In this investigation, high levels of PTSD symptoms after critical illness requiring mechanical ventilation were most likely to occur in female patients and in patients treated with high doses of lorazepam, whereas PTSD symptoms were less likely to occur in older patients. Understanding these risk factors may facilitate preventive strategies and direct screening for symptoms of PTSD after critical illness. In this study, 14% of patients evaluated six months after discharge reported high levels of symptoms consistent with PTSD. This coincides with the existing literature that reports a prevalence of 10% to 30% [4
]. Despite occurring frequently, PTSD goes unrecognized in many patients. The current study confirms previous work showing that high levels of PTSD symptoms are associated with impaired quality of life [4
], underscoring the importance of diagnosing and treating this disorder in survivors of critical illness.
In this study, women were significantly more likely than men to have high levels of PTSD symptoms after critical illness. The association between PTSD and female gender has been reported previously [5
], but few studies have evaluated the significance of gender on the development of PTSD after critical illness. Several studies have demonstrated that women are more vulnerable to PTSD, even after controlling for differences in the type of trauma [5
], and a higher incidence of pre-existing anxiety and/or depression disorders is postulated to play some role in the difference in PTSD rates between the sexes [6
This study reveals a significant relationship between age and PTSD symptoms, with older patients being less likely to experience high levels of PTSD symptoms after critical illness. A non-linear relationship between age and PTSD symptoms was observed, but caution is appropriate in interpreting this finding because of the small number of younger patients studied and the results of previous research. For example, Scragg and colleagues [18
] evaluated 80 ICU patients for symptoms of PTSD and reported that scores on the screening instrument were inversely correlated with age (p
= 0.05). Rattray and colleagues [23
] similarly found that symptoms of anxiety (p
= 0.04) and avoidance (p
= 0.01) were inversely correlated with age 12 months after discharge in 80 ICU survivors. In the current study, older patients were significantly less likely than middle-aged patients to have high levels of PTSD symptoms. Several possible explanations for this relationship exist. Although each patient studied was mechanically ventilated, older patients are less likely to receive aggressive interventions that may predispose them to the development of PTSD [26
]. Additionally, given that older patients may have multiple comorbidities and a history of hospitalization, they may be less likely to view critical illness as a traumatic event.
We hypothesized that patients who experienced longer periods of delirium would be more likely to develop high levels of PTSD symptoms after critical illness, but the data do not support this hypothesis. Jones and colleagues [8
] have demonstrated that the recall of delusions rather than factual memories of the ICU experience is associated with the development of PTSD symptoms. Their study assessed 45 patients after ICU discharge and revealed that patients with delusional memories and no recall of factual events in the ICU were more likely to develop PTSD symptoms than those patients with factual memories (p
< 0.0001). These data [8
] suggest that periods of delirium, with associated delusions, may predispose patients to PTSD whereas periods of alertness, which allow for the consolidation of factual memories, may protect patients from developing PTSD-related symptoms after discharge. However, the association between days of delirium and PTSD symptoms in this study was not statistically significant (p
Cumulative lorazepam dose correlated with PTSD symptoms. Although Nelson and colleagues [27
] studied 24 survivors of acute respiratory distress syndrome and noted that days of sedation correlated with symptoms of both PTSD (p
= 0.006) and depression (p
= 0.007), the current investigation is the first to report an association between sedative dose and PTSD symptoms. However, it cannot be concluded from these analyses that lorazepam causes PTSD. The possibility exists that ICU patients who demonstrate symptoms of anxiety during their ICU stay are likely to receive higher sedative doses than those patients who are not anxious. Therefore, high lorazepam doses may identify those ICU patients with acute stress disorder, a known risk factor for PTSD [28
Although the administration of lorazepam may lead to more PTSD symptoms or alternatively may identify anxious ICU patients, the daily interruption of sedatives may facilitate periods of alertness and reduce the risk of PTSD. Kress and colleagues [7
] evaluated 32 patients who were randomly assigned to the daily interruption of sedatives or standard sedation to determine the long-term psychological effects of this intervention. The 13 patients who had been treated with daily interruption of sedation had better Impact of Events scores (11.2 versus 27.3, p
= 0.02) and a lower incidence of PTSD (0% versus 32%, p
= 0.06). Further study of the effect of the daily interruption of sedation on the development of PTSD is needed.
Limitations of the current study warrant comment. Because the PTSS-10 does not make a formal diagnosis of PTSD, the results of this study may not be generalizable to the clinical syndrome of PTSD. Also, the PTSS-10 does not assess for delusional memories. Data on the frequency of delusional memories, such as those provided by the ICU Memory tool [29
], would have allowed for more in-depth analysis regarding the relationship between delusional memories, delirium, and PTSD symptoms. There were a significant number of patients lost to follow-up. Analysis suggests that baseline and outcome characteristics were similar between those patients lost to follow-up and those evaluated at six months (Table ), but this does not rule out the possibility of selection bias. In fact, 'avoidance of activities prompting recall of traumatizing events' is a symptom of PTSD, and patients experiencing this symptom may have been less likely to return for follow-up testing. Thus, this study may underestimate the prevalence of PTSD after critical illness. Also, the findings regarding risk factors might have differed if all survivors had been evaluated. It was not systematically determined whether any patient sought psychiatric care prior to the six-month follow-up, and follow-up was limited to a single visit. Therefore, it is possible that some patients experienced PTSD symptoms prior to follow-up but that psychiatric treatment resulted in the resolution of such symptoms prior to testing at six months. Also, no evidence exists to define the ideal follow-up interval after which to screen for PTSD symptoms. Therefore, it is possible that screening after a shorter interval would have identified a higher number of patients with PTSD symptoms. Because of the non-elective nature of critical illness, it could not be prospectively confirmed that patients did not have PTSD prior to ICU admission. This diagnosis was not reported by family members and was not recorded in the medical record for the patients in this study. Finally, no data were collected regarding corticosteroid and beta-blocker administration, two possible confounders [30
]. This planned pilot investigation was limited by a small sample size, and a larger study to confirm these findings is warranted.