study.

Diagnosis (Gold Standard)

The "gold standard" for evaluating research is a systematic, critical appraisal using established frameworks (e.g., CONSORT guidelines for RCTs). On the exam, this translates to correctly identifying the study type, biases, strengths/weaknesses, and interpreting the numerical findings (e.g., Confidence Intervals, P-values, NNT/NNH, Sensitivity/Specificity) in the context of the clinical question.

Management (First Line)

Apply principles of evidence-based medicine:

• Identify the most appropriate level of evidence for a clinical question.
• Critically appraise the methodology and statistical analysis of studies before applying findings to patient care.
• Use NNT/NNH to communicate risks and benefits to patients in an understandable way.
• Be aware of how prevalence affects the utility of diagnostic tests (PPV, NPV).
• Understand the difference between statistical significance (p-value) and clinical significance (effect size, NNT).

Exam Red Flags