Abstract
Although sleep disturbance is common in chronic obstructive pulmonary disease (COPD), relatively little is known on the effect of the exacerbation on sleep quality. Accordingly, we longitudinally assessed sleep variables during exacerbations and clinical stability. This is a sub-study of a larger observational analysis. Inclusion criteria were clinically stable COPD and two or more clinical exacerbations in the preceding 12 months. Patients were followed for approximately 6 months and during this time the following were recorded daily: (1) COPD exacerbations, which were determined in two ways, clinically and symptom defined using the exacerbations of chronic pulmonary disease tool (EXACT); (2) daytime sleepiness, which was measured using the Stanford Sleepiness Scale; (3) subjective awakenings, which was measured from a sleep diary; and (4) sleep duration, efficiency, and objective awakenings, which was measured from actigraphy. These variables for exacerbation and non-exacerbation days were compared. Seventeen patients (9 male, age 63 ± 12 years, forced expiratory volume in 1 second 52 ± 20%) entered data over 135 ± 18 days. During this time, 15 patients had 27 symptom-defined exacerbations and 8 had 9 clinically reported exacerbations. Symptom-defined exacerbation days were 26% of the total study days. More daytime sleepiness, decreased total sleep time (TST), and decreased sleep efficiency (SE) were present during exacerbations compared with clinical stability (p < 0.001). These disturbances tended to be greater during clinically reported exacerbations than during unreported events (p < 0.05). Increased daytime sleepiness, less TST, and poorer SE are present during COPD exacerbations.
Introduction
Chronic obstructive pulmonary disease (COPD) is a progressive disease punctuated by exacerbations. The frequency of exacerbations in this disease ranges from approximately 0.85 to 2.00 per year for events associated with clinic visits 1 and 3.2 per year in those sent home following a hospitalization for an exacerbation. 2 Not only are exacerbations common, they result in substantial deteriorations in pulmonary function, 3 respiratory symptoms, 3,4 health-related quality of life, 5 and physical activity and functional status. 6,7 Despite the pervasive and substantial effect of the COPD exacerbation on the patient, relatively little information is known regarding its relationship to sleep variables. Accordingly, we assessed subjective daytime sleepiness and nighttime awakenings together with objective sleep measures (total sleep time (TST), sleep efficiency (SE), and number of awakenings) longitudinally in COPD patients, comparing these variables during exacerbation and non-exacerbation (exacerbation-free) days.
Methods
Study design
This investigation was part of a longitudinal study that evaluated the effects of COPD exacerbations on physical activity and sleep. The relationship between exacerbations and directly measured physical activity—activity is significantly decreased on exacerbation days—has been reported. 6 We now present the longitudinal data on sleep variables in this study.
The study was approved by our hospital’s Institutional Review Board, and informed consent was obtained from all subjects. Inclusion criteria included (1) a clinical diagnosis of COPD, (2) a post-bronchodilator forced expiratory volume in 1 second (FEV1)/forced vital capacity <0.70 and an FEV1 <80% of predicted, (3) a history of two or more clinical COPD exacerbations in preceding 12 months, and (4) clinical stability in the month before study entry. Exclusion criteria included (1) a significant movement disorder (that would affect activity measurement), (2) severe lung disease or comorbidity that, in the investigators’ opinion, might preclude completing 6 months of testing, and (3) the inability to complete a daily diary or wear a motion device on the wrist.
The purpose of this component of the longitudinal study was to compare sleep parameters during COPD exacerbations and during clinical stability (exacerbation-free days). Each patient was followed for approximately 6 consecutive months (if no exacerbation occurred) or for 4 weeks following a documented clinical exacerbation.
COPD exacerbations
COPD exacerbations could be either symptom defined or clinically diagnosed. For the former, the exacerbation, including its onset, intensity, and termination, was defined using scoring from the exacerbations of chronic pulmonary disease tool (EXACT), 4,8,9 a patient-reported outcome instrument designed to capture symptom-defined exacerbations. Patients were instructed to complete a paper-based EXACT diary every evening. The 14-item EXACT symptom assessment includes breathlessness, cough, sputum, chest symptoms, sleep disturbance, tired or weak, and scared or worried. Scores range from 0 to 100, with higher values indicating increased symptoms. Increases of (1) greater than 9 points sustained for 3 days or (2) 12 points sustained for 2 days indicate the onset of a symptom-defined exacerbation. A decline (i.e. improvement) in score of at least 9 points from the maximum observed value during an exacerbation event that was sustained for 7 days indicated recovery from that exacerbation. 10
Clinically defined COPD exacerbations were based on physician judgment made independently of the EXACT scores. In our study, patients were instructed to call the investigator physicians if they had an increase in symptoms that might indicate an exacerbation. Supplementing this, follow-up visits were scheduled every 4 weeks, and patients were queried about changes in symptoms and medications at this time. We considered symptom (EXACT)-defined exacerbations and clinically reported exacerbations to be concordant if they had onsets within 5 days. Previous analyses showed a perfect correspondence for the clinically reported exacerbations in this study, with each reaching the EXACT scoring threshold for a symptom-defined event. 6 Recovery from exacerbations and exacerbation-free days was defined symptomatically, using the EXACT.
Sleep variables
In the evening, patients completed the single-item Stanford Sleepiness Scale, 11,12 rating how alert they felt during the day on an 8-point scale, ranging from 1 (“feeling active, vital, alert, or wide awake”) to 8 (“asleep”). In the morning, they entered into the same diary the number of times they woke up and got out of bed the preceding night. They also reported when they went to bed the previous evening and got out of bed that morning. This diary information was used to determine daytime sleepiness, subjective awakenings, and total time in bed.
Objective sleep parameters were assessed using a triaxial accelerometer, the ActiGraph GT3X+ activity monitor, worn on the nondominant wrist like a wrist watch. Actigraphy has been considered a useful research tool for the study of sleep, 13 and a wrist site for the motion detector has more agreement with polysomnography variables than a hip site. 14 The ActiGraph GT3X+ has a battery life and memory storage that permit approximately 30 days of continuous use. Patients were asked to wear the devices continuously, day and night. The ambient light sensor and the inclinometer on the device were not utilized. Return visits were scheduled at monthly intervals. At those times, the battery was recharged and data were downloaded.
Our extraction from ActiGraph data (based on methodology by Sadeh 15 ) and diary entries included estimated TST (using the visual display of the actigraph output and diary information), SE (TST divided by time in bed × 100), and the number of objective awakenings. 6 A review of wrist actigraphy suggests that this method is useful in the estimation of TST, sleep percentage, and wake after sleep onset; however, it is less consistent in the estimation of sleep latency. 16
Statistical analysis
Descriptive statistics are presented as means ± standard deviations. Sleep data collected longitudinally over the course of the study were analyzed using linear mixed models that accounted for repeated measures within subjects. Each model included the covariates of age and gender. The EXACT was used to define exacerbation and non-exacerbation days, as described earlier. Sleep variables during exacerbation days and exacerbation-free days were compared using this mixed model analysis (statistical analysis software, SAS) and least squares means. Regression analyses were performed to explore how other demographic and clinical factors (i.e. age, gender, FEV1% predicted, and 6-minute walk distance (6MWD)) predicted sleepiness. The criterion for significance was p < 0.05. All analyses were performed with the SAS software package (release 9.3; SAS Institute, Inc., Cary, NC, USA).
Results
Eighteen patients were recruited for this study. Of the 18 patients, 1 did not return for testing and was subsequently dropped from the study sample. Baseline disease severity and demographic data of the 17 patients are provided in Table 1. Nine patients were male, three were current smokers, and four had been prescribed supplemental oxygen therapy. The data on directly measured physical activity have been previously reported. 6
Patient characteristics (N = 17).a
BMI: body mass index; FEV1: forced expiratory volume in 1 second; 6MWD: 6-minute walk distance; EXACT: Exacerbations of Chronic Pulmonary Disease Tool; SD: standard deviation.
aValues represent numbers or means ± SD. All data with the exception of EXACT severity scores were gathered at baseline.
In the study, mean participation in the 17 patients was for 135 ± 18 days. Over this time, sleep diary information was available on 2067 days, and actigraphy sleep data were available on 1649 nights. Over the course of the study, 15 patients had 27 EXACT-defined (i.e. symptom-defined) exacerbations and 8 patients had 9 clinically reported exacerbations. All clinically reported exacerbations met EXACT criteria for symptom-defined exacerbations. Two exacerbations required hospitalization. In these cases, activity monitoring was maintained during the hospital stays. Of the 27 symptom-defined exacerbations, 18 (67%) were not reported. Median time from study entry to first exacerbation was 64 days. Of the total number of study days, 26% were exacerbation days.
The Stanford Sleepiness Scale score was significantly higher during exacerbations than during clinically stable (exacerbation-free) days: 2.96 ± 0.30 (SE) versus 2.47 ± 0.29, respectively, p < 0.0001 (see Table 2). Significantly less TST and poorer SE were also observed during exacerbations compared with clinical stability: sleep time, 351 ± 12 versus 368 ± 11 minutes, respectively (p < 0.0001) and SE, 70.0 ± 2.2% versus 73.4 ± 2.1%, respectively (p < 0.0001). Objective awakenings greatly exceeded subjective awakenings, both during exacerbation-free days and during exacerbation days. However, exacerbation-free days and exacerbation days were not significantly different with respect to either objective or subjective awakenings.
Sleep variables during clinical stability and during exacerbations.
TST: total sleep time.
aDefinitions: TST
bVersus exacerbation free, p < 0.0001.
cVersus reported exacerbations, p < 0.001.
dVersus exacerbation free, p < 0.001.
eVersus exacerbation free, p < 0.01.
fVersus exacerbation free, p < 0.05.
gVersus reported exacerbations, p < 0.05.
Table 2 lists sleep variables during exacerbations that were clinically reported (n = 9; all of these also met the scoring threshold for symptom-defined events) and not clinically reported (n = 18, symptom (EXACT)-defined only). Mean sleep values from exacerbation-free days are also listed. Greater impairment in sleepiness, TST, and SE were observed during reported exacerbations than during unreported, symptom-defined exacerbations. There was no significant difference in subjective or objective nighttime awakenings between the health states.
In regression analysis for the entire time period (including both exacerbation and exacerbation-free days), younger age predicted increased overall daytime sleepiness (p = 0.008). After dichotomizing age to either 60 years old or younger or over 60 years, the mean Stanford Sleepiness Scale was 3.59 ± 0.54 in the former category (n = 6) and 2.07 ± 0.40 in those in the latter category (n = 11; p = 0.02). Gender, FEV1 percent predicted and 6MWD were not significantly related to sleepiness. Actigraphy-based TST (p = 0.64), SE (p = 0.05), and number of objective awakenings (p = 0.64) were not related to overall daytime sleepiness. Subjects reporting more frequent nighttime awakenings (i.e. subjective awakenings) also reported higher levels of daytime sleepiness (p = 0.049).
Discussion
Exacerbations of COPD cause disturbances in lung physiology, 17,18 respiratory symptoms, 3 health status, 19 and physical activity. 6 Our study extends these results by demonstrating that increased daytime sleepiness, reduced TST, and poorer SE are additional negative consequences of the exacerbation. The clinical meaningfulness of the change in mean Stanford Sleepiness Scale from 2.48 to 2.96 in our study is not clear, but text anchors for scores of 2 and 3 may be beneficial in this regard: 2: Functioning at high levels, but not at peak; able to concentrate and 3: Awake, but relaxed; responsive but not fully alert. While the negative effects of the exacerbation on health status and fatigue probably reflect multiple causal factors, increased daytime sleepiness may also be contributing to these impairments.
Compared with historical normative populations, sleep quality is decreased in stable COPD. 20 While our study did not compare COPD patients with individuals without disease, TST and SE found in our sample appear to be substantially worse than those reported for persons with insomnia (TST (in minutes) = 413 ± 71; SE = 86 ± 9%) and non-insomnia controls (TST = 456 ± 57; SE = 94 ± 3; wrist actigraphy). 21 They were also poorer than those found in older adults (mean age 68 ± 7 years): TST (in hours): 6.4 men and 6.65 women; SE (%): 77.8 men and 79 women. 22 Comparisons with these studies, however, must be interpreted with caution since different devices and probably different methodologies were utilized.
We demonstrate in our within-subject longitudinal analysis that these sleep parameters are significantly worse during exacerbations. Whether these disturbances account for some or all of the increased sleepiness during the exacerbation is not at all clear. For instance, would a decrease in TST of 18 minutes or a decrease in SE of 3.4% during the exacerbation account for the observed increased sleepiness? Adding to the uncertainly, none of four sleep variables we tested—sleep time, SE, objective awakenings, and subjective awakenings—were strongly associated with daytime sleepiness. Since our study did not measure sleep variables in depth, it is certainly possible that further investigation might uncover more significant causal associations.
The frequency of symptom-defined exacerbations (15 of 17 patients over 135 days) and the total number of exacerbation days (26%) in our study is high. This probably reflects, in part, our selection criterion of two or more clinical exacerbations in the preceding 12 months. Additionally, the EXACT questionnaire, filled out daily by the patient, appears to detect more exacerbations than clinical assessments that typically require patient recall and judgment or changes in medical therapy. In a post hoc analysis of three phase II clinical trials (two of which recruited frequent exacerbators) of 1581 COPD patients, 10 747 had EXACT defined while 328 had medically treated exacerbations, giving a ratio of approximately 2:1. EXACT-defined exacerbations corresponded well with alternative indicators of worsening, and unreported events (i.e. EXACT-defined only) were similar in other measures of severity but longer in duration than medically defined events of moderate severity. In our study, the ratio of EXACT defined to clinically reported exacerbations was 3:1. One major reason for using the EXACT was that it provided a clear-cut end of the exacerbation, a necessary requirement for comparing outcome variables during exacerbation days with those on non-exacerbation days.
Of interest, sleepiness was greater in clinically reported than in unreported (EXACT symptom-defined) only exacerbations, with two of the nine clinically reported events treated in hospital, accompanying this were greater decreases in sleep time and SE. This was somewhat unexpected, since in our analysis reporting directly measured physical activity outcome, clinically reported, and EXACT-only exacerbations had the same negative impact on physical activity, 6 and a recent study showed similar negative effects on health status. 23
Our study has several important limitations. First, while the total number of data points evaluated over 135 days was great, we only studied 17 patients. Furthermore, we made no attempt to evaluate or control for the baseline presence of sleep disorders in our patients. For example, overlap syndrome—coexistence of COPD and obstructive sleep apnea—is not uncommon, 24 and our exclusion criterion of movement disorder did not encompass restless leg syndrome. Thus, the inductive extension of these results to a larger population is relatively weak.
Second, by necessity, we performed only limited measurements of sleep variables using wrist actigraphy. Although there is evidence to suggest wrist actigraphy is superior to actigraphy from a hip site in terms of correspondence with polysomnography, wrist actigraphy is based on upper extremity movement during sleep and tends to overestimate the amount of sleep and underestimate wakefulness. 14 Polysomnography, of course, would be necessary to capture more fully the sleep disturbances during stable and exacerbating states. However, the American Academy of Sleep Medicine does state that, in when polysomnography is not available, actigraphy is indicated as a method to estimate TST in patients with obstructive sleep apnea. 25
Another limitation of our study is that we did not record nap time during daytime hours. It is certainly possible that, because of the prominent fatigue associated with the exacerbation, 26 patients sleep more in the daytime during these periods. This could have major effects on our nighttime-measured sleep variables. Future studies evaluating sleep and sleepiness during exacerbations should take this potential bias into consideration.
Finally, as stated earlier, we cannot determine whether the difference in Stanford Sleepiness Scale scores between exacerbations and clinical stability is clinically meaningful, or even whether the relatively small differences in sleep time or efficiency can account for this difference. Our findings, therefore, should be considered hypothesis generating, perhaps serving as the basis for a larger study.
In summary, our longitudinal study demonstrated that COPD patients reported greater sleepiness, decreased TST, and poorer SE during exacerbations compared with periods of clinical stability. Whether these disturbances can be considered clinically meaningful or whether the changes in sleep time or efficiency are causally related to sleepiness is not determined. Furthermore, our study with only a small number of subjects cannot determine the influence of other variables, such as age, cognitive ability, and comorbid illness, on sleep variables. A larger study with more a thorough evaluation of sleep parameters and with a direct comparison with age-matched normal subjects would be necessary to make firm conclusions.
Footnotes
Funding
This study was funded, in part, by an investigator-initiated study grant by Boehringer Ingelheim Pharmaceuticals, Inc.