Cost-Utility of HPV for Prevention of Cervical Cancer in the State of Roraima (Brazil): A Markov Model Approach.

1. Cost-Utility of HPV for Prevention of Cervical
Cancer in the State of Roraima (Brazil):
A Markov Model Approach
Giacomo Balbinotto Neto (Ph.D)
Allex Jardim da Fonseca (M.D)
Universidade Federal do Rio Grande do Sul – Brazil.
Institute for Health Technology Assessment – Brazil

2. Introduction
In Brazil, Cervical Cancer (CC) represents an important public health
problem.
It is estimated that 22.000 new cases of CC will arise in 2012, corresponding
to an incidence rate of 17.5 cases per 100.000 women and to a mortality
rate due to CC of 10.2 deaths per 100,000 women.
In the Brazilian Amazon Region, the problem is even more serious.
Due to a low screening coverage for CC in target population (less than 25%),
a high incidence of CC has been registered in Amazonia (up to 46
cases/100,000 women), similar to the incidence rates in low-income
countries, such as Uganda and Mali.
2

3. Roraima
Nacional Institute of Cancer – Brazil, 2011.
3

4. Brazilian government programs to control CC are
based on secondary prevention and have historically
failed to overcome:
(i) geographical isolation in the rain forrest ;
(ii) cultural barriers.
4

5. Objective
To assess cost-utility of the HPV vaccination on the
prevention of ICC in Brazilian Amazon Region (State
of Roraima).
Methods
A Markov cohort model was developed to simulate
the natural history of HPV and its progress to ICC,
considering the current preventive programs and
treatment costs.
5

6. Analytical Decision Model
The analysis was performed from the provider’s perspective (Brazil's Unified
Health System - SUS).
The target population was preteen girls under 12 years old.
The cohort time horizon was lifetime.
The model simulated the natural history of HPV infection until its progression
to invasive cervical cancer, taking into account the prevention programs (Pap
test) that are current in Brazil.
For each strategy (screening plus vaccination or screening only), the model
incorporated health state transition probabilities and the target population
was followed from adolescence until death in a hypothetical cohort.
6

7. The 1-year transition
probabilities were
based mainly on
empirical data of
local and national
studies.
The model evaluated
the addition of the
vaccine to 3 cervical
cancer screening
scenarios (0, 3 or 10
exams throughout
life).
7

8. Analytical Decision Model
The model incorporated the transition probabilities
of mutually exclusive health states that refer to one-
year cycles.
The model simulated the transition probabilities for
70 years from the age of vaccination (12 years old).
At each transition, the model attributed the costs,
quality of life, and death expectation according to
the individual’s health condition.
8

9. Analytical Decision Model
The transition probabilities were based on empirical
data from medical literature and referred to
transitions from a healthy state to a possible HPV
infection and low-grade squamous intraepithelial
lesion (LSIL) induction, which could regress over
time to normality, persist, or progress to high-grade
squamous intraepithelial lesion (HSIL).
This, in turn, could persist, regress to normality, or
progress to localized, regional, or metastatic invasive
cancer.
9

10. Analytical Decision Model
At each screening event, cervical lesions would be
found according to the criteria (Pap test sensitivity
and specificity) described by national studies.
The detection of cervical lesions would require
follow-up evaluations or treatment (colposcopy,
cryosurgery, and/or surgery), for which they have
been assigned a likelihood of success, costs, and
implications for the quality of life of the individuals of
the cohort model.
10

11. Analytical Decision Model
The economic analysis has adopted a 3% annual
discount rate for the cost and outcome, with the
intent to convert future values into present values.
The model was calibrated by adjusting the incidence
of precursor lesions of CC to adequately simulate the
results of cancer incidence as recorded in the
Brazilian Amazon Region.
11

12. Analytical Decision Model
The fundamental structure of the model is based on clinical
practice that is consistent with the clinical program procedures
advocated by the VIVA MULHER, program from the Brazilian
Ministry of Health.
Abnormal screening examinations were forwarded to
colposcopy, and tissues were evaluated by biopsy. If HSIL were
histologically confirmed, then the patient would be subjected to
cryotherapy treatment or surgery.
The costs related to each procedure were taken from the funds
allocation table of the Ministry of Health of Brazil.
12

13. Analytical Decision Model
The probabilities of death by cancer at each stage were
extracted from the global survival curves of longitudinal studies.
The five-year survival rate varied from 92.0% for localized
cancer to 55.7% for regional cancer and 16.5% for metastatic
cancer.
The annual incremental costs were estimated at 10% of the
initial value of the cancer treatment and refer to screening
examinations, control of sequelae, and treatment-related
toxicities or costs related to tumor recurrence. Only direct costs
that were assigned to cancer were computed and are
represented in international dollars (US$).
13

14. Analytical Decision Model
The model multiplies the years of life by the utility
implicated in the health status to adjust survival by
quality of life, with the final outcome of effectiveness
being quality-adjusted life years (QALYs).
In the base case, vaccination coverage was assumed
at 70% of the target population.
14

15. Analytical Decision Model
In this model, it was determined that the vaccine provides
immunity throughout life after three doses in the course of
one year. However, simulations on the need for booster doses
throughout life to maintain immunity (one booster dose every
ten years) were also performed.
The reduction in the incidence of CC-inducing lesions as a
result of vaccination was based on studies that originally
reported the effectiveness of the quadrivalent vaccine .
15

16. Analytical Decision Model
The cost of vaccination (three doses + implementation costs)
for the base case was estimated at US$ 268.
There are no references for the price of the vaccine in Brazil
for the large-scale public sector; since the vaccine has not
been incorporated into public health protocols yet.
However, it was reported that the average price of vaccination
(three doses) in the American market is US $360 (LIPPMAN,
2007). For the public sector, the value negotiated by the
Centers for Disease Control in the United States was US $290
(CDC, 2010).
16

17. Preventive Strategies Cost per Quality- Incremental QALYs ICER
individual adjusted life cost (US$) saved per
(US$/QALY)
years (QALYs) individual
(US$)
Non-screening scenario
Vaccination 467 24.80 -42 0.2 Dominant
No vaccination (natural history) 510 24.60
Scenario of 3 screenings
throughout the lifetime (base
case)
Vaccination + screening 511 29.6 188 0.2 1,374
Only screening 322 29.4
Scenario of 10 screenings
throughout the lifetime
Vaccination + screening 772 34.5 438 0.2 2,518
Only screening 334 34.3 17

18. Preventive Strategies Cost per Quality- Incremental QALYs ICER
individual adjusted life cost (US$) saved per
(US$/QALY)
years (QALYs) individual
(US$)
Non-screening scenario
Vaccination 467 24.80 -42 0.2 Dominant
No vaccination (natural history) 510 24.60
Scenario of 3 screenings
throughout the lifetime (base
case)
Vaccination + screening 511 29.6 188 0.2 1,374
Only screening 322 29.4
Scenario of 10 screenings
throughout the lifetime
Vaccination + screening 772 34.5 438 0.2 2,518
Only screening 334 34.3
18

19. Preventive Strategies Cost per Quality- Incremental QALYs ICER
individual adjusted life cost (US$) saved per
(US$/QALY)
years (QALYs) individual
(US$)
Non-screening scenario
Vaccination 467 24.80 -42 0.2 Dominant
No vaccination (natural history) 510 24.60
Scenario of 3 screenings
throughout the lifetime (base
case)
Vaccination + screening 511 29.6 188 0.2 1,374
Only screening 322 29.4
Scenario of 10 screenings
throughout the lifetime
Vaccination + screening 772 34.5 438 0.2 2,518
Only screening 334 34.3
19

20. Preventive Strategies Cost per Quality- Incremental QALYs ICER
individual adjusted life cost (US$) saved per
(US$/QALY)
years (QALYs) individual
(US$)
Non-screening scenario
Vaccination 467 24.80 -42 0.2 Dominant
No vaccination (natural history) 510 24.60
Scenario of 3 screenings
throughout the lifetime (base
case)
Vaccination + screening 511 29.6 188 0.2 1,374
Only screening 322 29.4
Scenario of 10 screenings
throughout the lifetime
Vaccination + screening 772 34.5 438 0.2 2,518
Only screening 334 34.3
20

21. Sensitivity Analysis
All economic assessments show a certain degree of uncertainty,
inaccuracy, or methodological controversy (PETITI, 2000;
MUENNING, 2002; RASCATI, 2010).
Therefore, sensitivity analyses (one-way) were performed for
variables with uncertainty over the base case values to assess
the robustness of the present study findings.
These analyses recalculate the ICER considering the variations
in a given parameter.
21

22. Sensitivity Analysis
The variables that were evaluated were the cost of
vaccination, the effectiveness of vaccination, the
scenario of the pre-existing screening program,
vaccination coverage, time of immunity, annual
discount rate and the characteristics of the Pap test
(sensitivity).
For such analyses, the variation values represent our
judgment regarding the uncertainty of the study
parameter or the variations in the results that have
been published in the medical literature.
22

23. Results
The primary outcome for the calibration of the model was the
incidence of invasive cancer.
In the scenario of the natural history of HPV infection, without
screening exams, the model simulated a 4.2% lifetime risk of
cancer, which equates to 34.1 invasive CC cases per 100,000
women, considering the demographic structure of the region
studied (IBGE, 2007b).
In the scenario with screening three times throughout an
individual’s lifetime, the risk of cancer was estimated at 3.4%
(equivalent to 27.5 cases per 100,000 women).
23

24. Results
Figure below illustrates the CC incidence rate adjusted
for each age group, according to the data of the
incidence recorded for the population studied
(described as "observed") and compares the
simulations of this incidence rate (from the Markov
model developed) considering 2 baseline strategies:
no screening (natural history) and three screenings
throughout a woman’s lifetime.
24

25. The scenario of three Pap tests resulted
in satisfactory calibration (base case).
25

26. Prediction
The prediction in the three-screenings
scenario corresponded satisfactorily to the
gross incidence rate of invasive CC as recorded
in Amazon region in 2010 (FONSECA, 2010)
(28.2 cases per 100,000 women) and will be
considered as the baseline strategy to be
compared with the addition of vaccination.
26

27. Base Case Analysis
With a vaccination coverage rate of 70%, the
vaccination strategy for preteen girls of the Brazilian
Amazon Region would reduce the lifelong incidence
of CC by 35% in this population and would reduce
the mortality due to CC by approximately 33.4%.
The addition of the vaccine generated an incremental
cost of approximately US$ 188 per woman to the
current strategy. The incremental cost-effectiveness
ratio was US$ 1,374/QALYs saved.
27

28. Base Case Analysis
Figure below compares the reduction in the incidence of CC
for the various strategies (combination of cytological
screening and vaccination), given the different vaccine
coverage levels simulated by this model.
We note that the goal of a 50% reduction in CC incidence
could be achieved by combining high vaccination coverage
(>70%) with existing screening procedures (>three lifetime
Pap tests).
28

29. Base Case
The addition of HPV vaccination would reduce by
45% the incidence of invasive CC (70% vaccination coverage).
29

30. Results
The sensitivity analyses reveal that vaccination
tends to provide a favorable profile from a
cost-effectiveness point of view despite
changes in the base case parameters
proposed by the sensitivity analysis.
30

31. Base Case Analysis
The population vaccination coverage implies wide variations
in ICER, surpassing US$ 2,150/QALYs for vaccine coverage
levels of less than 50%.
The vaccination strategy tends to dominate the cytological
screening (3x) in isolation, i.e., it is less costly and more
effective (ICER ≤ 0) for vaccination costs lower than US$ 53 (all
doses for primary immunization).
For vaccination costs in excess of US$ 537, the vaccination
strategy requires approximately US$ 3,225to save 1 QALY.
31

32. Base Case Analysis
A vaccination effectiveness (reduction in the
incidence of precursor lesions) of above 40%
maintains the ICER below US$1,612/QALY compared
to the basal strategy. Increases in the sensitivity of
the Pap test tends to modestly increase the ICER of
added vaccination by improving the efficiency of the
baseline strategy and leading to a relative reduction
in the additional benefit of the vaccine.
32

33. 33

34. Discussion
The present study revealed that the addition of HPV vaccination
to the existing preventive strategy exhibits a favorable cost-
effectiveness and cost- utility profile in the Brazilian Amazon
Region (State of Romaima).
Except for the simulation of vaccination coverage rates below
30%, the ICER of the addition of vaccination does not exceed
the conventional limit of the GDP value per capita (about US$
10,000 for Brazil) for any other uncertainty simulation. If the
cost of vaccination is reduced to US$53 or less, with a
vaccination coverage rate of 70%, then adding vaccination
tends to dominate the cytological screening strategy alone.
34

35. Limitations to the Present Study
(i) First, because of the lack of national data, some
parameters have been calibrated based on international data.
(ii) Second, the adherence to strategic programs in Brazil
tends to be opportunistic (VALE, 2010), favoring the
vaccination of women who would have good adherence to
existing cytological screening to the detriment of non-
participating women in higher-risk situations.
(iii) Finally, the Markov assumption itself establishes that
transition probabilities depend exclusively on the current
health state, not on a sequence of past health states
(DRUMMOND, 2001).
35

36. Therefore …
The results of the present study should be considered as a
conservative estimate of ICER of HPV vaccination in a Brazilian
Amazon Region (State of Roraima) with a high incidence of CC.
However, the satisfactory calibration of the proposed model
parameters and the alignment with other researchers’ studies
strengthen the results that vaccination is cost-effective in the
Brazilian Amazon Region (GOLDIE, 2007; GOLDIE, 2008;
COLANTÔNIO, 2009).
36

37. Political Implications
The high risk of invasive CC in Amazon Region implies
an urgent need to rethink the current CC preventive
policy, especially for underprivileged regions of the
country.
The present study was the first cost-effectiveness
analysis of a CC preventive strategy that was directed
toward a specific region of the country (State of
Roraima).
37

38. Political Implications
The cost-efectiveness analysis of HPV vaccine for Amazon
Region showed a better profile when compared to studies
addressing this topic to Brazil as a whole, as in the analysis
published by Colantônio et al. (ICER = US$10,181/QALY) and
Goldie et al. (ICER = US$9,600/QALY).
Regarding public health, these results lead to the conclusion
that public policies on women's health, particularly on CC
prevention programs, should be decentralized (adjusted to
territorial reality) and not uniform since the heterogeneity
inherent in a country of continental proportions, such as
Brazil.
38

39. Political Implications
Large-scale preteen vaccination in the Amazon region can be considered
an investment in the future to prevent the premature deaths of hundreds
of women in the coming decades, who have historically been neglected in
preventive government programs.
Although it is difficult to accurately measure and evaluate this benefit, it
would translate into families and children who would not lose their
mothers to CC and resources that would not be allocated to the treatment
of this disease (direct and indirect costs) (COLANTÔNIO, 2008).
These results may provide the basis for accelerating the incorporation of
the vaccine for HPV 16 and 18 in preventive programs that serve
underprivileged women in disadvantaged regions of Brazil like Amazon
Region.
39

40. Conclusions
Vaccination has a favorable profile in terms of
cost-utility in the State of Roraima (Amazon
Region in Brazil).
Its inclusion in the immunization schedule
would result in substantial reduction in
incidence and mortality of invasive CC in
Brazilian Amazon region.
40

41. Thanks !
Obrigado !
Gracias !
giacomo.babinotto@ufrgs.br
41

42. 42