Transient exposures and the risk of childhood injury: a case-crossover study in Greece

We used a case-crossover design to evaluate short-term effects of several exposures on the risk of childhood accident. One hundred fifty-six hospitalized children with injuries responded to an interviewer-administered questionnaire that included, among other variables, information concerning transient exposures that had terminated within 26 hours before the occurrence of the accident. We considered the 2-hour interval preceding the accident as the likely effect period and made within-individual comparisons between this period and the remaining 2-hour intervals during which the child was awake. We used conditional logistic regression analyses to evaluate the effect of the transient exposures on the occurrence of childhood accidents. We present univariate and multivariate analyses adjusting for possible within-person confounding by exposure to more than one exposure simultaneously and for clock time. The adjusted odds ratios (ORs) and 95% confidence intervals (CIs) were as follows: for strenuous physical activity, OR = 24.2, 95% CI = 10.8-54.4; for intellectual exertion, OR = 9.0, 95% CI = 1.9-25.8; for involvement in family quarrels, OR = 2.6, 95% CI = 0.4-16.9; for school examination, OR = 3.8, 95% CI = 1.5-9.4; and for a pleasing event, OR = 3.4, 95% CI = 1.5-8.2. Other transient exposures were not associated with increased accident risk. Comparison of the overall frequency of reported transient events between the first of the control intervals (3rd and 4th hours before the accident) and the 2-hour interval covering the 25th and 26th hours before the accident suggested that information bias may have led to slight overreporting of transient exposures during the period most proximal to the accident; this bias, however, was too small to explain the marked risk elevations associated with the indicated transient exposures. We conclude that several transient exposures are important component causes in the occurrence of childhood accidents.