One hundred (87.0%) of 115 Iowa hospitals participated in season 1. (There are actually 116 hospitals in Iowa, but 2 hospitals are part of the same health system and reported aggregated rates). Individual hospital vaccination rates ranged from 43.5% to 99.2% (mean, 72.4%; median, 73.1%). Because hospitals vary in size, this represents a 67.9% vaccination rate for HCWs employed by acute care hospitals in Iowa. In season 2, 115 (100%) of 115 Iowa hospitals participated. Individual hospital vaccination rates ranged from 53.6% to 100% (mean 79.5%; median 82.0%). This represents a 76% vaccination rate for HCWs employed by acute care hospitals in Iowa.
On the basis of data from the 99 hospitals participating in both seasons, the overall vaccination rate increased a mean of 7.5 percentage points per hospital (P < .001; 1-sided test), ranging from a decrease of 9.9 percentage points to an increase of 33.5 percentage points from season 1 to season 2. Eighty-one hospitals (82%) had increased rates, and 18 (18%) had decreased rates, from season 1 to season 2.
Univariable regression results for both seasons are presented in . All variables with coefficients that were significantly different than zero with 90% confidence intervals are included in the table. The only variables to meet this criterion were urban location, more than 1,000 full-time-equivalent employees, more than $45 million in annual revenue, and community access hospital type. The first 3 variables have negative coefficients, indicating that large or urban hospitals have lower vaccination rates than other Iowa hospitals. Specifically, in season 1, urban hospitals had vaccination rates that were 6.3 percentage points lower than hospitals in other locations; during season 2, this difference was 7.6 percentage points.
Results of Univariable Regression Analysis for Influenza Vaccination Rates of Healthcare Workers
All of the significant variables are highly correlated. The first 3 variables are positively correlated: urban hospitals are also most likely to be large hospitals. In contrast, community access hospitals are least likely to be large and located in urban locations, producing a negative correlation with the 3 other variables. Because of these correlations, no multivariable model that included all significant variables from the univariable regressions for season 1 had any coefficients that differed significantly from zero.
The third survey was only administered during season 2, and 15 variables were collected (). However, all hospitals that provided complete data gave free vaccination to all HCWs, so the first 2 items, “Influenza vaccine free of charge to all HCWs with direct patient care” and “Influenza vaccine free of charge to all HCWs” were uninformative. These variables were subsequently dropped from further analysis. The remaining 13 items were used as univariable predictors in the model of season 2 immunization rates. Some of the 115 hospitals did not respond to the third survey, so the number of responses in these models was 80 (69.6%), except for declination statements, for which the number of responses was 79 (68.7%). Most of the questionnaire items were statistically insignificant in these models, except for the presence of declination statements (P < .001). The univariable results for these items are presented in . A variable that contained the sum of the number of questions to which the hospital replied “yes” was also considered, and its coefficient was significantly different than zero (P = .042).
Using adjusted R2 as the criterion, the best multivariable model for the season 2 data included urban location, the “sum” variable, and the use of declination statements. The results of this model are presented in . As in the univariable results, urban location is associated with lower levels of staff influenza vaccination. The perception of respondents that their hospital administration was committed to staff vaccination and the use of declination forms were both factors associated with higher levels of staff influenza vaccination. Specifically, controlling for the other variables, declination statements were associated with an increase in influenza vaccination rates of 12.6 percentage points.
Results of Multivariable Regression Analysis for Influenza Vaccination Rates of Healthcare Workers for Season 2