Early Identification of Post-Traumatic Stress Disorder in Trauma Patients: Development of a Multivariable Risk Prediction Model

Data Collection

Data for this study were prospectively collected at an urban, Level 1 Trauma Center from 2018 to 2019. Patients 18 and older presenting with an operative orthopedic trauma injury were eligible for enrollment in Institutional Review Board-approved comprehensive social determinants of health study. Data at the time of injury were collected through an initial 325 question in-person survey, medical records, and the hospital’s trauma registry database. These factors included gender, race, age, tobacco history, relationship with the health care system, insurance status, injury mechanism, and injury severity score (ISS). The Screener and Opioid Assessment for Patients with Pain (SOAPP) and the Patient Health Questionnaire-2 (PHQ-2) were also included.^19,20 Acute Stress Disorder was assessed at the time of initial admission using a validated 5-point PTSD criteria questionnaire (Primary Care PTSD Screen for DSM-5) at the time of injury. ²¹ Frailty scores were calculated through a 5-point modified frailty index (mFI-5) which incorporates congestive heart failure, diabetes, functionally dependent health status, hypertension requiring medication, and chronic obstructive pulmonary disease. ²² Distressed Community Index (DCI) was determined based on residential zip code. ²³

One year later, study subjects underwent follow-up phone screening, including a reiteration of the 5-item validated questionnaire for PTSD. ²¹ A 1 year follow-up was used to capture patients who have persistent PTSD symptoms because this time frame is typical for a postoperative follow-up time point. Respondents screened positive for probable PTSD if they answered “yes” to any 3 of the 5 PTSD questions. The follow-up survey also collected information regarding patients’ on-going relationship with the health care system, community support, and financial, relational, and housing stability throughout the previous year.

Data Analysis

Patients were binarily grouped based on PTSD screening results. Univariate analysis examined associations between injury-related, personal, and community factors and probable PTSD at 1 year (≥3 factors on PTSD-5). Differences were evaluated using χ², Fischer exact tests, t-tests, or Wilcoxon rank-sum tests, with a P-value < .05 considered significant. Several multivariable logistic regression models were then developed utilizing forward selection techniques, although nonsignificant parameters were incorporated if they exhibited synergistic effects on model performance. Regression models were assessed based on Akaike Information Criterion (AIC) and Area Under the Receiver Operator Characteristic (AUC).

After determination of the best fit model, a novel PTSD risk prediction model was designed by assigning weights to each parameter based on odds ratios, a technique widely used in the development of risk prediction algorithms to facilitate clinical implementation.^24-27 In the current study, weights were assigned to each parameter based on methods outlined by Charlson et al ²⁸ Specifically, parameters with odds ratios between 1.2 and 1.5 were assigned a weight of 1; those between 1.5 and 2.5 were assigned a weight of 2; ratios between 2.5 and 3.5 led to a weight of 3; and parameters with odds ratios of 6 or greater were assigned a weight of 6. Regression analysis was completed on a complete-case basis, with all incomplete cases excluded.

Additional univariate analyses were conducted to examine which factors at the time of injury were associated with loss to follow-up. Personal and injury-related variables were investigated between patients who responded to the follow-up vs those who did not. χ², Fischer exact tests, t-tests, and Wilcoxon rank-sum tests were used to indicate significant differences between groups (P-value < .05). Data analyses were performed with R version 3.6.1 (2019-07-05). Copyright © 2019 The R Foundation for Statistical Computing. No external funding was obtained to complete this study.

The Effects of PTSD on Trauma Recovery

Post-traumatic stress disorder in the aftermath of physical trauma has widespread negative effects—some physical, including in-hospital complications and even mortality. In addition, there are significant financial implications—with prolonged hospital lengths-of-stay and increased cost-of-care in trauma patients with PTSD. ¹⁸ Some effects are longer-lasting and are more difficult to quantify in the aftermath of trauma—including work status, disruption of personal relationships, financial difficulties, social dysfunction, and risk of recidivism. All of these negative effects place stress on the patient, society, and the health care system.

Post-traumatic stress disorder has been shown to contribute more toward general health at 6 months than physical function or injury severity. ⁸ Psychiatric impairment has been shown to affect the perception of pain and to prolong recovery. ⁶ Depression alone has been shown to be an independent risk factor for postoperative complications. ⁷ One study of patients with traumatic brain injuries noted higher mortality at 6 months unrelated to the primary diagnosis (including suicide, injury, and assault); mortality was substantially elevated in patients with a psychiatric or substance abuse co-morbidity. ²⁹ Patients with assault injuries and PTSD are more likely to die by suicide. ³⁰ Persistent physical and financial problems are more important in long-term trauma outcomes than the initial injury severity. ¹⁰ These may represent chronic stressors that exceed a patient’s ability to cope, making it more difficult for overall trauma recovery. The negative effect PTSD can have on physical recovery can be easily overlooked by surgeons who are often over-worked.

In addition to the physical consequences that psychiatric comorbidities have on recovery, many patients face social challenges. In the setting of organized or informal communal activities, 45% of trauma patients report social dysfunction at 6-12 months. ³¹ Persistent PTSD had an extremely close relationship with social dysfunction, with an odds ratio of 16.25. Psychiatric diagnosis also predicts trauma recidivism and unintentional injury. ³² Lastly, the undertreatment of post-traumatic psychiatric disorders has an overwhelming financial impact on the health care system with an estimated 80% increase in hospital LOS and cost-of-care for patients with stress disorders. ¹⁸

While the nuances of interpersonal violence are beyond the scope of this paper, it must be recognized that interpersonal violence is a well-recognized public health crisis. ³³ In some particularly violent communities, on-going cycles of disadvantage, decreased community resilience, and intergenerational trauma are likely related to undiagnosed psychological trauma. A systemic review by Charlson et al ³⁴ estimated that 68.6 million people in the United Nations are displaced by violence and conflict, the highest number since World War II. Roughly 20% of this post-traumatic population suffered from depression, PTSD, or anxiety. ³⁴

Limitations of Current Care Delivery Models

Our current trauma care delivery model focuses heavily on physical recovery with limited attention directed toward early psychological recuperation following everyday trauma. One study of motor vehicle accidents and work-related injuries found only 18 of 113 (16%) patients with probable PTSD or other psychological problems received treatment for those conditions. ³⁵ A similar study found that only 33% of trauma survivors with psychological illnesses following injury received mental health treatment, and of those with PTSD, only 47.7% received the appropriate treatment. ³⁶

Recognizing antecedents of PTSD during the initial hospitalization can be challenging, making the need for systematic screening desirable. The hospital is a “safe” place for patients that is detached from their normal everyday life and is focused on physical recovery. This may set the tone for the patient to focus solely on their physical recuperation only to find themselves dealing with difficult psychological sequela when attempting to transition back into society. Though often overlooked by surgical teams, early identification of PTSD risk using a screening tool may help prioritize patients’ psychologic rehabilitation alongside their physical rehabilitation after acute trauma.

The Ideal Population-Based Whole Person Trauma Recovery Model

The current care model has made significant advances in physical treatment following trauma, but it must expand to incorporate psychological recovery. Bisson et al showed that early cognitive behavioral therapy in trauma patients presenting to the Emergency Department significantly decreased symptoms of PTSD at 13 months. ¹⁷ System-wide health care reform, as suggested by the World Health Organization (WHO), that addresses both new onset and pre-existing mental health disorders should be foundational to a comprehensive approach to caring for trauma patients. ³⁷ Early supportive interventions, including both personalized psychological intervention, as well as those that engage families and communities are needed to address the complex needs diverse trauma populations.

Recent efforts have been made to address the need for mental health care in low resource settings. The WHO along with the UN High Commissioner for Refugees designed mhGAP Humanitarian Intervention Guide, ³⁸ which addresses assessment and management of moderate and severe mental disorders in non-specialized settings. The WHO has also recently published Problem Management Plus, which provides guidance on psychological intervention for people exposed to adversity. ³⁹ The Common Elements Treatment Approach (CETA) was developed to reduce the burden of mental health disorders globally in low resource settings by providing treatments for mental health disorders that can be implemented by non-mental health providers. ⁴⁰ Lastly, the WHO has developed Self-Help Plus (SH+) as an intervention for managing stress and coping with adversity in low resource settings. ⁴¹ All of these interventions are directed towards decreasing the burden of mental health disease in a cost-effective manner for low resource settings. Findings from this study confirm that patients in low resource settings are at higher risk of mental health problems. Future trauma care models should direct efforts toward recognizing populations at risk of developing mental health disorders and implementing targeted interventions such as those mentioned above. Studies implementing this risk screening tool and various mental health interventions are needed to better understand potential health outcomes and cost savings in various populations of trauma patients.

Strengths and Weaknesses

This risk assessment tool was developed from data collected for a cross-sectional convenience sample of orthopedic patients at an urban level 1 trauma center assessing a comprehensive battery focused on social determinants of health. The level 1 trauma center serves patients with significant diversity of culture, race, education, and finances, and this survey is a cross-section of the population. The findings may be limited due to the 1 year follow-up rate of 26%, though this rate is similar to other reports of trauma patients. The sample includes a large proportion of under-resourced patients, and generalizability to other patient populations is not known. Additionally, this study focused on risk factors for orthopedic trauma patients and future studies may want to expand to all trauma patients. Through this study, we identified risk factors to poor follow-up including age, male gender, lack of PCP, and high SOAPP scores, which can hopefully inform strategies to improve follow-up in the future. Considering many factors correlated with loss to follow-up are the same as those that place patients at risk for PTSD, it is possible that with improved follow-up rates that the patient population would be represented differently and would present different risk factors for PTSD. Consequently, the current version of this scoring system may fail to capture a group of patients at risk for chronic PTSD. Future studies are needed to validate our risk assessment model prospectively and in various geographical regions. Future studies should also focus on targeting specific interventions that will be effective treatment options for patients at risk of developing PTSD.