Association of intensive care unit delirium with sleep disturbance and functional disability after critical illness: an observational cohort study

Critical illness encompasses a diverse group of diseases that confer a high risk of both mortality and morbidity [1, 2]. Impairment and morbidity after critical illness are believed to be particularly high and likely include worsening of medical comorbidities, decreased quality of life, and emotional and physical pain for patients. Understanding the extent of poor functional outcomes and their predictors after intensive care unit (ICU) admission is therefore clinically meaningful.

Sleep disturbance has elicited attention as one important outcome of critical illness. Sleep is highly deranged in the ICU, where patients regularly experience poor sleep quality, sleep fragmentation, and reduced sleep efficiency [3‐8]. Abnormal sleep may predispose patients to adverse medical outcomes of critical illness [9‐11], as well as poor sleep following hospital discharge. A systematic review from our group found that the prevalence of self-reported sleep disturbance in post-ICU patients was 50–66.7% in the first month, and as high as 61% more than 6 months after discharge [12]. Similarly, polysomnography studies in post-ICU patients have shown abnormal sleep parameters and sleep architecture that normalize by 6-month follow-up [12‐16]. These studies, however, have found few ICU factors associated with post-ICU sleep disturbance; those identified to date include number of chronic diseases, ICU severity of illness, substantial hospital acute stress symptoms, and hospital opiate use [17‐19].

Another post-ICU outcome of importance is functional disability, in which physiologic insults from critical illness cause persistent disability in meeting the physical or social demands of the patient’s life [20, 21]. This could include deficits in basic activities of daily living (BADLs) such as eating or dressing, or in instrumental activities of daily living (IADLs) such as managing finances or taking medications. A systematic review of IADLs in critical illness survivors found that patients in 69% of studies had new or worse IADL dependencies after critical illness [22], and studies specifically in older adults have found that post-ICU disability is often highly prevalent [20]. Risk factors identified for disability in critical illness survivors are diverse and include older age, depression, ICU severity of illness, mechanical ventilation, and pre-ICU sensory impairment [22‐24].

ICU delirium is a candidate risk factor with a plausible link to post-ICU impairment such as sleep disturbance and disability. Delirium is highly prevalent in the ICU. A recent meta-analysis estimated the prevalence in critically ill patients to be 31.8% [9, 25]. The etiology of delirium likely includes contributions of chronic medical comorbidity and cognitive impairment, sensory impairment, sedative/hypnotic use, and potentially sleep disturbance [9, 25, 26]. Overall, ICU delirium is associated with increased ICU length of stay, longer mechanical ventilation duration, and increased mortality [9, 25].

To our knowledge, no studies have assessed the influence of ICU delirium on sleep outcomes after critical illness, and its effects on post-hospital disability have been equivocal [24, 27, 28]. Overall, given the importance of sleep and functional recovery after critical illness, and the significant effect of delirium on other ICU outcomes, our aim was to evaluate both (1) the prevalence of self-reported sleep disturbance and increased disability post-ICU, and (2) the relationship between ICU delirium and post-ICU sleep disturbance and disability. Our secondary aim was to examine whether other previously studied ICU factors (such as severity of illness, length of stay, and mechanical ventilation) were related to these post-ICU outcomes.

Our study in a cohort of MICU survivors indicates that post-discharge impairment in sleep and functional disability is highly prevalent, with possible links to MICU factors such as delirium.

Subjective sleep disturbance was identified in 63% of post-discharge patients at an average of 4- to 5-month follow-up. This finding is consistent with other studies of post-ICU patients, in which the prevalence of sleep disturbance ranges from 22 to 57% from 3 to 6 months after discharge [12]. Furthermore, after adjusting for other important covariates, increasing total days of MICU delirium were significantly associated with increased post-discharge sleep disturbance—the observed rate ratio (1.11) suggests that each additional day of MICU delirium raises the PSQI score by 11%. While there is a hypothesized relationship between delirium and sleep disruption within the MICU [26], to our knowledge this is the first study to demonstrate delirium as a factor associated with poor sleep after critical illness. Given that delirium is a syndrome of acute brain dysfunction associated with long-term cognitive impairment, one hypothesis is that a similar persistent brain dysfunction increases sleep disruption in critical illness survivors [9, 37]. Overall, our results link a prevalent MICU neurocognitive condition to sleep disturbance in post-ICU patients.

Notably, regression analysis also identified that older age was associated with improved sleep scores at follow-up. This finding differs from several studies in post-ICU patients which found that older age was associated with worse sleep, albeit using different sleep instruments in heterogeneous critically ill populations [38‐40]. Furthermore, in non-critically ill populations, age is consistently associated with decreased total sleep time and sleep efficiency, less time spent in deep sleep stages, and worse self-reported sleep [41]. Although older age has been associated with improved sleep quality in some studies [42], this finding may reflect a self-report bias, wherein older adults normalize conditions (in this case, worse sleep) to which they have been chronically accustomed.

This study also assessed functional status in critical illness survivors using a composite of BADLs and IADLs. Accounting for baseline disability just prior to MICU admission, 37% of patients reported worsening disability at follow-up. In our model, patients who were older and nearer to their time of discharge were more likely to show increased disability at follow-up. Furthermore, lower baseline disability scores were significantly associated with having increases in disability at follow-up. This latter result possibly reflects a ceiling effect, wherein patients who are less disabled at baseline have greater capacity for worsening of their post-discharge disability relative to more disabled patients.

While total days of MICU delirium were not significantly associated with increased functional disability at follow-up (p = 0.240), delirium as a categorical variable (i.e., the presence of ICU delirium for at least one day) showed a trend toward association with increases in disability at follow-up (p = 0.077). The prior literature to date has been equivocal in assessing delirium as a risk factor for disability after critical illness. In a 2017 systematic review of post-ICU IADL dependencies, two of four studies assessing delirium found a significant association between delirium and worse post-ICU IADL scores [22, 24, 27, 28]. Several studies have also been conflicting regarding the association between BADLs and ICU delirium [24, 27, 28, 43]. Given our modest sample size, and the trend toward significance when using a categorical indicator for delirium, our results suggest a potential association between delirium and change in disability.

Overall, post-ICU impairments such as sleep disturbance and functional disability have been hypothesized to contribute to the “post-ICU syndrome,” in which patients have decreased physiologic reserve following critical illness and are at higher risk of hospital readmission [44]. Furthermore, these poor patient-centered outcomes likely interact with post-ICU comorbidities such as depression, anxiety, and reduced quality of life [19, 45‐47]. In our sample, post-ICU patients experienced a high prevalence of sleep disturbance and increased disability—the former being significantly associated with ICU delirium and the latter being marginally associated. These data reinforce the importance of both delirium prevention and the need to identify the patients most at risk of poor outcomes after surviving critical illness.

There are several limitations to this study. First, a sizeable number of potentially eligible patients were unable to be contacted (24%) or survived through discharge but died before follow-up (22%). As a result, our sample may not represent the sickest ICU survivors who are most likely to experience mortality within the first several months after discharge. Second, without baseline or within-ICU data on patient sleep, we cannot make conclusions regarding changes in sleep parameters that occur after critical illness admission and furthermore cannot ascertain whether baseline or in-ICU sleep disturbance accounted for delirium positivity in these patients. For example, patients with profound in-ICU sleep disturbance leading to acute confusion or disorientation may have met criteria for delirium either by our in-person or chart-based review methods, which could confound the observed relationship between ICU delirium and long-term sleep disturbance. Third, we did not have data on cognitive function post-ICU, which may be an important confounder in our results given that delirium and sleep are potentially linked to this variable. Patients that completed the follow-up PSQI, however, were subjectively deemed to have sufficient cognitive ability to communicate in the interview. Finally, while our modest sample size may have precluded the detection of an association between delirium and post-discharge disability, it comprises a respectably large sample relative to comparable studies of survivors of critical illness.

Critical illness represents a severe physiologic stress with a high degree of mortality and morbidity. In a cohort of MICU survivors assessed in the year following discharge, poor sleep quality and increased disability were highly prevalent. After adjusting for important covariates, total days of MICU delirium were significantly associated with increased sleep disturbance, and the occurrence of delirium showed a trend toward association with increased disability. Overall, these results have important implications for identifying MICU patients that are likely to experience difficulty with sleep and function in the year following hospital discharge. MICU delirium is an especially important factor to consider in planning for post-discharge care of these patients.