Document text
Economic analysis and health impacts of routine vaccination
with TAK- 003 dengue vaccine inSan Juan, Puerto Rico
Guido España, Manar Alkuzweny ,Alex Perkins
University ofNotre Dame
June 22nd,2023
Conflicts ofinterest & Disclosures
•GEand APhave previously received research funding from
GlaxoSmithKline tosupport unrelated research ondengue vaccine
development.
•APcurrently receives research funding and consulting fees from
Emergent Biosciences tosupport unrelated research on
chikungunya vaccine development.
•As of June 2023, GE is an employee of CDC. This work was
performed while working at the University of Notre Dame. The
findings and conclusions in this presentation are those of the
author(s) and do not necessarily represent the views of the CDC.
2
Abbreviation Full term/Meaning
VCD Virologically confirmed disease
VE Vaccine efficacy
DENV (e.g., DENV -3) Dengue virus (e.g., serotype 3 dengue virus)
HPV Human papillomavirus
PICO Policy question articulated as Population, Intervention,
Comparison, Outcomes
Additional hospitalizations Hospitalization induced by vaccine -enhanced disease
ICER Incremental cost -effectiveness ratio ($/QALY , $/hospitalization
averted)
QALY Quality adjusted life years
3Terminology
We modeled six scenarios to answer each of the
policy (PICO) questions.
•Age group scenarios:
•4 –16 years
•17 –60 years
•4 –60 years
•For each age group we modeled:
•Without required pre -vaccination screening (vaccinate
seropositives and seronegatives )
•With required pre -vaccination screening (vaccinate
seropositives only)
4
Outcomes of interest included benefits, harms, and
costs.
•Epidemiological outcomes:
•VCD* and hospitalizations averted
•Dengue hospitalizations among vaccinated seronegative
persons infected with DENV -3 and DENV -4.
•Health Outcomes:
•QALYs gained
•Economic outcomes:
•ICER ($/QALY)
•ICER ($/hospitalization averted)
*VCD refers to medically attended cases only 5
We calibrated an agent -based model to reproduce patterns
ofserotype circulation, force of infection, and age-specific
incidence.
•Wecreated asynthetic population torepresent demographic and
geographic characteristics ofSan Juan, Puerto Rico.
•Our model simulates individual daily activities ofhumans and
mosquitoes.
•Model calibrated using 30years ofhistorical dengue surveillance data
to reproduce observed force of infection and age -specific
incidence.
6
The serotype distribution simulated over 10 year time horizon of vaccination
was modeled on 30 (1986-2016) years of data from San Juan, Puerto Rico.
*Model based on 1,000 bootstraps with different proportion of serotypes in every simulation. Average reflects the mean propor tion of serotypes among
entire model. These estimates are preliminary and are subject to change.SerotypeAverage proportion in
model*
• DENV -1 14% (5% -25%)
• DENV -2 35% (17% -51%)
• DENV -3 29% (11% –46%)
• DENV -4 22% (9% -40%)
7
The estimated seroprevalence among persons 9 years of
age was approximately 40%.
•30% ofindividuals aged 5–9years were seropositive .
•People aged 20, 75% of people had ≥1 infection bytheend ofthe
calibration.Age group
8
Wesimulated the vaccine roll-out based onprevious
introduction ofHPV vaccine in the US*.
•Vaccine coverage increased gradually from 0% to ~40% in 10 years
and was distributed evenly across age ranges in the scenario.
•Wesimulated additional scenarios of2xand 0.5x thefinal coverage.
*https://www.cdc.gov/nchs/data/hus/2020 -2021/VaxTn.pdf 9
Vaccine efficacy and vaccine protection assumptions
10
Model and trial estimates of vaccine efficacy for VCD and
hospitalization among seropositive participants .
54100
70 686796
74100
020406080100
1 2 3 46385
43605680
5271
020406080100
1 2 3 4VCD HospitalizationVaccine efficacy
Takeda –ACIP WG presentation September13th –100% VE values are assumed. Biswal (Takeda). ACIP, February 17, 2023.11
Model and trial estimates of vaccine efficacy for VCD and
hospitalization among seronegative participants .
VCD Hospitalization
3096
30
-14588
-16
-106
-150-100-50050100
1 2 3 48599
-303978100
-88100
-150-100-50050100
1 2 3 4Vaccine efficacy
12
Takeda –ACIP WG presentation September 13th–100% VE values are assumed. Biswal (Takeda). ACIP, February 17, 2023.
Vaccine protection assumptions
•Protection varies foreach serotype (DENV 1 -4)and serostatus (seronegative
orseropositive).
•The duration of protection was estimated as
•VCD for seronegatives (13.1yrs) and for seropositives (17.4yrs)
•Hospitalization for seronegatives (19.3) and for seropositives (25.6).
•Equal level of protection against infection and disease given infection.
•In scenarios with pre -vaccination screening, the coverage of pre -vaccination
screening is the same as the vaccine rollout coverage (without screening). The
test has 80% sensitivity and 98% specificity for detecting previous dengue
infection.
•Wesimulated outcomes of VCD and hospitalizations over 10years in
scenarios with vaccination compared to scenarios without vaccination.
•We assumed that all individuals testing positive are vaccinated.
14
Epidemiological outcomes*: VCD and hospitalizations
averted (Benefits)
15*Results are preliminary and are subject to change.
No pre -vaccination screening increased the proportion
of total VCD averted compared to screening .
02468101214161820
1 2 3% of symptomatic VCD averted in the total
population
389
167
16*Results are preliminary and are subject to change.
No pre -vaccination screening increased the proportion of
total hospitalizations averted compared to screening .
02468101214161820
1 2 3% ofhospitalizations averted in the total
population
3 31315
1012
17*Results are preliminary and are subject to change.
Epidemiological outcomes: Additional hospitalizations
(Harms)
18
Outcomes: Additional hospitalizations (Harms)
•We used the ratio of averted hospitalizations to additional
hospitalizations as a relative measure of harm.
•Additional hospitalizations only occurred among seronegative persons
who received the vaccine and were infected with DENV -3 or DENV -4
post -vaccination.
19
Screening increased the ratio of averted to additional
hospitalizations compared to no pre-vaccination screening .
0102030405060708090100
1 2 3Ratio ofhospitalizations averted to
additional hospitalizations*
6:115:135:1
18:1102:1
69:1
*among seronegative persons who received the vaccine and were infected with DENV -3 post -vaccination. Note: Results are prelim inary and are subject to change.
20
Economic outcomes*
21
Weestimated theIncremental Cost -Effectiveness
Ratio to determine cost -effectiveness
•Weestimated thequality -adjusted life-years (QALYs) gained with
routine vaccination using disability weights (Zeng et al. 2018)
•The unit cost perfully vaccinated individual was varied from 100
to 600 USD (330 USD baseline).Costintervention–Costnointervention= ICER
QALYintervention–QALYnointervention
*The cost per fully vaccinated individual was estimated as 150 USD per dose with an additional 15 USD for vaccine administrat ioncosts per dose. All
costs were estimated for the year 2022.
22
Economic outcomes* by scenario for a public
payer perspective, over 10 years, 3% discounting.
Age
groupPre-vaccination
screeningQALYs
gainedICER $ / QALY
gainedICER $ /
hospitalization
averted
4-16 Yes 17 181,918 16,800
4-16 No 35 254,751 46,813
17-60 Yes 89 396,574 48,986
17-60 No 118 314,597 39,886
4-60 Yes 106 384,830 48,305
4-60 No 134 326,412 45,495
23*Results are preliminary and are subject to change.
Summary
•Our model simulations show some benefits from vaccination in terms of
symptomatic cases and hospitalizations averted.
•These benefits depend ontheserostatus ofthevaccinees and the
circulating serotypes.
•Overall, pre-vaccination screening reduces thepotential negative
outcomes inseronegative individuals.
•ICERs per QALY gained and ICER per hospitalization averted were higher
when implementing pre -vaccination screening except in the 4 –16 age
group where screening was more cost -effective.
•Cost -effectiveness of the intervention depended on the
seroprevalence, which is lower for younger age groups.
24
Limitations
•Our model projections are not predictions on the serotype circulation or
theburden of disease.
•Our results on the benefits and cost -effectiveness of the intervention
depend on the circulation of specific serotypes during the projection
period.
•We are not including QALYs loss due to death in the cost -effectiveness
analysis.
24
Acknowledgements
This analysis isbased onprevious analyses described inthe
manuscript byEspaña etal.:“Model -based assessment ofpublic
health impact and cost-effectiveness ofdengue vaccination
following screening for prior exposure. PLOS NTDs 2019” (España
etal.,2019). The manuscript was prepared incollaboration with
other authors: Yutong Yao, Kathryn B.Anderson, Meagan C.
Fitzpatrick, David L.Smith, Amy C.Morrison, Annelies
Wilder-Smith, and Thomas W.Scott.
References I
Zeng, W.,Halasa -Rappel, Y.A.,Durand, L.,Coudeville, L.,&
Shepard, D.S.(2018). Impact ofanonfatal dengue episode
ondisability -adjusted lifeyears: Asystematic analysis. The
American Journal ofTropical Medicine and Hygiene ,99(6),
1458 –1465. https://doi.org/10.4269/ajtmh.18 -0309
Halasa ,Y.A.,Shepard, D.S.,&Zeng, W.(2012). Economic cost of
dengue inPuerto Rico. The American journal oftropical
medicine and hygiene ,86(5),745–752.
España ,G.,Yao, Y.,Anderson, K.B.,Fitzpatrick, M.C., Smith, D.
L.,Morrison, A.C.,Wilder -Smith, A.,Scott, T.W.,&
Perkins, T.A.(2019). Model -based assessment ofpublic
health impact and cost-effectiveness of dengue vaccination
following screening forprior exposure. PLOS Neglected
Tropical Diseases ,13(7),1–21.
https://doi.org/10.1371/journal.pntd.0007482