Vaccine‐preventable disease hospitalized patients with heart failure with reduced ejection fraction

Abstract Background Over five million Americans suffer from heart failure (HF), and this is associated with multiple chronic comorbidities and recurrent decompensation. Currently, there is an increased incidence in vaccine‐preventable diseases (VPDs). We aim to investigate the impact of HF with reduced ejection fraction (HFrEF) in patients hospitalized with VPDs. Hypothesis Patient with HFrEF are at higher risk for VPDs and they carry a higher risk for in‐hospital complications. Methods Retrospective analysis from all hospital admissions from the 2016‐2018 National Inpatient Sample (NIS) using the ICD‐10CM codes for patients admitted with a primary diagnosis of VPDs with HFrEF and those without reduced ejection fraction. Outcomes evaluated were in‐hospital mortality, length of stay (LOS), healthcare utilization, frequency of admissions, and in‐hospital complications. Multivariate regression analysis was conducted to adjust for confounders. Results Out of 317 670 VPDs discharges, we identified 12 130 (3.8%) patients with HFrEF as a comorbidity. The most common admission diagnosis for VPDs was influenza virus (IV) infection (75.0% vs. 64.1%; p < .01), followed by pneumococcal pneumonia (PNA) (13% vs. 9.4%; p < .01). After adjusting for confounders, patients with HFrEF had higher odds of having diagnosis of IV (adjusted [aOR], 1.42; p < .01) and PNA (aOR, 1.27; p < .01). Patients with VPDs and HFrEF had significantly higher odds of mortality (aOR, 1.76; p < .01), LOS, respiratory failure requiring mechanical ventilation, and mechanical ventilation for less than 96 h. Conclusion Influenza and PNA were the most common VPDs admitted to the hospital in patients with a concomitant diagnosis of HFrEF. They were associated with increased mortality and in‐hospital complications.


| BACKGROUND
Vaccines are considered one of the greatest achievements in medicine and have helped decrease the incidence and even eradicate several transmissible diseases. Vaccination contributes substantially to global health by preventing morbidity and mortality. 1 Currently we are witnessing an increased incidence of VPDs in the United States which can be attributed to vaccine refusals, under-vaccination, waning immunity, and migration from countries outside the United States. 2,3 As an example the overall vaccination coverage among adults in racial minorities is lower when compared with non-Hispanic whites leading to racial and ethnic disparities. 4 In the United States, HF is common; over 5 million Americans have this diagnosis with an annual incidence of 800 000 cases. Heart failure (HF)-related hospitalizations are one of the most common reasons for admission in the United States, with more than one million hospitalizations annually for acute disease exacerbation. Patient outcomes remain poor, with an approximate 50% 5-year survival rate, making this a crucial public health issue. 5 HF has also been associated with inflammatory states with elevated serum proinflammatory cytokines. This suggests that diseases causing low-grade chronic inflammation may be important contributors to HF progression. 6 Thus infections can be an important cause of HF decompensation; these can be acquired in the community or during hospitalization, some leading to primary pulmonary infections. 7 Vaccination in HF varies in the United States, and this patient population is more susceptible to influenza-related complications like pneumonia, acute decompensation, and increased hospitalization complications. Data from a landmark clinical trial suggest that vaccination for influenza in HF was associated with a reduced risk of death. 8 The aim of this study is to identify the outcomes of patients hospitalized with VPDs and a concomitant diagnosis of HFrEF.

| Study patients
There is not a unique ICD-10-CM code for VPDs, but there are validated codes that may identify patients with these diseases. Only

| Study variables
Mortality, LOS, total hospitalization charges, and cost are provided within the NIS with each hospitalization. Potential confounders were gender, age, race, median yearly income per patient's zip code, CCI, hospital location (rural or urban), geographic region (Northeast, West, South, or Midwest), hospital teaching status, hospital beds, and individual patient comorbidities (OT, HIV, DM, or malignancy) that were identified with secondary ICD-10-CM.

| Outcomes
The primary outcomes were proportions and admission odds for VPDs in patients with and without HFrEF. Secondary outcomes were in-hospital mortality, healthcare utilization (total hospital charges and costs), length of stay provided by the NIS for each discharge, and inhospital complications (respiratory failure, mechanical ventilation for DEL CID FRATTI ET AL. | 475 less than 24 h, for 24−96 h, and >96 h, acute kidney injury, shock, and sepsis) that were extracted with specific ICD-10-CM/PCS codes. We performed a secondary analysis for the primary and secondary outcomes independently for race and gender. Due to the low incidence of HPV, MNC, TET, DIPHT, and WHC, a variable called Rare VPDs was created with these discharge codes.

| Statistical analysis
Analyses were performed using STATA version 16. The NIS is based on a complex sampling design that includes weighting, clustering, and stratification. STATA facilitates analysis to produce nationally representative unbiased results, variance estimates, and p values. We performed a multivariate logistic regression to adjust for confounding variables (diabetes mellitus, human immunodeficiency virus, history of organ transplant, malignancy, weekend admission, age, gender, race, income, CCI, hospital location, hospital region, teaching hospital, and hospital bed-size). We included an extensive multivariate logistic regression analysis to adjust for covariate imbalance, selection bias, and potential confounders which we deem noninferior to a propensity score with covariate adjustment based on prior studies comparing both methods. 11 Proportions were compared by using the Fischer exact test, and continuous variables were compared by using the Student's t test. All p values were two-sided with a .05 as the threshold for statistical significance. The study adheres to best methodological practices for the NIS analysis to minimize the NIS limitations and to provide reliable results. 12 (Figure 1 and Table 3).

| In-hospital mortality
The in-hospital mortality proportion for patients with VPDs and HFrEF was 2.6% and 0.9% without HFrEF, respectively. On univariable analysis and multivariable analysis, HFrEF was an independent predictor of overall in-hospital mortality compared with patients without HFrEF admitted with VPDs (adjusted OR, 1.76; p < .01) ( Table 4).  (Figure 1 and Table 4).

| VPD outcomes: Gender and race based analysis
After adjusting for confounders, there were no significant differences in VPD, complications, hospital LOS, total charges, and total costs between genders and race (these findings are included in the Supporting Information Appendices A and B).

| DISCUSSION
In this large national study with more than 317 540 VPDassociated hospitalizations, we found that in patients with HFrEF, IV and PNA admissions are more common. HF-related hospitalizations carry a great financial burden, can increase disease progression and death. Despite the decrease in the HF-related admissions from 2006 to 2014, the disease burden is still high. 14 With PNA being the second most common diagnosis in patients with HFrEF, preventing associated infections could lead to lower admissions for this group. 15 HF has been recognized as a risk factor for worse outcomes in patients with seasonal influenza infection, with a high rate of acute decompensation, cardiovascular morbidity, and all-cause mortality. 16 A subgroup analysis from PARADIGM-HF trial demonstrated that vaccination for influenza in patients with HFrEF was associated with a reduced risk of death, although this association is not well understood. 17 A recent self-controlled case series with a regression analysis done in Europe showed that in HF patients, influenza vaccination is associated with reduced risk of cardiovascular and all-cause hospitalizations. 18 In comparison, in our study by using national cohort data, we identified that patients admitted with VPDs and HFrEF had higher odds IV when compared with patients without HFrEF. VPD admissions and a concomitant diagnosis of HFrEF also carry a worse prognosis in terms of morbidity and mortality. Data from IAMI randomized trial demonstrated that influenza vaccination after acute myocardial infarction or high-risk stable coronary artery disease patients decreased the risk of all-cause mortality, cardiovascular mortality, myocardial infarction, and stent thrombosis at 12 months when compared with placebo that demonstrated the protective effect of influenza vaccination in high cardiovascular risk patients. 19 Patients with HFrEF also had more odds of being admitted for PNA when compared with those without HFrEF. Prior data suggest that mortality is higher in this high-risk cohort group. 20

| LIMITATIONS
Our study is limited by being a retrospective analysis of the NIS, that relies on ICD-10 codes for the population of interest, comorbidities/ outcomes extraction; thus there could be variable misdiagnosis based on coding mistakes. It was also not possible to verify the vaccination status of the patient, or the type, duration, and severity of HFrEF.
Patients with IV and PNA are more likely to be admitted to the hospital, and the rest of VPDs are usually managed as outpatient, and the NIS captures only inpatient records. This is an observational and retrospective study; there are potentials for selection bias and unmeasured confounders. In this study, we adhere to required practices for the NIS and performed an extensive multivariate regression analysis to try to mitigate these risks and deliver reliable results. We believe that this study reinforces the importance of vaccination, and the burden of HFrEF in this population.

| CONCLUSION
This retrospective study demonstrates that from VPDs that require

CONFLICTS OF INTEREST
The authors declare no conflicts of interest.

DATA AVAILABILITY STATEMENT
Data are available upon request.