The potential economic value of influenza vaccination for healthcare workers in The Netherlands

Background Despite the clinical evidence, influenza vaccination coverage of healthcare workers remains low. To assess the health economic value of implementing an influenza immunization program among healthcare workers (HCW) in University Medical Centers (UMCs) in the Netherlands, a cost‐benefit model was developed using a societal perspective. Methods/Patients The model was based on a trial performed among all UMCs in the Netherlands that included both hospital staff and patients admitted to the pediatrics and internal medicine departments. The model structure and parameters estimates were based on the trial and complemented with literature research, and the impact of uncertainty explored with sensitivity analyses. Results In a base‐case scenario without vaccine coverage, influenza‐related annual costs were estimated at € 410 815 for an average UMC with 8000 HCWs and an average occupancy during the influenza period of 6000 hospitalized patients. Of these costs, 82% attributed to the HCWs and 18% were patient‐related. With a vaccination coverage of 15.47%, the societal program’s savings were € 2861 which corresponds to a saving of € 270.53 per extended hospitalization. Univariate sensitivity analyses show that the results are most sensitive to changes in the model parameters vaccine effectiveness in reducing influenza‐like illness (ILI) and the vaccination‐related costs. Conclusion In addition to the decreased burden of patient morbidity among hospitalized patients, the effects of the hospital immunization program slightly outweigh the economic investments. These outcomes may support healthcare policymakers’ recommendations about the influenza vaccination program for healthcare workers.

(WHO) recommends vaccination of all healthcare workers worldwide to protect staff and prevent potential transmission to their patients, but the response to this recommendation differs significantly between countries, professional organizations, advisory committees, and employers. [11][12][13] In the United States, vaccination coverage among healthcare workers increased from 63.5% during the 2010-2011 season to 77.3% during the 2014-2015 season. [14][15][16][17] However, in Europe, healthcare workers are less compliant, with reported vaccine coverage as low as 30% or less. [18][19][20] In general, the vaccination rate in the Netherlands and the United Kingdom is among the highest in Europe, although the coverage rate has been decreasing in the Netherlands since 2008 among the target population according to age (60 years and older) and/or certain medical conditions. 21 Since 2008, the vaccination coverage decreased in previous years from 71.5% in 2008 to 52.8% in 2014. 19,21,22 With the decreasing trend in the vaccine coverage rate of patients, it might become more important focusing on the vaccination rate of healthcare workers.
In long-term care settings, four clinical trials are performed and despite the differences between the trials, they observed a decrease in patient morbidity or mortality after vaccine coverage increased. [2][3][4]8 For acute care settings, which treat patients during epidemics, the number of trials is limited. As these settings were not applicable to acute care settings, Riphagen et al performed a trial in acute care settings in the Netherlands. 7,23 Moreover, while influenza immunization is safe and relatively cheap, evidence on the economic benefits is not widely available for various healthcare settings, but this is an important aspect for hospital managers and policymakers to support such a program. 24 To get a better understanding of the health economic benefits of a vaccination program for healthcare workers, we performed a modeling study using as much input as possible from a clinical trial and complemented information with additional data not provided by the trial. 7,23 The basis for the modeling study is a clustered random-

| Trial design and participants
The trial study design has been reported earlier. 7 The study as-

| Healthcare workers
The primary outcome measure of the trial was the influenza vaccine uptake among all HCWs at UMC level. Vaccine uptake was measured using the data of vaccinated persons, and this was divided by the total HCW population. Secondary outcome measures were absenteeism rates among HCWs during the month December of each study year as this was the month in which influenza peaked. 25 The absenteeism rate was not extrapolated to the whole influenza season because of the rapid increase before, and the rapid decrease in the incidence after the month December, and to avoid substantial misclassification. 24 These outcomes have been included in the present cost-benefit study and completed with data from sources such as the average number of days of work absence, the number of GP visits following ILI, and the number of GP visit due to side effects following vaccination. The parameters are also expressed in monetary units as described in detail further below.

| Patients
As a secondary outcome, patient outcome data from two selected high-risk departments (ie, Pediatrics and Internal Medicine) in the trial were collected retrospectively for all patients who were hospitalized 3 days or more to ensure nosocomial exposure during both study epidemic seasons. The outcomes were laboratory-confirmed influenza and/or pneumonia, the length of hospital stay, use, and duration of intensive care and were collected by scrutinizing computerized discharge letters and laboratory outcome data from the microbiology laboratories by two reviewers. Influenza was defined as laboratory-confirmed influenza A (all subtypes) or influenza B during a hospital stay. Pneumonia was defined as any pneumonia which was clinically diagnosed during a hospital stay. Also, following the high mortality in risk groups during influenza epidemics, influenza mortality would be a valuable outcome measure. However, it appeared to be impossible to collect these data in the UMCs because of the absence of a good registration system for death. Thus, because it was impossible monitoring any mortality following hospital-acquired pneumonia in the participating centers, it was decided not to include this in the calculations, and, therefore, the estimated outcomes can be considered conservative.
To estimate the effects on the reduction in the incidence of healthcare-associated influenza and/or pneumonia for different vaccine coverage rates, a linear relationship between vaccine coverage rates of HCWs and the proportion of patients with outcomes was assumed according to the mathematical model by Van den Dool et al. 26 In the estimates, an average of 23

| Cost estimates
The cost estimates associated with the immunization program were based on Dutch guidelines for cost-effectiveness research. 27 The cost prices were indexed to the 2014 level. 28 For healthcare workers, the direct medical costs and indirect non-medical costs related to the research objective have been used and for patients the costs of the extended hospitalizations. For an overview of the model parameters used, see Table 1.

| Cost estimates for the healthcare workers
Costs associated with the immunization program, direct medical costs resulting from an influenza episode, and the effect vaccination has on the productivity are considered and the values used are substantiated further below.

| Costs associated with the immunization program and vaccine efficacy
The cost estimates of the influenza vaccination program were estimated at €15.00 per staff member and included the costs for the vaccine (approximately 5 euro), the communication, and implementation of the program. 24 In the study by Hak et al, 24 the potential cost savings were determined using plausible, but theoretical, effects in a UMC setting using the data from the University Medical Center Groningen. For the administration, a nurse gross salary (scale 9) per month was assumed with 5 minutes for vaccination of one staff member and another 5 minutes for correction of inefficiency (waiting time).
The assumed costs currently assume a linear relationship between the number of persons vaccinated and the total cost for the vaccination campaign. Indirect costs due to productivity loss for the administration of the vaccine were assumed to be virtually absent because of the elasticity in working hours.
The vaccine efficacy for preventing ILI was assumed to be 20% because only vaccination status was available for HCWs and this was linked to the absence registry, but no laboratory-confirmed influenza was measured. 6

| Direct medical effect cost estimates
The direct medical effects of immunizing staff members against influenza are associated with seeking medical care for influenza.

| Working days lost due to influenzalike illness
The productivity loss for the healthcare workers was calculated using

| Cost estimates for patients
The main patient-related outcome was hospital-acquired influenza and/or pneumonia and the costs related to the treatment and the

| Cost-benefit analysis
The decision tree was developed using Excel for Windows, version 2010. The deterministic probability analysis is based on trial data and on existing literature. In the analyses, both the societal perspective and the hospital perspective considered. Outcomes presented are incremental costs for the healthcare staff, the patients, total incremental costs, and the costs per extended patient hospitalization.
The results are presented from a societal perspective, and the hospital perspective was also explored to consider aspects specifically relevant for the hospital as an employer.
Further, univariate deterministic sensitivity analyses were conducted to obtain the most influential model parameters on the outcome measure using plausible ranges. A multivariate probabilistic sensitivity analysis (PSA) with 10 000 iterations was performed to review the uncertainty of the model parameters simultaneously. The outcomes of the PSA are described in the results paragraph with the 95% confidence intervals (95% CI). The values used in the sensitivity analyses are presented in Table 1.

| RE SULTS
For an average UMC with 8000 staff members and 6000 patients,

| Hospital/employer's perspective
From the hospital employer's perspective, when only total influenzarelated costs for the healthcare staff are included, the total incremental investment for the vaccination program is € 8508 and the total cost per prevented extended hospitalization € 808. 30. For an overview of the results from both perspectives, see Table 2.

| Sensitivity analyses
In the base case, when only the vaccination-related costs are varied, the breakeven point is reached when the vaccination-related costs are € 17.31 per vaccinated person. Above this cost, vaccinating healthcare workers is not cost saving from a societal perspective.
From a hospital perspective, the vaccination-related costs should not exceed € 8.13 to be cost saving. In case the work absence due to ILI is 25% lower (3 instead of 4 days), the vaccination program is breakeven (with a saving of € 289), but when the reduction in work absence due to ILI is lower than 3 days, the program is not cost saving anymore. The multivariate probabilistic sensitivity analyses on the incremental total costs of both healthcare workers and patients show that the 95% confidence interval (95% CI) of the outcome incremental costs for both patients and healthcare workers correspond to average costs of-€2861 with 95% CI from-€21 828 and €11 931. For the outcomes of the PSA, see Table 2. program among hospital staff. Main savings are the reduced productivity loss of the healthcare staff and in a minor part from reduced healthcare-associated influenza infections for patients.

| D ISCUSS I ON
To value these findings, some aspects need to be considered.
The input for the analysis was largely based on the established effects of the trial, and potential limitations and strengths of the trial have been discussed earlier but some relevant aspects for the costbenefit analyses are discussed here.
First, the absenteeism rate (4.6%) was estimated using the work absence registration from the hospitals participating in the trial. In the trial, a slight increase in absenteeism rates was reported in intervention as compared with control UMCs. It is likely a proxy for stricter regulations regarding working when staff has influenza and is not a result of the vaccination program. Therefore, it was decided to use the 4.6% for the situation where no vaccination was available.  47,48 They graded the available clinical evidence of influenza vaccination of healthcare workers on patients and healthcare workers and concluded that the quality of the evidence for the effect of HCW vaccination on mortality and influenza cases in patients was "moderate" and "low." Thus, including the mortality rate is expected to be of importance, but additional research is necessary to substantiate the influenzaassociated mortality rate further. Consequently, the patient-related results presented in this study are conservative estimates.
Finally, the trial included both an influenza pandemic and an epidemic. It is expected that the pandemic and the anxiety among healthcare workers have affected the trial results and, therefore, the robustness of the results presented here. However, this was the case in both the intervention and control group, and still, statistically significant differences were found between both groups. The influenza attack rate included in the calculations affects the outcomes of both patients and HCWs. In 2009, during the pandemic, the influenza attack rate was somewhat higher than generally, and during the 2010/2011 season, it was low. Because the average of the two seasons is used in the calculations, it is expected that a general epidemic was simulated.

| CON CLUS ION
The vaccination program is likely cost saving from a societal perspective. From the hospital perspective, it requires an investment by the hospital management, but the biggest return on investment is also for the hospital. However, these investments are not supported by financial incentives in the current system.
Studies are warranted that focus on the effect of vaccination programs in peripheral hospitals, on the costs for hospitals involved when preparing and organizing the influenza vaccination and on the effect of influenza vaccination on the mortality rate.