Serological response to influenza vaccination among adults hospitalized with community‐acquired pneumonia

Ninety‐five adults enrolled in the Etiology of Pneumonia in the Community study with negative admission influenza polymerase chain reaction (PCR) tests received influenza vaccination during hospitalization. Acute and convalescent influenza serology was performed. After vaccination, seropositive (≥1:40) hemagglutination antibody titers (HAI) were achieved in 55% to influenza A(H1N1)pdm09, 58% to influenza A(H3N2), 77% to influenza B (Victoria), and 74% to influenza B (Yamagata) viruses. Sixty‐six (69%) patients seroconverted (≥4‐fold HAI rise) to ≥1 strain. Failure to seroconvert was associated with diabetes, bacterial detection, baseline seropositive titers for influenza B (Yamagata), and influenza vaccination in the previous season.

TA B L E 1 Distribution of epidemiological and clinical factors among patients who did and did not have seroconversion, defined as at least one ≥4-fold rise in HAI antibody titers between acute and convalescent serum specimens.   sion for pathogen detection and tested using multiple modalities.
Patients included in this analysis were hospitalized, had radiographically confirmed pneumonia, and a naso/oropharyngeal (NP/ OP) swab negative for influenza by PCR. Acute (collected upon enrollment) and convalescent (obtained 3-10 weeks after enrollment) sera were tested for hemagglutination inhibition (HAI). Patients were also required to have received influenza vaccination during the current hospitalization (≥2 weeks before collection of convalescent serum).

| Laboratory studies
Hemagglutination inhibition assays using standardized methods were performed on paired sera for the following strains circulat-

| Serologic responses to influenza vaccination
A patient was considered to have seroconversion if a ≥ 4-fold rise in HAI antibody titers occurred between acute and convalescent serum specimens with a convalescent titer of ≥40 for influenza A(H1N1)pdm09, A(H3N2), influenza B Victoria lineage, or influenza B Yamagata lineage. 8,9 Seropositive titers were defined as HAI ≥ 1:40. 10

| Statistical methods
Descriptive statistics included number (%) and median (interquartile range, IQR) for categorical and continuous variables, respectively.
Demographic and clinical characteristics were compared between seroconversion and non-seroconversion groups using Pearson's chisquare test for categorical variables and Wilcoxon rank-sum test for continuous variables.
To identify variables associated with non-seroconversion in a setting with a relatively large number of covariates but a limited number of observations, we first identified a subset of potential variables using a least absolute shrinkage and selection operator approach (lasso) regression strategy. 11 We then applied a multivariable logistic regression approach to assess the association between the subset of factors identified by the lasso, and the probability of non-seroconversion. 12 The

| Patient population
Among the 2320 adults with radiographic pneumonia enrolled in the

| Factors associated with non-seroconversion
In comparison with patients who seroconverted, patients who did not seroconvert were more likely to have diabetes mellitus (P = 0.033) and a baseline HAI ≥ 40 for influenza B (Yamagata, P = 0.026) in bivariate analyses (Table 1). Median age, immunosuppression, PSI score, ICU admission, and duration of hospitalization did not differ between those who seroconverted and those who did not.

Influenza vaccination during CAP hospitalization resulted in sero-
conversion to ≥1 influenza virus strains in 69% of the 95 patients.
We lacked a directly comparable control group of non-hospitalized adults, but overall seroconversion rates observed here were suboptimal. We had anticipated that age, mild immunosuppression, or measures of CAP severity (eg, PSI score, duration of hospitalization, and ICU admission) might explain differences in seroconversion, but these associations were not observed (Table 1). Prior receipt of influenza vaccine and diabetes has been previously identified as risk factors for non-seroconversion after influenza vaccination in people who were not acutely ill. 3,4 Seropositive titers were achieved for >50% of influenza A strains and about 75% of influenza B strains following vaccination.

ACK N OWLED G EM ENTS
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.