Long term outcome after delirium in the intensive care unit
E-mail:bart.vanrompaey@ua.ac.be
Abstract
Aims and objectives. This research studied the long term outcome of intensive care delirium defined as mortality and quality of life at three and six months after discharge of the intensive care unit.
Background. Delirium in the intensive care unit is known to result in worse outcomes. Cognitive impairment, a longer stay in the hospital or in the intensive care unit and a raised mortality have been reported.
Design. A prospective cohort study.
Methods. A population of 105 consecutive patients was included during the stay at the intensive care unit in July–August 2006. The population was assessed once a day for delirium using the NEECHAM Confusion Scale and the CAM-ICU. Patients were visited at home by a nurse researcher to assess the quality of life using the Medical Outcomes Study Short-Form General Health Survey at three and six months after discharge of the intensive care unit. Delirious and non delirious patients were compared for mortality and quality of life.
Results. Compared to the non delirious patients, more delirious patients died. The total study population discharged from the intensive care unit, scored lower for quality of life in all domains compared to the reference population. The domains showed lower results for the delirious patients compared to the non delirious patients.
Conclusions. Mortality was higher in delirious patients. All patients showed lower values for the quality of life at three months. The delirious patients showed lower results than the non delirious patients.
Relevance to clinical practice. Nurses are the first caregivers to observe patients. The fluctuating delirious process is often not noticed. Long term effects are not visible to the interdisciplinary team in the hospital. This paper would like to raise the awareness of professionals for long term outcomes for patients having experienced delirium in the intensive care unit.
Introduction
Delirium is a common syndrome in the intensive care unit with incidences ranging from 11 to 87% (Aldemir et al. 2001, Ely et al. 2001, Bergeron et al. 2002, Immers et al. 2005). The definition is worldwide accepted using the Diagnostic and Statistical Manual of Mental Disorders criteria (American Psychiatric Association 2000). This acute, fluctuating disturbance in the cognitive or motoric functions is induced by a physical cause. The patient encounters periods of inattention in combination with disorganised thinking or a changed level of consciousness.
Delirium is observed as a hypoactive, hyperactive or mixed process. The hyperactive type is the least frequent one although it is the easiest to detect (Palmieri 2003, Miller & Ely 2006, Peterson et al. 2006). Recently, validated tools are worldwide available to assess patients for delirium in the intensive care unit without the need to consult a psychiatrist (Neelon et al. 1996, Bergeron et al. 2001, Ely et al. 2001, Devlin et al. 2007). Research showed, however, that delirium goes unrecognised in one third of the patients in the intensive care unit. The interdisciplinary team notices the hyperactive process, but fails in recognising the more frequent hypoactive syndrome (Peterson et al. 2006).
The attention of physicians and nurses is well focused on interventions for life threatening situations, but less on the poorly visible cognitive processes in the patient’s brain (Pandharipande et al. 2005). Screening for delirium as a standard procedure was often criticised. The clinical relevance of detecting delirious patients has been questioned because defined guidelines for the treatment of the diagnosed patient are still not generally accepted or evidence based (Lacasse et al. 2006). In current years, researchers and intensive care physicians agreed to focus on the incidence, the treatment, the prevention and the outcome of the syndrome (Angus & Carlet 2002).
Background
Several studies reported a worse outcome for patients developing delirium during the stay in the intensive care unit. Attention has been given to the physical, cognitive and social outcome of patients leaving this unit. A higher morbidity, a higher mortality, a longer stay, a deterioration in the cognitive processes and a higher cost of treatment have been linked to the delirious process in the intensive care unit and the hospital (McCusker et al. 2001, 2002, Jackson et al. 2003, Leslie et al. 2005, Thomason et al. 2005, Jackson 2006). The influence of delirium on the long term outcome of intensive care patients has been studied less.
The intensive care team has less contact with the patient once discharged from the intensive care unit or the hospital. Long term effects or complications are not visible to the interdisciplinary team. Therefore, the awareness of intensive care workers to the long term outcome of a delirious state developed during the stay in the intensive care unit must be stimulated.
One of the important aspects in the follow up of diseases is quality of life. The assessment of this outcome has been discussed as an indicator for the general health status of the patient. Most quality of life instruments assess several domains. Mental health and emotional well being are known to predict the quality of life. The general health status of a patient is more related to the physical functioning and the level of perceived pain. Quality of life and general health status are influenced by each other but are not synonyms. Therefore, an instrument covering several domains of the quality of life is advised to assess the status of patients after specific disorders or in specific situations (Smith et al. 1999). The choice for an instrument in research or in clinical practice depends on the situation, the availability and often the cost. The aim of this research is to study the long term outcome, defined as mortality and quality of life, of patients having experienced intensive care delirium at three and six months after discharge of the intensive care unit.
Methods
Study design
A sample of 105 patients was reached in an intensive care delirium study at the Antwerp University Hospital after a limited period of consecutive inclusion in July and August 2006. Non-intubated adult patients were included after a minimum stay of 24 hours in the intensive care unit and scoring a Glasgow Coma Scale greater than 10. All patients meeting the criteria and giving informed consent were included. The first objective of this study was to observe the incidence of delirium comparing two validated instruments. The recruitment of the patients was described in the original paper (Van Rompaey et al. 2008). Furthermore, the included patients were contacted by one nurse researcher for a follow up at three and six months after discharge of the intensive care unit to observe the long term outcome of delirium. A time span of two weeks before and after the exact date was accepted to visit the patient at home. The data collection was performed by a short questionnaire scoring mortality and quality of life.
Diagnosis of delirium in the intensive care unit
The diagnosis of delirium was established by nurse researchers. The mixed population was assessed once a day for delirium using both the Neelon and Champagne Confusion Scale (NEECHAM) (Neelon et al. 1996, Immers et al. 2005) and the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) (Ely et al. 2001). The CAM-ICU diagnoses the delirious state by a yes or no answer to a four point algorithm. A positive answer to this algorithm indicates delirium, a negative indicates a normal cognitive state (Fig. 1). The CAM-ICU was developed by Ely et al. (2001) based on de Confusion Assessment Method originally developed by (Inouye et al. (1990). The instrument is widely spread and translated in multiple languages. The assessment is based on an immediate assessment of the cognitive state of the patient (Fabbri et al. 2001, Laplante et al. 2005, Micek et al. 2005, Wei et al. 2008). The NEECHAM developed by Neelon et al. (1996) and Champagne et al. (1987) uses the nurses’ 24 hour assessment of the level of processing information, the level of behavior and the physiological condition to assess confusion. The instrument rates the patient on a 30-0 scale classifying him in one of four categories. The cut-off values, 30-27 ‘normal’, 26-25 ‘at risk’, 24-20 ‘early to mild confusion’ were standardised (Fig. 2). Delirium was diagnosed for the scores 19-0 ‘moderate to severe confusion’ (Csokasy & Pugh 1999, Matsushita et al. 2004, Milisen et al. 2005, Devlin et al. 2007). The instrument was validated for assessing delirium in intensive care patients (Immers et al. 2005). The assessments of delirium scoring either positive on the CAM-ICU or scoring lower than 20 for the NEECHAM, resulted in the calculation of the incidence for delirium in the studied population (Van Rompaey et al. 2008). In this paper, patients having experienced a delirious period during the stay in the intensive care unit were indicated as the delirious patients, the others as the non delirious patients.
The Confusion Assessment Method for the Intensive Care Unit (CAM-ICU).
The Neelon and Champagne Confusion Scale (NEECHAM).
Observation of mortality
A patient who died in the intensive care unit scored positive for ‘mortality in the intensive care unit’. A patient who died in the intensive care unit or between the discharge of the intensive care unit and the first visit at home was scored as ‘died within three months’. Both mentioned groups and patients who died between the first and the second visit were mentioned as ‘died within six months’.
Assessment of the quality of life
The quality of life was assessed using the Dutch Medical Outcomes Study Short-Form General Health Survey (SF-20) (Kempen et al. 1995). This tool, based on the RAND Health Insurance Study Questionnaire, was developed to decrease the burden on the patient using a shorter instrument (Stewart et al. 1988). The SF-20 assesses the quality of life in six domains: physical, role and social functioning, mental health, health perception and pain. The scale has been used in different groups of patients (Cooke et al. 1996, Hänninen et al. 1998, Carver et al. 1999). The 20 items of the questionnaire are computed to a 0–100 scale in each domain. A maximum score of 100 indicates the best possible functioning and a score of 0 the worst. Pain, however, scores 100 for a maximum possible pain and 0 for the total absence of pain. A total score for the quality of life is not calculated using the SF-20. The results obtained in this study were compared to the SF-20 reference scores for the Dutch population (Kempen et al. 1995). Differences in scores for delirious and non delirious patients were calculated.
Baseline data
Baseline data for the studied population were collected within the framework of the study in the intensive care unit. The Simplified Therapeutic Intervention Scoring System (TISS 28) (Reis Miranda et al. 1996, 2003) was collected daily for each patient during the stay in the intensive care unit. Based on these scores, a mean TISS 28 score was calculated for each individual patient. The Acute Physiology and Chronic Health Evaluation II (APACHE II) is a severity of disease baseline assessment tool in intensive care patients. The score, predicting mortality, is calculated once from the worst scores of routine measurements within 24 hours of admission to the intensive care unit (Knaus et al. 1985). Although the score has been studied recently in cardiac surgery or coronary care patients, the APACHE II was originally not developed to predict outcome in these patients (Knaus et al. 1985, Sawchuk et al. 2003, Kramer & Zimmerman 2008). Therefore, in our study the score was calculated for the internal medicine and non cardiac surgery patients only.
Statistical analysis
The scores for the NEECHAM, APACHE II, TISS 28 and age were explored for their relation with the six domains of the SF-20. The standard variation in the SF-20 scores of delirious and non delirious patients and the way those scores are computed, necessitated the use for non parametric statistics. Other data were analyzed with the Students T-test, One Way anova or the chi squared test where appropriate. Odds Ratio’s (Bland & Altman 2000) were calculated for mortality in the intensive care unit, at three and six months using binary logistic regression with adjustment for age and gender. Correlations were calculated using the non parametric Spearman’s Rho. A significance level of 0·05 was accepted. All statistics have been calculated using the Statistical Package for the Social Sciences 15.0 (SPPS Inc., Chicago, IL, USA).
Ethical considerations
The ethical board of the University Hospital of Antwerp approved the study design and an informed consent was obtained from all patients or their legal representative. Before the first and the second interview patients were formally asked if they did agree to continue their collaboration to the study.
Results
Characteristics of the studied population
A group of 105 consecutive patients was included in this study. Almost two third of the patients was male (63%), the mean age was 62 years (20–90). The population was admitted to the intensive care unit for ‘cardiac surgery’ (35%), ‘other surgery’ (32%) and ‘internal medicine’ (32%). The incidence of delirium was 19% during the stay on the intensive care unit. The CAM-ICU and the NEECHAM assessed each 18% of the patients as positive for delirium. The Kappa of the assessment of delirium for the used scales was 0·94 (p < 0·001). The mean TISS 28 score for all patients was 30 (18–41). The mean APACHE II score for the ‘other surgery’ and the ‘internal medicine’ group was 20 (7–47) (Table 1).
| Non delirium n = 85 | Delirium n = 20 | p* | ||
|---|---|---|---|---|
| Age | Years (range) | 61·5 (20–86) | 64·3 (32–90) | 0·46 |
| APACHE II† | n = 62 (range) | 19·7 (7–47) | 22·3 (9–36) | 0·47 |
| TISS 28 | (range) | 29·5 (18–41) | 31·1 (23–31) | 0·28 |
| Length of stay ICU | Days (range) | 2·5 (1–14) | 13·4 (2–49) | <0·001 |
| Gender | Male | 64·7% | 55·0% | 0·42 |
| Reason of admittance | Cardiac surgery | 37·6% | 35·2% | |
| Non-cardiac surgery | 35·3% | 32·4% | ||
| Internal medicine | 27·1% | 32·4% | 0·06 | |
- *p-value of difference.
- †The APACHE II score was not calculated for cardiac surgery patients.
For the follow up of this cohort, two patients refused further cooperation to the study and seven patients could not be located at the address registered in the hospital (Fig. 3). Mortality during the stay in the intensive care unit was 15% (n = 3) in the delirious population and 5% (n = 4) in the non delirious population. A sample of 78 patients was interviewed at three months. Only one patient in this population was readmitted to the intensive care unit between the interview at three and six months. Therefore, this patient was excluded for further follow up (Fig. 3).
Inclusion of patients.
Mortality after delirium
Three delirious and four non delirious patients died in the intensive care unit. After three months, seven delirious and 11 non delirious patients did not survive. Additionally, one non delirious patient died after six months. More included patients died in the delirium group compared to the non delirium group: 41% and 15% respectively (Table 2).
| Cumulative mortality study population | Non delirium n = 85 (%) |
Delirium n = 20 (%) |
p* | OR | 95% CI |
|---|---|---|---|---|---|
| Mortality in the intensive care unit | 4·7 | 15·0 | 0·097 | 3·03 | 0·57–16·19 |
| Mortality within three months | 13·9 | 41·2 | 0·009 | 4·31 | 1·26–14·68 |
| Mortality within six months | 15·4 | 41·2 | 0·016 | 3·80 | 1·11–13·05 |
- CI, confidence interval.
- *p-value of difference.
The Odds Ratio for mortality in the intensive care unit was 3·03 (0·57–16·19) after experiencing a delirium (Table 2). After discharge of the intensive care unit 11 patients died before the first interview at three months. The Odds Ratio for mortality within three months after a delirious period increased to 4·31 (1·26–14·68) after correction for age and gender. One patient died before the interview at six months. He had not experienced delirium during the stay in the intensive care unit. This influenced the Odds Ratio for mortality within six months after delirium to 3·80 (1·11–13·05) after correction for age and gender (Table 2).
Quality of life of the total study population
At three months the total intensive care unit study population scored lower in all domains of the SF-20 than the reference population (Table 3). Physical function, role function and health perception showed significant lower values. The social function and the mental health showed a trend towards a lower score. At six months, only the perception of pain was significantly different from the reference population.
| Domain | SF-20† (SD) | At three months | At six months | ||
|---|---|---|---|---|---|
| Mean (SD) | p* | Mean (SD) | p* | ||
| Physical function | 67·8 (29·6) | 56·8 (35·1) | 0·01 | 72·9 (31·2) | 0·16 |
| Role function | 73·7 (41·3) | 58·3 (58·3) | 0·01 | 73·3 (43·8) | 0·94 |
| Social function | 80·9 (25·7) | 72·3 (41·1) | 0·07 | 78·9 (36·2) | 0·64 |
| Mental health | 76·0 (18·9) | 71·4 (23·3) | 0·09 | 74·0 (22·8) | 0·44 |
| Health perception | 67·6 (24·8) | 54·6 (31·5) | <0·001 | 64·8 (28·3) | 0·40 |
| Pain | 30·4 (31·1) | 34·0 (36·0) | 0·38 | 22·3 (32·8) | 0·04 |
- *p-value of difference.
- †Mean scores for the Dutch reference population.
Quality of life after intensive care delirium
At three months, the delirium group showed lower values than the non delirium group in all domains of the SF-20 (Fig. 4). Only the role function scored significantly lower than the non delirium group (Table 4). At six months, all domains had lower results for the delirious patients than the non delirious patients also. The role function persisted in a significantly lower score. The non delirious patients had lower SF-20 values than the reference population at three months. In contrast to the delirious patients, the non delirious patients reached the values of the reference population at six months in five domains.
SF-20 scores for the Dutch population, delirium and non delirium.
| Domain | SF-20† | At three months | At six months | ||||
|---|---|---|---|---|---|---|---|
| Non delirium | Delirium | p* | Non delirium | Delirium | p* | ||
| Physical function | 67·8 | 59·1 | 41·7 | 0·13 | 73·8 | 66·7 | 0·56 |
| Role function | 73·7 | 62·5 | 30·0 | 0·05 | 78·5 | 40·0 | 0·01 |
| Social function | 80·9 | 75·3 | 52·0 | 0·20 | 81·5 | 62·0 | 0·24 |
| Mental health | 76·0 | 71·6 | 70·0 | 0·48 | 74·9 | 68·0 | 0·21 |
| Health perception | 67·6 | 55·7 | 47·5 | 0·40 | 68·9 | 57·5 | 0·42 |
| Pain | 30·4 | 36·4 | 17·5 | 0·14 | 20·8 | 32·5 | 0·30 |
- *p-value of difference.
- †Mean scores for the Dutch reference population.
Relation between quality of life scores and patient and clinical characteristics
Age was negatively related to most domains at three and six months (Table 5). The physical function at six months, however, was the only domain significantly related. Mental health showed a weak positively relation at three and at six months. The Spearman’s Rho showed a significant positive relation between the results of the NEECHAM Confusion Scale and the SF-20’s physical function, role function and the health perception at three months. At six months, the social function was additionally positively related. A higher degree of confusion, indicated by a lower a NEECHAM score, seemed related to lower values for the quality of life. The APACHE II score in the intensive care unit was significantly related to four domains at three months and to three domains at six months. The APACHE II score was still linked to the physical, the role and the social functioning at six months.
| Physical function | Role function | Social function | Mental health | Health perception | Pain | |
|---|---|---|---|---|---|---|
| Three months | ||||||
| Age | −0·21 | −0·17 | −0·19 | 0·03 | −0·13 | −0·10 |
| NEECHAM | 0·35* | 0·31* | 0·20 | 0·19 | 0·25* | −0·05 |
| APACHE II† | −0·34* | −0·38* | −0·40* | −0·31* | −0·19 | −0·08 |
| TISS 28 | −0·01 | 0·00 | 0·06 | 0·13 | 0·09 | −0·16 |
| Six months | ||||||
| Age | −0·31* | −0·16 | −0·15 | 0·11 | −0·11 | −0·01 |
| NEECHAM | 0·27* | 0·34* | 0·30* | 0·28* | 0·20 | −0·18 |
| APACHE II† | −0·33* | −0·34* | −0·40* | −0·27 | −0·13 | −0·04 |
| TISS 28 | 0·14 | 0·06 | 0·06 | 0·25* | 0·12 | −0·13 |
- *Significant at 0·05 level.
- †APACHE II is calculated for non cardiac surgery and internal medicine patients only.
The TISS 28 score in the intensive care unit seemed not related to the scores in the different domains at three months. At six months, only the domain mental health showed a significant relation. Multivariate analysis, taking into account delirium and possible confounding factors, did not find any relevant or significant relationship with the quality of life scores.
Discussion
In this study delirious patients showed a higher mortality within six months after discharge of the intensive care unit. At three months non delirious patients scored higher on the SF-20 as the delirious patients, but showed lower values than the reference population. At six months the delirious patients still scored significantly lower for the role function.
There seems no doubt that mortality is an important outcome after an intensive care unit delirium. In this study a larger amount of delirious patients died before and after the discharge of the intensive care unit. In spite of the small population, the calculated Odds Ratio’s showed a higher risk for mortality after delirium. This confirmed research on outcome of delirium outside the intensive care unit. Leslie et al. (2005) and McCusker et al. (2002) calculated hazard ratios of 1·62 and 2·11 for mortality in delirious patients after 12 months. Intensive care nurses and physicians should be aware that the life threatening situation of a patient could be extended for months by a delirium.
The scores in the six domains of the SF-20 in this study showed lower values for the included patients compared to the reference population. The results suggest the impact of critical illness on the quality of life. For non delirious patients this situation seems to normalise after six months. In contrast, delirious patients showed a delayed improvement of the situation.
Delirious patients scored lower at three and at six months for all domains. Jackson et al. (2003) did not find differences in quality of life at hospital discharge and at a six month follow-up after cognitive impairment in the intensive care unit. This should be verified in a larger cohort. The major problem for delirious patients in this study seems to be situated in the role function. This domain reflects on functioning in daily life activities such as work, housekeeping or daily activities (Kempen et al. 1995). Therefore, delirious patients seem to rehabilitate more difficult in the society after discharge of the intensive care unit. This was supported by the lower trend in the domain social function. The quality of life scores of the Dutch reference population were given separately for different decades for age, differences in education or specific chronic diseases in the original instrument (Kempen et al. 1995). Due to the small population, the findings in this study were not compared to this categorised SF-20 values.
The Spearman’s Rho pointed at two important factors explaining the scores in the different domains. First, the NEECHAM demonstrated a long term effect on the quality of life. A lower score, indicating more confusion or delirium, was correlated to a lower score in all domains. Second, the APACHE II might balance the long term effect of delirium. A higher APACHE II score resulted in lower scores for several domains also. It might be questioned whether either delirium or the severity of illness influenced the quality of life most. The severity of illness, however, was not calculated for the total population leaving a multivariate analysis for a smaller population not appropriate.
The SF-20 was selected for this research based on availability without cost and low burden for the patient. The instrument has proved to be a valid tool in the past (Cooke et al. 1996, Hänninen et al. 1998, Carver et al. 1999). The physical functioning is based on six questions, the mental health and health experience on five. The role function, however, is based on two questions and the social function and perceived pain on one. The transformation of a limited number of questions to a percentage results in a low number of available scores. The range of scores between 0–100 resulted in an interrupted scaling where only a few scores were possible within that range. Therefore, non parametric statistics were chosen to handle the scores of the SF-20.
Limitations
As already mentioned, the small population of this study limits the conclusions. The results must be confirmed in a larger population. Additionally, this research was confronted with a lack of total score of the quality of life limiting the analytical possibilities.
Relevance to clinical practice
Despite standardised screening protocols, intensive care delirium remains mainly undetected. Nurses are always present at the bedside of a patient and therefore best placed to observe and report. A long term effect on the outcome of the patient has been reported. This worse outcome, however, is not visible to the intensive care nurse or physician. Since delirium might be a preventable complication in the intensive care unit, preventive or therapeutic action must be developed. Therefore, this paper would like to raise the awareness of all caregivers for delirium in the intensive care unit.
Conclusions
Non delirious patients were more likely to survive three months after discharge of the intensive care unit. All included intensive care patients showed lower SF-20 results than the reference population three months after discharge. Moreover, the delirious patients still showed lower results than the non-delirious patients at six months. Further research in larger intensive care unit populations is needed to confirm the long term outcome of a delirium in the intensive care unit.
Contributions
Study design: BVR, ME; data collection and analysis: BVR, ME; manuscript preparation: BVR, ME, MS, LS, ST, LB.
Acknowledgements
The authors like to thank Yves Denis, nurse researcher, for the efforts in contacting the patients and assessing the SF–20, and Els Schakman, nurse researcher, for the assistance in the data collection in the intensive care unit. The physicians and nurses of the intensive care unit of the University Hospital of Antwerp supported the study.
Conflict of interest
The authors declare that they have no conflict of interests. No funds or grants were received for this study.
