Background/Aims: Hemodialysis (HD) patients are less active than their healthy counterparts and frequently experience poor sleep. Our aims were to objectively quantify activity and sleep quality in HD patients of an urban population and to determine the effect of providing feedback on activity. Methods: Activity parameters and sleep parameters were collected by a commercially available activity tracker in 29 chronic HD patients. Patients in the feedback group were provided with their activity and sleep data during each HD treatment. Questionnaires were administered at the beginning and at the end of the study. Results: On average, patients walked 8,454 steps/day and slept 349 min/night. Only 28% of the patients were sedentary, defined as walking <5,000 steps/day. Providing feedback did not increase the activity in this urban population. Patients walked significantly less on Sundays compared to other days of the week: 7,024 steps on Sundays vs. 8,633 steps on HD days and 8,732 on non-HD days. It was also found that patients experienced poor sleep quality. HD treatments during shift 1 (6 a.m. to 10 a.m.) interfered with sleep patterns. Most patients reported that physical activity became more important to them after the 5-week period. The tracking device was very well accepted. Conclusion: Interventions to increase physical activity on Sundays could improve physical activity levels overall. Prospective studies are necessary to further explore the use of tracking devices to identify patients at risk and to implement targeted interventions.
Previous research has established that hemodialysis (HD) patients are, on average, less active than their healthy peers, and that disparity in activity levels is more severe in older patients. Also, low physical activity levels may cause muscle atrophy, which in turn may cause reduced physical function in this population . In a study that used the human activity profile (HAP), a questionnaire to measure self-reported physical activity, both the maximum activity score (MAS) and the adjusted activity score (AAS) were directly correlated with albumin and pre-albumin levels, and inversely correlated with C-reactive protein (CRP) levels . Another study in HD patients, which objectively measured physical activity using a pedometer, found that there are specific phenotypes of patients with low physical activity levels. These phenotypes include patients with inflammation, cardiovascular disease, protein energy wasting, obesity, and diabetes . Sedentary HD patients are at a 62% greater risk of mortality than their non-sedentary counterparts. It was found that a sedentary lifestyle has an effect of the same magnitude as a 1 g/dl decrease in serum albumin, a well-established predictor of mortality in the dialysis population .
While it is obvious that activity levels should be increased in dialysis patients, what type of physical activity and how to increase it, is less obvious. An observational analysis of the 2003-2004 National Health and Nutrition Examination Survey examined the effect of replacing sedentary behavior with low intensity, light intensity, or moderate to vigorous intensity activity. It concluded that replacing time spent in a sedentary state with time spent performing light activity (e.g. walking), not low-intensity activity nor moderate to vigorous intensity activity, could result in a survival benefit in patients with chronic kidney disease .
Before interventions to increase physical activity levels can be tested, appropriate means to measure activity must be established. Traditionally, the 3 types of tools used to measure physical activity are self-report, pedometers, and accelerometers. Self-reported physical activity often fails to capture the level of physical activity in this cohort due to ‘floor effects', meaning that the questionnaires utilized cannot detect activity levels that are low . Pedometers, while cost effective and user friendly, are often inaccurate and cannot be worn at all times (e.g. when sleeping or bathing) . Accelerometers are very accurate tools for measuring physical activity levels that have been validated against calorimetric methods and double labeled water techniques . Unfortunately, accelerometers are expensive and burdensome to use. Recently, a new class of tools for measuring activity has emerged. Consumer-grade wearable tracking devices are cost effective, fairly accurate, and easily incorporated into daily life .
The 2 main objectives of this pilot study were to (1) assess physical activity levels of HD patients dwelling in an urban setting and (2) determine if providing feedback on activity during dialysis treatments will affect physical activity levels.
The activity and sleep data of 29 chronic HD patients were collected using the Fitbit® Flex™ (Fitbit, San Francisco, Calif., USA) device over the course of 5 weeks. The participants were randomly assigned to 2 groups after enrollment. The ‘feedback' group (n = 14) received a report of activity and sleep data during each HD treatment, while the control group (n = 15) did not. Questionnaires were administered at the beginning and end of the study period. The study was approved by Western IRB (Protocol No. 20141534) and registered at ClinicalTrials.gov (NCT02320513).
HD patients treated in 2 out-patient facilities located in Manhattan, New York, of the Renal Research Institute (RRI) were recruited to participate in this study. Patient recruitment occurred on a rolling basis from September 2014 to May 2015.
Patients were eligible to participate if they were receiving HD 3 times a week, were on HD for >3 months, and were between 18 and 75 years old. Patients were required to be able to walk without assistance or assistive devices to assure that the device is able to track activity. Patients were excluded if they had unstable health (e.g. acute infections, congestive heart failure (CHF) NYHA class 4 and/or unstable angina), were hospitalized within 3 months before enrollment for non-access related reasons, or were cognitively impaired. The clasp of the device contains trace amounts of nickel, so patients with a known nickel allergy were ineligible to participate. Additionally, patients who had previously worn activity-tracking devices were excluded so the total effect of providing feedback could be observed.
Thirty-three participants were enrolled in the study. Three of them were withdrawn due to non-compliance with study protocol and one due to hospitalization.
Physical Activity and Sleep Data
All participants were equipped with the Fitbit® Flex™ tracking bracelet and were instructed to wear the bracelet at all times, even when bathing, as it is waterproof. This device has been found to be an accurate and reliable method of measuring physical activity levels . The Fitbit® Flex™ measures activity parameters (steps taken, distance traveled) and sleep parameters (minutes asleep, total time in bed). A Fitbit® user account was created and the device was configured for each subject's height, weight, and gender. The device was worn on the non-vascular access arm, and the device settings were configured to reflect if this was the dominant or non-dominant hand. Data were downloaded, from the device to the user account, during each HD treatment. Participants were asked to keep a sleep log, in which they recorded the times when they went to bed and woke up. Sleep start and end times were entered into the user account by the staff in order to collect sleep data. The devices were charged during each HD treatment. Activity and sleep data were exported via Fitbit® Premium for analysis.
Based on their average daily step counts, participants were separated into 3 categories. Participants were considered sedentary, fairly active, or active if they walked <5,000 steps, between 5,000 and 10,000 steps, or >10,000 steps, respectively .
The National Institute of Health (NIH) recommends at least 7 h (420 min) of sleep per night . Sleep efficiency (in %) was calculated as 100 times the ratio of sleep duration to total time in bed. A sleep efficiency of 85% or above is considered good sleep . Based on sleep duration and sleep efficiency, patients were categorized into 3 groups: poor, intermediate, and good sleep quality. Patients who slept <420 min with a sleep quality of <85% were considered to have poor sleep quality. Patients who slept ≥420 min with a sleep efficiency of <85% or slept <420 min with a sleep efficiency of ≥85% were considered to have intermediate sleep quality. Patients who slept ≥420 min with a sleep efficiency of ≥85% were considered to have good sleep quality.
The HAP was administered at the beginning and at the end of the study. The MAS is the highest ranking, or the most rigorous, activity done. The AAS is the MAS less the number of activities, which fall below the MAS, that the patient had stopped doing. The HAP has been shown to effectively estimate physical activity and physical function in the end-stage renal disease population . The ‘Physical Activity Questionnaire' was developed to capture subject attitudes toward their physical activity, and was administered at the end of the study period. The statements ‘I was able to incorporate this device into my daily activities' and ‘I desire to continue wearing this device to track my own activity' were answered only by 22 participants.
Laboratory measurements were done at Spectra Laboratories (New Jersey, N.J., USA) and downloaded to the RRI data warehouse. For study purposes, clinical and laboratory data were imported electronically from the RRI data warehouse. In addition to collecting test results from routine monthly blood tests, pre-albumin, CRP, and interleukin-6 (IL-6) levels were also tested. Blood samples were collected before the start of the mid-week dialysis session.
Baseline (first week in the study) demographics, anthropometrics, comorbidities, treatment-related parameters, activity parameters, and sleep parameters were described by mean and SD for continuous variables and frequency distribution for categorical variables. Two independent sample t-test was used to test statistical significant differences between the feedback and control group for continuous variables; Fisher's exact test was used for categorical variables (table 1). Paired t tests were also performed to test the differences among the daily average steps on HD days, non-HD days, and Sundays. Sundays were excluded from non-HD days.
In order to assess the effect of HD scheduling on sleep quality, we also performed paired t test to compare the sleep duration and sleep efficiency between the night after HD treatment and the night before HD treatment, night after HD treatment and night between 2 non-HD days, and night before HD treatment and night between 2 non-HD days per shift.
Statistical analysis was performed in SAS version 9.4 and Rx64 3.2.0.
Demographics, anthropometrics, treatment-related parameters, and laboratory parameters of study participants are presented in table 1. On average, participants were 53 years old, with a body mass index of 25.8 kg/m2, and 173 cm tall. The study cohort was 55.2% male and 75.9% black; 17.2% of the participants were diabetic and 27.6% had CHF. Average HD treatment time was 218 min with an average eKdrt/V of 1.44, and dialysis vintage was 6.6 years. Average CRP and IL-6 levels were 7.4 mg/l and 9.6 pg/ml, respectively. Average levels of pre-albumin, serum albumin, and hemoglobin were 35.2 mg/dl, 4.1 and 11.3 g/dl, respectively.
Patients received treatments at the HD clinic during 4 different shifts, namely shift 1 from 6 a.m. to 10 a.m., shift 2 from 10 a.m. to 2 p.m., shift 3 from 2 p.m. to 6 p.m., and shift 4 from 6 p.m. to 10 p.m. The number of patients who were scheduled to receive HD treatments during shifts 1, 2, 3, and 4 were 9, 8, 6, and 6, respectively.
Participants walked an average of 8,454 steps and 6.0 km per day (table 1). Since there were no significant differences in activity and sleep between the feedback and control groups, nor between the first week of the study and weeks 2-5 of the study, all 29 patients were pooled for analysis. The participants were separated into 3 groups based on average steps/day walked. Of the 29 participants, 28% (n = 8) were sedentary, 24% (n = 7) were fairly active, and 48% (n = 14) were active (fig. 1). In addition, the average daily steps walked directly correlated (R2 = 0.43, p < 0.001) with AAS scores from HAP (fig. 2).
Individual patients' activity levels varied greatly. The patient with the highest average daily steps walked 16,116 steps/day, while the patient with the lowest average daily steps walked 2,578 steps/day (fig. 3; note that the scales of the y-axis differ between panels a and b).
On average, participants walked significantly less on Sundays compared to other days of the week, regardless of whether it was a HD or a non-HD day (fig. 4a). Participants walked 1,608 steps less on Sundays compared to HD days and 1,707 steps less on Sundays compared to non-HD days. However, when separated by activity category, that is, sedentary, fairly active, and active, a significantly lower step count on Sunday was only observed in the active group (fig. 4b-d).
On average, patients slept for 349 min/night with a sleep efficiency of 82%. The proportion of participants who experienced poor, intermediate, or good sleep quality were 58% (n = 17), 38% (n = 11), and 3% (n = 1), respectively (fig. 5).
When separated by treatment shift, patients who received treatments during shift 1 (6 a.m. to 10 a.m.), on average slept significantly less on nights before HD treatment than on nights after HD treatment and nights between 2 non-HD days (table 2a). This was also observed when patients were analyzed individually (fig. 6a). Though differences between night after HD, night before HD, and nights between 2 non-HD days in shifts 2 and 3 were statistically significant, these do not seem to be clinically relevant. Besides lower sleep efficiency in nights before HD than nights after HD in shift 1, there were no other significant differences in sleep efficiency within the different shifts (table 2b).
Physical Activity Questionnaire
At the end of the 5-week study period, participants were read several statements and asked how much they agreed with the statement. They could rank their level of agreement in 5 categories: ‘not at all', ‘somewhat', ‘moderately', ‘definitely', or ‘most definitely'. More than half the patients felt that they did walk more than usual during the study period. Furthermore, a majority of the patients felt that it was easy to incorporate the device into their lifestyle and felt that they wanted to continue wearing the activity-tracking device (table 3).
The main findings of our study are that (a) the proportion of HD patients who exceeded the WHO recommendation of 10,000 steps/day was greater than observed in other HD populations [3,15]; (b) patients walked significantly less on Sundays compared to other days of the week; and (c) that patients in our population experience poor sleep quality, corroborating findings in other HD populations .
The feedback provided to the participants about their physical activity levels did not appear to make a difference in the number of steps walked (table 1). One reason as to why patients in the feedback group did not walk more than the control group could be that someone keeping track of their physical activity motivated the control group to walk more (Hawthorne effect). The fact that more than half of the patients reported that they have walked more than usual supports this claim. Another reason could be that feedback was not given in real-time, and instead was given thrice weekly only at every HD treatment.
In a prospective national French epidemiological study, Panaye et al.  found that 94% of the HD population walked <10,000 steps/day and 64% of the population were sedentary, walking <5,000 steps/day. In our HD population, 52% of patients walked <10,000 steps/day, and only 28% of patients walked <5,000 steps/day. A possible explanation for this difference could be that our patients lived in an urban environment (New York, NY, USA), while the French study included patients from rural and suburban areas. Another possible reason could be that the average age of the patients in our study was almost 10 years younger than that of the sample studied by Panaye et al. [3.]
Majchrzak et al.  found in a cross-sectional single-center study that physical activity is lower on dialysis days compared to non-dialysis days; one reason given for this difference was the time spent inactive during dialysis treatments. Surprisingly, there was no difference in the steps walked on dialysis vs. non-dialysis days in our population. A reason for this could be that since our patients were located in New York City, many of them walked and/or took public transportation (i.e. subway, bus) to get to the dialysis clinic. This extra activity on dialysis days may have compensated for the inactivity during dialysis treatments. There was however a significant decrease in steps walked on Sundays. This effect, to the best of our knowledge, has not yet been observed in HD populations. Future interventions to increase activity levels could specifically target Sundays.
The majority of patients slept less than the NIH recommendation of 420 min (7 h) and experienced poor sleep efficiency (minutes asleep/total time in bed ×100 <85%) [12,13]. It is well known that HD patients frequently experience sleep disorders . A main cause of abnormal sleep in this population is restless leg syndrome . Another potential reason for poor sleep is disturbed circadian rhythm caused by uremia and dialysis treatments, although the exact mechanism(s) for this phenomena (are) still unknown . Patients who received dialysis treatments very early in the morning, during shift 1, slept for a significantly shorter duration on nights before HD treatments than nights after HD treatments or nights between 2 non-HD days. This irregular pattern of sleeping for a short interval on one night and for a much longer interval on another night could be an additional reason as to why HD patients experienced poor sleep.
It has been found that the Fitbit® Flex™ overestimates sleep duration and quality due to its limited ability in sensing when the wearer is awake . However, the device used does correlate directionally with actual sleep. Polysomnography, the gold-standard for measuring sleep, is time consuming, impractical on a routine basis, and expensive. The Fitbit® Flex™ can be used to measure sleep, as long as the user is aware that the data collected indicate an overestimation of sleep duration and quality. Consequently, our measurements of sleep duration and quality, which are already low, may have been overestimated.
There were several limitations to this pilot study. First, the sample size was small. Now that we know the tracking device is fairly easy to use in HD patients, we can include more patients in order to better observe the effect of feedback, if any, on physical activity levels. Moving forward, it will be also important to increase the diversity of patients enrolled. The second limitation was that patients were enrolled on a rolling basis over a year-long period. This means that activity levels of some of the patients were collected during the autumn/spring time while the activity levels of others were collected during the winter. In the winter of 2014-2015, there were several blizzards that hit New York City. This inclement weather could have caused lower physical activity levels in some patients. Another limitation was the potential selection bias in our cohort, since patients who were eligible to participate in this study were healthier than the average HD patient. They had not been hospitalized for at least 3 months before the study and were able to ambulate independently. Additionally, there was an inherent selection bias since not all patients who were approached agreed to participate in the study. Perhaps the patients who agreed to participate in the study were more interested in learning about their physical activity compared to their peers. Finally, the observation period of 5 weeks was relatively short. More patterns in physical activity and sleep of HD patients could be detected if the data were collected over a longer period of time.
Future research is required to address the clinical use of data gathered by tracking devices, in particular for identification of patients at risk for adverse outcomes and to deploy targeted interventions, for example, strategies to improve activity and sleep. While not insurmountable, operationalization and scaling of the routine clinical use of tracking devices will present additional layers of challenges and complexities.
HD patients are significantly less active on Sundays compared to the other days of the week. Interventions to increase physical activity on Sundays could improve physical activity levels overall. HD patients also experience poor quantity and quality of sleep. Prospective studies are necessary to further explore the use of tracking devices to identify patients at risk and to implement more targeted interventions.
PK holds stock in Fresenius Medical Care. The remaining authors declared no competing interest.