Study design and patient selection
We collected data from medical records in two large tertiary care centers in Thailand that are Ramathibodi hospital and Maharaj Nakorn Chiang Mai hospital between 2013-2021. We identified all patients diagnosed with new heart failure in internal medicine department clinic during January 1, 2013 to December 1, 2020 by International Classification of Diseases, Tenth Revision [ICD-10] codes I50 – heart failure, I500, I509, I42 – cardiomyopathy, I255 – ischemic cardiomyopathy. After we enrolled all patients with this ICD10 code, we reviewed all medical records to confirm the diagnosis of heart failure and collected the data. For missing data, we used telephone interviews to ascertain the data completion. Patients who age < 18 years of age were excluded.
Baseline characteristics
The following patient data were recorded on the case record form: age, gender, body weight, height, BMI, blood pressure, comorbidities such as hypertension, diabetes mellitus, dyslipidemia, chronic renal failure, etiology of heart failure, LVEF, medication such as ACEI, ARB, ARNI, beta blocker, MRA, statin, antiplatelet, laboratory test results, calculated MAGGIC heart failure scoreand influenza vaccine status. Medications use was defined as at least 1 filled prescription in the 6 months leading up to heart failure diagnosis or within 30 days following diagnosis.
Influenza vaccination status
We defined the following variables : (1) influenza vaccination exposure (yes/no), defined as whether a patient received at least 1 influenza vaccination within the follow-up period and thus after heart failure diagnosis, (2) the cumulative number of vaccinations during the follow-up periods, (3) year of vaccination
Outcomes
The primary endpoints of this study were composite endpoints of all-cause death or heart failure hospitalization. The secondary endpoints include all-cause death, cardiovascular death or heart failure hospitalization. HF hospitalization was defined as the presence of clinical signs or symptoms of HF, severe enough to require the use of intravenous furosemide in either a ward or emergency department for more than 24 hours. Cardiovascular death was defined as death due to acute coronary syndrome (ACS), heart failure or stroke. This information was retrieved from the medical records and was reviewed the correction of data by physicians. We had access to follow-up data on the outcome until June 30, 2021.
Sample Size Calculation
The sample size for the primary outcome was calculated assuming a 73.3% in ten years mortality in heart failure patients9 and influenza vaccine would reduce the incidence by at least 18%.7 Thus, considering a test of differences in proportion between 2 groups of 1:2 (vaccinated group: unvaccinated group), 80% power and 5% alpha, 422 participants were needed (159 patients in vaccinated group and 318 patients in unvaccinated group). Sample size calculation was performed in Epi Info software v5.5.5
Statistical analysis
Comparisons between groups of patients were made by using the chi-square test or Fisher’s exact test for categorical variables, independent sample t test for normally distributed continuous variables, and Mann-Whitney U test when the distribution was skewed. We used univariate and multivariate analysis to quantify the association of variables with all-cause death or HF hospitalization in an unadjusted model. Given the differences in the baseline characteristics among eligible participants in the treatment groups, a propensity score method was used to reduce this confounding effect. The propensity score was defined as the probability of treatment assignment subjective to observed baseline characteristics and the inverse probability weighting with regression adjustment (IPWRA) method on the basis of the propensity score was used to adjust confounding factors. We estimated the average treatment effect between two groups in propensity-adjusted model by inverse probability weighting with regression adjustment regression (IPWRA) technique. A p-value < 0.05 was considered to indicate statistical significance. All statistical procedures were performed using STATA software version 16.0
Sensitivity analysis
Because of the survival bias in the vaccination group, we performed a sensitivity analysis examining the association of influenza vaccine receipt within 12 months of heart failure diagnosis and clinical outcomes by divide patients into two groups that received flu vaccine less than 1 year and more than 1 year after HF diagnosis to evaluate the consistency of the result.