Cost-effectiveness analysis of quadrivalent influenza vaccine versus trivalent influenza vaccine for elderly in Hong Kong

Background Cost and quality-adjusted life-years (QALYs) gained by quadrivalent influenza vaccine (QIV) versus trivalent influenza vaccine (TIV) in Hong Kong elderly were estimated over 9 seasons. Methods TIV-unmatched influenza B infection rates with QIV versus TIV were estimated by an epidemiology model. Model parameters included percentages of influenza B lineages in circulation, influenza B-associated hospital admission, age-specific population, vaccine coverage and effectiveness. Incremental cost per QALY gained (ICER) by QIV versus TIV were estimated from Hong Kong’s societal perspective. Results Mean reduction in influenza B infection rate was 191.3 (95%CI 45.1-337.5) per 100,000 population aged ≥65 years. Highest cost savings and QALYs gained by QIV occurred in 2007 with high percentage of TIV-unmatched strain (92.9%) for age groups 65–79 years (USD266,473 and 22.8 QALYs) and ≥80 years (USD483,461 and 27.3 QALYs). ICERs of QIV were below willingness-to-pay for age group 65–79 years in 6, 5 and 3 years when QIV cost + USD1 + USD2 and + USD5 more than TIV, respectively. For age group ≥80 years, ICERs of QIV were below willingness-to-pay in 7 and 5 years when QIV cost + USD1 and + USD5, correspondingly. Conclusions Acceptance of QIV to be cost-effective in Hong Kong elderly was subject to QIV unit cost and percentage of circulating TIV-unmatched influenza B lineages.


Background
We recently reported the percentages of influenza B lineages (B/Victoria and B/Yamagata) in clinical isolates in 2000-2010 and found that co-circulation of both influenza B lineages was very common in Hong Kong [1]. Furthermore, one of the influenza B lineages predominated (accounting for over 80% of the circulating influenza B strains) in six of the ten years. These predominated years were associated with increased risk of hospitalization by 1.4-fold. The seasonal trivalent influenza vaccine (TIV) covers two subtypes of influenza A (A/H1N1 and A/H3N2) plus one of the two influenza B lineages circulating in humans, and TIV only matched the predominated influenza B strain in two of these six predominated years.
Findings in literature suggested that including two lineages of influenza B in a quadrivalent influenza vaccine (QIV) would have a positive impact in terms of cost saving and reducing hospitalization related to influenza infection [2,3]. To further quantify the clinical and economic impact of QIV, we estimated the difference in costs and qualityadjusted life-years (QALYs) gained by QIV, when compared to TIV, in elderly population over the period of 2001-2010 from a societal perspective of Hong Kong.

Methods
In the present analysis, we estimated the potential cost and QALYs difference of influenza B, when compared to TIV, from 2001-2010 years in elderly population (aged ≥65 years) in Hong Kong. Year 2009 was excluded as it was dominated by the novel pandemic influenza A 2009 H1N1. Events caused by infection of the TIVunmatched influenza B lineage in each year were calculated by an epidemiology model. The infections caused by the influenza B lineage or influenza A subtypes covered by both TIV and QIV were excluded, assuming that there were no change in the infection rates. The costs and QALYs loss for symptomatic patients who received medical care in outpatient and hospital setting were estimated from the event rates. The parameters used for estimation of clinical events, cost difference and QALYs gained by QIV versus TIV were listed in Table 1. All model parameters were obtained from published literature.

Clinical events analysis
For each year, the hospitalization rate of target B lineage was calculated using the influenza B hospital admission rate and the percentage of target influenza B lineage identified in the isolates reported in our previous cohort study [1]. Briefly, in this prior study we identified hospitalized patients with laboratory confirmed influenza virus A or B infection The intensive care unit (ICU) admission rate was calculated by hospitalization rate of target lineage together with ICU admission rate among hospitalized patients [4]. The mortality rate of influenza B among hospitalized patients was retrieved from patients aged ≥65 years in the same cohort of our previous study [1]. The expected annual infection rate of target lineage was estimated using the mortality rate and a case-fatality ratio of influenza [5]. The rate of outpatient care was further calculated as the difference of estimated annual infection rate and hospitalization rate. To estimate the reduction in event rates of outpatient care, hospitalization, ICU admission and death with the use of QIV versus TIV, the epidemiology model described by Reed  The VE TIV against the TIV-unmatched influenza B lineage was estimated to be reduced by 30% [6]. The VE QIV against the TIV-unmatched influenza B lineage was assumed to be the same as VE of TIV against the TIV-selected influenza B lineage, and the year-specific VE was retrieved from literature reported in the period of 2001-2010 [7][8][9][10][11][12][13][14]. The year 2012 age-specific influenza vaccine coverage rate (39.1%) in elderly published by the Department of Health of Hong Kong was applied in the present analysis for 9 seasons [15].
The reduction in case numbers of each event (outpatient care, hospitalization, ICU admission and death) were then calculated by multiplying the change in event rate and the population of ≥65 years for each year.

QALY analysis
The QALYs gained by QIV were calculated using the number of reduced infected cases, and the utility value and duration of time-spent in each of the four statuses: (1) Outpatient care, (2) hospitalization without ICU admission, (3) ICU care, and (4) death. The utilities were retrieved from health-related quality of life reported from literature (Table 1) [16][17][18][19][20][21][22]. The time-spent in outpatient care and hospitalization were the length of outpatient duration of illness [23] and length of hospital stay [4], correspondingly. The QALYs loss of death was estimated using life expectancy in Hong Kong and age-specific utility value for ≥65 years [24,25]. The QALYs loss in both vaccine arms was discounted to 2014 with 3% annual discounted rate.

Cost analysis
Both direct medical cost and indirect cost of reduced cases of the TIV-unmatched influenza B lineage were considered as cost savings of QIV, from the societal perspective of Hong Kong. Direct medical cost included costs of outpatient care and hospitalization. The cost per outpatient clinic visit and cost per hospital day (with and without ICU care) were estimated from the 2014 charges of the Hospital Authority (HA). HA is the largest, nonprofit-making public health organization in Hong Kong. Assuming the charges listed by the Hospital Authority represent only the cost components (including labor costs) with no addition of profits, the cost per general outpatient visit and daily cost of hospitalization were therefore approximated using the Hospital Authority charges [26].
Productivity loss of caregivers of elderly patients was estimated by the median daily income (non-gender specific) of the population in Hong Kong [24], and duration of hospitalization for inpatient care. For patients receiving only outpatient care, the duration of productivity loss was limited to days attending outpatient clinic. The number of clinic visit was assumed to be one time for a conservative estimation on both direct and indirect costs of outpatient care. The proportion of employed caregivers was estimated by the percentage of Hong Kong population in the labor force and the employment rate among the labor force [24].

Cost-effectiveness analysis
The incremental cost per QALY gained (ICER) by QIV versus TIV was calculated using the following equation: (Additional vaccine cost of QIV versus TIV -Costsavings of QIV)/QALYs gained by reduced events, where additional vaccine cost of QIV was calculated by: Additional unit cost of QIV versus TIV × age-specific coverage rate × population of age group The current unit cost of TIV in Hong Kong is USD7.7. Four levels of additional cost (the unit cost of QIV to be USD1, 2, 5 and 10 more than TIV) were considered in the cost-effectiveness analysis. As recommended by the World Health Organization (WHO), three-fold of gross domestic product (GDP) per capita was used as the threshold of willingness-to-pay per QALY [27]. A scenario with ICER of QIV less than 3-fold of GDP per capita of Hong Kong was considered as cost-effective.

Results
The reduction in expected events rates of TIV-unmatched influenza B by QIV versus TIV are shown in The GDP per capita of Hong Kong was USD36,557. 32 Using 3-time GDP per capita (USD109,671) as the threshold of willingness-to-pay, ICERs of QIV indicated that it was more cost-effective than TIV for age group 65-79 years in 6 years when QIV cost USD1 more than TIV (Table 4). When QIV was more costly than TIV by USD2, USD5 and USD10, it was more cost-effective than TIV in 5, 3 and 2 years, respectively. In the age group ≥80 years, the ICERs of QIV was below the threshold of willingness-to-pay in 7 years when QIV versus TIV cost level was + USD1. QIV remained to be cost-effective in 5 years when it cost up to USD5 more than TIV. It was cost-effective in 3 years when it cost USD10 more than TIV.    [2]. As demonstrated by the annual cost savings and QALYs loss averted by QIV (Table 2), and QIV would be more cost-effective than TIV if the unit cost of QIV is the same as TIV (additional unit cost = USD0) in all 9 years (except year 2002). The number of years in which QIV was more cost-effective than TIV decreased in both age groups 65-79 years and ≥80 year when the unit cost of QIV increased by USD1-10 (Table 3) In the present study, QIV was cost-effective in year 2007 at four cost levels (+USD1, +USD2, +USD5 and + USD10). The cost savings and QALYs gained by QIV were the highest in this year. It could be explained by the fact that a high percentage of circulating influenza B lineages in Hong Kong were not covered by TIV (92.9%) and the hospitalization rate for influenza B in the elderly was the highest in 2007 [1]. The cost-effectiveness of the influenza vaccine tends to be enhanced when the expected hospitalization rate without the vaccine is high [28]. The expected QALYs gained and cost savings of QIV were sufficient to overcome the economic impact of additional cost of QIV and it therefore remained costeffective over a broad range of additional vaccine cost.

Discussion
Our study was limited by using a single source (one teaching hospital) of influenza B hospital admission rates and circulating virus lineages for the calculation of population-based influenza B infection rate in Hong Kong elderly. The infection rate was estimated by the case-fatality ratio, yet this information was not available from Hong Kong. A case-fatality ratio derived from US was used, as the estimated influenza-related mortality rates were generally similar in Hong Kong and the US [29,30]. We assumed that patients in the outpatient setting would seek care at medical clinic for one time, as Chinese older people in Hong Kong tend to perceive influenza as a serious illness [31]. Some patients would use home remedies or over-the-counter preparation to alleviate influenza symptoms, and the cost-saving of QIV might be overestimated. On the other hand, the QALYs loss of caregivers was not included in the analysis and the QALY saved by QIV might therefore be underestimated. The current cost and QALYs were calculated based upon the vaccine coverage of 39.1% and herd immunity effect was not included in the analysis. The study findings are likely to change if the vaccine coverage is improved in elderly (and therefore generates herd immunity), probably showing more prominent saving in cost and QALYs with QIV comparing to TIV. Sensitivity analysis was not conducted in the present study. Expected cost and clinical outcomes were calculated for 9 seasons using year-specific key parameters. The change of cost and clinical outcomes was found to be sensitive to the variation of key year-specific parameters (namely the percentage of circulating TIVunmatched influenza B lineage and hospitalization rate).
Using year-specific inputs has therefore served the purpose of sensitivity analysis.

Conclusions
In conclusion, QIV could reduce direct and indirect costs of influenza B infection and gain higher QALYs when compared to TIV in the elderly population of Hong Kong. The acceptance of QIV to be cost-effective in Hong Kong elderly was highly subject to the unit cost of QIV versus TIV and the percentage of circulating TIV-unmatched influenza B lineages.

Competing interests
Chan PSK has participated in clinical trials supported by F. Hoffmann-La Roche and GlaxoSmithKline, received consultancy fees from F. Hoffmann-La Roche and GlaxoSmithKline, and served the Quadrivalent Influenza Vaccine Asia-Pacific Advisory Board of GlaxoSmithKline. You JHS and Ming WK have no conflict of interest to disclose.
Authors' contributions JHSY and PKSC: Study concept and design, and interpretation of data, preparation of manuscript. JHSY and WKM: Acquisition of data and data analysis. All authors read and approved the final manuscript.

Funding
This study did not receive any funding.