Incidence and Risk Factors for Intensive Care Unit-related Post-traumatic Stress Disorder in Veterans and Civilians

Abstract

Rationale: The incidence and risk factors of post-traumatic stress disorder (PTSD) related to the intensive care unit (ICU) experience have not been reported in a mixed veteran and civilian cohort.

Objectives: To describe the incidence and risk factors for ICU-related PTSD in veterans and civilians.

Methods: This is a prospective, observational, multicenter cohort enrolling adult survivors of critical illness after respiratory failure and/or shock from three Veterans Affairs and one civilian hospital. After classifying those with/without preexisting PTSD (i.e., PTSD before hospitalization), we then assessed all subjects for ICU-related PTSD at 3 and 12 months post hospitalization.

Measurements and main results: Of 255 survivors, 181 and 160 subjects were assessed for ICU-related PTSD at 3- and 12-month follow-up, respectively. A high probability of ICU-related PTSD was found in up to 10% of patients at either follow-up time point, whether assessed by PTSD Checklist Event-Specific Version (score ≥ 50) or item mapping using the Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV). In the multivariable regression, preexisting PTSD was independently associated with ICU-related PTSD at both 3 and 12 months (P < 0.001), as was preexisting depression (P < 0.03), but veteran status was not a consistent independent risk factor for ICU-related PTSD (3-month P = 0.01, 12-month P = 0.48).

Conclusions: This study found around 1 in 10 ICU survivors experienced ICU-related PTSD (i.e., PTSD anchored to their critical illness) in the year after hospitalization. Preexisting PTSD and depression were strongly associated with ICU-related PTSD.

Keywords: intensive care unit; post-traumatic stress disorder; veterans.

Figure 1.

Assessment of intensive care unit (ICU)-related post-traumatic stress disorder (PTSD). At 3 and 12 months after hospital discharge, we used the PTSD Checklist Event-Specific Version (PCL-S) anchored to the ICU experience as the traumatic event to evaluate survivors for new PTSD related to their ICU stay. We categorized patients as having a high probability of PTSD using two methods: a cut-off–based approach relying on PCL-S score of 50 or greater and an item-mapping approach based on Diagnostic and Statistical Manual of Mental Disorders-IV (DSM-IV) criteria (13, 22).

Figure 2.

Intensive care unit (ICU)-related post-traumatic stress disorder (PTSD) (i.e., PTSD symptoms related to the ICU experience) at 3- and 12-month follow-up. The box-and-whisker plots show the PTSD Checklist Event-Specific version (PCL-S) score related to the ICU experience at 3 months and 12 months separately for the civilian and veteran populations. The horizontal line within each box indicates the median PCL-S score, the upper and lower limits of the boxes indicate the 25th to 75th interquartile range, the ends of the vertical whiskers indicate 1.5 times the interquartile range, and the black dots indicate values outside these ranges. A score of 50 or greater on the PCL-S denotes high probability of ICU-related PTSD.

Figure 3.

Risk factors for intensive care unit–related post-traumatic stress disorder (PTSD) at 3- and 12-month follow-up. This forest plot denotes the association between risk factors of interest and the odds of having a higher PTSD Checklist Event-Specific version (PCL-S) score at 3 and 12 months. For each risk factor, the point estimate and 95% confidence interval (CI) are shown. Interpretative example: at both 3 and 12 months, preexisting PTSD was independently associated with higher odds of greater PCL-S scores (3-month odds ratio [OR], 7.7; 95% CI, 2.5–23.5; P < 0.001; 12-month OR, 10.7; 95% CI, 3.3–35.1; P < 0.001) as compared with a patient without preexisting PTSD. IQCODE = short Informant Questionnaire on Cognitive Decline in the Elderly; SOFA = modified Sequential Organ Failure Assessment score excluding the Glasgow Coma Scale score component. *Benzodiazepine and opiate doses were cube-root transformed to reduce the influence of extreme outliers.