[Is a Time Lag Between Diagnostic Information from Different Data Sources in Data Linkage Projects a Problem? an Empirical Evaluation Using ADHD Diagnosis Data from the INTEGRATE-ADHD Project].

Abstract:In the project INTEGRATE-ADHD, routine data from a German statutory health insurance company (DAK-Gesundheit) was linked with data from an online survey and clinical online diagnostics. The period between the documentation of a child's ADHD diagnosis in the routine data and the parent report of the diagnosis in the survey was at least nine and at most 32 months. Clinical online diagnostics according to the German AWMF-S3 guideline took place between three and five months after the survey. Only about two-thirds of the parents reported the ADHD diagnosis of their child in the survey and only just under two-thirds of the administrative ADHD diagnoses were clinically confirmed. Based on data from 1,355 children and adolescents with incident ADHD diagnosis, this study used descriptive statistics and logistic regression to examine whether the time lag between the date of diagnosis documentation and the parent report in the survey or the diagnosis in the clinical examination could explain the discrepancies between the different diagnosis data, for example, due to parents' recall bias or changes in in the children's symptom presentation. The date when the diagnosis was coded could be approximated with the start and end date of the treatment period in which the diagnosis was coded. Subsequently, the effect of the time lag was analyzed using descriptive and multivariate statistics. Even when the date of diagnostic coding was differently approximated, no statistically significant association was observed between the time lag and discrepancies between the diagnosis data. Accordingly, the present analyses did not provide evidence that a time-dependent recall bias of the parents or changes in the presentation of symptoms were associated with unreported or clinically unconfirmed ADHD diagnoses.