You went to medical school to learn medicine, and now you're staring at a 2x2 table wondering which box is "b."
If biostatistics makes your brain go quiet, you're in good company. Plenty of students who are strong in physiology and pathology freeze when a question asks for a positive predictive value or a number needed to treat.
Here's the encouraging part: biostats is one of the most predictable subjects on Step 1. The same handful of concepts come up again and again, and once you've drilled them, these questions become some of the fastest points on the exam.
What Step 1 Biostatistics Topics Are Tested?
Step 1 biostatistics centers on the 2x2 table (sensitivity, specificity, PPV and NPV), how prevalence changes predictive values, likelihood ratios, relative risk vs. odds ratio, absolute and relative risk reduction, NNT and NNH, study designs, common biases, type I and type II error, power, and confidence intervals. Most questions give you a short study description or a small data set and ask you to calculate one value or interpret a result. The math is basic arithmetic. The challenge is setting it up correctly under time pressure.
Biostatistics, epidemiology and interpretation of the medical literature make up a small but reliable slice of the exam. Check the content outline on usmle.org for the official breakdown. Because these questions are so formula-driven, they reward practice more than almost any other topic.
The 2x2 Table: A Worked Example
Imagine a new screening test used in 1,000 people. The disease prevalence is 20%, the test's sensitivity is 90%, and its specificity is 80%.
Step 1: Fill in the Table
20% of 1,000 means 200 people have the disease and 800 don't. With 90% sensitivity, 180 of the 200 sick people test positive (true positives) and 20 test negative (false negatives). With 80% specificity, 640 of the 800 healthy people test negative (true negatives) and 160 test positive (false positives).
| Disease Present | Disease Absent | Total | |
|---|---|---|---|
| Test Positive | a = 180 (TP) | b = 160 (FP) | 340 |
| Test Negative | c = 20 (FN) | d = 640 (TN) | 660 |
| Total | 200 | 800 | 1,000 |
Step 2: Calculate
- Sensitivity = a / (a + c) = 180 / 200 = 90%. Sensitive tests help rule a disease out when negative.
- Specificity = d / (b + d) = 640 / 800 = 80%. Specific tests help rule a disease in when positive.
- PPV = a / (a + b) = 180 / 340, or about 53%. Only about half of positive results are true positives.
- NPV = d / (c + d) = 640 / 660, or about 97%.
- Positive likelihood ratio = sensitivity / (1 minus specificity) = 0.90 / 0.20 = 4.5.
- Negative likelihood ratio = (1 minus sensitivity) / specificity = 0.10 / 0.80 = 0.125.
Step 3: Change the Prevalence
Now use the same test in a population where prevalence is only 2%. Of 1,000 people, 20 have the disease: 18 true positives and 2 false negatives. Of the 980 without it, 784 are true negatives and 196 are false positives. PPV drops to 18 / 214, or about 8%, while NPV rises to 784 / 786, or about 99.7%.
Sensitivity and specificity didn't change at all. That's the rule Step 1 loves: prevalence changes PPV and NPV, not sensitivity or specificity. Higher prevalence raises PPV and lowers NPV.
Cutoff Questions
If a test's cutoff is lowered so more people count as positive, sensitivity rises, specificity falls, false positives increase and false negatives decrease. Questions often show this with a graph of two overlapping populations.
Risk, Treatment Effect and Study Design
Relative Risk vs. Odds Ratio
Relative risk compares the incidence of disease in exposed vs. unexposed people and comes from cohort studies. Odds ratio compares the odds of exposure in cases vs. controls and comes from case-control studies. When a disease is rare, the odds ratio approximates the relative risk.
Quick odds ratio example: in a case-control study, 60 of 100 cases and 20 of 100 controls were exposed. OR = (60 x 80) / (20 x 40) = 4,800 / 800 = 6.
ARR, RRR, NNT and NNH
Suppose stroke occurs in 10% of a placebo group and 6% of a treatment group.
- Absolute risk reduction = 10% minus 6% = 4%, or 0.04.
- Relative risk = 0.06 / 0.10 = 0.6.
- Relative risk reduction = 0.04 / 0.10 = 40%, which is the same as 1 minus RR.
- Number needed to treat = 1 / ARR = 1 / 0.04 = 25.
If major bleeding occurs in 3% of the treatment group vs. 1% of the placebo group, the absolute risk increase is 0.02 and the number needed to harm is 1 / 0.02 = 50.
Study Designs at a Glance
| Design | How It Works | Key Measure |
|---|---|---|
| Cross-sectional | Exposure and disease measured at one point in time | Prevalence |
| Case-control | Starts with disease status, looks back for exposure | Odds ratio |
| Cohort | Starts with exposure status, follows for outcomes | Relative risk, incidence |
| Randomized controlled trial | Randomly assigns an intervention | Strongest evidence for causation |
| Meta-analysis | Pools results of multiple studies | Pooled estimate |
Bias, Errors and Confidence Intervals
Biases Step 1 Likes to Test
- Selection bias: the study population doesn't represent the target population, including loss to follow-up.
- Recall bias: cases remember past exposures differently than controls, a classic case-control problem.
- Lead-time bias: screening finds disease earlier, so survival from diagnosis looks longer even if death occurs at the same time.
- Length-time bias: screening tends to pick up slower-growing disease, making screen-detected cases look less aggressive.
- Hawthorne effect: subjects change behavior because they know they're being studied.
- Confounding: a third variable is linked to both exposure and outcome. Randomization, matching and stratification help reduce it.
Type I Error, Type II Error and Power
A type I error (alpha) means concluding there's an effect when there isn't one, a false positive. A type II error (beta) means missing a real effect, a false negative. Power is 1 minus beta, and it increases with a larger sample size, a larger true effect size, or a higher alpha.
Confidence Intervals
For a relative risk or odds ratio, a 95% confidence interval that includes 1 means the result isn't statistically significant. For a difference between means, the null value is 0. A narrower interval usually reflects a larger sample size.
A Two-Week Step 1 Biostatistics Study Plan
- Days 1 to 2: Learn the 2x2 table. Rebuild the example above by hand from a blank page until you can do it without notes.
- Days 3 to 4: Practice prevalence and cutoff questions. Change one number at a time and predict what happens before you calculate.
- Days 5 to 6: Cover RR, OR, ARR, RRR, NNT and NNH. Do every calculation twice, once with fractions and once with percentages.
- Days 7 to 9: Study designs and bias. For every question, name the design first, then the bias.
- Days 10 to 11: Type I and II error, power and confidence intervals.
- Days 12 to 14: Mixed, timed blocks. Then keep a few biostat questions in your daily mix for the rest of dedicated so the formulas stay fresh.
To see how biostats fits with the rest of your weak areas, take the free Step 1 Free 120 diagnostic on MedSchoolGuru. It's MedSchoolGuru's own question set, separate from the official NBME Free 120, and it gives you a scaled score estimate and pass likelihood with a free account.
What Most Students Get Wrong About Step 1 Biostatistics
The most common mistake is memorizing formulas without understanding the table. If you only memorize "PPV equals a over a plus b," a question with the columns flipped will catch you. Always draw the table yourself, in the same orientation, every time.
The second mistake is mixing up the measure and the study design. Odds ratios belong to case-control studies, relative risk to cohort studies. If you identify the design first, the right measure usually becomes obvious.
The third mistake is saving biostats for the last week. It's very learnable, but it fades quickly without practice. A few questions a day across dedicated beats a cram session.
And finally, students rush the arithmetic. These are easy points, so slow down, write out the numbers, and check your answer against common sense. A PPV of 95% in a rare disease should make you look again.
Turn Biostats Into Reliable Points
Biostatistics rewards repetition more than any other Step 1 subject. The Step 1 Question Bank from MedSchoolBro gives you 2,000+ NBME-style questions with clinical-reasoning walkthroughs and CBSE-style diagnostic self-assessments, so you can mix biostats into your daily blocks and track progress. If you'd like a single place to review concepts across all Step 1 subjects, The Complete USMLE Step 1 Guide from MedSchoolBro is built for that.
Frequently Asked Questions
How much biostatistics is on Step 1? It's a smaller share of the exam than the major organ systems, but it shows up reliably. The usmle.org content outline lists the official breakdown for biostatistics, epidemiology and interpretation of the medical literature. Because the questions are formula-based, they're some of the most learnable points on the test.
Is the math on Step 1 biostats hard? Not usually. Most questions use round numbers so the arithmetic stays simple, and the real challenge is setting up the problem correctly. Practice doing the calculations quickly by hand, and check usmle.org for the current list of test-day tools.
What's the difference between sensitivity and PPV? Sensitivity is the probability that a person with the disease tests positive. PPV is the probability that a person who tests positive actually has the disease. Sensitivity doesn't change with prevalence, but PPV does.
When should I start studying biostatistics for Step 1? Start early in dedicated, or even before it, and then keep a few questions in your daily mix. Biostats is easy to forget, so steady practice works better than one intense week.





