Intradialytic exercise: a controlled nationwide effectiveness-implementation study

ABSTRACT

Background

Efficacy studies have demonstrated the benefits of intradialytic exercise. However, real-world effectiveness trials are lacking. This study evaluated a nationwide clinical implementation of intradialytic exercise settled as a routine practice. Implementation of intradialytic exercise and its effectiveness regarding safety, physical function and body composition were investigated.

Methods

This was a retrospective analysis of the first year of implementation using the RE-AIM framework (Reach, Effectiveness, Adoption, Implementation and Maintenance). For effectiveness outcomes, participants (n = 347) were compared with patients who refused the intervention (n = 394), except for physical function in which a pre–post design was performed. Physical function tests included the 8-foot up and go, 30-s sit-to-stand, 5 times sit-to-stand, single leg stance and hand dynamometer. Body composition was determined by multifrequency bioimpedance.

Results

Adoption: 20 hemodialysis units adopted the intervention (55.6%). Reach: 1270 patients were eligible (55.8%). The main reason for non-eligibility was physical/cognitive incapacity (50.8%). Of those eligible (n = 1270), 811 (63.9%) started the intervention and 459 (36.1%) refused. Maintenance: attrition rate was 57.2%. Implementation: patients completed 86.3 ± 29.0 min/week of aerobic exercise and adherence was 75.0% ± 19.7%. Effectiveness: there was a lower incidence of cramps in the exercise group (P < .001). No significant differences were found for other adverse events. A significant improvement was observed for all physical function tests, except for handgrip strength. For body composition, significant differences favorable to the exercise group were found for lean tissue mass (P = .045), lean tissue index (P = .013) and body cell mass (P < .001).

Conclusions

Large-scale intradialytic exercise implementation is realistic, safe and effective. Nevertheless, implementation outcomes were limited, and complementary interventions may be needed.

Graphical Abstract

Graphical Abstract.

KEY LEARNING POINTS.

What was known:

• Intradialytic exercise is very convenient for hemodialysis patients and is the most typical form of exercise in this population.

• Some recent guidelines recommend its implementation in all hemodialysis units.

• Efficacy studies have demonstrated its benefits, however real-world effectiveness trials are lacking.

This study adds:

• Clinical implementation of intradialytic exercise is realistic, safe and effective to improve physical function and body composition.

• Nevertheless, its health impact is limited by patients who refused the intervention or who voluntarily dropped out.

• Moreover, the achieved exercise dose was not enough to meet the recommended physical activity levels.

Potential impact:

• Some patients may need additional support so that they can achieve higher exercise doses.

• Complementary exercise interventions that consider individual preferences and needs may improve the overall health impact of exercise.

INTRODUCTION

Despite the association of physical activity (PA) with better health outcomes in hemodialysis (HD) patients [1, 2], most remain inactive [3]. The catabolism worsens neuromuscular functioning and exercise tolerance. Over time, these lead to functional impairment, and further inactivity and consequent reduction in anabolic stimuli [4]. This vicious cycle makes PA promotion difficult in this population. Barriers to exercise are reported by 92% of HD patients, with fatigue being the most reported one [5]. Nevertheless, most patients recognize the benefits of exercise, notwithstanding the need for support to do so [6]. Intradialytic exercise (IDE) is an opportunity for the dialysis staff to support patients to exercise while on dialysis and is also very convenient for patients [7]. Several systematic reviews demonstrate its efficacy in improving cardiorespiratory fitness [8], physical function [9] and health-related quality of life [10]. Guidelines recommend that HD patients should aim for 150 min/week of moderate-intensity aerobic exercise [11]. This is difficult to achieve considering the time spent in HD. Despite the recommendation for IDE implementation in all HD units (e.g. by the UK Kidney Association) [11], no outcomes data are available from large-scale implementation studies, which may contribute to its low dissemination [12–14].

The time lag between research discovery and routine clinical uptake is inflated by a research focus on efficacy studies, instead of more generalizable effectiveness trials [15]. Thus, implementation studies are needed to inform the implementation of IDE in clinical settings and to investigate its effectiveness in real-world conditions. This study aims to evaluate a nationwide implementation of IDE settled as part of the routine care of HD patients and provided by the existing dialysis staff. This study examined implementation outcomes and its effectiveness on safety, physical function and body composition.

MATERIALS AND METHODS

This article follows the recommendation of the Standards for Reporting Implementation Studies statements [16] (Supplementary data, Table S1) and was approved by the NephroCare Portugal Ethics Committee (04/2021).

Study design

This is a hybrid type III implementation study focusing on both the success of the implementation, and its health impact [15]. We retrospectively analyzed the implementation of an IDE program at a nationwide level in the largest dialysis provider in Portugal (NephroCare). We used the RE-AIM framework (Reach, Effectiveness, Adoption, Implementation and Maintenance), used to evaluate public health interventions to produce a more balanced approach between internal and external validity [17].

All units that joined the program during the first year (from September 2016 to September 2017) were analyzed and patients were followed during 1 year for effectiveness outcomes (safety, physical function and body composition). For each effectiveness outcome, IDE patients (exercise group) were compared with a non-exercise group of patients that refused IDE. However, physical function was assessed only in the exercise group, thus a pre–post design was used for this outcome. Additionally, a subsequent analysis was performed comparing effects in a lower (<124 exercise sessions) vs higher (≥124 exercise sessions) exercise group. These groups were created based on the median number of exercise sessions performed.

Implementation context

Portugal has an integrated management of HD patients [18], with no inclusion of any exercise intervention. Thus, the IDE program was fully funded by NephroCare and free of charge for patients. In each unit, the program was conducted with the existing staff.

Targeted sites and participants

All the NephroCare Portugal HD units were targeted for implementation. Inclusion criteria were physical/cognitive capacity (defined after a first exercise trial with the patient); and HD vintage ≥2 months. Patients were excluded in case of vascular access in the lower limb or at high risk of hematoma (e.g. previous hematoma, first arteriovenous fistula punctures), cardiovascular risk, hemoglobin <8.5 g/dL or systemic infection. Cardiovascular conditions indicative of high risk for IDE are detailed in Supplementary data, Table S2.

Description of the implementation strategy

After a pilot experience in a single dialysis unit (September 2014), the medical and executive boards decided to expand the program to a national level. A national IDE coordinator was nominated and a guide to IDE implementation was developed and disseminated to all HD units. In each clinic, the decision to integrate the program was made by consensus of the medical and nursing directors who designated two members of the medical and nursing staff to be the local IDE coordinators. Local coordinators were trained by the national coordinator who had previous experience in IDE implementation. On the initial screening visit, all patients who met the inclusion criteria received an explanation of the purpose, risks and procedures of IDE and patients were invited to sign an informed consent. A digital platform was developed to monitor IDE implementation. For implementation quality assurance, every 3 months the national coordinator visited each of the HD units that joined the IDE program. Methods and techniques are described in Supplementary data, Table S3.

Description of the intervention

Participants were instructed to perform a simple exercise protocol in every HD session (generally three times per week), which was designed to be easily implemented by the existing dialysis staff (nurses and physicians) with low supervision requirements (Table 1). Aerobic exercise was performed in a cycle ergometer (MoVeS OxyCycle I Active Pedal Exerciser) and included a warm-up, an aerobic component and a cool-down period. Given the limited supervision provided, strength training started 3 months later to make it easier for participants to understand the entire exercise protocol. Strength training included upper and lower limb exercises using squeeze balls and ankle weights, respectively. Progression in the aerobic component was increased by 10 min every 2 weeks (up to 60 min). For strength exercises, firstly the number of sets increased from one up to four, followed by weight increases up to 4.5 kg. Information on the IDE equipment can be found in Supplementary data, Table S4. A checklist of exercise contraindications (e.g. fever, chest pain, dyspnea) was provided to each nurse. For any treatment session in which the patient did not perform the prescribed exercise, the nurse was required to document the reason.

Table 1:

IDE protocol.

Aerobic trainingCycle ergometer	Warm-up: 5 min; low load
	Aerobic component: up to 60 min (12–15 on the 6–20 Borg scale); 50–70 rpm; increased load
	Cool-down: 5 min; low load
	12–15 on the 6–20 Borg scale
Strength training	Upper limbs exercise:
3 months laterAfter aerobic training	Handgrip (squeeze balls with different resistance)
	Lower limbs exercises:
	Against dialysis chair—isometric: plantar flexion, leg extension
	Ankle weights—isotonic: knee extension, knee flexion, hip abduction, hip flexion

rpm: rotations per minute.

Implementation and effectiveness outcomes

Metrics for each dimension of the RE-AIM framework were adapted for the specific case of IDE (Table 2).

Table 2:

Implementation and effectiveness outcomes (RE-AIM dimensions).

Dimension	Definition	Measures
Adoption	Number, proportion and representativeness of settings and intervention agents who are willing to initiate the intervention	Proportion of HD units who accepted IDE
Reach	Number, proportion and representativeness of patients who decided to participate	Proportion of non-eligible patients with reasons
		Comparison of eligible and non-eligible patients
		Prevalence of non-eligibility across HD units
		Proportion of patients refusing the intervention
		Comparison of refusals with patients that accepted IDE
		Prevalence of patients refusing IDE across HD units
Maintenance (setting level)	Extent to which a program or policy becomes institutionalized or part of the routine organizational practices and policies	Proportion of HD units who continued the intervention
Maintenance (patient level)		Attrition rate with reasons and timing
		Comparison of voluntary withdrawals with completers
		Prevalence of voluntary withdrawals across HD units
Implementation (fidelity)a	Whether the intervention is delivered as intended	Adherence (exercise sessions among total HD sessions), number and proportion of fully, partially and non-performed exercise sessions with reasons
		Mean adherence levels to IDE across HD units
Implementation (dose)a	Amount of intervention successfully implemented	Weekly exercise dose: aerobic (sessions/week, duration/week) and strength (sessions/week, workload/week)
Effectivenessa	Impact of an intervention on health outcomes, including negative impacts	Safety, physical function and body composition

^aOutcomes studied in patients who completed the 1-year follow-up period.

Safety was studied examining the incidence of all reported intradialytic adverse events in every HD session during the study period (cramps, headache, needle displacement/dislodgement, dyspnea, dysrhythmia, abdominal pain, chest pain, fatigue, hypertension, hypotension, hypoglycemia in diabetics, nausea/vomiting and syncope). A composite outcome was computed by summing all adverse events. Aggregated data from each HD unit were used to study the correlation of the proportion of non-eligibility with the incidence of adverse events.

Physical function tests were performed only by the exercise group in the second or third weekly HD session, so that the patients’ performance was not affected by the fluid overload of the long interdialytic period. These tests included the 8-foot up and go (agility and dynamic balance), 30-s sit-to-stand (STS 30) (muscle resistance), 5 times STS (STS 5; muscle power), single leg stance (static balance) and handheld dynamometry (handgrip strength).

Body composition measures were taken before a second or third weekly HD treatment, by multifrequency bioimpedance using the body composition monitor (Fresenius Medical Care, Bad Homburg, Germany). This device is based on a three-compartment model, providing excess fluid mass, lean tissue mass and adipose tissue mass [19], and has been validated against reference methods [20].

Data collection and analysis

Data related to IDE participation were extracted from the IDE digital platform. Patient descriptive and clinical data were extracted from the NephroCare European Clinical Database [21].

Normality was checked and, if non-normally distributed, statistical analysis was performed on the logarithmically transformed data. Comparison between groups at baseline was performed using Mann–Whitney U test or the independent samples t-test. For categorical variables, groups were compared using the Chi-squared test.

Pre–post comparisons within groups were conducted using Wilcoxon Signed Rank test. Between-group comparisons of outcome changes were performed using both unadjusted tests (Mann–Whitney or independent samples t-test) and adjusted analysis (Quade’s or analysis of covariance tests), controlling for baseline values of the corresponding outcome. Statistical analyses were performed using the IBM^® SPSS^® statistics version 23 (IBM^® Corporation, Armonk, NY, USA).

RESULTS

RE-AIM dimension: Adoption

Proportion of HD units who accepted IDE

All NephroCare Portugal dialysis units at the time (n = 36) were invited. During the first year, 20 (55.6%) agreed to initiate the program.

RE-AIM dimension: Reach

Proportion of non-eligible patients with reasons

Of those assessed for eligibility (n = 2278), 1008 (44.2%) patients were excluded (Fig. 1). The main reasons for exclusion were physical/cognitive incapacity (50.8%) and cardiovascular risk (34.6%) (Table 3).

Figure 1:

Enrollment flowchart for IDE.

Table 3:

Reasons for exclusion from IDE.

Reason	n	%
Hemoglobin <8.5 g/dL	3	0.3
Vascular access in the lower limb	24	2.4
Risk of vascular access hematoma	39	3.9
Cardiovascular risk	349	34.6
Physical/cognitive incapacity	512	50.8
Other/unknown	81	8.0
Total	1008	100

Comparison of eligible and non-eligible patients

Non-eligible patients had poorer general health status (older, more comorbidities and worst body composition) (Supplementary data, Table S5).

Prevalence of non-eligibility across HD units

Non-eligibility to IDE in each unit ranged from 16.7% to 74.2% (Table 4).

Table 4:

Results across the different HD units: non-eligibility, refusals, voluntary withdrawal, adherence and incidence of adverse events.

	Non-eligible (%)	Refusals (%)a	Vol. withdrawal (%)b	Mean adherence (%)c	Adverse events/1000 HD sessionsd
Clinic 1	20	11.1	18.8	74.5	12.00
Clinic 2	28.2	44.4	15.4	75.1	1.50
Clinic 3	36.9	59.6	35.0	79.6	7.00
Clinic 4	21.3	20.6	8.0	70.0	12.80
Clinic 5	34.6	32.9	31.9	87.0	5.80
Clinic 6	57.5	39.2	51.6	63.6	9.30
Clinic 7	25.9	60.7	63.0	63.0	1.60
Clinic 8	47.5	12.5	14.3	69.1	9.90
Clinic 9	50.8	10.9	31.6	75.2	11.70
Clinic 10	61.8	6.9	25.9	92.0	9.00
Clinic 11	38.7	36.8	0.0	66.3	9.10
Clinic 12	74.2	8.3	18.2	87.2	19.30
Clinic 13	64.4	29.0	25.0	79.4	14.90
Clinic 14	53.6	26.8	21.2	72.8	7.20
Clinic 15	53.2	53.2	54.9	60.6	2.90
Clinic 16	53.1	28.9	40.7	71.9	10.50
Clinic 17	16.7	16.7	24.0	82.0	9.60
Clinic 18	67.1	15.4	13.6	73.3	5.50
Clinic 19	53.3	0.0	0.0	67.9	14.40
Clinic 20	59.6	30.6	28.0	79.0	10.00

^aCalculated among eligible patients.

^bCalculated among participants.

^cMean adherence of patients completing the 1-year follow-up.

^dCalculated among patients who completed the 1-year intervention period.

Proportion of patients refusing the intervention

Of those eligible (n = 1270), 36.1% (n = 459) refused and 63.9% (n = 811) accepted the intervention (Fig. 1).

Comparison of IDE participants with refusals

Both IDE participants and refusals were eligible for IDE. However, those who refused had poorer health condition (older, higher dialysis vintage, more comorbidities, worst body composition) (Supplementary data, Table S6).

Prevalence of patients refusing IDE across HD units

In each unit, the prevalence of patients who refused IDE ranged from 0% to 60.7% of all eligible patients (Table 4).

RE-AIM dimension: Maintenance

Proportion of HD units who continued the intervention

After 1 year, all 20 dialysis units continued the IDE program.

Attrition rate with reasons and timing

Attrition rate was 57.2% (n = 464) and voluntary withdrawal was the main reason (52.4%) (Fig. 2). Attrition with reasons over time is illustrated in Fig. 3. Most voluntary withdrawals occurred during the first six months (70.8%, n = 172).

Figure 2:

Reasons for attrition.

Figure 3:

Reasons for attrition over time.

Comparison of voluntary withdrawals with completers

Voluntary withdrawals and completers had no important differences for age, comorbidities and body composition, which suggests a similar health condition (Supplementary data, Table S7).

Prevalence of voluntary withdrawals across HD units

In each unit, prevalence of voluntary withdrawal ranged from 0.0% to 63.0% (Table 4).

RE-AIM dimension: Implementation

Implementation fidelity: adherence, number and proportion of fully, partially and non-performed exercise sessions with reasons

Adherence (%) to IDE sessions was 75.0 ± 19.7. Of all HD sessions (n = 50 356), 77.4% were fully performed, 3.2% partially performed and 19.4% were non-performed exercise sessions. As detailed in Supplementary data, Tables S8 and S9, the main reason for non-performed and for partially performed exercise sessions was voluntary (61.5% and 74.5%, respectively).

Mean adherence levels to IDE across HD units

Mean adherence to exercise sessions ranged from 60.6% to 92.0% (Table 4).

Implementation dose: mean weekly exercise dose

Weekly, patients performed 2.2 ± 0.6 aerobic exercise sessions completing 86.3 ± 29.0 min/week. Strength exercise was performed by 80.4% (n = 279) of the patients. Frequency per week was 1.5 ± 0.7 achieving a total workload/week of 217.9 ± 306.4 (sets.kg) (Supplementary data, Table S10).

RE-AIM dimension: Effectiveness

Compared with IDE completers (n = 347), the non-exercise group (n = 394) had a worse health status (older, higher dialysis vintage, more comorbidities and worst body composition) (Table 5).

Table 5:

Comparison between the non-exercise and the exercise groups.

	Non-exercise (n = 394)	Exercise (n = 347)	P
Age (years)	64.9 ± 14.5	61.5 ± 13.9	.001 a
Female, n (%)	147 (37.3)	114 (32.9)	.205b
Dialysis vintage (months)	86.9 (91.0)	61.3 (76.0)	<.001 a
Interdialytic weight gain (L)	2.7 ± 1.1	2.9 ± 1.1	.060a
Vascular access
Arteriovenous fistula, n (%)	296 (75.1)	294 (84.7)	.001 b
Arteriovenous graft, n (%)	42 (10.7)	32 (9.2)
Central venous catheter, n (%)	56 (14.2)	21 (6.1)
Treatment modality
Hemodiafiltration, n (%)	349 (88.6)	327 (94.2)	.527b
Hemodialysis, n (%)	45 (11.4)	20 (5.8)
Blood pressure
Systolic BP (mmHg)	140.5 ± 20.8	139.0 ± 19.5	.310c
Diastolic BP (mmHg)	67.8 ± 14.1	68.2 ± 13.0	.668c
Patients taking AH drugs, n (%)	237 (60.2)	212 (61.1)	.793b
Age-adjusted Charlson Comorbidity Index	6.0 ± 2.5	5.4 ± 2.4	<.001 a
Diabetes mellitus, n (%)	97 (24.6)	100 (28.8)	.197b
CV disease
CV disease, n (%)	335 (85.0)	266 (76.7)	.004 b
No. of CV diseases	2.1 ± 1.8	1.6 ± 1.4	<.001 a
Body composition
Body mass index (kg/m2)	26.3 ± 4.9	26.5 ± 4.5	.292a
Lean tissue index (kg/m2)	12.8 ± 3.1	13.7 ± 3.1	.002 c
Fat tissue index (kg/m2)	12.3 ± 5.6	12.2 ± 5.4	.664a
Overhydration (%)	9.0 ± 9.5	6.8 ± 7.1	<.001 a
Single pool Kt/V	1.6 ± 0.4	1.6 ± 0.3	.650c
Phosphate (mg/dL)	4.6 ± 1.3	4.4 ± 1.3	.108a
Hemoglobin (g/dL)	11.2 ± 1.3	11.3 ± 1.1	.063a

Data are mean ± standard deviation.

^aMann–Whitney U test.

^bChi-square test.

^cIndependent samples t-test.

BP, blood pressure; AH, antihypertensive; CV, cardiovascular.

Safety

Incidence of adverse events is presented in Table 6. There was a significant lower incidence of cramps in the exercise group even after adjustment for ultrafiltration volume (P < .001). There were no significant differences between groups for incidence of any other adverse events and in the total incidence of adverse events.

Table 6:

Incidence of adverse events per 1000 HD sessions.

	Non-exercise (n = 394)	Exercise (n = 347)	P*	P b
Cramps	2.0 ± 4.8	1.3 ± 4.5	.001	<.001 c
Headache	0.1 ± 1.1	0.2 ± 1.9	.816	.806c
Needle displacement	0.2 ± 1.2	0.2 ± 1.4	.464	.403d
Needle dislodgement	0.1 ± 1.0	0.2 ± 1.1	.065	.069d
Dyspnea	0.4 ± 2.6	0.2 ± 1.2	.173	.406c
Dysrhythmia	0.1 ± 1.0	0.0 ± 0.0	.103	.381c
Abdominal pain	0.3 ± 2.1	0.1 ± 1.1	.479	.508c
Chest pain	0.2 ± 1.0	0.1 ± 0.8	.575	.624c
Fatigue	0.0 ± 0.6	0.1 ± 0.7	.870	.870c
Hypertension	0.4 ± 2.0	0.4 ± 1.9	.325	.323c
Hypotension	5.6 ± 12.7	5.9 ± 13.4	.717	.485c
Hypoglycemia (diabetics)a	0.1 ± 0.9	0.4 ± 2.8	.678
Nausea/vomiting	0.2 ± 1.2	0.3 ± 1.7	.991	.867c
Syncope	0.1 ± 0.7	0.1 ± 0.9	.832	.979c
Total adverse events	9.9 ± 16.3	9.2 ± 15.5	.398	.908e

Data are mean ± standard deviation.

^aAnalysis restricted to diabetic patients (n = 196).

^bAdjusted analysis using Quade’s test.

^cAdjustment for ultrafiltration volume.

^dAdjustment for vascular access type.

^eAdjustment for ultrafiltration volume and vascular access type.

*Unadjusted analysis using Mann–Whitney U test.

Across HD units, mean incidence of adverse events (per 1000 HD sessions) varied between 1.5 and 19.3 and was not correlated with the prevalence of non-eligibility (P = .141).

Physical function

A significant improvement in all physical function tests was observed after 1 year, except for handgrip strength that showed a slight reduction from 30.0 ± 11.4 kg to 29.1 ± 11.5 kg (P < .001) (Fig. 4).

Figure 4:

Physical function measures at baseline and 1 year. Data are median with 95% confidence intervals.

Compared with the lower exercise group, improvements in STS 30 were greater in the higher exercise group (P = .010; Fig. 5). Moreover, the reduction in handgrip strength was observed only for the lower exercise group (–1.6 ± 8.3 kg, P = .049; Fig. 5).

Figure 5:

Variation in physical function between a lower and a higher exercise group. Data are median with 95% confidence intervals.

Body composition and fluid status

Variation in body composition is shown in Fig. 6. The non-exercise group reduced lean tissue mass (kg) and index (kg/m²), and the exercise group increased both outcomes. After adjustment, significant differences between groups were observed for variations in lean tissue mass (P = .045) and lean tissue index (P = .013). Moreover, the exercise group had a significantly higher increase in body cell mass (P < .001). No significant differences were found for other body composition parameters.

Figure 6:

Variation in body composition measures between groups. Data are median with 95% confidence intervals.

DISCUSSION

To our knowledge, this is the first implementation study describing a large implementation of an IDE program at a nationwide level. Taken together, these are encouraging results demonstrating that IDE can be part of standard care for HD patients. Moreover, this is a safe intervention, and its effectiveness seems to be preserved in real-world conditions to improve physical function and body composition. Thus, compliance with the recommendation of implementing IDE in all HD units [11] is feasible. However some concerns emerged from this investigation: only about half of the HD units adopted the intervention; there was a high proportion of non-eligible patients; a high attrition rate mainly due to voluntary withdrawal; adherence levels were good but limited by voluntarily missed exercise sessions; and the exercise dose was not enough to meet the PA recommendations (>150 min/week) [11]. Yet, there was a wide variation between units, suggesting that better implementation outcomes are possible and could increase the health impact of IDE.

During its first year, IDE was adopted by 55.6% of the HD units. We have no data addressing the reasons why the medical and nursing directors declined to participate. Possibly, safety concerns, the perceived patient disinterest and staff workload may have discouraged adoption by these decisive actors [22]. However, these barriers are not consistent with our results. The safety of IDE was confirmed by the non-increase in the incidence of adverse events. Patient interest was demonstrated by the fact that half of the patients completed the 1-year intervention with a good adherence (75.0 ± 19.7%). The staff workload increase was minimized by the simplistic exercise protocol with low supervision requirements and by selecting patients who were able to cycle without continuous support.

Reach was affected by the proportion of patients excluded from the intervention (44.2%), mainly due to physical/cognitive incapacity (50.8%) and cardiovascular risk (34.6%). There was a broader disparity between HD units for eligibility (16.7% to 74.2%) meaning that some units adopted a more flexible while others a more conservative patient selection. There was no correlation between the prevalence of non-eligibility and the occurrence of adverse events, suggesting that non-eligibility could be safely reduced using a more flexible patient screening. This would be important since those who were non-eligible were also those who may benefit most from exercise (older and sickest patients). Moreover, a more inclusive approach would increase the representativeness of IDE participants, thus improving the reach and its overall health impact. A shift towards a more liberal screening for exercise was adopted for the general population and diabetic patients [23]. Our results support its application for IDE too. Non-eligibility due to physical/cognitive incapacity may also be explained by the low-supervision nature of our exercise protocol which was delivered by the existing dialysis staff. Support from an exclusively dedicated exercise professional has been advocated [24, 25] and may increase eligibility to IDE. Among those eligible to IDE, a substantial proportion of patients refused the intervention (36.1%). The wider prevalence of refusals between HD units (0% to 60.7%) indicates that the way IDE candidates are approached may be decisive for their acceptance. To increase PA engagement, a combination of proven approaches that are grounded in relevant theories (e.g. focus on small quantities of exercise and increase self-regulation) can lead to better outcomes [26]. The approach to candidates to integrate IDE programs should also address important motivators such as improvements in quality of life and physical function, and the will to have a healthier lifestyle [27]. Likewise, family involvement and support from dialysis staff are important promoters of exercise in HD patients [28]. Patients refusing IDE were older and had more comorbidities, indicative of a misperception that having too many medical conditions inhibits one’s ability to exercise. This misperception is common in HD patients [22] and should be considered when approaching patients for IDE participation.

Maintenance was affected by a considerable attrition rate (57.2%) and voluntary withdrawal was, by far, the most common reason for it (52.4%). This is a lost opportunity to improve health outcomes in such an at-risk population. Wide variation between units (0.0% to 63.0%) suggests that improvements can be made. Strategies to enhance long-term sustainability of exercise interventions are described for older adults [29] and could be valuable for HD patients as well. Self-efficacy is associated with dropouts from exercise interventions in HD patients [30]. Thus, self-efficacy assessment and management could be a way to prevent voluntary withdrawal. The simplicity of the exercise protocol may have rendered it unchallenging for some patients. Thus, implementing a progressive structure with greater exercise variety, preferably supported by an exercise professional, could enhance maintenance outcomes. Adherence to exercise sessions (75.0 ± 19.7%) was superior to that observed in a previous multicentric and pragmatic study that had a median (interquartile range) adherence of 47% (29–77) [31]. Nevertheless, opportunity for improvement is demonstrated by the high heterogeneity between HD units (60.6% to 92.0%) and because the participants’ own decision was the main reason for missing an exercise session (61.5%). A smaller study considering only 900 HD sessions found that patients were more than twice as likely to exercise if an exercise professional was present [32].

The implementation dose for aerobic exercise was 86.3 min/week, achieving 57.5% of the recommendation [11]. Strength exercise was accepted by 80% of the participants with a lower frequency/week and a minimal volume. The higher impact on the dialysis staff workload may explain the lower implementation of strength exercise. Taken together, our data suggest that a substantial aerobic exercise volume can be achieved through IDE. Nevertheless, IDE alone was far from reaching the recommended PA levels, mainly for strength exercise. Thus, a more comprehensive and individualized strategy has been advocated, which may include low-intensity activities (e.g. housework and gardening) and IDE, and progression to higher-intensity activities and recreational sports [33].

Despite not being implementation studies, two previous multicentric exercise trials in dialysis patients have reported important implementation outcomes [31, 34]. In the EXCITE (EXerCise Introduction to Enhance Performance in Dialysis) trial, which had a modest home-based exercise (walking) prescription [34], there was a significant dropout (31% at 6 months). Moreover, about 50% of those completing the EXCITE trial had a low adherence (<60%), and 36% of those who were eligible refused to participate. Similarly, the PEDAL (PrEscription of intraDialytic exercise to improve quAlity of Life in patients with chronic kidney disease) trial exercise prescription (intradialytic cycling twice per week) had a median adherence of 47% and a similar withdrawal at 6 months (33.7%) [31]. This limited implementation suggests that exercise protocols should be complemented by motivational and self-monitoring strategies to increase IDE acceptability and to reduce voluntary withdrawals [35, 36]. In routine practice, the dialysis staff must accommodate IDE within their other existing HD-related tasks, and this may generate dissatisfaction. Thus, behavior change techniques at the staff level could also improve IDE implementation [37].

IDE was demonstrated to be safe, with no increase in the incidence of adverse events. This is an important result since there is insufficient evidence to assess the safety of exercise in HD patients [38], which may discourage IDE adoption. Moreover, there was no correlation between the proportion of non-eligibility and the incidence of adverse events, suggesting that a more flexible patient selection could be safely adopted.

IDE patients significantly improved STS 30, 8-foot up and go, STS 5 and single leg stance. However, there was a slight reduction in handgrip strength. These findings are concordant with a previous study [39] and may be due to the exercise protocol focusing mainly on aerobic exercise using the lower limbs. However, an additional analysis comparing patients with a lower and a higher exercise frequency demonstrated that only the lower exercise frequency group had a significant handgrip strength reduction. Also, this group had a significant lower improvement in STS 30. Accordingly, these results suggest that physical function improvement is exercise-dose dependent. Improvements in physical function are encouraging, since most of these measures are mortality predictors [40]. Yet, there is evidence that higher improvements are achieved combining IDE with an exercise prescription guided by an exercise professional [41].

The non-exercise group had decreased lean tissue mass and lean tissue index, whereas the exercise group had increases in these measures. Adjusted analysis revealed a significant difference of variations from baseline to 1-year for both outcomes. Additionally, the exercise group had a higher increase in body cell mass leading to significant differences in variations of both groups. Previous evidence of the effect of exercise on body composition of HD patients is inconclusive [42]. Nevertheless, exercise (particularly strength exercise) is known to improve muscle mass in HD patients [43]. Our encouraging findings demonstrate that, even with a modest exercise dose, the benefits of IDE on body composition persist under real-world conditions. The long follow-up period could have made muscle wasting observable in the non-exercise group, while exercise could have attenuated its occurrence in the exercise group. Previous investigations demonstrated that, in older adults, aerobic exercise mitigates long-term decrements in muscle mass [44]. Considering the deleterious evolution of body composition in this population [45], maintaining patient’s condition should, per se, be regarded as a positive result of exercise [33]. Evidence of the benefits on muscle mass and body cell mass with IDE are noteworthy findings for HD patients because these are predictors of mortality [46, 47]. Nevertheless, IDE with high-load resistance training elicits greater improvements in skeletal muscle index [48] and could be a worthy addition to this exercise protocol.

An important limitation of this study is the absence of qualitative data involving the different stakeholders for a better understanding of reasons for HD units’ non-adoption, patients’ refusal, voluntary withdrawal and missed exercise sessions. This would inform future quality improvement projects. The control group in this study was generally in poorer health condition. However, it was not possible to include a matched control group from other HD units that did not offer the intervention, which may limit the comparability of outcomes between groups. Another limitation is the absence of a cost-effectiveness analysis, which could provide valuable insights to support wider adoption of IDE programs. In addition, data on other important outcomes such as mortality, hospitalizations and quality of life were not included in the current analysis and should be examined in future real-world studies.

Overall, IDE was demonstrated to be realistic, safe and effective in routine practice. Variability across HD units suggests room to improve in several dimensions of the RE-AIM framework. Adoption could be improved, raising awareness of medical and executive boards for the importance of IDE. Reach could be enhanced by assuming a more flexible patient selection and by finding strategies to reduce IDE rejection. Maintenance was severely affected by voluntary withdrawal. Implementation fidelity (adherence) was mostly affected by voluntarily missed exercise sessions. Despite a substantial implementation dose, IDE did not meet the recommended PA levels. Hence, more personalized approaches could complement IDE, helping HD patients to meet the current PA recommendations with higher health impacts, an objective more likely to be achieved through the involvement of qualified exercise professionals.

FUNDING

This work was funded by National Funds by FCT - Foundation for Science and Technology under the following project UID/04045: Research Center in Sports Sciences, Health Sciences, and Human Development.

DATA AVAILABILITY STATEMENT

The datasets used and analyzed during the current study are available from the corresponding author on reasonable request.

CONFLICT OF INTEREST STATEMENT

None declared.