Alimentary Pharmacology and Therapeutics

Sleep and physical activity measured by accelerometry in Crohn’s disease D. R. van Langenberg, M. C. Papandony & P. R. Gibson

Department of Gastroenterology and Hepatology, Eastern Health Clinical School, Monash University, Box Hill, Vic., Australia.

Correspondence to: Dr D. van Langenberg, Department of Gastroenterology & Hepatology, Eastern Health Clinical School, Monash University, Level 2, 5 Arnold Street, Box Hill, Vic. 3128, Australia. E-mail: daniel.van-langenberg@ monash.edu

Publication data Submitted 3 December 2014 First decision 30 December 2014 Resubmitted 4 February 2015 Resubmitted 21 February 2015 Accepted 22 February 2015 EV Pub Online 17 March 2015 This article was accepted for publication after full peer-review.

SUMMARY Background Sleep and physical activity are inherent to human living, yet appear affected by Crohn’s disease (CD), resulting in fatigue and disability. Aim To objectively assess sleep quality and physical activity and their associations using accelerometers, comparing CD vs. matched healthy control (HC) subjects. Methods Exactly 49 CD and 30 HC subjects completed surveys encompassing selfreported fatigue and sleep quality, pathology testing and wore an accelerometer for 7 days, measuring physical activity and sleep. In this cross-sectional observational study, per-group comparisons were performed and in CD, factors associated with reduced activity and/or sleep quality were assessed via multivariate analyses. Results Regarding physical activity, CD subjects overall performed less total accelerometer counts (median 1.3 9 106 vs. 2.0 9 106), were more sedentary (97.7% vs. 96.2%) and completed fewer bouts of moderate-vigorous intensity exercise (1.0 vs. 5.0, each P < 0.01 (Mann–Whitney) than HC over 7 days. Factors associated with poor physical activity in CD included elevated serum CRP (OR = 22.6), lower vitamin D3 (OR = 13.1) and longer disease duration (OR = 1.2 per year, each P < 0.05). Regarding sleep, the CD group had similar total sleep time (median 458 vs. 447 min, P = 0.56), but more awakenings post-sleep onset (22 vs. 11, P = 0.01). Factors associated with severe sleep dysfunction in CD included lower haemoglobin (OR = 6.7) concurrent anti-TNF (OR = 6.5, each P < 0.05) and opioid therapy (OR = 6.6, P = 0.09). Conclusion Utilising objective measurement in a habitual context over 7 days, patients with Crohn’s disease exhibited poorer sleep quality and less physical activity than well-matched healthy controls. Aliment Pharmacol Ther 2015; 41: 991–1004

ª 2015 John Wiley & Sons Ltd doi:10.1111/apt.13160

991

D. R. van Langenberg et al. INTRODUCTION Chronic diseases such as Crohn’s disease (CD) have wide-ranging effects on health and normal function beyond the induction of typical symptoms. For instance, while the diurnal routine of physical work by day and restorative sleep at night is fundamental to human life, preliminary and anecdotal evidence suggests that both are adversely affected in those with CD. Moreover, although it is accepted that CD is associated with poorer quality of life and higher rates of disability compared with healthy controls (HC), the mechanisms underlying this remain under-explored.1 It is hypothesised, therefore, that these impairments could be in part due to the insidious effects of CD on sleep and physical activity. Increasingly, sleep is recognised for its role in immune homoeostasis, yet patients with CD frequently report unrefreshing and/or broken sleep that does not appear to be simply attributable to nocturnal symptoms.2 Sleep dysfunction is reportedly common in patients with CD,3, 4 similar to other intestinal disorders such as irritable bowel syndrome (IBS),5, 6 though the pathogenesis remain unclear.1, 7 Furthermore, in inflammatory bowel disease (IBD), there appears to be a link with poor sleep quality and disease activity/relapse,4, 8 and poor sleep has been associated with severe fatigue and impaired QoL in several recent studies.9–12 Nevertheless, studies hitherto have relied on patient self-report surveys to assess sleep, which are inherently subjective. Data on the nature of and potential contributors to sleep dysfunction in CD are sparse. Moreover, physical activity in patients with CD has received even less attention. However, interest in physical activity research is piqued by the global rise of obesity, diabetes and associated chronic diseases, which have been attributed (along with excesses and changes in dietary intake) to an progressively sedentary lifestyle.13–15 Even in patients with IBD, there is a trend to equivalent or even higher fat mass and/or body mass index compared with age- and sex-matched HC in contrast with the archetypal, malnourished and cachectic IBD patient of times past.16 Research into sleep quality and physical activity is hampered in this field by the difficulties of accurate, objective measurement, the expense and inconvenience of tools such as polysomnography and for the purposes of clinical applicability, the need to measure these in a habitual context with subjects performing their normal daily activities. Triaxial accelerometers overcome many if not all of these obstacles and strongly correlate with

992

‘gold standard’ testing such as formal polysomnography.17, 18 This study aimed first to assess both habitual sleep quality and physical activity over a typical 7-day period utilising accelerometers in CD subjects, comparing these with age- and sex-matched HC, and secondly, to begin to explore potential pathogenic mechanisms associated with hypothesised reduced physical activity and sleep dysfunction in those with CD.

MATERIALS AND METHODS Setting and recruitment Subjects with a confirmed diagnosis of CD according to standard histological, endoscopic and/or imaging criteria were consecutively recruited from the Box Hill Hospital Inflammatory Bowel Disease Clinic, a metropolitan outpatient clinic providing secondary and tertiary IBD care. Relevant clinical and demographical data were extracted from the patients’ medical records and the IBD clinic database. Healthy volunteers were consecutively recruited via advertisement in local hospital and university newsletters/bulk email notices and the local community newspaper. All study participants were aged between 18 and 65 years, were able to understand English and give written, informed consent. Participants were excluded if they had any known major medical comorbidity including neurological/neuromuscular disorders, rheumatological disorders, active ischaemic heart disease or congestive cardiac failure, chronic obstructive lung diseases or active malignancy (excluding skin only) resulting in significant functional limitation, pregnancy and/or consumed >40 g of alcohol per day on average. To avoid bias, healthy volunteers were excluded from recruitment if at initial screening interview they typically engaged in regular high intensity exercise for an equivalent of more than 30 min three or more times per month. This study was approved by the Eastern Health Research & Ethics Committee on 15 June 2009. Study survey The contents and execution of the survey are described elsewhere.9 Briefly, the survey encompassed demographics, medical, psychiatric and medication history and, for CD subjects, further clinical aspects including surgical, admission and medication data. The survey also included the Pittsburgh Sleep Quality Index (PSQI) to measure sleep quality via patient self-report. The PSQI has been widely used, validated and reported in the literature, including in other IBD cohorts.3, 4, 19, 20 It covers

Aliment Pharmacol Ther 2015; 41: 991–1004 ª 2015 John Wiley & Sons Ltd

Sleep and physical activity in Crohn’s disease domains including subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, use/reliance on sleep medications and daytime dysfunction. Item scores are summed to produce a global score with a range of 0–21, where the higher the score, the worse the sleep quality. A total PSQI score ≥6 is considered to be suggestive of significant sleep dysfunction.21 In addition, the survey included assessment of fatigue using the Fatigue Impact Scale (FIS),22 and assessment of concurrent mood disorder with the Hospital Anxiety Depression Scale (HADS)23 as previously applied to IBD populations.10, 24

Study visit After completing the survey, all recruited subjects completed a study visit with an investigator (DvL) where height and weight were measured, and body mass index (BMI) was derived. In CD subjects, disease activity was assessed via the Harvey Bradshaw Index (HBI), with active disease defined as an HBI score ≥5.25 Accelerometer During this same visit, participants were instructed on the use of, and fitted with, a triaxial accelerometer (GT3X, Actigraph, Pensacola FL, USA), which was worn via an elastic belt firmly at the waist. The GT3X records accelerations and converts this digitally into arbitrary measures of movement called ‘counts’, where one count is one occurrence when the magnitude of acceleration reaches a certain threshold per unit time.26 These data are then summed per 1-min intervals (‘epochs’). Thus, the device can accurately and sensitively detect the intensity, extent and duration of movements in three planes. Each device was tested for accuracy prior study commencement. Participants were requested to wear the accelerometer for seven consecutive days, 24 h per day (including during sleep times but not when bathing, swimming or showering as the device is not waterproof) incorporating both week and weekend days. They were also asked to conduct activities representative of their typical routine and lifestyle during the 7-day period. An accelerometer instruction and log sheet was given to each participant, so that the exact times of commencement and completion of wearing the accelerometer, plus occurrences of nonwearing, strenuous exercise and times of retiring and rising from bed for each 24-h period were recorded, enabling cross-referencing with accelerometer data. At completion, participants either returned the accelerometer via mail or directly in person to the investigator within 1 month. The data were uploaded and analysed Aliment Pharmacol Ther 2015; 41: 991–1004 ª 2015 John Wiley & Sons Ltd

using the manufacturer’s proprietary Actilife version 5.1 software (Pensacola, FL, USA). The derived measures of habitual physical activity relevant to this study are outlined in Table 1. Levels of physical activity categorised by accelerometer counts are shown in Table 2. The accelerometer output included measurement of bouts of moderate-vigorous activity, defined as an episode of physical activity at an intensity level of 1953 counts or above for at least 8 min of a consecutive 10-min period. For this study, ‘lower physical activity’ was denoted as those who did not achieve any bouts of moderate-vigorous activity in the 7 days of wearing an accelerometer.

Sleep assessment with accelerometry To assess sleep, the accelerometer data were assessed using the algorithm devised by Sadeh et al.,27 which is provided within the Actilife 5.1 software. The algorithm is computed as follows: PS ¼ 7:601  0:065  MW5  1:08  NAT  0:056  SD6  0:073  lnðACTÞ where PS is the probability of sleep; MW5 is the average number of activity counts during the scored epoch and a window of five epochs preceding and following it; NAT is the number of epochs with activity level equal to or higher than 50 but lower than 100 activity counts in a window of 11 min, including the scored epoch and the five epochs preceding and following it; SD6 is the standard deviation of the activity counts during the scored epoch and the five epochs preceding it; and ln (ACT) is the logarithm of the number of activity counts during the scored epoch + 1. If PS is zero or greater, the specific epoch is scored as sleep; otherwise, it is scored as awake.17 This algorithm gives high agreement rates of around 90% with the gold standard polysomnography in prior calibration studies, including sensitivity for sleep over wakefulness of 97%, with similar output measures as shown in Table 1.17, 28 In this study, ‘severe sleep dysfunction’ was defined as those who with a higher number of awakenings per night, arbitrarily determined by the number of awakenings greater than the 95th percentile of that of the HC group.

Laboratory testing In peripheral blood samples taken at the study visit, complete blood examination, electrolytes including calcium, magnesium, zinc, selenium and iron studies (including ferritin), and serum albumin were measured. Endocrine 993

D. R. van Langenberg et al. Table 1 | Output variables of physical activity and sleep derived from wearing a GT3X accelerometer for 7 days Output variable

Comments/definition

Physical activity Total counts (over 7-day period)*,† Total time per activity level category‡ (in % of total time) Total bouts§ (number achieved in 7-day period)

Measure of overall physical activity Differential breakdown of gradings of physical activity and sedentary activity, establishing patterns of activity Analyses whether subjects are achieving bouts of sustained exercise intensity which are relevant to health and important in activities of daily living/manual work activities

Total time in bouts§ (mins, over 7-day period) Total counts in bouts§ (over 7-day period) Sleep Sleep onset (time/date)* Total sleep time (min)* Sleep latency (min) Awakenings after sleep onset Sleep efficiency (%) Total counts during sleep (7 days)

First epoch (minute) that the device/algorithm senses that subject is asleep Total number of minutes algorithm calculates that subject is asleep Number of minutes from time of going to bed until time of sleep onset Number of awakenings scored by device/algorithm The proportion of time spent asleep divided by the total time in bed (%) The number of activity counts during the time spent asleep

* Also cross-checked with the subject’s log sheet times, adjustments made as appropriate. † Counts = arbitrary units of movement, based on each time that the magnitude of acceleration exceeds a given threshold per unit time. ‡ Accelerometer count activity level categories as described in Table 2.46 § Bouts = defined time periods within which exercise activity is maintained within a range of activity levels. For this study, a participant was deemed to have completed one bout if they had maintained a moderate-vigorous activity level of 1953 counts or above for at least 8 min (80%) of a consecutive 10-min period.

Table 2 | Cut-off levels of activity as per accelerometer counts and MET equivalents with examples for adults as per Actigraph (and used in this study)46, 47 Actigraph activity level/category

Minimum cut-off*

Maximum cut-off*

Sedentary

0

100

Lifestyle Light Moderate

101 760 1953

759 1952 5724

Vigorous

5725

9498

Very vigorous

9499

Equivalent to

Example of activity (METs†)

‘Light’ 6 METs

And higher

* In terms of accelerometer counts † MET = metabolic equivalent of tasks, or the energy cost of physical activities (1 kcal/kg/h).

tests including testosterone, estradiol (for females), sex hormone-binding globulin, thyroid function tests and insulin growth factor-1 (IGF-1) were also performed. All tests were performed using routine methodology at the 994

study centre (Eastern Health Pathology). The thiobarbituric acid-reactive species (TBARS) assay (Sigma-Aldrich, Melbourne, Vic., Australia) was used to assess lipid peroxidation (surrogate marker of oxidative stress) using the Aliment Pharmacol Ther 2015; 41: 991–1004 ª 2015 John Wiley & Sons Ltd

Sleep and physical activity in Crohn’s disease method of Bar-Or et al.29 Serum highly sensitive C-reactive protein (hsCRP) and faecal calprotectin [on faecal sample brought by each subject to study visit (Buhlmann EK-Cal, Sch€ onenbuch, Switzerland)] were tested according to manufacturers’ instructions as markers of systemic and intestinal mucosal inflammation respectively. Cutoffs for active disease as a categorical variable in this study, thus included a HBI ≥5, serum CRP >3 mg/L and/ or faecal calprotectin >100 lg/g.

Statistical analysis Normality of data was assessed by the Shapiro–Wilk test (where P > 0.05 was deemed to represent normality). Since the vast majority of physical activity and sleep variables did not conform to a normal distribution, nonparametric statistics such as medians and Mann–Whitney tests were used throughout to compare the CD and HC groups. Proportions between groups were compared with Fisher exact tests. Bivariate logistic regression analyses [with odds ratio (OR) and corresponding 95% confidence intervals (CI)] and subsequently, multivariate logistic regression analyses were conducted to assess factors associated with dependent variables, lower physical activity and poor sleep quality respectively, in the CD subjects only. IBM-SPSS version 19.0 statistical software (Chicago, IL, USA) was used for all the statistical analyses above. A P value ≤0.05 was deemed to be statistically significant throughout this study. RESULTS Comparison of group characteristics The characteristics of CD (n = 48) and HC subjects (n = 30) who completed all the study investigations are shown in Table 3. Two CD subjects and one HC subject were excluded from these analyses as they did not provide a full 7 days of accelerometer data. There were no statistically significant differences in demographics between the groups, other than gainful employment. Compared to the HC group, those with CD had significantly higher global and physical fatigue scores (FIS survey), as well as higher depression and anxiety scores (HADS) and poorer sleep quality (PSQI). There was no difference in body mass index (BMI) between the groups (Table 4). As expected, the CD participants typically had or tended to have higher markers associated with inflammatory activity. However, there were no statistically significant differences between the groups in haematological, biochemical or micronutrient levels other than serum Aliment Pharmacol Ther 2015; 41: 991–1004 ª 2015 John Wiley & Sons Ltd

zinc and IGF-1 levels, which were both lower in those with CD compared to HC (each P < 0.05).

Accelerometer-derived physical activity All physical activity measures were impaired in the CD compared with the HC group, including a greater time duration spent in sedentary, light or lifestyle activity categories in preference to time in moderate-vigorous activity (as percentage of time worn), median 97.7% (range 91.3–99.7) vs. 96.2% (91.7–99.2), median 2.3% (0.3–8.7) vs. 3.8% (0.8–8.3), and reduced total physical activity (as per total accelerometer counts) over 7 days, median 1.32 9 106 (3.45 9 104–4.13 9 106) vs. 1.95 9 106 (1.10 9 106–3.70 9 106), each P < 0.01. As shown in Figure 1, this was also reflected in comparisons of the frequency of bouts (i.e., 10-min periods of predominantly moderate-vigorous activity). The CD group completed fewer bouts, less total time and less total counts in bouts [median 1.0 (0–25) vs. 5.0 (0–20) bouts, median 7.7 (0–120) vs. 18.5 (0–100) min, median 92 652 (0–2.61 9 106) vs. 437 912 (0–1.36 9 106) counts, each P < 0.05 respectively]. Accelerometer-derived sleep quality Sleep output variables as measured by accelerometer are shown in Figure 2. Total sleep latency (median time taken to get to sleep once in bed per night) was shorter in the CD than the HC group [median 10 (range 0–98) vs. 22 (1–57) min respectively, P = 0.005]. There was no significant difference in total sleep time per night between the groups [median 458 (327–649) vs. 447 (372–507) min respectively, P = 0.56]. However sleep efficiency, defined as the proportion of time spent asleep divided by total time in bed, was lower in CD compared to HC subjects [94.9% (85.6–99.1) vs. 96.2% (90.7–99.4), P = 0.03]. Furthermore, those with CD had a median of 22 (range 1–81) awakenings after sleep onset per night compared to 11 (1–35) awakenings in HC (P = 0.01). Also the median number of minutes awake after sleep onset per night tended to be higher in CD subjects [7 (1–17) min] than in HC subjects [4 (1–15) min per night; P = 0.051]. Factors associated with physical activity and/or sleep dysfunction in Crohn’s disease Within the CD group, there was no significant relationship between measures of disease activity (serum CRP, faecal calprotectin or HBI), either as continuous or dichotomised variables (as per typical cut-offs for active/ 995

D. R. van Langenberg et al. Table 3 | Clinical and demographical characteristics of participants by group Variable Number of subjects Female sex Median age (range) years Married/de facto relationship Completed secondary school Gainful employment Documented psychiatric comorbidity Current smoker Crohn’s disease characteristics* Location L1 ileal L2 ileocolonic L3 colonic L4 upper GI P perianal Behaviour B1 nonstricturing/penetrating B2 stricturing B3 penetrating Median duration IBD (range) years Active disease (as per HBI ≥5)† Prior bowel resection Current medical therapy Oral corticosteroid Thiopurine or methotrexate Anti-TNF therapy Anti-TNF therapy naive

CD

HC

P value – 0.81 0.88 0.14 0.12 0.02 0.19 0.14

48 30 44 28 38 29 15 8

(61%) (21,65) (57%) (80%) (61%) (31%) (17%)

30 20 46 23 28 27 5 1

11 17 20 3 11

(23%) (35%) (42%) (6%) (23%)

– – – – –

– – – – –

29 11 8 14 20 19

(60%) (23%) (17%) (1.46) (42%) (40%)

– – – – – –

– – – – – –

6 26 12 29

(13%) (54%) (25%) (60%)

– – – –

– – – –

(67%) (21,63) (77%) (93%) (90%) (17%) (3%)

* As per Montreal classification. † Harvey Bradshaw Index.

inactive), and physical activity or sleep dysfunction (Figure 3). Furthermore, in a subgroup analysis comparing those CD subjects in deep remission (HBI 3 mg/L or calprotectin alpr lg/g) at the time of study visit, there were no significant differences in any important accelerometer-derived physical activity or sleep quality indices (Table 5). To further assess for factors associated with lower physical activity in CD, bivariate and multivariate logistic regression analyses were performed (Table 6). Lower physical activity, i.e. those who did not achieve one or more bouts of moderate-vigorous activity during 7 days of accelerometry, acted as the dependent variable. Selfreported measures of fatigue were associated with lower physical activity, with mean OR of 5.7 (95% CI = 1.1– 29.1) for global fatigue (total FIS I40) and 3.9 (1.1–14.5) for physical fatigue (physical FIS ≥18). Subsequently, a multivariate analysis was performed with statistical significant (and/or those trending towards, P < 0.10) vari996

ables from the bivariate analyses included (forced entry model used). This showed a longer duration since CD diagnosis (per additional year, despite controlling for age), presence of systemic inflammation (as per serum CRP >3 mg/L) and low vitamin D3 (52 years) was associated with severe sleep dysfunction, but those Aliment Pharmacol Ther 2015; 41: 991–1004 ª 2015 John Wiley & Sons Ltd

Sleep and physical activity in Crohn’s disease Table 4 | Survey, anthropometric and laboratory variables in patients with Crohn’s disease (CD) and healthy controls (HC) Variable* (normal range/units)

CD n = 48

HC n = 30

P value†

Body mass index (kg/m2) Global fatigue score (total FIS, range 0–160) Physical fatigue score (FIS phys, range 0–40) Sleep quality score (PSQI, range 0–21) Depression score (HADS, range 0–21) Anxiety score (HADS, range 0–21) CRP (

Sleep and physical activity measured by accelerometry in Crohn's disease.

Sleep and physical activity are inherent to human living, yet appear affected by Crohn's disease (CD), resulting in fatigue and disability...
322KB Sizes 4 Downloads 9 Views