Epidemiology Exam Questions And Answers

Getting Through an Epidemiology Exam Without Losing Your Mind

I spent three semesters as a graduate teaching assistant for an epidemiology course, which means I saw every version of these exams over and over again. The questions are rarely trick questions, but they do test whether you actually understand the concepts or just memorized definitions. I have compiled study guides, reviewed old exams, and watched students struggle with the same topics repeatedly. This is what actually works. Epidemiology Exam Questions And Answers from university courses tend to fall into a few predictable categories. The first is study design — you will be given a scenario and asked whether it is cross-sectional, cohort, case-control, or randomized controlled trial, then asked to justify your choice and identify the associated measures of association. The second category is biostatistics applied to epidemiology, covering confidence intervals, p-values, sensitivity, specificity, positive and negative predictive values, and relative risk calculations. The third is causal inference and bias — selection bias, information bias, confounding, and how to detect or control for them. Here is something most review guides do not emphasize: the hardest part of these exams is not calculating odds ratios. It is correctly identifying which measure applies in which situation. Students routinely write the formula for relative risk when the question describes a case-control study where you can only compute an odds ratio. Once you lock in the study design, everything else follows mechanically. That is where people lose points.

Common Epidemiology Exam Questions And Answers Breakdown

Question type 1 — Study design identification. You will see a description like "researchers identified 200 patients with lung cancer and 200 matched controls without lung cancer, then asked both groups about their smoking history." The answer is case-control. The measure is odds ratio. The direction of inquiry is retrospective. If the scenario said they identified smokers and non-smokers from a population and then followed them forward in time to see who developed lung cancer, that would be a prospective cohort study using relative risk. This distinction comes up on virtually every exam. Question type 2 — Screening test characteristics. A typical problem gives you a 2x2 table and asks for sensitivity, specificity, positive predictive value, and negative predictive value. The formulas are standard. Sensitivity is true positives divided by all actual positives. Specificity is true negatives divided by all actual negatives. PPV is true positives divided by all positive test results. NPV is true negatives divided by all negative test results. The trap here is confusing sensitivity with PPV. They are completely different. Sensitivity is a property of the test. PPV depends on disease prevalence in the population being screened. I once saw a student lose ten points because they reported PPV as a fixed test characteristic. Prevalence changes PPV constantly even when the test itself does not change. Question type 3 — Bias and confounding. Exams love asking you to distinguish between confounding and effect modification. Confounding is a mixing of effects where a third variable distorts the observed association. You control for it by stratification, matching, or regression. Effect modification is when the magnitude or direction of an association genuinely differs across levels of a third variable. You do not adjust for effect modification. You report it separately. A classic exam question will present data showing that a drug reduces mortality in men but increases it in women. That is not confounding. That is effect modification by sex. Students who answer "confounding" on this get it wrong almost every time.

Measures of Association and When to Use Each One

Relative risk compares incidence in exposed versus unexposed groups. It applies to cohort studies and randomized trials. Odds ratio compares the odds of exposure in cases versus controls. It applies to case-control studies. Risk ratio is essentially the same as relative risk but uses cumulative incidence rather than incidence rate. Rate ratio uses person-time denominators. Hazard ratio comes from survival analysis with time-to-event data. The nuance that separates passing students from top marks: odds ratios approximate relative risks only when the disease is rare. When prevalence exceeds about ten percent, the odds ratio diverges noticeably from the relative risk, and exam questions that ask you to interpret the magnitude will expect you to note that. I have a specific memory of a practice exam where a question described a common condition and presented an odds ratio of 3.2. The incorrect answer choices included "the risk is 3.2 times higher" and the correct choice was "the odds of exposure are 3.2 times higher in cases than controls." Missing that distinction is an easy way to waste points.

Calculating Confidence Intervals for Rate Ratios

This shows up frequently and most students skip it because they do not remember the log-transformation steps. For a rate ratio, you take the natural logarithm of the rate ratio, calculate the standard error as the square root of one over the cases in the numerator plus one over the cases in the denominator, multiply by 1.96 for a 95% confidence interval, then exponentiate the bounds back to the original scale. The shortcut is using an epidemiology calculator or spreadsheet formula instead of doing it by hand under exam pressure. If your program allows calculators during the exam, use it. If not, practice the log method until you can do it in under two minutes. I timed myself during practice sessions and it took about ninety seconds once I stopped second-guessing the exponentiation step. Bias questions on exams are really testing whether you can think through a study critically. The most common framework is: identify the comparison group, determine whether selection into the study differs by exposure or outcome status, and check whether measurement of exposure or outcome differs systematically between groups. Here is an edge case that caught me off guard during my TA years. A question described a nested case-control study within a prospective cohort. The cases were identified from hospital records rather than from the cohort follow-up itself. The correct answer about bias was that this introduces selection bias because hospital-based cases may have different exposure distributions than the source cohort. Most students read "nested" and automatically assumed the design protected against selection bias. It does not, not when you switch the case ascertainment source mid-study. I now flag this pattern immediately whenever I see a nested design with hospital-ascertained cases. Another common setup involves recall bias in case-control studies. Cases with a serious diagnosis remember past exposures more thoroughly than healthy controls. The exam answer is usually to suggest using objective records like prescription databases or employment records rather than relying on self-report. This is also the practical solution in real research, not just on paper.

Study Design Comparison Table

Cross-sectional studies measure prevalence at a single point in time. They cannot establish temporality. Cohort studies follow exposed and unexposed groups forward and measure incidence. They establish temporality but can be expensive and slow. Case-control studies start with disease status and look backward for exposure. They are efficient for rare diseases but vulnerable to recall and selection bias. Randomized controlled trials assign exposure randomly and are the strongest design for causal inference, though they are not always ethical or feasible for harmful exposures. The exam question that trips people up is mixing up ecological studies with cross-sectional studies. An ecological study measures exposure and outcome at the population level. A cross-sectional study measures both at the individual level in a snapshot. The ecological fallacy is when you infer individual-level associations from group-level data. I once graded an exam where a student described an ecological study as cross-sectional and wrote that individual risk could be estimated from it. Wrong on both counts. Ecological data cannot estimate individual risk. That is the ecological fallacy, plain and simple.

A Note on Incidence vs Prevalence Problems

These are among the most straightforward questions on epidemiology exams, which is why students sometimes overthink them. Point prevalence is the number of existing cases at a specific time divided by the population at risk. Period prevalence includes all cases during a defined interval. Incidence proportion uses a fixed population at risk over a specified period. Incidence rate uses person-time. The conversion between prevalence and incidence only holds at steady state, and you need to know that assumption explicitly. The formula P equals I times D applies when incidence and duration are constant and the population is stable. Exams occasionally test whether you recognize when the steady-state assumption is violated, such as during an emerging outbreak where incidence is changing rapidly. In those situations, prevalence and incidence will not track each other predictably. Do not memorize definitions in isolation. Work through full problems from end to end. The exam rewards procedural fluency, not vocabulary recall. Practice setting up 2x2 tables from word problems before you touch any formula. If you can set up the table correctly, the calculation is trivial. Focus your energy on the things that cause errors: misidentifying the study design, confusing odds ratio with relative risk, mixing up confounding with effect modification, and applying prevalence-incidence relationships outside steady state. The resource I recommended most often to my students was a combination of past exam papers from your own department and the CDC's principles of epidemiology course materials. The CDC modules are free and cover the foundational material thoroughly. Pair them with your professor's old exams because local exams reflect local emphases. Some programs weight biostatistics heavily. Others weight study design and bias. Your syllabus and lecture slides are the best indicator of what will appear on the exam.

When These Approaches Break Down

Standard epidemiology exam preparation assumes you have access to a basic calculator and that numerical answers are expected in a reasonable range. If your exam requires complex survival analysis or multivariable regression interpretation, the standard shortcuts stop working. You will need familiarity with statistical software output reading, particularly log-transformed coefficients and censoring adjustments. This material is more common in advanced graduate courses than in introductory undergraduate exams. If you are in that tier, spend more time interpreting hazard ratios and proportional hazards assumptions than memorizing screening test formulas. The foundational skills are still necessary, but the exam focus shifts. Another limitation worth noting: no amount of practice with standard questions prepares you well for exams that include novel research abstracts as the basis for questions. Some professors now write questions directly from recent publications. The underlying concepts are the same, but the context is unfamiliar. The workaround is to practice reading and critiquing abstracts from journals like Epidemiology, American Journal of Epidemiology, and International Journal of Epidemiology. Identify the study design, the population, the exposure, the outcome, the main measure of association, and the stated limitations. Doing this regularly builds the pattern recognition that novel questions depend on.