The aim of this assessment is to construct a simple decision tree
diagram to assess the cost-effectiveness of a new cervical screening
test when compared to the current screening test. The assessment task
requires you to complete a series of structured stages, and then to
write a brief health technology assessment report with the specified
headings summarizing the project, methods and outcomes. All stages
contribute to the assessment.
Please note that you DO NOT REQUIRE dedicated economic modelling
software to complete this task. All of the calculations can be
undertaken easily with a calculator or in Excel.
Your completed assessment should be submitted through TURNITIN – via UTS Online. Select Assessment from the menu
Label your attachment with your name and student number as follows:
23787_Assessment_3_First name_Last name_Student Number
Your completed assessment task should have
1. An executive summary that represents a brief health technology
assessment report based on your answers: summarizing the project,
methods and outcomes. This part should be no more than 500 words. (10
marks)
The required headings for the executive summary are:
• Objectives
• Methods
• Data
• Results
• Key areas of uncertainty
• Discussion
• Recommendation
2. An attachment that shows your answers to Parts One to Six of the
task, including calculations where required. (Contribution to overall
marks are shown for each part) (50 marks in total)
Part One: Calculating the accuracy of the two test alternatives (10 marks)
Screening tests for cervical cancer aim to identify pre-cancerous
changes in the cervix that could develop into cervical cancer. If the
pre-cancerous tissues is identified early and removed, then cervical
cancer can be prevented for developing.
The Comparator – In the conventional Pap smear, the doctor collecting
the cells smears them on a microscope slide and applies a fixative. This
slide is then sent to a laboratory for evaluation. Studies of the
accuracy of conventional (current) Pap smear tests report:
• Sensitivity 72%
• Specificity 94%
The New Test – The new test works in exactly the same way as the current
test, however the manufacturer believes that the sensitivity of the new
test is better. Below are the results of a cohort study that tested the
new cervical screening test. Note that all women were 30 years of age
when tested.
New Test Disease Status Total
Cervical cancer (+ve) Cervical cancer (-ve)
Test (Positive) 44 36
Test (Negative) 6 564
Total
A) For the new cervical screening test define the following, and include the number of individuals in each group.
• True positive
• False positive
• True negative
• False negative
B) Calculate the sensitivity and specificity for the new test.
C) Compared with the current test, the new test was evaluated using a
different cohort of women and in a different laboratory. Does this
influence the sensitivity and specificity of the new test?
Part Two: Construct a decision tree (10 marks)
Your task is to assess the cost-effectiveness of screening women when
they reach the age of 30. We also assume that everyone who is invited to
participate in screening program receives a cervical screening test
(i.e. the uptake rate of the test is 100%).
Draw a decision tree to determine whether the new cervical screening
test is more cost-effective than the current test. To do this you need
to create a decision node with the option to accept the new test or the
current test. For each test, the terminal nodes should reflect the
possible outcomes of the test result (e.g. True positive etc…)
Part Three: Estimating the benefit of testing (5 marks)
To populate the decision tree we need to estimate the benefits and costs
of each test option. The benefits of screening are measured in terms of
quality adjusted life years (QALYs) gained (i.e. quality-of-life
multiplied by the number of years in that health state).
• Utility score – A time-trade off study conducted on the same cohort of women that received the new test demonstrated that;
o The average utility in the non-cancer group (test negative) was 0.92.
o The average utility in the non-cancer group (test positive) was 0.91
(slight reduction in utility due to further investigations and concern
of possible cancer)
o The average utility in the cancer group (not detected by the test) was
0.45 (This reduced utility is due to the side-effects of treatment and
the impact of the disease).
o The average utility in the cancer group (detected by the test and
treated early) is 0.87 (there is a slight reduction in quality of life
due to early treatment.
• Survival – Long-term registry data were used to estimate the
additional survival (note that this is the additional survival beyond 30
years of age, which is the age when a person would be screened in this
model)
o The average survival of a 30 year old woman with cervical cancer (not detected early) is an additional 5 years.
o The average survival of a 30 year old woman with cervical cancer that
is detected early and treated (i.e. detected with a positive test
results) is an additional 40 years.
o For all other 30 year old women (no cancer) the average survival is an additional 40 years.
A) Calculate the average additional QALYs gained for individuals with the following possible test outcomes:
• True positive
• False positive
• True negative
• False negative
B) In this model, all outcomes (costs and benefits) are undiscounted.
Why do we discount future costs and benefits? Why might discounting
costs and benefits at the same rate penalize preventative health
programs?
Part Four: Estimating Costs (5 marks)
The tables below were taken from a longitudinal cohort study of women
that participated in the current screening program. The unit costs are
provided in Table 1. Table 2, contains an inventory of all the resources
used, on average, by an individual depending upon their test result.
• For example, an individual identified as being ‘true positive’ (using
the current test) would require the following resources – 1xcurrent
test, 2 x GP visits, 1 x further examination – early treatment.
Therefore their treatment would cost – 1x$50 + (2x$35) + 1 x $2000 =
$2,120
Combine the information from Tables 1 and 2 to generate the total cost
of each screening outcome. Do this for both the current test and the new
test scenarios.
Table 1: Unit costs
Description Cost
Current test $50
New Test $300
GP appointment $35
Further examination – No treatment $500
Further examination – Early treatment $2000
Delayed treatment $50,000
Table 2: Resources use for each possible alternative
Current test New Test GP visit Further exam – no treat Further exam – early treat Delayed treatment
Current test True Positive 1 2 1
False positive 1 2 1
True negative 1 1
False negative 1 1 1
New Test True Positive 1 2 1
False positive 1 2 1
True negative 1 1
False negative 1 1 1
NOTE: All costs calculated should be presented to two decimal places.
Part Five: Cost-utility analysis (10 marks)
You should now have the following information:
• Accuracy of the current and new cervical screening tests
• A decision tree that reflects the possible outcomes of both tests
• An estimate of the QALYs gains for each alternative
• An estimate of the resource use (cost) of each alternative
The final information that you need to complete to complete the analysis
is the prevalence of cervical cancer in this population. In this
example we are screening women 30 years of age; the prevalence of
cervical cancer in this cohort is 1 in 1000 or (0.001)
A) Complete Table 3: Model Parameters using the information Part One-Part Four.
Parameter description Current Test New Test
Prevalence of cervical cancer 0.001 0.001
Sensitivity of test 0.72
Specificity of test 0.94
Cost – True Positive
Cost – False Positive
Cost – True Negative
Cost – False negative
QALYs – True Positive
QALYs – False Positive
QALYs – True Negative
QALYs – False negative
B) You now need to combine this information into your decision tree to
determine the cost-effectiveness of the new test relative to the current
test. Provide your answer as an incremental cost-effectiveness ratio
(ICER) (i.e. cost/QALY gained). Also, provide the diagram of your
decision tree at this stage.
• Hint: Remember that you need to calculate the expected value (costs
and QALYs) of each alternative before you can estimate the
cost-effectiveness. It is easier to calculate the expected value if you
start at the end of the tree, rather than the beginning (i.e. you need
to roll-back the decision tree – see lecture notes for example)
C) If the decision maker has set an explicit threshold of $50,000 / QALY
gained, would you say the new test is cost-effective? Explain your
answer.
Part 6: Sensitivity Analysis (10 marks)
The decision maker would like you to determine the cost-effectiveness of
the new test in a population of women with a family history of cervical
cancer. In this high-risk cohort of women, the prevalence of cervical
cancer is 1 in 100 (0.01).
A) Calculate the ICER of the new test relative to the current test in this high-risk population of women.
B) Why do you think the cost-effectiveness of the new test is
sensitivity to prevalent risk of cervical cancer in the population?
C) In the original model (prevalence = 0.001), we assumed a 20 min GP
appointment costs $35. However, an audit of General Practices conducting
the new test shows that 60% of GPs charge patients a double appointment
(2x20mins). How does this change you ICER? Explain your answer.
D) What would be the ICER for the new test in the high-risk cohort
(prevalence of cervical cancer is 0.01), if 60% of GPs charge patients a
double appointment (2x20mins).
E) What type of sensitivity analysis was carried out in sub-question
(D)? What is the advantage of this over what was conducted in
sub-question (C)?
.
