Welcome, PharmD students, to this insightful MCQ quiz on Decision Analysis in healthcare! When faced with complex choices between different health interventions, especially under conditions of uncertainty, decision analysis provides a structured and quantitative approach to compare alternatives. This quiz will test your understanding of the components of decision trees and Markov models, how probabilities and outcomes are incorporated, the calculation of expected values, and the critical role of sensitivity analysis. We’ll explore how these modeling techniques are used in pharmacoeconomics to inform policy and clinical decisions. Let’s begin!
1. Decision analysis in healthcare is best defined as a systematic approach to:
- a) Diagnosing complex diseases.
- b) Making choices under conditions of uncertainty by explicitly structuring the problem, probabilities, and outcomes.
- c) Only calculating drug dosages.
- d) Conducting clinical trials.
Answer: b) Making choices under conditions of uncertainty by explicitly structuring the problem, probabilities, and outcomes.
2. In a decision tree, a square node typically represents a(n):
- a) Chance event
- b) Decision node (a point where a choice is made)
- c) Terminal outcome
- d) Probability
Answer: b) Decision node (a point where a choice is made)
3. A circular node in a decision tree typically represents a(n):
- a) Decision point
- b) Chance node (an uncertain event with associated probabilities)
- c) Final outcome value
- d) Starting point of the tree
Answer: b) Chance node (an uncertain event with associated probabilities)
4. The branches emanating from a chance node in a decision tree must have probabilities that sum to:
- a) 0
- b) 0.5
- c) 1.0 (or 100%)
- d) The expected value of the node.
Answer: c) 1.0 (or 100%)
5. Terminal nodes in a decision tree represent:
- a) The initial decision to be made.
- b) Uncertain events.
- c) The final outcomes of a particular pathway, to which values (e.g., costs, QALYs) are assigned.
- d) The point where probabilities are assigned.
Answer: c) The final outcomes of a particular pathway, to which values (e.g., costs, QALYs) are assigned.
6. The process of “folding back” or “averaging out and folding back” a decision tree is used to calculate the:
- a) Number of branches in the tree.
- b) Expected value (e.g., expected cost, expected QALYs) for each decision alternative.
- c) Sensitivity of the model to parameter changes.
- d) Time horizon of the analysis.
Answer: b) Expected value (e.g., expected cost, expected QALYs) for each decision alternative.
7. Markov models are particularly useful in decision analysis for modeling:
- a) Single, acute events with no long-term consequences.
- b) Chronic diseases or situations where patients transition between different health states over multiple time cycles.
- c) Only diagnostic test accuracy.
- d) The cost of medications only.
Answer: b) Chronic diseases or situations where patients transition between different health states over multiple time cycles.
8. Key components of a Markov model include defined health states, cycle length, time horizon, and:
- a) Decision nodes only.
- b) Only terminal outcomes.
- c) Transition probabilities between health states for each cycle.
- d) A single chance event.
Answer: c) Transition probabilities between health states for each cycle.
9. An “absorbing state” in a Markov model is a health state that, once entered:
- a) Can transition to any other state.
- b) Cannot be exited (e.g., death).
- c) Has a utility value of 1.0.
- d) Is only temporary.
Answer: b) Cannot be exited (e.g., death).
10. The “time horizon” in a decision analysis or Markov model refers to the:
- a) Duration of each cycle in a Markov model.
- b) Total period over which costs and outcomes are evaluated in the model.
- c) Time it takes to build the model.
- d) Point at which all patients are cured.
Answer: b) Total period over which costs and outcomes are evaluated in the model.
11. Decision analysis is often used within which types of pharmacoeconomic evaluations to estimate long-term costs and consequences?
- a) Only Cost-Minimization Analysis.
- b) Cost-Effectiveness Analysis (CEA) and Cost-Utility Analysis (CUA).
- c) Only Cost-Benefit Analysis.
- d) Only Cost-of-Illness studies.
Answer: b) Cost-Effectiveness Analysis (CEA) and Cost-Utility Analysis (CUA).
12. Data for probabilities and outcome values used in decision models are ideally derived from:
- a) The modeler’s personal opinion only.
- b) High-quality sources such as systematic reviews of clinical trials, meta-analyses, large observational studies, or well-conducted preference studies.
- c) Pharmaceutical company advertisements.
- d) A single small case series.
Answer: b) High-quality sources such as systematic reviews of clinical trials, meta-analyses, large observational studies, or well-conducted preference studies.
13. An advantage of using decision analysis in healthcare is its ability to:
- a) Eliminate all uncertainty in decision-making.
- b) Make the assumptions and logic of a decision explicit and transparent.
- c) Always provide a single, definitive answer.
- d) Replace the need for clinical judgment.
Answer: b) Make the assumptions and logic of a decision explicit and transparent.
14. A significant limitation of decision analysis is that:
- a) It cannot incorporate probabilities.
- b) The results are highly dependent on the model structure, data inputs, and underlying assumptions, which may be uncertain or simplified.
- c) It can only be used for acute diseases.
- d) It does not allow for sensitivity analysis.
Answer: b) The results are highly dependent on the model structure, data inputs, and underlying assumptions, which may be uncertain or simplified.
15. “One-way sensitivity analysis” performed on a decision model involves:
- a) Simultaneously varying all parameters in the model.
- b) Varying one input parameter at a time across a plausible range to examine its impact on the model’s results.
- c) Only considering the best-case scenario.
- d) Changing the structure of the decision tree.
Answer: b) Varying one input parameter at a time across a plausible range to examine its impact on the model’s results.
16. The output of a one-way sensitivity analysis is often presented graphically as a:
- a) Cost-effectiveness plane.
- b) Kaplan-Meier curve.
- c) Tornado diagram.
- d) Receiver Operating Characteristic (ROC) curve.
Answer: c) Tornado diagram.
17. Probabilistic Sensitivity Analysis (PSA) in decision modeling involves:
- a) Assigning fixed values to all parameters.
- b) Assigning probability distributions to uncertain parameters and running the model many times (simulations) to generate a distribution of expected outcomes.
- c) Testing only two extreme scenarios.
- d) Discounting costs but not outcomes.
Answer: b) Assigning probability distributions to uncertain parameters and running the model many times (simulations) to generate a distribution of expected outcomes.
18. The “cycle length” in a Markov model should be chosen based on:
- a) The total time horizon of the model.
- b) The clinical relevance of the time interval for transitions between health states and accrual of costs/outcomes.
- c) The number of decision nodes.
- d) The preference of the software user.
Answer: b) The clinical relevance of the time interval for transitions between health states and accrual of costs/outcomes.
19. When critiquing a decision analysis study, it is important to assess whether the chosen model structure (e.g., decision tree vs. Markov model) is:
- a) The most complex possible.
- b) Appropriate for the disease process and decision problem being addressed.
- c) Published in a high-impact journal.
- d) Visually appealing.
Answer: b) Appropriate for the disease process and decision problem being addressed.
20. If a decision analysis calculates an “expected utility” for different treatment strategies, it is likely part of a:
- a) Cost-Minimization Analysis.
- b) Cost-Effectiveness Analysis using natural units.
- c) Cost-Utility Analysis (where utility reflects preference for health states).
- d) Cost-Benefit Analysis.
Answer: c) Cost-Utility Analysis (where utility reflects preference for health states).
21. A “base-case analysis” in a decision model refers to the results obtained using:
- a) The most optimistic assumptions.
- b) The most pessimistic assumptions.
- c) The most likely or best estimates for all input parameters.
- d) Only data from a single patient.
Answer: c) The most likely or best estimates for all input parameters.
22. The results of a probabilistic sensitivity analysis are often displayed as a:
- a) Single ICER point.
- b) Cost-effectiveness acceptability curve (CEAC).
- c) Simple bar graph of costs.
- d) Pie chart of probabilities.
Answer: b) Cost-effectiveness acceptability curve (CEAC).
23. A Cost-Effectiveness Acceptability Curve (CEAC) shows the probability that an intervention is cost-effective compared to an alternative, across a range of:
- a) Discount rates.
- b) Time horizons.
- c) Willingness-to-pay (WTP) thresholds for an additional unit of effect (e.g., QALY).
- d) Patient ages.
Answer: c) Willingness-to-pay (WTP) thresholds for an additional unit of effect (e.g., QALY).
24. Which of the following is a key assumption of a basic Markov model regarding transitions between states?
- a) The probability of transitioning to another state depends on the patient’s entire history in the model (memory).
- b) The “Markovian property” or “memoryless” property: the probability of transitioning to another state in the next cycle depends only on the current state, not on how the patient arrived in that state.
- c) Transitions can only occur in one direction.
- d) All transition probabilities are equal to 0.5.
Answer: b) The “Markovian property” or “memoryless” property: the probability of transitioning to another state in the next cycle depends only on the current state, not on how the patient arrived in that state.
25. “Tunnel states” might be used in a Markov model to:
- a) Represent absorbing states.
- b) Model a temporary condition or a sequence of events within a larger health state before transitioning out.
- c) Simplify the model by removing states.
- d) Only represent perfect health.
Answer: b) Model a temporary condition or a sequence of events within a larger health state before transitioning out.
26. When selecting probabilities for chance events in a decision tree, it is important that these probabilities are:
- a) Based on the modeler’s gut feeling.
- b) Derived from the best available evidence (e.g., clinical trials, epidemiological data).
- c) Always set to 50%.
- d) Only obtained from a single expert.
Answer: b) Derived from the best available evidence (e.g., clinical trials, epidemiological data).
27. One advantage of using a decision tree over a simple comparison of average outcomes is that it can explicitly model:
- a) Only costs.
- b) The sequential nature of decisions and the impact of chance events.
- c) Only patient preferences.
- d) The chemical structure of drugs.
Answer: b) The sequential nature of decisions and the impact of chance events.
28. If a decision analysis model shows that Strategy A has an expected cost of $5000 and an expected QALY of 2.5, and Strategy B has an expected cost of $4000 and an expected QALY of 2.0, the ICUR of A vs B is:
- a) $2000 per QALY
- b) $1000 per QALY
- c) $5000 per QALY
- d) Cannot be determined.
Answer: a) $2000 per QALY (ICUR = ($5000-$4000) / (2.5-2.0) = $1000 / 0.5 QALY = $2000/QALY)
29. The “validation” of a decision analysis model refers to processes that assess:
- a) How visually appealing the model is.
- b) The accuracy and reliability of the model’s structure, assumptions, and predictions (e.g., face validity, internal validity, external validity/calibration).
- c) The cost of building the model.
- d) The number of publications citing the model.
Answer: b) The accuracy and reliability of the model’s structure, assumptions, and predictions (e.g., face validity, internal validity, external validity/calibration).
30. A pharmacoeconomic study using decision analysis should clearly state its limitations, which might include:
- a) The model being too simple to be useful.
- b) Uncertainty in input parameters, simplifications made in the model structure, or limited generalizability.
- c) The fact that it was funded.
- d) The use of probabilities.
Answer: b) Uncertainty in input parameters, simplifications made in the model structure, or limited generalizability.
31. The main purpose of including “costs” associated with each pathway in a decision tree used for pharmacoeconomic analysis is to:
- a) Determine which pathway is clinically superior.
- b) Calculate the expected cost associated with each decision option.
- c) Only to identify direct medical costs.
- d) Make the tree more complex.
Answer: b) Calculate the expected cost associated with each decision option.
32. In a Markov model for a chronic disease, patients can _______ health states at the end of each cycle based on defined probabilities.
- a) only improve
- b) only worsen
- c) remain in the same state or transition to other defined
- d) always exit the model
Answer: c) remain in the same state or transition to other defined
33. The “expected value of perfect information” (EVPI) is an advanced concept in decision analysis that quantifies:
- a) The current value of the best decision.
- b) The maximum amount one would be willing to pay to obtain perfect information and eliminate all uncertainty about model parameters.
- c) The probability of making the wrong decision.
- d) The cost of conducting further research.
Answer: b) The maximum amount one would be willing to pay to obtain perfect information and eliminate all uncertainty about model parameters.
34. When critiquing the probabilities used in a decision model, a pharmacist should consider if they are:
- a) Based on outdated or irrelevant studies.
- b) Clearly referenced and appropriate for the population and context being modeled.
- c) Guessed by the authors.
- d) All equal to 1.0.
Answer: b) Clearly referenced and appropriate for the population and context being modeled.
35. If a decision analysis is used to compare three mutually exclusive treatment options, the option with the _______ expected utility (or _______ expected cost for equivalent outcomes) would generally be preferred.
- a) lowest; highest
- b) highest; lowest
- c) median; median
- d) negative; positive
Answer: b) highest; lowest (Highest expected utility, or lowest expected cost if outcomes are equivalent or ICER is favorable for effectiveness outcomes).
36. One reason decision analysis is favored over simply observing outcomes in a single cohort is its ability to:
- a) Eliminate all bias.
- b) Systematically compare multiple strategies and incorporate uncertainty from various data sources in a structured way.
- c) Provide definitive proof of causality.
- d) Be completed without any data.
Answer: b) Systematically compare multiple strategies and incorporate uncertainty from various data sources in a structured way.
37. The choice between using a decision tree and a Markov model often depends on:
- a) The number of authors on the paper.
- b) Whether the condition involves ongoing risk, recurring events, or transitions between chronic states over time (favoring Markov).
- c) The software available.
- d) The desired level of graphical complexity.
Answer: b) Whether the condition involves ongoing risk, recurring events, or transitions between chronic states over time (favoring Markov).
38. Sensitivity analysis helps address the _______ inherent in the input parameters of a decision model.
- a) certainty
- b) uncertainty
- c) simplicity
- d) color
Answer: b) uncertainty
39. A decision model might be used to estimate the cost-effectiveness of a new drug when long-term outcome data from clinical trials are:
- a) Abundant and cover a lifetime.
- b) Not yet available or limited in duration.
- c) Completely irrelevant.
- d) Only available for animal models.
Answer: b) Not yet available or limited in duration.
40. When reviewing a decision analysis in a research article, the “Methods” section should clearly describe:
- a) Only the results of the base-case analysis.
- b) The model structure, data sources for probabilities and outcomes, assumptions made, and types of analyses performed (e.g., sensitivity analysis).
- c) The authors’ personal opinions on the best treatment.
- d) The marketing strategy for the interventions.
Answer: b) The model structure, data sources for probabilities and outcomes, assumptions made, and types of analyses performed (e.g., sensitivity analysis).
41. If a decision node offers two choices, Treatment A and Treatment B, and the expected QALYs for A are 5.5 and for B are 5.2, then based solely on maximizing QALYs:
- a) Treatment B is preferred.
- b) Treatment A is preferred.
- c) Both are equally preferred.
- d) A cost-benefit analysis is required.
Answer: b) Treatment A is preferred.
42. A major strength of Markov models is their ability to simulate disease progression and treatment effects over:
- a) A single point in time only.
- b) Multiple discrete time cycles, allowing for changes in health states and accumulation of costs/outcomes.
- c) Only the duration of a typical clinical trial.
- d) The lifetime of the decision analyst.
Answer: b) Multiple discrete time cycles, allowing for changes in health states and accumulation of costs/outcomes.
43. The “half-cycle correction” in Markov models is used to:
- a) Adjust for the fact that transitions are assumed to occur at the beginning or end of a cycle, to better approximate continuous time.
- b) Correct errors in probability estimates.
- c) Ensure the model runs twice as fast.
- d) Discount future outcomes more heavily.
Answer: a) Adjust for the fact that transitions are assumed to occur at the beginning or end of a cycle, to better approximate continuous time.
44. The complexity of a decision analysis model should be:
- a) As high as possible to impress readers.
- b) As simple as possible while still adequately capturing the essential features of the decision problem and clinical reality.
- c) Dictated by the software used.
- d) The same for all decision problems.
Answer: b) As simple as possible while still adequately capturing the essential features of the decision problem and clinical reality.
45. Pharmacists may use the principles of decision analysis (even informally) when:
- a) Counting pills.
- b) Weighing different treatment options for a patient by considering probabilities of success, risks of side effects, and patient preferences.
- c) Stocking shelves.
- d) Processing insurance claims only.
Answer: b) Weighing different treatment options for a patient by considering probabilities of success, risks of side effects, and patient preferences.
46. When different data sources provide conflicting estimates for a probability in a decision model, a common approach is to:
- a) Use the estimate that favors the desired outcome.
- b) Use an average or median value, and then test the impact of this uncertainty using sensitivity analysis.
- c) Ignore that parameter in the model.
- d) Only use data from the oldest study.
Answer: b) Use an average or median value, and then test the impact of this uncertainty using sensitivity analysis.
47. If a decision analysis comparing two drugs shows that Drug A has a higher expected cost but also higher expected QALYs than Drug B, the next step is often to:
- a) Conclude Drug A is always preferred.
- b) Conclude Drug B is always preferred.
- c) Calculate an Incremental Cost-Utility Ratio (ICUR) for Drug A vs. Drug B to assess its value.
- d) Assume both are equivalent.
Answer: c) Calculate an Incremental Cost-Utility Ratio (ICUR) for Drug A vs. Drug B to assess its value.
48. The transparency of a decision analysis model is crucial for its critique. This means that:
- a) The model should be a “black box” with no details provided.
- b) All assumptions, data sources, calculations, and the model structure should be clearly reported and accessible for scrutiny.
- c) Only the final ICER should be reported.
- d) The model should only be understandable by expert modelers.
Answer: b) All assumptions, data sources, calculations, and the model structure should be clearly reported and accessible for scrutiny.
49. One of the most important skills for a pharmacist when encountering a study based on decision analysis is the ability to:
- a) Re-program the entire model from scratch.
- b) Critically appraise its methodology, assumptions, and the relevance of its conclusions to their specific patient population or practice setting.
- c) Memorize all the input probabilities.
- d) Contact the original study authors for clarification on every detail.
Answer: b) Critically appraise its methodology, assumptions, and the relevance of its conclusions to their specific patient population or practice setting.
50. Decision analysis provides a framework to combine evidence on _______ and _______ to support rational decision-making in healthcare.
- a) only drug costs; only drug benefits
- b) probabilities of events; values of outcomes (costs and/or effectiveness/utility)
- c) patient preferences; physician opinions only
- d) drug manufacturing processes; marketing strategies
Answer: b) probabilities of events; values of outcomes (costs and/or effectiveness/utility)
I am a Registered Pharmacist under the Pharmacy Act, 1948, and the founder of PharmacyFreak.com. I hold a Bachelor of Pharmacy degree from Rungta College of Pharmaceutical Science and Research. With a strong academic foundation and practical knowledge, I am committed to providing accurate, easy-to-understand content to support pharmacy students and professionals. My aim is to make complex pharmaceutical concepts accessible and useful for real-world application.
Mail- Sachin@pharmacyfreak.com