Cost-effectiveness of a pentavalent human–bovine reassortant rotavirus vaccine for children ≤5 years of age in Taiwan

Abstract

Objective:

A Markov model was used to assess the impact of RV5, a pentavalent (G1, G2, G3, G4, P1A[8]) human bovine (WC3 strain) reassortant rotavirus vaccine, on reducing the healthcare burden and cost associated with rotavirus gastroenteritis (RGE) in Taiwan. Other cost-effectiveness analyses for rotavirus vaccination in industrialized countries have produced varying results depending on the input parameters assumed.

Methods:

Vaccination with RV5 is compared to no vaccination in a hypothetical cohort of Taiwanese children during their first 5 years of life to determine the per dose prices at which vaccination would be cost neutral or provide good value based on established standards from the healthcare (direct medical care costs only) and societal (all RGE-related costs) perspectives. The effects of vaccination on RGE healthcare utilization and days of parental work loss missed are based on results from the Rotavirus Efficacy and Safety Trial.

Results:

Without vaccination there would be 122,526 symptomatic episodes of RGE. Universal vaccination would reduce RGE-related deaths, hospitalizations, emergency department, and outpatient visits by 91.7%, 92.1%, 83.7%, and 73.4%, respectively. The price per dose at which vaccination would be cost-neutral is US$ 21.80 (688 NTD) and US$ 26.20 (827 NTD) from the healthcare and societal perspectives, respectively. At $25 per dose, the cost per QALY gained is US$ 2261 (71,335 NTD) from the healthcare perspective and cost saving from the societal perspective.

Key limitations:

The model only assesses the effect of RV5 on vaccinated children and does not account for herd immunity. However, given that high levels of coverage are anticipated in Taiwan, the effects of herd immunity are likely to be short-term.

Conclusion:

A pentavalent rotavirus vaccination program is likely to substantially reduce the healthcare burden associated with rotavirus gastroenteritis at a cost per QALY ratio within the range defined as cost-effective.

Key words::

• Cost-effectiveness
• Rotavirus gastroenteritis
• Rotavirus vaccination
• RV5
• Taiwan

Introduction

Rotavirus gastroenteritis (RGE) is the leading cause of severe diarrheal disease worldwide in children under 5 years of age1. Children with RGE usually experience a combination of diarrhea, vomiting, and/or fever, often leading to dehydration in severe cases2,3. In industrialized countries deaths due to RGE are relatively rare but healthcare costs are substantial. In Taiwan, estimates of RGE-related annual healthcare costs range from US$10–18.5 million4,5.

Currently there are two rotavirus vaccines licensed in Taiwan. One is a pentavalent (G1, G2, G3, G4, P1A[8]) human bovine (WC3 strain) reassortant rotavirus vaccine (RV5; RotaTeq®, Merck & Co., Inc., Whitehouse Station, NJ, USA). The other is a monovalent G1P1A[8] human rotavirus vaccine (RV1; Rotarix®, GlaxoSmithKline Biologicals, Rixensart, Belgium)6. Until now, vaccine use has been limited to the private sector and studies are currently being conducted in preparation for public sector reimbursement7,8. Government reimbursement would significantly help to increase the coverage rates in Taiwan because nearly all children are covered by the Taiwanese government’s national health insurance program4,9,10.

The purpose of this study is to assess the potential cost-effectiveness of universal vaccination with RV5 in Taiwan from the healthcare and societal perspectives. The healthcare perspective is limited to direct medical care costs and the cost of the immunization program. The societal perspective encompasses all RGE-related costs including travel to healthcare facilities, oral rehydration therapies given at home, extra diapers, and parental work days missed to care for sick children.

Other cost-effectiveness analyses have been published for high and middle-income countries throughout North America, South America, Europe, Australia, New Zealand, and Asia11–20. Many of these cost-effectiveness analyses have demonstrated that rotavirus vaccination is cost-effective and should be added to the immunization schedule12,13,15,16,18,19. These results are supported by post-marketing surveillance studies that have reported dramatic reductions in RGE-related healthcare utilization since rotavirus vaccination was added to the immunization schedule21–25. However, there are studies that have concluded rotavirus vaccination is not cost-effective17,20. The differences in study results, often within the same country, are largely due to differences in the input parameters assumed26. Although another cost-effectiveness analysis for rotavirus vaccination was published in Taiwan, the estimates of hospitalizations and ambulatory visits included in the base case in that study were lower than other published studies in Taiwan3–5,10. In addition, the study did not include the cost per quality adjusted life year (QALY) as an outcome measure.

The healthcare utilization and cost estimates for the base case scenario in this study are derived from a recently published prospective, cross-sectional surveillance study for rotavirus gastroenteritis conducted in hospitals and outpatient pediatric clinics in Taiwan8. These estimates are then varied based on other published studies from Taiwan3–5,10. The results are presented for the prices at which universal vaccination would be cost-neutral as well as the cost per quality adjusted life years gained for prices ranging from US$20–30 per dose.

Methods

Model design

The cost-effectiveness analysis is based on a Markov model using EXCEL software to assess the cost-effectiveness of universal vaccination with RV5 against RGE for a hypothetical birth cohort of 220,260 Taiwanese children during their first 5 years of life (Figure 1). Epidemiology studies have shown that although children may develop multiple episodes of RGE, the first two infections are generally the most severe, and subsequent infections are generally very mild or asymptomatic because heterotypic protection develops with successive episodes of RGE27. This is why we allow children to contract up to two symptomatic episodes in the model.

Figure 1.  Flow chart of Markov model.

Children may move between different health states over a discrete time period known as a Markov cycle28. The different health states reflect whether the child has ever experienced a rotavirus infection, the number of prior rotavirus infections, the serotypes causing the episodes, and whether the child survived or died of the illness. Although deaths due to rotavirus gastroenteritis are relatively rare in developed countries, cost-effectiveness analyses should capture all possible outcomes. The number of deaths assumed should reflect that rotavirus deaths are uncommon. During the first 6 months of life, monthly cycles are used to allow flexibility in the vaccination schedule. Quarterly cycles are employed after 6 months to track a child’s health state. The transition probabilities, or the likelihood that the child will move from one health state to another, reflect the age-specific probability of developing first and second episodes of rotavirus and the percentage of the episodes that are symptomatic. These data are based on a prospective, population-based study in Mexico that closely monitored children for the development of symptomatic and asymptomatic episodes27,29. This study was unique because study personnel collected stool specimens from all children born in the catchment area during the study period regardless of symptoms every 2 weeks for the first 2 years of life. Almost all other rotavirus studies were conducted in specific settings and usually once the children became symptomatic. The age distribution of episodes involving deaths and healthcare encounters by type are based on data from Taiwan in which the cumulative frequency of rotavirus is 78% and 73% for pediatric and hospital settings, respectively, by 36 months of age8.

Analytic perspectives and outcome measures

This cost-effectiveness analysis compares infants receiving RV5 as a 3-dose series to no vaccination in Taiwan. All of the scenarios are evaluated with per dose prices ranging from US$20–30. The QALY data is based on a Canadian study that includes the impact on the child as well as the caregiver where the utility weights for children with rotavirus gastroenteritis are based on the HUI2 and the Visual Analog Scale. The utility weights for the parents are based on the EQ-5D and the Visual Analog Scale30. Given that there are no established benchmarks in Taiwan on which to evaluate the cost-effectiveness of an intervention, we use the World Health Organization’s benchmark based on recommendations from the WHO Commission on Macroeconomics and Health31,32. This is consistent with the criteria used in other cost-effectiveness analyses conducted in Taiwan5,33. Interventions are classified as highly cost-effective if the cost per QALY gained is lower than the per capita GDP, may be cost-effective if the cost is less than 2–3-times the per capita GDP, and are not cost-effective if the cost is higher than 3-times the per capita GDP. In Taiwan, the per capita GDP was US$18,303 for 201034.

Model inputs

General overview

Table 1 lists the key model inputs along with the values for the base case scenario. All medical and non-medical care costs are presented in US dollars (US$) and adjusted for inflation to 2010 dollars using the Consumer Price Index for Taiwan35. Costs are given in US dollars as well as new Taiwanese dollars (NTD) shown in parentheses based on an exchange rate of US$1 = NTD 31.55 in 201036. RV5 is administered as a 3-dose vaccine at 2, 4, and 6 months of age, which is consistent with the schedule for DTwP in Taiwan37. The base case scenario assumed that 96% of children would receive three doses of RV5, 2% would receive two doses, and 2% would receive one dose based on the immunization coverage level for DTP3 reported in the National Immunization Information System in Taiwan, and this assumption is varied in sensitivity analyses37.

Table 1.  Summary of key model inputs for base case scenario.

Parameter	Value	References
Size of birth cohort	220,260^a	Statistical Yearbook of the Republic of China, 200946
Healthcare and family burden of RGE
Deaths, n	7	Wu et al.5
Hospitalizations, n	15,400	Chen et al.10
ED visits, n^b	20,322	Mast et al.8; Wu et al.5^b
Outpatient visits, n^b	215,978	Mast et al.8; Wu et al.5^b
Length of hospital stay, days	5.7	Mast et al.8
Work days missed because of an episode involving hospitalization and/or death	4.0	Mast et al.8
Work days missed because of an episode involving ED visits or office visits	1.3	Mast et al.8
QALY loss per RGE episode^c	0.01052	Brisson M et al. 201030
Costs of care^d
Cost per day of hospitalization^e	$103.88	Mast et al.8
Non-medical care costs for episodes involving  death or hospitalization^f	$68.60	Mast et al.8
Cost of ED visit	$59.36	Wu et al.5
Cost of office visit	$33.02	Mast et al.8
Non-medical care costs for episodes not involving  death or hospitalization^f	$13.91	Mast et al.8
Cost of a work day missed	$23.00	Basic Wage Rate, 200743
Vaccination characteristics
Coverage rate	96% 3 doses 2% 2 doses 2% 1 dose	Centers for Disease Control6
% Reduction in rate of healthcare encounters
	88% in the 3rd month of life with full protection thereafter	Dennehy et al.39
ED visits	47% in the 3rd and 4th months of life with full protection thereafter	Dennehy et al.39
Outpatient visits	30% in the 3rd and 4th months of life, 51% in the 5th and 6th months of life with full protection thereafter	Assumption
Work days lost	24% in the 3rd and 4th months of life, 47% in the 5th and 6th months of life with full protection thereafter	Assumption

^aThe size of the hypothetical birth cohort reflects an average of the birth cohorts between 2002–2006.

^bEstimate of 236,300 ambulatory care visits given in Mast et al.8 of which 8.6% were ED visits and 91.4% were office visits based on Wu et al.5

^cThis value was derived from the MIRAGE study and assumes a QALY loss of 7.1/1000 children based on the Visual Analog Scale as well as a QALY loss of 1.8/1000 parents for 1.9 parents per episode based on the EQ5D.

^dAll costs were adjusted to 2010 dollars based on the Consumer Price Index.

^eCost per day in the hospital reflects the total cost of the hospitalization divided by average hospital length of stay given in Mast et al.8

^fBased on non-medical care costs given in Mast et al.8 with value of work loss deducted.

The effect of vaccination on reducing the rate of hospitalizations, emergency department (ED) visits, office visits, and days of parental work loss missed due to RGE for those receiving all three doses are based on results from the Rotavirus Efficacy and Safety Trial (REST), a large-scale, placebo-controlled, Phase III trial that enrolled nearly 70,000 children38. The REST trial provided efficacy data for the first 2 years of life and different assumptions regarding efficacy were extended for the 3rd–5th years of life for different scenarios. Since there are no efficacy data against death, the effect of vaccination on hospitalizations is also applied to RGE deaths. The efficacy prior to completion of the 3rd dose shown in Table 1 is consistent with post-hoc analyses of the clinical trial data from REST39. Given that there is limited data on the efficacy of RotaTeq® for children not completing the three dose regimen, we assume the efficacy is 25% and 50% of a 3-dose regimen for one and two doses, respectively, to be conservative. These assumptions are then varied in sensitivity analyses based on available effectiveness data of the vaccine after administration of only one and only two doses40,41. The efficacy estimates are applied to all episodes of RGE because efficacy has been established against all of the major serotypes circulating in Taiwan8.

The cost of vaccination is based on the price per dose of RV5 with no additional administration fees charged by physicians in the public sector. Although there is a required registration fee that parents must pay for a physical examination prior to the administration of vaccines, RV5 is given on the same schedule as DTwP and the registration fee is already being paid for that vaccine. Therefore, it is not considered a marginal cost associated with RV5. There are also no costs associated with the adverse effects of vaccination since the clinical trial results found RV5 to be generally safe and well tolerated38,42.

Base case scenario

For the base case scenario, we rely on the surveillance study conducted by Mast et al.8 for the average hospital length of stay, the average number of days missed from work and the percentage of parents missing time from work to care for sick children as well as the average medical costs for hospitalizations and outpatient visits (Table 1). We assume that, among children with RGE involving hospitalization or death, 56% of working parents miss an average of 4 days of work. For episodes involving ED visits and office visits, 21% of working parents miss an average of 1.3 days of work8. The cost estimate for an ED visit is based on the estimate used in the Wu et al.5 cost-effectiveness analysis.

The number of ambulatory visits is based on an estimate provided in the Mast et al.8 study. The number of hospitalizations is based on the results from the Asian Rotavirus Surveillance Network10. The number of deaths is consistent with Wu et al.5 In the Wu et al. cost-effectiveness analysis, the number of emergency department visits is 8.6% of all ambulatory care visits. This percentage is applied to the total number of ambulatory visits estimated in the Mast et al. study to distinguish between outpatient visits and emergency department visits for RGE. We also apply the Basic Wage Rate in Taiwan for 2007 to be consistent with the Wu et al. cost-effectiveness analysis and the Basic Wage rate has not been adjusted for inflation since 200743.

Finally, we use the utility weight based on the Visual Analog Scale to estimate the QALY loss for the child and the utility weight based on the EQ-5D for the parents in the base case scenario and then vary these values in the sensitivity analysis.

Sensitivity analyses

These sensitivity analyses are designed to evaluate the impact of using different assumptions for healthcare utilization, vaccine coverage and efficacy as well as the QALY loss associated with episodes of rotavirus gastroenteritis. The estimates for the number of deaths, hospitalizations, ambulatory care visits, and work days missed are based on various published RGE surveillance studies in Taiwan. The number of annual deaths due to rotavirus varies from 1–2211. For inpatient care two different types of sensitivity analyses are performed. In one set of scenarios, all of the assumptions in the base case scenario remain the same except the number of hospitalizations. In another set of scenarios a three-way sensitivity analysis varies the number of hospitalizations, the average hospital length of stay, and the cost per day in the hospital to reflect the assumptions in two published studies, with the lowest and highest number of hospitalizations for RGE reported in Taiwan4,5.

Similarly, in the sensitivity analyses for outpatient care all of the assumptions in the base case scenario remain the same except the number of ambulatory care visits including emergency department (ED) and outpatient visits. In another set of scenarios the sensitivity analysis varies the number of ambulatory care visits and the cost of an outpatient visit based on the two published studies with the lowest and highest number of outpatient visits. The cost estimate for the ED visit does not change because only Wu et al.5 provided a cost estimate for ED visits. Lu et al.4 estimated that the number of RGE hospitalizations varies from 21,800–28,500, and the number of ambulatory care visits ranges from 255,600–596,400. The 25,150 hospitalizations and the 426,000 ambulatory visits (36,636 ED visits and 389,364 outpatient visits) in the sensitivity analysis reflect the average of these ranges. We again assume that 8.6% of the ambulatory visits are ED visits and 91.4% are physician outpatient visits5.

Another set of sensitivity analyses is designed to evaluate the effect of the number of doses administered, the efficacy of receiving only one or two doses, and the duration of protection. In one set of sensitivity analyses the coverage is not changed but the efficacy of only one and only two dose regimens are 81% and 69%, respectively, for hospitalizations and emergency department visits40. In another scenario the coverage with three doses is reduced to 90%, with another 5% receiving two doses and 5% receiving one dose where the efficacy of an incomplete regimen is 0.

In the base case scenario we extend the efficacy of the vaccine in the 2nd year after vaccination up to 5 years of age. In the sensitivity analysis we assume a 10% drop in the relative rate reduction each year for healthcare encounters in the 3rd, 4th, and 5th years after vaccination.

We also vary the value of a day missed from work and the utility loss associated with episodes of RGE. In one scenario we increase the value of a day missed from work to US$58.16 (1897 NTD). This reflects the average monthly earnings and monthly working hours of industrial and service-sector employees as reported by the government of Taiwan8.

Finally, we use the utility weight based on the HUI2 for the child with no QALY loss assumed for the parents as a conservative assumption and then use the VAS for two parents as well as the child as an upper bound.

Results

Base case scenario

The model predicts that in the absence of vaccination there would be 122,526 symptomatic episodes of RGE in a single birth cohort over 5 years with rotavirus related costs of US$20.5 million (646.4 million NTD) including US$16.6 million (525.5 million NTD) in direct medical care costs of which 42% are for hospitalizations (Table 2). When the price per dose is US$25 (789 NTD), the cost of vaccinating the birth cohort with a three dose regimen of RV5 would be US$16.2 million (510.7 million NTD).

Table 2.  Effect of vaccination on the healthcare burden and medical care costs of RGE at a cost of US$25 per dose for the base case scenario.

Events	No intervention	RV5	Net difference	% Change
Symptomatic cases	122,526	48,958	−73,568	−60.0
Deaths	7	0.58	−6.42	−91.7
Hospitalizations	15,400	1212	−14,188	−92.1
ED visits	20,322	3310	−17,012	−83.7
Outpatient visits	215,978	57,541	−158,437	−73.4
Medical care costs
Hospitalizations	8,684,363	567,913	−8,116,451	−93.5
ED visits	1,153,376	187,541	−965,835	−83.7
Outpatient visits	6,818,426	1,808,047	−5,010,379	−73.5
Direct non-medical costs (e.g., transportation, extra diapers, oral rehydration)	2,432,885	711,378	−1,721,507	−70.8
Indirect (work days lost)	1,399,785	259,834	−1,139,951	−81.4
Societal costs	20,488,836	3,534,713	−16,954,122	−82.7
Cost of vaccination	0	16,189,110
Net direct medical care costs^a	16,656,165	18,752,611	2,096,446	12.6
Net societal costs^b	20,488,836	19,723,823	−765,013	−3.7

^aNet direct medical care costs include hospitalization and ambulatory care (ED and outpatient visits) costs as well as the cost of vaccination with and without a vaccination program instituted.

^bNet societal costs include medical care costs, direct non-medical care costs, and indirect costs, as well as the cost of vaccination with and without a vaccination program instituted.

Universal vaccination would result in reductions of 91.7%, 92.1%, 83.7%, and 73.4% for deaths, hospitalizations, emergency department, and outpatient visits, respectively. The price per dose at which vaccination would be cost-neutral is US$21.80 (688 NTD) and US$26.20 (827 NTD) from the healthcare and societal perspectives, respectively. At $25 per dose, the cost per QALY gained would be US$2261 (71,335 NTD) from the healthcare perspective and cost saving from the societal perspective.

Sensitivity analyses

When the price per dose ranges from US$20–30, universal vaccination with RV5 would either be cost saving or highly cost-effective under all the scenarios examined (Table 3). From the healthcare perspective, the results are most sensitive to the number of hospitalizations and ambulatory visits (Figures 2 and 3). When 12,900 hospitalizations are assumed, universal vaccination is cost neutral at US$19.75 (623 NTD). When 25,150 hospitalizations are assumed, universal vaccination is cost neutral at US$29.70 (937 NTD.). Similarly, the prices at which universal vaccination would be cost neutral ranges from US$18.30 (577 NTD) when 148,325 ambulatory visits are assumed to US$29.15 (920 NTD) when 426,000 ambulatory visits are assumed.

Figure 2.  Sensitivity analysis for costs of vaccination at which vaccination is cost neutral.

Figure 3.  Sensitivity analysis for the cost per QALY gained at a cost of US$25 per dose.

Table 3.  Sensitivity analyses for incremental cost-effectiveness ratios with costs of vaccination ranging from US$20–30.

Sensitivity analysis analytic perspective	Cost per QALY gained^a
	US$ 20	US$ 25	US$ 30
1-way sensitivity analyses
A: 1 Death
Healthcare	Cost saving	Highly cost-effective (US$2726)	Highly cost-effective (US$6936)
Societal	Cost saving	Cost saving	Highly cost-effective (US$3216)
B: 22 Deaths
Healthcare	Cost saving	Highly cost-effective (US$1586)	Highly cost-effective (US$4035)
Societal	Cost saving	Cost saving	Highly cost-effective (US$1869)
C: 12,900 Hospitalizations
Healthcare	Highly cost-effective (US$189)	Highly cost-effective (US$3681)	Highly cost-effective (US$7172)
Societal	Cost saving	Highly cost-effective (US$832)	Highly cost-effective (US$4324)
D: 25,150 Hospitalizations
Healthcare	Cost saving	Cost saving	Highly cost-effective (US$211)
Societal	Cost saving	Cost saving	Cost saving
E: 12,755 ED visits; 135,570 OP visits^b
Healthcare	Highly cost-effective (US$1169)	Highly cost-effective (US$4661)	Highly cost-effective (US$8154)
Societal	Cost saving	Highly cost-effective (US$1528)	Highly cost-effective (US$5020)
F: 36,636/389,364 ED/OP visits^b
Healthcare	Cost saving	Cost saving	Highly cost-effective (US$579)
Societal	Cost saving	Cost saving	Cost saving
G: US$61.71/work day missed
Healthcare	Cost saving	Highly cost-effective (US$2199)	Highly cost-effective (US$5660)
Societal	Cost saving	Cost saving	Highly cost-effective (US$526)
H: 10% Reduction in efficacy, years 3–5 post-vaccination
Healthcare	Cost saving	Highly cost-effective (US$3441)	Highly cost-effective (US$7317)
Societal	Cost saving	Highly cost-effective (US$285)	Highly cost-effective (US$4161)
I: Utility weight based on the HUI2 for the child without any QALY loss for the parents
Healthcare	Cost saving	Highly cost-effective (US$6171)	Highly cost-effective (US$15,702)
Societal	Cost saving	Cost saving	Highly cost-effective (US$7279)
J: Utility weight based on the VAS for the child and two parents
Healthcare	Cost saving	Highly cost-effective (US$1866)	Highly cost-effective (US$4748)
Societal	Cost saving	Cost saving	Highly cost-effective (US$2201)
2- and 3-way sensitivity analyses
K: 90% 3 doses, 5% 2 doses, 5% 1 dose; 0 efficacy for incomplete regimens
Healthcare	Cost saving	Highly cost-effective (US$3194)	Highly cost-effective (US$6857)
Societal	Cost saving	Highly cost-effective (US$116)	Highly cost-effective (US$3779)
L: 96% 3 doses, 2% 2 doses, 2% 1 dose; 80% efficacy against hospitalizations and ED visits for 2 dose regimen, 70% efficacy against hospitalizations and ED visits for 1 dose regimen
Healthcare	Cost saving	Highly cost-effective (US$2044)	Highly cost-effective (US$5565)
Societal	Cost saving	Cost saving	Highly cost-effective (US$2435)
M: 12,900 hospitalizations; average length of stay 4.4 days; cost per hospital day US$173.21
Healthcare	Cost saving	Highly cost-effective (US$1577)	Highly cost-effective (US$5069)
Societal	Cost saving	Cost saving	Highly cost-effective (US$2220)
N: 25,150 hospitalizations; average length of stay 4 days; cost per hospital day US$92.22
Healthcare	Cost saving	Highly cost-effective (US$2109)	Highly cost-effective (US$5601)
Societal	Cost saving	Cost saving	Highly cost-effective (US$1587)
O: 12,755 ED visits; 135,570 OP visits^c; cost per outpatient visit US$25.88
Healthcare	Highly cost-effective (US$1902)	Highly cost-effective (US$5395)	Highly cost-effective (US$8887)
Societal	Cost saving	Highly cost-effective (US$2262)	Highly cost-effective (US$5754)
P: 36,636 ED visits; 389,364 OP visits^c; cost per outpatient visit US$11.86
Healthcare	Cost saving	Highly cost-effective (US$3331)	Highly cost-effective (US$6823)
Societal	Cost saving	Highly cost-effective (US$255)	Highly cost-effective (US$3748)

^aIn Taiwan, the per capita GDP was US$18,303 for 201034; the WHO Commission on Macroeconomics and Health suggested that interventions are highly cost-effective if the ICER is lower than the per capita GDP; may be cost-effective if the ICER is less than 2–3-times the per capita GDP (US$36,606–54,909) and are cost-ineffective if the ICER is higher than 3-times the per capita GDP.

^bCurrency conversion US$1 = 31.55 NTD.

^cWe assumed that 8.6% of the ambulatory visits were ED visits and 91.4% were physician outpatient visits5.

The utility weights assumed do not affect the prices at which vaccination is cost neutral, but they do affect the cost/QALY gained. If we limit the quality-of-life impact of rotavirus gastroenteritis to the sick child without considering the impact on the parents and use the utility weight derived from the HUI2, vaccination is again cost saving at US$20 per dose and still highly cost-effective at US$30 per dose. The incremental cost-effectiveness ratios are US$15,702 (495,398 NTD) and US$7,279 (229,653 NTD) from the healthcare and societal perspectives, respectively. If we include the effect of rotavirus gastroenteritis on the sick child as well as the child’s parents and use the utility weights derived from the Visual Analog Scale the incremental cost-effectiveness ratios are US$4,748 (149,800 NTD) and US$2,201 (69,442 NTD) from the healthcare and societal perspectives. Nevertheless, vaccination is expected to be cost-effective or cost-saving regardless of the utility weights assumed at prices ranging from US$20–30.

It should also be noted that if high levels of coverage are expected in Taiwan, the efficacy associated with incomplete regimens has little effect on the results. In a worst case scenario, if 90% of the population receives three doses, with the remaining 10% receiving only one or two doses, and we assume no efficacy for partial vaccination, the price at which vaccination is cost neutral is US$20.65 (652 NTD) and US$24.80 (782 NTD) from the healthcare and societal perspectives, respectively.

Finally, we also conducted a series of 2- and 3-way sensitivity analyses because the study that reported the lowest healthcare utilization also assumed the highest unit costs and the study with the lowest unit costs assumed the highest rates of healthcare utilization4,5. We therefore wanted to compare the total burden for hospitalizations and ambulatory visits from each of these studies. When the number of hospitalizations, the average hospital length of stay, and the average cost per day in the hospital are changed at the same time to reflect the assumptions in the Wu et al. cost-effectiveness analysis, universal vaccination is cost neutral at US$22.70 (716 NTD) (Table 4) and at US$25 per dose, the cost per QALY gained is US$1,577 (49,754 NTD) from the healthcare perspective (Table 3)5. In comparison, if we use the average of the range given for the number of RGE hospitalizations found in the Lu et al.4 study, universal vaccination is cost neutral at US$22.00 (694 NTD) and at US$25 per dose, the cost per QALY gained is US$2,109 (66,539 NTD) from the healthcare perspective. Both scenarios would be cost saving from the societal perspective. When the number of ambulatory visits and the cost of an outpatient visit are changed at the same time to reflect the assumptions in the Wu et al.5 cost-effectiveness analysis, universal vaccination is cost neutral at US$17.30 (546 NTD) and at US$25 per dose, the cost per QALY gained is US$5,395 (170,212 NTD). If we use the average of the range given for the number of ambulatory visits found in the Lu et al.4 study, universal vaccination is cost neutral at US$20.20 (637 NTD) and at US$25 per dose, the cost per QALY gained is US$3,331 (105,093 NTD).

Table 4.  Sensitivity analyses for costs of vaccination at which universal vaccination is cost neutral.

Scenario	Prices at which universal vaccination is cost neutral
	Healthcare perspective	Societal perspective
1-way sensitivity analyses
Scenario A: 1 Death	US$21.80; NTD^b 687.80	US$26.15; NTD 825.03
Scenario B: 22 Deaths	US$21.80; NTD 687.80	US$26.15; NTD 825.03
Scenario C: 12,900 hospitalizations	US$19.75; NTD 623.11	US$23.80; NTD 750.89
Scenario D: 25,150 hospitalizations	US$29.70; NTD 937.03	US$35.50; NTD 1120.03
Scenario E: 12,755 ED visits; 135,570 OP visits^a	US$18.30; NTD 577.37	US$22.80; NTD 719.34
Scenario F: 36,636 ED visits; 389,364 OP visits^a	US$29.15; NTD 919.68	US$33.60; NTD 1060.08
Scenario G: US$ 61.71 per day missed from work	US$21.80; NTD 687.80	US$29.25; NTD 922.84
Scenario H: 10% drop in efficacy in the 3rd–5th years after vaccination	US$20.60; NTD 649.93	US$24.60; NTD 776.13
Scenario I: Utility weight based on the HUI2 for the child without any QALY loss for the parents	US$21.80; NTD 687.80	US$26.15; NTD 825.03
Scenario J: Utility weight based on the VAS for the child and two parents	US$21.80; NTD 687.80	US$26.15; NTD 825.03
2- and 3-way sensitivity analyses
Scenario K: 90% 3 doses, 5% 2 doses, 5% 1 dose; 0 efficacy for incomplete regimens	US$20.65; NTD 651.51	US$24.80; NTD 782.44
Scenario L; 96% 3 doses, 2% 2 doses, 2% 1 dose; 80% efficacy against
hospitalizations and ED visits for 2 dose regimen, 70% efficacy against hospitalizations and ED visits for 1 dose regimen	US$22.05; NTD 695.68	US$26.50; NTD 836.08
Scenario M: 12,900 hospitalizations; average length of stay 4.4 days; cost per hospital day US$173.21	US$22.70; NTD 716.19	US$26.80; NTD 845.54
Scenario N: 25,150 hospitalizations; average length of stay 4 days; cost per hospital day US$92.22	US$22.00; NTD 694.10	US$27.70; NTD 873.94
Scenario O: 12,755 ED visits; 135,570 OP visits^a; cost per outpatient visit US$25.88	US$17.30; NTD 545.82	US$21.80; NTD 687.79
Scenario P: 36,636 ED visits; 389,364 OP visits^a; cost per outpatient visit US$11.86	US$20.20; NTD 637.31	US$24.60; NTD 776.13

^aWe assumed that 8.6% of the ambulatory visits were ED visits and 91.4% were physician outpatient visits5.

^bCurrency conversion US$1 = 31.55 NTD.

Discussion

Like many industrialized countries, Taiwan needs to balance access to quality healthcare with cost. Nearly the entire population is covered by national health insurance and healthcare costs are rising rapidly. Many new products are being developed and the government needs to select which technologies will contribute most to the overall health of the population within budgetary constraints in order to contain healthcare costs.

This cost-effectiveness analysis demonstrates that universal vaccination with RV5 will most likely be cost saving or highly cost-effective at prices ranging from US$20–30 per dose in Taiwan. (From the healthcare perspective, a cost-saving strategy would lower the total healthcare expenditure by the national health insurance, as the decrease in costs associated with RGE would cover the cost of the vaccine program. Even if there is an increase in cost to the national health insurance program, vaccination might be cost-saving from the societal perspective when offsets in cost to the parents are also included.)

Although it is unlikely all children will receive three doses of RV5, very high levels of coverage are expected in Taiwan if the vaccine is publicly reimbursed based on the percentage of children receiving three doses of DTwP at 2, 4, and 6 months of age37. At high levels of coverage with ≥90% completing the three dose regimen, the assumptions regarding the efficacy of incomplete regimens has little effect on the results.

There are several published studies on the RGE burden in Taiwan, with varying estimates of the number of hospitalizations, ambulatory care visits, and medical care costs. The assumptions related to healthcare resource utilization in the base case scenario are conservative, but they are between the highest and lowest estimates reported in the literature. The base case scenario assumed that in the absence of vaccination there would be 122,526 symptomatic episodes of RGE in a single birth cohort followed for the first 5 years of life with 15,400 hospitalizations and 236,300 ambulatory visits. In comparison, Wu et al.5 assumed 185,149 symptomatic episodes of RGE. Based on other published studies, the estimates of RGE hospitalizations ranged from 12,900–28,4924,5 and the estimates of RGE ambulatory visits ranged from 148,325–596,4004,5.

The results demonstrate that when the healthcare utilization and cost estimates from the published studies are applied together in multi-way sensitivity analyses, the higher number of hospitalizations and ambulatory visits in the Lu et al.4 study are partially offset by much lower cost estimates for medical care. Conversely, the lower number of hospitalizations is offset by higher cost estimates in the Wu et al.5 cost-effectiveness analysis. As a result, the incremental cost-effectiveness ratios and the cost neutral prices for the 2- and 3-way sensitivity analyses are closer to the results for the base case scenario than varying the number of hospitalizations and ambulatory care visits without changing the cost estimates.

This study also provides the cost per QALY gained in addition to the price at which vaccination is cost neutral because the latter measure does not allow us to evaluate the value of the product if the cost of vaccination exceeds the benefits. Most new technologies are cost additive, but they may represent good value for the money spent. The QALY measure assesses this and provides a common metric by which decision makers can compare different interventions. However, the derivation of utility weights for young children is particularly challenging, especially since they are often not able to answer questions for themselves and their cognitive abilities change over time44,45. The issue of whether to include the impact of a condition on caregivers is also controversial44. The results are sensitive to these choices and each of the methods to derive utility weights has different strengths and weaknesses. The HUI2 and the EQ-5D are choice-based valuation methods consistent with economic theory, but they are focused on functional limitations that may not apply to young children. Studies that ask parents to complete the HUI2 for their young children describe parental difficulties in interpreting the meaning of items given the child’s developmental age4,5. In contrast, the VAS is not based on economic theory, but it is also not limited to specific functional limitations and parents have more freedom to determine how the manifestations of rotavirus gastroenteritis apply to their situation.

Although the results are not dramatically different from the results of the Wu et al.5 cost-effectiveness analysis, the per dose prices evaluated in this study are lower than those evaluated in the earlier publication and it includes an extensive sensitivity analysis to account for the range of healthcare utilization estimates derived from published studies. Like the Wu et al.5 cost effectiveness analysis, our study does not account for herd immunity, an indirect effect of vaccination whereby children are less susceptible to an infectious condition due to an overall decrease in the amount of circulating virus in the environment. However, given that high levels of coverage are anticipated in Taiwan, the effects of herd immunity are likely to be short-term because nearly all children will be protected through vaccination.

Conclusion

Universal vaccination would most likely result in substantial reductions in morbidity and healthcare costs in Taiwan. At a price of US$20 per dose, vaccination with RV5 is likely to be cost-saving from the healthcare and societal perspectives. Although vaccination may not be cost saving at prices ranging from US$25–30 per dose from the healthcare perspective, a pentavalent rotavirus vaccination program is likely to reduce the healthcare burden associated with rotavirus gastroenteritis at a cost per QALY ratio within the range defined as cost-effective.

Transparency

Declaration of funding

This study was funded by Merck & Co., Inc., North Wales, Pennsylvania, USA.

Declaration of financial/other relationships

R.F.I., A.C.E., and J.R.C. are employees of and own Merck & Co. stock. C.L. is an employee of MSD, Taipei, Taiwan. P.Y.C. was an investigator who enrolled subjects at the study sites and coordinated the clinic sites for the Mast et al. 2010 study. As an investigator, he was compensated by Merck & Co. for all activities related to the execution of the study.

Acknowledgments

The authors would like to thank Karen Collins of JK Associates, Inc., Conshohocken, PA, for assistance with manuscript preparation on behalf of Merck & Co., Inc.