Natural Language Processing - A Surveillance Stepping Stone to Identify Child Abuse

OBJECTIVE

We aimed to refine a natural language processing (NLP) algorithm that identified injuries associated with child abuse and identify areas in which integration into a real-time clinical decision support (CDS) tool may improve clinical care.

METHODS

We applied an NLP algorithm in "silent mode" to all emergency department (ED) provider notes between July 2021 and December 2022 (n = 353) at 1 pediatric and 8 general EDs. We refined triggers for the NLP, assessed adherence to clinical guidelines, and evaluated disparities in degree of evaluation by examining associations between demographic variables and abuse evaluation or reporting to child protective services.

RESULTS

Seventy-three cases falsely triggered the NLP, often due to errors in interpreting linguistic context. We identified common false-positive scenarios and refined the algorithm to improve NLP specificity. Adherence to recommended evaluation standards for injuries defined by nationally accepted clinical guidelines was 63%. There were significant demographic differences in evaluation and reporting based on presenting ED type, insurance status, and race and ethnicity.

CONCLUSIONS

Analysis of an NLP algorithm in "silent mode" allowed for refinement of the algorithm and highlighted areas in which real-time CDS may help ED providers identify and pursue appropriate evaluation of injuries associated with child physical abuse.

Inadequate sociomedical evaluation of possible abusive head trauma in Yokohama

BACKGROUND

There have been no previous studies on the adequacy of combined evaluation of possible abusive head trauma cases by frontline medical personnel, hospital-based child protection teams, and child protective services in local districts of Japan.

METHODS

We conducted a questionnaire survey of hospitalized patients under 24 months old with a diagnosis of intracranial hemorrhage (ICH) from January 2011 to December 2013. Eleven large-scale general hospitals in Yokohama, Japan were surveyed, which provide centralized inpatient care to moderately-severely ill children.

RESULTS

A total of 51 ICH patients were listed from eight hospitals. Median patient age was 7 months, and 84% were younger than 12 months. The most common diagnosis on computed tomography was subdural hematoma (n = 26; 51%). Of a total of 51 cases, 31 (61%) occurred inside the home; the injury scene was unknown in six cases (12%). We reviewed these 37 cases from the viewpoint of evaluation with concern for suspected child abuse. Three out of 37 patients (8%) were not examined for inflicted skin lesions, and skeletal surveys and funduscopy were not conducted in 14 (38%) and 15 (41%), respectively. Thirteen out of 37 cases (35%) were not reported to hospital-based child protection teams and 22 (59%) were not reported to regional child protective services.

CONCLUSION

The sociomedical evaluation of possible child abuse appears to be systematically inadequate in Yokohama.

A novel self-evaluation tool to assess the team function of a child protection team

OBJECTIVE

The aim of this study was to develop a reliable and valid self-evaluation tool for use by child protection team (CPT) members.

METHODS

An online survey was administered to members of 10 CPTs. The survey included the following 3 sections: 1) initial conditions (eg, team composition, resources), 2) enabling conditions (eg, team effort, strategy), and 3) team effectiveness (eg, team cohesion, meeting performance standards). Each section contained multiple subscales. Internal consistency was calculated using Cronbach α. To evaluate construct validity, the subscale scores of the most advanced teams who qualified as centers of excellence (n = 3) were compared with the subscale scores of the other teams (n = 7) to determine whether the tool could distinguish between the two.

RESULTS

Of 116 team members, 83 (72%) completed the survey. The subscales exhibited good internal consistency (α = .71-.97). The 3 centers of excellence had significantly higher mean scores than the other 7 CPTs on the following subscales: incentives (in the initial conditions section, 61.46 vs 38.89; P = .003), effort (in the enabling conditions section, 79.31 vs 67.70; P = .003), and professional growth (in the team effectiveness section, 83.89 vs 80.40; P = .004).

CONCLUSIONS

This novel survey demonstrates satisfactory test characteristics and can be used to assess CPT performance and identify areas for improvement.

Factors that influence the effectiveness of child protection teams

OBJECTIVES

More than $55 million is spent on hospital-based child protection teams (CPTs) annually, but there is no consensus on what makes CPTs effective. The objective of this study was to create expert consensus on tasks that CPTs should perform and factors that contribute to effectiveness.

METHODS

A modified Delphi approach was used to create expert consensus among professionals with experience working on or with hospital-based CPTs. Three initial rounds of surveys were conducted; a first round of open-ended questions generated topics related to CPT tasks and factors related to team effectiveness. A Likert scale (range: 1-7) determined rank. In the fourth round, participants ranked the top 5 variables associated with effectiveness.

RESULTS

Twenty-six (90%) of 29 participants completed the first 3 rounds, and 20 (67%) completed the final ranking. Experts believed that CPTs should provide communication of findings to appropriate agencies (mean Likert score: 7.0), court testimony (7.0), medical consultations (6.9), multidisciplinary case review (6.6), and forensic interviews (6.0). CPT success should be determined by professionals who use CPT services (6.6) and CPT members (6.5). Variables that were ranked most often as critical to effectiveness included interdisciplinary collaboration (95% of participants), provision of resources (80%), and team collegiality (75%). Variables that were ranked as most detrimental included inadequate staffing (85%) and lack of collegiality (80%).

CONCLUSIONS

A multidisciplinary team working in a collegial atmosphere seems to be the major key to CPT effectiveness. In addition to providing services, CPTs should focus on improving collegiality and interdisciplinary collaboration and should seek performance feedback from referring professionals and CPT members.