Comparison of Methods for Classifying Persistent Post-Concussive Symptoms in Children.

Abstract

Pediatric mild traumatic brain injury (pmTBI) has received increased public scrutiny over the past decade, especially regarding children who experience persistent post-concussive symptoms (PPCS). However, several methods for defining PPCS exist in clinical and scientific literature, and even healthy children frequently exhibit non-specific, concussive-like symptoms. Inter-method agreement (six PPCS methods), observed misclassification rates, and other psychometric properties were examined in large cohorts of consecutively recruited adolescent patients with pmTBI (n = 162) 1 week and 4 months post-injury and in age/sex-matched healthy controls (HC; n = 117) at equivalent time intervals. Six published PPCS methods were stratified into Simple Change (e.g., International Statistical Classification of Diseases and Related Health Problems, 10th revision [ICD-10]) and Standardized Change (e.g., reliable change indices) algorithms. Among HC, test-retest reliability was fair to good across the 4-month assessment window, with evidence of bias (i.e., higher symptom ratings) during retrospective relative to other assessments. Misclassification rates among HC were higher (>30%) for Simple Change algorithms, with poor inter-rater reliability of symptom burden across HC and their parents. A 49% spread existed in terms of the proportion of pmTBI patients "diagnosed" with PPCS at 4 months, with superior inter-method agreement among standardized change algorithms. In conclusion, the self-reporting of symptom burden is only modestly reliable in typically developing adolescents over a 4-month period, with additional evidence for systematic bias in both adolescent and parental ratings. Significant variation existed for identifying pmTBI patients who had "recovered" (i.e., those who did not meet individual criteria for PPCS) from concussion across the six definitions, representing a considerable challenge for estimating the true incidence rate of PPCS in published literature. Although relatively straightforward to obtain, current findings question the utility of the most commonly used Simple Change scores for diagnosis of PPCS in clinical settings.