Does the PEBC exam test biostatistics and evidence-based medicine?

Yes. Study design, interpreting trial results, and core statistics like number needed to treat and confidence intervals are tested, because pharmacists are expected to critically evaluate the evidence behind a therapy. The exact mix is randomized from one sitting to the next.

What biostatistics topics should I focus on for the PEBC?

Focus on the study-design hierarchy, relative versus absolute risk and number needed to treat, the meaning of p-values and confidence intervals, the difference between statistical and clinical significance, and the common sources of bias and confounding.

Do I need to memorize statistical formulas for the PEBC?

Mostly no. The exam tests whether you can interpret results, not derive formulas. You should be able to calculate absolute risk reduction and number needed to treat from a simple two-by-two table, and explain what relative risk, p-values, and confidence intervals mean, but heavy formula memorization is not the point.

How should I study biostatistics for the PEBC?

Focus on understanding what each statistic means and practicing interpretation questions, rather than rote memorization.

PEBC Biostatistics and Evidence-Based Medicine Questions

Evidence-based medicine and biostatistics are high yield on the PEBC exam, and they reward understanding over memorization. Study design, interpreting trial results, and the core statistics like number needed to treat and confidence intervals all show up, because pharmacists are expected to evaluate evidence.

These free questions are drawn from across those areas. The reasoning covers every option, not just the right one, so the concept that settled the question is on the page.

Written and reviewed against current Canadian guidelines and references.

free questions here

no signup to start

All 4

options explained

full clinical case

How to approach a PEBC biostatistics question

1Know the study-design hierarchy and what each design can and cannot prove: a randomized controlled trial shows causation, while cohort and case-control studies show association.
2Be fluent in the core measures: relative risk, absolute risk reduction, and number needed to treat, and always notice whether a result is framed in relative or absolute terms.
3Know what a p-value and a confidence interval tell you, and when a result is statistically significant but not clinically meaningful.
4Watch for bias and confounding, and understand how randomization and blinding are meant to control them.
5These questions reward reasoning, so work from what each statistic means rather than trying to recall a formula.

Practice questions

Tap an answer to see why each option is right or wrong.

Question 1EBM & Study Design

A pharmacist is reviewing an oncology RCT that compared a new immunotherapy to standard chemotherapy for advanced lung cancer. The study reports a hazard ratio (HR) of 0.72 (95% CI 0.58-0.89, p = 0.003) for overall survival. Which of the following statements about this result is NOT correct?

Incorrect · Answer: D

PEBC Yield: Highevidence-appraisal

Tempting because HR comes from time-to-event analysis is close to the tested list. It is disqualified because it is aligned with the source rule, while the stem asks for the exception; HR comes from time-to-event analysis is related to the same topic area.

Tempting because CI excluding 1.0 supports significance is close to the tested list. It is disqualified because it is aligned with the source rule, while the stem asks for the exception; CI excluding 1.0 supports significance uses nearby biostatistics or trial-analysis language.

Tempting because HR below 1 favours the immunotherapy group is close to the tested list. It is disqualified because it is aligned with the source rule, while the stem asks for the exception; HR below 1 favours the immunotherapy group is related to the same topic area.

D) (Correct)

Correct. A hazard ratio comes from time-to-event analysis and is not simply the cumulative end-of-study risk ratio. The stem clue is: A pharmacist is reviewing an oncology RCT that compared a new immunotherapy to standard chemotherapy for advanced lung cancer. The study reports a hazard ratio (HR) of 0.... Key point: HR is an end-of-study cumulative risk ratio wins only when it matches that specific source rule, not merely because it is a related term.

Question 2EBM & Study Design

A pharmacist is evaluating an industry-sponsored RCT for a new lipid-lowering agent. She notes the trial was funded entirely by the drug manufacturer, and the lead investigators hold equity in the sponsoring company. Which of the following BEST describes the concern regarding conflicts of interest in clinical research?

Incorrect · Answer: C

PEBC Yield: Highevidence-appraisalethical-reasoning

Tempting because Detection bias is the main sponsorship concern is close to the topic. It is disqualified because the stem points to A conflict of interest occurs when a secondary interest, financial or non-financial, while Detection bias is the main sponsorship concern names a real appraisal concept.

Tempting because Non-financial interests are outside COI scope is close to the topic. It is disqualified because the stem points to A conflict of interest occurs when a secondary interest, financial or non-financial, while Non-financial interests are outside COI scope is related to the same topic area.

C) (Correct)

Correct. A conflict of interest occurs when a secondary interest, financial or non-financial, can influence professional judgment about the primary research interest. The stem clue is: A pharmacist is evaluating an industry-sponsored RCT for a new lipid-lowering agent. She notes the trial was funded entirely by the drug manufacturer.... Key point: Secondary interests can bias professional judgment wins only when it matches that specific source rule, not merely because it is a related term.

Tempting because Financial sponsorship confirms invalid results is close to the topic. It is disqualified because the stem points to A conflict of interest occurs when a secondary interest, financial or non-financial, while Financial sponsorship confirms invalid results is related to the same topic area.

Question 3EBM & Study Design

A pharmacist reads a drug information bulletin stating that a new antihypertensive lowers blood pressure by 5 mmHg in a large trial. A colleague argues this is evidence, while another says it is merely information. Which statement best distinguishes evidence from information?

Incorrect · Answer: B

PEBC Yield: Highevidence-appraisal

Tempting because Raw data before appraisal is close to the topic. It is disqualified because the stem points to In EBM, information is raw facts before judgment, while evidence is organized information judged to have meaning and to support a claim, while Raw data before appraisal is related to the same topic area.

B) (Correct)

Correct. In EBM, information is raw facts before judgment, while evidence is organized information judged to have meaning and to support a claim. The stem clue is: A pharmacist reads a drug information bulletin stating that a new antihypertensive lowers blood pressure by 5 mmHg in a large trial. A colleague argues this is evidence.... Key point: Organized data with recognized significance wins only when it matches that specific source rule, not merely because it is a related term.

Tempting because Isolated facts without interpretation is close to the topic. It is disqualified because the stem points to In EBM, information is raw facts before judgment, while evidence is organized information judged to have meaning and to support a claim, while Isolated facts without interpretation is related to the same topic area.

Tempting because Trial measurements before synthesis is close to the topic. It is disqualified because the stem points to In EBM, information is raw facts before judgment, while evidence is organized information judged to have meaning and to support a claim, while Trial measurements before synthesis uses a real evidence or study-design term.

Question 4EBM & Study Design

A pharmacist reads a study that surveyed 5,000 adults in a single city on one day in March 2025, measuring both statin use and the presence of diagnosed type 2 diabetes. The study found that statin users had a higher rate of diabetes than non-users. Which of the following conclusions can this study design validly support?

Incorrect · Answer: B

PEBC Yield: Highevidence-appraisal

Tempting because Incidence differs after statin initiation is close to the topic. It is disqualified because the stem points to A cross-sectional study measures exposure and outcome at one time point, so it can support prevalence associations but not incidence or causality, while Incidence differs after statin initiation is related to the same topic area.

B) (Correct)

Correct. A cross-sectional study measures exposure and outcome at one time point, so it can support prevalence associations but not incidence or causality. The stem clue is: A pharmacist reads a study that surveyed 5,000 adults in a single city on one day in March 2025, measuring both statin use and the presence of diagnosed type 2 diabetes.... Key point: Prevalence association is present wins only when it matches that specific source rule, not merely because it is a related term.

Tempting because Lifetime cumulative risk is increased is close to the topic. It is disqualified because the stem points to A cross-sectional study measures exposure and outcome at one time point, so it can support prevalence associations but not incidence or causality, while Lifetime cumulative risk is increased uses nearby biostatistics or trial-analysis language.

Tempting because Statin exposure causes diabetes is close to the topic. It is disqualified because the stem points to A cross-sectional study measures exposure and outcome at one time point, so it can support prevalence associations but not incidence or causality, while Statin exposure causes diabetes is related to the same topic area.

Question 5EBM & Study Design

A new rapid strep test is evaluated in 500 patients. The following results are obtained: Disease present: 90 tested positive, 10 tested negative Disease absent: 40 tested positive, 360 tested negative What is the positive predictive value (PPV) of this test?

Incorrect · Answer: A

PEBC Yield: Highevidence-appraisalcalculation

A) (Correct)

Correct. PPV = TP/(TP+FP) = 90/(90+40) = 90/130 = 0.692 or approximately 69%. PPV tells us the proportion of patients with a positive test result who truly have the disease. This is distinct from sensitivity, which uses the disease column rather than the positive test row.

This value (90/100) represents the sensitivity of the test, not the PPV. Sensitivity measures the proportion of truly diseased patients who test positive: TP/(TP+FN) = 90/(90+10) = 90%. PPV instead looks at how many of those who tested positive actually have the disease.

This value (360/450) does not correspond to any standard diagnostic test metric from the 2x2 table. It incorrectly divides true negatives by total disease-absent patients, which is actually the specificity denominator approach but with the wrong numerator arrangement.

This value (360/370) represents the negative predictive value (NPV), not the PPV. NPV = TN/(TN+FN) = 360/(360+10) = 97%. NPV tells us the proportion of patients with a negative test who are truly non-diseased.

Question 6EBM & Study Design

A clinical researcher wants to determine whether Drug X or Drug Y provides better blood pressure control. Each participant is randomly assigned to receive Drug X first, followed by a washout period, then Drug Y (or vice versa). Each participant's results on both drugs are compared. Which of the following is NOT an advantage of this study design compared to a parallel-group RCT?

Incorrect · Answer: C

PEBC Yield: Highevidence-appraisal

Incorrect. Fewer participants is a recognized advantage of crossover designs. Because each subject receives both treatments, the within-subject comparison increases statistical power with smaller sample sizes.

Incorrect. Eliminating between-subject variability is a direct consequence of each subject serving as their own control, which is a key advantage of the crossover design.

C) (Correct)

Correct. Crossover studies actually require a longer time commitment, not shorter, because participants must complete both treatment periods with an adequate washout period between them. This extended duration is a recognized disadvantage that can increase dropout rates.

Incorrect. Having each subject act as their own control is the defining advantage of a crossover study. This eliminates confounding from individual patient differences.

Question 7EBM & Study Design

A pharmacist is evaluating an evidence hierarchy diagram and notes that each tier occupies a different width on the pyramid. Which relationship does the horizontal axis of the evidence hierarchy pyramid represent?

Incorrect · Answer: C

PEBC Yield: Highevidence-appraisal

Quality of evidence is represented by the vertical axis of the pyramid, not the horizontal axis. Higher tiers have higher quality but fewer available studies.

Clinical applicability is an important concept in evidence appraisal (external validity), but it is not depicted on either axis of the standard evidence hierarchy pyramid.

C) (Correct)

Correct -- the horizontal axis of the evidence hierarchy pyramid represents information volume. The base is widest because case reports, expert opinions, and background information are the most abundant, while meta-analyses at the top are fewer in number.

Research cost is not a dimension represented on the evidence hierarchy pyramid. The two axes are quality (vertical) and volume (horizontal).

Question 8EBM & Study Design

A pharmacist is reviewing the results of a randomized controlled trial comparing a new antibiotic to standard therapy for community-acquired pneumonia. The trial enrolled 800 patients, but 120 patients in the intervention arm discontinued therapy early due to gastrointestinal side effects. The published analysis excluded these 120 patients entirely. Which type of analysis was used in this trial?

Incorrect · Answer: A

PEBC Yield: Highevidence-appraisal

A) (Correct)

Correct. Per-protocol analysis includes only patients who completed the trial according to the protocol. By excluding the 120 patients who discontinued early, the researchers performed a per-protocol analysis. This approach can overestimate treatment effectiveness because it removes patients who may have dropped out due to treatment-related issues.

Modified ITT typically excludes patients who never received any study drug but still retains those who started and later discontinued. The scenario describes exclusion of patients who started but did not complete therapy, which fits per-protocol analysis.

Intention-to-treat analysis includes ALL randomized patients regardless of adherence, dropout, or protocol violations. Since the 120 patients who discontinued were excluded, this is not ITT.

As-treated analysis groups patients based on the treatment they actually received, regardless of initial randomization. It does not simply exclude non-completers but rather reassigns them based on actual treatment received.

Question 9EBM & Study Design

A pharmacist is reviewing a double-blind RCT that evaluated a new oral anticoagulant. She finds that the intervention group had a 22% dropout rate while the control group had a 6% dropout rate over 12 months. The investigators only reported per-protocol results. Which type of bias is MOST concerning in this study?

Incorrect · Answer: A

PEBC Yield: Highevidence-appraisal

A) (Correct)

Correct. Attrition bias occurs when there is a difference between groups due to dropout rates. A 22% vs 6% dropout rate represents a substantial imbalance that can distort results, especially when only per-protocol analysis is reported. ITT analysis should also be presented to minimize this bias.

Publication bias refers to the tendency for studies with positive findings to be published more readily than those with negative findings. This is a bias in the literature, not within a single study's conduct.

Performance bias involves systematic differences in care provided based on group allocation. While unequal care could contribute to dropouts, the double-blind design mitigates performance bias. The primary concern here is the large imbalance in dropout rates between groups.

Volunteer bias refers to the concept that individuals who volunteer for a study are inherently different from those who would not. This occurs at enrollment, not during the study when participants drop out.

Question 10EBM & Study Design

A pharmacist is critically appraising a randomized controlled trial comparing two antihypertensives. She notices that the computer-generated sequence used to assign participants to groups was predictable, and the research coordinator was able to anticipate allocations before they occurred. Which type of bias is MOST likely present in this study?

Incorrect · Answer: C

PEBC Yield: Highevidence-appraisal

Selection bias occurs when the subjects studied are not representative of the target population. While a flawed allocation process can contribute to selection bias, the primary issue here is that the randomization process itself was not truly random, which defines randomization bias specifically.

Performance bias refers to systematic differences in care provided to participants depending on their allocation, such as unblinded clinicians treating groups differently. The scenario describes a flaw in the assignment process, not in the care delivered.

C) (Correct)

Correct. Randomization bias occurs when the process for randomization is not completely random. A predictable sequence that allows anticipation of allocations violates the principle of allocation concealment and constitutes a direct failure of the randomization process.

Detection bias occurs when groups are assessed for outcomes differently. This involves measurement of outcomes, not the allocation of participants to study arms.

Want the full bank?

These are a sample. The full bank adds full-length cases and timed mock exams.

Sample full case

One patient with several linked questions, the format the real exam uses for case clusters.

A pharmacy resident is preparing to present a recently published cardiovascular outcomes trial at hospital journal club. The trial enrolled 4,012 adults with symptomatic heart failure with reduced ejection fraction (NYHA II-IV, LVEF <=40%) on guideline-directed medical therapy, and allocated them 1:1 to a new SGLT2 inhibitor 10 mg once daily or matched placebo using a centralized computer-generated sequence with concealed allocation. Participants, treating clinicians, outcome assessors, and the statistical team were masked to assignment until database lock. Median follow-up was 24 months. The pre-specified primary outcome was a composite of cardiovascular death, hospitalization for heart failure, or an urgent heart failure visit requiring intravenous diuretic therapy. Secondary outcomes included each component analyzed separately, all-cause mortality, and KCCQ quality-of-life score. Analysis was by intention-to-treat. The trial was funded by the manufacturer, with an independent academic steering committee and data safety monitoring board. Exclusion criteria included eGFR <30 mL/min/1.73 m^2, type 1 diabetes, recent acute coronary syndrome within 30 days, and systolic blood pressure <95 mmHg. The primary result was a statistically significant reduction in the composite endpoint (HR 0.76, 95% CI 0.65-0.89, p<0.001).

Question 1EBM & Study Design

A pharmacy resident is leading journal club on the trial described above. The senior preceptor opens by asking the group to classify the study design before any results are discussed. Which of the following best describes the design of this trial?

Incorrect · Answer: A

PEBC Yield: Highstudy-design-identificationrctevidence-hierarchyjournal-club-appraisalinternal-validity

A) (Correct)

Randomization to intervention or placebo with a pre-specified prospective outcome defines an experimental RCT, the design capable of establishing causality. Canadian EBM teaching and the evidence hierarchy place RCTs immediately below systematic reviews/meta-analyses because allocation concealment plus blinding control both selection and detection bias. The described centralized 1:1 computer-generated sequence, masking of participants, clinicians, assessors and statisticians, and ITT analysis are the hallmark RCT features. Would be incorrect if participants had not been randomly allocated to intervention versus comparator or if the investigators only observed an existing exposure.

Cohort studies compare exposed versus unexposed groups who self-select or are selected by clinical circumstance, without investigator-assigned intervention. The trial here actively assigns the SGLT2 inhibitor versus placebo via randomization, which moves it up the hierarchy to an RCT rather than an observational cohort. Would be appropriate if the investigators had simply followed patients already taking an SGLT2 inhibitor in clinical practice and compared them to non-users.

Cross-sectional designs measure exposure and outcome at a single point in time and estimate prevalence, with no follow-up and no intervention. The 24-month follow-up, allocated intervention, and time-to-event primary endpoint all exclude this design. Would be appropriate if the study had surveyed heart failure patients on one clinic day to estimate the proportion currently prescribed SGLT2 inhibitors.

Case-control studies start with cases who have the outcome and controls who do not, then look backward for differential exposure, and are typically retrospective. This trial instead starts with an exposure assignment and follows forward for outcomes, which is the opposite temporal direction. Would be appropriate if the design had identified patients who had already died of cardiovascular causes and compared their prior SGLT2 inhibitor exposure to matched survivors.

Question 2EBM & Study Design

During the discussion, a junior learner asks why the investigators chose a three-part composite primary endpoint rather than cardiovascular mortality alone. The resident is preparing the response. Which of the following is the BEST rationale for the use of a composite primary endpoint in this trial?

Incorrect · Answer: D

PEBC Yield: Highcomposite-endpointstatistical-powertrial-design-rationalesample-sizejournal-club-appraisal

Composite endpoints typically make interpretation harder, not easier, because clinicians must consider whether the treatment effect is uniform across components or driven by a single less severe component. The composite yields one p-value but obscures the clinical meaning of the result. Would be appropriate if the composite contained only clinically equivalent events with similar patient impact, which is rarely true in practice.

A statistically significant composite result does not require any single component to reach significance on its own; frequently the composite is driven by the most common and softest component while mortality remains non-significant. This is in fact a well-known limitation of composites rather than a rationale for using them. Would be appropriate if the question asked about a weakness or interpretive pitfall of composites rather than a rationale for their use.

No Canadian or FDA regulation requires a composite primary endpoint for heart failure drug approval; single endpoints such as cardiovascular mortality or heart failure hospitalization are acceptable when adequately powered. This distractor is tempting because many landmark cardiovascular trials happen to use composites, but that reflects statistical practicality rather than a regulatory mandate. Would be correct if a specific regulatory guidance document required the exact composite used, which is not the case for this indication.

D) (Correct)

Combining several related clinically meaningful events raises the pooled event rate, which increases statistical power for a fixed sample size or reduces the sample size and duration needed to detect a given effect. This is the standard EBM rationale taught for composite endpoints and is why they are common in cardiovascular outcomes trials where hard mortality events alone would require very large and long studies. It also preserves a single pre-specified primary test, protecting the alpha level from multiple-comparisons inflation. Would be incorrect if the components were not of similar clinical importance and direction, in which case a composite can dilute rather than enhance the signal.

Question 3EBM & Study Design

The resident presents the component breakdown: HF hospitalization HR 0.67 (95% CI 0.55-0.82), urgent HF visit HR 0.71 (95% CI 0.52-0.96), and cardiovascular death HR 0.94 (95% CI 0.76-1.16). All-cause mortality was nonsignificant. The preceptor asks how the pharmacy team should interpret the significant primary composite result for a patient counselling discussion. Which interpretation is most appropriate?

Incorrect · Answer: C

PEBC Yield: Highcomposite-endpoint-interpretationcomponent-analysiscritical-appraisalclinical-significancejournal-club-appraisal

A non-significant mortality component does not invalidate a trial whose pre-specified primary outcome was a composite and which met that primary endpoint with adequate power and clear benefit on symptomatic events. Invalidation would require internal validity problems, not a component not reaching significance. Would be appropriate if the trial had declared mortality as its primary outcome and failed to reach it.

This is the classic composite-endpoint trap: a significant overall composite does not mean every component is individually significant, especially when one component (HF hospitalization) has a much higher event rate and drives the pooled result. Reading the component HRs here shows CV death and all-cause mortality are non-significant. Would be appropriate if each component's confidence interval excluded 1 independently, which is not the case here.

C) (Correct)

A significant composite reflects the weighted sum of its components; inspecting the forest of component hazard ratios and confidence intervals shows whether the effect is uniform or driven by specific parts. Canadian critical appraisal teaching emphasizes that a composite should never be assumed to imply benefit on each component, and the clinically most important component (mortality) should be examined explicitly. Here the confidence intervals for CV death and all-cause mortality cross 1, so the honest counselling message is a reduction in heart failure events without proven mortality benefit. Would be incorrect if the mortality component's confidence interval were entirely below 1, in which case a mortality claim would be supported.

Composite results do not imply equal relative effects across components; the component HRs here clearly differ (0.67 for HF hospitalization versus 0.94 for CV death). Assuming uniform benefit misrepresents the evidence to patients and prescribers. Would be appropriate only if the components showed near-identical hazard ratios and overlapping confidence intervals, which is not the case.

Full-length cases like this, across every PEBC domain, live in the question bank.

Frequently asked questions

Does the PEBC exam test biostatistics and evidence-based medicine?: Yes. Study design, interpreting trial results, and core statistics like number needed to treat and confidence intervals are tested, because pharmacists are expected to critically evaluate the evidence behind a therapy. The exact mix is randomized from one sitting to the next.
What biostatistics topics should I focus on for the PEBC?: Focus on the study-design hierarchy, relative versus absolute risk and number needed to treat, the meaning of p-values and confidence intervals, the difference between statistical and clinical significance, and the common sources of bias and confounding.
Do I need to memorize statistical formulas for the PEBC?: Mostly no. The exam tests whether you can interpret results, not derive formulas. You should be able to calculate absolute risk reduction and number needed to treat from a simple two-by-two table, and explain what relative risk, p-values, and confidence intervals mean, but heavy formula memorization is not the point.
How should I study biostatistics for the PEBC?: Focus on understanding what each statistic means and practicing interpretation questions, rather than rote memorization.