An incremental economic evaluation of targeted and universal influenza vaccination in pregnant women.
Skedgel, Chris ; Langley, Joanne M. ; MacDonald, Noni E. 等
A recent population-based study in Nova Scotia demonstrated that
women at all stages of pregnancy are at increased risk of serious
respiratory illness during influenza season, even in the absence of
pre-existing co-morbid conditions known to increase the risk of
influenza-associated morbidity. (1) The rate of hospitalization observed
in this study among healthy pregnant women exposed to influenza season
during their third trimester (65/100,000) was comparable to observed
rates among American adults aged 15-44 years with co-morbid medical
conditions (56-110/100,000), (2) a group for whom annual influenza
immunization is already recommended. (3) However, while the US Advisory
Committee on Immunization Practices (3) and the Canadian National
Advisory Committee on Immunization (4) both recommend annual influenza
immunization for all pregnant women, uptake in Nova Scotia is low
(<3%). (5)
Inactivated influenza vaccine has a long record of use in pregnant
women and is considered safe in all stages of pregnancy. (6,7) Influenza
vaccination in pregnant women may help prevent influenza-related
physician and hospital utilization, but the economic implications of
such a policy are unclear. We developed an economic model to estimate
the incremental cost-effectiveness of targeted and universal vaccination
strategies relative to a no-vaccination strategy in pregnant women in
Nova Scotia, Canada.
METHODS
The evaluation compared targeted vaccination of pregnant women with
one or more co-morbidities, universal vaccination of all pregnant women,
and no-vaccination strategies. The evaluation was performed in Excel
(Microsoft; Redmond, Washington) using Palisade Decision Tools
(Palisade; Newfield, New York) to construct the decision tree (Figure 1)
and to perform the probabilistic sensitivity analysis. The decision tree
characterized costs and consequences over a one-year horizon, including
the acquisition and administration costs of vaccination and the costs
and quality-of-life consequences of influenza-related events and
vaccination-related adverse effects. As all events in the evaluation
occurred within one year, neither costs nor outcomes were discounted.
The research project was approved by the Capital District Health
Authority Research Ethics Board.
Event rates
Baseline event rates were derived from an analysis of a
population-based cohort of 134,188 pregnancies extracted from health
administrative databases in Nova Scotia, covering the period 1990-2003
by Dodds et al. (1) These rates were consistent with a Canada-wide study
using different data and methods. (8) The Dodds study stratified women
into two subgroups: those with no co-morbidities and those with one or
more co-morbidities, including pre-existing diabetes, respiratory
disease (including asthma), heart disease, renal disorder or anaemia.
Event rates were calculated as the number of physician visits or
hospital admissions for an influenza-related diagnosis (Table 1, adapted
from Neuzil et al. (9)) concurrent with a pregnancy code, divided by the
number of women in the stratified cohort.
[FIGURE 1 OMITTED]
The evaluation also included a risk of Guillain-Barre syndrome
(GBS) following influenza vaccination or an influenza event with or
without vaccination. GBS risk in the absence of vaccination or an
influenza event, risk following vaccination and risk following an
influenza event were represented as ranges derived from published
literature reviews. (10,11) Pregnant women receiving vaccination and
experiencing an influenza event were assigned the higher of the two
risks.
The effectiveness of vaccination was taken from a single randomized
prospective study of laboratory-confirmed influenza in pregnant women
with and without vaccination. (12) The severity and duration of an
influenza event was assumed to be the same whether or not a woman was
vaccinated.
Costs
The cost of vaccine acquisition was based on costs to the Nova
Scotia Department of Health and Wellness. * Family practitioner (FP)
delivery costs were based on the 2010 Nova Scotia fee schedule. Public
health delivery costs were based on the average cost per vaccination at
clinics conducted by the Department of Health and Wellness * and were
consistent with previously published Canadian costs. (13) The cost of
influenza-related physician utilization was derived from the 2005/06
Nova Scotia physician claims database. Hospital costs were derived from
the 2006/07 Ontario Case Cost Initiative database. The annual cost of
GBS was taken from a US evaluation of influenza vaccination. (14) Costs
were adjusted to 2010 Canadian dollars based on the consumer price
index, health component. (15) The evaluation took a health system payer
perspective as this was felt to be most relevant to public health
authorities considering an immunization strategy.
Quality of life
As there was no mortality observed in the study cohort, the key
outcome in the evaluation was the quality-of-life improvement due to
influenza-related events prevented. Baseline utility (no influenza) was
0.95, reflecting the average utility of all individuals more than 12
years of age with no chronic conditions. (16) Utility weights for
influenza-related illnesses were derived from a Canadian study by
O'Brien et al. (17) As we were unable to find estimates of the
utility associated with GBS, we assumed a conservative weight of 0.25.
Economic analysis
The evaluation was conducted as an incremental cost-effectiveness
analysis, comparing each vaccination strategy to the next best
alternative. Key economic outcomes were the net cost of vaccination
(vaccination costs less event costs avoided), net quality-adjusted life
years (QALYs) gained and the incremental cost per QALY gained.
Estimating cost-effectiveness in terms of cost per QALY gained
facilitates economic comparisons across different programs and diseases.
All costs and outcomes were reported on the basis of the average annual
cohort of pregnant women, calculated as the total number of pregnancies
observed over 1990-2003 divided by the number of years of observation.
(1) Reflecting a decision-making perspective, all conclusions were based
on expected values. (18)
Sensitivity analysis
The base-case scenario assumed all vaccinations were delivered by a
family practitioner (FP) as part of a routine visit, but sensitivity
analyses considered alternative modes of delivery (extra FP visit,
public health clinic). Threshold analyses were conducted on key
parameters to identify the threshold values necessary to meet specific
cost-effectiveness targets. One-way sensitivity analyses were conducted
on other key parameters. Probabilistic sensitivity analysis was used to
incorporate uncertainty into the economic evaluation. (19,20) Parameter
point estimates, standard deviation and probability distributions are
shown in Table 2. Uncertainty was expressed in terms of confidence
intervals around point estimates.
RESULTS
The average annual cohort of pregnant women was 10,316, but for
simplicity this was rounded to 10,000. Relative to a no-vaccination
strategy, a targeted vaccination strategy for pregnant women with at
least one co-morbidity delivered as part of a routine FP visit was
cost-saving. Relative to the targeted strategy, the universal strategy
had an incremental cost-effectiveness of $39,942 per QALY gained. The
expected budget impact of the targeted strategy was -$9,485, while the
universal strategy had a net budget impact of $81,658 relative to the
no-vaccination strategy. Results and associated 95% confidence intervals
are shown in Table 3.
Threshold analysis on vaccine effectiveness showed that the
targeted strategy would be economically dominant (cost saving, improved
outcomes) over no vaccination with a vaccine effectiveness (i.e.,
relative risk) less than 0.76 and would meet a $50,000 per QALY gained
threshold with an effectiveness less than 0.84. A universal strategy
would meet a $50,000 threshold with an effectiveness less than 0.68 and
a $100,000 threshold with an effectiveness less than 0.80.
One-way sensitivity analysis on mode of delivery suggests that the
targeted strategy would remain dominant relative to the novaccination
strategy when delivered by public health clinics, while the universal
strategy would be strongly cost-effective, bordering on dominant,
relative to the targeted strategy. If vaccination required an additional
FP visit, the targeted vaccination strategy would lose its dominance and
have a cost-effectiveness of $62,796, while the cost-effectiveness of
the universal strategy would increase to more than $150,000. Other
one-way sensitivity analyses are shown in Figure 2, sorted by their
impact on the cost-effectiveness of a universal strategy. The targeted
strategy remained dominant across all ranges.
DISCUSSION
Two recent evaluations of universal influenza vaccination in
pregnant women compared to a "no vaccination" strategy have
found universal strategies to be economically attractive. Roberts et al.
(14) reported universal vaccination of all pregnant women was dominant
relative to no vaccination, while Beigi et al. (21) reported a universal
vaccination strategy ranged from dominant to strongly cost-effective
relative to no vaccination, depending on prevalence and mortality
assumptions. However, we suggest that comparison of a universal strategy
to a no-vaccination alternative risks overstating the benefit of the
universal strategy, since it is generally accepted that targeted
programs (influenza immunization of pregnant women with risk factors for
complicated influenza) are already standard practice. Our analysis is
unique in conducting an incremental comparison of the costs and benefits
of a universal strategy relative to both a targeted and a no-vaccination
strategy. It is methodologically more appropriate to consider the
incremental costs and benefits of a universal strategy relative to a
targeted strategy rather than the dominated no-vaccination comparator.
(22)
[FIGURE 2 OMITTED]
The use of a population-based cohort was a strength of our
analysis. This dataset included 134,188 pregnancies over 14 years and
allowed us to track individual-level physician and hospital utilization.
It was not possible to reliably identify and exclude vaccinated women
from the cohort when calculating baseline event rates, but since only
2.6% of pregnant women in Nova Scotia were vaccinated over the period
covered by our data, (5) this is unlikely to have influenced our
estimates. The inclusion of vaccinated women would likely underestimate
the baseline risk of an influenza-related event and result in a more
conservative estimate of benefit from vaccination. Also, although
illnesses attributed to influenza were not laboratory-confirmed in this
study, our incidence rates are consistent with estimates from another
study with a different design. (8)
Evidence around the precise risk reduction associated with
influenza vaccination in pregnant women is limited. In the only
prospective randomized controlled trial, Zaman et al. (12) reported a
relative risk (RR) of 0.64 for laboratory-confirmed influenza in
vaccinated pregnant women. Their estimate is similar to a comprehensive
Cochrane review of vaccine effectiveness in healthy adults that found a
RR of 0.77 (95% CI 0.69-0.86) for clinically-defined influenza and an RR
of 0.65 (95% CI 0.34-1.22) for influenza-related hospitalizations. (23)
A 2010 review by Jefferson et al. (24) reported lower effectiveness but
excluded all industry-funded studies. Sensitivity analysis on this
critical parameter suggests that a targeted strategy would be dominant
or strongly cost-effective, and a universal strategy would meet a
$100,000 per QALY gained threshold, at all these RR point estimates,
although a universal strategy would not be cost-effective at the upper
extremes of the 95% confidence intervals.
The risk of GBS following influenza vaccination is uncertain, but
clinical evidence suggests that vaccination is protective against GBS,
supporting vaccination strategies. (11) Owing to the protective benefit
associated with vaccination, the relative cost-effectiveness of both the
targeted and universal strategies improve as either the baseline risk of
GBS or the associated costs increase. As such, our one-year horizon
presents a conservative estimate of the broader benefits of vaccination.
Our estimates also exclude savings related to prevented complications
related to the child or the pregnancy itself--even though a recent
case-control study estimated influenza vaccination was 91.5% effective
at preventing influenza-related hospitalization within the first 6
months of life (25)--as well as the value of lost productivity and the
social and emotional costs associated with illness in pregnancy. It is
also worth noting that although our estimates of budget impact are
relative to a no-vaccination alternative, Nova Scotia does have an ad
hoc vaccination program. The budget impact of moving to a universal
strategy, therefore, would likely be less than our estimates suggest.
Our analysis suggests that a strategy of targeted vaccination in
pregnant women with at least one co-morbidity could be economically
dominant, and that a strategy of universal vaccination of all pregnant
women could be cost-effective relative to a targeted strategy. It also
shows that a universal strategy could be delivered with a reasonable
budget impact by public health clinics or by FPs as part of a routine
office visit. Precise cost-effectiveness will vary by jurisdiction, but
sensitivity analysis suggests that our results are robust across a range
of costs and risks. Both strategies have potential advantages and
disadvantages. For example, targeted strategies may be associated with
less vaccine uptake than universal strategies, while universal
strategies will have greater costs. The economically preferred strategy
must be considered within the context of affordability, potential
coverage rates and the overall health benefit.
Conflict of Interest: None to declare.
Received: August 11, 2010
Accepted: June 3, 2011
REFERENCES
(1.) Dodds L, McNeil SA, Fell DB, Allen VM, Coombs A, Scott J, et
al. Impact of influenza exposure on rates of hospital admissions and
physician visits because of respiratory illness among pregnant women.
CMAJ2007;176(4):463-68.
(2.) Barker WH, Mullooly JP. Impact of epidemic type A influenza in
a defined adult population. Am J Epidemiol 1980;112(6):798-811.
(3.) Centers for Disease Control and Prevention. Prevention and
Control of Seasonal Influenza with Vaccines: Recommentations of the
Advisory Committee on Immunization Practices (ACIP), 2009. MMWR
2009;58(No. RR-8):1-52.
(4.) Addendum. National Advisory Committee on Immunization (NACI)
Statement on Influenza Vaccination for the 2007-2008 Season. Can Commun
Dis Rep 2007;33(11):23-24.
(5.) McNeil SA, Dodds L, Allen VM, Scott J, Halperin B, MacDonald
N. Influenza vaccine programs and pregnancy: New Canadian evidence for
immunization. J Obstet Gynaecol Can 2007;29(8):674-76.
(6.) Munoz FM, Greisinger AJ, Wehmanen OA, Mouzoon ME, Hoyle JC,
Smith FA, et al. Safety of influenza vaccination during pregnancy. Am J
Obstet Gynecol 2005;192(4):1098-106.
(7.) Mak TK, Mangtani P, Leese J, Watson JM, Pfeifer D. Influenza
vaccination in pregnancy: Current evidence and selected national
policies. Lancet Infect Dis 2008;8(1):44-52.
(8.) Schanzer DL, Langley JM, Tam TW. Influenza-attributed
hospitalization rates among pregnant women in Canada 1994-2000. J Obstet
Gynaecol Can 2007;29(8):622-29.
(9.) Neuzil KM, Reed GW, Mitchel EF, Simonsen L, Griffin MR. Impact
of influenza on acute cardiopulmonary hospitalizations in pregnant
women. Am J Epidemiol 1998;148(11):1094-102.
(10.) Hughes RAC, Rees JH. Clinical and epidemiologic features of
Guillain-Barre Syndrome. J Infect Dis1997;176:S92-S98.
(11.) Lehmann HC, Hartung H, Kieseier BC, Hughes RA. Guillain-Barre
syndrome after exposure to influenza virus. Lancet Infect Dis
2010;10(9):643-51.
(12.) Zaman K, Roy E, Arifeen SE, Rahman M, Raqib R, Wilson E, et
al. Effectiveness of maternal influenza immunization in mothers and
infants. N Engl J Med 2008;359(15):1555-64.
(13.) Mercer NJ. Cost analysis of public health influenza vaccine
clinics in Ontario. Can J Public Health 2009;100(5):340-43.
(14.) Roberts S, Hollier LM, Sheffield J, Laibl V, Wendel GD, Jr.
Cost-effectiveness of universal influenza vaccination in a pregnant
population. Obstet Gynecol 2006;107(6):1323-29.
(15.) Consumer Price Index, Health and Personal Care, Nova Scotia
[homepage on the Internet]. Available at:
http://www40.statcan.ca/l01/cst01/econ161d.htm (Accessed 28 July 2010).
(16.) Schultz SE, Kopec JA. Impact of chronic conditions. Health
Rep 2003;14(4):4153.
(17.) O'Brien BJ, Goeree R, Blackhouse G, Smieja M, Loeb M.
Oseltamivir for treatment of influenza in healthy adults: Pooled trial
evidence and cost-effectiveness model for Canada. Value Health
2003;6(2):116-25.
(18.) Claxton K. The irrelevance of inference: A decision-making
approach to the stochastic evaluation of health care technologies. J
Health Econ 1999;18(3):341-64.
(19.) Briggs A, Claxton K, Sculpher MJ. Decision Modelling Methods
for Health Economic Evaluation. Oxford: Oxford University Press, 2006.
(20.) O'Brien BJ, Briggs AH. Analysis of uncertainty in health
care cost-effectiveness studies: An introduction to statistical issues
and methods. Stat Methods Med Res 2002;11(6):455-68.
(21.) Beigi RH, Wiringa AE, Bailey RR, Assi TM, Lee BY. Economic
value of seasonal and pandemic influenza vaccination during pregnancy.
Clin Infect Dis 2009;49(12):1784-92.
(22.) Drummond MF, Sculpher MJ, Torrance GW, O'Brien BJ,
Stoddart GL. Methods for the Economic Evaluation of Health Care
Programmes, 3rd Ed. Oxford: Oxford University Press, 2005.
(23.) Demicheli V, Rivetti D, Deeks JJ, Jefferson TO. Vaccines for
preventing influenza in healthy adults. Cochrane Database Syst Rev
2004(3):CD001269.
(24.) Jefferson T, Di Pietrantonj CD, Rivetti A, Bawazeer GA,
Al-Ansary LA, Ferroni E. Vaccines for preventing influenza in healthy
adults. In: Jefferson T (Ed.), The Cochrane Collaboration, Cochrane
Database of Systematic Reviews. Chichester, UK: John Wiley & Sons,
Ltd., 2010.
(25.) Benowitz I, Esposito DB, Gracey KD, Shapiro ED, Vazquez M.
Influenza vaccine given to pregnant women reduces hospitalization due to
influenza in their infants. Clin Infect Dis 2010;51(12):1355-61.
* Personal Communication. D Vaughan, Nova Scotia Dept. of Health
Promotion & Protection, 2009. July 27, 2010).
Correspondence: Chris Skedgel, Centre for Clinical Research, Room
242, 5790 University Avenue, Halifax, NS B3H 1V7, Tel: 902-473-3303,
Fax: 902-4251611, E-mail: chris.skedgel@cdha.nshealth.ca
Funding: Capital Health Research Fund.
Chris Skedgel, MDE, [1,2] Joanne M. Langley, MD, [1,3,4] Noni E.
MacDonald, MD, [1,3] Jeff Scott, MD, [1,3,5] Shelly McNeil, MD [1-3]
Author Affiliations
[1.] Canadian Center for Vaccinology, IWK Health Centre, Capital
District Health Authority and Dalhousie University, Halifax, NS
[2.] Department of Medicine, Dalhousie University & Capital
District Health Authority, Halifax, NS
[3.] Department of Pediatrics, Dalhousie University & IWK
Health Centre, Halifax, NS
[4.] Department of Community Health and Epidemiology, Dalhousie
University, Halifax, NS
[5.] Department of Emergency Medicine, Dalhousie University,
Halifax, NS
Table 1. Influenza-related Respiratory Diagnoses
Disease ICD-9 ICD-10
Viral pneumonia 480 J12
Pneumonia due to adenovirus 480.0 J12.0
Pneumonia due to respiratory 480.1 J12.1
synctial virus
Pneumonia due to 480.2 J12.2
parinfluenza virus
Pneumonia due to other 480.8 J12.8
virus, not classified
elsewhere
Viral pneumonia, unspecified 480.9 J12.9
Pneumococcal and other 481,482 J13, J15
bacterial pneumonia
Pneumonia due to other 483 J16.8
specified organism
Pneumonia in infectious 484 J17
diseases classified elsewhere
Bronchopneumonia, organism 485 J18.0
unspecified
Pneumonia, organism 486 J18.9
unspecified
Influenza 487 J10, J11
With pneumonia 487.0 J10.0, J11.0
With other respiratory 487.1 J10.1, J11.1
manifestations
With other manifestations 487.8 J10.8, J11.8
Acute nasopharyngitis 460 J00
Acute sinusitis 461 J01
Acute pharyngitis 462 J02
Acute tonsillitis 463 J03
Acute laryngitis and 464 J04
tracheitis
Acute upper respiratory 465 J06
infections of multiple sites
Acute bronchitis and 466 J20, J21
bronchiolitis
Bronchitis, not specified as 490 J22
acute or chronic
Chronic bronchitis 491 J42
Acute myocarditis 422 I40
Heart failure 428 I50
Adapted from Neuzil et al., 1998 (see ref. 9).
Table 2. Model Inputs
Deterministic Parameters Expected
Value
Vaccine acquisition cost $3.64
Vaccine delivery costs
FP office visit $29.64
Tray fee $3.42
Injection fee $13.68
Public health clinic, $6.75
cost per vaccination
Probabilistic Expected Std Dev Distribution
Parameters Value or Range
Proportion of 0.101 0.001 Beta
pregnant women with
[greater than or
equal to]1 co-
morbidity
Baseline probability 0.21 0.001 Beta
of physician event,
0 co-morbidities
Baseline probability 0.28 0.004 Beta
of physician event,
[greater than or
equal to]1 co-
morbidity
Expected cost of $31.82 $4.70 Log Normal
physician event
Baseline probability 0.003 0.000 Beta
of hospital event, 0
co-morbidities
Baseline probability 0.014 0.001 Beta
of hospital event,
[greater than or
equal to]1 co-
morbidity
Expected LOS per 2.97 2.64 Log Normal
influenza-related
event
Expected cost of $4,464 $5,995 Log Normal
hospital event
Relative risk of an 0.11 Beta/Beta
event with
vaccination (vaccine
effectiveness) 0.64
Baseline utility 0.95 1.00-0.90 Triangle
weight
Relative utility 0.58 0.97-0.77 Triangle
weight, influenza *
Relative utility 0.25 0.10-0.40 Triangle
weight, Guillain-
Barre Syndrome *
Guillain-Barre 1.90 0.4-4.0 Triangle
Syndrome per 100,000
| No Vaccination (1)
Guillain-Barre 0.54 0-1.07 Uniform
Syndrome per 100,000
| Vaccination
Relative risk of GBS 15.32 8.09 Beta
| Influenza Event
(2)
Guillain-Barre 29.10 21.28 Derived from
Syndrome per 100,000 (1) & (2)
| Influenza
Annual Cost, $135,464 $30,000 Log Normal
Guillain-Barre
Syndrome
Deterministic Parameters Source
Vaccine acquisition cost 13
Vaccine delivery costs
FP office visit NS Fee Schedule
Tray fee NS Fee Schedule
Injection fee NS Fee Schedule
Public health clinic, 13
cost per vaccination
Probabilistic Lower Upper Source
Parameters 95% CI 95% CI
Proportion of 0.099 0.102 1
pregnant women with
[greater than or
equal to]1 co-
morbidity
Baseline probability 0.24 0.24 1
of physician event,
0 co-morbidities
Baseline probability 0.33 0.34 1
of physician event,
[greater than or
equal to]1 co-
morbidity
Expected cost of $23.70 $41.70 PHRU
physician event
Baseline probability 0.002 0.003 1
of hospital event, 0
co-morbidities
Baseline probability 0.012 0.016 1
of hospital event,
[greater than or
equal to]1 co-
morbidity
Expected LOS per 0.44 10.24 OCCI
influenza-related
event
Expected cost of $315 $17,773 OCCI
hospital event
Relative risk of an 0.45 0.90 12
event with
vaccination (vaccine
effectiveness) 0.64
Baseline utility 0.91 0.99 17
weight
Relative utility 0.50 0.66 18
weight, influenza *
Relative utility 0.13 0.37 Assumption
weight, Guillain-
Barre Syndrome *
Guillain-Barre 0.68 3.50 10
Syndrome per 100,000
| No Vaccination (1)
Guillain-Barre 0.27 10.43 11
Syndrome per 100,000
| Vaccination
Relative risk of GBS 6.80 38.20 11
| Influenza Event
(2)
Guillain-Barre 7.90 87.30
Syndrome per 100,000
| Influenza
Annual Cost, $92,237 $189,539 15
Guillain-Barre
Syndrome
* Health-related utility is calculated assuming a multiplicative
utility function, where the utility associated with any particular
state is the product of baseline utility and specific health state
utility (e.g., the health-related utility associated with influenza is
0.95 x 0.58 = 0.551). The utility penalty associated with influenza is
calculated as the difference between baseline utility and the
multiplicative utility associated with influenza, weighted by
influenza duration (i.e., (0.551-0.95)*(2.97/365)=-0.003).
LOS=length of stay; PHRU=Population Health Research Unit,
Dalhousie University; OCCI=Ontario Case Cost Initiative
Table 3. Results
Women Cohort Incremental
Vaccinated Cost Cost
(95% CI)
No Vaccination 0 $344,878 --
Targeted Strategy 1002 $335,392 -$9,485
(-$65,993-$14,177)
Universal Vaccination 10,000 $426,536 $91,143
(-$22,546-$152,454)
Total Incremental Cost per
Cohort QALYs QALYs QALY
(95% CI)
No Vaccination 9,492.23 -- --
Targeted Strategy 9,492.55 0.32 Dominant
(0.06-0.88)
Universal Vaccination 9,494.83 2.28 $39,942
(0.44-6.18)
Gained
Budget
Impact *
No Vaccination --
Targeted Strategy -$9,485
Universal Vaccination $81,658
* Budget impact is relative to a no-vaccination strategy and may
therefore be less than incremental cost.
QALY=quality-adjusted life year.
