N C H S - Publications and Information Products - Response Errors in Surveys of Children's Immunizations

Series 6, No. 8. Response Errors in Surveys of Children's Immunizations. 56 pp. (PHS) 99-1083. GPO stock number and price forthcoming
View/download PDF 300 KB

Survey of children’s immunizations often require reports by adults who may not have access to records such as a child’s shot card. This report describes research to explore whether, in the absence of records, a) parental memory is sufficient for obtaining data of adequate quality, and b) if performance can be improved through the use of retrieval cues that enhance access to the relevant memories. Findings from the report concluded that parental reports were generally poor, and that attempts to improve memory performance were unsuccessful. To the extent possible, immunization surveys should rely on the use of medical provider record data.

Overall, there was little relationship between parental reports and the child’s vaccination status according to provider records. Parent reports of overall (4-3-1-3-3) up-to-date status revealed net overreporting (net bias of +53 percent). Accuracy for number of individual vaccines received was significantly above chance only for MMR, and in contrast to reports concerning the global questions, reports involving individual vaccines tended to represent net underreports. Finally, over- reporting was observed for vaccines that the child would never have received (smallpox, varicella, and pneumococcus).

The medical events calendar seemed more helpful than the personal events calendar because it encouraged respondents to draw specifically on memories of doctor visits and illnesses. In particular, its structure was helpful in laying out the 2-year reference period clearly. This calendar seemed to help respondents recall the months in which their children received vaccinations, but not which vaccines were given. Overall, only the medical events calendar appeared to be worthy of further investigation.

The use of the calendar, show card, and variation in presentation order had no significant affects on accuracy of reporting of either overall up-to-date status or number of individual vaccines. Overall, attempts to improve reporting through the use of recall aids were unsuccessful.

The only variable found to be related to reporting accuracy was an independent measure of the parent’s knowledge concerning the functions of each vaccine, parents who were about the median level concerning such knowledge were more accurate than those who were below the median.