SENSITIVITY ANALYSIS
MARKOV MODELING
MILIN THOMAS
SEMESTER 2 MPHARM
PHARMACY PRACTICE
SENSITIVITY ANALYSIS
• A sensitivity analysis is an integral component of any sound
pharmacoeconomic study.
• It is a 'process through which the robustness of an economic model is
assessed by examining the changes in results of the analysis when key
variables are varied over a specified range’.
• Sensitivity analysis represents a method of testing the validity of
statements made from results that were based on certain assumptions
in the pharmacoeconomic study.
• It is concerned with exploring the consequences of the uncertainty in
both the parameter estimations and assumptions made in the
pharmacoeconomic model
• Sensitivity analysis should be done for all kinds of economic
evaluation—whether cost minimization, cost-effectiveness, or some
other type.
• This technique is used to quantify the uncertainty in an economic
evaluation by varying the values given to a key variable in the
valuation.
• Sensitivity analyses are advocated in pharmacoeconomic research
because a number of assumptions and uncertainties are incorporated
into the analysis, and thus the integrity of the results may be
compromised if those assumptions prove to be inaccurate
The areas of uncertainty relate to 4 general components, namely: –
1. The sample data
2. The generalizing ability of the results
3. The extrapolation of the results
4. Uncertainty in the analytical methods
• In order to confirm the influence of such uncertainty, sensitivity
analyses are conducted in pharmacoeconomic analyses using models.
• A sensitivity analysis is a way of examining any change in results due
to changes in the parameter values applied within a certain scope.
• A sensitivity analysis using only one parameter is termed a one way
sensitivity analysis while one that simultaneously varies two
parameters is a two-way sensitivity analysis.
• Examples of parameters in a cost-analysis study include cost of
physician visits, price of medicines, cost estimate of adverse drug
reactions (ADRs) as well as the number of ADRs experienced,
laboratory tests required in the treatment, and duration of treatment
Types of sensitivity analysis
Univariate sensitivity analysis
techniques
• It involves varying one parameter at a time while keeping all others constant.
• For example researcher may assess how changes in drug price or treatment effectiveness impact
cost effectiveness result
Multivariate sensitivity analysis
• Multiple parameters are varied simultaneously to examine the joint effect of uncertainty.
• It allows for the estimation of interaction between parameters and provides a comprehensive
understanding of how changes in multiple factors collectively affect the study.
Threshold analysis
• It determine values of certain parameters or variables at which the study outcome changes from
one conclusion to other.
• For example, it identifies the threshold drug price or threshold effectiveness level required for a
treatment to be costeffective.
Method
Selection of parameter :
• dentify the specific parameter that will be varied in the analysis. This
parameter can be cost value, treatment efficacy, discount rate,
duration of treatment or any other variable.
Define a range of values
• Determine a plausible range or values for the selected parameter.
• This range can be based on available data, expert opinion, published
literature or clinical guidelines.
Conduct a scenario analysis
• Perform the analysis by systematically varying the chosen parameter
within the defined range.
• Calculate the study outcome or result for each value of parameter.
• Eg: in a cost effective analysis, calculation of incremental cost
effectiveness ratio for different values of the cost parameter
Interpret the results
• Examine the changes in study outcome as the parameter is varied.
• Assess how sensitive the results are to changes in the parameter
value.
• Identify threshold or key value where outcome change significantly
Present findings
• Report the results in clear concise manner.
• This may include graphical representation like tornado diagram or
spider plots to visualise the impact of parameter
MARKOV MODELING
• A Markov model is a Stochastic method for randomly changing
systems where it is assumed that future states do not depend on past
states. These models show all possible states as well as the
transitions, rate of transitions and probabilities between them.
• Markov models are often used to model the probabilities of different
states and the rates of transitions among them. The method is
generally used to model systems. Markov models can also be used to
recognize patterns, make predictions and to learn the statistics of
sequential data.
STEPS IN MARKOV MODELING
There are 5 steps for Markov modeling:
1. Choose the health states that represent the possible outcomes from
each intervention
2. Determine possible transitions between health states
3. Choose how long each cycle should be and how many cycles will be
analyzed
4. Estimate the probabilities associated with moving(i.e., transitioning)
in and out of health states
5. Estimate the costs and outcomes associated with each option.
BUBBLE DIAGRAM FOR A GENERAL MARKOV MODEL
EXAMPLE :BUBBLE DIAGRAM FOR DIABETES
STEP 1: CHOOSE HEALTH STATES
• First, a delineation of mutually exclusive health states should be
determined by listing different scenarios a patient might reasonably
experience. These are referred to as Markov states.
• Patients cannot be in more than one health state during each cycle.
• A simple general example is “well, sick, or dead.”
• Graphically, by convention, each health state is placed in an oval or
circle in a bubble diagram (Fig. 10.1).
• Time cycles are depicted on the left of the graph.
• For the DM example, we are concerned with two health states: IGT and
DM (Fig. 10.2).
STEP 2: DETERMINE TRANSITIONS
• Possible transitions between states are determined based on clinical information.
Determine whether patients can move (i.e., transition) from one health state to another
• For example, if the patient dies, this is called an absorbing state.
• An absorbing state indicates that patients cannot move to another health state in a later
cycle. Graphically, arrows are used to indicate which transitions are allowed.
• In the general example given in Figure 10.1, we assume that everyone starts out in the
well state. For cycle 1, each patient can stay well, or can move to the sick or dead states.
For the next cycle, patients in the well state can again stay well or move to the sick or
dead states.
• Those in the dead state cannot move back to the other 2 states
• For the diabetes example, we will assume that if a patient transitions from the
prediabetic state to DM that he or she cannot return to the prediabetic state. Thus, in
the diabetes Markov model (Fig. 10.2), DM is an absorbing state
STEP 3: CHOOSE THE CYCLE LENGTH AND NO: OF
CYCLES
• The cycle length depends on the disease being modeled
• For chronic diseases, a cycle length of 1 year is commonly used.
• Again, the number of cycles depends on clinical relevance.
• Eg:, For the diabetes example (Fig. 10.2), five 1-year cycles were used
to determine the impact of the diet and exercise program on
progression to diabetes
• Sometimes the model is run for the natural lifetime of the patients or
until a certain percent of the cohort is in the absorbing state
STEP 4: ESTIMATE TRANSITION PROBABILITIES
• Transition probabilities are used to estimate the percent of patients
who are likely to move from one health state to another during each
cycle.
• These probability values usually come from previous research or
expert panel estimates
• Eg: For the diabetes example, estimates were used for those with and
without the diet and exercise program.
• For patients who did not receive the specific diet and exercise
program, there was a 10% probability per year (cycle) that they would
transition from IGT to DM (90% would stay in the IGT state).
• For patients who received the program, the probability of DM was
reduced to 5% per year (95% would stay in the IGT state).
• It was assumed that after patients were diagnosed with DM, they
could not transition back to IGT.
STEP 5: CALCULATE COSTS AND OUTCOMES
• Outcomes for each health state should be estimated and given a value.
• If the outcome of interest is years of life gained or saved and each cycle
is for 1 year, then each person who is alive during a cycle gets a value of
1.0 as his or her outcome for that cycle.
• Eg:- In Figure 10.1, for each year in a well state, the value is 1.0, and for
each year in the sick state, the value is 0.5, and the value for the dead
state is 0.
• Costs in each health state should be estimated as with simple decision
analyses.
• The total costs and outcomes are then summed for all cycles.
DISADVANTAGES OF MARKOV
MODELING
• By their nature, Markov models can be more complex than simple decision
trees and therefore less transparent to decision makers.
• A commonly cited disadvantage of Markov modeling is that it is
“memoryless” because the Markovian assumption is that the probability of
moving from state to state is not based on the previous experiences from
former cycles. In practice, a patient’s medical history is an important
determinant in the probability of his or her future health.
• Another disadvantage is that the data needed to estimate probabilities and
costs, especially in the long term, are often unavailable.
• Most clinical studies measure outcomes for a short time, and extrapolation
into the future may compound errors in estimations.
APPLICATIONS OF MARKOV
MODELING
The greater availability & access to processing power through
computers allow that these models can be used more often to
represent clinical structures
Estimates cost & clinical outcomes ,especially in case of chronic
disease
Estimate cost-effectiveness of treatment programmes(prevention,
hospital care, etc)
It also has wide application in applied fields such as economics,
logistics, marketing
REFERENCES
• Karen L Rascati , Essentials of Pharmacoeconomics, 2nd edition ,page
no: 193- 203
