Table 1.
Bivariate and multiple logistic regression models of parental survey responses evaluating barriers to influenza vaccination by child's vaccination statusa
| Independent variable | Unadjusted logistic models | Multiple logistic models | |
|---|---|---|---|
| Unadjusted odds ratio (95% CI) | Adjusted odds ratio (95% CI) | P value | |
| My child is too sick to receive the flu vaccine | 1.37 (1.03, 1.82) | 1.35 (1.00, 1.82) | .051 |
| Flu vaccines work | 1.25 (0.95, 1.64) | 1.32 (0.993, 1.75) | .056 |
| Flu vaccines are safe | 2.59 (1.85, 3.66) | 2.50 (1.76, 3.58) | <.0001 |
| I would like my child to get the vaccine at the pediatrician's office | 1.22 (0.90, 1.66) | 1.16 (0.84, 1.61) | .356 |
| A family member has had a bad experience with flu vaccines | 0.79 (0.58, 1.07) | 0.81 (0.59, 1.10) | .177 |
| Flu vaccines are needed every year | 3.25 (2.28, 4.69) | 3.30 (2.30, 4.81) | <.0001 |
| My child already gets enough shots | 0.43 (0.27, 0.65) | 0.43 (0.27, 0.66) | .0002 |
For all models, the dependent variable was the vaccination status of the patient (vaccinated/not vaccinated). Parental survey responses with a significance level of <0.2 in the bivariable analysis were modeled independently of each other using logistic regression. Unadjusted logistic regression was used to assess the unadjusted association between each independent variable and vaccination status. Multiple logistic regression was used to adjust the association for baseline imbalances of insurance status, high‐risk status, and age. The following table presents the results of the unadjusted and adjusted odds ratios (OR) and the corresponding 95% CI values. For each independent variable, the referent group was the “do not agree” response.