Intensive care unit environment may affect the course of delirium
- 3.9k Downloads
Delirium is a common disorder in intensive care unit (ICU) patients. It is unclear whether ICU environment affects delirium. We investigated the influence of ICU environment on the number of days with delirium during ICU admission.
In this prospective before–after study, ICU delirium was compared between a conventional ICU with wards and a single-room ICU with, among others, improved daylight exposure. We included patients admitted for more than 24 h between March and June 2009 (ICU with wards) or between June and September 2010 (single-room ICU). Patients who remained unresponsive throughout ICU admission were excluded. The presence of delirium in the preceding 24 h was assessed daily with the confusion assessment method for the ICU (CAM-ICU) by research physicians combined with evaluation of medical and nursing charts. The number of days with delirium was investigated with Poisson regression analysis.
We included 55 patients (449 observation days) in the ICU with wards and 75 patients (468 observation days) in the single-room ICU. After adjusting for confounding, the number of days with delirium decreased by 0.4 days (95 % confidence interval 0.1–0.7) in the single-room ICU (p = 0.005). The incidence of delirium during ICU stay was similar in the ICU with wards (51 %) and in the single-room ICU (45 %, p = 0.53).
This study is the first to show that ICU environment may influence the course of delirium in ICU patients.
KeywordsDelirium Delirium management Intensive care Intensive care environment Nursing care
Acute physiology and chronic health evaluation
Confusion assessment method in intensive care unit
Charlson co-morbidity index
Delirium severity index
Glasgow coma scale
Intensive care unit
Length of stay
Richmond agitation and sedation score
Sequential organ failure assessment
University Medical Centre Utrecht
Delirium is a common condition in intensive care unit (ICU) patients [1, 2], characterized by a disturbance of consciousness and attention with a change in cognition and fluctuating course . Delirium in the ICU is associated with complications and adverse outcomes including prolonged hospital stay [4, 5] and increased morbidity and mortality [6, 7, 8].
In non-ICU patients, non-pharmacological strategies could prevent delirium [9, 10, 11]. These include proactive geriatric consultation  and education of medical and nursing staff with special attention to known risk factors [10, 11]. In ICU patients, non-pharmacological measures have hardly been studied in relation to delirium . Still, the ICU environment with continuous light and noise, around the clock personnel and lack of orientation points is assumed to play a role in the development of delirium [12, 13, 14, 15]. A previous study in ICU patients suggested that isolation and the absence of daylight were associated with an increased risk of delirium . However, this study may have been biased as isolation and bed assignment may have been dependent on disease characteristics.
The ICU design can ameliorate healthcare outcomes, and lead to improved patients’ sleep and a reduced frequency of hospital-acquired infections and medical errors by influencing circadian rhythm, the immune response, and patient and staff behaviour . The ICU of our institution recently moved to a new location, providing a unique opportunity to study the effects of a changed ICU environment on the course of delirium.
The aim of this study was to investigate the influence of ICU environment on the number of days with delirium.
Design and study population
This prospective, before–after study was performed in the 32-bed mixed ICU of the University Medical Centre Utrecht (UMCU), the Netherlands. Both in the ICU with wards and in the single-room ICU, the ICU has been divided into three units with 11 beds. All three units are equal in staffing, admission criteria and facilities. A waiver of informed consent was obtained from the local medical ethics committee.
We included all patients admitted for more than 24 h of one out of the three ICU units between March and June 2009 (ICU with wards) or between June and September 2010 (single-room). We excluded patients who were unable to speak Dutch and English, and those who remained unresponsive [defined as a Richmond agitation and sedation scale  (RASS) less than −3 and/or a Glasgow coma scale score  (GCS) no greater than 8] throughout ICU admission.
Data were collected prospectively, on a daily basis, 7 days a week. Standard demographic data were registered at inclusion. Co-morbidity at hospital admission was registered with the Charlson co-morbidity index . Severity of illness at ICU admission was assessed using the acute physiology and chronic health evaluation (APACHE) version II score . Severity of illness during ICU admission was estimated daily with the sequential organ failure (SOFA) score [22, 23]. We further recorded daily the use of physical restraints at any moment during the preceding day, defined from 10 a.m. to 10 a.m.
Delirium in the preceding day was assessed daily, 7 days a week, by two research physicians (IJZ and CFS) who received theoretical and bedside training by a neurologist-intensivist (AJCS). This classification of mental status was as follows: (1) awake and not delirious during the preceding day, or (2) delirious at any moment during the preceding day, or (3) always unresponsive during the preceding day. The research physicians administered the Dutch version of the confusion assessment method for the ICU  (CAM-ICU) at a predefined time, between 10 and 12 a.m. If the patient was inaccessible, the evaluation was repeated between 3 and 5 p.m. Delirium often has a fluctuating course. To minimize the influence of this fluctuation the research physicians inspected the medical and nursing charts, including the results of the twice-daily CAM-ICU screening by trained bedside ICU nurses . In case of doubt, a neurologist or psychiatrist was consulted who had the decisive vote with regard to the classification of mental status. The number of days with delirium was counted cumulatively during ICU admission, hence without taking different periods of delirium into account. The delirium severity index (DSI) was used to register daily the severity of delirium , based on the highest absolute RASS score in the preceding day. To determine interobserver variability in the mental status classification, the two research physicians and neurologist-intensivist assessed 36 patients simultaneously, where every observer was blinded to other observers’ conclusions.
Medication use was recorded with a computerized patient data monitoring system. For this study, we collected data on the use of medication that is mentioned in our protocols for sedation, analgesia and delirium, i.e. propofol, midazolam, oxazepam, temazepam, zopiclon, fentanyl, morphine, clonidine, haloperidol and quetiapine. During the study period, dexmedetomidine was not available in the Netherlands. In the statistical analyses, benzodiazepine dosages were converted in diazepam equivalents  and opioid dosages into fentanyl equivalents .
Light intensities were measured in September 2009 (ICU with wards) and in September 2010 (single-room ICU) with a light sensor (OSD15-E photodiode). The light sensor was attached 1 m from a randomly assigned patient’s head to measure light intensity in 30-s intervals. After 3 days, the light sensor was moved to another, randomly assigned, patient. This cycle was repeated six times in both the old and the new setting. The Royal Netherlands Meteorological Institute (KNMI) headquarters is located in De Bilt, the Netherlands, just 2,700 m north of the UMCU. For the days of our light measurements we consulted the KNMI database and retrieved the daily observations of the weather station in De Bilt . In this database the daily sunshine duration is calculated according to an algorithm developed and published by the KNMI . The cloud cover is recorded in a graded scale from zero, indicating clear weather, to nine, completely overclouded.
Outcomes and statistical analysis
The primary outcome was the number of days with delirium during ICU admission. Secondary outcomes were the occurrence rate and severity of delirium.
The Student’s t test was used to study independent samples of continuous, normally distributed data and the Mann–Whitney U test for continuous, skewed data. The Chi-square test was used to analyse categorical data. With Poisson regression analysis, we compared the cumulative number of days with delirium during ICU admission between the two settings, adjusted for the following confounding variables: age, gender, APACHE II, Charlson co-morbidity index, highest SOFA score during ICU admission, admission type (urgent vs. elective) and admitting discipline [in four categories: (1) medicine, (2) general surgery, (3) cardiology or cardiothoracic surgery, (4) neurology or neurosurgery]. We used logistic regression analysis to study the association between the two settings and the occurrence of delirium, adjusted for the possible confounders mentioned above. We used linear regression analysis to compare the severity of delirium, based on the cumulative DSI per patient divided by the total number of days with delirium for the particular patient, adjusting for the same covariates. We tested for multicollinearity in our multivariable regression analyses. Interobserver variability was expressed as κ score.
We computed total light intensity over 24 h by summing the measurements of the 30-s intervals. Dividing this total light intensity by the number of measurements provides a mean light intensity over 24 h. Similarly, we calculated the mean light intensity during the day (07.00 a.m.–10.00 p.m.) and the night (10.00 p.m.–07.00 a.m.).
All statistical analyses were performed using SPSS 17.0®, SPSS inc., Chicago, Illinois, USA. A two-sided p value less than 0.05 was considered statistically significant. Variance inflation factors (VIF) of at least 10 and tolerance values (1/VIF) less than 0.1 were considered as values indicating problems of multicollinearity.
Characteristics of study population
Old ICU (n = 55)
New ICU (n = 75)
Age, mean (SD)
Male gender, n (%)
36 (66 %)
44 (59 %)
APACHE II, mean (SD)
CCI, median (IQR)
SOFA max, mean (SD)
Admitting discipline, n (%)
14 (26 %)
22 (29 %)
4 (7 %)
15 (20 %)
27 (49 %)
18 (24 %)
10 (18 %)
20 (27 %)
Admission type, n (%)
33 (60 %)
62 (83 %)
22 (40 %)
13 (17 %)
ICU LOS; days, median (IQR)
Mechanical ventilation, n (%)
52 (95 %)
63 (84 %)
Days of mechanical ventilation, median (IQR)
Mortality, n (%)
2 (4 %)
7 (9 %)
Main outcomes and multivariate regression analysis
Old ICU (n = 28)
New ICU (n = 34)
Crude number of days with delirium, median (IQR)
Adjusted difference in the number of days with delirium, (95 % CI)b
−0.4 (−0.7 to −0.1)
Mean DSI per day with delirium, mean (SD)
Adjusted difference in DSI, (95 % CI)b
0.3 (−0.2 to 0.7)
Days spent comatose, median (IQR)
Mortality, n (%)
1 (4 %)
3 (9 %)
Old ICU (n = 55)
New ICU (n = 75)
Crude risk of delirium during ICU admissiona, n (%)
28 (51 %)
34 (45 %)
Adjusted odds ratio for delirium, OR (95 % CI)a,b
Mean RASS, mean (SD)
Days spent comatose, median (IQR)
Haloperidol, sedatives and analgesics use during ICU admission
Old ICU (n = 55)
New ICU (n = 75)
Number of patients using haloperidol, n (%)
16 (21 %)
20 (27 %)
Average amount of haloperidol per patient, mg/day, median (IQR)
Number of patients using propofol, n (%)
34 (62 %)
41 (55 %)
Average amount of propofol per patient, mg/day, median (IQR)
Number of patients using opioids, n (%)
47 (85 %)
59 (79 %)
Average amount of opioids per patient in fentanyl equivalents, mg/day, median (IQR)
Number of patients using benzodiazepines, n (%)
35 (64 %)
43 (47 %)
Average amount of benzodiazepines per patient in diazepam equivalents, mg/day, median (IQR)
Number of patients using clonidine, n (%)
12 (22 %)
20 (27 %)
Average amount of clonidine per patient, mg/day, median (IQR)
This study suggests that a change in ICU environment can decrease the number of days with delirium during ICU admission. In an ICU with all single rooms, we found that patients spent fewer days with delirium than in a conventional ICU with wards. Reducing the duration of delirium is of pivotal importance to ICU patients as each additional day with delirium in the ICU has been found to be associated with a 10 % increased risk of death [6, 31].
Since the development of ICUs in the 1950s, the main focus of intensive care medicine was the survival of the patients, whereas less attention was paid to the environment in which this was established. However, the nursing environment, and especially excessive noise or light, appeared to influence several healthcare outcomes . Post-surgical patients with a view onto a natural scene had a shorter hospital stay than patients with windows facing a brick wall . Studies focusing on ICU environment and patient outcomes are, however, sparse [15, 33]. This is the first study comparing different ICU environments and delirium in intensive care patients.
The strengths of this study include the 7 days a week assessments by research physicians with high interobserver agreement, which ensured good quality of the evaluations. Delirium often has a fluctuating course. Using both CAM-ICU assessments and chart evaluations we tried to minimize the influence of fluctuation of the symptoms on our daily mental status classification. Because of the strict follow-up, there were no missing data. The study populations included a wide spectrum of diseases and conditions representing a typical mixed-type adult ICU, ensuring generalization of the results. Although the sample size of our study population (n = 130) was relatively small, the number of observation days was high (n = 917).
However, this study has some limitations. Most important is the before–after study design and the possibility that other factors than ICU environment have changed over time and influenced our findings. Between the two settings there were, however, no changes in doctor–patient or nurse–patient ratios, nor in practice or protocols for sedation, analgesia and delirium. We further adjusted for all differences in patient characteristics between the two ICU settings.
A source of bias may be that the two research physicians were aware of the change in ICU environment and the aim of the study. Theoretically, blinding would overcome this limitation, but this is impossible to apply in a study on the effects of ICU environment. Because of the high specificity of the CAM-ICU performed in daily practice by bedside nurses , who were not aware of the aim of the study, their CAM-ICU screening was taken into account in the daily mental status classification.
Up until now it was unclear how to quantify the severity of delirium; therefore, there was no golden standard for rating the severity of delirium. Without a golden standard it is not possible to validate severity instruments. In our study we made use of the DSI, which has also not been validated yet. The use of the Charlson co-morbidity index in studies regarding delirium is limited because it does not register relevant co-morbidity for delirium, such as cognitive decline, drug or alcohol abuse and other psychiatric disorders.
We have used a multicomponent intervention. However, our results could be subject to confounding by non-environmental factors or any other environmental change. The use of specific medication or physical restraints, undertreated pain and the duration of mechanical ventilation are possible risk factors for the development of delirium [7, 13, 34, 35]. In our study population, the duration of mechanical ventilation and the use of psychoactive medication, opioids and physical restraints were not different between the two settings. Unfortunately, we did not register pain in our study. However, possible differences in pain may be related to the changes in the ICU environment, as shown in patients after cholecystectomy . Differences in pain between the two settings may therefore be one of the intermediate factors in the causal relationship between the intensive care unit environment and delirium. In our study we have measured light intensities as one part of the multicomponent intervention. In the literature light levels are generally expressed in either volts or lux. Whereas lux is correlated to the light perceived by the human eye, a photodiode (measuring light in volts) is much more sensitive. Therefore, linear conversion from volts to lux is only possible within the visible spectrum. In both the ICU with wards and the single-room ICU, we measured light beyond the visible spectrum, hampering conversion to lux. Unfortunately, in our study we did not incorporate measurement of other environmental factors, such as noise levels.
Finally, we used Poisson regression to analyse differences in the number of days with delirium and did not include the time-varying nature of delirium. We cannot exclude potential immortal time bias, as the sickest patients are most likely to die early during critical illness with less opportunity to develop delirium . However, in the two settings there was no difference in mortality, nor in the occurrence rate of delirium in the patients who died. Therefore, the risk of immortal time bias in our data was considered low.
This study is the first to assess the influence of ICU environment on delirium in the ICU. Our study suggests that environmental factors can influence the number of days with delirium during ICU admission, broadening the evidence for the effectiveness of non-pharmacological measures in the treatment of delirium. Future research should focus on single interventions in the ICU environment to determine the influence of ICU environmental factors on delirium. As the ICU environment appears to influence the course of delirium, non-pharmacological anti-delirium measures deserve more attention in intensive care medicine.
The authors thank A.W. van der Kooi, MSc, Department of Intensive Care Medicine, University Medical Centre Utrecht, Utrecht, the Netherlands, for her assistance in analysing the data concerning light intensities in the ICU.
Conflicts of interest
The authors declare that they have no competing interests.
- 3.American Psychiatric Association (2000) Diagnostic and statistical manual of mental disorders, 4th edn. Revision, Washington DC, pp 135–147Google Scholar
- 17.Ely EW, Truman B, Shintani A, Thomason JW, Wheeler AP, Gordon S, Francis J, Speroff T, Gautam S, Margolin R, Sessler CN, Dittus RS, Bernard GR (2003) Monitoring sedation status over time in ICU patients: reliability and validity of the Richmond agitation-sedation scale (RASS). JAMA 289:2983–2991PubMedCrossRefGoogle Scholar
- 19.Kesecioglu J, Schneider MME (2012) The intensive care unit of tomorrow: a case study of patient-centred care. ICU Management 12:12–13Google Scholar
- 22.Vincent JL, de MA, Cantraine F, Moreno R, Takala J, Suter PM, Sprung CL, Colardyn F, Blecher S (1998) Use of the SOFA score to assess the incidence of organ dysfunction/failure in intensive care units: results of a multicenter, prospective study. Working group on “sepsis-related problems” of the European Society of Intensive Care Medicine. Crit Care Med 26:1793–1800PubMedCrossRefGoogle Scholar
- 24.Vreeswijk R, Toornvliet A, Honing MLH, Bakker K, De Man T, Daas G, Spronk PE, de Jonghe JF, Kalisvaart KJ (2009) Validation of the Dutch version of the confusion assessment method (CAM-ICU) for delirium screening in the intensive care unit. Neth J Crit Care 13:73–78Google Scholar
- 25.van Eijk MM, van den Boogaard M, van Marum RJ, Benner P, Eikelenboom P, Honing ML, van der Hoven B, Horn J, Izaks GJ, Kalf A, Karakus A, Klijn IA, Kuiper MA, de Leeuw FE, De Man T, van der Mast RC, Osse RJ, de Rooij SE, Spronk PE, van der Voort PH, van Gool WA, Slooter AJ (2011) Routine use of the confusion assessment method for the intensive care unit: a multicenter study. Am J Respir Crit Care Med 184:340–344PubMedCrossRefGoogle Scholar
- 29.Royal Netherlands Meteorological Institute (2012) KNMI weather database of weather station De Bilt. Available via KNMI. http://www.knmi.nl/datacentre. Accessed 28 June 2012
- 30.Groen G, Sluijter R, Stammes P, Ilderda R (2012) Implementation of the Hinssen-Knap algorithm for the calculation of sunshine duration. KNMI publication TR-319. Available via KNMI. http://www.knmi.nl. Accessed 28 June 2012