We used data from the 2003 National Immunization Survey (NIS) for our study. Data are collected in the NIS in two phases: a telephone survey of households with landline telephones that have children aged 19 to 35 months, followed by a survey mailed to those children's vaccination providers. We used provider-reported vaccination histories to determine vaccination status in our study. In 2003, 30,909 households with 19- to 35-month-old children completed the NIS telephone interview; among those, we obtained provider-reported vaccination histories for 21,291 children. In 2003, the response rate of the telephone portion of the NIS was 70%, and the percentage of children with an adequate provider-reported vaccination history from the mailed survey was 78%. In 2003, 3,576 households with 19- to 35-month-old children were randomly selected to be administered a series of questions regarding parents' concerns about vaccine safety. Among those households, 3,403 completed that series of questions, and 2,921 19- to 35-month-old children had provider-reported vaccination histories returned from sampled children's vaccination providers.
In our study, we analyzed data from the 2,921 19- to 35-month-old children who had provider-reported vaccination histories returned in the mailed survey to their vaccination providers, and whose parents completed the NIS telephone interview and were administered questions about their vaccine safety concerns. Among the 2,921 completed telephone interviews that yielded provider-reported vaccination coverage histories for sampled children, 96% were conducted with the child's parent, nearly 3% were conducted with the child's grandparent, and a little more than 1% were conducted with another family member who was ≥18 years of age and knowledgeable about the child's vaccination history. For brevity, we refer to the respondent in the telephone portion of the NIS interview as the child's parent.
Because the vaccination schedule13
specifies that children are to be administered all recommended vaccine doses by 19 months of age, we defined “timely” vaccination coverage as vaccination coverage at 19 months of age and assessed coverage by a review of the provider-reported vaccination history of all sampled children. We defined “catch-up” vaccination coverage as vaccination coverage at 24 months of age and assessed this coverage by a review of the provider-reported vaccination histories of all sampled children ≥24 months of age. Children were defined to be up-to-date (UTD) at those milestone ages if their vaccination history indicated that they were administered ≥4 doses of diphtheria, tetanus, and pertussis (DTaP and/or DTP) vaccine; ≥3 doses of polio vaccine; ≥1 dose of measles, mumps, and rubella (MMR) vaccine; ≥3 doses of Haemophilus influenzae
type b (Hib) vaccine; ≥3 doses of hepatitis B (Hep B) vaccine; and ≥1 dose of varicella vaccine. Children were defined to be 4:3:1:3:3:1 UTD at those milestone ages if they were UTD for DTaP/DTP, polio, MMR, Hib, Hep B, and varicella vaccines.
In the survey, parents were asked if they ever decided to delay a vaccine dose and if they ever decided to not allow the administration of a dose. In our study, parents who answered “yes” to either of these two questions were categorized as intentionally delaying vaccine administration. Parents who reported delaying administration were asked to provide one reason for the delay, and parents who reported deciding to not allow the administration of a dose were asked to provide one reason for the delay. Because of this, parents who were categorized in our study as intentionally delaying vaccine administration could provide up to two reasons for the delay.
To evaluate the association between a parental report of hearing or reading unfavorable information about vaccines and the parents' decision to intentionally delay vaccine administration, NIS interviewers asked parents if they had heard or read about vaccines sometimes not preventing disease, not being safe or having serious side effects, being opposed by groups for political or religious reasons, and being opposed by groups that oppose vaccines for health reasons.
We used SUDAAN® software14
in all of our statistical analyses to account for the NIS design and sampling weights. Estimated percentages are reported along with their 95% confidence intervals (CIs). We considered differences between estimated percentages to be significantly different if a z-score test used to compare the estimates had a p
-value ≤0.05. Smith et al.15
provide a detailed description of the statistical methods used in the NIS. The NIS has been approved by the Centers for Disease Control and Prevention (CDC) Institutional Review Board.