Validity of the mother's recall of her child's antibiotic use.

We compared the mother's recall of her child's antibiotic use with the information obtained from the child's health records, while studying the association between antibiotic use and prevalence of mutans streptococci among 102 (5-12 years) inner-city black children. Poor recall of the mother (concordance ratio = 62 percent; kappa = 0.18, p = 0.06) and the lower frequency reported by her (p = 0.02) biased the positive association between child's antibiotic use during the second year of life and oral presence of mutans streptococci toward the null value (Odds Ratio [OR] based on health records = 3.1, 95 percent Confidence Interval [CI] = 1.03-9.3; OR based on mother's recall = 1.6, 95 percent CI = 0.44-5.8). Multiple logistic regression analysis failed to identify any significant determinant of the accuracy of the mother's recall. Since mothers considerably underestimate child's use of antibiotics, at least among inner-city low income blacks, this method of data collection should be limited to circumstances where no other options are available, and such results should be interpreted with caution.