Sensitivity analysis and Markov modeling MCQs With Answer

Introduction

Sensitivity analysis and Markov modeling are central to pharmacoeconomic evaluations and long-term pharmacoepidemiology studies. This set of MCQs is designed for M.Pharm students to deepen understanding of model structure, transition probabilities, cycle length, half-cycle correction, cohort versus microsimulation, and interpretation of outputs like QALYs, ICERs and cost-effectiveness acceptability curves. It also covers deterministic and probabilistic sensitivity analyses, parameter distributions, tornado diagrams, and value-of-information concepts. These questions emphasize practical decision-making, model validation, and how uncertainty affects policy conclusions, preparing students to critically appraise and construct robust cost-effectiveness models in pharmaceutical research and regulatory contexts.

Q1. Which element is NOT a required component when constructing a Markov model for a chronic disease?

• Explicit health states representing mutually exclusive clinical conditions
• Transition probabilities between states for each cycle
• Immediate event sequencing for intra-cycle events without approximation
• A defined cycle length and time horizon

Correct Answer: Immediate event sequencing for intra-cycle events without approximation

Q2. What is the primary purpose of applying a half-cycle correction in a Markov cohort model?

• To change transition probabilities over time
• To adjust for costs and utilities accrued mid-cycle rather than at the cycle boundary
• To model rare instantaneous events more accurately
• To convert a cohort model into a microsimulation

Correct Answer: To adjust for costs and utilities accrued mid-cycle rather than at the cycle boundary

Q3. In probabilistic sensitivity analysis (PSA), which distribution is commonly used to model probabilities bounded between 0 and 1?

• Normal distribution
• Gamma distribution
• Beta distribution
• Uniform distribution over positive real numbers

Correct Answer: Beta distribution

Q4. A Markov model includes an absorbing state. Which statement best describes an absorbing state?

• A state that permits rapid cycling back to the initial state
• A state from which no transitions to other states are possible
• A state with time-varying transition probabilities only
• A transient state with temporary utility impact

Correct Answer: A state from which no transitions to other states are possible

Q5. Which sensitivity analysis technique systematically varies one parameter at a time to assess impact on model outcomes?

• Probabilistic sensitivity analysis
• Threshold analysis
• One-way deterministic sensitivity analysis
• Value-of-information analysis

Correct Answer: One-way deterministic sensitivity analysis

Q6. When performing PSA using Monte Carlo simulation in a Markov model, increasing the number of iterations primarily affects which aspect?

• The structural form of the transition matrix
• The computational need for half-cycle correction
• The precision (Monte Carlo error) of estimated mean costs and QALYs
• The choice of utility values for health states

Correct Answer: The precision (Monte Carlo error) of estimated mean costs and QALYs

Q7. In a Markov microsimulation model, what advantage does individual-level simulation have over cohort simulation?

• Microsimulation always requires fewer computations
• It allows capturing patient heterogeneity, history-dependent transitions and tunnel states
• It eliminates the need to define transition probabilities
• It guarantees deterministic results without stochastic variation

Correct Answer: It allows capturing patient heterogeneity, history-dependent transitions and tunnel states

Q8. Which graphical output is most appropriate to demonstrate the impact of multiple one-way sensitivity analyses on the ICER?

• Cost-effectiveness acceptability curve (CEAC)
• Tornado diagram
• Kaplan–Meier curve
• Transition probability matrix heatmap

Correct Answer: Tornado diagram

Q9. In the context of PSA, what does a cost-effectiveness acceptability curve (CEAC) show?

• The cumulative survival probability over time
• The probability that an intervention is cost-effective across a range of willingness-to-pay thresholds
• The distribution of transition probabilities used in the model
• The net benefit for individual patients over time

Correct Answer: The probability that an intervention is cost-effective across a range of willingness-to-pay thresholds

Q10. Which parameter distribution is most appropriate for modeling positive, right-skewed continuous costs in PSA?

• Normal distribution with mean zero
• Beta distribution
• Gamma distribution
• Binomial distribution

Correct Answer: Gamma distribution

Q11. What is the main goal of deterministic (scenario) sensitivity analysis in a Markov model?

• To explore uncertainty by assigning probability distributions to all parameters
• To identify whether model conclusions change under alternative plausible assumptions or scenarios
• To estimate the sampling distribution of the ICER directly
• To perform subgroup analysis using individual patient data

Correct Answer: To identify whether model conclusions change under alternative plausible assumptions or scenarios

Q12. Which concept describes the maximum value a decision-maker should be willing to pay for one additional QALY, often used to interpret ICERs?

• Discount rate
• Willingness-to-pay (WTP) threshold
• Half-cycle correction
• Transition probability

Correct Answer: Willingness-to-pay (WTP) threshold

Q13. In Markov modeling, what is a ‘tunnel state’ used to represent?

• A state with instantaneous death
• A series of temporary states that allow history-dependent effects without expanding the number of core health states
• An absorbing state with variable utilities
• A state representing aggregated costs only

Correct Answer: A series of temporary states that allow history-dependent effects without expanding the number of core health states

Q14. Which output quantifies the additional cost per additional unit of health effect gained (e.g., per QALY) when comparing two strategies?

• Net monetary benefit (NMB)
• Incremental cost-effectiveness ratio (ICER)
• Transition probability
• Discounted survival time

Correct Answer: Incremental cost-effectiveness ratio (ICER)

Q15. If two parameters are correlated in PSA, what is the best practice when sampling from their distributions?

• Sample each parameter independently from marginal distributions
• Ignore correlation because it has negligible effect on ICER
• Use joint sampling methods (e.g., copulas or multivariate distributions) to preserve correlation
• Fix one parameter at its mean and vary the other

Correct Answer: Use joint sampling methods (e.g., copulas or multivariate distributions) to preserve correlation

Q16. Which of the following is a key advantage of performing probabilistic sensitivity analysis over one-way deterministic analysis?

• PSA identifies which single parameter has the largest influence
• PSA provides a coherent estimate of joint parameter uncertainty and its impact on decision uncertainty
• PSA eliminates the need for structural sensitivity analysis
• PSA produces simpler graphical outputs than deterministic analyses

Correct Answer: PSA provides a coherent estimate of joint parameter uncertainty and its impact on decision uncertainty

Q17. In a Markov model evaluating a new drug, which approach helps estimate the value of reducing parameter uncertainty before making a funding decision?

• Deterministic one-way sensitivity analysis
• Half-cycle correction
• Expected value of perfect information (EVPI)
• Kaplan–Meier estimation

Correct Answer: Expected value of perfect information (EVPI)

Q18. What does the Markov trace represent in cohort modeling?

• The sequence of individual patient histories simulated over time
• The distribution (proportion) of the cohort across health states at each cycle
• The list of cost inputs used in each cycle
• The parametric survival function used to derive transitions

Correct Answer: The distribution (proportion) of the cohort across health states at each cycle

Q19. When converting survey-derived health-state utilities into parameters for PSA, which method preserves mean and variance and is appropriate for bounded utilities?

• Fit a normal distribution directly to untransformed utility scores
• Use a beta distribution fitted from the sample mean and variance after ensuring utilities are within 0–1
• Use a gamma distribution because utilities are always positive
• Assume a uniform distribution between 0 and 100

Correct Answer: Use a beta distribution fitted from the sample mean and variance after ensuring utilities are within 0–1

Q20. Which statement best explains threshold analysis in the context of sensitivity analysis?

• It identifies the probability distribution best fitting parameter data
• It determines the value of a parameter at which the preferred strategy changes (e.g., ICER equals WTP)
• It quantifies stochastic variability in microsimulations
• It assesses the long-term structural validity of model states

Correct Answer: It determines the value of a parameter at which the preferred strategy changes (e.g., ICER equals WTP)

Download