Biostatistics Practice Questions

Q: A research study comparing two groups shows a statistically significant difference (p=0.04) but the confidence interval is very wide (0.5 to 4.2). How should this result be interpreted?

The result is statistically significant but imprecise. ***The result is statistically significant but imprecise*** - A **p-value of 0.04** indicates **statistical significance** at the conventional ɑ=0.05 level, meaning the observed difference is unlikely due to chance. - A **wide confidence interval (0.5 to 4.2)** suggests that while the true effect is likely positive, its magnitude is **highly uncertain** or **imprecise**, possibly due to a small sample size. *Additional statistical tests are needed* - While more analysis is often valuable, the current statistical outputs (p-value and CI) are sufficient to interpret the **significance** and **precision** of the original finding. - Simply adding more tests without addressing the underlying **imprecision** (e.g., through larger studies) would not fundamentally change the interpretation of the current result. *The result shows no meaningful difference between groups* - The **p-value of 0.04** suggests there *is* a statistically significant difference, so stating there's "no meaningful difference" is incorrect based on the statistical evidence. - However, the **precision of the effect** with the large confidence interval means the *clinical meaningfulness* is still very uncertain, as the effect could be anywhere from small to quite large. *The result is both statistically and clinically significant* - The result is **statistically significant** (p=0.04). - However, the **wide confidence interval (0.5 to 4.2)** means the **clinical significance** is unknown, as the true effect size could be small (0.5) or large (4.2); it is not definitively "clinically significant."

Q: A randomized controlled trial of a new diabetes medication is stopped early when interim analysis shows significant benefit. The trial enrolled only 60% of the planned participants. What is the most important consideration about this early termination?

The result may overestimate the true treatment effect. ***The result may overestimate the true treatment effect*** - Early termination of a trial due to a strong interim signal of benefit often leads to an **overestimation of the treatment effect** observed in the final results. This phenomenon is known as **"stopping early for benefit bias"** or **"truncation bias"**. - This occurs because trials are stopped when random fluctuations happen to show a particularly large effect, and if the trial continued, the effect might regress to the mean. - This is the **most important methodological concern** when trials are terminated early based on interim analyses. *Early termination is always appropriate when benefit is shown* - While showing benefit is a common reason for early termination, it is not "always appropriate" without careful consideration of the **magnitude of benefit**, **potential harms**, and the impact on **statistical inference**. - **Ethical considerations** to prevent exposing more patients to an inferior treatment must be balanced against the **scientific rigor** of completing the study as planned. *Early termination violates research protocols* - **Research protocols** often include pre-specified rules for early termination based on interim analyses, such as a **stopping boundary** for efficacy or harm (e.g., O'Brien-Fleming boundaries). - If a protocol includes such rules, termination is within the protocol, not a violation. However, if no such rules are in place and termination is ad hoc, it could be problematic. *Statistical significance justifies stopping the trial* - While **statistical significance** is necessary for early stopping, it is not the sole justification. - **Clinical significance**, the magnitude of the effect, ethical considerations, and the potential for **overestimation bias** are also crucial factors in the decision to stop early.

Q: A screening test for a rare disease has 99% sensitivity and 95% specificity. In a population where the disease prevalence is 0.1%, what is the most important consideration for implementing this test?

Low prevalence will result in many false positives. ***Low prevalence will result in many false positives*** - In populations with **low disease prevalence**, even a test with high specificity can yield a large number of **false positives** because the vast majority of tested individuals do not have the condition. - For example, with a 95% specificity and 0.1% prevalence, 5% of healthy individuals will test positive, which is significantly more than the true positives from a 0.1% prevalence. - The **positive predictive value (PPV)** will be very low (~1.9%), meaning most positive results will be false positives. *High specificity ensures accurate results* - While high specificity (95%) means a low rate of false positives among healthy individuals, in a **rare disease** scenario, the absolute number of healthy people is very large, leading to many healthy individuals being incorrectly flagged. - This high number of false positives can lead to unnecessary follow-up procedures, anxiety, and resource strain, undermining the screening program's effectiveness. *High sensitivity makes it excellent for screening* - **High sensitivity** (99%) ensures that most true cases are identified, which is crucial for a good screening test. - However, in a low prevalence setting, high sensitivity alone does not address the issue of **false positives** among the vast majority of healthy individuals, which can overwhelm the system and cause patient harm. *The test performance is ideal for any population* - The performance of a screening test is highly dependent on the **prevalence of the disease** in the target population, especially the predictive values (PPV and NPV). - A test that performs well in a high-prevalence population may be unsuitable for screening a low-prevalence population due to a high rate of **false positives** and low PPV, making it impractical for population-wide screening.

Q: A meta-analysis of 10 studies shows significant benefit from a new treatment (p=0.001), but there is high heterogeneity between studies (I² = 78%). Individual studies show conflicting results. How should this evidence be interpreted?

Inconclusive evidence requiring additional studies. ***Inconclusive evidence requiring additional studies*** - High **heterogeneity (I² = 78%)** indicates that the studies are measuring different effects, making a combined statistical result unreliable even with a low p-value. - **Conflicting individual study results** despite a statistically significant meta-analysis result means that the overall conclusion might be misleading or only applicable to a specific, unidentifiable subgroup. *Evidence supports treatment in all patient populations* - **High heterogeneity** suggests that treatment effects vary significantly between studies, making it inappropriate to universally apply the findings to all patient populations. - The presence of **conflicting individual results** contradicts the idea of a universal benefit across all populations. *Statistical significance overrides heterogeneity concerns* - While a **low p-value (p=0.001)** indicates statistical significance for the overall effect, **high heterogeneity (I² = 78%)** fundamentally questions the validity of pooling such diverse results. - Ignoring significant heterogeneity can lead to **misleading conclusions** about the consistency and generalizability of the treatment effect. *Treatment harmful due to conflicting individual results* - While conflicting results do raise concerns about the consistency of the treatment effect, they do not automatically imply that the treatment is **harmful**. - The data merely indicates **inconsistency** rather than a detrimental effect, suggesting a need for further investigation to understand the variability.

Q: A clinical trial shows that a new cardiac medication reduces mortality from 20% to 15%. The relative risk is 0.75 and the number needed to treat is 20. How should the clinical significance be interpreted?

Modest benefit requiring individual risk-benefit assessment. ***Modest benefit requiring individual risk-benefit assessment*** - A reduction in mortality from 20% to 15% represents an absolute risk reduction of 5% (20% - 15% = 5%). This corresponds to a **Number Needed to Treat (NNT)** of 20 (1/0.05 = 20), meaning 20 patients must be treated to prevent one additional death. - While any reduction in mortality is beneficial, an NNT of 20 suggests a **modest benefit** rather than a dramatic one, necessitating an assessment of potential side effects, cost, and patient preferences to determine the overall utility. *Not clinically meaningful despite statistical significance* - A 5% absolute reduction in mortality and an NNT of 20 for a life-threatening condition like heart disease is generally considered **clinically meaningful**, as it directly impacts patient survival. - The implication of "statistical significance" without clinical meaning usually applies when the effect size is very small, which is not the case here given the mortality outcome. *Highly significant due to mortality reduction* - Although the medication reduces mortality, a **5% absolute risk reduction** and an **NNT of 20** are not typically considered "highly significant" in the context of immediately revolutionizing clinical practice. - "Highly significant" would imply a much larger reduction in adverse outcomes or a much lower NNT (e.g., NNT of 2-5). *Excellent result warranting immediate adoption* - An "excellent result" would imply a more substantial impact (e.g., a significantly lower NNT or a much larger absolute risk reduction) that clearly outweighs potential harms and costs for most patients. - The need to treat 20 patients to save one life suggests that while beneficial, the medication may not be universally suitable without considering individual patient factors and potential side effects in comparison to its benefits.

Q: A surgeon notices that their complication rates are higher than expected. Hospital administration suggests this might be due to taking more complex cases. Analyze the statistical concept that explains this phenomenon.

Confounding by case complexity. ***Confounding by case complexity*** - The higher complication rates are explained by the increased inherent risk associated with **more complex cases**, which the surgeon is undertaking. The complexity of cases is acting as a **confounding variable**, influencing both the surgeon's choice of cases and the complication rate. - To accurately assess the surgeon's performance, the analysis must account for the baseline risk disparities between patient groups, rather than simply comparing crude complication rates. *Observer bias in outcome assessment* - This bias occurs when the **observation or recording of outcomes** is systematically influenced by the observer's expectations or knowledge, not by the true characteristics of the cases themselves. - In this scenario, the issue is not how complications are observed, but rather the underlying patient population driving the observed rates. *Selection bias in case assignment* - Selection bias occurs when there are **systematic differences** between participant groups in a study that can distort the results. While the surgeon is "selecting" more complex cases, the term "selection bias" in statistical research typically refers to issues in forming comparative groups (e.g., in clinical trials). - Here, the issue is not a flaw in study design for comparison, but rather a characteristic of clinical practice where the surgeon's case mix inherently impacts outcomes. *Regression to the mean* - This statistical phenomenon describes the tendency for **extreme measurements** to be followed by less extreme measurements closer to the average. For example, an exceptionally high or low score is likely to be closer to the average on subsequent measurements. - While it explains why an unusually high measurement might decrease over time, it does not explain why consistently taking on more complex cases would lead to higher complication rates as a sustained trend.

Question 1

A research study comparing two groups shows a statistically significant difference (p=0.04) but the confidence interval is very wide (0.5 to 4.2). How should this result be interpreted?

Accepted Answer

The result is statistically significant but imprecise

Answer

Additional statistical tests are needed

Answer

The result shows no meaningful difference between groups

Answer

The result is both statistically and clinically significant

Question 2

A randomized controlled trial of a new diabetes medication is stopped early when interim analysis shows significant benefit. The trial enrolled only 60% of the planned participants. What is the most important consideration about this early termination?

Accepted Answer

The result may overestimate the true treatment effect

Answer

Early termination is always appropriate when benefit is shown

Answer

Early termination violates research protocols

Answer

Statistical significance justifies stopping the trial

Question 3

A depression screening tool developed in specialty clinics performs poorly when used in primary care settings. The positive predictive value drops from 85% to 25%. What is the most likely explanation?

Accepted Answer

Disease prevalence is lower in primary care than specialty clinics

Answer

The tool is fundamentally flawed

Answer

Patients in primary care are less honest about symptoms

Answer

Primary care physicians are using the tool incorrectly

Question 4

A cohort study follows 5,000 nurses for 20 years to study night shift work and breast cancer risk. After 10 years, 30% of participants have been lost to follow-up. What is the most significant threat to study validity?

Accepted Answer

Loss to follow-up may introduce bias if related to exposure or outcome

Answer

Twenty years is too long for meaningful results

Answer

Nurses are not representative of the general population

Answer

The study should be converted to a case-control design

Question 5

A screening test for a rare disease has 99% sensitivity and 95% specificity. In a population where the disease prevalence is 0.1%, what is the most important consideration for implementing this test?

Accepted Answer

Low prevalence will result in many false positives

Answer

High specificity ensures accurate results

Answer

High sensitivity makes it excellent for screening

Answer

The test performance is ideal for any population

Question 6

A meta-analysis of 10 studies shows significant benefit from a new treatment (p=0.001), but there is high heterogeneity between studies (I² = 78%). Individual studies show conflicting results. How should this evidence be interpreted?

Accepted Answer

Inconclusive evidence requiring additional studies

Answer

Evidence supports treatment in all patient populations

Answer

Statistical significance overrides heterogeneity concerns

Answer

Treatment harmful due to conflicting individual results

Question 7

A clinical trial shows that a new cardiac medication reduces mortality from 20% to 15%. The relative risk is 0.75 and the number needed to treat is 20. How should the clinical significance be interpreted?

Accepted Answer

Modest benefit requiring individual risk-benefit assessment

Answer

Not clinically meaningful despite statistical significance

Answer

Highly significant due to mortality reduction

Answer

Excellent result warranting immediate adoption

Question 8

A researcher wants to study a new antidepressant in patients with severe depression. Standard antidepressants are available and effective. The pharmaceutical company insists on a placebo-controlled design, arguing it provides the clearest evidence of efficacy. What is the most appropriate study design?

Accepted Answer

Active-controlled trial comparing to standard antidepressant

Answer

Placebo-controlled trial as requested by the company

Answer

Crossover design with placebo and active drug periods

Answer

Observational study without intervention

Question 9

A screening program implements two tests in parallel (positive if either test is positive) for a disease. Test A has 80% sensitivity and 90% specificity. Test B has 70% sensitivity and 95% specificity. Analyze the expected performance of the combined approach.

Accepted Answer

Combined sensitivity will be higher but specificity will be lower

Answer

Both sensitivity and specificity will be higher

Answer

Combined sensitivity will be lower than either individual test

Answer

Combined specificity will be higher but sensitivity will be lower

Question 10

A surgeon notices that their complication rates are higher than expected. Hospital administration suggests this might be due to taking more complex cases. Analyze the statistical concept that explains this phenomenon.

Accepted Answer

Confounding by case complexity

Answer

Observer bias in outcome assessment

Answer

Selection bias in case assignment

Answer

Regression to the mean

Biostatistics — MCQs

Biostatistics — MCQs

On this page

Practice by Chapter

Want unlimited practice?