Visual Abstract
Abstract
Background and objectives High-dose influenza vaccine, which contains fourfold more antigen than standard dose, is associated with fewer cases of influenza and less influenza-related morbidity in the elderly general population. Whether the high-dose influenza vaccine benefits patients on dialysis, whose immune response to vaccination is less robust than that of healthy patients, is uncertain.
Design, setting, participants, & measurements We compared hospitalizations and deaths during the 2015–2016 and 2016–2017 influenza seasons by vaccine type (standard trivalent, standard quadrivalent, and high-dose trivalent influenza vaccine) administered within a national dialysis organization. The association of vaccine type with outcomes was estimated using Cox proportional hazards regression with adjustment for patient factors and “center effect.” Analyses were stratified by age and dialysis modality.
Results Between September 1 and December 31, 2015, standard dose trivalent, standard dose quadrivalent, and high-dose trivalent influenza vaccines were administered to 3057 (31%), 5981 (61%), and 805 (8%) patients, respectively. The adjusted rates of first hospitalizations by vaccine type during the influenza season were 8.43, 7.88, and 7.99 per 100 patient-months, respectively, and the adjusted rates of death were 1.00, 0.97, and 1.04, respectively. These differences were not significant. In 2016, 3614 (39%) received quadrivalent vaccine, and 5700 (61%) received high-dose trivalent vaccine. The adjusted rates of first hospitalization by vaccine type were 8.71 and 8.04 per 100 patient-months, respectively, and the adjusted rates of death were 0.98 and 1.02, respectively. Receipt of high dose was associated with a significant reduction in hospitalization (hazard ratio, 0.93; 95% confidence interval, 0.86 to 1.00; P=0.04); there was no significant association with death. There was no significant heterogeneity of either association by age group or dialysis modality.
Conclusions Receipt of high-dose compared with standard dose influenza vaccine in 2016–2017 was associated with lower rates of hospitalization in patients on dialysis, although that was not seen in 2015–2016.
- infection
- dialysis
- Aged
- Humans
- Influenza Vaccines
- Influenza, Human
- Seasons
- renal dialysis
- Vaccination
- Immunization
- hospitalization
Introduction
Influenza is a major contributor to hospitalization and death in elderly patients, including among patients with chronic conditions, such as kidney failure (1–3). Rates of vaccination against influenza have increased among US Medicare beneficiaries treated with dialysis, rising from 52% in 2006 to 71% in 2015 (4). The effectiveness of influenza vaccination in the dialysis population remains uncertain. Serologic studies show a substandard immune response compared with the age-matched general population (5–15). For example, the seroprotection rate (defined as achievement of >1:40 hemagglutinin titers) measured 3 weeks after receipt of standard dose trivalent vaccine among ≥60 year olds was 47.3% in patients on dialysis versus 81.3% in patients not on dialysis, and among those <60 years of age, 67.1% versus 92.4%, respectively (16). A population-based analysis of the 2013 influenza season found no significant reduction in morbidity among dialysis patients after receipt of the standard dose trivalent influenza vaccination (17). Strategies for enhancing the immune response are needed.
A high-dose trivalent influenza vaccine, which contains four times the antigen dose (60 μg) per influenza strain as the standard dose, was approved for use in the United States in 2009 (18). In 2014, a randomized clinical trial of >31,000 adults age 65 years old or older from the general population found a 24.2% reduction in laboratory-confirmed influenza cases with high dose compared with the standard dose trivalent vaccine (19). Consequently, the Centers for Disease Control and Prevention (CDC) recommends consideration of use of the high-dose vaccine in older adults (20). Whether high-dose vaccine benefits patients on dialysis, including those below age 65 years old, given their impaired serologic responses to vaccine has not been assessed.
In this study, we assessed hospitalization and mortality rates in the 2015–2016 and 2016–2017 influenza seasons by vaccine type (standard dose trivalent, standard dose quadrivalent, and high-dose trivalent) in a not-for-profit United States dialysis organization.
Materials and Methods
Study Population
Separate analyses of the 2015–2016 and 2016–2017 influenza seasons were conducted across Dialysis Clinic Inc. (DCI), a not-for-profit dialysis provider with approximately 230 clinics in the United States. The study population for each season included all patients receiving maintenance in-center hemodialysis (HD) or peritoneal dialysis (PD) who had received an influenza vaccine at their dialysis facility between August 1 and December 31 of the respective year. Patients who were not vaccinated at a DCI clinic were excluded given uncertainty as to whether they were vaccinated elsewhere and if vaccinated, the vaccine type administered. This study was conducted in adherence with the Declaration of Helsinki. It was reviewed by the Tufts University Institutional Review Board and considered exempt due to the use of deidentified data.
Outcomes
The primary outcomes included all-cause hospitalization occurring at least 14 days after vaccination and all-cause death occurring at least 14 days after vaccination, with patients censored on June 30 of the respective influenza season. Hospitalizations are entered by clinic staff into “Darwin,” DCI’s proprietary electronic medical information system. Facility staff reconcile missed dialysis sessions with a reason for missed treatment on a monthly basis; accordingly, the rate of capture of hospitalizations lasting >24 hours in Darwin is high. The reason for hospitalization is not reliably entered at all facilities, and thus, cause-specific hospitalization was not examined.
Covariates
The patient-level demographic and clinical factors were obtained from Darwin. In the case of the laboratory values, those closest to and within 30 days of the date of administration of vaccine were used. To account for differences in facility-level practices that could affect outcomes, facility was a random intercept (each facility had its own intercept) in the models.
Influenza Vaccines
The primary exposure of interest was the type of inactivated influenza vaccine administered; these included (1) standard dose trivalent (15 μg each of the H1N1, H3N2, and B influenza strains anticipated that year), (2) high-dose trivalent (60 μg each of the three antigens in the standard dose trivalent vaccine), or (3) standard dose quadrivalent (15 μg of an additional B strain over the trivalent vaccine). The vaccines administered to nearly all patients were manufactured by Sanofi Pasteur, except for 10% in 2015 and 5% in 2016 who received the quadrivalent vaccine from GlaxoSmithKline. The decision about vaccine to be administered was made at the unit level by the medical director.
Statistical Analyses
Primary analyses evaluated the association between vaccine type and time to all-cause hospitalization or death using Cox proportional hazards regression. Time at risk started from day 15 after the date of vaccination until hospitalization, death, a censoring event (transplant, transfer out of a DCI facility, loss to follow-up, or renal recovery), or June 30 of the respective year of analysis, whichever came first. Multivariable models adjusted for the baseline characteristics are listed in Table 1. There were no missing data for all variables except serum albumin and race. Missing albumin was imputed using the average albumin in the full sample, and race (a categorical variable) was set to nonblack, which was more common than black among those who had reported race. Models were also derived after excluding patients with missing data (“complete case analysis”). In a secondary analysis of the outcome of total hospitalization count, we used negative binomial regression models for each with the logarithm of the total time at risk for each patient as an offset parameter. In all models, we adjusted for correlations of outcomes within dialysis facility using a normally distributed frailty for the Cox proportional hazards and a random effect intercept for the negative binomial models. We tested for effect modification by age (<65 and ≥65 years of age) and dialysis modality (PD versus HD). To obtain estimates of the absolute effects of the intervention on first hospitalization and death, event rates expressed per 100 patient-months were derived using Poisson regression with the logarithm of time at risk used as an offset parameter.
Baseline characteristics of patients on dialysis administered influenza vaccine within Dialysis Clinic Inc. facilities in 2015–2016 and 2016–2017
Results
Population Derivation
Between September 1 and December 31 of 2015 and 2016, 10,018/13,852 (73%) and 9733/13,819 (70%) patients, respectively, dialyzing at a DCI unit were vaccinated against influenza at their dialysis facility. Per the flow diagrams (Figure 1), an additional 1729 (12%) in 2015 and 1767 (13%) in 2016 were offered but declined the vaccine, 0.5% in each year had a medical contraindication, and the remainder reported that they had been vaccinated elsewhere. A small number of patients, largely from one center, received more than one dose of vaccine (181 [1%] in 2015 and 209 [1%] in 2016). These patients were excluded, because the sample size was too small to assess effects of a booster. Other exclusions were <14 days of follow-up after vaccination, which applied to 84 (0.8%) in 2015 and 83 (0.9%) in 2016. The comparison in 2016–2017 is between high-dose trivalent and quadrivalent, because <1% (n=127) of patients received the standard trivalent vaccine in 2016–2017. After exclusions, the 2015–2016 study population consisted of 9843 patients, of whom 61% received standard dose trivalent vaccine, 31% received quadrivalent vaccine, and 8% received high-dose trivalent vaccine; the 2016–2017 study population consisted of 9314 patients, of whom 61% received quadrivalent vaccine and 39% received high-dose trivalent vaccine. The proportions of subjects who left the study cohort before the end of follow-up were the same by cause in 2015–2016 and 2016–2017, and were as follows: death (10%), kidney transplant (2%), renal recovery (0.4%), loss to follow-up (0.1%), and transfer to a nonparticipating unit (4%). The median (interquartile range [IQR]) follow-up times of the 2015–2016 and 2016–2017 study populations were 253 (IQR, 239–262) and 246 (IQR, 228–254) days, respectively.
Flow diagram of the derivation of the study populations. DCI, Dialysis Clinic Inc.; N/A, not applicable.
Baseline Characteristics
Baseline characteristics and differences by vaccine type administered are shown in Table 1. Patients receiving high dose trivalent vaccine were more likely to be ≥65 years old. In 2015, the proportion of patients receiving standard trivalent, quadrivalent and high dose trivalent who were ≥65 years of age was 42.8%, 40.6%, and 63.5%, respectively. In 2016, 32.6% of patients vaccinated with the standard dose quadrivalent were ≥65 years of age versus 49.8% of those receiving high dose trivalent vaccine.
Receipt of high dose was also associated with white race in both years and care provided within an End-Stage Renal Disease Seamless Care Organization in 2015. The number of patients who had been hospitalized in the 2 months before vaccination and serum albumin did not differ by vaccine type in either cohort year. Use of the high-dose vaccine varied by region (West, South, Northwest, and Northeast), ranging from 42% to 68% among the <65 year olds and from 61% to 84% among the ≥65 year olds. In both age groups, use tended to be less in the South versus other regions (Supplemental Table 1). Regional variation was not examined in 2015–2016, because a minority received high dose that year. There were small differences in the distribution of cause of ESKD, dialysis vintage, and vascular access by vaccine type that were adjusted for in the analyses.
2015–2016 Influenza Season.
All-cause hospitalization and mortality rates through June 30, 2016 were 8.15 and 1.32 events per 100 patient-months, respectively. In unadjusted analyses, there were no differences in the time to first hospitalization by vaccine type among patients <65 (Figure 2A) and ≥65 years old (Figure 2B). In adjusted models of the full cohort (Table 2) and age-stratified analyses (Supplemental Table 2), there were no statistically significant differences in time to first hospitalization, death, or the time to a composite of first hospitalization or death. The hazard ratios (HRs) and 95% confidence intervals (95% CIs) for first hospitalization in the ≥65-year-old age group with high-dose trivalent (HR, 0.86; 95% CI, 0.73 to 1.03) and standard dose quadrivalent vaccine (HR, 0.96; 95% CI, 0.87 to 1.07) compared with standard dose trivalent vaccine differed in magnitude from the corresponding estimates in the <65-year-old group (high dose: HR, 1.09; 95% CI, 0.87 to 1.38 and quadrivalent: HR, 0.94; 95% CI, 0.84 to 1.05), although an interaction of age with vaccine type was not significant (P=0.46). There were no differences in the outcome of the total count of hospitalizations by vaccine type either for the full cohort or within age subgroups (Supplemental Table 3). Results were not different for patients treated with PD versus HD (data not shown).
Kaplan–Meier survival estimates of the associations of vaccine type with hospitalization stratified by age. Receipt of High Dose as Compared with Standard Dose Influenza Vaccine Associated with Reduced Hospitalization in 2016-17, not in 2015-16. In 2015–2016, there was no significant association of vaccine type (standard dose trivalent, standard dose quadrivalent, and high-dose trivalent) with time to first hospitalization among patients <65 years of age (A) or ≥65 years of age (B). In 2016–2017, there was a significant association of high dose with reduced hospitalization in the full population (Table 3). When stratified by age, the association was significant in patients ≥65 years of age, but not in those <65 years of age. However, an interaction term of age group with age did not reach statistical significance (details are in the text).
Events by vaccine type in years 2015–2016 in the full study population
2016–2017 Influenza Season.
All-cause hospitalization and mortality rates from September 1, 2016 to June 30, 2017 were similar to the 2015–2016 influenza season at 8.40 and 1.31 events per 100 patient-months, respectively. In unadjusted analyses, there was no difference in time to first hospitalization by vaccine type in the <65-year-old age group (Figure 2C); however, there was a significant reduction in the time to first hospitalization among those 65 years old or older (Figure 2D) associated with receipt of the high-dose trivalent vaccine as compared with the standard dose quadrivalent vaccine. In the adjusted models of the full cohort, receipt of high-dose versus quadrivalent vaccine was associated with a reduction in all-cause hospitalization (HR, 0.93; 95% CI, 0.86 to 1.00) (Table 3). When stratified by age (Table 4), the effect seemed to be larger in patients ≥65 versus <65 years of age (HR, 0.88; 95% CI, 0.79 to 0.97 versus HR, 0.95; 95% CI, 0.86 to 1.05, respectively), although an interaction of age with vaccine type was not significant (P=0.50). There was no difference in death rate by vaccine type, and there were no differences in the outcome of the total count of hospitalizations by vaccine type either for the full cohort or within age subgroups (Supplemental Table 4). There was also no difference in the association of vaccine type among patients treated with PD versus HD (data not shown). Results were similar when models were estimated using “complete cases” (Supplemental Tables 5–8).
Events by vaccine type in years 2016–2017 in the full population
Events by vaccine type stratified by age during the 2016–2017 influenza season
Discussion
Receipt of high-dose trivalent influenza vaccine was associated with a lower risk of hospitalization compared with standard dose quadrivalent vaccine in patients on dialysis in the 2016–2017 influenza season. Vaccine type had no association with outcomes in 2015–2016. The inconsistency in results across years may be due to limited statistical power in 2015–2016, where only 8% received high dose, or differences in strain virulence or vaccine effectiveness. Also, it is possible that the 2016–2017 results are spurious. The CDC reports similar vaccine effectiveness in successive years (47% versus 40%, respectively); virulence has not, as of yet, been reported for 2016–2017 (21). We had postulated that the effect of the high-dose vaccine would be greater in older individuals in light of a number of studies that have shown a decline in the seroresponse to the influenza vaccine with age (7,13,15,22). Although the magnitude of reduced risk of high-dose versus standard dose vaccine with hospitalization was larger in the ≥65 versus <65-year-old age groups in both influenza seasons, an interaction of age with vaccine type was not statistically significant. If the association found in the 2016–2017 influenza season is causal, use of the high-dose vaccine compared with standard dose quadrivalent vaccine would be associated with 5.3 fewer hospitalizations per 100 patients per influenza season (on the basis of adjusted rate by vaccine type of 8.71 versus 8.04 hospitalizations per 100 patients-months per Table 3). Centers for Medicare Services reimbursements per dose administered in 2017–2018 (which are established to cover vaccine costs for most providers) were $49, $20, and $19 for high-dose trivalent, standard dose quadrivalent, and standard dose trivalent vaccines, respectively (23).
The magnitude of the association on hospitalizations of the high-dose vaccine compared with the standard dose vaccine is modest, although very consistent with estimates from high-quality clinical trials in the elderly general population. In a cluster randomized pragmatic trial of >38,000 nursing home residents, high-dose versus standard dose trivalent vaccine was associated with a 13% relative risk reduction in all-cause hospitalization (HR, 0.87; 95% CI, 0.78 to 0.98) in the 2013-2014 influenza season (25). A second randomized clinical trial, enrolling 31,989 participants from 126 sites in the United States and Canada, showed a 7% (HR, 0.93; 95% CI, 0.87 to 1.00) lower risk of all-cause hospitalization with high-dose versus standard dose vaccine in the same flu season (19,24), whereas a retrospective analysis of Medicare beneficiaries found a 21.9% (95% CI, 15.0 to 28.7) reduction in influenza related admissions and emergency department visits among ≥65 year olds receiving high versus standard dose vaccine in the 2012–2013 influenza season (26). The pooled odds ratio for influenza-related hospitalization with receipt of the high-dose vaccine compared with standard dose trivalent vaccine in a recent meta-analysis of seven randomized and nonrandomized trials was 0.82 (95% CI, 0.74 to 0.92). The pooled odds ratio for pneumonia-related hospitalization was 0.76 (95% CI, 0.67 to 0.86), and the pooled odds ratio for all-cause hospitalization was 0.91 (95% CI, 0.85 to 0.98) (27). Cause-specific hospitalizations could not be discerned in our study, although in each of the above-mentioned studies, the magnitude of the effect on influenza-specific events was greater than on all-cause hospitalizations (28); accordingly, the certainty about a causal association of high-dose influenza vaccine with influenza cases and influenza-related morbidity in dialysis patients may be greater than is estimated in this study.
Other strategies to enhance the immune response that have been tested in patients on dialysis include the use of adjuvants and boosters. Adjuvanted vaccines contain a high purified protein (e.g., squalene or tocopherol) that stimulates the immune response, and they have an advantage over the high-dose vaccine of being able to be produced in large quantities relatively quickly, because they do not require as much influenza antigen. Studies in the dialysis population have shown a greater seroresponse to an adjuvanted influenza vaccine compared with the standard dose vaccine (9,22,29,30), although no study has assessed effects on clinical outcomes (22). A comparison of adjuvanted and high-dose vaccines has also not been done in the dialysis population or the general population. Both are needed.
The value of booster vaccination against influenza in dialysis patients is also unclear; the limited number of trials that have been done have found little to no effect (31–34). In a meta-analysis of six studies involving 497 patients on HD receiving a booster (usually 4 weeks later, with range of 21–90 days) versus 368 receiving a single dose of the influenza vaccine (spanning different years across studies), seroprotection was achieved in 74% (at an undeclared time point) and only 2% higher after booster administration (31). A subsequent study with longer follow-up found no difference after receipt of a single dose of the standard trivalent vaccine compared with a booster dose administered 3 weeks later in seroprotection rates (for H1N1) at 6 weeks (58.3% versus 62.5%, respectively) or 18 weeks (25.0% versus 25.0%, respectively) (16). In a study of 51 patients treated with a double dose of standard dose influenza vaccine versus a single dose plus booster given at 5 weeks, seroprotetection rates for H3N2 were higher at 8 weeks in the group that had received the double dose (38% versus 20%, respectively) (32). These results suggest that it may be better to give a higher initial dose than a booster, although additional study is needed.
This is the first study testing clinical efficacy of high-dose vaccine compared with standard dose influenza vaccines in a large, nationally representative dialysis population. Although vaccine type was not randomized, vaccine type was allocated at the facility level rather than at the patient level, which should reduce the potential for bias. Furthermore, “center effect” was adjusted for via each center having its own random intercept in the models. The lack of cause-specific hospitalization and influenza events is a weakness, although if results of prior randomized, controlled trials extrapolate, effects on these outcomes would be greater than those found for all-cause hospitalization. We also did not measure the serologic response in this study, although clinical outcomes are more important. Comparing vaccine safety would require careful collection of serious adverse events in the days after receipt of administration of the vaccine and was not done in this study. Large clinical trials in the elderly population have noted very low rates of adverse events and no increase over the standard dose trivalent vaccine (19).
In summary, our study finds that administration of high-dose vaccine compared with standard dose influenza vaccine to patients on dialysis was associated with a reduction in all-cause hospitalization in the 2016–2017 influenza season but not in 2015–2016. This may be because a minority of patients (8%) received the high dose in 2015–2016, although we cannot discount the possibility that the 2016–2017 results are due to chance. A randomized clinical trial would be more definitive. Studies comparing other strategies for boosting the immune response, such as adjuvants and booster doses, are also needed.
Disclosures
None.
Footnotes
Published online ahead of print. Publication date available at www.cjasn.org.
This article contains supplemental material online at http://cjasn.asnjournals.org/lookup/suppl/doi:10.2215/CJN.03390318/-/DCSupplemental.
See related editorial, “Increasing Protection of Dialysis Patients against Influenza,” on pages 1624–1626.
- Received March 15, 2018.
- Accepted August 24, 2018.
- Copyright © 2018 by the American Society of Nephrology