Cognition in intensive care unit survivors:
correlation between the subjective and
objective components
Author: M. Cassee
Research group: Brain @ Risk
Division: Vital Functions
Department: Intensive Care Unit
Research name: Cognition & ICU: The virtual supermarket (adjusted*)
Supervisor: L. Kok
Location: University Medical Center Utrecht (UMC Utrecht)
Function Article: Thesis
Words: 5734
* Due to unforeseen circumstances the original design/method has been adjusted midway in the study. More information about this can be found in the prologue.
Abstract
Introduction Long-term cognitive impairment plays a major role in the post-intensive care syndrome (PICS). It affects 35 to 81% of the intensive care unit (ICU) survivors, can persist for years after discharge, and has major rehabilitation consequences. Since the relation between objective and subjective cognitive functioning could be of importance, this study aimed to determine whether and how subjective and objective cognitive domains correlate in an ICU population. Moreover, the prevalence of mild cognitive impairment (MCI) was measured. Methods Long-term cognitive impairment was assessed in 39 ICU survivors at three months, one year, three years, and five years after ICU discharge. Cognitive assessment included one subjective questionnaire (i.e. Cognitive Complaints-Participation [CoCo-P]) and two objective neuropsychological test batteries (i.e. Montreal Cognitive Assessment [MoCA] and Repeatable Battery for the Assessment of Neuropsychological Status [RBANS]). The correlation between comparable subjective and objective cognitive domains was tested. Moreover, the prevalence of MCI was determined with the MoCA in the different follow-up groups.
Results There were no significant correlations found between subjective and objective cognitive domains. Furthermore, the prevalence of MCI was between the 50 and 58.3% in the different follow up groups. No association was present between the follow-up duration and the prevalence of MCI (χ2 (df, n)=0.215 (3, 39), p=0.975).
Conclusion Subjective and objective cognitive domains do not seem to be correlated in cognitive impairment. Combined with the fairly high occurrence of MCI in ICU survivors, assessment of long-term cognitive impairment and potential therapies should be selected with care and tailor-made to individual complaints.
Keywords: Intensive care unit, long-term cognitive impairment, post-intensive
care syndrome, cognitive functioning, objective cognition, subjective cognition
Prologue
Adjustments
At the start of my internship, a supermarket-based virtual reality technique was available to assess cognitive impairment of intensive care unit survivors. Unfortunately, due to financial and lawful reasons this software could no longer be used in this study. As a suitable alternative, the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) was included in the study which is a comprehensive neuropsychological test battery. This test was conducted next to another test battery and questionnaires; the RBANS replaced the virtual reality task. Shortly after, the COVID-19 measures prohibited face-to-face RBANS assessment which resulted in too few RBANS results for a reliable use in the analysis. Therefore, the RBANS results will not be part of the primary analysis, but will be described for preliminary purposes. This also implies more use of previously collected data in order to answer the research questions posed in this thesis. Nevertheless, consensus about the abovementioned adjustments was received quickly between all involved parties. The changes to the original study design and the precautionary COVID-19 measures are not expected to negatively affect the quality of this thesis.
Cognitive impairment after Intensive Care Unit admission
Between 35% and 81% of the population surviving the intensive care unit (ICU) develops long-term complaints related to reduced neuropsychological functioning (Honarmand et al., 2020). This post-ICU cognitive impairment is characterized by a dysfunctional mental capacity and present in one or more cognitive domains (e.g., memory, attention, visuospatial abilities, language, orientation, executive functions, processing speed, sensorimotor control). The most regularly affected cognitive areas are attention, memory, and executive functions (Patel, Morandi, & Pandharipande, 2015). Executive functions can be divided into cognitive inhibition, working memory, and mental flexibility (Diamond, 2013). Moreover, cognitive impairment can be present for years after hospital discharge (Rawal, Yadav, & Kumar, 2017). Although the responsible underlying neurological mechanism has not yet been found, there are some important theories regarding the pathophysiology. One recognized possible explanation is global atrophy in the brain (Rengel, Hayhurst, Pandharipande, & Hughes, 2019). Furthermore, results of prior studies showed a decline in white matter integrity (Morandi et al., 2012). Lastly, biomarkers indicating inflammatory responses and alterations in blood-brain barrier permeability are associated with the development of cognitive impairment (Hughes et al., 2018). Although the mechanisms behind cognitive impairment development after ICU admission are not yet fully known, we do know that it is part of the post-intensive care syndrome (PICS) (Rawal, Yadav, & Kumar, 2017). PICS is characterized by deficiencies in three specific facets: cognition, physical- and psychological health. The most frequently observed form of reduced physical health is neuromuscular weakness, resulting in an reduced endurance, instability, and infirmity of the body. Psychological deterioration expresses itself by the development of mental conditions, such as post-traumatic stress disorder, anxiety and symptoms related to depression (Rawal, Yadav, & Kumar, 2017; American Psychiatric Association, 2013). In addition, also kin related to the individual suffering from PICS can be affected. The increased pressure on family members, friends, and caregivers of ICU survivors can result in the development of severe stress-related symptoms in family members. This form of the syndrome is called PICS-Family (Davidson, Jones, & Bienvenu, 2012).
Specifically, cognitive impairment as part of PICS can broadly interfere with a patient’s independence and recovery, and thereby affecting family members and caregivers as well. Adequate cognitive abilities are necessary to function both individually and socially. Daily activities such as personal care, planning, and grocery shopping, but also leisure activities and maintaining meaningful social relationships are potentially affected. Moreover, the number of ICU survivors in today’s society is rapidly expanding. This is mainly due to the increasing life expectancy among the population and the continuing improvement of ICU treatment (Ehlenbach et al., 2010; Wolters, Slooter, Van Der Kooi, & Van Dijk, 2013). As a result, more individuals are hospitalized for critical illness while the chances of surviving are increasing (Ehlenbach et al., 2010, Wolters et al., 2013). With the number of ICU survivors rising and potentially at risk for cognitive impairment, we decided to focus on a comprehensive assessment of their cognitive functioning.
Risk factors and impact of cognitive impairment
There are many risk factors associated with the development of long-term post-ICU cognitive impairment. These can be categorized into patient-related (non-modifiable) or hospital-related (modifiable) factors. Possibly the most important non-modifiable aspect is age, with older age increasing the risk of developing cognitive impairment. Nevertheless, impairment is seen in ICU survivors of all ages (Hopkins et al., 2017). Other non-modifiable factors associated with long-term cognitive impairment are cognitive impairment prior to admission, mental or physical problems prior to admission, level of education, and chronic diseases (Rawal et al., 2017; Hopkins et al., 2017; Sakusic & Rabinstein, 2018). As the name implies, hospital-related factors are processes and actions that took place during ICU admission. This includes the occurrence of delirium, hypotension, hypoxia, mechanical ventilation, and experiencing acute stress (Rawal et al., 2017; Hopkins et al., 2017; Sakusic & Rabinstein, 2018). Acute stress includes distressing intrusive, avoidance, and arousal symptoms. The wide variety of potential risk factors shows the complexity of cognitive impairment development.
As mentioned above, impaired cognitive functioning before hospitalization increases the risk of long-term cognitive decline after ICU admission. Unfortunately, no clear information can be found about the precise contribution of pre-existing cognitive impairment. Nevertheless, without prior cognitive problems present, cognitive impairment can also be newly acquired (Hopkins, Wade, & Jackson, 2017). Focusing on the duration, the difference between short and long-term impairment is of big importance. Cognitive impairment during or short after hospital discharge is more common than long-term cognitive impairment (Hopkins & Jackson, 2009). However, long-term cognitive impairment can be present months to years after hospital discharge and can result in lasting mild cognitive impairment (MCI) (Wolters et al., 2013). MCI is a neurocognitive disorder and a pre-stage of dementia. The difference between MCI and dementia lies in the severity of the cognitive impairment (American Psychiatric Association, 2013). The most unacceptable consequences of long-term MCI are the major limitations concerning someone’s ability to participate in society (Spreij, Sluiter, Gosselt, Visser-Melly, & Nijboer, 2019). As a result, cognitive impairment often comes with a decreased quality of life (Hopkins et al., 2005, Wolters et., 2013). The assessed ‘quality of life’ levels in the study of Hopkins et al. (2005) were health-, physical-, emotional-, and functional related. For as long as two years after discharge, quality of life scores were below the values of the control population. Furthermore, cognitive impairment strongly affects social healthcare expenses (Sakusic & Rabinstein, 2018). For caregivers cognitive impaired patients can be very costly in terms of time with possible loss of labor or other financial consequences as a result (Sakusic & Rabinstein, 2018). Together with a potentially frequent occurrence of cognitive impairment in ICU survivors, these consequences mark the wide variety and severity of implications.
Occurrence of cognitive impairment in ICU survivors
Although the occurrence of long-term post-ICU cognitive impairment has been assessed in quite a few studies, there still exists uncertainty about the exact prevalence. Prior studies use a
wide variety of cognitive assessment tools, either subjective or objective, differ in follow-up periods, and include patients of different ages. The wide range of results can possibly be partly attributed to all these variations (Honarmand et al., 2020; Wolters et al., 2013).
The BRAIN-ICU study can be considered as an influential study in ICU cognitive impairment prevalence. Cognitive functioning was assessed three and 12 months after discharge in 821 surgical- and medical ICU survivors (Pandharipande et al., 2013). The cognitive test battery showed that 40% of the ICU survivors lived with global cognitive problems three months after hospital discharge, of which 26% was classified as MCI. Similar proportions of cognitive complaints and MCI were found in the 12-months follow-up group: in 34% and 24% of the ICU survivors, respectively. However, higher rates were found in a systematic review that performed a meta-analysis including 19 studies with ICU survivors with follow-up groups varying from two months up to eight years. Cognitive impairment was reported in the broad range of 4% to 62% of the patients. In addition, 15 of these studies indicated MCI or worse in more than 10% of their patients (Wolters et al., 2013).
Similar fluctuating results were found in a more recent review including 41 prevalence studies (Honarmand et al., 2020). At three months, between 35% and 81% of the patients was diagnosed with cognitive impairment. Moreover, new clinical studies found results that are approximately similar to these of Pandharipande and colleagues. Marra et al. (2018) tested 406 ICU survivors three or 12 months after hospital discharge. From this group, 38% suffered from cognitive impairment at three months , whereas this was 33% for the 12 months follow-up. Lower values were found by Brück and colleagues (2019), who included a pool of 58 ICU survivors. 34% Lived with cognitive impairment three months after discharge. However, this prevalence of cognitive impairment was decreased to 18% at six months and 16% at 12 months. Although both studies assessed cognition with neuropsychological tests, the difference between the test batteries may account for the dissimilar results.
Concluded, measuring the prevalence of cognitive impairment in ICU survivors is complex and patient symptoms may vary over time. Nevertheless, cognitive impairment does occur in a significant part of the studied populations on every given time point, despite the differences between studies.
Innovative methods for assessing cognitive functioning in ICU survivors: a
narrative review
Recently, cognitive functioning has received a lot of attention from researchers. As a result, a large number of tools is available to assess cognitive functioning and/or impairment. This variety of methods can also be found in post-ICU cognitive impairment research, as shown by Honarmand et al. (2020). All included studies obtained results with either self-report subjective methods or by conducting an objective test(-battery). The subjective method is found to be less accurate in finding cognitive impairment than objective methods, but the objective method is more costly in terms of time, effort and expenses. There are multiple reasons why these
these instruments are composed of relative simple tasks and questions. The term ‘simple’ here only implies the instrument itself, since the cognitive condition of a patient can make the assessment and diagnoses more complicated. Another reason is the fact that these instruments cover most of the important cognitive domains. However, a major limitation of these ‘standard’ methods is the lack of ecological validity, since they all consist out of questions or tasks testing the mental capacity. Ecological validity concerns the correspondence and relevance between what is tested in an experimental setting and how this relates to reality (Schmuckler, 2001; Lewkowicz, 2001). This possibly causes differences between the assessed cognitive functions and cognitive functioning in reality. Therefore, the aim of this brief narrative review was to determine which innovative methods were recently (i.e. in the last 15 years) developed to realize a more realistic assessment of cognitive functioning in ICU survivors (Brück et al., 2019; Freund & Kasten, 2012; Honarmand et al., 2020).
The process of the systematic literature selection can be found in Figure 1 as a flow diagram (p. 16) in appendix A. This flow diagram also contains the precise numbers, exclusion and inclusion criteria. Furthermore, appendix A includes the search string (p. 17) used for the literature search in the database of PubMed. In short, the selection process led from 195 to one article and was executed in two selection rounds, based on title/abstract (n=27) and based on full article (n=1), respectively. The single matching article was the only one to 1) use an innovative cognitive assessment method 2) measure long-term cognitive impairment 3) assess adult ICU survivors. A method was considered innovative if it could not be categorized as a ‘standard’ test, test battery or subjective self-report tool.
The articles included based on title and abstract most frequently used the following classical methods: Montreal Cognitive Assessment (4; MoCA), Mini-Mental State Examination (4; MMSE), Telephone Interview for Cognitive Status (3; TICS), Repeatable Battery for the Assessment of Neuropsychological Status (3; RBANS) and the Informant Questionnaire On Cognitive Decline in Elderly (3; IQCODE). When selecting for innovative methods, only one study was included. This study succeeded to develop a robot that assesses PICS in ICU survivors, including cognition (Wood et al. 2018). The aim was to test the feasibility of using this robotic technique for automatic data collection and analysis. To realise this, cognitive functioning, inter alia, was assessed in a total of 33 ICU survivors three and 12 months after hospital discharge. Cognitive functioning was measured by conducting the RBANS and by using the robotic technology, called the Kinesiological Instrument for Normal and Altered Reaching Movement (KINARM). The robot conducts several tasks (sensorimotor, visuospatial, executive functioning and memory) in which the individual participates by holding on and moving two handles in a horizontal axis. Both the task and the position of the hands is displayed by using virtual reality technique. Because data is automatically collected and analyzed, data-experts are unnecessary. Results showed that in most domains the robotic method was able to detect cognitive impairment equal to the RBANS. However, the KINARM failed to detect memory impairment as accurately as the RBANS. In my opinion, this method has a lot of potential to assess cognitive functioning with a higher ecological validity. Currently, the assessment is mainly focussed on sensorimotor functioning and less on other important cognitive domains.
However, for the reliability of detecting cognitive impairment, the tasks conducted by the KINARM should focus more on other cognitive domains, such as memory and attention. Another important aspect is the applicability in this group of patients. Since the group of ICU survivors contains mainly elderly, the usability of a high-tech robotic that relies on movements for gathering data is disputable. In addition, this method requires that patients are able and motivated to travel to a specific location, possibly decreasing the number of study participants (Wood et al. 2018). Hence, before this method could be successfully used in cognitive research, it has to be more expanded, accessible and easier to use, especially for older patients.
Besides this article, two other worth mentioning articles were found. Unfortunately, Van Beusekom and colleagues did not include cognition in their study focussed on PICS. Nevertheless, their method should be briefly discussed. They developed and used a web-based questionnaire software developed to detect and assess PICS in ICU survivors (Van Beusekom, Bakhshi-Raiez, De Keizer, Dongelmans & Van Der Schaaf, 2018). Comparing this with face-to-face research on location with experts, this online automated processing software is both more efficient and easier to participate in. As they did, multiple questionnaires can be used for assessing different symptoms and aspects of PICS. Although it did not become clear for me why no cognition related questionnaire was included in this study, it could easily be added. In my opinion, this method has potential for a more efficient use of cognition-questionnaires in the future, but it could be challenging that the patient group mainly exists of older adults that are not always used to modern technology. Besides, this method has a low ecological validity according to my point of view. The second excluded article did not conduct a study, but discussed usage of smartphone data to potentially assess cognitive functioning and impairment (Akeret et al., 2018). Since smartphones are excessively used in contemporary society, valuable information could be obtained for cognitive functioning ‘at baseline’ and possible changes to screen for impairment. Besides, it could give insight in progression of symptoms. Although this idea is very innovative and provides data closely related to someone’s cognitive abilities in real life, it is also a very novel method that has to be further explored (e.g. in terms of validity and privacy).
In short, this narrative review describes recent innovative cognitive assessment methods with a more optimal ecological validity compared to the standard methods. Currently, there is a wide variety of methods to measure cognitive functioning; many of these are questionnaire-based or standard tests. Innovative methods developed to measure cognitive functioning are sparsely applied in post-ICU cognitive impairment research, since only one article matched all inclusion criteria. This study assessed cognitive functioning with robotic technology and virtual reality. The number of full text studies that were included was low; this might be due to comorbities and lesser mobility of ICU survivors. This might make assessment of cognitive functioning difficult using the novel methods. Therefore, testing innovative cognitive measuring methods in other (i.e. younger and healthier) populations is easier to realize. Nevertheless, the greater ecological validity of these novel methods makes their use, with regard to cognitive functioning post-ICU, vital for future research in this field.
Subjective and objective cognitive functioning
As mentioned, almost all current cognitive research is based on either objective or subjective data gathering. Subjective methods categorically are questionnaires or other self-report techniques, while objective methods include neuropsychological performance-based tests. Prior studies that analyzed the correlation between the results from these two test methods have shown contradictory results (Uiterwijk et al., 2014; Maaijwee et al., 2014). As also stated by Uiterwijk et al. (2014), it is therefore still uncertain whether objective and subjective cognitive functions correlate with each other. A recent study did found a significant correlation between the psychological state and subjective psychometric functioning (Brück et al., 2019), suggesting that someone’s psychological condition affects self-rated cognitive functioning. However, the results showed only weak/non-significant correlations between objective and subjective cognitive results in ICU survivors. They obtained data with one self-report questionnaire and four domain specific tests from a more elaborated neuropsychological test battery. The correlations were calculated between the overall subjective score and the multiple domain-specific objective scores. Brück and colleagues eventually interpreted the results as support that the two aspects of cognition do not represent an identical function. Concluded, therapy against post-ICU cognitive impairment should be approached consciously and should take the different components of cognition into account (Brück et al., 2019).
The abovementioned suggests that results obtained from subjective self-estimated tests provide dissimilar information about cognitive functioning compared to objective tests. If this lack of correlation in ICU survivors proves to be persistent, this supports the interpretation done in the study of Brück et al. (2019). To our knowledge, the study from Brück et al. (2019) is the first and only study that investigated the correlation between subjective and objective cognition in ICU survivors. For this reason it is of big importance that this relationship is further investigated. Increased knowledge about the relation of these cognitive components makes it possible to adjust cognitive assessment methods and future treatment to maximize the effectiveness and validity. Therefore, the goal of this study was to determine how and whether subjective cognitive complaints are in line with objective cognitive functioning results in ICU survivors. In other words, how do subjective and objective cognitive functioning in ICU survivors correlate with each other?
To be able to answer this question, correlations between subjective and objective cognitive outcome measures were analyzed. This was done for global cognition and domain specific cognitive functions in ICU survivors from multiple long-term follow-up groups. The correlations were performed between subjective and objective domains. Based on the results from the abovementioned study by Brück et al. (2019), it was expected that only weak non-significant correlations will be found. The Cognitive Complaints-Participation (CoCo-P) is used for the subjective measurement, while the Montreal Cognitive Assessment (MoCA) was assessed for the objective cognitive results. Furthermore, this study looked into the prevalence of cognitive
impairment (measured by the MoCA) among the different follow-up groups of patients and analyzed whether there was a significant difference between groups.
Methods
Patient selection
All approached individuals have experienced tertiary care ICU admission at the University Medical Center of Utrecht. A selection of eligible patients was created every two months. Based on the time passed since their ICU discharge, they were classified in follow-up groups: three months, one year, three years, or five years at the moment of testing. To be selected, the individuals had to meet the following criteria: age > 18 years, speak Dutch fluently and ICU duration ≥ 48 hours. The patients were excluded from this selection if they were hospitalized for a neurological disorder or a physical neurological trauma including post-anoxic encephalopathy, brain injuries and cerebrovascular accidents. Moreover, patients with diagnosed epilepsy were excluded. A total of 243 individuals have been informed by an informative letter and invited to participate. Eventually, cognitive functioning was tested in 41 individuals (17%) between September 2019 and March 2020. The other individuals did not respond (16%) or had specific reasons to reject participation, such as physical/medical (10%), emotional (9%), travel (9%) or time-related (5%) motives.
Testing procedure
Subjective cognitive functioning was assessed with a self-report questionnaire, the Cognitive Complaints-Participation (CoCo-P). It was send together with the letter and the informed consent, making it possible to fill it in at home. The CoCo-P was originally created to detect cognitive impairment in patients who suffer from acquired brain injury, but it was found that it could also be used in other groups of patients (Spreij, Sluiter, Gosselt, Visser-Meily & Nijboer, 2019). It consists of statements all focusing on daily life activities and aims to measure someone’s ability to function in the society. They can be answered with ‘independent without trouble’, ‘independent with trouble’, ‘with help’, ‘not possible’ and ‘not applicable’, and scored from zero to four respectively. Higher scores indicate worse cognitive functioning. The covered domains include memory, attention, executive functioning, and global cognitive functioning, equal to the total score. Results from a control group showed a median of zero on all domains, and an overall median (0-100) of 0.95 (Spreij et al., 2019).
The objective tests took place in a quiet room in the University Medical Center of Utrecht (UMCU). The process consisted out of two validated neuropsychological test batteries: the Montreal Cognitive Assessment (MoCA) and the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS) (Randolph, Tierney, Mohr & Chase, 1998; Nasreddine et al., 2005).
The MoCA is a cognitive screening test consisting of multiple small tasks. It examines the domains short-term and working memory, attention, executive functioning and overall global cognition (total score), among other domains (Nasreddine et al., 2005). The global cognition score is the sum of points (range 0-30) scored on all tasks, with a higher sum indicating better
performance. It is education corrected, since a point is added to the total score in case an individual has a total educative period of 12 years or less.
The RBANS is a more extensive 30 minute cognitive assessment tool measuring global cognitive functioning, short- and long-term memory, attention, and other domains (Randolph, Tierney, Mohr & Chase, 1998). For all domains, the score gets converted to an age corrected index score. At the end, those are summed up and modified to a total scale index score, describing the global cognitive functioning. Scoring ranges from 40 to 160 for all indexes with higher scores representing better results (Randolph, Tierney, Mohr & Chase, 1998). It was initially developed to detect mild forms of dementia, but is often used for testing MCI, as done by the influential study of Pandharipande et al. (2013) (Randolph, Tierney, Mohr & Chase, 1998).
Statistical analysis
The analysis was executed in SPSS (version 25.0.0.2). The descriptive results of the three cognitive assessment tools are reported as median and interquartile range. This implies that for every tool four domain specific descriptive results have been calculated. The assumption of normality was violated by almost all domain-specific results of the CoCo-P, MoCA, and RBANS. Since some scores were positively skewed, while others were negatively skewed, data transformation was not possible. Hence, analyses were performed with the non-parametric Spearman’s correlation test.
Because the aim of the study was to determine the correlation between subjective and objective cognition, these domains have been tested separately for correlations. In al cognitive tools, the total score represented global cognitive functioning. For the analysis between scores obtained from the CoCo-P and the MoCA, the subjective and objective domains of memory, attention, executive functioning, and global functioning have been tested apart from each other. In the other examination between the CoCo-P and the RBANS, the subjective domain memory has been analyzed with short-term memory and long term memory, subjective attention with objective attention and subjective global functioning with objective global functioning. The follow-up group categorization was not taken into account in the correlation analysis, only when determining the prevalence of MCI. Unfortunately, due to protective measures taken against the COVID-19 pandemic, the RBANS was only conducted in a small number of patients (n=6). Therefore, the RBANS test results are not included in the primary analysis and will be considered as inconclusive. However, the objective RBANS results have been analyzed in a similar way as the MoCA results for preliminary purposes.
Finally, the prevalence of MCI in the ICU survivors was calculated according to the MoCA. Following the guidelines of the MoCA, an individual was diagnosed with MCI if the overall score was ≤ 25 (Nasreddine et al., 2005). The prevalence has been determined for all follow-up groups separated and for all patients taken together, ignoring the group categorization. To determine whether the variables ‘presence/absence of MCI’ and ‘follow-up group’ were dependent of each other, the Chi-Square test was performed in a 4x2 contingency table.
Results
Included patients
From the 41 patients that completed the CoCo-P and MoCA, two were excluded prior to the analysis. One patient appeared to be diagnosed with epilepsy and experienced post-anoxic encephalopathy, while the other patient was excluded because of a later ICU admission in another hospital. A flowchart of the selection procedure and patient in-/exclusion can be found in Figure 2. As a result of these final exclusions, both the correlation and prevalence analyses are performed with 39 patients included (male=26, median age (IQR)= 64 (56-72)). All demographic characteristics of the included patient group can be found in Table 1.
Descriptive results and the subjective-objective cognition correlation
The subjective cognition scores derived from the CoCo-P show that overall patients are, according to their own experiences and opinion, mostly affected in the domain of attention (median score (IQR)=13 (0-24)). The activities related to the domains memory (5 (0-17)), executive functioning (4 (0-21)) and global functioning (8 (1-19)) are experienced as less problematic, since these overall subjective scores are considerably lower. The objective results were derived from the MoCA and the RBANS. For the MoCA, the median total/global score was 25 (23-26). Importantly, the scores of memory (3 (2-4)), attention (6 (5-6)), executive functioning (4 (3-5)) and global functioning in the MoCA cannot be compared to each other, as is the case in the CoCo-P, due to the unequally distributed performance points among the domains. The outcome of the correlation analyses between these cognitive tools showed that none of the subjective domains was significantly correlated with the objective domains. However, global cognitive functioning/total score was close to significance with a negative correlation (rs (p)= -0.308 (0.057)). The values between memory (-0.122 (0.458)), attention
(0.049 (0.769)) and executive functioning (-0.219 (0.180)) are less close to significance and show a low correlation score. Scatterplots of the domain specific scores from the CoCo-P and MoCA are shown in Figure 3, 4, 5, and 6 in the same order as they are described above.
Focussing on the RBANS, the median and IQR scores of short-term (103 (96-112)) and delayed memory (102 (100-103)) are very close to each other, while these values for the domains attention (97 (82-103)) and global functioning (96 (87-107)) are slightly lower. The correlation scores between the CoCo-P and RBANS are in line with those between the CoCo-P and MoCA, since no correlation came close to or reached significant values. All median scores derived with the three cognitive tools can be found in Table 2, as well as the correlation coefficient scores and the corresponding p-values.
Prevalence of mild cognitive impairment
The prevalence of MCI is shown in Table 3. It was calculated for the different follow-up time groups and for all groups taken together. The prevalence of MCI was at least 50%, and the highest prevalence was found in the three year follow-up group (frequency (%)= 7 (58.3)),
(50.0)) and one year group (5 (50.0)) showed little difference. In fact, the Chi-Square test showed no significant association when the prevalence of MCI was assessed for the different follow-up time groups (χ2 (df, n)=0.215 (3, 39), p=0.975).
Table 1. Demographic characteristics of the study population Characteristics Patients (n=39) Age, years Median (IQR) Gender, n (%) Male Female Time to follow-up, n (%) three months one year three years five years
Highest completed education, n (%) 1. Elementary school
2. Lower than ‘More advanced Primary Education’ 3. Lower general secondary education
4. Senior general secondary education / Pre-university 5. Higher professional education / University education. Admission duration, hours, median (IQR)
64 (56-72) 26 (66.6) 13 (33.3) 8 (20.5) 10 (25.6) 12 (30.8) 9 (23.1) 1 (2.6) 8 (20.5) 15 (38.5) 2 (5.1) 11 (28.2) 132 (72-228)
All values are reported as median with interquartile range or as frequency with percentage. Not all sums of percentages reach a total of 100 due to either missing data or rounding of numbers. Duration of admission was rounded to 12 hours. Abbreviations: IQR=interquartile range. Additional description education:
1. Elementary school and additional education < two years; 3. Lower general secondary education / intermediate technical school / pre-vocational secondary education
Table 2. Descriptive results of variables and correlation scores between cognitive domains Cognitive domains
Overall test scores CoCo-P, median (IQR)
Memory Attention
Executive functioning Global/Total
MoCA, median (IQR) Memory
Attention
Executive functioning Global/Total
RBANS, median (IQR) Short-term memory Delayed memory Attention Global/Total 5 (0-17) 13 (0-24) 4 (0-21) 8 (1-19) 3 (2-4) 6 (5-6) 4 (3-5) 25 (23-26) 103 (96-112) 102 (100-103) 97 (82-103) 96 (87-107) Combined analyzed domains
Non-parametric correlation scores, rs (p)
Subjective domain Objective domain CoCo-P Memory Attention Executive functioning Global CoCo-P Memory Memory Attention Global MoCA Memory Attention Executive functioning Global RBANS
Short term memory Long term memory Attention Global -0.122 (0.458) 0.049 (0.769) -0.219 (0.180) -0.308 (0.057) 0.000 (1.000) 0.424 (0.402) -0.600 (0.208) -0.223 (0.671)
All values are reported as median with interquartile range or as spearman’s rho with two-tailed significance. The upper half shows the descriptive results of all variables, while the lower half shows the correlation results between the similar domains of the tests. The correlation score of 0,000 can be explained because all patients that also completed the RBANS scored 0 for memory in the P Abbreviations: IQR=interquartile range, CoCo-P=Cognitive Complaints-Participation, MoCA=Montreal Cognitive Assessment, RBANS=Repeatable battery for the Assessment of Neuropsychological Status, rs
0 10 20 30 40 50 17 19 21 23 25 27 29 31
CoCo-P global functioning score
M oC a gl ob al f un ct io ni ng s co re
Figure 3. A scatterplot of global cognitive functioning scores obtained with the Montreal Cognitive Assessment (Y-axis) and the Cognitive Complaints-Participation (X-axis). The blue dots indicate the scores assigned to the patients. The red dashed line is the trendline and shows a potential relation between scores of the two cognitive tools. Meaning of abbreviations:
MoCA=Montreal Cognitive Assessment, CoCo-P=Cognitive Complaints-Participation
Figure 4. A scatterplot of the memory scores obtained with the Montreal Cognitive Assessment (Y-axis) and the Cognitive Complaints-Participation (X-axis). The blue dots indicate the scores assigned to the patients. The red dashed line is the trendline and shows a potential relation between scores of the two cognitive tools. Abbreviations: MoCA=Montreal Cognitive Assessment, CoCo-P=Cognitive Complaints-Participation
0 10 20 30 40 50 60 70 0 1 2 3 4 5 6 7
CoCo-P memory score
M oC a m em or y sc or e
Figure 5. A scatterplot of the attention scores obtained with the Montreal Cognitive Assessment (Y-axis) and the Cognitive Complaints-Participation (X-axis). The blue dots indicate the scores assigned to the patients. The red dashed line is the trendline and shows a potential relation between scores of the two cognitive tools. Abbreviations: MoCA=Montreal Cognitive Assessment, CoCo-P=Cognitive Complaints-Participation
Figure 6. A scatterplot of the executive functioning scores obtained with the Montreal Cognitive Assessment (Y-axis) and the Cognitive Complaints-Participation (X-axis). The blue dots indicate the scores assigned to the patients. The red dashed line is the trendline and shows a potential relation between scores of the two cognitive tools. Abbreviations: MoCA=Montreal Cognitive Assessment, CoCo-P=Cognitive Complaints-Participation
0 10 20 30 40 50 60 70 0 1 2 3 4 5 6 7
CoCo-P attention score
M oC a at te nt io n sc or e 0 10 20 30 40 50 0 1 2 3 4 5 6 7
CoCo-P executive functioning score
M oC a ex ec ut iv e fu nc ti on in g sc or e
Table 3. Prevalence and significance of mild cognitive impairment in different follow-up groups Measured value Time to follow-up, n (%) three months one year three years five years All patients, n (%)
Difference between groups, χ2(df, n), p
4 (50.0) 5 (50.0) 7 (58.3) 5 (55.6) 21 (53.8) 0.215 (3, 39), 0.975 All values are reported as the frequencies and percentages of mild cognitive impairment in a specific follow-up group or as the result of the Chi-Square test. The frequencies show the number of patients that scored lower or equal to 25 on global cognition at the Montreal Cognitive Assessment. Abbreviations: χ2=Chi-Square statistic value, df=degrees of freedom,
Discussion
The goal of this study was to determine whether subjective cognitive complaints correlate with objective cognitive functioning in ICU survivors. Because cognitive impairment is one of the most determining components of the post-intensive care syndrome (PICS), this research focused on long-term cognitive impairment in ICU survivors. Correlation analyses were performed between subjective and objective cognitive domains. These domains were memory, attention, executive functioning, and global cognitive functioning. Furthermore, the prevalence of MCI was determined for different follow-up time groups.
Most of the correlation coefficient scores between the subjective and objective cognitive domains were negative. Since quality scoring on the cognitive tools was reversed between the subjective (i.e. higher scores represent worse performance) and the objective tools (i.e. higher scores represent better performance), these findings may indicate a relationship. However, only the correlation between the MoCA and the CoCo-P with regard to global cognitive functioning showed a trend towards statistical significance (rs (p) = -0.308 (0.057)).
This indicates that these cognitive assessment methods can be considered as non-interchangeable, as also stated by Brück et al. (2019) and Yoon, Lee & Shin (2017). Moreover, it could mean that subjective and objective cognitive functioning are not clearly statistically related, and that these are in fact different components of cognition (Byrne, Coetzer & Addy, 2017). However, this implication remains hypothetical. Nevertheless, statistical significance might not equal clinical relevance (Brück et al., 2019): a recent meta-analyses that included 50 studies did find a weak association between the subjective and objective cognitive functions (Burmester, Leathern & Merrick, 2016).
Our results are in line with the results found by Brück and colleagues (2019) and suggest that the subjective and objective assessment tools do not measure comparable functions in cognition. The absence of this correlation is thought to be a result of the subjective performance being affected by an individual’s mental/psychological condition (Brück et al., 2019; Byrne, Coetzer & Addy, 2017; Yoon, Lee & Shin, 2017; Burmester, Leathern & Merrick, 2016). In addition, studies found significant correlations between factors like depression, fatigue, and anxiety, and subjective cognitive functioning. Another explanation could be the dissimilarity in assessing cognition in real life (Brück et al., 2019). While most subjective questionnaires are based on reality, objective tests consist more out of abstract tasks. This contrast may be responsible for measuring different cognition domains, besides subjective or objective performance. A more radical explanation may be that the cognitive components do not functionally affect each other: when one component is impaired, the other component could function properly. However, this is a simplified idea and further research is needed to support or reject this hypothesis.
MCI was prevalent in 50 to 58% of the ICU survivors, with the highest prevalence three years after discharge from the ICU. Moreover, no significant association was found between the follow-up time-points. Compared to other literature, this prevalence is relatively high (Honarmand et al., 2020; Pandharipande et al., 2013; Marra et al., 2018; Brück et al., 2019).
This discrepancy demonstrates the existing uncertainty regarding the number of ICU survivors affected by MCI. Based on the findings of Honarmand and colleagues (2020), determining MCI prevalence with a more comprehensive test battery could result in even higher prevalence. In our study, the prevalence of MCI was higher in the groups with a longer time to follow-up. This is contradictory to what was found in a prior review (Honarmand et al., 2020). Since the assessed patients were admitted to the ICU for a wide variety of reasons, the reason remains difficult to elucidate. However, the number of patients in our different follow-up time groups might have been too little to obtain reliable data.
The outcome of this study indicates that the subjective and objective assessment tools do not reflect similar cognitive functions. This implies that the different cognitive assessment methods are non-interchangeable. Hence, cognitive assessment and therapies of long-term cognitive impairment should be approached and selected with care: the chosen methods should fit the experienced signs and complaints. The reported prevalence of MCI in this study confirms the necessity of these adequate assessments and therapies, since the occurrence of cognitive impairment may be higher than initially expected. The CoCo-P (subjective test), MoCA and RBANS (objective tests) have proved to be easily accessible and practically feasible to use in ICU survivors. However, whether subjective functioning is independent of mental health and objective assessment provides a reliable view of cognition in reality, remains uncertain.
Strengths and limitations
Compared to existing literature, the current study is of value because of multiple aspects. First of all, comparable correlation analyses was executed using different neuropsychological test batteries. We showed that the CoCo-P, MoCA, and RBANS have great potential to comprehensively assess cognitive impairment in ICU survivors, both on global and domain-specific levels. Furthermore, our included patients varied in age and ICU admission reasons. This made our results well-generalizable with regard to both the prevalence of MCI, and the relationship between subjective and objective cognitive domains.
However, we also encountered limitations during this study. Firstly, a selection bias might have influenced the participation rate of de ICU survivors. A possible solution in future research would be to assess patients at home. This method has been proven feasible in the past and results were reliable (Rentz et al., 2016; Chodosh et al., 2008). Secondly, some patients have been admitted to the ICU multiple times which might have influenced our results. Furthermore, we had no information about the cognitive condition of the patients at the moment of ICU admission. It is possible that patients included in this study already experienced cognitive decline, cognitive problems, or MCI before their hospitalization, which could have affected their test results. Lastly, the reliability of the MCI prevalence is potentially low in this study, because of the small number of patients included.
Our results show the importance for future research of (long-term) cognitive impairment in ICU survivors. It is crucial for cognitive impairment recognition and treatment that more knowledge about subjective and objective domains of cognition is generated. Although a follow-up study should include more patients, more extensive use of the RBANS could be of added value. Furthermore, modern cognitive tools could improve their ecological validity to achieve a more realistic and thereby reliable assessment of cognitive functioning and impairment.
Concluding, this study indicates that subjective and objective cognitive assessment tools measure dissimilar functions of cognition, based on cognitive performance in ICU survivors. Therefore, the method for determining cognitive performance should be selected with care. Possibly, these components are independent and unrelated to each other, but more research is required before this can be concluded. Furthermore, the prevalence of long-term MCI might be higher than reported in many studies, underlining the importance of future research. More knowledge about cognitive impairment after ICU admission is necessary to determine the occurrence, comprehensively assess cognitive impairment and thereby lead to successful therapies. Especially with the current COVID-19 pandemic in mind, this is the moment to intensify research into PICS and post-ICU cognitive impairment in order to minimalize the consequences for patients and society (Stam, Stucki & Bickenbach, 2020).
References
Akeret, K., Vasella, F., Geisseler, O., Dannecker, N., Ghosh, A., Brugger, P., ... & Stienen, M. N. (2018). Time to be “smart”—Opportunities Arising From Smartphone-Based Behavioral Analysis in Daily Patient Care. Frontiers in behavioral neuroscience, 12, 303.
American Psychiatric Association. (2013). Diagnostic and statistical manual of mental disorders (DSM-5®). American Psychiatric Pub.
Brück, E., Larsson, J. W., Lasselin, J., Bottai, M., Hirvikoski, T., Sundman, E., ... & Olofsson, P. S. (2019). Lack of clinically relevant correlation between subjective and objective cognitive function in ICU survivors: a prospective 12-month follow-up study. Critical Care, 23(1), 253.
Burmester, B., Leathem, J., & Merrick, P. (2016). Subjective cognitive complaints and objective cognitive function in aging: a systematic review and meta-analysis of recent cross-sectional findings. Neuropsychology Review, 26(4), 376-393.
Byrne, C., Coetzer, R., & Addy, K. (2017). Investigating the discrepancy between subjective and objective cognitive impairment following acquired brain injury: the role of psychological affect. NeuroRehabilitation, 41(2), 501-512.
Chodosh, J., Edelen, M. O., Buchanan, J. L., Yosef, J. A., Ouslander, J. G., Berlowitz, D. R., ... & Saliba, D. (2008). Nursing home assessment of cognitive impairment: development and testing of a brief instrument of mental status. Journal of the American Geriatrics Society, 56(11), 2069-2075.
Davidson, J. E., Jones, C., & Bienvenu, O. J. (2012). Family response to critical illness: Postintensive care syndrome–family. Critical care medicine, 40(2), 618-624.
Diamond, A. (2013). Executive functions. Annual review of psychology, 64, 135-168.
Ehlenbach, W. J., Hough, C. L., Crane, P. K., Haneuse, S. J., Carson, S. S., Curtis, J. R., & Larson, E. B. (2010). Association between acute care and critical illness hospitalization and cognitive function in older adults. Jama, 303(8), 763-770.
Freund, P. A., & Kasten, N. (2012). How smart do you think you are? A meta-analysis on the validity of self-estimates of cognitive ability. Psychological bulletin, 138(2), 296.
Girard, T. D., Jackson, J. C., Pandharipande, P. P., Pun, B. T., Thompson, J. L., Shintani, A. K., ... & Ely, E. W. (2010). Delirium as a predictor of long-term cognitive impairment in survivors of critical illness. Critical care medicine, 38(7), 1513.
Honarmand, K., Lalli, R. S., Priestap, F., Chen, J. L., McIntyre, C. W., Owen, A. M., & Slessarev, M. (2020). Natural History of Cognitive Impairment in Critical Illness Survivors: A Systematic Review. American Journal of Respiratory and Critical Care Medicine, 10.1164/rccm.201904-0816CI.
Hopkins, R. O., & Jackson, J. C. (2006). Long-term neurocognitive function after critical illness. Chest, 130(3), 869-878.
Hopkins, R. O., & Jackson, J. C. (2009). Short-and long-term cognitive outcomes in intensive care unit survivors. Clinics in chest medicine, 30(1), 143-153.
Hopkins, R. O., Wade, D., & Jackson, J. C. (2017). What’s new in cognitive function in ICU survivors. Intensive care medicine, 43(2), 223-225.
Hopkins, R. O., Weaver, L. K., Collingridge, D., Parkinson, R. B., Chan, K. J., & Orme Jr, J. F. (2005). Two-year cognitive, emotional, and quality-of-life outcomes in acute respiratory distress syndrome. American journal of respiratory and critical care medicine, 171(4), 340-347.
Hughes, C. G., Patel, M. B., Brummel, N. E., Thompson, J. L., McNeil, J. B., Pandharipande, P. P., ... & Girard, T. D. (2018). Relationships between markers of neurologic and endothelial injury during critical illness and long-term cognitive impairment and disability. Intensive care medicine, 44(3), 345-355.
Lewkowicz, D. J. (2001). The Concept of Ecological Validity: What Are Its Limitations and Is It Bad to Be Invalid?. Infancy, 2(4), 437-450.
Maaijwee, N. A., Schaapsmeerders, P., Rutten-Jacobs, L. C., Arntz, R. M., Schoonderwaldt, H. C., van Dijk, E. J., ... & de Leeuw, F. E. (2014). Subjective cognitive failures after stroke in young adults: prevalent but not related to cognitive impairment. Journal of neurology, 261(7), 1300-1308.
Marra, A., Pandharipande, P. P., Girard, T. D., Patel, M. B., Hughes, C. G., Jackson, J. C., ... & Brummel, N. E. (2018). Co-occurrence of post-intensive care syndrome problems among 406 survivors of critical illness. Critical care medicine, 46(9), 1393.
Morandi, A., Rogers, B. P., Gunther, M. L., Merkle, K., Pandharipande, P., Girard, T. D., ... & Miller III, R. R. (2012). The relationship between delirium duration, white matter integrity, and cognitive impairment in intensive care unit survivors as determined by diffusion tensor imaging. Critical care medicine, 40(7), 2182.
Nasreddine, Z. S., Phillips, N. A., Bédirian, V., Charbonneau, S., Whitehead, V., Collin, I., ... & Chertkow, H. (2005). The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. Journal of the American Geriatrics Society, 53(4), 695-699.
Pandharipande, P. P., Girard, T. D., Jackson, J. C., Morandi, A., Thompson, J. L., Pun, B. T., ... & Moons, K. G. (2013). Long-term cognitive impairment after critical illness. New England Journal of Medicine, 369(14), 1306-1316.
Patel, M. B., Morandi, A., & Pandharipande, P. P. (2015). What’s new in post-ICU cognitive impairment?. Intensive care medicine, 41(4), 708-711.
Randolph, C., Tierney, M. C., Mohr, E., & Chase, T. N. (1998). The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS): preliminary clinical validity. Journal of clinical and experimental neuropsychology, 20(3), 310-319.
Rawal, G., Yadav, S., & Kumar, R. (2017). Post-intensive care syndrome: an overview. Journal of translational internal medicine, 5(2), 90-92.
Rengel, K. F., Hayhurst, C. J., Pandharipande, P. P., & Hughes, C. G. (2019). Long-term cognitive and functional impairments after critical illness. Anesthesia & Analgesia, 128(4), 772-780.
Rentz, D. M., Dekhtyar, M., Sherman, J., Burnham, S., Blacker, D., Aghjayan, S. L., ... & Maruff, P. (2016). The feasibility of at-home iPad cognitive testing for use in clinical trials. The journal of prevention of Alzheimer's disease, 3(1), 8.
Sakusic, A., & Rabinstein, A. A. (2018). Cognitive outcomes after critical illness. Current opinion in critical care, 24(5), 410-414.
Schmuckler, M. A. (2001). What is ecological validity? A dimensional analysis. Infancy, 2(4), 419-436.
Spreij, L. A., Sluiter, D., Gosselt, I. K., Visser-Meily, J. M., & Nijboer, T. C. (2019). CoCo-participation: The development and clinical use of a novel inventory measuring cognitive complaints in daily life. Neuropsychological Rehabilitation, 1-23.
Stam, H., Stucki, G., & Bickenbach, J. (2020). Covid-19 and post intensive care syndrome: A call for action. Journal of Rehabilitation Medicine.
Uiterwijk, R., Huijts, M., Staals, J., Duits, A., Gronenschild, E., Kroon, A. A., ... & van Oostenbrugge, R. J. (2014). Subjective cognitive failures in patients with hypertension are related to cognitive performance and cerebral microbleeds. Hypertension, 64(3), 653-657. Van Beusekom, I., Bakhshi-Raiez, F., de Keizer, N. F., Dongelmans, D. A., & van der Schaaf,
M. (2018). Lessons learnt during the implementation of a web-based triage tool for Dutch intensive care follow-up clinics. BMJ open, 8(9), e021249.
Wolters, A. E., Slooter, A. J., van der Kooi, A. W., & van Dijk, D. (2013). Cognitive impairment after intensive care unit admission: a systematic review. Intensive care medicine, 39(3), 376-386.
Wood, M. D., Maslove, D. M., Muscedere, J., Scott, S. H., Boyd, J. G., & Canadian Critical Care Trials Group. (2018). Robotic technology provides objective and quantifiable metrics of neurocognitive functioning in survivors of critical illness: A feasibility study. Journal of critical care, 48, 228-236.
Yoon, B. Y., Lee, J. H., & Shin, S. Y. (2017). Discrepancy between subjective and objective measures of cognitive impairment in patients with rheumatoid arthritis. Rheumatology international, 37(10), 1635-1641.
Figure 1. Flow diagram of the literature search conducted for the narrative review. Search string narrative review
"cognitive dysfunction"[MeSH Terms] OR ("cognitive"[Title/Abstract] AND "dysfunction"[Title/Abstract]) OR "cognitive dysfunction"[Title/Abstract] OR ("cognitive"[Title/Abstract] AND "impairment"[Title/Abstract]) OR "cognitive impairment"[Title/Abstract] OR ("mild"[Title/Abstract] AND "cognitive"[Title/Abstract] AND "impairment"[Title/Abstract]) OR "mild cognitive impairment"[Title/Abstract] OR "cognition"[MeSH Terms] OR "cognition"[Title/Abstract] OR ("cognitive"[Title/Abstract] AND "function"[Title/Abstract]) OR "cognitive function"[Title/Abstract] OR ("cognitive"[Title/Abstract] AND "decline"[Title/Abstract]) OR "cognitive decline"[Title/Abstract] OR (neuropsychological[Title/Abstract] AND ("physiology"[Title/Abstract] OR "function"[Title/Abstract] OR "physiology"[MeSH Terms])) AND ("long-term"[Title/Abstract] OR "months"[Title/Abstract]) AND ("critical care"[MeSH Terms] OR ("critical"[Title/Abstract] AND "care"[Title/Abstract]) OR "critical care"[Title/Abstract] OR ("intensive"[Title/Abstract] AND "care"[Title/Abstract]) OR "intensive care"[All fields] OR "intensive care units"[MeSH Terms] OR ("intensive"[Title/Abstract] AND "care"[Title/Abstract] AND "units"[Title/Abstract]) OR "intensive care units"[Title/Abstract] OR ("intensive"[Title/Abstract] AND "care"[Title/Abstract] AND "unit"[Title/Abstract]) OR "intensive care unit"[Title/Abstract] OR "critical illness"[MeSH Terms] OR ("critical"[All Fields] AND "illness"[All Fields]) OR "critical illness"[All Fields] OR ("intensive care"[Title/Abstract] AND ("survivors"[MeSH Terms] OR "survivors"[Title/Abstract]))) AND ("new"[Title/Abstract] OR "innovative"[Title/Abstract] OR “alternative”[Title/Abstract] OR “novel”[Title Abstract] OR “modern”[Title/Abstract] OR “experimental”[Title/Abstract] OR “technology”[Title/Abstract) AND ("2005/01/01"[PDAT] : "2020/04/01"[PDAT])
