Introduction
Physical activity levels are low among patients with ESKD and are an important contributor to reduced quality of life and poor cardiovascular outcomes (1,2). Although exercise interventions have improved activity levels and physical function, those designed for the general population may not be feasible for many patients with ESKD (3,4); in contrast, low intensity activity is feasible and has demonstrated benefit (5). Furthermore, a complete understanding of the factors contributing to low activity in the dialysis population is lacking. This can be addressed using highly granular data collection to achieve a thorough understanding of activity patterns and their determinants in patients receiving dialysis.
Increasing adoption of wearable technologies provides a unique opportunity to understand activity patterns and their relation to clinical outcomes. Prior studies in the hemodialysis population have used wearable devices to examine physical activity levels (6), but only one has attempted long-term use (7), and few have utilized the detailed data that such devices provide (8,9). We hypothesized that highly granular data collection coupled with patient-centered outcomes measures would yield new insights into patterns of behavior and activity among patients receiving hemodialysis. We sought to test the feasibility of continuous long-term physical activity monitoring using wrist-worn accelerometers in a representative cohort of patients receiving hemodialysis and to examine daily patterns of activity in the context of patient-reported symptoms and physical functioning.
Materials and Methods
Study Population
This pilot study was conducted at a single outpatient dialysis unit in the Bronx, New York, from May 2018 through January 2019. Patients were eligible if they had ESKD and had been receiving thrice-weekly hemodialysis for at least 3 months, were ambulatory, were ≥18 years of age, and were willing to wear a wrist-worn accelerometer for 6 months. The use of an assistive device for walking was permitted. The study sought to enroll 50 participants from the first three shifts on each thrice-weekly schedule (first patient on at 4:30am, last patient off between 5:00pm and 7:00pm), which included a maximum of 144 patients if all spots were filled. The fourth shift was not included due to resource limitations. Sixty patients were assessed for eligibility, and 52 met the eligibility criteria and were enrolled. Four patients were excluded from this analysis due to lack of sufficient data (defined as <1 month of accelerometry data): two patients withdrew from the study within 2 weeks of enrollment, and the other two were not adherent with wearing the accelerometer. Thus, there were 48 patients in the final study cohort (Figure 1). The study was approved by the Institutional Review Board of the Albert Einstein College of Medicine and conducted in adherence to the Declaration of Helsinki, and all participants provided written informed consent.
;)
Figure 1.: Flow diagram of study participation.
Study Design
Data regarding demographic information, including age, sex, self-reported race and ethnicity, medical history, and dialysis treatments, were collected from patients’ medical records. Ultrafiltration volume was defined as the difference between pre- and postdialysis weight, and ultrafiltration rate as ultrafiltration volume divided by dialysis duration and predialysis weight. Baseline ultrafiltration rate was defined as the mean value during the first 2 weeks of the study. Race and ethnicity were self-identified. During enrollment, patients were asked about employment status and transportation to and from the dialysis center, and physical function was assessed using the Short Physical Performance Battery (SPPB) (10). A commercially available activity monitor (Garmin Vivofit 2, Garmin Ltd., Olathe, KS) was placed on each participant’s non-dominant or non-access arm during the enrollment visit. Each device was set to display the time rather than step counts. Participants were asked to wear the device continuously throughout the course of the study, which was 6 months in duration. No device charging was required due to the monitor’s 1-year battery life. To facilitate automated data retrieval when patients came to the dialysis unit for treatment, a wireless hub was installed to which each Garmin device automatically connected when within range.
Accelerometer Data
Accelerometry data were used to calculate total daily step counts and the number of steps taken during each 15-minute epoch during the study. Data were excluded from any day with a total step count below 200, indicative of nonadherence wearing the accelerometer. Baseline levels of activity were defined as the mean daily step count during the first 2 weeks of study participation. A small number of steps were recorded during participants’ dialysis sessions; as these were considered implausible and likely due to arm movement, activity level was set to zero during hemodialysis treatments. Time intervals on nondialysis days were defined relative to the time of day that dialysis ended on the previous dialysis day. Adherence was defined as number of days with step counts recorded divided by total number of days in the study. At the conclusion of the study, participants were asked about their experience wearing the activity monitor, including barriers to adherence, “How much did it bother you?” (1–10 scale, where 1=“not at all” and 10=“very much”), and “When did you take it off?”
Recovery Time Assessment
Time to recovery from a dialysis treatment was assessed during enrollment and at monthly study visits in the dialysis unit by asking participants “How long does it take you to recover from a dialysis session?” (11). Responses were categorized as 0–15 minutes, >15 minutes–2 hours, >2–6 hours, and >6 hours, and were further classified as fast recovery (≤2 hours) or slow recovery (>2 hours).
Statistical Analyses
Baseline characteristics were compared between participants categorized by median step count or recovery time using the chi-squared test or Fisher’s exact test for categorical variables and two-tailed t tests or Wilcoxon rank-sum tests for continuous variables. Differences in activity levels between dialysis and nondialysis days were analyzed using mixed-effects models specifying random intercepts to account for within-person correlation and adjusted for time. Covariates were selected a priori as possible confounders of the associations of the dialysis treatment, physical function, and recovery time with activity level and included age, sex, race and ethnicity, body mass index (BMI), diabetes status, and ultrafiltration rate, for which the value for each dialysis treatment was applied to subsequent interdialytic nondialysis days. Bar and box plots were created by calculating the within-person mean across all study days and then calculating summary statistics for each category. Heat maps were constructed using the mean category step count within each 15-minute interval relative to the end of dialysis. All analyses were performed using Stata v13.1 (StataCorp, College Station, TX). A P value of <0.05 was considered statistically significant.
Results
Participant Characteristics
The mean age of the study population was 60±14 years, 50% were women, 60% were Black, and 15% were Hispanic. The majority had ESKD due to diabetes or hypertension (Table 1). The overwhelming majority of patients reported not working. Most were driven to dialysis via paratransit or by a family member or acquaintance. Participants with lower levels of physical activity were older, had higher BMI, and were more likely to have coronary artery disease and congestive heart failure. They were also more likely to use paratransit and to require an assistive walking device, and had worse performance on the SPPB, with 52% having scores of 6 or lower.
Table 1. -
Baseline characteristics by median daily step count
| Characteristic |
Overall (N=48) |
≤3808 Steps per Day (N=23) |
>3808 Steps per Day (N=25) |
P
|
| Age (yr), mean±SD |
60±14 |
65±14 |
55±11 |
0.004 |
|
Sex, n (%)
|
|
|
|
0.77 |
| Women |
24 (50) |
11 (48) |
13 (52) |
|
| Men |
24 (50) |
13 (52) |
11 (48) |
|
|
Race and ethnicity, n (%)
|
|
|
|
0.21 |
| Asian |
3 (6) |
3 (13) |
0 (0) |
|
| Black |
29 (60) |
14 (61) |
15 (60) |
|
| Hispanic |
7 (15) |
2 (9) |
5 (20) |
|
| Multiracial |
5 (10) |
1 (4) |
4 (16) |
|
| Other |
1 (2) |
1 (4) |
0 (0) |
|
| White |
3 (6) |
2 (9) |
1 (4) |
|
| BMI (kg/m2), mean±SD |
28.8±6.9 |
30.3±6.5 |
27.5±7.1 |
0.06 |
| Employed (N=46), n (%) |
2 (4) |
0 (0) |
2 (9) |
0.15 |
|
Transportation modality to/from dialysis, n (%)
|
|
|
|
0.003 |
| Paratransit |
19 (40) |
15 (65) |
4 (16) |
|
| Someone drives me |
21 (44) |
7 (30) |
14 (56) |
|
| I drive myself |
5 (10) |
1 (4) |
4 (16) |
|
| Other |
3 (6) |
0 (0) |
3 (12) |
|
|
Assistive walking device (N=47), n (%)
|
|
|
|
<0.001 |
| None |
33 (70) |
11 (48) |
22 (92) |
|
| Cane |
8 (17) |
6 (26) |
2 (8) |
|
| Walker |
6 (13) |
6 (26) |
0 (0) |
|
|
Smoker, n (%)
|
|
|
|
0.82 |
| Former |
23 (48) |
12 (52) |
11 (44) |
|
| Current |
4 (8) |
2 (9) |
2 (8) |
|
| Never |
21 (44) |
9 (39) |
12 (48) |
|
|
ESKD etiology, n (%)
|
|
|
|
0.96 |
| Diabetes mellitus |
18 (38) |
9 (39) |
9 (36) |
|
| Hypertension |
9 (19) |
5 (22) |
4 (16) |
|
| Glomerulonephritis |
9 (19) |
4 (17) |
5 (20) |
|
| Polycystic kidney |
4 (8) |
2 (9) |
2 (8) |
|
| Other |
8 (17) |
3 (13) |
5 (20) |
|
|
Dialysis access, n (%)
|
|
|
|
0.32 |
| AVF |
33 (69) |
13 (57) |
20 (80) |
|
| AVG |
11 (23) |
7 (30) |
4 (16) |
|
| Permacath |
4 (8) |
3 (13) |
1 (4) |
|
| Dialysis duration (min), mean±SD |
231±16.9 |
227±19.4 |
235±13.5 |
0.09 |
| Ultrafiltration rate (ml/kg per hour), mean±SD |
7.43±2.5 |
6.96±2.62 |
7.87±2.28 |
0.2 |
| Diabetes, n (%) |
22 (46) |
12 (52) |
10 (40) |
0.4 |
| Hypertension, n (%) |
46 (96) |
22 (96) |
24 (96) |
0.95 |
| Coronary artery disease, n (%) |
19 (40) |
11 (48) |
7 (28) |
0.09 |
| Congestive heart failure, n (%) |
8 (17) |
6 (26) |
2 (8) |
0.09 |
| Peripheral vascular disease, n (%) |
11 (23) |
6 (26) |
5 (20) |
0.62 |
| History of cancer, n (%) |
7 (15) |
4 (17) |
3 (12) |
0.6 |
| HIV, n (%) |
4 (8) |
0 (0) |
4 (16) |
0.11 |
| Stroke, n (%) |
7 (15) |
4 (17) |
3 (12) |
0.6 |
| SPPB (continuous; N=46) |
7.2±2.6 |
6.3±2.8 |
8.2±1.8 |
<0.001 |
|
SPPB (N=46), n (%)
|
| 0–6 |
14 (30) |
12 (52) |
2 (9) |
0.005 |
| 7–9 |
22 (48) |
7 (30) |
15 (65) |
|
| 10–12 |
10 (21) |
4 (17) |
6 (26) |
|
| Daily step count, mean±SD |
4495±2952 |
2313±969 |
6590±2686 |
<0.001 |
Percentages may not sum to 100% due to rounding. BMI, body mass index; AVF, arteriovenous fistula; AVG, arteriovenous graft; SPPB, Short Physical Performance Battery.
Adherence
The median number of days wearing the accelerometer among the 48 participants was 199 (interquartile range [IQR] 169–214) with a median adherence of 0.95 (IQR 0.86–1). Data regarding feasibility were available from 43 patients at the conclusion of the study. Nearly all (n=42) reported wearing the accelerometer every day: of these, 31 patients continuously wore the device; of the 11 who removed it, four did so only when showering, two removed it while sleeping, three during hospitalizations, and the remainder for miscellaneous reasons. Thirty-nine patients (of 42 responders) reported that the accelerometer did not bother them at all (1 on a 10-point scale); the remaining responses were 2 (n=1) and 7 (n=2). Thirty-five patients reported no barriers to adherence; among the remaining patients, barriers included itchiness or skin irritation, difficulty reading the numbers on the device, and difficulty putting it on after it had come off.
Physical Activity Levels Relative to Dialysis Treatment
Total daily step counts throughout the study period were lower on dialysis days compared with nondialysis days (3991 [95% confidence interval (CI), 3187 to 4796] versus 4561 [95% CI, 3757 to 5365] steps per day; P<0.001). We expected this would be driven not only by sedentary time during dialysis but also by a reduction in activity after dialysis; however, during the 6 hours immediately after dialysis, participants took only 13 fewer steps per hour (95% CI, 3 to 23) compared with the corresponding time period on nondialysis days (Figure 2A). Further investigation revealed an unexpected effect of time (P<0.001 for interaction): during the hour immediately after dialysis, step-count intensity was 74% higher on dialysis days than during the corresponding time on nondialysis days (increase of 188 steps per hour [95% CI, 170 to 205]); in contrast, during the ensuing 5 hours after dialysis, step-count intensity was lower compared with nondialysis days (Figure 2A). Graphical representation of step-count intensity relative to completion of dialysis confirmed increased activity levels immediately after dialysis treatment; these further demonstrated that greater intensity was achieved during this time period than at any other time during the week (Figure 2, B and C). Of note, a similar, more modest increase was also observed in the hours before a dialysis treatment: among patients on the second and third shifts, step-count intensity was 50 steps greater (95% CI, 29 to 72) during the hour before dialysis compared with the corresponding time on nondialysis days (patients on the first shift were excluded because the nondialysis comparator would be sleep time). These differences remained significant after multivariable adjustment (Table 2). Because low daily physical activity was strongly associated with poor physical function, we hypothesized the postdialysis increase in activity would be greatest among patients with better preserved physical function. Indeed, despite lower activity overall on dialysis days among patients with an SPPB of 7 or more (Figure 3A), the increment in step count after dialysis was largest in these patients and was independent of demographic characteristics, BMI, diabetes status, and ultrafiltration rate (Figure 3B, Table 2). Nevertheless, even patients with a SPPB of 6 or less increased step-count intensity after dialysis (Figure 3B, Table 2), and all SPPB groups became less active during the ensuing 5 hours (Figure 3C, Table 2). In addition, patients who used paratransit had a smaller increase in step counts during the hour after dialysis (88 steps per hour; 95% CI, 68 to 109) than patients who did not use paratransit (257 step per hour; 95% CI, 231 to 282; P<0.001 for interaction).
Figure 2.: Step-count intensity varies with time after hemodialysis treatment. (A) Comparison of within-person mean step counts throughout the duration of the study during the 6 hours immediately after a dialysis treatment. Boxes represent median and interquartile range; whiskers represent 95% confidence intervals. (B) Mean within-person step counts for each 15-minute epoch throughout the study relative to the end of dialysis for days with and without dialysis treatment. (C) Heat map of mean step counts throughout the study relative to the end of dialysis for days with and without dialysis treatment. Time intervals on nondialysis days were defined relative to the time of day that dialysis ended on the previous dialysis day. **P<0.01; ***P<0.001.
Table 2. -
Step-count difference between dialysis and nondialysis days relative to time post dialysis
| Time Period, hr |
Step-count Difference (per Hour) |
P
|
P for interaction |
|
Time post dialysis
|
|
|
<0.001b |
|
0–1 |
188 (171–205) |
<0.001 |
|
|
1–6 |
−54 (−62 to −46) |
<0.001 |
|
|
SPPB categories by time
|
|
0–1 |
|
|
<0.001c |
|
0–6
|
84 (52–115) |
<0.001 |
|
|
7–9
|
262 (237–286) |
<0.001 |
|
|
10–12
|
189 (149–230) |
<0.001 |
|
|
1–6 |
|
|
<0.001c |
|
0–6
|
−21 (−36 to −7) |
0.005 |
|
|
7–9
|
−49 (−61 to −38) |
<0.001 |
|
|
10–12
|
−113 (−131 to −94) |
<0.001 |
|
|
Recovery categories by timea |
|
0–1 |
|
|
<0.001c |
|
Fast
|
225 (203–248) |
<0.001 |
|
|
Slow
|
134 (107–161) |
<0.001 |
|
|
1–6 |
|
|
<0.001c |
|
Fast
|
−61 (−71 to −50) |
<0.001 |
|
|
Slow
|
−44 (−56 to −31) |
<0.001 |
|
Multivariable model adjusted for age, sex, race and ethnicity, date, diabetes, body mass index, and ultrafiltration rate. SPPB, Short Physical Performance Battery.
aRecovery time: fast, ≤2 hours; slow, >2 hours.
bEffect modification by time post dialysis of the association of dialysis day status (versus nondialysis day) with activity level.
cEffect modification by SPPB score or recovery time, respectively, of the association of dialysis day status (versus nondialysis day) with activity level.
Figure 3.: Step counts between dialysis and nondialysis days separated by Short Physical Performance Battery (SPPB) category. (A) Daily step counts by SPPB category for days with and without dialysis treatment. (B) Step-count intensity during the first hour after dialysis by SPPB category for days with and without dialysis treatment. (C) Step-count intensity during hours 1–6 after dialysis by SPPB category for days with and without dialysis treatment. Boxes represent median and interquartile range; whiskers represent 95% confidence intervals. Time intervals on nondialysis days were defined relative to the time of day that dialysis ended on the previous dialysis day. SPPB category 0–6 (n=14), 7–9 (n=22), 10–12 (n=10). ***P<0.001.
Physical Activity Levels and Postdialysis Recovery Time
After dialysis treatments, many patients report fatigue and other symptoms that may be slow to resolve, which is captured by the time to recovery; therefore, we hypothesized that a postdialysis surge in activity would be limited to patients who experienced little burden from such symptoms and reported fast recovery time. Baseline characteristics were not meaningfully different on the basis of recovery time, except patients who reported >2 hours of recovery time were more likely to be men and to have diabetes and had modestly lower SPPB scores compared with patients reporting ≤2 hours recovery time (Table 3). Patients who reported slow recovery were also less active overall than other patients (3400 [95% CI, 2171 to 4628] versus 4894 [95% CI, 3900 to 5889] steps per day; P<0.001). However, patients with both fast and slow recovery times increased activity level during the hour after dialysis (Figure 4A) by 225 steps (95% CI, 203 to 248) and 134 steps (95% CI, 107 to 161), respectively. Notably, this resulted in a level of intensity for the slow recovery group during the first hour after dialysis that was greater than the average intensity over the 6 hours post dialysis for the fast recovery group and similar to their activity level on nondialysis days (Figure 4A). Activity level decreased during the subsequent 5 hours in both groups. These associations were unchanged after multivariable adjustment (Table 2). Highly granular visual representation of the data demonstrated that patients who reported the fastest recovery times demonstrated the largest increases in activity; nevertheless, even patients who reported >6 hours recovery time achieved their highest intensity of physical activity in the hour immediately after a dialysis treatment (Figure 4, B and C).
Table 3. -
Baseline characteristics by dialysis recovery time
| Characteristic |
≤2 Hours (N=29) |
>2 Hours (N=19) |
P
|
| Age (yr), mean±SD |
61±13 |
57±14 |
0.33 |
|
Sex, n (%)
|
|
|
0.14 |
|
Women |
12 (41) |
12 (63) |
|
|
Men |
17(59) |
7 (37) |
|
|
Race and ethnicity, n (%)
|
|
|
0.73 |
|
Asian |
2 (7) |
1 (5) |
|
|
Black |
17 (59) |
12 (63) |
|
|
Hispanic |
4 (14) |
3 (16) |
|
|
Multiracial |
3 (10) |
2 (10) |
|
|
Other |
0 (2) |
1 (5) |
|
|
White |
3 (10) |
0 (0) |
|
| BMI (kg/m2), mean±SD |
28.6±7.6 |
29.2±5.9 |
0.51 |
| Employed (N=46), n (%) |
1 (4) |
1 (6) |
|
|
Transportation modality to/from dialysis, n (%)
|
|
|
0.79 |
|
Paratransit |
12 (41) |
7 (37) |
|
|
Someone drives me |
13 (45) |
8 (42) |
|
|
I drive myself |
2 (7) |
3 (16) |
|
|
Other |
2 (7) |
1 (5) |
|
|
Assistive walking device (N=47), n (%)
|
|
|
0.97 |
|
None |
20 (71) |
13 (68) |
|
|
Cane |
4 (14) |
4 (21) |
|
|
Walker |
4 (14) |
2 (11) |
|
|
Smoker, n (%)
|
|
|
0.57 |
|
Former |
15 (52) |
8 (42) |
|
|
Current |
3 (10) |
1 (5) |
|
|
Never |
11 (38) |
10 (53) |
|
|
ESKD etiology, n (%)
|
|
Diabetes mellitus |
|
|
0.11 |
|
Hypertension |
9 (31) |
9 (47) |
|
|
Glomerulonephritis |
8 (28) |
1 (5) |
|
|
Polycystic kidney |
7 (24) |
2 (11) |
|
|
Other |
1 (3) |
3 (16) |
|
|
Dialysis access, n (%)
|
|
|
0.09 |
|
AVF |
17 (59) |
17 (84) |
|
|
AVG |
8 (28) |
3 (16) |
|
|
Permacath |
4 (14) |
0 (0) |
|
| Dialysis duration (min), mean±SD |
231±18 |
232±16 |
0.78 |
| Ultrafiltration rate (ml/kg per hour), mean±SD |
7.59±2.5 |
7.19±2.5 |
0.59 |
| Diabetes, n (%) |
11 (38) |
11 (58) |
0.18 |
| Hypertension, n (%) |
29 (100) |
17 (89) |
0.07 |
| Coronary artery disease, n (%) |
12 (41) |
7 (37) |
0.75 |
| Congestive heart failure, n (%) |
5 (17) |
3 (16) |
0.9 |
| Peripheral vascular disease, n (%) |
7 (24) |
4 (21) |
0.8 |
| History of cancer, n (%) |
5 (17) |
2 (11) |
0.52 |
| HIV, n (%) |
3 (10) |
1 (5) |
0.53 |
| Stroke, n (%) |
3 (10) |
4 (21) |
0.3 |
| SPPB (continuous; N=46), mean±SD |
7.6±2.2 |
6.7±2.9 |
<0.001 |
|
SPPB (N=46), n (%)
|
|
|
0.32 |
|
0–6 |
6 (22) |
8 (42) |
|
|
7–9 |
15 (56) |
7 (37) |
|
|
10–12 |
6 (22) |
4 (21) |
|
| Daily step count, mean±SD |
5062±3334 |
3677±2142 |
<0.001 |
Percentages may not sum to 100% due to rounding. BMI, body mass index; AVF, arteriovenous fistula; AVG, arteriovenous graft; SPPB, Short Physical Performance Battery.
Figure 4.: Step-count intensity between dialysis and nondialysis days separated by dialysis recovery time status. (A) Step-count intensity during the 6 hours after a dialysis treatment for days with and without dialysis, separated by recovery time status. Boxes represent median and interquartile range; whiskers represent 95% confidence intervals. (B) Heat map of mean step counts relative to the end of dialysis for dialysis days separated by recovery time category. (C) Heat map of mean step counts relative to the end of dialysis for nondialysis days separated by recovery time category. Time intervals on nondialysis days were defined relative to the time of day that dialysis ended on the previous dialysis day. *P<0.05; **P<0.01; ***P<0.001.
Discussion
In this proof-of-concept study, we found that long-term continuous use of wrist-worn accelerometers was feasible and well tolerated in patients receiving hemodialysis. Adherence with the study devices was high, enabling continuous monitoring of physical activity patterns in a real-world setting. Remarkably, although patients were less active overall on dialysis days, they were most active during the hour immediately after dialysis treatments. Increases in intensity of activity were most pronounced among patients with relatively preserved physical function and those reporting fast postdialysis recovery time but were not limited to these groups. Thus, continuous long-term monitoring coupled with highly granular data collection identified an unexpected period of increased physical activity after dialysis. Importantly, these findings are likely generalizable to many other patients receiving hemodialysis in the United States: the study was conducted in an urban dialysis unit with a diverse population and high burden of disability and comorbidity; the vast majority of patients approached about the study were open to participating; and we enrolled more than one third of all patients receiving treatment during the eligible dialysis shifts. Taken together, these findings highlight the potential of wearable technologies to yield new insights into patient behaviors, drivers of inactivity, and opportunities to promote physical activity. Such insights could be useful not only for researchers but also for clinicians attempting to understand the effect of dialysis on their patients’ lives and lived experiences. Therefore, this work may provide impetus for additional research testing the adoption of wearable devices in ESKD care.
The incorporation of physical activity monitoring into clinical care has the potential to improve outcomes (12,13), but it requires consistent use of a monitoring device by patients over a long time period. However, the majority of studies quantifying activity levels in patients receiving hemodialysis have been short term, in the order of 1–2 weeks or less (6). Many used pedometers, which assess uniaxial motion of the hip (6). An important limitation of pedometers is undercounting steps in individuals with short step length and slow gait speed (14). This is particularly relevant among individuals with advanced kidney disease, in whom such gait abnormalities are common (151617–18). More recently, several studies have used triaxial wrist-worn accelerometers and monitored patients for 4–5 weeks (19202122–23). Only one prior study, a randomized trial of an intervention to increase daily walking, attempted a longer monitoring period of 3 months using pedometers worn around the waist (7). Thus, our study is the first to test long-term use of a device that would be acceptable to most patients in day-to-day life.
Wearable technologies are of particular interest in ESKD patients due to their substantial burden of chronic disease and frequent contact with the medical system. Modern accelerometers allow for automated data transfer, enabling ease of use by patients and straightforward monitoring by clinicians. Our ability to extract data reliably over a 6-month period was enabled by automated device syncing during dialysis treatments; in radiation oncology patients, another population with high frequency of contact with the medical system, continuous activity monitoring with automated syncing enhanced prediction of long-term outcomes (24,25). Device fatigue was not reported among our participants, and most had little or no difficulty with long-term adherence. The presence of an upper extremity arteriovenous access could present a barrier to accelerometer use, but 92% of our patients had an arteriovenous fistula or graft, and this did not affect adherence. The high rate of adherence was also facilitated by devices with long battery life, so patients did not need to charge them. Overall, our results demonstrate that low-maintenance wrist-worn devices are acceptable to patients for long-term use, and an automated data-collection process can be instituted to maintain high data fidelity.
Our study is unique in combining long-term accelerometry use with highly granular data visualization and analysis. Only one prior study examining step counts in patients receiving hemodialysis has taken advantage of the tremendously detailed data that can be extracted from triaxial accelerometers. In the HDFIT trial comparing hemodiafiltration with hemodialysis, patients wore Actigraph devices around their waists for 1-week periods; interestingly, physical activity levels were highest during the first 60–90 minutes after treatment and declined markedly after 2 hours post dialysis (8,26,27). The HDFIT participants were younger and had less comorbidity than our cohort, and one third used public transit to travel home after dialysis. In contrast, most of our participants were driven to dialysis by paratransit or a family member, yet we also observed increased activity levels after dialysis treatments, often to intensity levels they did not achieve at any other time. Although the increase among paratransit users may have been solely due to transportation to and from the dialysis unit, the larger increase in the remainder of the cohort, coupled with the increased step counts noted throughout the hour after dialysis, suggests other sources of activity as well. Thus, our study extends the paradoxical finding of increased physical activity post dialysis to a patient population with substantial functional limitation. We also demonstrate that this pattern of activity is maintained across many weeks of dialysis treatment.
We hypothesized the increase in activity would be restricted to patients who were least negatively affected by the dialysis procedure. Patients with longer recovery time did have lower activity levels overall and in the hour after dialysis, but even in this group, intensity increased in the postdialysis period. Furthermore, these patients reached a level of intensity greater than the average postdialysis intensity of the fast recovery time group. In the HDFIT trial, activity levels were not associated with postdialysis recovery time, but only 17% of participants reported recovery times >2 hours, in contrast with 40% in our cohort (26). Thus, even patients most likely affected by postdialysis fatigue (28) were capable of consistently increasing activity levels after completing dialysis treatments.
The increase in activity in the hour after dialysis highlights a potentially unrecognized opportunity to promote physical activity. In a randomized controlled trial, personalized walking exercise on nondialysis days improved functional status, indicating that even low-intensity physical activity is effective in improving quality of life (5). The peridialysis setting may be an ideal time to increase activity level due to the requisite interaction with health care professionals and could be paired with the formation of “exercise teams” composed of dialysis unit staff (29). Harnessing patients’ ability to increase their activity level, even for short periods of time after dialysis, has the potential to improve functional outcomes (5). Importantly, this could be performed by patients, with encouragement by dialysis staff, in the dialysis unit or at home.
Several limitations of this study should be noted. First, participants were recruited from a single dialysis center in the Bronx. However, a relatively high proportion of potential participants were enrolled, supporting the generalizability of our findings. Enrollment occurred over a 3-month time period; hence, future studies are needed to investigate seasonal effects on physical activity post dialysis. We were not able to determine the extent to which the increase in step-count intensity after dialysis was driven by requisite walking leaving the dialysis center as opposed to positive effects of hemodialysis such as improved volume status and clearance of uremic toxins. It is possible that wearing an accelerometer could itself affect patients’ activity levels; however, we set the devices to display time, and providing patients with feedback about their step counts has not increased physical activity (19,23). Finally, given our cohort’s modest size, larger studies are needed to confirm our findings and to elucidate the contribution of recovery time to overall activity. However, a multicenter study including 13 dialysis centers in Brazil also noted high levels of activity after dialysis treatments relative to the remainder of the postdialysis period.
In conclusion, in patients receiving hemodialysis, continuous measurement of step counts over a long time period is feasible and, coupled with highly granular data collection and analysis, can yield new insights into patterns of physical activity.
Disclosures
M.K. Abramowitz reports consultancy agreements with Tricida, and ownership interest in Aethlon Medical, Inc. N. Ohri reports consultancy for AstraZeneca, Genentech, and Merck; and research funding from AstraZeneca, Celldex, and Merck. All remaining authors have nothing to disclose.
Funding
This research was supported by K23DK099438 from the National Institutes of Health (NIH). Its contents are solely the responsibility of the authors and do not necessarily represent the official views or policies of the NIH.
Author Contributions
M.K. Abramowitz and B.S. Cohen wrote the draft of the manuscript; M.K. Abramowitz, B.S. Cohen, S. Kotwani, S. Munugoti, and L.S. Randhawa were responsible for data analysis/interpretation; M.K. Abramowitz, S. Dalezman, and N. Ohri were responsible for the research idea and study design; S. Dalezman, A.C. Elters, J.S. Ibarra,S. Kotwani, S. Munugoti, K. Nam, W. Paredes, L.S. Randhawa, and S. Venkataraman were responsible for data acquisition; and all authors read and approved the final manuscript and agree both to be personally accountable for their own contributions and to ensure that questions related to the accuracy or integrity of any part of the work, even ones in which each author was not personally involved, are appropriately investigated, resolved, and the resolution documented in the literature.
Data Sharing Statement
Due to concerns over maintenance of participant confidentiality related to the granularity of data collection, the datasets generated and/or analyzed during the current study may be available from the corresponding author on reasonable request.
References
1. Johansen KL, Kaysen GA, Dalrymple LS, Grimes BA, Glidden DV, Anand S, Chertow GM: Association of physical activity with survival among ambulatory patients on dialysis: The Comprehensive Dialysis Study. Clin J Am Soc Nephrol 8: 248–253, 2013
https://doi.org/10.2215/CJN.08560812
2. O’Hare AM, Tawney K, Bacchetti P, Johansen KL: Decreased survival among sedentary patients undergoing dialysis: Results from the dialysis morbidity and mortality study wave 2. Am J Kidney Dis 41: 447–454, 2003
https://doi.org/10.1053/ajkd.2003.50055
3. Delgado C, Johansen KL: Barriers to exercise participation among dialysis patients. Nephrol Dial Transplant 27: 1152–1157, 2012
https://doi.org/10.1093/ndt/gfr404
4. Roshanravan B, Gamboa J, Wilund K: Exercise and CKD: Skeletal muscle dysfunction and practical application of exercise to prevent and treat physical impairments in CKD. Am J Kidney Dis 69: 837–852, 2017
https://doi.org/10.1053/j.ajkd.2017.01.051
5. Manfredini F, Mallamaci F, D’Arrigo G, Baggetta R, Bolignano D, Torino C, Lamberti N, Bertoli S, Ciurlino D, Rocca-Rey L, Barillà A, Battaglia Y, Rapanà RM, Zuccalà A, Bonanno G, Fatuzzo P, Rapisarda F, Rastelli S, Fabrizi F, Messa P, De Paola L, Lombardi L, Cupisti A, Fuiano G, Lucisano G, Summaria C, Felisatti M, Pozzato E, Malagoni AM, Castellino P, Aucella F, ElHafeez SA, Provenzano PF, Tripepi G, Catizone L, Zoccali C: Exercise in patients on dialysis: A multicenter, randomized clinical trial. J Am Soc Nephrol 28: 1259–1268, 2016
https://doi.org/10.1681/ASN.2016030378
6. Zhang F, Ren Y, Wang H, Bai Y, Huang L: Daily step counts in patients with chronic kidney disease: A systematic review and meta-analysis of observational studies. Front Med (Lausanne) 9: 842423, 2022
https://doi.org/10.3389/fmed.2022.842423
7. Sheshadri A, Kittiskulnam P, Lazar AA, Johansen KL: A walking intervention to increase weekly steps in dialysis patients: A pilot randomized controlled trial. Am J Kidney Dis 75: 488–496, 2020
https://doi.org/10.1053/j.ajkd.2019.07.026
8. Larkin JW, Han M, Han H, Guedes MH, Gonçalves PB, Poli-de-Figueiredo CE, Cuvello-Neto AL, Barra ABL, de Moraes TP, Usvyat LA, Kotanko P, Canziani MEF, Raimann JG, Pecoits-Filho R; HDFIT Study Investigators: Impact of hemodialysis and post-dialysis period on granular activity levels. BMC Nephrol 21: 197, 2020
https://doi.org/10.1186/s12882-020-01853-2
9. Majchrzak KM, Pupim LB, Chen K, Martin CJ, Gaffney S, Greene JH, Ikizler TA: Physical activity patterns in chronic hemodialysis patients: Comparison of dialysis and nondialysis days. J Ren Nutr 15: 217–224, 2005
https://doi.org/10.1053/j.jrn.2004.08.002
10. Guralnik JM, Simonsick EM, Ferrucci L, Glynn RJ, Berkman LF, Blazer DG, Scherr PA, Wallace RB: A short physical performance battery assessing lower extremity function: Association with self-reported disability and prediction of mortality and nursing home admission. J Gerontol 49: M85–M94, 1994
https://doi.org/10.1093/geronj/49.2.M85
11. Lindsay RM, Heidenheim PA, Nesrallah G, Garg AX, Suri R; Daily Hemodialysis Study Group London Health Sciences Centre: Minutes to recovery after a hemodialysis session: A simple health-related quality of life question that is reliable, valid, and sensitive to change. Clin J Am Soc Nephrol 1: 952–959, 2006
https://doi.org/10.2215/CJN.00040106
12. Lunney M, Wiebe N, Kusi-Appiah E, Tonelli A, Lewis R, Ferber R, Tonelli M; Alberta Kidney Disease Network: Wearable fitness trackers to predict clinical deterioration in maintenance hemodialysis: A prospective cohort feasibility study. Kidney Med 3: 768–775.e1, 2021
https://doi.org/10.1016/j.xkme.2021.04.013
13. Sheshadri A, Kittiskulnam P, Lai JC, Johansen KL: Effect of a pedometer-based walking intervention on body composition in patients with ESRD: A randomized controlled trial. BMC Nephrol 21: 100, 2020
https://doi.org/10.1186/s12882-020-01753-5
14. Le Masurier GC, Tudor-Locke C: Comparison of pedometer and accelerometer accuracy under controlled conditions. Med Sci Sports Exerc 35: 867–871, 2003
https://doi.org/10.1249/01.MSS.0000064996.63632.10
15. Kimura A, Paredes W, Pai R, Farooq H, Buttar RS, Custodio M, Munugoti S, Kotwani S, Randhawa LS, Dalezman S, Elters AC, Nam K, Ibarra JS, Venkataraman S, Abramowitz MK: Step length and fall risk in adults with chronic kidney disease: A pilot study. BMC Nephrol 23: 74, 2022
https://doi.org/10.1186/s12882-022-02706-w
16. Tran J, Ayers E, Verghese J, Abramowitz MK: Gait abnormalities and the risk of falls in CKD. Clin J Am Soc Nephrol 14: 983–993, 2019
https://doi.org/10.2215/CJN.13871118
17. Zemp DD, Giannini O, Quadri P, de Bruin ED: Gait characteristics of CKD patients: A systematic review. BMC Nephrol 20: 83, 2019
https://doi.org/10.1186/s12882-019-1270-9
18. Zemp DD, Giannini O, Quadri P, Rabuffetti M, Tettamanti M, de Bruin ED: Signatures of gait movement variability in CKD patients scheduled for hemodialysis indicate pathological performance before and after hemodialysis: A prospective, observational study. Front Med (Lausanne) 8: 702029, 2021
https://doi.org/10.3389/fmed.2021.702029
19. Han M, Williams S, Mendoza M, Ye X, Zhang H, Calice-Silva V, Thijssen S, Kotanko P, Meyring-Wösten A: Quantifying physical activity levels and sleep in hemodialysis patients using a commercially available activity tracker. Blood Purif 41: 194–204, 2016
https://doi.org/10.1159/000441314
20. Han M, Ye X, Preciado P, Williams S, Campos I, Bonner M, Young C, Marsh D, Larkin JW, Usvyat LA, Maddux FW, Pecoits-Filho R, Kotanko P: Relationships between neighborhood walkability and objectively measured physical activity levels in hemodialysis patients. Blood Purif 45: 236–244, 2018
https://doi.org/10.1159/000485161
21. Malhotra R, Kumar U, Virgen P, Magallon B, Garimella PS, Chopra T, Kotanko P, Ikizler TA, Trzebinska D, Cadmus-Bertram L, Ix JH: Physical activity in hemodialysis patients on nondialysis and dialysis days: Prospective observational study. Hemodial Int 25: 240–248, 2021
https://doi.org/10.1111/hdi.12913
22. Oishi D, Koitabashi K, Hiraki K, Imai N, Sakurada T, Konno Y, Shibagaki Y, Yasuda T, Kimura K: Physical activity is associated with serum albumin in peritoneal dialysis patients. Adv Perit Dial 28: 148–152, 2012
23. Williams S, Han M, Ye X, Zhang H, Meyring-Wösten A, Bonner M, Young C, Thijssen S, Marsh D, Kotanko P: Physical activity and sleep patterns in hemodialysis patients in a suburban environment. Blood Purif 43: 235–243, 2017
https://doi.org/10.1159/000452751
24. Ohri N, Halmos B, Bodner WR, Cheng H, Guha C, Kalnicki S, Garg M: Daily step counts: A new prognostic factor in locally advanced non-small cell lung cancer? Int J Radiat Oncol Biol Phys 105: 745–751, 2019
https://doi.org/10.1016/j.ijrobp.2019.07.055
25. Ohri N, Kabarriti R, Bodner WR, Mehta KJ, Shankar V, Halmos B, Haigentz M Jr, Rapkin B, Guha C, Kalnicki S, Garg M: Continuous activity monitoring during concurrent chemoradiotherapy. Int J Radiat Oncol Biol Phys 97: 1061–1065, 2017
https://doi.org/10.1016/j.ijrobp.2016.12.030
26. Leme J, Guedes M, Larkin J, Han M, Barra ABL, Canziani MEF, Cuvello Neto AL, Poli-de-Figueiredo CE, de Moraes TP, Pecoits-Filho R; HDFIT Study Investigators: Patient perception of vitality and measured physical activity in patients receiving haemodialysis. Nephrology (Carlton) 25: 865–871, 2020
https://doi.org/10.1111/nep.13758
27. Pecoits-Filho R, Larkin J, Poli-de-Figueiredo CE, Cuvello-Neto AL, Barra ABL, Gonçalves PB, Sheth S, Guedes M, Han M, Calice-Silva V, de Castro MCM, Kotanko P, de Moraes TP, Raimann JG, Canziani MEF; HDFIT Study Investigators: Effect of hemodiafiltration on measured physical activity: Primary results of the HDFIT randomized controlled trial. Nephrol Dial Transplant 36: 1057–1070, 2021
https://doi.org/10.1093/ndt/gfaa173
28. Bossola M, Tazza L: Postdialysis fatigue: A frequent and debilitating symptom. Semin Dial 29: 222–227, 2016
https://doi.org/10.1111/sdi.12468
29. Capitanini A, Lange S, D’Alessandro C, Salotti E, Tavolaro A, Baronti ME, Giannese D, Cupisti A: Dialysis exercise team: The way to sustain exercise programs in hemodialysis patients. Kidney Blood Press Res 39: 129–133, 2014
https://doi.org/10.1159/000355787
