Ensemble machine learning prediction of posttraumatic stress disorder screening status after emergency room hospitalization.

Posttraumatic stress disorder (PTSD) develops in a substantial minority of emergency room admits. Inexpensive and accurate person-level assessment of PTSD risk after trauma exposure is a critical precursor to large-scale deployment of early interventions that may reduce individual suffering and societal costs. Toward this aim, we applied ensemble machine learning to predict PTSD screening status three months after severe injury using cost-effective and minimally invasive data. Participants (N = 271) were recruited at a Level 1 Trauma Center where they provided variables routinely collected at the hospital, including pulse, injury severity, and demographics, as well as psychological variables, including self-reported current depression, psychiatric history, and social support. Participant zip codes were used to extract contextual variables including population total and density, average annual income, and health insurance coverage rates from publicly available U.S. Census data. Machine learning yielded good prediction of PTSD screening status 3 months post-hospitalization, AUC = 0.85 95% CI [0.83, 0.86], and significantly outperformed all benchmark comparison models in a cross-validation procedure designed to yield an unbiased estimate of performance. These results demonstrate that good prediction can be attained from variables that individually have relatively weak predictive value, pointing to the promise of ensemble machine learning approaches that do not rely on strong isolated risk factors.

Prospective Evaluation of Posttraumatic Stress Disorder and Depression in Orthopaedic Injury Patients With and Without Concomitant Traumatic Brain Injury.

OBJECTIVES: Psychological morbidities after injury [eg, posttraumatic stress disorder (PTSD) and depression] are increasingly recognized as a significant determinant of overall outcome. Traumatic brain injury (TBI) negatively impacts outcomes of patients with orthopaedic injury, but the association of concurrent TBI, orthopaedic injury, and symptoms of PTSD and depression has not been examined. This study's objective was to examine symptoms of PTSD and depression in patients with orthopaedic trauma with and without TBI. DESIGN: Longitudinal prospective cohort study. SETTING: Urban Level I Trauma Center in the Southwest United States. PATIENTS/PARTICIPANTS: Orthopaedic trauma patients older than 18 years admitted for ≥24 hours. MAIN OUTCOME MEASUREMENTS: Questionnaires examining demographics, injury-related variables, PTSD, and depression were administered during hospitalization and 3, 6, and 12 months later. Orthopaedic injury and TBI were determined based on ICD-9 codes. Generalized linear models determined whether PTSD and depression at follow-up were associated with TBI. RESULTS: Of the total sample (N = 214), 44 (21%) sustained a TBI. Those with TBI had higher rates of PTSD symptoms, 12 months postinjury (P = 0.04). The TBI group also had higher rates of depressive symptoms, 6 months postinjury (P = 0.038). CONCLUSIONS: Having a TBI in addition to orthopaedic injury was associated with significantly higher rates of PTSD at 12 months and depression at 6 months postinjury. This suggests that sustaining a TBI in addition to orthopaedic injury places patients at a higher risk for negative psychological outcomes. The findings of this study may help clinicians to identify patients who are in need for psychological screening and could potentially benefit from intervention. LEVEL OF EVIDENCE: Prognostic Level III. See Instructions for Authors for a complete description of levels of evidence.

Discrepancy in caregiving expectations predicts problematic alcohol use among caregivers of trauma injury patients six months after ICU admission.

This prospective study examined the influence of caregiving variables on the development of problematic alcohol use among family members of patients admitted to an urban Level I trauma center. Data were collected from 124 caregivers 48 hrs after initial hospitalization of their family member. The final sample included 81 participants (24.6% male; Mage = 47.8) who completed their follow-up assessment at 6 months. Hierarchical linear and logistic regression analyses assessed increases in consumption and odds of a positive screen for problematic alcohol use in association with caregiver burden, actual time spent in the caregiving role, and caregiving differential (i.e., anticipated time spent caregiving at baseline in relation to actual time caregiving at 6 months). At 6 months, 24.7% of caregivers screened positive for problematic alcohol use. Results uniquely highlighted caregiving differential as a significant predictor of both increases in general alcohol consumption (ΔR2 = .06, p < .01) and odds of screening positive for problematic alcohol use at 6 months (Odds Ratio = 1.05, 95% CI [1.02-1.09]). More specifically, our adjusted model found that providing 10% more time caregiving, relative to expectations at baseline, was associated with an increase in the probability of problematic alcohol use by 22% (95% CI: 8-37%) at 6 months. These results suggest that a discrepancy in expectations regarding anticipated time caregiving and actual time caregiving, rather than solely the amount of caregiving or perceived caregiver burden, may be an important predictor of caregiver alcohol use 6 months after a family member's ICU hospitalization. (PsycINFO Database Record

Secondary traumatic stress in emergency medicine clinicians.

Previously called Secondary Traumatic Stress (STS), secondary exposure to trauma is now considered a valid DSM-5 Criterion A stressor for posttraumatic stress disorder (PTSD). Previous studies have found high rates of STS in clinicians who treat traumatically injured patients. However, little research has examined STS among Emergency Medicine (EM) physicians and advanced practice providers (APPs). The current study enrolled EM providers (N = 118) working in one of 10 hospitals to examine risk factors, protective factors, and the prevalence of STS in this understudied population. Most of the participants were physicians (72.9%), Caucasian (85.6%), and male (70.3%) with mean age of 39.7 (SD = 8.9). Overall, 12.7% of the sample screened positive for STS with clinical levels of intrusion, arousal, and avoidance symptom clusters, and 33.9% had at least one symptom cluster at clinical levels. Low resilience and a history of personal trauma were positively associated with positive STS screens and STS severity scores. Borderline significance suggested that female gender and spending ≥10% of one's time with trauma patients could be additional risk factors. Findings suggest that resilience-building interventions may be beneficial.

The effect of depressive symptoms on social support one year following traumatic injury.

BACKGROUND: Depression is a common mental health outcome after traumatic injury, negatively impacting physical outcomes and increasing the cost of care. Research shows that the presence and quality of support is a leading protective factor against depression post-injury; however, research is vague on the directional effects of both factors over the course of recovery. METHODS: 130 patients admitted to a Level I Trauma Center were recruited to a prospective study examining overall outcomes one-year after injury. Effects of social support and depression at baseline and 12-months post-injury were examined using correlational and cross-lagged path model analyses. Additional follow-up analyses were conducted for depression on specific types of social support. RESULTS: Findings replicated previous research suggesting depression and social support were inversely related. Initial depression at time of traumatic injury was predictive of social support 12-months after their injury, but initial social support levels did not significantly predict depression at 12-months. Additionally, initial depression significantly predicted attachment, social integration, reassurance of worth, and guidance 12-months later. LIMITATIONS: Findings of the analyses are limited by lack of experimentation and inability to control for other related variables. CONCLUSIONS: Findings of the present study support the notion that initial depression predicts poorer social support in recovery, in lieu of prevailing theory (i.e., initial support buffers against later depression) in a sample of trauma patients. These findings highlight the need for medical staff to target specific factors during inpatient stay, such as addressing depressive symptoms and preparing family members and caregivers prior to discharge.

Prospective Evaluation of Posttraumatic Stress Disorder in Injured Patients With and Without Orthopaedic Injury.

OBJECTIVES: The study purposes were to prospectively evaluate occurrence of posttraumatic stress (PTS) symptoms at hospital admission and 6 months later in patients with orthopaedic injury; to explore differences in PTS symptoms in those with and without orthopaedic injury; and to determine whether PTS symptoms are influenced by orthopaedic injury type. DESIGN: Prospective, longitudinal observational study. SETTING: Level 1 Trauma Center. PATIENTS/PARTICIPANTS: Two hundred fifty-nine participants admitted for at least 24 hours. MAIN OUTCOME MEASUREMENTS: The Primary Care Posttraumatic Stress Disorder (PTSD) Screen (PC-PTSD) measured PTSD symptoms during hospitalization. The PTSD Checklist-Civilian Version (PCL-C) measured PTS symptoms at 6 months. RESULTS: In orthopaedic patients, 28% had PTS at 6 months, compared with 34% of nonorthopaedic patients. Odds ratios (ORs) were calculated to determine the influence of pain, physical and mental function, depression, and work status. At 6 months, if the pain score was 5 or higher, the odds of PTS symptoms increased to 8.38 (3.55, 19.8) (P < 0.0001). Those scoring below average in physical function were significantly more likely to have PTS symptoms [OR = 7.60 (2.99, 19.32), P < 0.0001]. The same held true for mental functioning and PTS [OR = 11.4 (4.16, 30.9), P < 0.0001]. Participants who screened positive for depression had a 38.9 (14.5, 104) greater odds (P < 0.0001). Participants who did not return to work after injury at 6 months were significantly more likely to have PTS [OR = 16.5 (1.87, 146), P = 0.012]. CONCLUSIONS: PTSD is common in patients after injury, including those with orthopaedic trauma. At 6 months, pain of 5 or greater, poor physical and mental function, depression, and/or not returning to work seem to be predictive of PTSD. Orthopaedic surgeons should identify and refer for PTSD treatment given the high incidence postinjury. LEVEL OF EVIDENCE: Prognostic Level I. See Instructions for Authors for a complete description of levels of evidence.

Psychological factors predicting outcome after traumatic injury: the role of resilience.

BACKGROUND: Increasingly, studies have examined the psychological impact on individuals who survive a traumatic physical injury. The primary aim of this study was to determine the stability of resilience and its association with depressive symptoms. METHODS: This study included 110 adults admitted to a Level I trauma center. Resilience and depression were measured at baseline and 12 months. Injury-related variables included Glasgow Coma Scale, Injury Severity Score, etiology of injury, and type of injury. RESULTS: Analysis revealed that resilience remained stable over 12 months regardless of injury severity, etiology, or type. Negative correlations were found between baseline resilience and 12-month depression (P < .01), as well as Glasgow Coma Scale and depression (P = .001). CONCLUSIONS: Injured individuals with low resilience are more likely to be depressed at 12 months. Assessing resilience at the time of injury may be useful in identifying those at risk for depression 1 year later.