Cognitive and psychosocial outcomes of mechanically ventilated intensive care patients with and without delirium

With improvements in critical care and declining intensive care unit (ICU) mortality, the number of ICU survivors is increasing. These survivors are frequently left with significant long-term complications [1]. Cognitive impairment is an important long-term complication and is associated with both a reduced quality of life and increased healthcare costs and caregiver needs, although the magnitude of the problem is uncertain [2]. In addition, poor mental health and functional disability including depression and post-traumatic stress disorder (PTSD) are common in ICU survivors [3].

Delirium is an acute organic brain dysfunction characterised by disturbances of attention and cognition with a fluctuating course as a direct consequence of an underlying medical condition [4]. It occurs in different healthcare settings [5], affecting between 15 and 20% of general hospital patients, and up to 80% of patients in an ICU [6, 7]. Delirium has been associated with long-term disability following non-ICU hospitalisations [8‐10] and with poor outcomes following ICU admission including prolonged length of stay, cognitive impairment after hospital discharge [11‐14], and increased odds of long-term disability in activities of daily living [15]. Although suggested otherwise in the past, delirium is not associated with short-term mortality in critically ill patients, except for an increase in 90-day mortality associated with the mixed delirium subtype [16‐18]. Defining the extent of the association between delirium and persistent cognitive impairments in critically ill patients has been identified as an important research priority due to the high prevalence of both conditions [19].

We hypothesised that the occurrence of delirium in ICU is associated with long-term effects on cognition and psychosocial function and with symptoms of PTSD [20].

A total of 103 patients were enrolled. The trial was stopped because of low recruitment rates. The main reasons for recruitment failure were staffing issues and unresolvable organisational problems at one recruitment site. The study flowchart is shown in Fig. 1. There were 26 patients lost to follow up, with 21 patients who were not contactable by telephone despite a minimum of 3 attempts, and 5 patients who withdrew from the study.

In this prospective long-term follow-up study of intensive care survivors, we found an association between the occurrence of delirium during ICU admission and cognitive impairment at ICU discharge, but not after 6 or 12 months. Cognitive impairment was prevalent and severe in a significant proportion of intensive care survivors, with 1 in 6 patients at ICU discharge and 1 in 9 patients after 12 months exhibiting symptoms of severe cognitive impairment consistent with dementia. We found the prevalence of cognitive impairment improved at 12 months compared to at ICU discharge, similar to a previous report [38]. However, this observation may be confounded by the possibility that early cognitive assessment reflects residual pain, the effects of analgesic and sedative drugs, and/or residual delirium [38, 39].

Interestingly, according to scores obtained from the caregiver, a much larger proportion—almost 1 in 4 patients—showed severe cognitive impairment after 12 months. The observation that patients assessed their cognition differently to their caregiver, usually rating themselves higher, suggests they may have had an unrealistic view of their cognitive state. This phenomenon, called ‘anosognosia’ (the unawareness of deficits), is common in cognitive impairment and dementia. Alternatively, their caregiver may have underestimated the patient’s true cognitive ability. Regardless, this observation underlines the importance of interviewing patients’ caregivers when assessing long-term cognitive outcomes. This observation is consistent with another study which showed that patients assessed their cognition differently to the ICU team whilst in the ICU and highlights that the link between cognition and ICU memories may not be not reliable [26].

In addition, we found post-traumatic stress symptoms were common after 6 and 12 months, and often severe. The occurrence of delirium during ICU admission was associated with PTSD symptoms after 12 months, independent of age, sex, and disease severity,

Our findings are different from those of a previous study, which did not show an association between ICU delirium and PTSD symptoms [40]. There are several important methodological differences that could provide possible explanations for this discrepancy. First, the studies used different tools for the detection of PTSD symptoms. Second, we compared the distribution of IES-R scores between the groups rather than simply assigning groups based on a fixed cut-off (PTSD present or absent). In our opinion, because the IES-R is a continuous variable, a shift in the distribution of scores is the most relevant outcome because higher scores are consistent with more symptoms, regardless of which cut-off is used to define PTSD. Third, the study by Svenningsen et al. used a shorter duration of follow-up (6 months). In our cohort, there was no statistically significant difference in PTSD symptom scores after 6 months. However, after 12 months the patients who had ICU delirium had higher IES-R scores than patients who never had ICU delirium. Finally, there was a difference in the use of physical restraints between the studies. Although we did not collect data on this variable, in both hospitals where our study was performed, the use of physical restraints for agitated patients is reportedly common and standard practice; physical restraints were not used in the study by Svenningsen et al. The use of physical restraints has been associated with a higher incidence of delirium, development of PTSD and recall of delusions [41, 42].

PTSD is likely a serious problem in survivors of critical illness, and could significantly impact on long-term health-related quality of life [43, 44]. Our study suggests an association between delirium during ICU admission and later development of symptoms of PTSD, but this observation needs to be interpreted with caution. The screening test employed does not represent an expert clinical diagnosis of PTSD (which requires a qualified mental health professional diagnostic assessment), but rather an indication that the patients met criteria for such a diagnosis based on the existing literature for the IES-R test. We dichotomised patients in groups based on whether they had an episode of delirium in ICU but did not assess the duration or burden of ICU delirium. The latter has been independently associated with functional long-term outcomes [15, 45].

Regardless, it is conceivable that ICU delirium as a marker of organic brain dysfunction may be a risk factor for the development of PTSD. For example, the development of acute PTSD-related symptoms in the first months after ICU admission appears to be related to recall and memories of delusions, whereas memories of real events during critical illness may give some protection from development of anxiety [46]. Other studies also suggest memories of frightening and/or psychotic experiences consistently predicted post-ICU PTSD [43]. Detection, prevention and early treatment of potentially modifiable factors, e.g. by offering psychological support in ICU or by providing patients with ICU diaries, may have the potential to prevent PTSD and its associated burden [47, 48].

Both the development of ICU delirium and post-ICU PTSD have been linked to the administration of benzodiazepines in ICU [43, 49, 50]. This may provide a potential link between the occurrence of these disturbances, although causality would be difficult to prove. We did not assess the type and dose of sedation used in our cohort. Importantly, the severity of critical illness was not associated with the development of long-term PTSD in our study, which is consistent with the literature [43]. We also did not find an association between age, delirium and PTSD symptoms. It has been suggested that whilst older age is an independent risk factor for the development of delirium in ICU [51‐53], older age may also provide a degree of protection against the development of PTSD in medically ill patients [54].

The strengths of our study include the employment of a set of standardised validated assessments of patients’ cognitive and psychosocial function, and the extended follow-up after ICU discharge, which enabled us to complete the long-term assessment of cognitive and psychosocial outcomes necessary to explore PTSD symptomatology [4, 55] and to allow PTSD symptoms to surface. In addition, by incorporating cognitive evaluation by the patients’ caregivers, we were able to obtain a more complete picture of the long-term cognitive function in these intensive care survivors.

Our study also has several limitations. First, as previously indicated, we were unable to enrol the sample needed to have sufficient power to detect expected differences in long-term cognitive function between patients with and without delirium. In addition, there was a poor retention rate at 12 months because of death and lost to follow up, further decreasing the power of our analysis. Consequently, the results of our study should be viewed as hypothesis-generating. Second, we were unable to test patients’ cognition before ICU admission and critical illness. We addressed baseline cognition before ICU admission by excluding patients whose cognition was known to be compromised before their admission, as well as patients from CALD background who were likely to, in times of stress, revert to their native language [56, 57]. We were also unable to test for symptoms of depression or anxiety before ICU admission. Third, we may have underestimated the prevalence of cognitive impairment because we did not use extensive neuropsychological testing [2]. Fourth, as mentioned earlier, we did not assess the duration (or burden) of ICU delirium, which has been independently associated with functional long-term outcomes. We also did not collect data on the cumulative doses of sedatives and opiates, which have been shown to have a potential impact of delirium recognition [25]. Finally, as with any observational study, we were not able to address the possibility of confounding by death or withdrawal, and/or exclude the possibility of bias due to unmeasured confounders.

In conclusion, in a cohort of mechanically ventilated ICU patients, both cognitive impairment and PTSD symptoms were common. Delirium during ICU admission was independently associated with short-term but not long-term cognitive function, and with long-term PTSD symptoms. Further studies are needed to determine if delirium directly affects the trajectory of development of PTSD and to identify potentially treatable and modifiable factors.