Person-Specific Methods for Characterizing the Course and Temporal Dynamics of Concussion Symptomatology: A Pilot Study

Amanda R Rabinowitz,Aaron J Fisher

doi:10.1038/s41598-019-57220-1

Abstract

Better characterization of acute concussion symptomatology is needed in order to advance clinical and scientific understanding of persistent concussion symptoms. This paper aims to illustrate a novel framework for conceptualizing, collecting, and analyzing concussion symptom data. To that end, we describe the temporal and structural dynamics of acute concussion symptoms at the individual-patient level. Ten recently concussion adolescents and young adults completed 20 days of ecological momentary assessment (EMA) of post-concussion symptoms. Follow-up assessments were completed at 3 months post-injury. Network modeling revealed marked heterogeneity across participants. In the overall sample, temporal patterns explained the most variance in light sensitivity (48%) and the least variance in vomiting (5%). About half of the participants had symptom networks that were sparse after controlling for temporal variation. The other individualized symptom networks were densely interconnected clusters of symptoms. Networks were highly idiosyncratic in nature, yet emotional symptoms (nervousness, emotional, sadness), cognitive symptoms (mental fogginess, slowness), and symptoms of hyperacusis (sensitivity to light, sensitivity to noise) tended to cluster together across participants. Person-specific analytic techniques revealed a number of idiosyncratic features of post-concussion symptomatology. We propose applying this framework to future research to better understand individual differences in concussion recovery.

Highlights

The multifactorial and individual nature of post-concussion symptoms is not well captured by a latent disease model, which assumes that symptoms arise as the result of a singular pathological process—in the case of concussion, presumably a neurometabolic cascade of events triggered by transient deformation of brain tissue in response to trauma
Network theory stands in contrast to the latent variable and disease models, which purport that disorders arise from single underlying disease entities
The current conceptualization of post-concussion symptoms implies a latent disease model, wherein symptoms arise as the result of a singular pathological process—presumably transient neurometabolic disruption caused by mechanical deformation of brain tissue

Summary

Introduction

The multifactorial and individual nature of post-concussion symptoms is not well captured by a latent disease model, which assumes that symptoms arise as the result of a singular pathological process—in the case of concussion, presumably a neurometabolic cascade of events triggered by transient deformation of brain tissue in response to trauma. This assumption underlies the common practice of using the sum total of symptom-checklist items as an index of severity, which treats symptoms as though they (i) represent the same underlying process across individuals, (ii) are interchangeable indicators of injury severity, and (iii) are relatively stable over time. This paper represents the first instance of time series network models that include longitudinal trends and diurnal cycles within the estimation of the network

Objectives

Results

Conclusion