An economic assessment of high-dose influenza vaccine
van Aalst, Robertus
DOI:
10.33612/diss.127973664
IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.
Document Version
Publisher's PDF, also known as Version of record
Publication date: 2020
Link to publication in University of Groningen/UMCG research database
Citation for published version (APA):
van Aalst, R. (2020). An economic assessment of high-dose influenza vaccine: Estimating the vaccine-preventable burden of disease in the United States using real-world data. University of Groningen. https://doi.org/10.33612/diss.127973664
Copyright
Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).
Take-down policy
If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.
Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.
4
Comparing the impact of high-dose versus standard dose influenza vaccines on hospitalization cost for cardiovascular and respiratory diseases: economic assessment in the US Veteran population during 5
respiratory seasons using an instrumental variable method Robertus van Aalst a,b, Ellyn M. Russo c, Nabin Neupane c, Salaheddin M. Mahmud d,e, Jan Wilschut f, Sandrine I. Samson g, Ayman Chit b,h, Maarten Postma a,i,j, Yinong Young-Xu c,k
a Department of Health Sciences, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands
b Regional Epidemiology and Health Economics, Sanofi Pasteur, Swiftwater, PA, USA
c Clinical Epidemiology Program, Veterans Affairs Medical Center, White River Junction, Vermont, USA d Department of Community Health Sciences, College of Medicine, University of Manitoba, Winnipeg, MB,
Canada
e George & Fay Yee Center for Healthcare Innovation, University of Manitoba, Winnipeg Regional Health Authority, Winnipeg, MB, Canada
f Department of Medical Microbiology, University Medical Center Groningen, University of Groningen, Groningen, the Netherlands
g Global Medical affairs, Sanofi Pasteur, Lyon, France
h Leslie Dan Faculty of Pharmacy, University of Toronto, Toronto, Ontario, Canada
i Unit of PharmacoTherapy, -Epidemiology & -Economics (PTE2), University of Groningen, Department of Pharmacy, Groningen, the Netherlands
j Department of Economics, Econometrics & Finance, University of Groningen, Faculty of Economics & Business, Groningen, the Netherlands
k Department of Psychiatry, Geisel School of Medicine at Dartmouth, Hanover, New Hampshire, USA
ABSTRACT
Objectives
Cost savings associated with high-dose (HD) as compared to standard-dose (SD) influenza vaccination in the United States (US) Veteran’s Health Administration (VHA) population have been attributed to better protection against hospitalization for cardiac and respiratory diseases. The relative contribution of each of these disease categories to the reported savings remains to be explored.
Methods
During a recently completed study of HD versus SD vaccine effectiveness (conducted in the VHA over five respiratory seasons from 2010/11 through 2014/15), we collected cost data for all healthcare services provided at both VHA and Medicare-funded facilities. In that analysis, we compared the costs of vaccination and hospital care for patients admitted with either cardiovascular or respiratory disease. Treatment selection bias and other confounding factors were adjusted using an instrumental variable (IV) method. In this brief report we use the same study cohort and methods to stratify the results by patients admitted for cardiovascular disease (CVD) and those admitted for respiratory disease.
Results
We analyzed 3.5 million SD and 0.16 million HD person-seasons. The IV-adjusted rVEs were 14% (7% - 20%) against hospitalizations for CVD and 15% (5% - 25%) against respiratory hospitalizations. Net cost savings per HD recipient were $138 ($66 - $200) for CVD related hospitalizations and $62 ($10 - $107) for respiratory disease related hospitalizations.
Conclusions
In the US VHA population, the reduction in hospitalizations for CVD over five respiratory seasons contributed twice the cost savings (per HD recipient) of the reduction in hospitalizations for respiratory disease.
BACKGROUND
Adults 65 years and older (hereinafter referred to as seniors) are at an increased risk for complications caused or triggered by an influenza infection [1]. Young-Xu and colleagues estimated the range of annual direct medical costs of influenza-attributable hospitalizations at Department of Veterans Affairs (VA) Medical Centers for senior Veterans Health Administration (VHA) enrollees over five respiratory seasons (2010/11 through 2014/15) to be between 24 and 34 million US dollars [2]. One of the vaccination options available to the VHA during this period was the injectable high-dose inactivated trivalent influenza vaccine (Fluzone® High-Dose, Sanofi Pasteur, PA, US, licensed in the US in 2009 for people aged 65 years and older; hereinafter referred to as the high-dose vaccine (HD)). HD contains four times more influenza hemagglutinin antigen than standard-dose trivalent influenza (SD) vaccines (60 μg vs. 15 μg per strain), improving immune response and protection, in seniors [3]. In a recent study [4] we estimated that, with an average HD coverage rate of 4.4% of all influenza vaccines administered to seniors seeking care at VHA facilities during this five season period, HD was associated with a 14% (95% CI: 8% - 19%) additional reduction in hospitalizations for either cardiovascular (CVD) or respiratory disease as compared to SD. As a result of the reduced hospitalizations, we estimated that the US taxpayer achieved averaged annual net cost savings of 6.4 million US dollars (95% CI: $3.6 – $8.8 million); however, the relative contribution of each of the disease categories to the reported savings remains to be explored. While the literature is quite decisive on the causal relationship between influenza vaccination and prevention of respiratory complications, evidence for this relationship on especially the magnitude of
the prevention of CVD is still developing [5, 6]. The objective of this brief report is, to assess the relative contribution of CVD and respiratory diseases to the aggregate vaccine effectiveness and cost savings.
METHODS
Study Design, Population and Data Sources
The Van Aalst et al. (2019) study [4], a retrospective cohort study with approximately 700,000 patients included in each of the five respiratory seasons, compared hospitalizations between those who received HD versus SD at a VA facility. Patients
were included when they were at least 65 years old at vaccination, had received only one HD or SD vaccine in the seasons of interest, and had sought medical care at a VA facility in the six months before vaccination. We used the same population and methods of Van Aalst et al. (2019) to calculate rVEs for the present study. In summary, for each study participant at each season, the baseline period (during which baseline characteristics were measured) was defined from the beginning of each respiratory season in week 27 (beginning of July) until his or her influenza vaccination date. The observation period (during which study outcomes were measured) was defined from two weeks after vaccination until the end of the respiratory season in week 26 (end of June). Crude rVE rates were adjusted for treatment selection bias (confounding by indication) using differences in observable baseline characteristics between the cohorts that included demographics, comorbidities adapted from the Deyo-Charlson comorbidity score [7], and VA priority group, a surrogate measure for socio-economic status (Appendix 1) [8]. In addition, the same instrumental variable (IV), a facility’s preference for HD use defined as the proportion of HD recipients at a certain facility in a given respiratory season, was used to act as a pseudo-randomizer of unobservable differences (Appendix 12) [4, 9].
For the cost of vaccination in VA facilities, we obtained data from the National Acquisition Center Contract Catalog Search Tool [10]. Hospitalizations, and their reimbursement costs, of VHA enrollees that occurred in non-VA facilities were obtained from the Centers for Medicare and Medicaid Services (CMS) administrative fee-for-service claims. These records supplement those in the VHA database as many patients seek healthcare outside VA once eligible for CMS benefits. While VHA applies a system of cost allocation, costs of non-VA hospitalizations are based on insurance reimbursements, which do not necessary reflect true costs for the healthcare provider [11]. The study received institutional review board approval from the Veteran’s Institutional Review Board of Northern New England at the White River Junction VA Medical Center.
Outcomes and IV-adjusted rVEs
Our outcomes were an acute hospitalization for CVD, defined by its principal discharge diagnosis (International Classification of Diseases, Ninth Revision, [ICD-9]: 390-459) and acute hospitalizations for respiratory disease (ICD-9: 460-519, Appendix 2). For ease of comparison, we report the earlier published aggregated outcome, a
hospitalization for either CVD or respiratory disease (ICD-9: 390-519). Additionally, we explored stratification of outcomes by more specific disease groups (e.g. hospitalizations for pneumonia, acute myocardial infarction, Appendix 11). To adjust for measured and unmeasured confounding, crude rVEs for each outcome were adjusted using an instrumental variable (IV) analysis (Appendix 12).”
Economic Assessment
The need to adjust the crude rVE for treatment selection bias prevented us from a straight comparison of costs incurred by the HD recipients to those incurred by the SD recipients. We applied the model described by Van Aalst et al. (2019) and included the same sensitivity analyses (Appendix 6 – 10). In summary, the total observed number of hospitalizations were assigned to the HD and SD recipients using the IV-adjusted rVE. After adjusting the observed outcomes with the season and outcome-specific rVE, we calculated the absolute risk reduction [ARR] by subtracting the incidence rate in the HD cohort from the rate in the SD cohort. The multiplicative inverse of ARR results in the number needed to vaccinate (NNV = 1/ARR): the number of patients that need to be vaccinated with HD instead of SD to prevent one additional hospitalization. To evaluate cost savings of HD vaccination, we estimated the difference in costs per SD recipient as if they had received HD instead. This was calculated as the average cost of a hospitalization for an SD recipient divided by the NNV minus the average cost difference of administering the two vaccines. Vaccine cost included the cost of the vaccine itself as well as the cost of administrating it; ascertained by their current procedural terminology (CPT) code (Appendix 3). We calculated the total realized cost savings by multiplying the total number of HD recipients by the cost savings per patient. The potential savings were calculated under the assumption that 10% of the study population had received HD (assuming a continuation of the upward trend: 3.3% in 2013/14 and 7.7% in 2014/15, Appendix 4).
RESULTS
During the five-season study period, we analyzed 3.6 million person-seasons (Table 1). We observed 314,014 hospitalizations for CVD in our study cohort. We estimated the rVE (HD vs SD) for acute CVD hospitalizations at 14% (95% CI: 7% - 20%). In each study season HD was associated with reduced hospitalizations for CVD (Appendix
4). IV-adjusted hospitalization rates (per person-season) were 0.087 (95% CI: 0.087 – 0.087) for SD recipients and 0.075 (95% CI: 0.070 – 0.080) for HD recipients. Based on these rates, we calculated an NNV with HD instead of SD of 84 (95% CI: 59 – 160) to prevent one additional hospitalization for CVD. We observed 164,948 hospitalizations for respiratory disease – about half the number of those hospitalized for CVD (Table 1). We estimated an rVE for hospitalizations for respiratory disease of 15% (95% CI: 5% – 25%) and an NNV of 144 (95% CI: 89 – 473) to prevent one additional hospitalization for respiratory disease.
Table 1. Number of influenza vaccinations, hospitalization rates for cardiovascular, respiratory and
either cardiovascular or respiratory disease, and number needed to vaccinate (NNV) to prevent one additional hospitalization for VHA enrollees vaccinated during respiratory seasons 2010/11 through 2014/15
Study cohort 3,638,924
HD recipients 158,636 4.4%
SD recipients 3,480,288 95.6%
Cardiovascular Respiratory Either*
Observed hospitalizations 314,014 164,948 478,962
Applied rVE 14% (7% - 20%) 15% (5% - 25%) 14% (8% - 19%)
Hospitalization incidence rates
Rate among HD recipients 0.075 (0.070 - 0.080) 0.039 (0.035 - 0.0433) 0.114 (0.108 - 0.121)
Rate among SD recipients 0.087 (0.087 - 0.087) 0.046 (0.046 - 0.0454) 0.132 (0.132 - 0.133)
Vaccine effect
Number needed to
vaccinate (NNV) 84 (59 - 160) 144 (89 - 473) 55 (40 - 93)
*Data in the ‘Either’ column has been published previously [4], and is added for ease of comparing the stratified results with the aggregate.
The average cost for SD recipients of a VA hospitalization was $16,523 (95% CI: $16,269 – $16,781) for CVD and $15,497 (95% CI: $15,136 – $15,872) for respiratory disease (Table 2). Average CMS reimbursement to a non-VA facility was $10,320 (95% CI: $10,231 – $10,411) per hospitalization for CVD and $8,720 (95% CI: $8,636 – $8,803) for respiratory disease.
Table 2. Mean cost and reimbursement per hospitalization, in US dollars, for cardiovascular,
respiratory and either cardiovascular or respiratory disease among vaccinated VHA enrollees during the 2010/11 through 2014/15 respiratory seasons
Cardiovascular Respiratory Either*
Hospitalization Mean (95% CI) Wt.1 Mean (95% CI) Wt.1 Mean (95% CI) Wt.1
VHA cost 16,523 (16,269 - 16,781) 35% 15,497 (15,136 - 15,872) 26% 16,220 (16,009 - 16,430) 32%
CMS
reimbursement 10,320 (10,231 - 10,411) 65% 8,720 (8,636 - 8,803) 74% 9,716 (9,652 - 9,781) 68%
Average cost2 12,490
(12,343 - 12,639) 10,499 (10,342 - 10,659) 11,796 (11,685 - 11,907)
1Weight(Wt) is based on observed hospitalizations in VHA and non-VHA facilities incurred by HD and SD recipients. 2Average cost of one hospitalization is the weighted average of the VHA cost and CMS reimbursements for an SD recipient. *Data in the ‘Either’ column has been published previously [4], and is added for ease of comparing the stratified results with the aggregate.
We estimated the savings per HD-vaccinated VHA patient to be $138 (95% CI: $66 – $200, Table 3) due to reduced hospitalizations for CVD. Estimated total savings were $138 x 158,636 HD recipients = $22 million (95% CI: $11 – $32 million) based on an HD coverage rate of 4.4%. Estimated potential savings under the assumption that 10% of the study population had received HD are $138 x 363,892 HD recipients = $50 million (95% CI: $24 – $73 million). Reduced hospitalizations for CVD contributed for 69% to the cost savings due to reduced hospitalizations for either CVD or respiratory disease. The remaining 31% of cost savings were realized by reduced hospitalizations for respiratory disease. Where the rVEs for hospitalizations related to CVD ranged from 10% in seasons 2011/13 to 15% in season 2013/14, the rVEs for hospitalizations related to respiratory disease ranged from 3% in 2011/12 (not statistically significant) to 23% in 2012/13 (Appendix 4). As a result, vaccination with HD instead of SD was cost saving in each of the five seasons in our study as a result of reduced hospitalizations for CVD (Appendix 5). In contrast, vaccination with HD instead of SD was cost saving in some seasons and on average over five seasons due to reduced hospitalizations for respiratory disease. In the seasons HD was not cost saving, the net savings per patient were positive but not statistically significant.
Table 3. Estimation of realized and potential net cost savings, in US dollars, among vaccinated VHA
enrollees due to reduced hospitalizations for cardiovascular, respiratory and either cardiovascular or respiratory disease during the 2010/11 through 2014/15 respiratory seasons
Cardiovascular Respiratory Either*
Net cost savings Mean (95% CI) Mean (95% CI) Mean (95% CI)
Per patient SD➝HD 138 (66 - 200) 62 (10 -107) 202 (115 - 280) Total 22M (11M - 32M) 10M (2M - 17M) 32M (18M - 44M) Potential VHA 18M (8.4M - 26M) 5.7M (1.0M - 10M) 23M (13M - 33M) Potential Medicare 32M (16M - 47M) 16M (3.0M - 29M) 50M (29M - 69M) Potential Total 50M (24M - 73M) 22M (4M - 39M) 73M (42M - 102M) Contribution 69% 31% 100%
*The data in the ‘Either’ column has been published previously [4], and is added for ease of comparing the stratified results with the aggregate.
DISCUSSION
We estimated the average rVE (HD vs SD) over five years for prevention of acute CVD hospitalizations at 14% (95% CI: 7% - 20%), Table 1. Although the rVEs of HD vs SD for prevention of CVD and respiratory disease hospitalizations were similar, the average cost savings due to prevented hospitalizations for CVD were significantly higher than the average cost savings due to prevented hospitalizations for respiratory disease. Net savings per vaccinated patient (HD instead of SD) were $138 (95% CI: $66 - $200) for CVD hospitalizations compared to $62 (95% CI: $10 - $107) for hospitalizations for respiratory disease. This difference was mainly due to the higher incidence rate of CVD hospitalizations, which was almost twice as high as the incidence rate of hospitalizations for respiratory disease (Table 1).
As previously reported, the average costs of an HD and SD vaccination during this period were $23.48 (95% CI: $21.29 – $25.85) and $12.21 (95% CI: $11.49 – $13.00) per vaccinated patient, respectively [4]. We compared the actual cost of the HD and SD vaccines in each of the five seasons with the avoided costs due to hospitalizations prevented in each season (Appendix 5) and calculated the weighted averages for the cost analysis over five seasons. Assuming a linear cost increase, the half-way point of the study (2012/13 season) can be considered as the base year of the costs presented in Table 2.
We observed that a significantly higher proportion of hospitalizations for CVD took place in VA-facilities compared to hospitalizations for respiratory disease: 35% of all hospitalizations for CVD took place in VA facilities (65% in non-VA facilities) compared to 26% of all hospitalizations for respiratory disease (74% in non-VA facilities). In other words, VHA pays for a greater share of the more expensive CVD admissions compared to the Medicare-funded non-VA admissions. In this context, preventing one admission for CVD by vaccinating 84 patients with HD instead of SD leads to significant cost savings, for VHA especially. If HD uptake was to increase from 4.4% to 10%, and assuming that increased HD vaccination will not significantly change a patient’s propensity of being hospitalized in a VH versus non-VA facility, we estimate annual net cost savings of 3.5 million for VHA (18M / 5 years). Under similar assumptions, another annual 1.1 million would be saved by reduced hospitalizations for respiratory disease (5.7M / 5 years). Although the literature has traditionally focused on the cost-effectiveness of influenza vaccination due to reduced respiratory complications, our study suggests that the majority of the cost savings are associated with the reduction of cardiovascular complications. Evidence for a causal relationship between influenza vaccination and the magnitude of the prevention of CVD is still developing [5, 6]. Our study suggests an association between influenza vaccination (HD versus SD) and reduced hospitalizations for CVD of the same magnitude as the reduction of hospitalizations for respiratory disease. The association of HD with reduced hospitalizations for Acute Myocardial Infarction (AMI) was even higher (rVE of 21% [13% - 29%], Appendix 11), suggesting HD to have a stronger treatment effect on a more influenza specific CVD outcome [12]. Caution must be exercised when interpreting our findings. Although we used an instrument (IV) that fulfills the underlying assumptions based on accepted analytical methods (Appendix 12), some residual bias caused by treatment selection bias cannot be ruled out. Like in any retrospective study relying on routinely collected data, misclassification of treatment, outcomes and baseline characteristics (Appendix 1) and missing data (information bias) cannot be ruled out [13]. We classified outcomes using the principal discharge diagnosis, defined as the ‘condition established after study to be chiefly responsible for occasioning the admission of the patient to the hospital for care’ [14]. This does not rule out the possibility that some patients might have received treatment for both cardiovascular and respiratory disease during a hospitalization that we classified as Cardiovascular (or Respiratory).
Because HD rVE estimates highly depend on specificity and severity of the outcome, seasonal heterogeneity of viral activity and vaccine strain match, population under observation, and choice of comparator vaccine, comparing our results with other studies is challenging. A recent study comparing HD with a trivalent adjuvanted SD vaccine (aIIV3) in the 2016/17 and 2017/18 seasons reported a pooled rVE of 12.0% (95% CI: 3.3% - 20%) against hospitalizations for respiratory disease and 5.7% (0.3% - 11%) against hospitalizations for cardiovascular disease [15]. Izurieta and colleagues reported that HD was associated with lower hospitalization rates for probable influenza (hospitalization with an administrative ICD-10 code of 489 on any position on the claim) when compared to aIIV3, with an rVE of 7.7%, (5.1% - 10.2%) in the 2017/18 season [16]. Compared with the standard of care, a quadrivalent SD vaccine (IIV4-SD), the rVE of HD was estimated to be 10.0% (7.8% – 12.3%) against hospitalization for probable influenza. These recent studies add to the substantial body of evidence demonstrating that HD is more effective in the prevention of influenza associated outcomes than trivalent SD [17], and are likely to be repeated when quadrivalent HD becomes available in the 2020/21 season for the U.S. population.
Strengths and limitations of the methods, including rVE estimation, have been discussed in detail elsewhere [4, 9]. Briefly, strengths include the size and longitudinal observation of the cohort over multiple seasons. Seasonal variation in influenza viral activity and vaccine efficacy portends seasonal variation in the severity of influenza; therefore, incorporating multiple seasons in this analysis increases confidence in our assessment as an average economic effect. IV estimation can adjust for selection bias caused by measured and unmeasured preferential treatment based on patient characteristics such as “frailty”. To achieve that, IV estimation requires an instrument that satisfies the several assumptions including that the instrument is not associated with the outcome. Our instrument, a facility’s treatment preference for HD, targets patients who would have received a different vaccine if they had gone to a different VA facility. Although we can’t identify these “marginal patients” in the study population, it is likely that the estimates apply to the majority of the study population. Another limitation is that our study population is not representative of the general VHA-enrolled population: we included patients who had sought medical care at a VA facility in the six months before vaccination, which excluded approximately 30% of enrollees who received an HD or SD vaccine in a VA facility. This is, however, the population that has the biggest impact on VHA’s resources, and thus, most interesting from a
policy perspective. Because VHA stopped offering HD to its patient population after the 2016/17 season, we were unable to estimate cost savings for this population in recent influenza seasons.
CONCLUSION
In the US VHA population, the reduction in hospitalizations for CVD over five respiratory seasons contributed twice the cost savings (per HD recipient) of the reduction in hospitalizations for respiratory disease.
ACKNOWLEDGEMENTS
Support for VA/CMS data provided by the Department of Veterans Affairs, VA Health Services Research and Development Service, VA Information Resource Center (Project Numbers SDR 02-237 and 98-004). This study was funded by Sanofi Pasteur.
REFERENCES
1. Lambert ND, Ovsyannikova IG, Pankratz VS, Jacobson RM, Poland GA. Understanding the immune response to seasonal influenza vaccination in older adults: a systems biology approach. Expert Rev Vaccines. 2012;11(8):985-94.
2. Young-Xu Y, van Aalst R, Russo E, Lee JK, Chit A. The Annual Burden of Seasonal Influenza in the US Veterans Affairs Population. PLoS One. 2017;12(1):e0169344.
3. DiazGranados CA, Dunning AJ, Kimmel M, Kirby D, Treanor J, Collins A, Pollak R, Christoff J, Earl J, Landolfi V, et al. Efficacy of high-dose versus standard-dose influenza vaccine in older adults.
N Engl J Med. 2014;371(7):635-45.
4. van Aalst R, Russo EM, Neupane N, Mahmud SM, Mor V, Wilschut J, Chit A, Postma M, Young-Xu Y. Economic assessment of a high-dose versus a standard-dose influenza vaccine in the US Veteran population: Estimating the impact on hospitalization cost for cardio-respiratory disease. Vaccine.
2019;37(32):4499-503.
5. Poudel S, Shehadeh F, Zacharioudakis IM, Tansarli GS, Zervou FN, Kalligeros M, van Aalst R, Chit A, Mylonakis E. The Effect of Influenza Vaccination on Mortality and Risk of Hospitalization in Patients With Heart Failure: A Systematic Review and Meta-analysis. Open Forum Infect Dis.
2019;6(4):ofz159.
6. Udell JA, Zawi R, Bhatt DL, Keshtkar-Jahromi M, Gaughran F, Phrommintikul A, Ciszewski A, Vakili H, Hoffman EB, Farkouh ME. Association between influenza vaccination and cardiovascular outcomes in high-risk patients: a meta-analysis. Jama. 2013;310(16):1711-20.
7. Deyo RA, Cherkin DC, Ciol MA. Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases. Journal of clinical epidemiolog y. 1992;45(6):613-9.
8. Petersen LA, Byrne MM, Daw CN, Hasche J, Reis B, Pietz K. Relationship between clinical conditions and use of Veterans Affairs health care among Medicare‐enrolled veterans. Health Serv
Res. 2010;45(3):762-91.
9. Young-Xu Y, Snider JT, van Aalst R, Mahmud SM, Thommes EW, Lee JKH, Greenberg DP, Chit A. Analysis of relative effectiveness of high-dose versus standard-dose influenza vaccines using an instrumental variable method. Vaccine. 2019;37(11):1484-90.
10. U.S. Department of Veterans Affairs. National Acquisition Center (CCST) [cited 2018 8/4/2018]. Available from: https://www.va.gov/nac/Pharma/List.
11. Association AH. Underpayment by Medicare and Medicaid fact sheet. January 2016 [6/12/2019]. Available from: https://www.aha.org/2016-jan-underpayment-medicare-medicaid.
12. Kwong JC, Schwartz KL, Campitelli MA, Chung H, Crowcroft NS, Karnauchow T, Katz K, Ko DT, McGeer AJ, McNally D. Acute myocardial infarction after laboratory-confirmed influenza infection.
New England Journal of Medicine. 2018;378(4):345-53.
13. Rothman KJ, Greenland S, Lash TL. Modern epidemiolog y: Lippincott Williams & Wilkins; 2008.
14. Centers for Medicare and Medicaid Services. ICD-10-CM Official Guidelines for Coding and Reporting FY 2019 [2/28/2019]. Available from: https://www.cms.gov/Medicare/Coding/ICD10/ Downloads/2019-ICD10-Coding-Guidelines-.pdf.
15. van Aalst R, Gravenstein S, Mor V, Mahmud SM, Wilschut J, Postma M, Chit A. Comparative effectiveness of high dose versus adjuvanted influenza vaccine: A retrospective cohort study. Vaccine.
2020;38(2):372-9.
16. Izurieta HS, Chillarige Y, Kelman J, Wei Y, Lu Y, Xu W, Lu M, Pratt D, Chu S, Wernecke M, et al. Relative effectiveness of cell-cultured and egg-based influenza vaccines among the U.S. elderly, 2017-18. J Infect Dis. 2018:jiy716-jiy.
17. Lee JKH, Lam GKL, Shin T, Kim J, Krishnan A, Greenberg DP, Chit A. Efficacy and effectiveness of high-dose versus standard-dose influenza vaccination for older adults: a systematic review and meta-analysis. Expert Rev Vaccines. 2018;17(5):435-43.
SUPPORTING INFORMATION
Appendix 1. Patient Characteristics during Study Period by Vaccine Type Appendix 2. Ascertainment and data sources of the cost of acute
hospitalizations
Appendix 3. Ascertainment and data sources of vaccination costs Appendix 4. Number of influenza vaccinations, hospitalization rates, and
number needed to vacinate (NNV) to prevent one additional hospitalization for VHA enrollees vaccinated during the 2010-11 through 2014-15 respiratory seasons
Appendix 5. Mean cost and reimbursement per hospitalization and
vaccination, and estimation of realized savings to US taxpayers among vaccinated VHA enrollees during the 2010-11 through 2014-15 respiratory seasons
Appendix 6. Sensitivity analysis of estimated savings to US taxpayers
among vaccinated VHA enrollees for which the observed location of the hospitalization is raised and lowered 25 percentage points
Appendix 7. Sensitivity analysis of estimated savings to US taxpayers
among vaccinated VHA enrollees for which the incremental cost of vaccinating with HD is doubled without and with a 5 percentage point increase to the cost of VHA hospitalization and CMS reimbursement
Appendix 8. Sensitivity analysis of estimated savings to US taxpayers
among vaccinated VHA enrollees for which the observed location of the hospitalization is raised and lowered 25 percentage points with a 10 percentage point reduction in the cost of VHA hospitalization and a 10 percentage point increase in CMS reimbursement
Appendix 9. Sensitivity analysis of estimated savings to US taxpayers
among vaccinated VHA enrollees for which the observed location of the hospitalization is raised and lowered 25 percentage points with a 20 percentage point reduction in the cost of VHA hospitalization and a 20 percentage point increase in CMS reimbursement
Appendix 10
Table 1. Mean cost and reimbursement per hospitalization for cardiovascular
disease among vaccinated VHA enrollees during the 2010-11 through 2014-15 respiratory seasons in US dollars, outliers included
128 130 130 131 132 133 134 135 136 137
137
138
139
Table 2. Estimation of realized and potential savings for hospitalizations
among vaccinated VHA enrollees for cardio-respiratory disease in US dollars), cost and reimbursement outliers included
Appendix 11. Number of influenza vaccinations, hospitalization rates
per 10,000 person-seasons, and number needed to vaccinate (NNV) to prevent one additional hospitalization for VHA enrollees vaccinated during respiratory seasons 2010-11 through 2014-15
Appendix 12. Basis for and Validation of the Instrument
Appendix 1. Patient Characteristics during Study Period by Vaccine Type HD SD Total Number of Person-Seasons 158,636 (4%) 3,480,288 (96%) 3,638,924 (100%) Age 65-69 53,646 (34%) 1,130,630 (32%) 1,184,276 (33%) 70-74 29,004 (18%) 645,611 (19%) 674,615 (19%) 75-79 27,740 (17%) 643,189 (18%) 670,929 (18%) 80-84 25,379 (16%) 566,571 (16%) 591,950 (16%) 85+ 22,867 (14%) 494,287 (14%) 517,154 (14%) Male 156,378 (99%) 3,427,402 (98%) 3,583,780 (98%) Race Black 24,456 (15%) 328,621 (9%) 353,077 (10%) Hispanic 7,611 (5%) 149,218 (4%) 156,829 (4%) Other 2,563 (2%) 57,943 (2%) 60,506 (2%) White 114,538 (72%) 2,652,463 (76%) 2,767,001 (76%) Unknown 9,468 (6%) 292,043 (8%) 301,511 (8%) Region Northeast 28,062 (18%) 440,419 (13%) 468,481 (13%) Midwest 27,169 (17%) 570,764 (16%) 597,933 (16%) South 48,956 (31%) 1,140,606 (33%) 1,189,562 (33%) West 15,479 (10%) 509,927 (15%) 525,406 (14%) Unknown 38,970 (25%) 818,572 (24%) 857,542 (24%) Priority Level 1 41,068 (26%) 782,576 (22%) 823,644 (23%) Level 2 10,594 (7%) 212,977 (6%) 223,571 (6%) Level 3 18,888 (12%) 360,781 (10%) 379,669 (10%) Level 4 4,996 (3%) 127,520 (4%) 132,516 (4%) Level 5 61,921 (39%) 1,390,391 (40%) 1,452,312 (40%) Level 6-8 21,169 (13%) 606,043 (17%) 627,212 (17%) Comorbidity Any malignancy 24,188 (15%) 441,466 (13%) 465,654 (13%) Metastatic solid tumor 1,165 (1%) 19,448 (1%) 20,613 (1%) Congestive heart failure 13,538 (9%) 233,461 (7%) 246,999 (7%) Chronic pulmonary disease 30,026 (19%) 561,231 (16%) 591,257 (16%) Cerebrovascular disease 12,591 (8%) 225,186 (6%) 237,777 (7%) Dementia 3,393 (2%) 50,679 (1%) 54,072 (1%) Diabetes with complications 12,460 (8%) 235,526 (7%) 247,986 (7%)
Appendix 1 (continued). Patient Characteristics during Study Period by Vaccine Type
HD SD Total
Diabetes w/o chronic complications 68,075 (43%) 1,393,512 (40%) 1,461,587 (40%)
AIDS 945 (0.6%) 9,394 (0.3%) 10,339 (0.3%)
Mild liver disease 3,442 (2.2%) 46,183 (1.3%) 49,625 (1.4%) Moderate/severe liver disease 375 (0.2%) 5,607 (0.2%) 5,982 (0.2%) Myocardial infarction 2,102 (1.3%) 43,722 (1.3%) 45,824 (1.3%) Hemiplegia/paraplegia 1,131 (0.7%) 19,748 (0.6%) 20,879 (0.6%) Peptic ulcer disease 1,114 (0.7%) 20,006 (0.6%) 21,120 (0.6%) Peripheral vascular disease 12,277 (8%) 233,269 (7%) 245,546 (7%) Rheumatoid disease 2,815 (2%) 53,870 (2%) 56,685 (2%) Renal disease 16,510 (10%) 316,305 (09%) 332,815 (9%)
Ap pe nd ix 2 . A sc er ta in m en t a nd d at a s ou rc es o f t he c os t o f a cut e h os pi ta liz at io ns Asc er ta in m en t o f “ A cu te” C os t ( V H A) & R ei m bu rs em en t ( C M S) Ty pe D ia gn os is C od e V H A s ou rc e t ab le C M S s ou rc e t ab le V H A s ou rc e t ab le C M S s ou rc e t ab le Pne um on ia or infl ue nza IC D -9 : 480 -48 8 IC D -1 0: J0 9-J1 8 Inp at .Inp at ie nt , e xc lu de w he n In pat .P at ie nt Tr an sfe r.S pe ci al ty or Inp at .P at ie nt Tr an sfe r. Be dS ec tio n i s ‘ N H ’, ‘ D O M ’, ‘R EH A B’ o r ‘H O SP IC E’ M ED PA R, in clude w he n PR V N U M i s b et w ee n 0 00 1 a nd 0 99 9 and S SL SS N F = S ( Sh or t-S ta y) a nd SP C LU N IT i s m iss in g ( no n-sw in g-be d c la ss ifi cat io n) a nd L O SC N T <3 66 , O R P RV N U M i s b et w ee n 13 00 -13 99 D SS .D IS CH . To tC os t M ED PA R .P M T_ A M T + .C O IN _A M T + . D ED_ A M T + . PR PA YA M T + CA .L IN EP M T + .C O IN A M T + . LD ED A M T + . LP RP DA M T C ard io -re sp ir at or y IC D -9 : 3 90 -51 9 IC D -1 0: I-J Ap pe nd ix 3 . A sc er ta in m en t a nd d at a s ou rc es o f v ac ci na tio n c os ts Va cc in at io n t yp e Pr oc ed ur e C od e V H A s ou rc e t ab le H ig h-do se (H D ) CPT : 90 66 2 D SS .P H A .vs _c os t + . fix di r St and ar d-do se ( SD ) CP T: 9 065 5-90 65 9; Q 20 34 -Q 20 39
pe nd ix 4 . N umb er o f i nfl ue nz a v ac ci na tio ns , h os pi ta liz at io n r at es , a nd n umb er n ee de d t o v ac ci na te ( N N V ) t o p re ve nt o ne a dd iti on al h os pi ta liz at io n f or V H A le es v ac ci na te d d ur in g t he 2 01 0-11 t hr ou gh 2 01 4-15 r es pi ra to ry s ea so ns 20 10 -11 20 11 -1 2 20 12 -1 3 20 13 -14 20 14 -1 5 N % N % N % N % N % ud y c oh or t 67 8, 32 7 10 0% 68 6, 34 7 10 0% 73 6, 637 10 0% 76 5, 81 0 10 0% 77 1, 80 3 10 0% D re ci pi en ts 20 ,8 78 3. 1% 27, 56 1 4.0% 25 ,35 4 3. 4% 25, 30 7 3. 3% 59, 53 6 7.7 % ci pi en ts 65 7,4 49 96 .9 % 65 8, 78 6 96 .0% 711 ,2 83 96 .6 % 74 0, 50 3 96 .7 % 71 2, 26 7 92 .3% bs er ve d h os pi ta liz at io ns i n s tu dy c oh or t spi ra to ry 32 ,7 79 4. 8% 32 ,2 06 4.7 % 34 ,39 0 4.7 % 31 ,31 7 4. 1% 34 ,25 6 4. 4% rd io va scu la r 62 ,18 2 9% 63 ,4 52 9% 63 ,2 74 9% 62 ,59 6 8% 62 ,51 0 8% dio -re sp ir at or y 94 ,9 61 14 % 95, 65 8 14 % 97, 66 4 13% 93 ,91 3 12% 96 ,76 6 13% pp lie d r el at iv e v ac ci ne e ffe ct iv en es s ( rV E) spi ra to ry 14 % ( 1% - 2 5% ) 3% ( -1 1% - 1 4% ) 23 % ( 13 % - 3 2% ) 9% ( -5 % - 2 2% ) 20 % ( 13 % - 2 7% ) rd io va scu la r 14 % ( 6% - 2 1% ) 10 % ( 3% - 1 7% ) 10 % ( 2% - 1 7% ) 15 % ( 8% - 2 2% ) 14 % ( 9% - 1 8% ) dio -re sp ir at or y 14 % ( 6% - 1 9% ) 8% ( 2% - 1 3% ) 13 % ( 7% - 1 9% ) 13 % ( 7% - 2 0% ) 15 % ( 11 % - 1 9% ) spit al iz at io ns fo r r es pi ra to ry d is ea se e H D 1 0. 04 2 ( 0. 03 7 - 0 .0 47 9) 0. 04 6 ( 0. 04 1 0. 05 19 ) 0. 03 6 ( 0. 03 2 - 0 .0 40 8) 0. 03 7 ( 0. 03 2 - 0 .0 429 ) 0. 03 6 ( 0. 03 3 0 .03 90 ) e S D 2 0.0 49 (0 .0 49 - 0 .0 48 3) 0. 04 7 ( 0. 04 7 - 0 .0 46 7) 0. 04 7 ( 0. 04 7 - 0 .0 46 9) 0. 04 1 ( 0. 04 1 - 0 .0 40 8) 0. 04 5 ( 0. 04 5 - 0 .0 44 8) V 3 14 7 ( 82 - 2 ,0 69 ) 71 0 ( 15 1 t o -195 ) 92 ( 66 - 1 64 ) 27 1 ( 11 0 t o -49 0) 11 1 ( 82 - 1 72 ) os pi ta liza tio ns fo r ca rdi ov as cu la r di se as e e H D 0. 07 9 ( 0. 07 3 - 0 .0 86 ) 0. 08 4 ( 0. 07 7 - 0 .0 90 ) 0. 07 8 ( 0. 07 2 - 0 .0 84 ) 0. 07 0 ( 0. 06 4 - 0 .0 75 ) 0. 07 0 ( 0. 06 7 - 0 .0 74 ) e S D 0.0 92 (0 .0 92 - 0 .0 92 ) 0. 09 3 ( 0. 09 3 0 .0 93 ) 0. 08 6 ( 0. 08 6 - 0 .0 86 ) 0. 08 2 ( 0. 08 2 - 0 .0 82 ) 0. 08 2 ( 0. 08 2 - 0 .0 82 ) V 78 ( 52 - 1 81 ) 10 8 ( 63 - 3 60 ) 11 6 ( 68 - 5 82 ) 81 ( 55 - 1 53 ) 87 ( 68 - 1 36 )
4
Ap pe nd ix 4 (c on tin ued ). N umb er o f i nfl ue nz a v ac ci na tio ns , h os pi ta liz at io n r at es , a nd n umb er n ee de d t o v ac ci na te (N N V ) t o p re ve nt o ne a dd iti on al ho sp ita liz at io n f or V H A e nr ol le es v ac ci na te d d ur in g t he 2 01 0-11 t hr ou gh 2 01 4-15 r es pi ra to ry s ea so ns 20 10 -11 20 11 -1 2 20 12 -1 3 20 13 -14 20 14 -1 5 N % N % N % N % N % H os pi ta liza tio ns fo r ca rdi o-re sp ir at or y di se as e Rat e H D 0.1 21 (0 .11 4 0 .13 2) 0.1 29 (0 .1 22 - 0 .13 7) 0.1 16 (0 .10 8 0 .1 24 ) 0.1 07 (0 .0 99 - 0 .11 4) 0.1 08 (0 .10 3 0 .11 3) Rat e S D 0.1 41 (0 .14 1 0 .14 0) 0.1 40 (0 .14 0 0 .13 9) 0.1 33 (0 .13 3 0 .13 3) 0. 123 (0. 123 - 0. 123 ) 0.1 27 (0 .1 27 - 0 .1 26 ) NN V 51 ( 37 - 1 19 ) 89 ( 55 - 3 58 ) 58 ( 39 - 1 07 ) 62 ( 41 - 1 16 ) 53 ( 41 - 7 2) H ig h-do se i nfl ue nz a v ac ci ne r ec ip ie nt s. 2St an da rd -d os e i nfl ue nz a v ac ci ne r ec ip ie nt s. 3N umb er n ee de d t o v ac ci na te ( N N V ): t he n umb er o f p at ie nt s t ha t n ee d t o b e va cc in at ed w ith h ig h-do se i ns te ad o f s ta nd ar d-do se t riv al en t i nfl ue nz a v ac ci ne t o p re ve nt o ne a dd iti on al h os pi ta liz at io n. Ap pe nd ix 5 . M ea n c os t a nd r ei mb ur se m en t p er h os pi ta liz at io n a nd v ac ci na tio n, a nd e st im at io n o f r ea liz ed s av in gs t o U S t ax pa ye rs a m on g v ac ci na te d V H A en rol le es d ur in g t he 2 01 0-11 t hr ou gh 2 01 4-15 r es pi ra to ry s ea so ns 20 10 -11 20 11 -1 2 20 12 -1 3 20 13 -14 20 14 -1 5 Ho spit al iz at io ns fo r r es pi ra to ry d is ea se V H A co st 15 ,5 14 (15 ,3 10 - 15 ,7 18 ) 15 ,0 36 ( 14 ,8 41 - 1 5, 23 3) 15 ,7 76 (15 ,5 73 - 15 ,9 86 ) 16, 78 1 ( 16, 56 8 16, 99 0) 17, 74 7 ( 17, 51 2 17, 97 9) C M S re im bu rs em en t 9, 54 1 ( 9, 48 1 9, 60 1) 9, 62 4 ( 9, 56 1 9, 68 8) 9, 71 6 ( 9, 65 3 - 9 ,7 82 ) 9, 90 3 ( 9, 83 6 9, 97 0) 9, 82 6 ( 9, 75 8 9, 89 3) Va lu e 1 9, 94 9 ( 9, 80 7 - 1 0, 09 3) 10 ,2 51 (10 ,0 96 - 10 ,4 04 ) 10 ,2 77 (10 ,1 31 - 10 ,4 27 ) 11 ,010 (10 ,8 40 - 1 1,1 88 ) 11 ,0 35 ( 10 ,8 62 - 1 1, 21 0) Sa vi ngs p er p at ien t 2 62 ( -5 t o 1 16 ) 10 ( -6 0 t o 6 4) 10 2 ( 54 - 1 45 ) 26 ( -3 9 t o 8 4) 83 ( 47 - 1 17 ) H os pi ta liza tio ns fo r ca rdi ov as cu la r di se as e V H A co st 15 ,8 61 ( 15, 61 4 - 1 6, 10 7) 15 ,14 6 ( 14 ,9 16 - 1 5, 38 2) 16 ,2 27 ( 15, 97 5 - 1 6, 48 2) 17 ,0 03 ( 16 ,74 9 - 1 7,2 61 ) 18 ,17 4 ( 17 ,8 87 - 1 8, 465 ) C M S re im bu rs em en t 10 ,16 3 ( 10 ,0 77 - 10 ,2 49 ) 10 ,1 21 (10 ,0 35 - 10 ,2 08 ) 10 ,3 28 (10 ,2 39 - 10 ,4 20 ) 10 ,4 86 (10 ,3 93 - 10 ,5 77 ) 10 ,5 45 (10 ,4 51 - 10 ,6 41 ) Va lu e 11 ,9 65 ( 11 ,8 28 - 1 2, 10 2) 11 ,7 92 ( 11 ,65 8 - 1 1, 929 ) 12,3 78 (1 2, 23 3 1 2, 52 7) 12 ,9 41 ( 12 ,7 87 - 1 3, 095 ) 13 ,4 27 (13 ,2 60 - 13 ,5 97 )
pe nd ix 5 (c on tin ued ). M ea n c os t a nd r ei mb ur se m en t p er h os pi ta liz at io n a nd v ac ci na tio n, a nd e st im at io n o f r ea liz ed s av in gs t o U S t ax pa ye rs a m on g v ac ci na te d A e nr ol le es d ur in g t he 2 01 0-11 t hr ou gh 2 01 4-15 r es pi ra to ry s ea so ns 20 10 -11 20 11 -1 2 20 12 -1 3 20 13 -14 20 14 -1 5 vi ngs p er p at ien t 14 8 ( 57 - 2 26 ) 10 5 ( 26 - 1 82 ) 98 ( 12 - 1 72 ) 14 4 ( 70 - 2 18 ) 13 7 ( 82 - 1 81 ) os pi ta liza tio ns fo r ca rdi o-re sp ir at or y di se as e H A co st 15 ,5 14 (15 ,3 10 - 15 ,7 18 ) 15 ,0 36 ( 14 ,8 41 - 1 5, 23 3) 15 ,7 76 (15 ,5 73 - 15 ,9 86 ) 16, 78 1 ( 16, 56 8 16, 99 0) 17, 74 7 ( 17, 51 2 17, 97 9) S re im bu rs em en t 9, 54 1 ( 9, 48 1 9, 60 1) 9, 62 4 ( 9, 56 1 9, 68 8) 9, 71 6 ( 9, 65 3 - 9 ,7 82 ) 9, 90 3 ( 9, 83 6 9, 97 0) 9, 82 6 ( 9, 75 8 9, 89 3) lu e 11 ,2 62 (11 ,16 0 11 ,3 63 ) 11 ,2 67 (11 ,16 4 11 ,3 72 ) 11 ,6 27 (11 ,5 19 - 11 ,7 38 ) 12 ,29 0 ( 12 ,17 2 - 1 2, 40 6) 12 ,5 66 (12 ,4 40 - 12 ,6 90 ) vi ngs p er p at ien t 21 6 ( 86 - 295 ) 12 2 ( 25 - 2 01 ) 19 2 ( 99 - 2 84 ) 18 2 ( 91 - 2 86 ) 22 2 ( 15 8 - 2 85 ) gh te d a ve ra ge o f t he V H A c os t a nd C M S r ei mb ur se m en t p er h os pi ta liz at io n. 2Sa vi ng s p er p at ie nt v ac ci na te d w ith h ig h-do se i ns te ad o f s ta nd ar d-do se t riv al en t en za v ac ci ne . pe nd ix 6 . S en sit iv ity a na ly sis o f e st im at ed s av in gs t o U S t ax pa ye rs a m on g v ac ci na te d V H A e nr ol le es f or w hi ch t he o bs er ve d l oc at io n o f t he h os pi ta liz at io n i s d a nd l ow er ed 2 5 p er ce nt ag e p oi nt s spit al iz at io n s ite Obs er ve d Sc en ar io 1 Ch an ge Sc en ar io 2 Ch an ge HA 35 % 60% +2 5% 10 % -2 5% n-V H A 65 % 40% -2 5% 90% +2 5% 95 % C I 95 % C I 95 % C I in gs M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r r p at ie nt S D ➝ HD 13 8 66 20 0 15 6 76 22 6 11 9 56 17 5 ta l 22 M 11 M 32 M 25M 12 M 36M 19 M 9M 28M te nt ia l 50M 24 M 73M 57 M 28M 82 M 43M 21 M 64M
4
Ap pe nd ix 7 . S en sit iv ity a na ly sis o f e st im at ed s av in gs t o U S t ax pa ye rs a m on g v ac ci na te d V H A e nr ol le es f or w hi ch t he i nc re m en ta l c os t o f v ac ci na tin g w ith H D i s dou bl ed w ith out a nd w ith a 5 p er ce nt ag e p oi nt i nc re as e t o t he c os t o f V H A h os pi ta liz at io n a nd C M S r ei mb ur se m en t C ost Obs er ve d Sc en ar io 1 Ch an ge Sc en ar io 2 Ch an ge V H A h os pi ta liz at io n 1 16 ,5 23 16 ,5 23 0% 17, 34 9 +5% C M S re im bu rs em en t 2 10 ,3 20 10 ,3 20 0% 10 ,8 36 +5% Va cc in at io n ( inc re m en ta l) 3 11 .2 7 22 .5 4 +1 00 % 22 .5 4 +1 00 % 95 % C I 95 % C I 95 % C I Sav in gs M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r Pe r p at ie nt S D ➝ HD 202 11 5 28 0 12 6 53 191 13 4 57 201 To ta l 32 M 18 M 44M 20M 8M 30M 21 M 9M 32 M Po te nt ia l 73M 42 M 10 2M 46M 19 M 69M 49M 21 M 73M Av er ag e c os t o f o ne V H A h os pi ta liz at io n f or c ar di ov as cu la r d ise as e. 2Av er ag e C M S r ei mb ur se m en t o f o ne h os pi ta liz at io n f or c ar di ov as cu la r d ise as e. 3Av er ag e in cr em en ta l ( ad di tio na l) c os t o f v ac ci na tin g w ith h ig h-do se i ns te ad o f t riv al en t s ta nd ar d-do se i nfl ue nz a v ac ci ne a t a V H A f ac ili ty .
pe nd ix 8 . S en sit iv ity a na ly sis o f e st im at ed s av in gs t o U S t ax pa ye rs a m on g v ac ci na te d V H A e nr ol le es f or w hi ch t he o bs er ve d l oc at io n o f t he h os pi ta liz at io n ai se d a nd l ow er ed 2 5 p er ce nt ag e p oi nt s w ith a 1 0 p er ce nt ag e p oi nt r ed uc tio n i n t he c os t o f V H A h os pi ta liz at io n a nd a 1 0 p er ce nt ag e p oi nt i nc re as e i n C M S bu rs em en t spit al iz at io n s ite Obs er ve d Sc en ar io 2 Ch an ge Sc en ar io 3 Ch an ge HA 35 % 60% +2 5% 10 % -2 5% n-V H A 65 % 40% -2 5% 90% +2 5% ost Sc en ar io 1 Ch an ge H A h os pi ta liz at io n 1 14 ,8 71 -1 0% 14 ,8 71 -1 0% 14 ,8 71 -1 0% S re im bu rs em en t 2 11 ,3 52 +1 0% 11 ,3 52 +1 0% 11 ,3 52 +1 0% 95 % C I 95 % C I 95 % C I in gs M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r r p at ie nt S D ➝ HD 13 9 67 202 14 9 72 216 12 8 61 18 8 ta l 22 M 11 M 32 M 24 M 11 M 34 M 20M 10 M 30M te nt ia l 50M 24 M 74 M 54 M 26M 79M 47 M 22 M 68M er ag e c os t o f o ne V H A h os pi ta liz at io n f or c ar di ov as cu la r d ise as e. 2Av er ag e C M S r ei mb ur se m en t o f o ne h os pi ta liz at io n f or c ar di ov as cu la r d ise as e. T he er ic an H os pi ta liz at io n A ss oc ia tio n r ep or te d a n 1 1% u nd er pa ym en t b y M ed ic ar e i n 2 01 4 [ 1] . U sin g t hi s r ep or t a s a n a ve ra ge f or ou r s tu dy w hi ch r an f ro m /1 1 t o 2 01 5/ 16 , w e i nc re as ed t he M ed ic ar e r ei mb ur se m en t w ith 1 0% , w hi le d ec re as in g t he V H A c os t w ith 1 0% .
4
Ap pe nd ix 9 . S en sit iv ity a na ly sis o f e st im at ed s av in gs t o U S t ax pa ye rs a m on g v ac ci na te d V H A e nr ol le es f or w hi ch t he o bs er ve d l oc at io n o f t he h os pi ta liz at io n is r ai se d a nd l ow er ed 2 5 p er ce nt ag e p oi nt s w ith a 2 0 p er ce nt ag e p oi nt r ed uc tio n i n t he c os t o f V H A h os pi ta liz at io n a nd a 2 0 p er ce nt ag e p oi nt i nc re as e i n C M S re im bu rs em en t Ho spit al iz at io n s ite Obs er ve d Sc en ar io 2 Ch an ge Sc en ar io 3 Ch an ge V HA 35 % 60% +2 5% 10 % -2 5% no n-V H A 65 % 40% -2 5% 90% +2 5% C ost Sc en ar io 1 Ch an ge V H A h os pi ta liz at io n 1 13 ,21 9 -2 0% 13 ,21 9 -2 0% 13 ,21 9 -2 0% C M S re im bu rs em en t 2 12,3 84 +2 0% 12,3 84 +2 0% 12,3 84 +2 0% 95 % C I 95 % C I 95 % C I Sav in gs M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r M ea n ( U SD) Low er Up pe r Pe r p at ie nt S D ➝ HD 14 0 67 20 4 14 2 69 207 13 7 66 201 To ta l 22 M 11 M 32 M 23M 11 M 33 M 22 M 10 M 32 M Po te nt ia l 51 M 25M 74 M 52 M 25M 75 M 50M 24 M 73M Av er ag e c os t o f o ne V H A h os pi ta liz at io n f or c ar di ov as cu la r d ise as e. 2Av er ag e C M S r ei mb ur se m en t o f o ne h os pi ta liz at io n f or c ar di ov as cu la r d ise as e. W e d ou bl ed co st -a dj us tm en ts t o 2 0% t o r efl ec t p os sib le u nd er /o ve re st im at io n o f t he s en sit iv ity a na ly sis i n A pp en di x 8 .
Appendix 10
Table 1. Mean cost and reimbursement per hospitalization for cardiovascular disease among
vaccinated VHA enrollees during the 2010-11 through 2014-15 respiratory seasons in US dollars, outliers included
Hospitalization Mean (95% CI) Weight Value SD recipients
VHA cost 19,292 (18,805 - 19,895) 32%
13,151 (12,922 - 13,417)
CMS reimbursement 10,264 (10,157 - 10,372) 68%
Table 2. Estimation of realized and potential savings for hospitalizations among vaccinated VHA
enrollees for cardio-respiratory disease in US dollars), cost and reimbursement outliers included 95% CI
Savings Mean (USD) Lower Upper
Per patient SD➝HD 226 132 311
Total 36M 21M 49M
Potential 82M 48M 113M
Appendix 11. Number of influenza vaccinations, hospitalization rates per 10,000 person-seasons,
and number needed to vaccinate (NNV) to prevent one additional hospitalization for VHA enrollees vaccinated during respiratory seasons 2010-11 through 2014-15
N % Rate (95% CI)
Study cohort 3,638,924 100 Pneumonia or Influenza
HD recipients 158,636 4.4 Rate HD1 132 (122 – 143) SD recipients 3,480,288 95.6 Rate SD2 146 (146 – 147)
Observed hospitalizations in study cohort NNV3 683 (401 – 3,429)
Pneumonia or Influenza 53,013 1.5 Respiratory
Respiratory 164,948 4.5 Rate HD 387 (345 – 433) Acute Myocardial Infarction (AMI) 44,190 1.2 Rate SD 456 (458 – 454) Heart Failure 113,346 3.1 NNV 144 (89 – 473) Cardiovascular 314,014 8.6 Acute Myocardial Infarction (AMI)
Cardiovascular or respiratory 478,962 13.2 Rate HD 97 (87 – 106)
Applied relative vaccine
effectiveness rVE (95% CI) Rate SD 123 (122 – 132)
Pneumonia or Influenza 10% (2% - 17%) NNV 389 (280 – 630) Respiratory 15% (5% - 25%) Heart Failure
Acute Myocardial Infarction (AMI) 21% (13% - 29%) Rate HD 236 (215 – 260) Heart Failure 25% (17% - 32%) Rate SD 315 (314 – 316) Cardiovascular 14% (7% - 20%) NNV 127 (99 – 187) Cardiovascular or respiratory 14% (8% - 19%) Cardiovascular
1High-dose trivalent inactivated influenza vaccine (HD) 2Standard-dose trivalent inactivated influenza vaccine (SD)
3Number needed to vaccinate (NNV): the number of patients that need to be vaccinated with high-dose instead of standard-dose trivalent influenza vaccine to prevent one additional hospitalization.
Rate HD 749 (700 – 803) Rate SD 868 (870 – 866) NNV 84 (59 – 160) Cardiovascular or respiratory Rate HD 1,143 (1,079 – 1,213) Rate SD 1,324 (1,327 – 1,321) NNV 55 (40 – 93)
Appendix 12. Basis for and Validation of the Instrument
Because we used the same study population, years of analysis and instrumental as reported by Young-Xu et al.[2], we hereby provide a summary of Appendix A to that study.
Instrument: The instrument used in this study was the proportion of HD recipients at
each of the 1,347 VHA facilities during each influenza season, defined as the number of HD recipients divided by the total number of both HD and SD recipients in that season at each VHA facility.
Basis: We were not aware of any single, central decision-making or oversight agent
responsible for determining the quantity of HD to order or dispense at each VHA medical center or clinic during the study period. We conducted a survey and found that the primary decision-makers for influenza vaccine orders at VHA facilities were pharmacy leads and/or a multidisciplinary committee consisting of representatives from various specialties, most commonly Infection Control and Infectious Diseases departments. For some facilities (28%), pharmacy was the sole decision-maker for determining the amount of HD to be ordered. We, nevertheless, experimented with using states and Veteran Integrated Service Networks (VISN; these are large VHA patient care networks, like regions) to derive our instrumental variable. Unfortunately, quality of care and complexity level are assessed at the facility- level in the VHA, and we used these measures to perform exclusion restriction tests. Adding to this the independence of the numerous facilities in making vaccine purchasing decisions provided greater variations in terms of proportion of HD recipients; thus, we found that facility-level to be the most appropriate for our instrumental variable.
Validation: HD proportion ranged from 0% to 99% across VHA facilities. To be
considered valid, an instrument must meet two requirements. First, the instrument must correlate with (or predict) the explanatory variable of interest – in this case, provision of HD. Second, the instrument must not affect outcomes except through its effect on the explanatory variable of interest; this is known as the exclusion restriction. In other words, our instrument must not be correlated with other factors that would affect the individual-level outcomes of patients, such as facility quality or the overall
health status of patients treated by the facility. To test whether our instrument meets these two criteria, we performed three sets of instrument validity tests.
Test 1: Correlation
To test whether our instrument met the first requirement, we evaluated the F statistics of the first stage equations. The F statistic measures whether the instrument was sufficiently correlated with the endogenous variable (in this case, provision of HD). An F statistic greater than 10 is generally considered sufficient [3].The F statistic for our sample was greater than 1,000, which easily satisfied this condition. We also performed a logistic regression with provision of HD as the outcome and HD proportion as the independent variable, in unit of 10%. We found that, for every increment of 10% of HD adoption, the odds ratio was 2 (2.06, 95% CI, 2.05-2.07, p<0.0001). In other words, every increment of 10% in a facility’s HD proportion was associated with doubling the likelihood of a patient being provided with HD. Based on these tests, we concluded that the correlation requirement was satisfied.
Test 2: Exclusion Restriction
We compared patient characteristics across facilities to examine whether facilities with high propensity to use HD serve healthier patients, on average. To do this, we measured characteristics of each patient in the study. Next, we ranked facilities based on their propensity to provide HD in the previous season and compared facilities above and below the median (below 50th versus above 50th percentile), with those below
considered “low-HD” and those above considered “high-HD” facilities. Although the median proportion of facility HD, a binary variable, was used for this Exclusion Restriction test, our instrumental variable remained as the proportion of HD recipients, a continuous variable. The distributions of measures of patient risk were consistent across the two groups for all influenza seasons (Table 12A). The balance in the distribution of the measured risk factors across facilities provides reasonable evidence to infer that unmeasurable risk factors are also likely balanced across facilities. This inference lends support to facility-level HD influenza vaccine coverage being a valid IV.
Test 3: Exclusion Restriction
We then tested the correlation of the instrument with facility quality. We used VHA hospital “report card” data as a direct measure of hospital quality. We measured the correlation between measures on the report card and the study instrument. Again,
we found the distribution of facility quality to be similar between the “low-HD” and “high-HD” groups, lending further support that facility-level HD influenza vaccine coverage serves as a valid IV (Table 12A).
Statistical Modeling
We used the Stata ivpoisson command to conduct our IV analysis because we modeled count outcomes (i.e., number of hospitalizations) [4]. This Stata command “estimates the parameters of a Poisson regression model in which some of the regressors are endogenous. The model is also known as an exponential conditional mean model in which some of the regressors are endogenous.” The generalized method of moments (GMM) estimators are implemented in this command. In our study, we modeled the number of hospitalizations, e.g., influenza/pneumonia-associated hospitalizations. Exogenous regressors included the baseline demographic and clinical characteristics (Appendix 1). We suspect that the endogenous regressor, whether a patient receives a HD or SD vaccine (flu_vaccine), is correlated with unobserved factors that affect the number of influenza/pneumonia hospitalizations. The proportion of HD recipients at each facility (HD_proportion) was observed and used as an instrument for HD versus SD. In our statistical modeling, we did not explore potential effect modification, except in the case of seasonal variability where we included an interaction term between HD and seasons.
Table 12A. Exclusion restriction tests
Below Median (%) Median+ (%) SMD*
Patient Characteristics
Male 98 98 0
Married 61 60 2
White 76 75 2
Admitted to a Nursing Home 2 1 8
Vaccinated in the Previous Season 80 77 7
Patient Conditions
Chronic cardiac disease 28 27 2
Chronic pulmonary 13 13 0
Neurological/musculoskeletal 5 5 0
Other metabolic and immunity disorders 1 1 0
Diabetes mellitus 34 33 2
Liver diseases 1 1 0
Malignancies 14 14 0
Immunosuppressive disorders 4 5 -5
Chronic renal disease 6 6 0
Hemoglobinopathies 0.3 0.3 0
At least 3 risk factors 9 9 0
Facility-Level Comparison
Complexity (1 most complex, 3 least) P-value
1 73 80
2 15 14
3 12 7 0.213
Quality of Care (1 lowest, 5 highest)
1 5 2 2 16 16 3 41 38 4 24 32 5 14 11 0.311 Region Midwest 30 32 Northeast 20 19 South 32 31 West 17 18 0.96 Rurality Rural 27 29 Urban 73 71 0.541
*A standardized mean difference (SMD) less than 10 in absolute value suggests no important difference between the two cohorts [5]
REFERENCES
1. Association AH. Underpayment by Medicare and Medicaid fact sheet. January 2016 [6/12/2019]. Available from: https://www.aha.org/2016-jan-underpayment-medicare-medicaid.
2. Young-Xu Y, Snider JT, van Aalst R, Mahmud SM, Thommes EW, Lee JKH, Greenberg DP, Chit A. Analysis of relative effectiveness of high-dose versus standard-dose influenza vaccines using an instrumental variable method. Vaccine. 2019;37(11):1484-90.
3. Glymour MM, Tchetgen Tchetgen EJ, Robins JM. Credible Mendelian Randomization Studies: Approaches for Evaluating the Instrumental Variable Assumptions. American journal of epidemiolog y.
2012;175(4):332-9.
4. Stata.com. “ivpoisson – Poisson regression with endogenous regressors” [cited Stata Manual 13]. Available from: https://www.stata.com/manuals13/rivpoisson.pdf.
5. Austin PC, Stuart EA. Moving towards best practice when using inverse probability of treatment weighting (IPTW) using the propensity score to estimate causal treatment effects in observational studies. Stat Med. 2015;34(28):3661-79.
