gms | German Medical Science

Background: Children with attention deficit / hyperactivity disorder (ADHD) have a higher risk of accidents and injuries than children without ADHD. Stimulants (e.g. methylphenidate) reduce the symptoms of ADHD, but it is not clear if they also prevent accidents. Case-control or cohort studies to assess reduction of accidents are complicated by the fact that medication can be also an indicator of more severe disease (confounding by indication). We used two case-only designs to account for 1) confounding by time invariant characteristics of the subjects with ADHD and 2) age trends in exposure (medication) and outcome (accidents).

Methods: We used the German Pharmacoepidemiological Research Database (GePaRD) to identify children suffering from ADHD who were hospitalised due to injuries. To identify valid ADHD cases we requested a single inpatient ADHD diagnosis, two outpatient ADHD diagnoses or a combination of an outpatient diagnosis and specific medication within one year. We selected all 3 to 17 years old children with ADHD in 2005 and 2006 (N=37,650) and identified among them 2,186 children with an inpatient treatment with diagnoses indicating an injury (ICD-10 codes from the groups S and T, selected on the basis of the Injury Mortality Matrix) in 2004 to 2008. For statistical analysis we applied case-crossover method and subsequently the self-controlled case series design. Based on prescriptions of methylphenidate or atomoxetine we ascertained the treatment status of the children at the date of hospitalization and at the control time. A person was under treatment at a given time point if the days covered by the last prescription before that time point included the respective time point. Number of days covered by a prescription were calculated as packet size multiplied by number of packets prescribed. The control time point for the case cross-over was chosen 90 days before the event. The control time of the case series was the entire study period without an event.

Results: When all 2,186 children with ADHD and inpatient treatment for injuries were considered, medication was associated with a risk reduction (odds ratio of 0.80 (95% CI 0.65–0.99)) by the case series method. When the analysis was restricted to the age group 9 to 10 years old, the protective effect increased to 0.56 (0.35–0.89). The odds ratios estimated in the case-crossover design were 0.60 (0.42–0.89) for the full sample and 0.54 (0.27–1.02) for the age group 9 to 10 years old.

Conclusion: Methylphenidate and Atomoxetine reduced the risk of inpatient treatment with diagnoses of severe injuries by circa 40%. Both case-only designs indicate a similar risk reduction.