FPGEE Biostatistics 5 — Questions and Answers
Question 1: In a pharmacoeconomic study, incremental cost-effectiveness ratio (ICER) is calculated as $80,000 per QALY. Using the commonly cited US willingness-to-pay threshold, how should this be classified?
- Highly cost-effective
- Borderline cost-effective (Correct answer)
- Not cost-effective
- Dominant strategy
Correct answer: Borderline cost-effective
The conventional US WTP threshold is approximately $50,000–$150,000 per QALY; $80,000/QALY falls within this range, making it borderline or acceptable.
Question 2: In a cohort study, person-years are used as the denominator for incidence rates. Why is this approach used?
- It accounts for varying lengths of follow-up among participants (Correct answer)
- It increases statistical power
- It corrects for confounding variables
- It normalizes for differences in baseline risk
Correct answer: It accounts for varying lengths of follow-up among participants
Person-years of observation account for the fact that participants may enter or exit the study at different times, ensuring the denominator reflects actual time at risk.
Question 3: A publication reports a forest plot in which most individual study confidence intervals overlap zero, but the pooled diamond does not touch zero. What is the most accurate interpretation?
- The meta-analysis result should be ignored due to heterogeneity
- The pooled analysis has sufficient power to detect a statistically significant overall effect (Correct answer)
- Each individual study confirms the treatment works
- The funnel plot shows publication bias
Correct answer: The pooled analysis has sufficient power to detect a statistically significant overall effect
Pooling studies in meta-analysis increases effective sample size, giving the pooled estimate greater power to detect effects that individual underpowered studies could not.
Question 4: A researcher wants to assess agreement between two pharmacists measuring pill-cutting accuracy using a continuous scale. Which statistical method is most appropriate?
- Pearson correlation coefficient
- Bland-Altman plot and limits of agreement (Correct answer)
- Kappa coefficient
- Intraclass correlation coefficient only
Correct answer: Bland-Altman plot and limits of agreement
The Bland-Altman method assesses agreement between two measurement methods by plotting the difference against the mean, revealing systematic bias and variability.
Question 5: Cox proportional hazards regression produces a hazard ratio (HR) of 0.60 for a new antifungal vs. placebo (95% CI: 0.40–0.89). How should this be interpreted?
- The new drug increases the event hazard by 40%
- The new drug reduces the instantaneous risk of the event by 40% at any given time (Correct answer)
- Patients on the new drug survive 60% longer on average
- The result is not statistically significant
Correct answer: The new drug reduces the instantaneous risk of the event by 40% at any given time
HR < 1 means the treatment group has a lower instantaneous hazard; HR = 0.60 means a 40% reduction in the hazard of the event at any point in time.
Question 6: Which study design provides the strongest evidence for establishing causality between a drug exposure and an outcome?
- Prospective cohort study
- Randomized controlled trial (Correct answer)
- Systematic review of case-control studies
- Well-designed cross-sectional survey
Correct answer: Randomized controlled trial
The RCT, through random allocation, minimizes confounding and allows for direct causal inference — it sits at the top of the hierarchy for intervention evidence.
Question 7: In logistic regression used to predict a binary outcome (adverse drug reaction: yes/no), the model output is best expressed as:
- A regression coefficient (β) directly interpreted as probability
- An odds ratio derived from the exponentiated regression coefficient (e^β) (Correct answer)
- A hazard ratio comparing event times
- A correlation coefficient between predictors and the outcome
Correct answer: An odds ratio derived from the exponentiated regression coefficient (e^β)
In logistic regression, exponentiating the β coefficient (e^β) yields the odds ratio for each one-unit increase in the predictor variable.
FPGEE Biostatistics 5
In a pharmacoeconomic study, incremental cost-effectiveness ratio (ICER) is calculated as $80,000 per QALY.
Using the commonly cited US willingness-to-pay threshold, how should this be classified?