Cost Influenza
Cost Influenza
Abstract
Background: In July 2000, the province of Ontario, Canada, initiated a universal influenza immunization program (UIIP) to
provide free seasonal influenza vaccines for the entire population. This is the first large-scale program of its kind worldwide.
The objective of this study was to conduct an economic appraisal of Ontario’s UIIP compared to a targeted influenza
immunization program (TIIP).
Methods and Findings: A cost-utility analysis using Ontario health administrative data was performed. The study was
informed by a companion ecological study comparing physician visits, emergency department visits, hospitalizations, and
deaths between 1997 and 2004 in Ontario and nine other Canadian provinces offering targeted immunization programs.
The relative change estimates from pre-2000 to post-2000 as observed in other provinces were applied to pre-UIIP Ontario
event rates to calculate the expected number of events had Ontario continued to offer targeted immunization. Main
outcome measures were quality-adjusted life years (QALYs), costs in 2006 Canadian dollars, and incremental cost-utility
ratios (incremental cost per QALY gained). Program and other costs were drawn from Ontario sources. Utility weights were
obtained from the literature. The incremental cost of the program per QALY gained was calculated from the health care
payer perspective. Ontario’s UIIP costs approximately twice as much as a targeted program but reduces influenza cases by
61% and mortality by 28%, saving an estimated 1,134 QALYs per season overall. Reducing influenza cases decreases health
care services cost by 52%. Most cost savings can be attributed to hospitalizations avoided. The incremental cost-
effectiveness ratio is Can$10,797/QALY gained. Results are most sensitive to immunization cost and number of deaths
averted.
Conclusions: Universal immunization against seasonal influenza was estimated to be an economically attractive
intervention.
Please see later in the article for the Editors’ Summary.
Citation: Sander B, Kwong JC, Bauch CT, Maetzel A, McGeer A, et al. (2010) Economic Appraisal of Ontario’s Universal Influenza Immunization Program: A Cost-
Utility Analysis. PLoS Med 7(4): e1000256. doi:10.1371/journal.pmed.1000256
Academic Editor: Joshua A. Salomon, Harvard School of Public Health, United States of America
Received August 28, 2009; Accepted February 25, 2010; Published April 6, 2010
This is an open-access article distributed under the terms of the Creative Commons Public Domain declaration which stipulates that, once placed in the public
domain, this work may be freely reproduced, distributed, transmitted, modified, built upon, or otherwise used by anyone for any lawful purpose.
Funding: This study was supported by an operating grant from the University of Toronto Research Program, which provided fellowship support for BA, and by
an operating grant from the Public Health Agency of Canada. JCK was supported by a Canadian Institutes of Health Research Fellowship Award at the time of the
study and is currently supported by an Ontario Ministry of Health and Long Term Care (MOHLTC) Career Scientist Award and a University of Toronto Department
of Family and Community Medicine Research Scholar Award. This work was made possible with the support of the Institute for Clinical Evaluative Sciences (ICES),
which is funded in part by the Ontario MOHLTC. The opinions, results and conclusions reported in this paper are those of the authors and are independent from
the funding sources. No endorsement by ICES or the Ontario MOHLTC is intended or should be inferred. The funders had no role in the design and conduct of the
study; collection, management, analysis, and interpretation of the data; or preparation, review, or approval of the manuscript. The researchers are independent
from the funders.
Competing Interests: BS has served as consultant to F. Hoffmann La-Roche for economic evaluations of Tamiflu (unrestricted funding). BS has received travel
grants and honoraria for speaking at meetings on pandemic influenza mitigation strategies from F. Hoffmann La-Roche.
Abbreviations: CMG, Case Mix Group; DAD, Discharge Abstract Database; ED, emergency department; GBS, Guillain-Barré syndrome; ICUR, incremental cost-
utility ratio; MOHLTC, Ministry of Health and Long-Term Care; NICE, National Institute for Health and Clinical Excellence; OHIP, Ontario Health Insurance Plan;
QALY, quality-adjusted life year; RIW, resource intensity weight; RR, relative rate; SIR, Susceptible-Infected-Removed; TIIP, targeted influenza immunization
program; UIIP, universal influenza immunization program
* E-mail: [email protected]
Table 1. Mean annual influenza-related event rates and relative rates comparing post-UIIP to pre-UIIP event rates.
Other Provinces
Pre-UIIP Post-UIIP Ontario Combined
Office visits 813.58 (807.90; 819.20) 173.00 (169.90; 176.10) 0.21 (0.21; 0.22) 0.52 (0.51; 0.53)
ED visits 139.57 (137.50; 141.70) 43.57 (42.20; 44.90) 0.31 (0.30; 0.32) 0.69 (0.67; 0.70)
Hospitalizations 33.38 (32.20; 34.60) 8.49 (7.70; 9.30) 0.25 (0.23; 0.28) 0.44 (0.42; 0.46)
Deaths 12.00 (11.05; 12.96) 2.79 (2.15; 3.42) 0.23 (0.18; 0.30) 0.33 (0.28; 0.39)
31% for ED use, and 79% versus 48% for doctors’ office visits To calculate QALYs lost due to premature death (Table 2), we
[13]. estimated influenza-associated deaths using mortality data from
Statistics Canada’s Mortality Database [13]. We considered deaths
Quality-Adjusted Life Years (QALYs) Lost from Influenza due to all respiratory and circulatory conditions for seven influenza
The administrative data used to estimate resource use do not seasons covering the years from 1997 to 2004. After UIIP
yield a count of influenza cases. The number of cases of influenza introduction, influenza-associated mortality decreased 77% in
requiring health care was estimated using the number of office visits Ontario compared to 46% in other provinces [13].
and ED visits and excluding repeat visits by the same patient for the The average life expectancy was estimated by age [19] and
same diagnosis within a 21-d window following an initial visit. adjusted for quality of life, using utility scores from a community-
According to this approximation, 90% of all influenza-associated dwelling population, ranging from 0.88 to 0.94 depending on age
office visits, 85% of all influenza-associated ED visits, and all [20].
influenza-associated hospitalizations were deemed to be discrete Quality-adjusted life expectancy was discounted at 3% per year
new cases of influenza. The remainder of visits were considered to in the base case analysis [21].
be repeat visits. Multiple visits across settings by the same patient
were not removed since the number of those was expected to be Unit Costs: UIIP Program, Physician Services, and
small (for example, only 0.07% of all health care contacts were Hospitalizations
hospitalizations, some of which may have had an office and/or ED All costs were obtained from Ontario sources and are expressed
visit before being hospitalized). It was assumed that patients who die in 2006 Canadian dollars. Cost data are summarized in Table 2.
due to influenza had at least one health care contact before death. The total program cost for the universal program was $40
Therefore, these were also not separately counted as cases as they million per year (approximately $3.96 per dose and $7.55 in total
were presumed to have been accounted for. per dose distributed) for each year since introducing the program
Quality weights as reported by Turner [16] were used to (Nancy Peroff-Johnston, Ministry of Health and Long-Term Care
estimate QALYs lost per case of influenza due to morbidity. In this [MOHLTC], personal communication, May 11, 2007). This
recent assessment by the National Institute for Health and Clinical included the cost of the vaccine, health care provider reimburse-
Excellence (NICE) of antiviral treatment of influenza, utilities for ment for vaccination, communications strategies, and direct
influenza were estimated from patient health state valuations operating expenditures for the MOHLTC including staffing.
reported daily for 21 d in oseltamivir clinical trials. Utilities were Approximately 50% of the total budget was spent on the vaccine
generated by recalibrating the Likert scores as obtained in the and the remaining 50% on all other cost items. To calculate the
clinical trials to mean visual analogue scale scores, which were expected targeted program (TIIP) cost if universal immunization
then transformed to time trade-off scores. Quality of life data were (UIIP) had not been implemented, the relative change in vaccine
available for ‘‘otherwise healthy adults,’’ ‘‘adults with co- coverage (pre-2000 versus post-2000) observed in other provinces
morbidities,’’ and the ‘‘elderly.’’ In the absence of utility data for was applied to Ontario’s pre-2000 vaccine coverage rate. Vaccine
children, it was assumed that children have similar utility weights coverage rates for the population 12 y or older more than doubled
as ‘‘otherwise healthy adults’’ as both groups are similar in other (18% to 42%) in Ontario and other provinces (13% to 28%)
health outcome measures, such as time to return to normal activity between 1996/1997 and 2005. The Ontario pre-2000 TIIP
[17]. The utility weight for adults with co-morbidities as obtained program cost (Nancy Peroff-Johnston, MOHLTC, personal
in the clinical trials was applied to the 18.8% of the Ontario communication, May 11, 2007) per person vaccinated was used,
population aged 12 to 64 y deemed high risk because of chronic inflated to 2006, and applied to the population who would have
conditions [18]. Patients in the trials were recruited from Europe been covered under a targeted program [22]. This assumes that
and North America. We assume that the clinical trial population is the average cost per person (including vaccine, vaccine delivery,
sufficiently similar to the Ontario population (high income and other cost to the Ministry) remains the same but is applied to a
Western countries with similar life expectancy and population larger number of persons receiving influenza immunization. For
health status) so that we can apply the utility estimates to the an expected TIIP coverage rate of 40%, the total program cost for
population under consideration here. The QALY gain per targeted immunization was expected to be $20 million if TIIP had
influenza case prevented (Table 2) was calculated by multiplying been continued.
the duration of the symptomatic period by the utility decrement Unit costs for physician services (office visits, ED visits, and in-
associated with influenza-related illness. hospital services) were obtained from the OHIP dataset. A location
Base Case
Parameter Analysis, Mean Deterministic Sensitivity Analysis
Source: quality of life with influenza: Turner [16]; life expectancy: Statistics Canada [19]; quality of life adjustment by age in the absence of influenza: Mittmann [20];
office visits, ED visits, hospitalizations: Ontario Health Insurance Plan (OHIP) dataset; TIIP: calculated based on influenza immunization program cost before
implementation of UIIP, vaccine coverage rates in Ontario pre-UIIP, and relative increase of influenza immunization coverage in other provinces [13,56] (Nancy Peroff-
Johnston, MOHLTC, personal communication, May 11, 2007); UIIP: Ministry of Health and Long-Term Care [56] (Nancy Peroff-Johnston, MOHLTC, personal
communication, May 11, 2007).
ED, emergency department; TIIP, targeted influenza immunization program; UIIP, universal influenza immunization program.
doi:10.1371/journal.pmed.1000256.t002
code in the dataset indicates whether the services were provided in Hospitalization costs were costed using the resource intensity
the physician office, ED, or hospital. weight (RIW) approach [24,25]. The RIW is the ratio of the cost of
Costs reflect the fee paid by condition type (pneumonia and a case in a Case Mix Group (CMG) to the average cost of all cases.
influenza). Mean fees paid ($35 per office visit, $54 per ED visit) The mean RIWs for pneumonia- or influenza-related hospitaliza-
were used in the base case analysis; 95% CI defined the lower and tions were extracted from the acute care Discharge Abstract
upper limit for deterministic sensitivity analysis. For probabilistic Database (DAD). The provincial cost per weighted case of $4,732
analysis the primary data were sampled. An additional $166 per was applied to the mean RIW of influenza-associated hospitaliza-
ED visit was added for non-physician costs to calculate total cost tions of 1.33 [23]. Mean physician charges for hospital inpatient
per ED visit [23]. services for pneumonia or influenza (obtained from the OHIP
database using location code) were added. On average, there were analysis. For probabilistic analysis, the primary data were sampled.
1.65 claims for physician services per hospitalization at a cost of Distributions for utility weights were based on confidence intervals
$66.41 each. The average length of stay for pneumonia and in published reports [16]. The probability of Ontario’s UIIP being
influenza is 6 d [26]. The total mean cost of a pneumonia- and cost-effective for a range of willingness to pay thresholds ($0 to
influenza-related episode of hospitalization was therefore $6,418. $100,000 per QALY gained) was calculated by sampling from
these distributions running 1,000 trials. To ensure internal
Analyses consistency, the same random number was used to sample from
Cost-effectiveness. Cost-effectiveness of UIIP versus TIIP distributions for the different age brackets within the same group
was measured by the incremental cost-utility ratio (ICUR), defined of variables: pre-UIIP events, post-UIIP events, RR, and QALY
as the additional cost per QALY gained. The health benefits of gain due to morbidity averted.
universal immunization were estimated as number of influenza An additional analysis, using the diagnostic codes for all
cases, health services utilization (physician visits, ED visits, respiratory conditions to extract health care utilization data, is
hospitalizations), and deaths prevented in a typical influenza presented in the Text S2. This analysis was performed because
season. A time horizon of a lifetime was adopted to calculate many influenza-related health care interactions are coded under
QALYs lost attributable to influenza-related death. Time different diagnostic codes (e.g., common cold, asthma, COPD,
preference was incorporated by discounting QALYs lost at 3% otitis media, etc.), therefore underestimating the health care
annually in the base case analysis [21]. Costs are not discounted resource use due to influenza if using the diagnostic codes for
because all costs occur within 1 y. The analysis was performed from pneumonia and influenza only (as in all other analyses).
the perspective of the health care payer, the Ontario MOHLTC.
Sensitivity analysis. The impact of data uncertainty was Results
explored with one-way and probabilistic sensitivity analyses. One-
way sensitivity analyses were performed for numbers of events, Base Case
number of cases, event costs, immunization program cost, QALY In Ontario, 22,457 cases of influenza were observed on average
penalty associated with influenza, and discount rate (undiscounted per season after the introduction of the universal immunization
and 5%). Grouped sensitivity analysis was performed by program. If TIIP had been continued, the expected average
simultaneously adjusting the values of (1) events (all events and number of cases per season was estimated to be 56,998. UIIP
deaths separate), (2) relative rate (RR) of events (office visits, ED therefore prevented 34,541 influenza cases (61% of all cases each
visits, hospitalizations, and deaths, each event separate), (3) costs of season).
events (office visits, ED visits, and hospitalizations, each event Ontario’s UIIP also prevented 111 deaths, a 28% reduction in
separate), and (4) QALY gain due to morbidity (illness) averted. mortality. This resulted in a projected 1,134 QALYs gained in
This was achieved by the use of scaling parameters that link total or 0.09 quality-adjusted life days per person vaccinated.
variables within a group and allow the simultaneous adjustment of Approximately half of all health gains (QALYs) were associated
their values over a defined range. The base case value was with a reduction of influenza mortality; the other half was
therefore multiplied with scaling parameter values ranging from 0 associated with reduction in influenza-related morbidity.
to 2. This range covers the 95% CI of all parameters except for The program costs of UIIP are high, approximately double that
deaths observed post-UIIP in persons under 65 y of age. The of a targeted program ($40 million versus $20 million). However,
range for post-UIIP deaths is 24 to 6 for ,65 y old. This age UIIP was estimated to prevent 786 influenza-related hospitaliza-
group (,65 y) accounts for approximately 2% of all deaths (7 out tions, 7,745 influenza-related ED visits, and 30,306 office visits per
of 309 deaths per season). We therefore assumed the range from 0 season. Preventing influenza cases effectively reduced influenza-
to 2 for scaling parameters to be acceptable for varying post-UIIP related health care costs by 52%, saving the health care system
deaths for all age groups combined (not individually). Two approximately $7.8 million per season, so that the net cost of the
extreme (best and worst) case scenarios were also performed UIIP program is $12.2 million, or $2.60 per person vaccinated.
(Tables 1 and 2; more detailed information can be found in Tables The cost per QALY gained is $10,797/QALY (discount rate
S1 and S2) biasing the analysis for and against UIIP. The 95% CIs 3%). All base case results are presented in Table 3 and in more
were used as extreme values for pre- and post-UIIP number of detail in Table S3.
events (cases, physician services, ED visits, hospitalizations,
deaths), RRs of events (to calculate number of events if TIIP Sensitivity Analysis
would have been continued), unit costs, and QALY loss Deterministic one-way sensitivity analysis (Figure 1, Table S4)
attributable to influenza morbidity. revealed that results are highly sensitive to the RR of deaths if
For probabilistic analysis, distributions were assigned to key TIIP had been continued, the pre-UIIP and post-UIIP number of
variables in the model (Tables S1 and S2). Distributions for events deaths, and moderately sensitive to UIIP and TIIP immunization
(office visits, ED visits, hospitalizations, and deaths), RRs, and unit program cost. However, the ICER remains below $50,000 per
cost for events were drawn from the original data. For unit costs QALY under the following conservative individual conditions: (1)
for physician services, the actual administrative data were sampled UIIP costs are twice the current program cost, (2) TIIP costs are
using $10.00 increments, with the last bin being $220 per visit and zero, (3) the RR of death is not smaller than 45% of the base case
greater (fees paid: $0 to ,$10, $10 to ,$20, $20 to ,$30, … , value, (4) the pre-UIIP number of deaths is not smaller than 45%
$210 to ,$220, $220 and greater). The midpoint for the top and of the base case value, or (5) the post-UIIP number of deaths is not
bottom categories was adjusted by the minimum and maximum greater than 190% of the base case value.
fee charged. The midpoint for the bottom category was $5.50 for It is highly unlikely that these thresholds for key parameters are
office visits, $9.00 for ED visits, and $6.50 for inpatient services. exceeded. Program costs were obtained from the Ontario Ministry
The midpoint for the top category was $225 for office visits and of Health and thus reflect the true costs to the Ministry. A RR of
inpatient services and $222 for ED visits. The midpoint of each bin death or a pre-UIIP number of deaths smaller than 45% of the
was used for calculation. For hospitalization cost, the RIW 95% base case value is far outside the 95% CI for these values. The
CI defined the lower and upper limit for deterministic sensitivity 95% CIs range from 92% to 108% of the base case value for pre-
ED, emergency department; ICER, incremental cost-effectiveness ratio; $M, $ million; QALY, quality-adjusted life years; TIIP, targeted influenza immunization program;
UIIP, universal influenza immunization program.
doi:10.1371/journal.pmed.1000256.t003
UIIP deaths (all ages) and from 85% to 118% of base case value Discussion
for RR of death (all ages). Similarly a post-UIIP number of deaths
greater than 190% of the base case value is outside the 95% CI Our analysis suggests that Ontario’s UIIP is economically
range for this variable (2260% to 120% of the base case value). attractive. UIIP reduces the number of influenza cases and deaths
An analysis of extremes was also performed by examining best- and reduces health services resource use. The additional costs of
and worst-case scenarios (Table S5). In the worst-case scenario (UIIP UIIP are partly (39%) offset by savings in health care costs.
does not provide much health benefit over TIIP, health service Compared to TIIP, UIIP is cost-effective at an ICUR of $10,797
resource unit costs are low), UIIP increases program costs by $20 per QALY gained.
million, saves only $1.8 million in health care service cost, and incurs The main limitation in determining the impact of UIIP is the
a QALY loss. UIIP becomes unattractive because it is not associated ecologic study design of the underlying effectiveness study. Causality
with health gain, and also increased costs compared to TIIP. cannot be established with this type of study design. The effect size
In the best-case scenario (UIIP is highly effective, health service may be exaggerated if there are province-specific epidemiological or
resource unit costs are high), UIIP increases program costs by health service factors that have in more recent years reduced
$19.1 million, saves $40.4 million in health care services, and gains influenza events in Ontario or increased it in other provinces.
5,619 QALYs (discount rate 3%). UIIP dominates TIIP in this Similarly, since the outcome measure is influenza-like illness rather
scenario; i.e., UIIP is less costly and more effective than TIIP. than confirmed influenza, it is possible that this could make
Probabilistic sensitivity analysis (Table 4, Figure 2) shows that estimates of outcomes sensitive to changes in the epidemiology of
the probability of UIIP being cost-effective at a willingness-to-pay non-influenza respiratory illness. However, this design is appro-
threshold of $50,000 per QALY is exceeding 90%. The mean priate for assessing the public health impact of a population-wide
(95% CI) QALYs gained are 1,263 [2456; 3,082], and the mean intervention [13]. The results of Kwong’s analysis of observational
(95% CI) net cost is $12.10 million [2$0.01 million; $16.11 ecologic data in four provinces over 7 y are congruent with the
million]. Finally, results for the additional analysis, using the results of randomized controlled trial data in targeted groups.
diagnostic codes for all respiratory conditions to extract health Targeted immunization has been shown to be effective in
care utilization data, are presented in Table S7. preventing laboratory-confirmed influenza infection in healthy
Figure 1. Tornado diagram comparing the relative importance of variables. The grey vertical line corresponds to all the uncertain
parameters being at their respective base values. The width of the bars indicates the uncertainty associated with each parameter (ranging from lower
to upper limit). The red segments of the bars correspond to result values increasing the base case ICER, and the blue segments of the bars correspond
to result values decreasing the base case ICER. ED, emergency department; QALY, quality-adjusted life year; RR, relative rate; TIIP, targeted influenza
immunization program; UIIP, universal influenza immunization program.
doi:10.1371/journal.pmed.1000256.g001
children 2 y and older [27], in healthy adults, especially when there been responsible for the decrease in cases and deaths in Ontario, in
is a good match and virus circulation is high [28], and older adults the absence of other obvious causal mechanisms. Stronger
(65 y and older), especially in long-term care facilities [28]. The randomised clinical trial evidence of the effectiveness of universal
weight of evidence seems to support the hypothesis that UIIP has vaccination in large populations is unlikely to be available in the
future. Finally, to test whether the disproportionate decrease in
influenza events from a 9% difference in coverage improvement
may be attributable to herd immunity, we analyzed a simple
Susceptible-Infected-Removed (SIR) compartmental model (for a
Table 4. Results—probabilistic sensitivity analysis. description of the model and results, see Figures S1 and S2, Tables
S7 and S8, and Text S3). This basic analysis suggests that the 40%–
60% decrease in influenza events is plausible.
Analysis Net Cost ($M) Net QALYsa
Other study limitations relate to the definition of cases and
Deterministic $12.24 1,134 quality of life estimates used. Cases were defined based on health
Probabilistic care resource use for pneumonia and influenza and therefore
Mean $12.10 1,263 represent influenza-like illness cases. Symptomatic cases not
Median $13.46 1,190
requiring health care contacts are not included, potentially
underestimating the benefit of the program. Utility weights to
2.5th percentile 2$0.01 2456
estimate QALYs were obtained from an analysis of clinical trial
97.5th percentile $16.11 3,082 data from influenza patients in Europe and North America [16].
a
Discounted at an annual rate of 3%.
While these data are not Canada-specific, the populations are
$M, $ million; QALY, quality-adjusted life years. similar. Furthermore, sensitivity analysis was performed and
doi:10.1371/journal.pmed.1000256.t004 reported for utility weights used in the analysis.
This study has several strengths. The model is robust and is Supporting Information
based on administrative health care resource use and cost data
covering all residents of Ontario. Data were extracted and Figure S1 Mean filtered relative reduction in Ontario for a
analyzed in detail to estimate influenza-associated events and range of vaccine coverage under UIIP.
costs as accurately as possible to assess the impact of this Found at: doi:10.1371/journal.pmed.1000256.s001 (0.20 MB
intervention on the population of interest rather than estimating TIF)
parameter values from other, smaller subpopulations that may not Figure S2 Proportion of simulations with a relative reduction
be representative. Finally, extensive sensitivity analyses demon- below 0.01%, 0.1%, 1%, and 5% in Ontario for a range of vaccine
strated the program to be effective and cost-effective under very coverage under UIIP.
conservative assumptions. Found at: doi:10.1371/journal.pmed.1000256.s002 (0.26 MB
The cost-effectiveness of influenza immunization programs TIF)
has been demonstrated by numerous economic evaluations of
TIIPs, many of which are directly based on clinical trial data. Table S1 Mean annual influenza-related event rates by age
TIIPs have been shown to be cost-effective in children 6 mo and group.
older [29–36], adults 50 y and older [37–41], working adults Found at: doi:10.1371/journal.pmed.1000256.s003 (0.14 MB
[42–44], working adult cancer patients [45], pregnant women DOC)
[46], health care workers [47,48], high-risk individuals [49], Table S2 Relative rates by age groups.
and older adults (65 y and older) [39,50–55] from a health care Found at: doi:10.1371/journal.pmed.1000256.s004 (0.10 MB
payer perspective. Most economic evaluations found TIIPs
DOC)
to be not only cost-effective but cost-saving from a societal
perspective. Table S3 Results (base case, disaggregated).
Policy makers in many jurisdictions considering the implemen- Found at: doi:10.1371/journal.pmed.1000256.s005 (0.12 MB
tation of universal immunization have expressed interest in an DOC)
economic evaluation of Ontario’s program. The program appears
Table S4 Deterministic sensitivity analysis.
to offer some health benefits, but the relationship between cost and
Found at: doi:10.1371/journal.pmed.1000256.s006 (0.29 MB
health benefits of universal immunization had not been evaluated.
DOC)
Our study provides evidence that a universal program is
economically attractive in jurisdictions with influenza epidemiol- Table S5 Aggregated worst-case and best-case results.
ogy and health care costs that are broadly similar to that of Found at: doi:10.1371/journal.pmed.1000256.s007 (0.08 MB
Ontario. DOC)
A UIIP may be an appealing intervention in high-income
Table S6 Results of the regression analysis.
jurisdictions with comparable demographic characteristics (age
Found at: doi:10.1371/journal.pmed.1000256.s008 (0.35 MB
distribution, risk profile, density) where influenza transmission can
be expected to be reasonably similar to the population analyzed. A DOC)
health care system similar to Ontario’s (i.e., health care systems Table S7 Results - sensitivity analysis using all respiratory
with one major payer), where the costs of the immunization conditions as outcome.
program and the costs of treating influenza cases are both in the Found at: doi:10.1371/journal.pmed.1000256.s009 (0.05 MB
payer’s budget, will enable the universal program costs to be partly DOC)
offset by savings in health care cost.
Table S8 Input parameter ranges.
Found at: doi:10.1371/journal.pmed.1000256.s010 (0.04 MB
Conclusion
DOC)
This analysis indicates that compared to a TIIP, Ontario’s UIIP
reduces influenza illness attack rates, morbidity, and mortality at Table S9 Results of empirical versus predicted relative
reasonable cost to the health care payer. reduction.
Found at: doi:10.1371/journal.pmed.1000256.s011 (0.03 MB We also thank the Toronto Health Economics Network and fellow
DOC) colleagues and students for comment on earlier drafts.
Previous presentation of the information reported in this
Text S1 Effectiveness of Ontario’s UIIP. manuscript:
Found at: doi:10.1371/journal.pmed.1000256.s012 (0.06 MB N Sander B, Kwong J, Bauch C, Maetzel A, McGeer A, Raboud J,
DOC) Krahn, M. The Cost-Effectiveness of Ontario’s Universal Influenza
Immunization Program. Society for Medical Decision Making, 30th
Text S2 Additional analysis using all respiratory conditions as Annual Meeting. 2008.
outcome. N Sander B, Kwong J, Bauch C, Maetzel A, McGeer A, Raboud J,
Found at: doi:10.1371/journal.pmed.1000256.s013 (0.05 MB Krahn, M. The Cost-Effectiveness of Ontario’s Universal Influenza
DOC) Immunization Program. The European Scientific Working Group on
Influenza, 3rd European Influenza Conference. 2008.
Text S3 Dynamic transmission model demonstrating herd
immunity effects.
Found at: doi:10.1371/journal.pmed.1000256.s014 (0.07 MB Author Contributions
DOC) ICMJE criteria for authorship read and met: BS JCK CTB AM AM JMR
MDK. Agree with the manuscript’s results and conclusions: BS JCK CTB
Acknowledgments AM AM JMR MDK. Designed the experiments/the study: BS CTB AM
AM JMR MDK. Analyzed the data: BS. Collected data/did experiments
We thank Deva Thiruchelvam and Brandon Zagorski for assistance with for the study: BS JCK. Wrote the first draft of the paper: BS. Contributed
data preparation and analyses of Ontario cost data and Nancy-Peroff to the writing of the paper: JCK CTB AM AM JMR MDK.
Johnston for providing data on Ontario’s influenza immunization program.
References
1. (2003) Influenza. Fact Sheet: World Health Organization. 22. (2008) Consumer price index (CPI), Ontario, Health Care, 1979–2007.
2. Jefferson TO, Rivetti D, Di Pietrantonj C, Rivetti A, Demicheli V (2007) Canadian socio-economic information management system (CANSIM) Statistics
Vaccines for preventing influenza in healthy adults. Cochrane Database Syst Canada et al.
Rev. CD001269 p. 23. (2008) JPPC Rate Model results based on 2006/07 data. Ontario Joint Policy
3. Govaert TM, Thijs CT, Masurel N, Sprenger MJ, Dinant GJ, et al. (1994) The and Planning Committee.
efficacy of influenza vaccination in elderly individuals. A randomized double- 24. Pink GH, Bolley HB (1994) Physicians in health care management: 4. Case mix
blind placebo-controlled trial. JAMA 272: 1661–1665. groups and resource intensity weights: physicians and hospital funding. CMAJ
4. Neuzil KM, Dupont WD, Wright PF, Edwards KM (2001) Efficacy of 150: 1255–1261.
inactivated and cold-adapted vaccines against influenza A infection, 1985 to 25. Pink GH, Bolley HB (1994) Physicians in health care management: 3. Case mix
1990: the pediatric experience. Pediatr Infect Dis J 20: 733–740. groups and resource intensity weights: an overview for physicians. CMAJ 150:
5. Heikkinen T, Ruuskanen O, Waris M, Ziegler T, Arola M, et al. (1991) 889–894.
Influenza vaccination in the prevention of acute otitis media in children. Am J Dis 26. OCCI (Ontario Case Costing Initiative) (2009) Costing analysis tool.
Child 145: 445–448. 27. Jefferson T, Rivetti A, Harnden A, Di Pietrantonj C, Demicheli V (2008)
6. Clements DA, Langdon L, Bland C, Walter E (1995) Influenza A vaccine Vaccines for preventing influenza in healthy children. Cochrane Database Syst
decreases the incidence of otitis media in 6- to 30-month-old children in day Rev. CD004879 p.
care. Arch Pediatr Adolesc Med 149: 1113–1117. 28. Rivetti D, Jefferson T, Thomas R, Rudin M, Rivetti A, et al. (2006) Vaccines for
7. Colombo C, Argiolas L, La Vecchia C, Negri E, Meloni G, et al. (2001) preventing influenza in the elderly. Cochrane Database Syst Rev 3: CD004876.
Influenza vaccine in healthy preschool children. Rev Epidemiol Sante Publique 29. Esposito S, Marchisio P, Bosis S, Lambertini L, Claut L, et al. (2006) Clinical
49: 157–162. and economic impact of influenza vaccination on healthy children aged 2–5
8. (2008) Statement on influenza vaccination for the 2008–2009 season. An years. Vaccine 24: 629–635.
Advisory Committee Statement (ACS). Can Commun Dis Rep 34: 1–46.
30. Hibbert CL, Piedra PA, McLaurin KK, Vesikari T, Mauskopf J, et al. (2007)
9. (1993) Canadian Consensus Conference on Influenza. Can Commun Dis Rep
Cost-effectiveness of live-attenuated influenza vaccine, trivalent in preventing
19: 136–142, 145–137.
influenza in young children attending day-care centres. Vaccine 25: 8010–8020.
10. Squires SG, Pelletier L (2000) Publicly-funded influenza and pneumococcal
31. Li S, Leader S (2007) Economic burden and absenteeism from influenza-like
immunization programs in Canada: a progress report. Can Commun Dis Rep
illness in healthy households with children (5–17 years) in the US. Respiratory
26: 141–148.
Medicine 101: 1244–1250.
11. Fiore AE, Shay DK, Broder K, Iskander JK, Uyeki TM, et al. (2008) Prevention
32. Marchetti M, Kuhnel UM, Colombo GL, Esposito S, Principi N (2007) Cost-
and control of influenza: recommendations of the Advisory Committee on
effectiveness of adjuvanted influenza vaccination of healthy children 6 to 60
Immunization Practices (ACIP), 2008. MMWR Recomm Rep 57: 1–60.
months of age. Human Vaccines 3: 14–22.
12. (2000) Ontario invests $38 million to ease emergency room pressures with
universal vaccination program. Ministry of Health and Long-Term Care. 33. Navas E, Salleras L, Dominguez A, Ibanez D, Prat A, et al. (2007) Cost-
13. Kwong JC, Stukel TA, Lim J, McGeer AJ, Upshur RE, et al. (2008) The effect of effectiveness analysis of inactivated virosomal subunit influenza vaccination in
universal influenza immunization on mortality and health care use. PLoS Med children aged 3–14 years from the provider and societal perspectives. Vaccine
5: e211. doi:10.1371/journal.pmed.0050211. 25: 3233–3239.
14. Moran K, Maaten S, Guttmann A, Northrup D, Kwong JC (2009) Influenza 34. Schmier J, Li S, King JC, Jr., Nichol K, Mahadevia PJ, et al. (2008) Benefits and
vaccination rates in Ontario children: implications for universal childhood costs of immunizing children against influenza at school: an economic analysis
vaccination policy. Vaccine. based on a large-cluster controlled clinical trial. Health Affairs 27: w96–
15. Juurlink DN, Stukel TA, Kwong J, Kopp A, McGeer A, et al. (2006) Guillain- w104.
Barre syndrome after influenza vaccination in adults: a population-based study. 35. Teufel IRJ, Basco WT, Jr., Simpson KN (2008) Cost effectiveness of an inpatient
Arch Intern Med 166: 2217–2221. influenza immunization assessment and delivery program for children with
16. Turner D, Wailoo A, Nicholson K, Cooper N, Sutton A, et al. (2003) Systematic asthma. J Hosp Med 3: 134–141.
review and economic decision modelling for the prevention and treatment of 36. Salo H, Kilpi T, Sintonen H, Linna M, Peltola V, et al. (2006) Cost-effectiveness
influenza A and B. Health Technol Assess 7: iii–iv, xi–xiii, 1–170. of influenza vaccination of healthy children. Vaccine 24: 4934–4941.
17. Reisinger K, Greene G, Aultman R, Sander B, Gyldmark M (2004) Effect of 37. Aballea S, Chancellor J, Martin M, Wutzler P, Carrat F, et al. (2007) The cost-
influenza treatment with oseltamivir on health outcome and costs in otherwise effectiveness of influenza vaccination for people aged 50 to 64 years: an
healthy children. Clin Drug Investig 24: 395–407. international model. Value in Health 10: 98–116.
18. Kwong JC, Stukel TA, McGeer AJ, Manuel DG (2007) Appropriate measures of 38. Aballea S, De Juanes JR, Barbieri M, Martin M, Chancellor J, et al. (2007) The
influenza immunization program effectiveness. Vaccine 25: 967–969. cost effectiveness of influenza vaccination for adults aged 50 to 64 years: a
19. (2008) Mortality rate per 1,000 population. Canadian socio-economic informa- model-based analysis for Spain. Vaccine 25: 6900–6910.
tion management system (CANSIM) Statistics Canada et al. . 39. Maciosek MV, Solberg LI, Coffield AB, Edwards NM, Goodman MJ (2006)
20. Mittmann N, Trakas K, Risebrough N, Liu BA (1999) Utility scores for chronic Influenza vaccination. Health impact and cost effectiveness among adults aged
conditions in a community-dwelling population. PharmacoEconomics 15: 50 to 64 and 65 and older. Am J Prev Med 31: 72–79.
369–376. 40. Newall AT, Scuffham PA, Kelly H, Harsley S, MacIntyre CR (2008) The cost-
21. (2006) HTA guidelines for the economic evaluation of health technologies. effectiveness of a universal influenza vaccination program for adults aged 50–64
Ottawa. years in Australia. Vaccine 26: 2142–2153.
41. Turner DA, Wailoo AJ, Cooper NJ, Sutton AJ, Abrams KR, et al. (2006) The 49. Ryan J, Zoellner Y, Gradl B, Palache B, Medema J (2006) Establishing the
cost-effectiveness of influenza vaccination of healthy adults 50–64 years of age. health and economic impact of influenza vaccination within the European
Vaccine 24: 1035–1043. Union 25 countries. Vaccine 24: 6812–6822.
42. Colombo GL, Ferro A, Vinci M, Zordan M, Serra G (2006) Cost-benefit 50. Cai L, Uchiyama H, Yanagisawa S, Kamae I (2006) Cost-effectiveness analysis
analysis of influenza vaccination in a public healthcare unit. Ther Clin Risk of influenza and pneumococcal vaccinations among elderly people in Japan.
Manag 2: 219–226. Kobe J Med Sci 52: 97–109.
43. Rothberg MB, Rose DN (2005) Vaccination versus treatment of influenza in 51. Gilman BH, Bonito AJ, Eicheldinger C (2007) Impact of influenza immunization
working adults: a cost-effectiveness analysis. Am J Med 118: 68–77. on medical expenditures among medicare elderly, 1999–2003. Am J Prev Med
44. Samad AH, Usul MHBHJ, Zakaria D, Ismail R, Tasset-Tisseau A, et al. (2006) 32: 107–115.
Workplace vaccination against influenza in Malaysia: does the employer benefit? 52. Hoshi SL, Kondo M, Honda Y, Okubo I (2007) Cost-effectiveness analysis of
J Occup Health 48: 1–10. influenza vaccination for people aged 65 and over in Japan. Vaccine 25:
45. Avritscher EB, Cooksley CD, Geraci JM, Bekele BN, Cantor SB, et al. (2007) 6511–6521.
Cost-effectiveness of influenza vaccination in working-age cancer patients. 53. Wang ST, Lee LT, Chen LS, Chen THH (2005) Economic evaluation of
Cancer 109: 2357–2364.
vaccination against influenza in the elderly: an experience from a population-
46. Roberts S, Hollier LM, Sheffield J, Laibl V, Wendel GD (2006) Cost-
based influenza vaccination program in Taiwan. Vaccine 23: 1973–1980.
effectiveness of universal influenza vaccination in a pregnant population. Obstet
54. Postma MJ, Baltussen RM, Heijnen ML, de Berg LT, Jager JC (2000)
Gynecol 107: 1323–1329.
Pharmacoeconomics of influenza vaccination in the elderly: reviewing the
47. Burls A, Jordan R, Barton P, Olowokure B, Wake B, et al. (2006) Vaccinating
healthcare workers against influenza to protect the vulnerable-Is it a good use of available evidence. Drugs Aging 17: 217–227.
healthcare resources? A systematic review of the evidence and an economic 55. Postma MJ, Jansema P, van Genugten ML, Heijnen ML, Jager JC, et al. (2002)
evaluation. Vaccine 24: 4212–4221. Pharmacoeconomics of influenza vaccination for healthy working adults:
48. Chan AL, Shie HJ, Lee YJ, Lin SJ, Chan ALF, et al. (2008) The evaluation of reviewing the available evidence. Drugs 62: 1013–1024.
free influenza vaccination in health care workers in a medical center in Taiwan. 56. Kunji K (2004) The Ontario experience with universal vaccination. National
Pharm World Sci 30: 39–43. Influenza Vaccine Summit. Atlanta.
Editors’ Summary
Background. Annual outbreaks (epidemics) of influenza—a calculated by multiplying the time spent in a certain health
viral disease of the nose, throat, and airways—make millions state by a measure of the quality of that health state. The
of people ill and kill about 500,000 individuals every year. In researchers report that the cost of Ontario’s UIIP was about
doing so, they impose a considerable economic burden on twice as much as the cost of a TIIP for the province. However,
society in terms of health care costs and lost productivity. the introduction of the UIIP reduced the number of influenza
Influenza epidemics occur because small but frequent cases by nearly two-thirds and reduced deaths from
changes in the viral proteins to which the immune system influenza by more than a quarter compared with what
responds mean that an immune response produced one year would have been expected had the province continued to
by exposure to an influenza virus provides only partial offer a TIIP, an overall saving of 1,134 QALYs. Furthermore,
protection against influenza the next year. Annual the reduction in influenza cases halved influenza-related
immunization with a vaccine that contains killed influenza health care costs, mainly because of reductions in
viruses of the major circulating strains can boost this natural hospitalization. Overall, this means that the additional cost
immunity and greatly reduce a person’s chances of catching to Ontario of saving one QALY through the introduction of
influenza. Consequently, many countries run seasonal the UIIP was Can$10,797, an ‘‘incremental cost-effectiveness
influenza vaccine programs. These programs usually target ratio’’ of $10,797 per QALY gained.
people at high risk of complications from influenza and
individuals likely to come into close contact with them, and What Do These Findings Mean? In Canada, an
people who provide essential community services. So, for intervention is considered cost-effective from the point of
example, in most Canadian provinces, targeted influenza view of a health care purchaser if it costs less than Canadian
immunization programs (TIIPs) offer free influenza $50,000 to gain one QALY. These findings indicate, therefore,
vaccinations to people aged 65 years or older, to people that for Ontario the introduction of the UIIP is economically
with chronic medical conditions, and to health care workers. attractive. Indeed, the researchers calculate that even if the
costs of the UIIP were to double, the additional cost of
Why Was This Study Done? Some experts argue, saving one QALY by introducing universal immunization
however, that universal vaccination might provide would remain below $50,000. Other ‘‘sensitivity’’ analyses
populations with better protection from influenza. In 2000, undertaken by the researchers also indicate that universal
the province of Ontario in Canada decided, therefore, to immunization is likely to be effective and cost-effective in
introduce a universal influenza immunization program (UIIP) Ontario if other key assumptions and/or data included in the
to provide free influenza vaccination to everyone older than calculations are varied within reasonable limits. Given these
6 months, the first large program of this kind in the world. A findings, the researchers suggest that a UIIP might be an
study published in 2008 showed that, following the appealing intervention in other Canadian provinces and in
introduction of the UIIP, vaccination rates in Ontario other high-income countries where influenza transmission
increased more than in other Canadian provinces. In and health-care costs are broadly similar to those in Ontario.
addition, deaths from influenza and influenza-related use
of health care facilities decreased more in Ontario than in Additional Information. Please access these Web sites via
provinces that continued to offer a TIIP. But is universal the online version of this summary at http://dx.doi.org/10.
influenza vaccination good value for money? In this study, 1371/journal.pmed.1000256.
the researchers evaluate the cost-effectiveness of the Ontario
UIIP by comparing the health outcomes and costs associated N A PLoS Medicine Research Article by Kwong and colleagues
with its introduction with the health outcomes and costs describes how the introduction of universal influenza
associated with a hypothetical continuation of targeted immunization in Ontario altered influenza-related health
influenza immunization. care use and deaths in the province