Sleep Loss the night before surgery and incidence of postoperative delirium in adults 65–95 years of age

Abstract

Study objectives:

To describe the association between preoperative sleep disruption and postoperative delirium.

Methods:

Prospective cohort study with six time points (3 nights pre-hospitalization and 3 nights postsurgery). The sample included 180 English-speaking patients ≥65 years old scheduled for major noncardiac surgery and anticipated minimum hospital stay of 3 days. Six days of wrist actigraphy recorded continuous movement to estimate wake and sleep minutes during the night from 22:00 to 05:59. Postoperative delirium was measured by a structured interview using the Confusion Assessment Method. Sleep variables for patients with (n = 32) and without (n = 148) postoperative delirium were compared using multivariate logistic regression.

Results:

Participants had a mean age of 72 ± 5 years (range 65–95 years). The incidence of postoperative delirium during any of the three postoperative days was 17.8%. Postoperative delirium was significantly associated with surgery duration (OR = 1.49, 95% CI 1.24–1.83) and sleep loss >15% on the night before surgery (OR = 2.64, 95% CI 1.10–6.62). Preoperative symptoms of pain, anxiety and depression were unrelated to preoperative sleep loss.

Conclusions:

In this study of adults ≥65 years of age, short sleep duration was more severe preoperatively in the patients who experienced postoperative delirium as evidenced by sleep loss >15% of their normal night’s sleep. However, we were unable to identify potential reasons for this sleep loss. Further investigation should include additional factors that may be associated with preoperative sleep loss to inform potential intervention strategies to mitigate preoperative sleep loss and reduce risk of post-operative delirium.

1. Introduction

Sleep disruption and delirium are two frequent phenomena that occur in older hospitalized patients. Although sleep disruption has been frequently cited as an important etiological factor associated with the development of delirium [1], few studies have systematically described the relationship between sleep disruption and delirium in older adults.

Sleep disruption is commonly observed in older adults living at home. While hospitalized, environmental factors such as noise and continuous ambient light, and health care practices such as frequent vital sign measures and medical tests further contribute to sleep disruption [2]. Several small prior studies showed that patients reported sleeping poorly after surgery and this poor sleep may be related to postoperative delirium [3–5]. Most prior studies have consisted only of case reports, and there have been few studies that have systematically evaluated sleep using an objective measure before and after surgery. A more recent study showed that early postoperative wrist actigraphy measures were poor indicators of concurrent or subsequent hypoactive delirium [6]. As a result, the contribution of sleep disruption to the occurrence of postoperative delirium remains controversial.

In a prior pilot study, we showed that patients >40 years of age who experienced postoperative delirium had more severe preoperative sleep disruption [7]. Only 7 of the 50 patients developed postoperative delirium. Despite that small sample size and wide age range (43–91 years), our results suggested a new paradigm showing that sleep disruption in surgical patients may start even before hospitalization and subsequent surgery. Accordingly, we planned a follow up prospective study in a larger cohort to investigate the relationship between the timing of sleep disruption and postoperative delirium in older patients. We hypothesized that patients with sleep disruption before surgery would be at increased risk of postoperative delirium, after adjusting for known risks of postoperative delirium. In addition to age, race, and sex, we also aimed to evaluate additional psychosocial factors such as anxiety and depression that may be associated with preoperative sleep disruption.

2. Methods

2.1. Patient recruitment

The study was approved by the University of California, San Francisco Committee on Human Research, and written informed consent was obtained preoperatively from each study patient. The study took place at the University of California, San Francisco Medical Center between July 2013 and December 2020). Inclusion criteria included English-speaking patients ≥65 years of age undergoing noncardiac surgery requiring anesthesia, with an anticipated hospital stay for 3 days or longer after surgery. Exclusion criteria included a history of restless leg syndrome, periodic limb movement disorder, or obstructive sleep apnea. Potential participants who matched the inclusion criteria were contacted one to two weeks before the planned procedure either by phone or in person if they had an in-person appointment in the preoperative clinic.

2.2. Wrist actigraphy

Participants wore one of two types of wrist actigraph - either a Mini Motionlogger Actigraph (Ambulatory Monitoring, Inc., Ardsley, NY) or Philips Actiwatch 2 (Philips Respironics, Murrysville, Pennsylvania) for 72 h at home before the planned surgery. After surgery, before leaving the post-anesthesia recovery unit, the wrist actigraph was reattached and the patient continued to wear the same device for 72 h after surgery.

Preoperative and postoperative actigraphy data for both the Motion Logger and Philips Actiwatch 2 were dichotomized as night (22:00 to 05:59) and day (06:00 to 21:59) to reflect the hospital’s routine practice with “lights out” at 22:00 and vital sign assessments beginning at 06:00 for elective surgical patients. To obtain actigraphy values, trained research assistants used the autoscoring Cole-Kripke algorithm provided by Action4 software (Ambulatory Monitoring, Inc., Ardsley, NY) for the Motion Logger Actigraphy and the autoscoring default algorithm provided by Actiware software (Version 6.0.9, 2020) for the Philips Actiwatch in order to minimize researcher sleep scoring bias. Time of initial onset and final offset of sleep were measured by the usage of an “event marker” located on the wrist actigraph and confirmed with self-reported sleep diary entries.

Details of the Motion Logger Actigraphy analysis protocol were reported previously [7]. Data from the Philips actigraph were uploaded for analysis onto the Philips Respironics Actiware Software (Version 6.0.9, 2020). Like the Action4 software of the Motion Logger Actigraphy, the sleep variables calculated by the Actiware software’s sleep interval included total sleep time (TST), sleep onset latency, wake time in minutes, number of awakenings, and wake after sleep onset as a percentage of TST (%WASO). Sleep efficiency was calculated by dividing the amount of time spent asleep (in minutes) by the total amount of time in bed (in minutes). Consecutive epochs of inactivity for two or more hours during the daytime were excluded under the assumption that the wrist device was not worn. Trained research assistants also visually examined the actograms to ensure the software’s algorithm properly captured sleep.

Sleep loss the night before surgery was calculated as a percentage of the difference between the mean of TST from the two nights before (preoperative day 1 and 2) and TST on the night before surgery (preoperative day 3), as shown in this equation:

$$\frac{\mathit{mean}\text{(}\mathit{preop}\mathit{day}2\mathit{TST}\text{,}\mathit{pretop}\mathit{day}1\mathit{TST}\text{) -}\mathit{preop}\mathit{day}3\mathit{TST}}{\mathit{mean}\text{(}\mathit{preop}\mathit{day}2\mathit{TST}\text{,}\mathit{pretop}\mathit{day}1\mathit{TST}\text{)}}.$$

We computed receiver operating curve (ROC) analysis to translate values from the above ratio into a cutoff score with the highest sensitivity and specificity. The area under the curve (AUC) determines the accuracy of the score to discriminate between those who will and will not develop postoperative delirium. From these analyses, the cut-off score with the highest sensitivity and specificity was 15%. Sleep loss was then dichotomized at ≤15% and >15% [8,9].

2.3. Self-report sleep measures

In addition to wearing the actigraph, the patients completed two preoperative sleep questionnaires: 1) General Sleep Disturbance Scale (GSDS) consisting of 21 items that query for information about a person’s quality and quantity of sleep in the past week, daytime sleepiness and fatigue, and types of sleep medications [10] and 2) the Pittsburgh Sleep Quality Index (PSQI) [11]. The PSQI asks about perception of sleep quality during the past month and is a valid and reliable measure of habitual sleep patterns.

2.4. Cognitive and delirium assessments

Each patient was interviewed preoperatively and on each of the first three postoperative days. The preoperative interview typically occurred less than one week prior to surgery in the preoperative clinic. Cognitive status was measured before surgery using the Telephone Interview of Cognitive Status (TICS). The TICS is an 11-item screening test that was originally developed to assess cognitive function in patients with Alzheimer’s dementia who were unable to be evaluated in person [12]. The TICS has been compared to the Mini Mental State exam and has similar scores that allowed for standardized comparison [13]. During both the preoperative and the three postoperative interviews, the presence of delirium was measured using the Confusion Assessment Method (CAM) [14]. The CAM assessments were performed daily on the first three days after surgery between 0900 and 1200, using a structured interview. The CAM assessment was developed as a screening instrument based on operationalization of Diagnostic and Statistical Manual of Mental Disorders (DSM)–III–R criteria for use by non-psychiatric clinicians in high-risk settings. Based on a structured interview, the CAM algorithm consists of four clinical criteria: acute onset and fluctuating course, inattention, disorganized thinking, and altered level of consciousness. The determination of delirium requires that both the first and second criterion be present, and either the third or fourth criterion must also be evident. CAM has a sensitivity of 94–100% and specificity of 90–95%, has high inter-observer reliability [14], and has convergent agreement with four other mental status tests. During the interviews, trained interviewers determined the presence of delirium using the CAM. All assessments of postoperative delirium were validated by a second investigator (JML). We defined the occurrence of delirium as the patient meeting CAM criteria for delirium on any of the three postoperative day assessments.

2.5. Demographic factors, symptoms, and surgical risk

The patient’s past health history and other demographic information were abstracted from the medical chart or in-person interview. Current pre- and post-operative pain at rest was measured prospectively by a Visual Analog Scale (range 0 = no pain, to 10 = maximum pain). The 15-item Geriatric Depression Scale (GDS) was used to measure preoperative symptoms of depression [15]. Preoperative anxiety and depression were measured by Hospital Anxiety and Depression Scale (HADS) [16]. Perioperative risk was estimated using the Charlson Comorbidity index [17], and American Society of Anesthesiologists classification [18]. Surgical risk was estimated using guidelines from the American College of Cardiology and American Heart Association update for the perioperative cardiovascular evaluation for noncardiac surgery, which considers type of surgery, intraoperative blood loss, and surgical duration [19].

2.6. Statistical analysis

Initial associations between postoperative delirium status and preoperative and patient characteristics were first examined using unpaired Student t-tests for continuous variables (age, Charlson Comorbidity Index, preoperative TICS scores, GDS scores, surgery duration) and Chi-square tests for categorical preoperative characteristics (gender, race, history of CNS disorders, ASA classification, surgery type, surgery risk, surgery type, and sleep loss determined as above (Table 1). Some patients’ GDS and TICS assessments were incomplete. To determine whether eliminating those with incomplete scores for the TICS and GDS influenced the findings, we imputed their total scores based on available data. The estimates provided in Table 1 provide evidence that the imputed values are very close to scores for participants who had completed every item, providing evidence that missingness for the TICS and GDS did not affect the interpretation of the findings.

Table 1.

Descriptive comparisons for preoperative demographic and surgical characteristics (N = 180).

Characteristic	No Delirium	Delirium	p value
Total	148 (82.2%)	32 (17.8%)
Age (years), mean ± SD	72.0 ± 5.4	73.4 ± 5.6	0.17
Gender			0.30
Female	70 (47%)	19(59%)
Male	78 (53%)	13(41%)
Race			0.12
White	113 (76%)	29 (91%)
Non-white	35 (24%)	3 (9%)
Charlson Comorbidity Index, mean ± SD	1.8 ± 2.0	1.6 ± 1.6	0.88**
TICS scores (preop)*	35.2 ± 3.5	33.7 ± 3.1	0.003**
GDS scores*^	2.7 ± 2.1	3.3 ± 2.1	0.025
HADS depression	5.8 ± 4.3	4.5 ± 3.6	0.92
HADS anxiety	4.7 ± 4.1	4.3 ± 3.5	0.67
History of CNS disorders			0.47
Yes	43 (29%)	12 (37.5%)
No	105 (71%)	20 (62.5%)
ASA classification			0.37
I–II	76 (51%)	13 (41%)
III–IV	72 (49%)	19 (59%)
Preoperative use of sleep aids	38 (26%)	14 (44%)	0.04
Surgical type			0.08
Hip/knee	23 (16%)	3 (9%)
Spine	40 (27%)	15 (47%)
Other**	85 (57%)	14 (44%)
Type of anesthesia
General only	84 (57%)	20 (63%)	0.74
General + regional	49 (33%)	10 (31%)
Regional only	15 (10%)	2 (6%)
Surgical risk			0.002
I-II	132 (89%)	21 (65.6%)
III	16 (11%)	11 (34.4%)
Surgery duration	4.9 ± 2.1	7.1 ± 2.9	<0.001
PSQI, mean ± SD	7.05 ± 3.65	7.95 ± 3.82	0.86
GSDS, mean ± SD	16.2 ± 5.3	19.9 ± 5.4	0.98
Sleep loss the night before surgery (compared to nights 2 and 3 prior)
≤15%	92 (62%)	11(34%)	0.007
>15%	56 (38%)	21(66%)

^**Results from Wilcoxon nonparametric test; type of test based on visual inspection of the distribution of the data.

Note: ASA = American Society of Anesthesiologists; CNS = central nervous system; GDS = geriatric depression scale; GSDS = General Sleep Disturbance Scale, PSQI = Pittsburgh Sleep Quality Index; TICS = Telephone Interview of Cognitive Status.

^**Other = general, urologic, vascular, thoracic.

Univariate logistic regression models were then calculated to determine the association between each preoperative characteristic and the occurrence of postoperative delirium (Table 2) separately for each variable. Surgical type was not included in Table 2 because it is highly associated with surgery risk and surgery duration. A multivariate logistic regression model was then used to determine whether sleep loss was associated with postoperative delirium after considering all variables included in Table 2. Backward elimination was used to determine the best set of predictor variables for occurrence of delirium according to Akaike Information Criterion (AIC). All statistical analyses were done in R software version 4.0.4.

Table 2.

Wrist actigraphy sleep characteristics the night immediately before surgery (preoperative day 1) stratified by postoperative delirium status (N = 180).

Sleep characteristic	No Delirium	Delirium	p value
Sleep onset latency (minutes)	34.5 ± 50.6	49.8 ± 58.6	0.14a
Total sleep time (minutes)	310 ± 83.9	283 ± 89.7	0.11a
Wake time (minutes)	63 ± 48.6	70 ± 51.6	0.39a
Sleep efficiency (%)	58.66 ± 25.06	39.77 ± 31.67	0.0006
Wake after sleep onset (%)	17.1 ± 12.1	20.5 ± 15.2	0.34a
Number of awakenings	23.2 ± 13.7	19.7 ± 12.5	0.18
Mean awakening (minutes)	3.4 ± 3.8	5.3 ± 6.0	0.10^a

^aResults from Wilcoxon nonparametric test; type of test based on visual in-spection of the distribution of the data.

3. Results

The analyses included 180 patients with complete data on sleep and delirium. Participant recruitment, exclusions, and reasons for exclusion are shown in Fig. 1. The mean age was 72.2 ± 5.45 years (range 65–95 years). Approximately half (49.4%) were women. No patient had preoperative delirium. The overall incidence of postoperative delirium observed during any of the three postoperative days was 17.8% (32/180 patients). Patients with postoperative delirium were similar to patients without delirium, with the exception of: 1) preoperative TICS cognitive status scores, which were lower in patients who subsequently developed postoperative delirium; 2) preoperative GDS depression scores, which were also higher in patients with postoperative delirium; and 3) more likely to use sleep aids in the past week before surgery (Table 1). The values in the table show the means and standard deviations for those who completed these inventories. To determine whether missing data may have affected the findings, we imputed scores for those with missing values. Imputed values for TICS scores from available data (n = 161) are 35.3 ± 3.7 for patients without delirium and 33.6 ± 3.2 for patients with delirium; imputed values for GDS from available data (n = 154) are 2.6 ± 2.3 for patients without delirium and 3.4 ± 2.18 for patients with delirium. Therefore, the imputed values are very close to the values reported in Table 1. In addition, patients with delirium had longer surgery duration compared to patients without delirium and were more likely to undergo higher risk surgery.

Fig. 1. Participant flow chart.

Adults 65 years of age and older were recruited and 199 patients scheduled for surgery consented to participate. After excluding incomplete cases due to equipment failure or missing clinical evaluations of delirium, the final sample size was 180.

The sleep characteristics are shown in Table 2. We evaluated the potential difference in results between the two types of watches and found no significant difference between the two hence the results are combined here. Preoperative sleep characteristics including sleep onset latency, TST, wake time, sleep ratio, and number of awakenings were similar between the two groups with and without delirium. Wake after sleep onset (%WASO) was not significantly different between patients without delirium (17 ± 12.1%) vs. with delirium (20 ± 15.2%, p = 0.34). The sleep characteristics for the two and three days before surgery were also similar between patients with and without delirium (Supplementary Tables 1a and 1b). However, the proportion of patients with at least 15% more sleep loss the night before surgery compared to their prior two nights was significantly higher in the group with delirium (66%) than the group without delirium (38%, p = 0.007). This result was also supported by the lower sleep efficiency during the night immediately before surgery in the subjects who subsequently developed postoperative delirium (Table 2). In univariate analysis, preoperative TICS score, GDS score, surgery risk, surgery type, surgery duration, and sleep loss ≥15% were associated with postoperative delirium at p values ≤ 0.10 (Table 3). After backward selection, the following variables were associated with delirium: TICS score, surgery duration, and sleep loss the night before surgery (Table 4). The use of preoperative sleep aids was eliminated via backward selection procedure used for the multivariable model.

Table 3.

Univariate associations with postoperative delirium.

Variable	OR	95% CI	p value
Age	1.05	[0.98, 1.12]	0.18
Gender (Female)	1.63	[0.76, 3.61]	0.22
Race (Non-white)	0.33	[0.08, 1.01]	0.08
Charlson Comorbidity Index	0.94	[0.75, 1.15]	0.60
CNS History	1.47	[0.64, 3.23]	0.35
TICS scorea	0.90	[0.80, 1.00]	0.06
GDS scorea	1.14	[0.97, 1.35]	0.11
PSQI	1.07	[0.95, 1.21]	0.29
ASA III-IV	1.54	[0.72, 3.41]	0.27
Use of preoperative sleep aids	2.25	[1.02 4.96]	0.047
Surgery Risk (III)	4.32	[1.74, 10.58]	0.001
Surgery Duration	1.43	[1.22, 1.71]	<0.001
Sleep loss night before surgery (≥15%)b	3.14	[1.43, 7.21]	0.005

Note: ASA = American Society of Anesthesiologists; CNS = central nervous system; GDS = geriatric depression scale; TICS = Telephone Interview of Cognitive Status.

^aAnalysis done on data with imputation. Sensitivity analysis also done with complete data to ensure consistency of results.

^bSleep loss on night before surgery compared with nights 2–3 prior to surgery.

Table 4.

Multivariate regression analysis of predictors of postoperative delirium after backward selection using AIC criteria (N = 180).

Variable	OR	95% CI	p value
	Logistic
Age	1.08	[1.00, 1.16]	0.06
Female	2.29	[0.94, 5.88]	0.07
Non-white race/ethnicity	0.34	[0.07, 1.18]	0.12
TICSa	0.89	[0.78, 0.99]	0.04
Surgery duration	1.49	[1.24, 1.83]	<0.001
Loss of sleep the night before (>15%)	2.57	[1.07, 6.45]	0.04

Note: both surgery duration and loss of sleep the night immediately before surgery is associated with delirium. TICS = Telephone Interview of Cognitive Status; Age and TICS score show significance with delirium also.

^aAnalysis done on data with imputation. Sensitivity analysis done with complete data to ensure consistency of results.

The preoperative clinical characteristics for sleep loss before surgery are compared in Table 5. Overall, no differences were seen between groups for characteristics of preoperative pain or anxiety measures. To investigate further the potential etiology of acute sleep loss, we evaluated whether the timing of surgery on the next day was related to TST the evening prior to hospital arrival. We found that 91% (70 of 77) of patients with sleep loss the evening prior to surgery were scheduled for the first surgical case of the day compared to 65% 67 of 103) patients without sleep loss (p < 0.001).

Table 5.

Comparison of patients with sleep loss the night before surgery.

Characteristic	≤15% loss	>15% loss	p value
Total	103 (57%)	77 (43%)
Age (years), mean (SD)	72.2 ± 5.7	72.3 ± 5.1	0.88
Gender			0.67
Female	49 (48%)	40 (52%)
Male	54 (52%)	37 (48%)
Race			1
White	81 (79%)	61 (79%)
Non-white	22 (21%)	16 (21%)
Charlson Comordity Index, mean ± SD	1.8 ± 2.1	1.7 ± 1.7	0.79b
TICS scores (preop)a	35.3 ± 2.9	34.5 ± 4.5	0.29b
GDS score (preop)a	2.6 ± 2.2	3.1 ± 2.4	0.16b
History of CNS disorders			0.33
Yes	75 (73%)	50 (65%)
No	28 (27%)	27 (35%)
ASA classification			0.86
I–II	52 (50.5%)	37 (48%)
III–IV	51 (49.5%)	40 (52%)
Anxiety (HADS) scorea			0.55
Normal	65 (82%)	47 (86%)
Borderline	10 (13%)	4 (7%)
Abnormal	4 (5%)	4 (7%)
Depression (HADS) scorea			0.62
Normal	60 (86%)	42 (79%)
Borderline	6 (8%)	6 (11%)
Abnormal	4 (6%)	5 (9%)
Preoperative pain scorea	3.1 ± 3.1	3.2 ± 2.9	0.60
PSQIa	7.4 ± 3.8	7.1 ± 3.6	0.55
GSDSa	16.7 ± 5.5	16.6 ± 5.3	0.88

Note: ASA = American Society of Anesthesiologists; CNS = central nervous system; GDS = geriatric depression scale; GSDS = General Sleep Disturbance Scale; PSQI = Pittsburgh Sleep Quality Index; TICS = Telephone Interview of Cognitive Status.

^aSample size are as follows: TICS n = 161, GDS n = 154, Anxiety A-score n = 140, Anxiety D-score n = 123, Pre-op pain n = 153, PSQI n = 153, GSDS n = 122.

^bResults from Wilcoxon nonparametric test; type of test based on visual inspection of the distribution of the data.

4. Discussion

Our study demonstrates that sleep loss in older surgical patients, specifically at least 15% of normal TST the evening before hospital arrival was a significant predictor of postoperative delirium. Our approach is novel in that we investigated sleep loss as a percentage of the individual’s average normal TST rather than the mean changes in more conventional sleep variables.

Few researchers have measured sleep continuously using actigraphy in the days before major elective surgery in older adults. Two separate meta-analyses reported sleep disturbance to be a predictor of postoperative delirium [20,21]. However, many of the studies cited were in patients with pre-existing obstructive sleep apnea [22–24]. Since obstructive sleep apnea has been shown to increase the risk of postoperative delirium [25], sleep disturbance is an expected experience in these patients given the pathophysiology of obstructive sleep apnea. In contrast to many of these previous studies, our present results may differ because we excluded patients with a known history of obstructive sleep apnea. Furthermore, many studies in the past only focused on evaluating sleep after surgery [26–28].

The present novel finding of substantial sleep loss just before surgery provides an alternative insight as to when sleep disruption begins and how it relates to risk of postoperative delirium. In three studies [29–31] that assessed sleep with the PSQI, patients with self-reported worse sleep quality prior to cardiac surgery [28], arthroplasty [29], or femoral fracture [30] were more likely to develop postoperative delirium. In contrast to these previous studies, our patients with and without postoperative delirium reported similar PSQI scores before surgery but exhibited different sleep loss the evening before surgery. Therefore, although these commonly used sleep surveys such as PSQI may be useful, they may not be sensitive enough to accurately depict acute sleep loss.

If sleep loss occurs before the onset of surgery, potential reasons for loss need to be explored. Older adults undergoing surgery may have symptoms of pain, anxiety, or depression. In our study, pain levels did not differ between patients with and without preoperative sleep loss. In a small study of women scheduled for surgery to diagnosis or treat breast cancer, Wright et al. [5], reported on actigraphy results the night before surgery. They found that intrusive thoughts, anxiety and emotional well-being were each related to sleep duration the night before surgery. In our cohort, preoperative anxiety level was not significantly different between patients with vs. without preoperative sleep loss but depressive symptoms did differ (p = 0.025). Furthermore, depressive symptoms (whether measured by the GDS or HADS) were similar in patients with preoperative sleep loss and without sleep loss. In an exploratory analysis, we showed that patients who were scheduled for the first surgical case of the day likely had to wake up early for same day admission. This experience is likely to alter sleep duration the night prior to surgery. This result differs from findings in a small study that utilized the Athens Insomnia Scale (AIS) to measure sleep in morning vs. afternoon surgical cases in which no significant differences between the two patient groups on preoperative AIS scores [32]. Whether there are other factors contributed to acute sleep loss in our study remain to be determined.

In distinction from prior studies, we evaluated sleep parameters at an individual level as a function of their sleep time prior to surgery rather than as a group comparison such as change in WASO minutes or other more conventional sleep measures. Our approach offers a more robust evaluation of sleep loss at a patient-centered level not previously evaluated.

4.1. Potential limitations

There were several potential limitations in our study. First, we measured delirium only once daily. Given the fluctuating nature of delirium, we might have underestimated its incidence. Second, although we excluded patients with known obstructive sleep apnea, patients may have unrecognized sleep disordered breathing. Finally, we chose wrist actigraphy to measure sleep continuously because it can be used non-invasively to estimate sleep and wake time continuously over an extended period. Although the gold standard for measuring sleep stages is polysomnography, considerable limitations exist for this technology for surgical patients, particularly the limited tolerability by patients over an extended monitoring period. In addition, actigraphy could be used to measure pre- and post-operative sleep in this study because of the feasibility of wearing the device at home and in the hospital. Wrist actigraphy has been shown to have good agreement with polysomnography for total sleep time and wake time during the night [33]. Finally, our results should not be generalized to adults under 65 years of age, and with our limited representation from nonWhite patients, our results have limited generalizability beyond Caucasian older adults.

4.2. Clinical implications

The preponderance of studies in the literature describes how sleep is disrupted for hospitalized patients. In particular, the emphasis has been on describing how hospital practices and environmental factors create conditions detrimental to achieving a good night’s sleep. Our results provide additional evidence to suggest that sleep loss occurs even before hospitalization for some patients. Although we have identified preoperative sleep loss to be prognostically important in its relationship with postoperative delirium, we were unable to identify any patient symptoms of pain, anxiety, or depression associated with this sleep loss. We have preliminary data to suggest that the timing of surgery may affect sleep loss the evening prior to surgery. Whether preoperative sleep loss may also be a marker of more chronic condition in older patients awaiting surgery, and how this can be minimized will need to be investigated in future studies. In addition, poor sleep quality and insufficient sleep quantity impact quality of life. However, despite their frequency and importance, such conditions often go unnoticed, as patients may not report their sleep problems. Our findings need to be replicated but would suggest that sleep disruption should be included in the clinical evaluation of the older hospitalized patients and minimized by helping the patient to prioritize sleep prior to surgery.

In conclusion, using wrist actigraphy to estimate sleep and wake time in a group of older patients before major non-cardiac surgery, we identified acute preoperative sleep loss to exist before surgery when patients were still sleeping in their home environment. Moreover, preoperative sleep loss was associated with the occurrence of postoperative delirium.

List of abbreviations

ASA

American Society of Anesthesiologists

CNS

central nervous system

GDS

geriatric depression scale

GSDS

General Sleep Disturbance Scale

PSQI

Pittsburgh Sleep Quality Index

TICS

Telephone Interview of Cognitive Status