Effects of aquatic high-intensity interval training on patient-reported outcome measures and quality-adjusted life-years in adults with rheumatic and musculoskeletal diseases: a secondary analysis of the AquaHigh randomised controlled trial

Abstract

Objectives

To examine the effects of 12 weeks group-based peer-led aquatic high-intensity interval training (AHIIT) compared with aquatic moderate continuous training (AMICT) on patient-reported outcome measures (PROMs) and quality-adjusted life-years (QALYs).

Design

A single-blind, parallel-group, randomised trial with a 1:1 allocation ratio.

Setting

Community-based setting.

Participants

89 participants (mean age 62 (SD 13) years) with rheumatic and musculoskeletal diseases, including hip and knee osteoarthritis, fibromyalgia, rheumatoid arthritis, psoriatic arthritis and axial spondyloarthritis, were randomly allocated to an AHIIT (n=44) or an AMICT (n=45) group.

Interventions

The intervention consisted of AHIIT (four intervals of 4 min at high intensity, Borg scale 14–18) or AMICT (Borg scale 12–13), conducted twice weekly for 12 weeks.

Main outcome measures

Outcomes included disease activity (measured by the Patient Global Assessment), fatigue, pain and health-related quality of life (HRQoL), measured by the EQ-5D utility index (five-dimensional health status measure) and EQ VAS (self-rated overall health scale) for overall health, physical and social activities. All outcomes were assessed at baseline, 3 months and 6 months. To compare the overall benefit of these interventions, QALYs were estimated based on HRQoL. Linear mixed models for repeated measures were used to estimate the mean difference (95% CI) in outcomes.

Results

No statistically significant differences between the groups were found in any outcomes at either three or 6 months (p>0.05).

Conclusion

No difference between the groups was found on PROMs and QALYs. Future research should include larger sample sizes and a non-exercising control group to better determine the efficacy of AHIIT and clarify the role of exercise intensity in symptom management.

Trial registration number

NCT05209802.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• The study compared two different exercise intensities (aquatic high-intensity interval training (AHIIT) and aquatic moderate continuous training (AMICT)) that were peer-led.

• The study was conducted across multiple sites and included participants with various rheumatic and musculoskeletal diseases (RMDs), enhancing the generalisability of the findings to common clinical settings.

• The randomised controlled trial (RCT) design and intention-to-treat analysis reduce bias and provide an accurate assessment of interventions’ effectiveness.

• Wide confidence intervals, a small sample size and participant dropouts together suggest uncertainty in the data and increase the risk of a type II error, potentially affecting the reliability of the findings.

Introduction

Rheumatic and musculoskeletal diseases (RMDs) significantly impact physical function and health-related quality of life (HRQoL).^{1 2} Furthermore, their global burden is growing.¹,³ These diseases, including osteoarthritis (OA), fibromyalgia (FM), rheumatoid arthritis (RA), psoriatic arthritis (PsA), axial spondyloarthritis (AxSpA) and other autoimmune, degenerative or inflammatory diseases, often lead to chronic pain, fatigue and reduced life expectancy.⁴

Unpredictable disease activity and fluctuating fatigue are common, disabling and challenging symptoms that significantly affect daily life and HRQoL in people with RMDs.⁵,⁷ Fatigue may impact a person’s disease remission.⁵,⁷ Disease-modifying medications alone do not always succeed in reducing symptoms such as fatigue and pain.⁸ While some RMDs have options for disease-modifying medications, many, including systemic lupus erythematosus⁹ and OA, have no effective disease-modifying treatments.¹⁰ Additionally, disease activity flares are associated with both poor clinical and patient-reported outcomes and can occur even after prolonged periods in remission.^{11 12} Therefore, additional strategies beyond pharmacological treatments are necessary to manage these symptoms effectively.⁸¹³,¹⁶ There is some evidence indicating that exercise can be a strategy to reduce disease activity and alleviate symptoms in individuals with inflammatory RMDs, with exercises typically being of low to moderate intensity.¹⁷

High-intensity interval training (HIIT) offers superior and time-efficient health benefits compared with moderate-intensity training for various disease groups, particularly in enhancing aerobic capacity.¹⁸,²² However, clinicians have to some extent avoided HIIT to prevent flare-ups of disease activity in people with inflammatory RMDs.^{23 24} While there are a limited number of land-based HIIT studies for inflammatory RMDs, these studies have shown either improvements²⁵ or no change²⁶,²⁸ in disease activity and fatigue. Furthermore, these land-based HIIT studies were conducted in specialised healthcare settings²⁵,²⁷ or in the community,²⁸ and all were led by physiotherapists.²⁵,²⁸ Given that these studies only have a non-exercising control group, there is a dearth of studies with a moderate-intensity or low-intensity exercise group for comparison. Comparing HIIT to moderate intensity training may be key to understanding the effect of exercise intensity on patient-reported outcome measures (PROMs). Additionally, including quality-adjusted life-years (QALYs) may allow for a more comprehensive assessment of the impact of exercise intensity on both the quality and quantity of life.

Aquatic therapy is highly valued by individuals with RMDs,^{29 30} often because it provides a low-impact environment that reduces pain and improves mobility.^{31 32} It is often offered as group-based exercise programmes which may enhance commitment by fostering a sense of belonging, support, mutual learning³³ and enjoyment,³⁴ particularly among older adults who value social interaction.^{35 36} The AquaHigh randomised controlled trial (RCT) demonstrated that group-based aquatic HIIT (AHIIT) significantly improved aerobic capacity compared with aquatic moderate continuous-intensity training (AMICT), led by peers as exercise leaders.³⁷ A recent systematic review also found AHIIT to significantly enhance exercise capacity in individuals with various chronic conditions, with effects comparable to land-based HIIT.³⁸ Building on these findings, we hypothesised that higher levels of aerobic capacity achieved through AHIIT may have a positive impact on several health outcomes and yield higher QALYs than moderate levels of exercise. To our knowledge, no studies have compared the effect of AHIIT and AMICT in a broader population of individuals with RMDs on PROMs. Aquatic exercise leverages hydrodynamics and hydrostatics for higher-intensity workouts with less cardiovascular stress and joint impact.³¹ Buoyancy decreases joint load, while hydrostatic pressure provides resistance support.^{31 32} Warm water may alleviate muscle spasms and pain.^{31 32} For adults with RMDs, the preference for water-based exercise³⁹ and the supportive environment⁴⁰ may contribute to long-term compliance.

Exercise guidelines for RMDs recommend considering patients’ needs, preferences and abilities,^{41 42} which is crucial given the recent evidence supporting aquatic therapy. Recent European Alliance of Associations for Rheumatology (EULAR) recommendations show strong evidence for aquatic therapy in reducing the burden of RMDs,⁴³ particularly in OA (with regards to pain, function and HRQoL),⁴⁴,⁴⁶ moderate evidence in RA (for pain and function)^{47 48} and low evidence in AxSpA (for pain, function and disease activity).⁴⁹,⁵¹ However, effect sizes are small⁴³ and various types of exercise face limitations in achieving clinically meaningful improvements in knee OA outcomes.⁵² Only one AHIIT RCT (in RA) showed significant improvement in disease activity,⁵³ while another found no difference in HRQoL between AHIIT and control group (in OA).⁵⁴ Given the complexity of symptoms, the barriers to traditional land-based exercise programmes, and the constraints on healthcare resources, it is essential to assess the impact of a peer-led AHIIT exercise programme on patient-reported outcomes. Involving trained peer leaders in group-based community settings could be a more appealing, sustainable and resource-effective alternative to delivering exercise services. Therefore, the aim of this secondary analysis was to examine the effects of a 12-week AHIIT compared with AMICT on PROMS including disease activity, fatigue, pain, HRQoL, physical and social activities in people with RMDs. Additionally, the aim was to compare the overall benefit of these interventions, integrating change in HRQoL and estimating QALYs.

Methods

A supporting Consolidating Standards of Reporting Trials (CONSORT) for this trial is available as supplementary information (online supplemental file 1).

Study design

This study was a secondary analysis of the AquaHigh study, examining the effects of a 12-week AHIIT compared with AMICT.³⁷ The study protocol was registered at ClinicalTrials.gov (NCT05209802) and approved by the Regional Committee for Medical and Health Research Ethics (272749).

The exercise interventions were supervised by volunteer peers affiliated with the local chapters of the Norwegian Rheumatism Association (NRF).

Participants

Participants meeting the inclusion criteria ≥18 years old, diagnosed with (any kind of) RMDs (confirmed by a doctor), able to walk with or without a walking aid and able to understand Norwegian, were recruited from three municipalities in Eastern Norway. Exclusion criteria were medical contraindications to high intensity exercise, current or recent participation (within the last 3 months) in AHIIT or land based HIIT programmes or trials. Local chapters (locations) from the patient organisation of the NRF recruited adult members with RMDs (ie, hip and knee OA, FM, RA, PsA, RA, systemic sclerosis, Sjögren’s syndrome, systemic lupus erythematosus, large vessel vasculitis, mixed connective tissue disorders). Recruitment methods included flyers, emails, local webpages and promotion visits in local chapters of NRF from a researcher (HB-N). Additionally, participants were also recruited from Healthy Life Centres and local physiotherapy institutes. Participants were initially screened via telephone, and the outcome measures were assessed at baseline, at 3 months and further at 6 months.

Patient and public involvement

The project was initiated by the NRF and developed based on user experiences, feedback and evidence-based knowledge. The instructors are experienced users and have completed instructor courses through the NRF. Local chapters of the NRF and the instructors have been pivotal in the recruitment, planning and execution of the project. Additionally, we have involved a reference group with a user representative to evaluate outcome measures. Users have been central to many aspects of the project.

Exercise interventions

Group-based exercise sessions were led by experienced volunteer peer instructors affiliated with the local NRF chapters, using five different aquatic facilities located within three municipalities. Before the start of the intervention, the instructors attended a half-day workshop led by the lead researcher (HB-N). The Borg 6–20 Scale was used to monitor exercise intensity by the rating of perceived exertion (RPE).⁵⁵ Heart rate was assessed using a watch with a chest strap (Swim 2, Garmin, USA). These watches were utilised during four of the visits to ensure the exercise intensity was maintained or increased as per protocol. They served to monitor the level of exertion and to guide RPE and encourage participants to maintain or increase their intensity. The sessions were predominantly accompanied by music and conducted with the instructor either in the pool or on land. The AMICT intervention had a limited focus on the perception of exertion during the sessions, emphasising more on the performance of strength, balance and flexibility exercises. The lead researcher (HB-N) supervised the instructors throughout the intervention period, providing both online and face-to-face guidance, visiting each site on a bi-weekly basis.

The AHIIT intervention has been detailed in a previous study.³⁷ The programme included a warm-up, followed by four high-intensity intervals (RPE of 14–18) each lasting 4 min. These exercises activated large muscle groups in the upper and lower limbs. Light recovery exercises lasting 2–3 min were performed between the high-intensity intervals. The AMICT group intervention followed the usual structure of the local chapter’s aquatic exercise sessions, alternating between flexibility, endurance, strength and balance exercises to maintain a moderate continuous intensity (RPE 12–13).

Each group consisted of 5–15 participants who undertook exercises in water temperatures between 32 and 34°C, with immersion up to the xiphoid process level. Both groups completed training sessions lasting between 45 and 60 min, held twice a week on non-consecutive days for 12 weeks. Following the initial 12 weeks, participants were encouraged to continue an active lifestyle and/or participate in aquatic exercise at the local NRF chapters.

All participants were familiarised with the RPE during baseline assessment. Participants were encouraged to document their training participation, RPE and any additional physical activities they engaged in a diary after each session. The total number of completed sessions was counted from diaries. Attendance was calculated as a percentage of the total sessions (24 sessions=100%).

Participants were encouraged to document any adverse events, pain or exercise-related symptoms in a diary after each session. Instructors were instructed to report adverse events and abnormal exercise-related symptoms to the researcher (HB-N).

Outcome measures

Disease activity was measured using the Patient’s Global Assessment (PGA).⁵⁶ The PGA of the American College of Rheumatology (ACR) and EULAR is based on the question ‘Considering all ways your RMD affect you, how do you feel your RMD has affected you the past week?’. The PGA was reported on a 100-mm visual analogue scale (VAS), with higher scores indicating higher levels of disease activity. A PGA score of <20 indicates low disease activity.^{57 58} A reduction of ≥8.4 points is seen as a meaningful improvement, whereas an increase of ≥11.5 is considered a meaningful worsening.⁵⁹

Fatigue was evaluated using the Bristol RA Fatigue Multi-Dimensional Questionnaire (The BRAF-MDQ). Respondents were asked to rate their fatigue over the past 7 days on a numeric rating scale (NRS) from 0 (no fatigue) to 10 (worst fatigue). The NRS scores covered various aspects of fatigue: physical fatigue (ranging from 0 to 22), living with fatigue (ranging from 0 to 21), cognitive fatigue (ranging from 0 to 15), and emotional fatigue (ranging from 0 to 12). In addition, a total BRAF-MDQ score was calculated, which could range from 0 to 70.^{60 61} The questionnaire has demonstrated both construct and criterion validity,⁶⁰ and good reliability and sensitivity to change.⁶² A 17.5% relative change or a reduction of 7.43 points was considered a clinically important improvement in BRAF total.⁶²

Pain was assessed using NRS, ranging from 0 to 10, where 0 denotes no pain and 10 represents the worst imaginable pain. This measurement was taken both at rest and during activity. A 30% relative change or a reduction of 2 points is considered a clinically important change in NRS pain.^{63 64}

HRQoL was measured by the EQ-5D-5L, a standardised instrument developed by the EuroQol Group.⁶⁵ The EQ-5D-5L consists of two parts: a descriptive system and a visual analogue scale (EQ VAS). The descriptive system assesses HRQoL across five dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression, with each dimension having five levels of severity, ranging from no problems (0) to extreme problems (5). These five health states were converted into the EQ-5D health utility index, on a scale of 1 to 0 to reflect how ‘good’ (1) or ‘bad’ (0) a health state is according to the preferences of the general population using a value set with UK tariffs as recommended by national health economic evaluation standards.⁶⁶

QALYs were calculated using the EQ-5D health utility index.⁶⁷ QALYs are a two-dimensional concept combining both the quality and quantity of life (ie, a result of the health-related utility state and the duration or frequency of being in that state).⁶⁸

Overall health was reported on the EQ VAS between 0 (‘worst imaginable health’) and 100 (‘best imaginable health’).⁶⁵ For the EQ-5D utility index, a change of ≥0.07 points is considered a meaningful improvement, while a worsening change of ≥0.05 points is considered meaningful.⁶⁹ For the EQ-5D VAS, a change of ≥7 points is defined as a meaningful change.⁶⁹

The current ability to participate in social roles and activities was measured with the 8-item Patient-Reported Outcomes Measurement Information System (PROMIS Item Bank, Ability to Participate in Social Roles and Activities 8a Short Form (V.2.0), APA-8a). The PROMIS APA-8a is an 8-item questionnaire. A higher score (>50) represents a better ability to participate and was also reported as a T-score.⁷⁰ PROMIS APA-8a is responsive to changes in RA patients⁷¹ and has adequate measurement properties.⁷² A score of 50 (SD 10) represents the mean of a relevant reference population.⁷³

Physical activity level was evaluated with The University of California, Los Angeles (UCLA) Activity Scale. It is a 10-point scale based on 10 descriptive activity levels ranging from 0 (wholly inactive) to 10 (regular participation in impact sports).⁷⁴,⁷⁶ UCLA activity scale is valid at group level⁷⁷ and reliable⁷⁶ for patients following joint replacement.⁷⁸

Physical activity habits were quantified by questions included grading frequency, intensity and duration of physical activity.⁷⁹ The physical activity index is considered valid to assess leisure time physical activity.⁸⁰ Furthermore, the activity index is reported to have high test-retest reliability. A product of frequency, intensity and duration gave a summary index (ranging from 0 to 45).⁸¹

Demographic information including age, sex, body mass index (BMI, kg/ height in m²), educational status, employment and marital status, together with details on disease-related variables, was collected at baseline.

Sample size

The sample size estimation was based on the primary outcome of aerobic capacity in the AquaHigh study.³⁷ Accordingly, our goal was to recruit 84 patients. Power calculations were not conducted for the outcome measures reported in this study.

Randomisation and blinding

Participants were randomly allocated to either the intervention group (AHIIT) or the control group (AMICT) using a computer-generated sequence created by an independent statistician. Block randomisation was used to ensure a balanced distribution of participants across three distinct locations. A researcher, who was not involved in participant recruitment, testing or intervention, prepared the sealed envelopes. Group assignment took place after the initial testing phase, conducted by another researcher who was kept unaware of the participants’ test outcomes.

Eight trained physiotherapists or personal trainers, who were unaware of group assignments, assisted with questionnaires and were outcome assessors. After baseline testing, the participants contacted the researcher, who had no knowledge of the test results. The researcher unsealed envelopes and randomly assigned the participants to their respective groups in a 1:1 ratio. Concealing the group identities from both the participants and instructors was not feasible.

Statistical analysis

Summaries of descriptive statistics were provided using mean values and SD, or percentages and frequencies. The comparison of mean values and proportions among those who completed the study versus those who did not was performed using the student’s t-test and the χ² test. Our outcome analyses were conducted following an intention-to-treat approach, utilising all accessible data at every time point. To estimate the mean difference with a 95% CI, we used a linear mixed model for repeated measures. Fixed effects included in the model were treatment, time, interaction between treatment and time, location and the baseline value of the outcome variable. Mean differences were estimated at 3 and 6 months, with the associated 95% CIs and p values. We executed a pre-post analysis as a within-group analysis for sensitivity analysis, both using the mixed model approach, similar to the main analysis. All statistical analyses for the main analysis were performed using IBM SPSS Statistics V.29. Throughout the analyses, a 5% significance level was applied.

QALYs were estimated using a linear mixed effects model proposed by Gabrio et al,⁸² which is a variation of the traditional area-under the curve method that summarises individual measurements over time and handles missing data under the missing data at random assumption. Health utilities, derived from the EQ-5D-5L at baseline, 3 and 6 months, were weighted to account for the importance of health states over time. The weighted utility scores were multiplied by the duration of each health state and summed to estimate total QALYs. QALY analyses were conducted using R Core Team V. 2023.06.1+524.

Results

Participants and baseline characteristics

From February 2022 to June 2022, 102 individuals with RMDs were screened for eligibility, and 89 participants were enrolled in the AquaHigh trial (figure 1). The mean age was 62 (SD 13) years, and 91% of the participants were women (table 1). The enrolled participants had one or more concurrent diagnoses, including 53 (60%) with OA, 34 (38%) with FM, 24 (27%) with RA, 11 (12%) with PsA and 6 (7%) with AxSpA. Additionally, 8 participants (9%) had other diagnoses, such as systemic sclerosis, Sjögren’s syndrome, systemic lupus erythematosus, large vessel vasculitis and mixed connective tissue disorder, either individually or in combination with other RMDs.

Figure 1. Flow (CONSORT) of participants through the randomised controlled trial. AHIIT, aquatic high-intensity interval training; AMICT, aquatic moderate-intensity continuous training; CONSORT, Consolidating Standards of Reporting Trials.

Table 1. Participants’ demographics and health-related conditions^*.

	Total n=89	AHIIT n=44	AMICT n=45
Gender, Women, n (%)	81 (91)	41 (93)	40 (89)
Men	8 (9)	3 (7)	5 (11)
Age, y (mean, SD)	62 (13)	60 (12)	64 (13)
BMI	29 (4.9)	28.5 (5)	29.6 (5)
Relationship status, cohabits, n (%)	56 (63)	28 (64)	28 (62)
Education, >12 years, n (%)	38 (43)	17 (39)	21 (47)
Full-time employment, n (%)	17 (19)	11 (25)	6 (13)
Current smoker, yes, n (%)	8 (9)	3 (7)	5 (11)
Health-related conditions, yes, n (%)
Inflammatory disease	42 (47)	21 (48)	21 (47)
Joint replacement	17 (19)	7 (16)	10 (22)

^*There were no statistically significant differences between the AHIIT (aquatic high-intensity interval training) and the AMICT (aquatic moderate continuous training) group on any of the descriptive variables at baseline. BMI Body Mass Index.

Adherence to programme, training intensity and adverse events

A total of 66 participants completed the 3 months assessments after the 12 week intervention. In the AHIIT group, 28 diaries were returned compared with 21 in the AMICT group. A total of 16 cases of SARS-CoV-2 infection (COVID-19) (with interruptions in training) were reported during the interventions, equally distributed between groups. The reported training intensity in the AHIIT was a mean Borg RPE of 15 (SD 2) and in the AMICT, it was 13 (SD 2) (the between group differences were p<0.001). There was no difference between groups in numbers of completed sessions. Three participants in the AMICT group reported a training intensity of RPE >15, while seven in the AHIIT group reported a lower intensity than encouraged, RPE <15. There were no adverse events reported during the intervention by either the participants or the instructors.

Patient-reported outcome measures from baseline to 3 months

There were no statistically significant differences between groups in any outcomes (p>0.05). However, there were statistically significant within-group improvements in the AHIIT group in disease activity and in the five HRQoL dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression (p<0.05). In addition, there was a statistically significant within-group improvement in the AMICT group in fatigue and in EQ VAS in both groups (p<0.05). See table 2 for unadjusted means, tables3 4 for adjusted means (SDs) and figure 2 for the graph of results.

Table 2. Descriptive statistics of patient-reported outcome measures at baseline, 3 months and 6 months^*.

	Total n=89	AHIIT n=44	AMICT n=45	AHIIT 3 months n=35	AMICT 3 months n=31	AHIIT 6 months n=29	AMICT 6 months n=24
PROMs (mean, SD)
Disease activity (PGA)(0–100 mm) 0=no disease activity	47.7 (25.2)	45.8 (25.7)	49.5 (24.8)	32.9 (23.7)	40.2 (27.4)	35.4 (25.5)	47.3 (27.5)
Fatigue (BRAF physical) (0–22) 0=no fatigue	10.4 (5.9)	9.8 (5.8)	11.0 (5.9)	8.0 (4.9)	8.8 (6)	7.5 (5.1)	9.0 (6.4)
Fatigue (BRAF living) (0–21) 0=no fatigue	7.1 (5.2)	6.6 (4.8)	7.5 (6)	4.1 (4.1)	5.8 (5.4)	4.7 (3.8)	5.2 (5.2)
Fatigue (BRAF cognitive) (0–15) 0=no fatigue	5.3 (4.1)	4.9 (3.6)	5.8 (4.6)	4.0 (3.8)	4.13 (3.8)	4.5 (3.9)	4.7 (4.3)
Fatigue (BRAF emotion) (0–12) 0=no fatigue	3.9 (3.2)	3.7 (2.9)	4.1 (3.4)	2.6 (2.9)	2.8 (2.9)	2.9 (3.1)	2.9 (3.1)
Fatigue (BRAF total) (0–70) 0=no fatigue	26.7 (16.8)	25 (15.6)	28.4 (17.8)	18.4 (13.4)	21.4 (16.4)	19.6 (14.4)	21.8 (17.7)
Pain (NRS 0–10), at rest 0=no pain	4.0 (2.1)	3.9 (2.0)	4.1 (2.3)	3.4 (2.0)	4.05 (2.4)	3.3 (1.9)	4.3 (2.1)
Pain (NRS 0–10), during activities, 0=no pain	5.3 (2.2)	5.0 (2.3)	5.6 (2.2)	4.6 (2.1)	5.0 (2.3)	4.5 (1.9)	4.8 (1.9)
EQ-5D-5L (mobility)	2.0 (.91)	2.0 (.92)	2.0 (.91)	1.7 (.73)	2.1 (.87)	1.7 (.77)	2.0 (.72)
EQ-5D-5L (self-care)	1.3 (.58)	1.4 (.66)	1.3 (.51)	1.2 (.49)	1.3 (.52)	1.2 (.64)	1.2 (.42)
EQ-5D-5L (usual activities)	2.0 (.81)	2.0 (.78)	2.0 (.85)	1.7 (.75)	1.9 (.81)	1.6 (.63)	1.8 (.92)
EQ-5D-5L (pain)	3 (.72)	3.0 (.66)	3.0 (.78)	2.5 (.74)	2.9 (.76)	2.7 (.81)	2.7 (.69)
EQ-5D-5L (anxiety)	1.7 (.86)	1.7 (.88)	1.7 (.85)	1.5 (.74)	1.6 (.85)	1.6 (.94)	1.6 (.83)
EQ VAS (overall health) (0–100 mm), 100=best	55.7 (17.6)	56.9 (16.2)	54.5 (19.3)	63.2 (17.6)	62.3 (17.5)	68.9 (14.7)	63.4 (23.9)
EQ-5D (index)	.716 (.185)	.718 (.178)	.715 (.193)	.801 (.122)	.739 (.173)	.785 (.162)	.772 (.158)
Physical and social activities
PROMIS sum score	27 (6.9)	28.3 (6.1)	26.0 (7.4)	29.2 (5.9)	28.3 (6.1)	31.6 (4.8)	28.2 (7.5)
PROMIS T score	47.5 (7.6)	48.7 (7.1)	46.2 (7.8)	48.2 (7.1)	46.9 (9.5)	51.8 (5.7)	49.9 (8.5)
Physical activity index (0–45 points) 45=best	9.6 (9)	9.5 (9.1)	9.6 (9.1)	14.4 (10.2)	11.6 (10.6)	11.3 (8)	11.3 (9.6)
Physical activity level (UCLA) (0–10 points) 10=best	5.7 (1.8)	5.7 (1.9)	5.6 (1.8)	6.5 (1.7)	6.1 (1.6)	6.6 (1.8)	6.0 (2.0)

^*There were no statistically significant differences between the AHIIT (aquatic high-intensity interval training) and the AMICT (aquatic moderate-intensity continuous training) group on any of the descriptive variables at baseline. BRAF the Bristol RA Fatigue Multi-Dimensional Questionnaire, EQ-5D-5L measure of health-related quality of life, NRS numeric rating scale, PGA Patient global assessment disease activity, PROMs patient-reported outcome measures, PROMIS Ability to participate in Social Roles and Activities (short form 8 a), University of California, Los Angeles (UCLA) physical activity scale.

Table 3. Effect of AHIIT compared with AMICT at 3 and 6 months.

	3 months Adjusted mean (95% CI)		6 months Adjusted mean (95% CI)		3 months		6 months
	AHIIT n=35	AMICT n=31	AHIIT n=29	AMICT n=24	Mean difference AHIIT-AMICT (95% CI)	p	Mean difference AHIIT – AMICT (95% CI)	p
PROMs (mean, SD)
Disease activity (PGA) (0–100 mm) 0=no disease activity	33.0 (25.0, 41.0)	39.5 (30.9, 48.0)	35.9 (27.1, 44.6)	48.1 (38.5, 57.7)	−6.46 (−18.16, 5.25)	0.276	−12.2 (−25.21,.78)	0.065
Fatigue (BRAF total) (0–70 points) 0=no fatigue	19.6 (16.2, 22.9)	19.4 (15.9, 23.0)	22.4 (18.8, 26.1)	19.6 (15.6, 23.5)	0.12 (−4.81, 5.04)	0.963	2.892 (−2.55, 8.33)	0.294
Pain, rest (NRS 0–10) 0=no pain	3.4 (2.8, 4.0)	4.0 (3.3, 4.7)	3.5 (2.8, 4.2)	4.3 (3.6, 5.1)	−0.60 (−1.50,.30)	0.192	−0.81 (−0.1.81, 0.18)	0.109
Pain, during activities (NRS 0–10) 0=no pain	4.6 (4.0, 5.1)	4.9 (4.3, 5.5)	4.7 (4.0, 5.3)	5.0 (4.3, 5.6)	−0.35 (−1.17,.49)	0.412	−0.31 (−1.22,.61)	0.509
EQ VAS (overall health) (0–100 mm) 100=best	63.4 (58.5, 68.4)	62.6 (57.4, 67.9)	67.9 (62.5, 73.3)	63.5 (57.5, 69.5)	0.79 (−6.46, 8.03)	0.830	4.40 (−3.71, 12.51)	0.284
EQ-5D utility index	0.800 (.759,.840)	0.756 (.712,.800)	0.771 (.728,.815)	0.763 (.716,.811)	0.044 (−0.016,.103)	0.152	0.008 (−0.057, 0.73)	0.810
Physical and social activities
Ability to participate in social roles and activities (PROMIS T score)	48.6 (46.6, 50.6)	48.1 (46.0, 50.2)	50.5 (48.4, 52.7)	49.2 (46.9, 51.6)	0.503 (−2.40, 3.40)	0.731	1.31 (−4.49, 1.88)	0.418
Physical activity level (UCLA) (0–10 points) 10=best	6.5 (6.0, 7.1)	6.1 (5.5, 6.7)	6.6 (6.0, 7.2)	6.0 (5.4, 6.7)	0.47 (−0.34, 1.27)	0.254	0.59 (−0.31, 1.49)	0.193
Physical activity index (0–45 points) 45=best	13.88 (10.86, 16.91)	12.17 (8.96, 15.38)	11.41 (8.14, 14.68)	11.34 (7.76, 14.9)	1.71 (−2.72, 6.14)	0.445	0.066 (−4.81, 4.94)	0.978

AHIIT aquatic high-intensity interval training, AMICT aquatic moderate-intensity continuous training, BRAF the Bristol RA Fatigue Multi-Dimensional Questionnaire, EQ VAS and EQ 5D measures of health-related quality of life and utility, NRS numeric rating scale, Physical activity index physical activity habits, PGA Patient global assessment of disease activity, PROMIS T score ability to participate in social roles and activities (short form 8 a), Adjusted means (95% CI) at timepoint 3 months and 6 months and mean difference between groups (95% CI) and p-value at 3 months and 6 months assessed with longitudinal analysis of covariance.

Table 4. Within-group changes in AHIIT and AMICT from baseline to 3 months and 6 months.

	Adjusted means, 95% CI				Mean difference time (baseline-3 months)		Mean difference time (baseline-6 months)		Mean difference time (3 months-6 months)
	At 3 months		At 6 months
	AHIIT n=35	AMICT n=31	AHIIT n=29	AMICT n=24	Mean difference (95% CI)	p	Mean difference (95% CI)	p	Mean difference (95% CI)	p
					AHIIT	AHIIT	AHIIT	AHIIT	AHIIT	AHIIT
					AMICT	AMICT	AMICT	AMICT	AMICT	AMICT
PROMs (mean, SD)
Disease activity (PGA) (0–100 mm) 0=no disease activity	33.7 (27.3, 40.1)	39.8 (33.0, 46.6)	36.8 (29.8, 43.8)	48.8 (41.1, 56.5)	−12.49 (-20.8,–4.17)	0.004	−9.40 (-18.2,–0.59)	0.037	3.09 (−6.13, 12.3)	0.509
					−7.49 (−16.08, 1.11)	0.087	1.54 (−7.80, 10.89)	0.745	9.03 (−0.97, 19.0)	0.076
Fatigue (BRAF total) (0–70 points) 0=best	20.1 (17.5, 22.7)	19.5 (16.7, 22.2)	23.1 (20.2, 26.0)	19.5 (16.3, 22.6)	−5.88 (−9.15 to –2.62)	<0.001	−2.925 (−6.40, .545)	0.098	2.96 (.657, 6.57)	0.108
					−7.06 (-10.44,–3.68)	<0.001	−7.045 (-10.7,–3.36)	<0.001	0.010 (−3.91, 3.93)	0.996
Pain (NRS 0–10), rest 10=best	3.4 (2.9, 3.9)	4.0 (3.5, 4.6)	3.6 (3.0, 4.1)	4.3 (3.7, 4.9)	−0.451 (−1.09, .188)	0.165	−0.302 (−0.980, .375)	0.379	0.149 (−0.559, .856)	0.678
					0.079 (.581, .739)	0.814	0.378 (−0.341, 1.10)	0.301	0.299 (−0.468, 1.07)	0.442
Pain (NRS 0–10), during activities, 10=best	4.6 (4.1, 5.0)	4.9 (4.4, 5.4)	4.7 (4.2, 5.2)	5.0 (4.4, 5.5)	−0.627 (−1.21, −0.048)	0.034	−0.483 (−1.10, 0.129)	0.121	0.144 (−0.495, 0.782)	0.658
					−0.437 (−1.03, 0.159)	0.149	−0.361 (−1.01, 0.288)	0.274	0.076 (−0.617, 0.769)	0.829
EQ VAS (overall health) 100=best	63.2 (59.3, 67.2)	62.6 (58.4, 66.7)	67.42 (63.1, 71.7)	63.8 (59.1, 68.6)	7.019 (1.81, 12.23)	0.009	11.20 (5.70, 16.7)	<0.001	4.182 (−1.59, 9.95)	0.154
					7.020 (1.66, 12.39)	0.011	8.260 (2.45, 14.06)	0.006	1.24 (−5.01, 7.49)	0.695
EQ-5D utility index	0.796 (.763, 829)	0.760 (725, 795)	0.759 (723, 796)	0.769 (729, 808)	0.075 (.034, .117)	<0.001	0.039 (−0.006, .083)	0.086	−0.037 (−0.083, .009)	0.116
					0.040 (−0.003, .083)	0.068	0.048 (.002, .095)	0.043	0.008 (−0.041, .058)	0.739
Physical- and social activities
Ability to participate in social roles and activities (PROMIS T score)	48.2 (46.7, 49.7)	48.0 (46.5, 49.6)	50.3 (48.6, 51.9)	49.5 (47.7, 51.3)	0.133 (−1.84, 2.11)	0.894	2.20 (0.112, 4.29)	0.039	2.067 (−0.124, 4.26)	0.064
					0.498 (−1.54, 2.53)	0.630	1.93 (−0.278, 4.14)	0.086	1.431 (−0.943–3.81)	0.235
PA index (0–45 points) 45=best	13.9 (11.5, 16.2)	12.6 (10.2, 15.3)	11.1 (8.5, 13.7)	12.3 (9.4, 15.2)	4.64 (1.62, 7.66)	0.003	1.84 (−1.37, 5.04)	0.260	−2.80 (−6.15, 0.542)	0.100
					3.51 (0.378, 6.64)	0.028	3.03 (−0.380, 6.44)	0.081	−0.479 (−4.11, 3.15)	0.794
Physical activity level (UCLA) (0–10 points) 10=best	6.6 (5.2, 7.1)	6.1 (5.6, 6.6)	6.7 (6.2, 7.2)	6.1 (5.6, 6.7)	0.998 (.408, 1.59)	0.001	1.127 (.502, 1.75)	<0.001	0.129 (−0.524, .782)	0.697
					0.521 (−0.087, 1.13)	0.093	0.527 (−0.134, 1.19)	0.117	0.006 (−0.702, .714)	0.987

AHIIT aquatic high-intensity interval training, AMICT aquatic moderate continuous intensity training, BRAF the Bristol RA Fatigue Multi-Dimensional Questionnaire, EQ VAS and EQ 5D measures of health-related quality of life and utility, NRS numeric rating scale, PA index physical activity habits, PGA Patient global assessment of disease activity, PROMIS T score ability to participate in social roles and activities (short form 8 a), adjusted means (95% CI) at timepoint 3 months and 6 months and mean difference between groups (95% CI) and p-value at 3 months and 6 months assessed with longitudinal analysis of covariance.

Figure 2. Time trajectories: mean disease activity and fatigue for AHIIT (aquatic high-intensity interval training) and AMICT (aquatic moderate-intensity continuous training). figure 2b Time trajectories: mean pain at rest and during activities for AHIIT (aquatic high-intensity interval training) and AMICT (aquatic moderate-intensity continuous training). figure 2c Time trajectories: mean EQ ED utility index and EQ VAS for AHIIT (aquatic high-intensity interval training) and AMICT (aquatic moderate-intensity continuous training). figure 2d Time trajectories: mean social role and activities for AHIIT (aquatic high-intensity interval training) and AMICT (aquatic moderate-intensity continuous training). figure 2e Time trajectories: mean physical activity level (UCLA) and physical activity index for AHIIT (aquatic high-intensity interval training) and AMICT (aquatic moderate-intensity continuous training).

Patient-reported outcome measures from baseline to 6 months

There were no statistically significant differences between groups in any outcomes (p>0.05). However, the significant within-group improvement in the AHIIT group in disease activity was sustained (p<0.05). The improvement in EQ VAS observed after the interventions (at 3 months) was maintained in both groups at 6 months (p<0.05). See table 2 for unadjusted means, tables3 4 for adjusted means (SDs) and figure 2a–e for the graph of results.

QALYs

The AHIIT group generated 0.38 QALYs over the study period of 6 months, while the AMICT group generated 0.37 QALYs. The AHIIT group was associated with higher utilities compared with the AMICT group at all follow-ups, which leads to higher QALYs estimates. The mean difference in QALYs was 0.012 (95% CI −0.012 to 0.035), indicating no statistically significant difference between the groups (table 5).

Table 5. Linear mixed model results of QALYs for AHIIT and AMICT ^b.

	AHIIT		AMICT		Incremental
	Estimate	95% CI	Estimate	95% CI	Estimate	95% CI
3 months	0.795	(0.748, 0.841)	0.751	(0.702, 0.800)	0.044	(−0.109, 0.023)
6 months	0.764	(0.710, 0.818)	0.757	(0.699, 0.814)	0.007	(−0.082, 0.067)
QALYs	0.384	(0.364, 0.404)	0.372	(0.351, 0.392)	0.012	(−0.012, 0.035)

^*Point estimates and 95% CI for the marginal means and mean difference of the utilities derived from the EQ-5D at 3 and 6 months follow-up, and for the quality-adjusted life years (QALYs) over the duration of the study for AHIIT (aquatic high-intensity interval training) and AMICT (aquatic moderate continuous training).

Sensitivity analysis

No statistically significant difference was found between responders and non-responders at 3 months (p>0.05) (online supplemental table 6). However, at 6 months, responders were generally older, exhibited better outcomes than non-responders in terms of overall health measured by the EQ VAS, social role and functioning score, and pain during activities (online supplemental table 7). A higher number of participants with FM was categorised as non-responders at 6 months follow-up.

Discussion

This secondary analysis of the AquaHigh RCT found no significant differences between AHIIT and AMICT on disease activity, fatigue, pain, HRQoL, QALYs, and physical and social activities among adults with RMDs. However, both groups demonstrated statistically significant within-group improvements in fatigue and overall health after the intervention. For overall health, these improvements were clinically meaningful, irrespective of exercise intensity. Furthermore, AHIIT showed clinically meaningful improvements in disease activity and in the five HRQoL dimensions: mobility, self-care, usual activities, pain/discomfort and anxiety/depression, while AMICT showed similar improvements in fatigue. To our knowledge, this is the first study examining the effects of AHIIT on PROMs and QALYs in people with RMDs compared with AMICT. These findings may indicate that the supportive properties of water during exercise are valuable in aiding a range of exercise intensities that reduce fatigue. The aquatic environment may provide ease and comfort with exercise potentially facilitated more successful experiences to exercise regimen and improving overall well-being, as well as both the quality and quantity of life for adults with RMDs. Additionally, involving peers as exercise leaders may represent an accessible and sustainable model for delivering exercise services within the community.

Exercise intensity might not be the critical factor for improving disease activity for people with RMDs; however, AHIIT was well tolerated, did not provoke symptoms and did significantly increase aerobic capacity.³⁷ Both groups improved during the study period, indicating that water-based exercises are beneficial even with high baseline disease activity. From baseline to 3 months, the AHIIT group demonstrated clinically meaningful improvements and significant within-group differences. At 6 months, the between-group difference was clinically meaningful in favour of AHIIT. Recent RCTs on land-based HIIT for inflammatory RMDs show mixed results.¹⁷²⁵,²⁸ Sveaas et al reported significant improvements,²⁵ while others found no significant impact.²⁶,²⁸ All sessions were tailored and supervised by physiotherapists, using various low impact cardio equipment,²⁶,²⁸ and compared with a non-exercising control group.¹⁷²⁵,²⁸ However, Siqueira et al⁵³ found that AHIIT significantly improved disease activity compared with land-based HIIT and controls in people with RA. Extending the intervention to 4 weeks and increasing frequency to three times a week, as per Siqueira et al,⁵³ could yield more robust data. AHIIT can be recommended as a non-pharmacological intervention for managing disease activity in adults with RMDs, without concerns about flare-ups.

Participants in the present study commonly reported experiencing fatigue and pain at baseline, with both moderate and high-intensity aquatic exercise proving beneficial in managing these symptoms. Previous research suggests that increased aerobic capacity might reduce fatigue, which is linked to inflammation, pain and quality of life.⁸³ Kargarfard et al⁸⁴ found significant improvement in fatigue and aerobic capacity in people with MS after AHIIT, while Sveaas et al⁸⁵ found that land-based HIIT benefited fatigue in AxSpA patients, though not clinically meaningful. Contrary to these findings, the present study showed improvements in fatigue for both exercise groups, with the AMICT group showing clinically meaningful improvements at 3 months that sustained at 6 months. Christie et al²⁹ support our findings, noting reduced fatigue during moderate-intensity aquatic exercises. However, the mechanisms linking inflammatory RMDs and fatigue remain unclear.⁸⁶ Physical exercise may also induce stress-related analgesia by releasing cortisol and epinephrine, which increase the pain threshold, especially at higher intensities.⁸⁷ Exercise intensities above 50% of maximal oxygen uptake are thought to provide an analgesic effect in chronic low back pain.⁸⁸ However, the present study found similar small improvements in pain during activities for both AHIIT and AMICT, regardless of intensity. The benefits may be due to the unique properties of water, such as reduced joint loading and enhanced proprioceptive feedback,^{31 32} rather than increases in aerobic capacity and exercise intensity. The warmth of the water aids muscle relaxation, while buoyancy and hydrostatic pressure provide support around the body and aid movement.^{31 39} Thus, moderate-intensity aquatic exercise may be a useful self-management tool for reducing fatigue. The unique properties of water allow people with RMDs to perform AHIIT or AMICT without exacerbating their condition, indicating potential for long-term participation in aquatic exercise.

Improvements in utility and clinically meaningful enhancements in overall health suggested long-term health benefits from 12-week aquatic exercise interventions. Baseline overall health (EQ-VAS) values were 25% lower than age-matched healthy adults,⁸⁹ improving to a 12% difference after 6 months. This improvement is supported by literature, which indicates that AMICT increases HRQoL,⁴³ particularly for individuals with OA.⁴⁴,⁴⁶ Furthermore, the findings align with EULAR recommendations, highlighting the potential of aquatic therapy to improve HRQoL in a clinically significant manner in RMDs.⁴³ A recent systematic review supports this, particularly highlighting the benefits of localised water-based exercise, such as muscle strengthening in warm water, for individuals with OA with severe baseline values of pain, stiffness and functional limitation.⁹⁰ However, the improvements found in overall health in our study contrast with the findings of a single study that found no difference in HRQoL between AHIIT and control groups in individuals with OA.⁵⁴ One explanation for the gains in HRQoL in the present study may be the social interaction facilitated by group-based aquatic exercise.^{33 91} Exercising with peers can enhance well-being, support and commitment.³⁴ While these findings are promising, it is important to acknowledge that the clinical relevance of small, short-term improvements in function and symptoms has been questioned in recent literature.⁵² This underscores the need to also consider the broader health benefits of exercise, including its impact on general health and comorbidities, when making treatment decisions. Regardless of exercise intensity, aquatic exercise significantly improves PROMs and reduces symptoms, underscoring its importance in the management of RMDs. However, we cannot rule out the possibility of a Hawthorne effect,⁹² whereby participants may have altered their behaviour due to increased attention and awareness of being observed, potentially contributing to the observed improvements.

Strengths and limitations

A strength of the present study was that it was a multisite RCT. QALYs were also estimated. The use of an intention-to-treat analysis, incorporating all available data in a linear mixed model for repeated measures, was also a significant strength. Participants in this study likely represent the diverse population living in the municipality who could benefit from aquatic exercise sessions. However, the study has limitations. As participation in the exercise intervention was voluntary, it is possible that individuals who chose to take part were more motivated and engaged than those who did not. This self-selection, combined with a predominance of female participants, may limit the generalisability of the findings to the broader population. The small sample size and substantial dropout rate may have introduced attrition bias, particularly affecting the interpretation of secondary outcomes. While non-responders and responders were comparable at 3 months, significant differences at 6 months (eg, age, FM diagnosis, pain levels) suggest potential bias in long-term estimates. These limitations may have reduced statistical power and precision, increasing the risk of a type II error. Furthermore, the results were estimated with low precision, as reflected by the wide CIs, indicating substantial uncertainty or variability in the data, possibly related to population heterogeneity. Concurrent diagnoses may have confounded the results. The interpretation of clinically meaningful improvement follows a commonly used approach, focusing on mean values. However, it can be applied differently,⁹³ which may influence how clinical relevance is assessed. The COVID-19 pandemic, which ended in 2022 in Norway, also affected dropouts due to viral infections during the intervention, postponement of the testing time and post-viral fatigue. These factors may have affected the participants' quality of life. All outcome measures were self-reported, and the lack of blinding may have introduced detection bias.⁹⁴ Furthermore, most HIIT or AHIIT studies have been conducted with a non-exercising control group, which limits direct comparison with other studies. These limitations may have impacted the accuracy and precision of the QALY estimates, affecting the interpretation of the long-term benefits of the interventions.

Future directions and applications

The findings from this study illustrate the potential of using peers as instructors to enhance patient-reported outcomes, regardless of exercise intensity. This innovative approach to exercise delivery can be applied in various clinical and research settings to improve patient engagement and adherence to treatment plans. By leveraging the social support and relatability of peer instructors, healthcare providers can offer more personalised and effective interventions elsewhere. Additionally, these findings can inform patient counselling strategies, emphasising the importance of peer support in achieving better health outcomes. Estimating QALYs standardises the health benefits of interventions like AHIIT, assessing their impact on quality of life and longevity, and aids in comparing cost-effectiveness to inform healthcare decisions. Additionally, future studies should consider strategies to minimise dropout and assess reasons for attrition, as this would provide valuable insights into its impact on outcomes and improve reliability of cost-effectiveness analyses.

Conclusion

There were no significant overall group differences on PROMs and QALYs between AHIIT and AMICT. Future research should include larger sample sizes and a non-exercising control group to better determine the efficacy of AHIIT and clarify the role of exercise intensity in symptom management.

Data availability statement

The data supporting the findings of this study are not publicly available as participants have not consented to their data to be publicly available. However, data may be available from the corresponding author on reasonable request with permission from Oslo Metropolitan University and of an ethics committee.