Diagnostic Algorithms to Study Post-Concussion Syndrome Using Electronic Health Records: Validating a Method to Capture an Important Patient Population

Implementation of Return to Learn Protocols for Student Athletes with Sport and Recreation Related Concussion: An Integrative Review of Perceptions, Challenges and Successes

Concussion or mild traumatic brain injury (mTBI) is a common phenomenon in the United States, with up to 3.6 million sport-related mTBIs diagnosed annually. Return to learn protocols have been developed to facilitate the reintegration of students into school after mTBI, however, the implementation of return to learn protocols varies significantly across geographic regions and school districts. An integrative review of the literature was performed using Whittemore and Knalf's methodology. A search of published literature was conducted using the PRISMA checklist. Database searches were conducted from March 2,019 to October 2,021 using the terms "mild traumatic brain injury" and "return to learn." Twenty-eight publications were included. Three themes were derived from this review: lack of policy, poor staff education on concussion symptoms and stakeholder communication breakdown. The development of communication patterns and use of a return to learn protocol could facilitate a gradual return to full academic workload after concussion.

Utility of CVLT-3 response bias as a measure of performance validity after traumatic brain injury

OBJECTIVE

We sought to determine the utility of a new performance validity index that was recently proposed. In particular, we wanted to determine if this index would be associated with a specificity of at least .90, a sensitivity of at least .40, and an Area Under the Curve of at least .70 in a traumatic brain injury (TBI) sample.

METHOD

We used logistic regression to investigate how well this new index could distinguish persons with TBI (n = 148) who were evaluated within 1-36 months after injury. All participants had been classified on the basis of at least two independent performance validity tests as having provided valid performance (n = 128) or invalid performance (n = 20).

RESULTS

The new performance validity index had acceptable specificity (.96) but had suboptimal sensitivity (.35) and Area Under the Curve (.66). It was concerning that almost half (5/12) of the cases that were identified by this index as providing invalid effort were false positives. Although a slightly more liberal cut-off improved sensitivity, the problem with poor positive predictive power remained. The conventional Forced Choice index had relatively better classification accuracy.

CONCLUSION

Differences in base rates between the original sample of Martin et al. and the current one most likely affected positive predictive power of the new index. Although their performance validity has excellent specificity, the current results do not support the application of this index in the clinical evaluation of patients with traumatic brain injury when base rates of invalid performance differ markedly from those in the original study.

Helping the science speak for itself: lessons learned developing video summaries for a precision medicine and health disparities collaborative

Skills to effectively communicate research findings are important to expand meaningful inclusion of the public in research, but attempts to summarize findings may be challenging given increasing reliance on succinct communications. Led by our Ethics Advisory Board and within the context of this work with the Precision Medicine and Health Disparities Collaborative, the authors decided to engage in the iterative process of developing video summaries. Our stakeholders taught us to incorporate novel strategies to engage broader audiences, leading to the production of two video summaries, a public summary toolbox and an adapted process for developing video summaries. The authors refreshed the static concept of 'research summaries' and found ways of producing summaries which emphasized a dynamic reframing of the shared benefits of science.

Conditional Vasospasm-Free Survival Following Aneurysmal Subarachnoid Hemorrhage

BACKGROUND

Following aneurysmal subarachnoid hemorrhage (SAH), patients are monitored closely for vasospasm in the intensive care unit. Conditional vasospasm-free survival describes the risk of future vasospasm as a function of time elapsed without vasospasm. Conditional survival has not been applied to this clinical scenario but could improve patient counseling and intensive care unit use. The objective of this study was to characterize conditional vasospasm-free survival following SAH.

METHODS

This was a single institution, retrospective cohort study of patients treated for aneurysmal SAH between 1/1/2000-6/1/2020. The primary outcome was the development of vasospasm defined by the first instance of either radiographic vasospasm on computed tomography angiography, Lindegaard Index > 3.0 by transcranial doppler ultrasonography, or vasospasm-specific intraarterial therapy. Multivariable Cox regression was performed, and conditional vasospasm-free survival curves were constructed.

RESULTS

A total of 528 patients were treated for aneurysmal SAH and 309 (58.5%) developed vasospasm. Conditional survival curves suggest patients who survive to postbleed day 10 without vasospasm have a nearly 90% chance of being discharged without vasospasm. The median onset of vasospasm was postbleed day 6. Age more than 50 years was associated with a lower risk (hazard ratio [HR] = .76; 95% confidence interval [CI] 0.64-0.91; p < 0.001). Higher initial systolic blood pressure (HR = 1.18; 95% CI 1.046-1.350; p = .008), Hunt-Hess grades 4 or 5 (HR = 1.304; 95% CI 1.014-1.676), and modified Fisher scale score of 4 (HR = 1.808; 95% CI 1.198-2.728) were associated with higher vasospasm than the respective lower grades.

CONCLUSION

Conditional survival provides a useful framework for counseling patients and making decisions around vasospasm risk for patients with aneurysmal SAH, while risk factor-stratified plots facilitate a patient-centric, evidence-based approach to these conversations and decisions.

Association of Preinjury Medical Diagnoses With Pediatric Persistent Postconcussion Symptoms in Electronic Health Records

OBJECTIVE

To identify risk factors and generate hypotheses for pediatric persistent postconcussion symptoms (PPCS).

SETTING

A regional healthcare system in the Southeastern United States.

PARTICIPANTS

An electronic health record-based algorithm was developed and validated to identify PPCS cases and controls from an institutional database of more than 2.8 million patients. PPCS cases (n = 274) were patients aged 5 to 18 years with PPCS-related diagnostic codes or with PPCS key words identified by natural language processing of clinical notes. Age, sex, and year of index event-matched controls (n = 1096) were patients with mild traumatic brain injury codes only. Patients with moderate or severe traumatic brain injury were excluded. All patients used our healthcare system at least 3 times 180 days before their injury.

DESIGN

Case-control study.

MAIN MEASURES

The outcome was algorithmic classification of PPCS. Exposures were all preinjury medical diagnoses assigned at least 180 days before the injury.

RESULTS

Cases and controls both had a mean of more than 9 years of healthcare system use preinjury. Of 221 preinjury medical diagnoses, headache disorder was associated with PPCS after accounting for multiple testing (odds ratio [OR] = 2.9; 95% confidence interval [CI]: 1.6-5.0; P = 2.1e-4). Six diagnoses were associated with PPCS at a suggestive threshold for statistical significance (false discovery rate P < .10): gastritis/duodenitis (OR = 2.8; 95% CI: 1.6-5.1; P = 5.0e-4), sleep disorders (OR = 2.3; 95% CI: 1.4-3.7; P = 7.4e-4), abdominal pain (OR = 1.6; 95% CI: 1.2-2.2; P = 9.2e-4), chronic sinusitis (OR = 2.8; 95% CI: 1.5-5.2; P = 1.3e-3), congenital anomalies of the skin (OR = 2.9; 95% CI: 1.5-5.5; P = 1.9e-3), and chronic pharyngitis/nasopharyngitis (OR = 2.4; 95% CI: 1.4-4.3; P = 2.5e-3).

CONCLUSIONS

These results support the strong association of preinjury headache disorders with PPCS. An association of PPCS with prior gastritis/duodenitis, sinusitis, and pharyngitis/nasopharyngitis suggests a role for chronic inflammation in PPCS pathophysiology and risk, although results could equally be attributable to a higher likelihood of somatization among PPCS cases. Identified risk factors should be investigated further and potentially considered during the management of pediatric mild traumatic brain injury cases.

Machine Learning for Subtyping Concussion Using a Clustering Approach

Background: Concussion subtypes are typically organized into commonly affected symptom areas or a combination of affected systems, an approach that may be flawed by bias in conceptualization or the inherent limitations of interdisciplinary expertise.

Objective: The purpose of this study was to determine whether a bottom-up, unsupervised, machine learning approach, could more accurately support concussion subtyping.

Methods: Initial patient intake data as well as objective outcome measures including, the Patient-Reported Outcomes Measurement Information System (PROMIS), Dizziness Handicap Inventory (DHI), Pain Catastrophizing Scale (PCS), and Immediate Post-Concussion Assessment and Cognitive Testing Tool (ImPACT) were retrospectively extracted from the Advance Concussion Clinic's database. A correlation matrix and principal component analysis (PCA) were used to reduce the dimensionality of the dataset. Sklearn's agglomerative clustering algorithm was then applied, and the optimal number of clusters within the patient database were generated. Between-group comparisons among the formed clusters were performed using a Mann-Whitney U test.

Results: Two hundred seventy-five patients within the clinics database were analyzed. Five distinct clusters emerged from the data when maximizing the Silhouette score (0.36) and minimizing the Davies-Bouldin score (0.83). Concussion subtypes derived demonstrated clinically distinct profiles, with statistically significant differences (p < 0.05) between all five clusters.

Conclusion: This machine learning approach enabled the identification and characterization of five distinct concussion subtypes, which were best understood according to levels of complexity, ranging from Extremely Complex to Minimally Complex. Understanding concussion in terms of Complexity with the utilization of artificial intelligence, could provide a more accurate concussion classification or subtype approach; one that better reflects the true heterogeneity and complex system disruptions associated with mild traumatic brain injury.

Automated Phenotyping Tool for Identifying Developmental Language Disorder Cases in Health Systems Data (APT-DLD): A New Research Algorithm for Deployment in Large-Scale Electronic Health Record Systems

Purpose Data mining algorithms using electronic health records (EHRs) are useful in large-scale population-wide studies to classify etiology and comorbidities (Casey et al., 2016). Here, we apply this approach to developmental language disorder (DLD), a prevalent communication disorder whose risk factors and epidemiology remain largely undiscovered. Method We first created a reliable system for manually identifying DLD in EHRs based on speech-language pathologist (SLP) diagnostic expertise. We then developed and validated an automated algorithmic procedure, called, Automated Phenotyping Tool for identifying DLD cases in health systems data (APT-DLD), that classifies a DLD status for patients within EHRs on the basis of ICD (International Statistical Classification of Diseases and Related Health Problems) codes. APT-DLD was validated in a discovery sample (N = 973) using expert SLP manual phenotype coding as a gold-standard comparison and then applied and further validated in a replication sample of N = 13,652 EHRs. Results In the discovery sample, the APT-DLD algorithm correctly classified 98% (concordance) of DLD cases in concordance with manually coded records in the training set, indicating that APT-DLD successfully mimics a comprehensive chart review. The output of APT-DLD was also validated in relation to independently conducted SLP clinician coding in a subset of records, with a positive predictive value of 95% of cases correctly classified as DLD. We also applied APT-DLD to the replication sample, where it achieved a positive predictive value of 90% in relation to SLP clinician classification of DLD. Conclusions APT-DLD is a reliable, valid, and scalable tool for identifying DLD cohorts in EHRs. This new method has promising public health implications for future large-scale epidemiological investigations of DLD and may inform EHR data mining algorithms for other communication disorders. Supplemental Material https://doi.org/10.23641/asha.12753578.