Effect of different types of exercise in fibromyalgia syndrome: a network meta-analysis

Abstract

Background

Patients with fibromyalgia syndrome (FMS) often exhibit poor health and reduced physical function. Exercise interventions can effectively alleviate symptoms and improve function. This study aims to systematically evaluate and compare the efficacy of different exercise interventions on the key clinical outcomes to identify the most effective modality and inform evidence-based exercise recommendations for patients with FMS.

Methods

This study conducted a systematic search of four databases: PubMed, Embase, the Cochrane Library, and Web of Science. The primary outcome measures included health status, pain intensity, the number of tender points (TP), and aerobic capacity. Stata 15.0 and R software were used for network meta-analysis (NMA), and the effect size and its 95% confidence interval were calculated using a random-effects model. The methodological quality of the included studies was assessed using the Cochrane RoB 2.0 tool and the TESTEX scale.

Results

Stretching combined with balance training resulted in the greatest improvement in health status (MD = − 25.13; 95% CI [− 33.81, − 16.44]). For other symptom-related outcomes, aerobic exercise (AE) combined with strength and stretching showed the highest probability of alleviating pain intensity (MD = − 3.61; 95% CI [− 4.40, − 2.82]). AE alone was most effective in reducing the number of tender points (MD = − 2.28; 95% CI [− 3.57, − 1.00]), while whole-body vibration (WBV) produced the largest enhancement in aerobic capacity (MD = 55.0; 95% CI [26.14, 83.75]).

Conclusions

Stretching combined with balance training was found to be most effective in improving health status. AE combined with strength training and stretching demonstrated the greatest efficacy in reducing pain intensity, while AE was most effective in decreasing the number of tender points. WBV appeared to provide the largest improvement in aerobic capacity in patients with FMS. These findings suggest that healthcare providers should combine AE, strength, stretching, balance, WBV, and other exercises based on the individual circumstances of patients with FMS, while further high-quality RCTs are warranted to confirm these results and to standardize exercise protocols.

Trial registration

CRD420251036473.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13102-025-01432-8.

Background

FMS is a chronic, widespread, and non-inflammatory musculoskeletal disorder that profoundly affects patients’ quality of life and relationships [1, 2]. It is characterized by widespread musculoskeletal pain, accompanied by fatigue, cognitive dysfunction, sleep disturbances, decreased concentration and memory, negative mood, and a variety of somatic symptoms [3]. The etiology of the disease is currently unknown, and it is typically characterized as a disorder affecting the musculoskeletal and connective tissue systems [4]. Some studies attribute its origin to central sensitization processes [5], inflammation [6], or small fiber neuropathy [7]. It is the third most common musculoskeletal disorder, following low back pain and osteoarthritis, with a prevalence ranging from 3% to 10% in the global population [8]. It affects women more frequently than men [9], particularly those aged between 30 and 60 years [10], and it leads to significant disability and a substantial healthcare burden, either directly or indirectly [11]. There are no specific therapies for treating FMS patients. Instead, multimodal pharmacological and non-pharmacological treatments are recommended, each with its own potential risks and side effects. Only 10% to 25% of patients with FMS who receive medication experience a reduction in pain intensity of at least 50%. Many patients also fear addiction or adverse effects from the medication [12]. This inadequate treatment not only necessitates efficient healthcare services but also imposes a significant economic burden [13]. A Canadian study estimated the annual healthcare expenditure per individual to be approximately CAD 3800 [14]. Therefore, for both medical and economic reasons, incorporating exercise therapy into treatment programs is essential.

Regular exercise training is recommended as an effective treatment strategy to improve pain, quality of life, physical functioning, sleep, depression, and anxiety in patients with FMS [15]. Regimens for FMS typically include aerobic, resistance, and stretching exercises. Numerous studies have examined the effects of AE [16], stretching exercise [17], resistance exercise [18], AE combined with strengthening exercise [19], traditional Chinese exercise [20], and AE combined with resistance exercise [21] on patients with FMS, using pain symptoms as an outcome indicator in meta-analyses. A recent study analyzed the effect of different exercise doses of intervention based on frequency, intensity, type, and duration [22]. One study compared the efficacy of pool-based and land-based aerobic exercise interventions [23].

In recent years, some network meta-analyses have compared pain intensity in female FMS patients; their focus has mainly been on pain as a single outcome, and the interventions included have been limited [24]. Other systematic reviews and meta-analyses tend to use pairwise meta-analysis methods [15], which means they cannot be compared directly in the same model. Furthermore, there is still insufficient systematic evaluation of other key outcomes (e.g., health status, aerobic capacity) beyond pain.

Therefore, this study aims to construct an NMA framework that encompasses multiple forms of exercise interventions and compares their effects on improving health status, reducing pain intensity and the number of tender points, and enhancing aerobic capacity. This approach will provide FMS patients with more comprehensive evidence-based treatment recommendations and address the limitations of existing studies in terms of the scope of interventions and outcome dimensions.

Methods

Registration

This study follows the Preferred Reporting Items (PRISMA) statement for systematic reviews and meta-analyses [25]. The study is registered with PROSPERO under registration number CRD420251036473. The PRISMA checklist is provided in Supplementary Material 1.

Search strategy

Articles were obtained from PubMed, Embase, the Cochrane Library, and Web of Science through a systematic electronic search. The searches were performed on February 7, 2025, using a combination of subject terms and free-text keywords. Specific search strategies for each database are detailed in Supplementary Material 2.

Inclusion criteria

The inclusion criteria follow the PICOS (Participants, Intervention, Comparison, Outcome, Study Design) strategy.

1. Study design

   Randomized controlled trials are included, with no restrictions on blinding method or allocation concealment method. Only English language studies are included.

2. Population

   Inclusion criteria: adults (≥18 years old) with FMS, diagnosed according to the American College of Rheumatology (ACR) criteria; no restriction on disease duration; exclusion of patients with severe cardiovascular, respiratory, or neuromuscular diseases, or other diseases affecting physical function.

3. Intervention

   Any form of therapeutic exercise (TE), including single-mode (e.g., AE, strength training, stretching, balance training, WBV, exergame) or pre-defined fixed-combination mode (e.g., AE+ strength + stretching, AE + stretching, stretching + balance, resistance + stretching), with the content of the intervention clearly described in the study, including the elements of training, frequency, intensity, duration of each session, and intervention cycle.

4. Control group

   Conventional care, placebo intervention, no intervention, or other exercise intervention modes (used for comparing nodes in network meta-analyses). The type of control must be clearly stated in the included studies and categorized uniformly in the analysis.

5. Result indicators

   Health status (Fibromyalgia Impact Questionnaire, FIQ), pain intensity (Visual Analogue Scale, VAS), number of tender points (Tender Points, TP), and aerobic capacity (6-minute walk test, 6MWT).

Exclusion criteria

1. Repeatedly published studies, review articles, conference abstracts, animal experiments, laboratory-based research, and case reports.(1) Repeatedly published studies, review articles, conference abstracts, animal experiments, laboratory-based research, and case reports.

2. Studies that combine physical intervention with other interventions (e.g., educational therapy, psychological intervention) and cannot be analyzed separately for physical intervention effects.

3. Studies in which both the experimental and control groups only received the same form of AE intervention, making it impossible to perform group comparisons.

4. Studies where the outcome was not reported or where the necessary data could not be extracted.

5. Studies where the intervention measures were not clearly described in terms of training elements (frequency, intensity, duration, cycle), making it impossible to assess the intervention.

Data extraction

Two researchers conducted a thorough review of the literature, adhering to predefined inclusion and exclusion criteria, and independently extracted all relevant data. In cases of disagreement, the involved parties resolved their differences and reached a consensus through discussion, negotiation, or consultation with an additional researcher. The information extracted from the included studies encompassed essential details such as the first author’s name, publication year, country of study, sample size, gender distribution, age range, type of intervention employed, and outcome indicators assessed.

In cases where essential information was unavailable, the corresponding authors of the included studies were contacted via email to request the missing data.

Risk of bias assessment

Two researchers independently assessed the methodological quality of the included studies using the RoB 2.0 tool and the TESTEX scale. In cases of disagreement, the reviewers reached a consensus through discussion, and a third researcher was consulted to arbitrate when necessary.

The RoB 2.0 tool primarily assesses the risk of bias across five domains: randomization process, deviation from the intended intervention, missing outcome data, outcome measurement, and selective reporting. The results are categorized as ‘low risk’, ‘some concerns’, or ‘high risk’. The TESTEX scale scores the quality of study design (5 items) and reporting quality (10 items) in exercise intervention studies, with a total score of 15 points. A higher score indicates better methodological quality and reporting quality. Ultimately, the credibility of the study results will be determined by considering both the RoB 2.0 bias risk assessment and the TESTEX scale score.

Certainty of evidence

This study applied the GRADE (Grading of Recommendations Assessment, Development and Evaluation) framework to evaluate the quality of evidence for the primary outcome measures. The GRADE system evaluates evidence systematically across five domains: risk of bias, inconsistency, indirectness, imprecision, and publication bias, and classifies the quality of evidence into four levels: high, moderate, low, and very low. Two researchers independently conducted the assessments, and any disagreements were resolved through discussion or consultation with a third expert.

Statistical analysis

A Bayesian NMA of multiple trials was conducted using a priori random-effects random effects model with R software. Markov chain Monte Carlo methods were employed to derive the best combined estimates and probabilities for each treatment regimen. Trajectory plots and Brooks-Gelman-Rubin plots were utilized to assess model convergence, with the results presented as posterior mean differences (MD) and their corresponding 95% confidence intervals (CI). The percentage area under the cumulative ranking curve (SUCRA) was calculated to estimate the probability of the optimal intervention. Network and funnel plots were generated using Stata 15.0, while cumulative probability plots were created using the ggplot2 package.

Results

Study selection

A total of 3,169 potentially eligible articles were identified through an initial database search. After removing 1,029 duplicates, 2,140 records remained for screening, of which 2,049 were excluded after title and abstract review. After re-reading the full text and obtaining it again, a further 26 articles were removed, leaving 65 articles for quantitative synthesis. The flow diagram of the study selection process is presented in Fig. 1.

Fig. 1.

PRISMA flow diagram of the study process. PRISMA, Preferred Reporting Items for Systematic review and Meta-analysis

Study characteristics

This NMA included 65 randomized controlled trials published between 2001 and 2024, covering a variety of exercise interventions [26–87, 90]. The studies were conducted in Spain (21), Brazil (13), Turkey (10), the United States (8), Canada (3), Sweden (3), Italy (2), the United Kingdom (1), China (1), South Korea (1), the Netherlands (1), and Romania (1). A total of 52 studies reported FIQ as an outcome, 29 reported VAS, 18 reported TP, and 14 reported 6MWT. A total of ten studies included both male and female participants, while the remaining fifty-five studies focused exclusively on female subjects. In total, 3,764 patients diagnosed with FMS were randomly assigned to either the exercise group (n = 2,420) or the non-exercise group (n = 1,344). Among these participants, there were 3,707 females and 57 males, with a mean age ranging from 30 to 60 years.

The experimental group received a single exercise or a combination of multiple forms of exercise, while the control group received routine care or active treatment. This NMA compared eleven types of exercise interventions: AE, AE combined with strength and stretching, AE combined with stretching, exergaming, resistance training, resistance training combined with stretching, strength training, stretching, stretching combined with balance training, WBV, and balance training. Detailed characteristics of the included studies are presented in Table 1.

Table 1.

Key characteristics and intervention details of studies included in the Meta-Analysis

Study	Year	Country	Sample size	Gender(M/F)	Age(years)	Intervention	Outcomes
Gowans	2001	Canada	Aerobi:16 Control:15	3/28	Aerobi:46.7 Control:49.1	30 min 3times 23weeks	F1 F3 F4
Jones	2002	American	Strength:14 Stretch:14	0/28	Strength:49.2 Stretch:49.2	60 min 2times 12weeks	F1 F3
King	2002	Canada	Aerobic:46 Control:39	0/85	Aerobic:45.2 Control:47.3	40 min 3times 12weeks	F1 F3 F4
Richards	2002	Britain	Aerobic:69 Stretch:67	10/126	Aerobic:48 Stretch:45	60 min 2times 12weeks	F1 F3
Schachter	2003	Canada	Aerobic:51 Control:36	0/87	Aerobic:41.3 Control:42.5	60 min 4times 16weeks	F2
Valim	2003	Brazil	Aerobic:38 Stretch:38	0/76	Aerobic:46.05 Stretch:47.57	45 min 3times 20weeks	F2 F3
Altan	2004	Turkey	Aerobic:24 Control:22	0/46	Aerobic:43.1 Control:43.9	35 min 3times 12weeks	F1 F2 F3
Sencan	2004	Turkey	Aerobic:20 Control:20	0/40	Aerobic:35.4 Control:35.55	40 min 3times 6weeks	F2
Izquierdo	2007	Spain	Aerobic:29 Control:24	0/53	Aerobic:50 Control:46	60 min 3times 16weeks	F3
Rooks	2007	American	Aerobic_Stretch:35 Aerobic_Strength_Stretch:35	0/70	Aerobic_Stretch:48 Aerobic_Strength_Stretch:50	60 min 1times 16weeks	F1 F4
Carus	2007	Spain	Aerobic:17 Control:17	0/34	Aerobic:51 Control:51	60 min 3times 12weeks	F1
Tomas	2007	Spain	Aerobic:17 Control:17	0/34	Aerobic:51 Control:51	60 min 3times 12weeks	F1
Bircan	2008	Turkey	Aerobic:13 Strength:13	0/26	Aerobic:48.3 Strength:46.0	30 min 3times 6weeks	F2 F3 F4
Jones	2008	American	Aerobic:39 Control:41	2/78	Aerobic:49.62 Control:49.78	60 min 3times 24weeks	F2
Izquierdo	2008	Spain	Aerobic:29 Control:24	0/53	Aerobic:50 Control:46	60 min 3times 16weeks	F3
Altan	2009	Turkey	Aerobic:25 Control:24	0/49	Aerobic:48.2 Control:50	60 min 3times 24weeks	F1 F2 F3
Valencia	2009	Spain	Stretch:10 Control:10	0/20	Stretch:54.1 Control:56.4	60 min 2times 12weeks	F1 F3
Carson	2010	American	Aerobic:25 Control:28	0/53	Aerobic:51.4 Control:55.8	120 min 1times 8weeks	F1 F3
Wang	2010	American	Aerobic:33 Control:33	9/57	Aerobic:49.7 Control:50.5	60 min 2times 12weeks	F1 F2 F4
Sanudo	2011	Spain	Aerobic_Strength_Stretch:18 Control:20	0/38	Aerobic_Strength_Stretch:55.48 Control:56.15	60 min 2times 24weeks	F1
Hooten	2012	American	Strength:36 Aerobic:36	7/65	Strength:47.3 Aerobic:45.8	50 min 2times 3weeks	F1 F3
Baptista	2012	Brazil	Aerobic:40 Control:40	0/80	Aerobic:49.5 Control:49.1	45 min 2times 16weeks	F2 F4
Martinez	2012	Spain	Aerobic_Strength_Stretch:14 Control:14	0/28	Aerobic_Strength_Stretch:59.3 Control:58.6	60 min 2times 15weeks	F1
Jones	2012	American	Aerobic:51 Control:47	7/91	Aerobic:53 Control:54.8	90 min 2times 12weeks	F1
Rodriguez	2012	Spain	Aerobic:35 Stretch:35	0/70	Aerobic:55.41 Stretch:55.41	60 min 2times 12weeks	F1 F2
Sanudo	2012	Spain	Aerobic_Strength_Stretch:21 Control:20	0/41	Aerobic_Strength_Stretch:58.6 Control:58.6	60 min 2times 24weeks	F1 F4
Hustings	2013	Netherlands	Aerobic:47 Control:48	0/95	Aerobic:43.9 Control:42.9	60 min 2times 12weeks	F1
Gavi	2014	Brazil	Strength:35 Stretch:31	0/66	Strength:44.34 Stretch:48.65	45 min 2times 16weeks	F1
Genc	2015	Turkey	Aerobic:27 Stretch:27	0/54	Aerobic:35.1 Stretch:36.9	50 min 3times 6weeks	F1
Kibar	2015	Turkey	Stretch_Balance:28 Stretch:29	3/54	Stretch_Balance:48.11 Stretch:48.17	20 min 5times 4weeks	F1
Larsson	2015	Sweden	Resistance:67 Control:63	0/130	Resistance:50.81 Control:52.1	60 min 2times 15weeks	F1 F4
Ericsson	2016	Sweden	Resistance:67 Control:63	0/130	Resistance:50.81 Control:52.1	60 min 2times 15weeks	F1
Bongi	2016	Italy	Aerobic:22 Control:22	0/44	Aerobic:50.36 Control:54.3	60 min 2times 16weeks	F1 F3
Paolucci	2016	Italy	Aerobic:19 Control:20	0/39	Aerobic:50.4 Control:51.3	60 min 2times 5weeks	F1
Mateo	2017	Spain	Exergame:42 Control:41	0/83	Exergame:52.52 Control:52.47	60 min 2times 8weeks	F1
Ekici	2017	Turkey	Aerobic:15 Control:21	0/36	Aerobic:37.13 Control:36.86	60 min 2times 4weeks	F1 F2
Assumpcao	2018	Brazil	Stretch:14 Resistance:16 Control:14	0/44	Stretch:47.9 Resistance:45.7 Control:46.9	40 min 2times 12weeks	F1 F2 F3
Ernberg	2018	Sweden	Resistance:49 Control:43	0/92	Resistance:48.2 Control:48.2	2times 15weeks	F1 F4
Andrade	2019	Brazil	Resistance:25 Control:21	0/46	Resistance:52.0 Control:50.6	60 min 3times 4weeks	F1
AndradeS	2019	Brazil	Aerobic:27 Control:27	0/54	Aerobic:48 Control:47	16weeks	F1 F2
Fernandes	2019	Brazil	Resistance:41 Stretch:41	0/82	Resistance:47.8 Stretch:55.0	45 min 2times 14weeks	F2
Jiao	2019	China	Aerobic:31 Control:31	8/54	Aerobic:48.9 Control:53.5	60 min 2times 12weeks	F1 F3
Silva	2019	Brazil	Resistance:30 Control:30	0/60	Resistance:44.93 Control:49.40	40 min 2times 12weeks	F2 F4
Villafaina	2019	Spain	Exergame:28 Control:27	0/55	Exergame:54.04 Control:52.72	60 min 2times 24weeks	F1
Atan	2020	Turkey	Aerobic_Strength_Stretch:20 Aerobic_Strength_Stretch:20 Control:20	0/60	Aerobic_Strength_Stretch:46.57 Aerobic_Strength_Stretch:47.36 Control:52.7	60 min 5times 6weeks	F1 F2
Carvalho	2020	Brazil	Exergame:16 Control:19	0/35	Exergame:55.64 Control:47.70	60 min 3times 7weeks	F1
Alventosa	2020	Spain	Aerobic:16 Control:16	0/32	Aerobi:53.3 Control:53.3	60 min 2times 8weeks	F1 F2 F4
Sarmento	2020	American	Aerobic:10 Control:10	0/20	Aerobic:42.6 Control:56.1	25 min 7times 10weeks	F2
Valmana	2020	Spain	Aerobic:23 Control:25	0/48	Aerobic:55.01 Control:52.96	90 min 2times 12weeks	F1 F2
Villafaina	2020	Spain	Exergame:22 Control:15	0/37	Exergame:54.27 Control:53.44	60 min 2times 24weeks	F1 F4
Ardila	2021	Spain	Strength:36 Control:33	0/69	Strength:56.06 Control:54.39	13weeks	F1 F2
Garijo	2021	Spain	Aerobic:17 Control:17	0/34	Aerobic:51.81 Control:55.06	50 min 2times 15weeks	F1 F2 F3 F4
Alventosa	2021	Spain	Aerobic:16 Control:16	0/32	Aerobic:53.06 Control:55.13	60 min 2times 8weeks	F1 F2
Park	2021	Korea	Strength:20 Stretch:20	2/38	Strength:52.8 Stretch:50.5	30 min 2times 4weeks	F1
Patru	2021	Romania	Aerobic:34 Control:32	0/66	Aerobic:55.3 Control:56.3	60 min 3times 12weeks	F1
Mansilla	2021	Spain	Aerobic:29 Aerobic:33 Control:31	0/93	Aerobic:52.24 Aerobic:52.24 Control:52.24	45 min 2times 4weeks	F1 F2
Ribeiro	2021	Brazil	Whole-body vibration:17 Control:15	0/32	Whole-body vibration:56 Control:54	3times 6weeks	F1 F2 F4
Fernandez	2022	Spain	Aerobic:40 Control:40	6/74	Aerobic:50.6 Control:50.7	2weeks	F1
Lorena	2022	Brazil	Stretch:19 Control:21	0/40	Stretch:46.47 Control:46.38	90 min 1times 10weeks	F1 F2
Kolak	2022	Turkey	Aerobic_Stretch:13 Resistance_Stretch:13 Stretch:15	0/41	Aerobic_Stretch:48.3 Resistance_Stretch:46 Stretch:45.9	40 min 3times 12weeks	F1 F2
Marcen	2023	Spain	Aerobic:33 Control:31	0/64	Aerobic:46.9 Control:48.7	60 min 3times 16weeks	F1
Mansilla	2023	Spain	Aerobic:31 Control:29	0/60	Aerobic:52.24 Control:52.24	45 min 2 times 4weeks	F1
Yavuz	2023	Turkey	Aerobic:26 Balance:25	0/51	Aerobic:38.12 Balance:41	3times 6 weeks	F2
Fernandes	2024	Brazil	Resistance:41 Stretch:41	0/82	Resistance:47.8 Stretch:55	45 min 2 times 14 weeks	F2
Schulze	2024	Brazil	Stretch:13 Control:12	0/25	Stretch:45.92 Control:45.17	45 min 1times 8weeks	F1 F2 F3

F1:FIQ(Fibromyalgia lmpact Questionnaire)

F2:VAS(a 10 mm visual analog scale)

F3:TP(tender point)

F4:6MWT(6-minute walk)

Quality assessment

The RoB 2.0 results indicated that most studies had a low risk of bias concerning the randomization process and missing outcome data, but raised some concerns regarding blinding. Due to the unique nature of exercise interventions, nearly all studies were unable to implement blinding for participants and researchers, resulting in a high risk across this dimension. Overall, the methodological quality of the included studies was moderate, with some limitations remaining. Detailed assessment results for each study are provided in Supplementary Material 3.

The TESTEX evaluation results showed that the scores ranged from 9 to 13 points, with an average score of 10.37 points, indicating overall quality ranging from moderate to high. Most studies clearly defined inclusion criteria and ensured comparability between baseline groups, and all reported intergroup comparisons, point estimates, and variability measures. Additionally, they provided relatively thorough descriptions of the intensity, volume, and energy expenditure of the exercise interventions. Detailed assessment results for each study are provided in Supplementary Material 4.

Pairwise meta-analysis

In the pairwise meta-analysis, different exercise intervention methods demonstrated varying degrees of effectiveness compared to the control group across multiple outcome measures. Overall results indicated that certain exercise interventions significantly improved FIQ, VAS, and 6MWT, while the improvement in TP was relatively limited. The combined effect results for each outcome measure are presented in Supplementary Material 5.

Additionally, we conducted a publication bias analysis using Egger’s test for the primary outcomes. The results indicated significant publication bias for FIQ (P = 0.000) and VAS (P = 0.002), while no obvious publication bias was observed for TP (P = 0.288) and 6MWT (P = 0.354).

Network meta-analysis

Each circle in the diagram represents an exercise intervention, with the size of the circle corresponding to the number of patients involved. Lines connecting the nodes signify direct comparisons between two interventions. Thicker lines denote studies that were more frequently used for direct comparisons. The global inconsistency test revealed no significant inconsistency in the network (P > 0.05). A network diagram comparing various exercise interventions across four outcome measures is presented in Fig. 2.

Fig. 2.

Network plot of different exercise interventions for key outcomes in patients with fibromyalgia syndrome

The network diagram forms a closely connected loop, with the control group at its center. However, when the FIQ was employed as the outcome index for the NMA, we observed notable differences in the results of direct comparisons, indirect comparisons, and network comparisons between the strength and control groups, as well as between the stretch and control groups (P < 0.05). In contrast, when VAS and TP were employed as outcome indicators, the local inconsistency test yielded non-significant results, indicating that no data exhibited inconsistencies between direct and indirect comparisons (P > 0.05). Furthermore, since the network diagram using 6MWT as an outcome index did not form a closed loop, a local inconsistency test was deemed unnecessary. Local consistency test results for all outcomes are presented in Supplementary Material 6.

At the same time, we used the GRADE framework to grade the certainty of evidence for the primary outcome measures. The results showed that the evidence quality for FIQ was moderate, while that for TP, VAS, and 6MWT was low. This implies that while our conclusions have certain reference significance, they should be interpreted with caution and verified through further high-quality clinical trials. Detailed GRADE assessments for all primary outcomes are provided in Supplementary Material 7.

Efficacy outcomes

The findings indicate that FIQ, AE [MD = −10.23, 95% CI (−11.28, −9.19)], AE and Strength and Stretch [MD = −17.23, 95% CI (−20.93, −13.53)], AE and Stretch [MD = −15.92, 95% CI (−22.48, −9.31)], Exergame [MD = −6.80, 95% CI (−10.81, −2.78)], Resistance [MD = −6.90, 95% CI (−10.18, −3.62)], resistance and stretch [MD = −23.65, 95% CI (−34.43, −12.90)], Strength [MD = −15.87, 95% CI (−18.23,−13.51)], Stretch [MD = −12.44, 95% CI (−13.83, −11.05)], stretch and balance [MD = −25.13, 95% CI (−33.81, −16.44)], and WBV [MD = −23.70, 95% CI (−34.91,−12.47)] were found to be significantly more effective than the control group according to the results of the consistency NMA. Overall, all exercise interventions significantly improved the FIQ, with stretching combined with balance training, whole-body vibration training, and resistance combined with stretching showing the largest effect sizes, suggesting these interventions may represent the most effective exercise options.

For VAS, AE [MD = −1.26, 95% CI (−1.52, −1.01)], AE and Strength and Stretch [MD = −3.61, 95% CI (−4.40, −2.82)], AE and Stretch [MD = −2.71, 95% CI (−3.80, −1.62)], Balance [MD = −1.52, 95% CI (−2.58, −0.46)], Resistance [MD = −1.73, 95% CI (−2.35, −1.11)], resistance and stretch [MD = −3.12, 95% CI (−3.98, −2.26)], Stretch [MD = −1.12, 95% CI (−1.39, − 0.85)], and WBV [MD = − 2.45, 95%CI (–4.11, − 0.80)] were found to be more effective than the control group based on the results of a consistent NMA. In contrast, strength training alone [MD = −0.65, 95% CI (−1.47, 0.17)] did not show a significant difference. Overall, composite interventions (e.g., AE and strength and stretch, resistance and stretch) demonstrated superior effects compared with single-modality interventions.

For TP, AE showed a statistically significant advantage over the control group [MD = −2.28, 95% CI (−3.57, −1.00)]. In contrast, Resistance [MD = −1.00, 95% CI (−6.46, 4.47)], Strength [MD = −2.29, 95% CI (−4.59, 0.01)], and Stretch [MD = −1.83, 95% CI (−3.89, 0.28)] did not show statistically significant differences compared to the control group. Overall, AE was most effective in improving the number of tender points, while single resistance, strength, or stretching training did not show significant advantages.

For 6MWT, AE [MD = 53.18, 95% CI (35.01, 71.29)], Resistance [MD = 46.97, 95% CI (31.50, 62.49)], and WBV [MD = 55.00, 95% CI (26.14, 83.75)] were found to be more effective than the control group. In comparison, AE and Strength and Stretch [MD = 24.49, 95% CI (−18.72, 67.72)], AE and Stretch [MD = 5.62, 95% CI (−49.13, 60.12)], Exergame [MD = 18.08, 95% CI (−48.50, 84.52)], and Strength [MD = 44.30, 95% CI (−20.79, 108.93)] did not reach statistical significance compared to the control group. Overall, AE, resistance training, and WBV training were the most effective in improving patients’ exercise endurance. The meta-analysis effect sizes of various exercise movement types on post-intervention efficacy in patients with FMS are displayed in Tables 2, 3, 4 and 5.

Table 2.

Network Meta-Analysis matrix of Post-Intervention effects on health status (FIQ) across different exercise interventions in patients with FMS

MD95%CI
AE
6.99(3.16,10.84)*	AE_Strength_Stretch
5.68(−1.01,12.32)	−1.33(−7.26,4.6)	AE_Stretch
−10.23(−11.28,−9.19)*	−17.23(−20.93,−13.53)*	−15.92(−22.48,−9.31)*	Control
−3.44(−7.59,0.72)	−10.43(−15.91,−5)*	−9.11(−16.82,−1.4)*	6.8(2.78,10.81)*	Exergame
−3.34(−6.79,0.1)	−10.33(−15.28,−5.39)*	−9(−16.36,−1.64)*	6.9(3.62,10.18)*	0.1(−5.08,5.31)	Resistance
13.42(2.63,24.22)*	6.42(−4.58,17.49)	7.74(−3.51,19.06)	23.65(12.9,34.43)*	16.83(5.37,28.35)*	16.77(5.52,28.03)*	Resistance_Stretch
5.64(3.18,8.09)*	−1.35(−5.73,3.02)	−0.05(−6.94,6.92)	15.87(13.51,18.23)*	9.07(4.42,13.74)*	8.98(4.93,13.01)*	−7.78(−18.68,3.13)	Strength
2.21(0.6,3.82)*	−4.77(−8.7,−0.88)*	−3.47(−10.12,3.22)	12.44(11.05,13.83)*	5.64(1.39,9.9)*	5.54(1.99,9.11)*	−11.21(−21.94,−0.5)*	−3.43(−5.55,−1.3)*	Stretch
14.9(6.18,23.59)*	7.89(−1.54,17.3)	9.22(−1.64,20.12)	25.13(16.44,33.81)*	18.33(8.78,27.9)*	18.23(8.94,27.47)*	1.48(−12.23,15.16)	9.26(0.42,18.08)*	12.7(4.09,21.26)*	Stretch_Balance
13.46(2.21,24.74)*	6.46(−5.32,18.23)	7.8(−5.22,20.84)	23.7(12.47,34.91)*	16.9(4.99,28.8)*	16.81(5.17,28.47)*	0.04(−15.52,15.56)	7.81(−3.63,19.28)	11.25(−0.04,22.55)	−1.45(−15.54,12.71)	WBV

*meansP < 0.05

Table 3.

Network Meta-Analysis matrix of Post-Intervention effects on pain intensity (VAS) across different exercise interventions in patients with FMS

MD95%CI
AE
2.35(1.51,3.18)*	AE_Strength_Stretch
1.45(0.33,2.56)*	−0.89(−2.25,0.44)	AE_Stretch
0.26(−0.77,1.29)	−2.09(−3.41,−0.76)*	−1.19(−2.7,0.33)	Balance
−1.26(−1.52,−1.01)*	−3.61(−4.4,−2.82)*	−2.71(−3.8,−1.62)*	−1.52(−2.58,−0.46)*	Control
0.47(−0.2,1.13)	−1.88(−2.88,−0.88)*	−0.98(−2.2,0.23)	0.21(−1.02,1.43)	1.73(1.11,2.35)*	Resistance
1.86(0.96,2.75)*	−0.49(−1.65,0.68)	0.41(−0.71,1.54)	1.6(0.23,2.96)*	3.12(2.26,3.98)*	1.39(0.37,2.41)*	Resistance_Stretch
−0.62(−1.45,0.22)	−2.96(−4.1,−1.82)*	−2.07(−3.42,−0.69)*	−0.88(−2.2,0.45)	0.65(−0.17,1.47)	−1.08(−2.1,−0.05)*	−2.47(−3.67,−1.28)*	Strength
−0.14(−0.51,0.22)	−2.49(−3.32,−1.65)*	−1.59(−2.65,−0.54)*	−0.4(−1.5,0.69)	1.12(0.85,1.39)*	−0.61(−1.21,−0.01)*	−2(−2.82,−1.18)*	0.47(−0.4,1.33)	Stretch
1.19(−0.49,2.87)	−1.15(−3,0.68)	−0.26(−2.26,1.73)	0.93(−1.04,2.9)	2.45(0.8,4.11)*	0.72(−1.05,2.49)	−0.67(−2.54,1.21)	1.8(−0.05,3.66)	1.33(−0.35,3.02)	WBV

*meansP < 0.05

Table 4.

Network Meta-Analysis matrix of Post-Intervention effects on number of tender points (TP) across different exercise interventions in patients with FMS

MD95%CI
AE
−2.28(−3.57,−1)*	Control
−1.28(−6.91,4.33)	1(−4.47,6.46)	Resistance
0.02(−2.19,2.21)	2.29(−0.01,4.59)	1.29(−4.63,7.24)	Strength
−0.44(−2.55,1.59)	1.83(−0.28,3.89)	0.83(−5.02,6.65)	−0.46(−3.02,2.05)	Stretch

*meansP < 0.05

Table 5.

Network Meta-Analysis matrix of Post-Intervention effects on aerobic capacity (6MWT) across different exercise interventions in patients with FMS

MD95%CI
AE
28.63(−18.26,75.55)	AE_Strength_Stretch
47.59(−9.94,105.4)	18.98(−14.46,52.25)	AE_Stretch
53.18(35.01,71.29)*	24.49(−18.72,67.72)	5.62(−49.13,60.21)	Control
35.06(−33.76,104.07)	6.52(−72.41,85.87)	−12.47(−98.53,73.5)	−18.08(−84.52,48.5)	Exergame
6.16(−17.65,30.09)	−22.38(−68.39,23.54)	−41.41(−98.32,15.22)	−46.97(−62.49,−31.5)*	−28.93(−97.31,39.41)	Resistance
8.77(−53.34,71.3)	−19.73(−97.22,58.06)	−38.68(−123.49,45.64)	−44.3(−108.93,20.79)	−26.3(−119.42,66.74)	2.61(−63.83,69.55)	Strength
−1.83(−35.8,32.24)	−30.41(−82.5,21.41)	−49.32(−111.21,12.36)	−55(−83.75,−26.14)*	−36.91(−109.51,35.28)	−8.03(−40.73,24.53)	−10.68(−82.07,60.09)	WBV

*meansP < 0.05

Rank probability

The rankings based on SUCRA values for FIQ are as follows: stretch and balance (SUCRA = 90.43%) > resistance and stretch (SUCRA = 85.86%) > WBV (SUCRA = 85.35%) > AE and Strength and Stretch (SUCRA = 67.05%) > Strength (SUCRA = 59.54%) > AE and Stretch (SUCRA = 58.72%) > Stretch (SUCRA = 42.01%) > AE (SUCRA = 29.90%) > Resistance (SUCRA = 15.58%) > Exergame (SUCRA = 15.54%) > Control (SUCRA = 0.01%).

For VAS, the rankings are: AE and Strength and Stretch (SUCRA = 95.42%) > resistance and stretch (SUCRA = 85.63%) > AE and Stretch (SUCRA = 75.66%) > WBV (SUCRA = 68.78%) > Resistance (SUCRA = 53.07%) > Balance (SUCRA = 44.17%) > AE (SUCRA = 35.46%) > Stretch (SUCRA = 26.59%) > Strength (SUCRA = 14.49%) > Control (SUCRA = 0.73%).

For TP, the rankings are: AE (SUCRA = 71.06%) > Strength (SUCRA = 70.07%) > Stretch (SUCRA = 56.13%) > Resistance (SUCRA = 42.10%) > Control (SUCRA = 10.64%).For 6MWT, the rankings are: WBV (SUCRA = 78.59%) > AE (SUCRA = 77.58%) > Resistance (SUCRA = 67.18%) > Strength (SUCRA = 62.41%) > AE and Strength and Stretch (SUCRA = 43.07%) > Exergame (SUCRA = 36.68%) > AE and Stretch (SUCRA = 21.05%) > Control (SUCRA = 13.43%). The comparative efficacy of various interventions in enhancing pain relief and exercise capacity among patients with FMS is illustrated in Fig. 3.

Fig. 3.

Cumulative ranking probability plots of exercise interventions for improving outcomes in patients with fibromyalgia syndrome

Sensitivity analysis

To assess the robustness of the results, we conducted sensitivity analyses and heterogeneity tests for each outcome measure. Overall, the sensitivity analysis results showed that the effect sizes varied only slightly across different models and methods, and did not substantially alter the relative ranking of exercise interventions, suggesting that the findings of this study are highly robust.

Regarding heterogeneity, the I² values varied significantly across different outcomes and comparison pairs. The FIQ results indicated that some comparisons exhibited high heterogeneity, such as Control vs. AE (pairwise 36.6%, network 37.8%) and Stretch vs. Control (pairwise 20.4%, network 87.3%), suggesting substantial differences in effect sizes between exercise modalities. In the VAS results, Control vs. AE (pairwise 67.6%, network 65.7%) and Stretch vs. Control (pairwise 56.8%, network 75.0%) also exhibited moderate to high heterogeneity. In the TP results, heterogeneity was significant for Control vs. AE (pairwise 86.7%, network 85.9%) and Stretch vs. Strength (network 91.2%). Some comparisons (e.g., Strength vs. Control, Resistance vs. Control) demonstrate low heterogeneity (I² = 0%). In the 6MWT metric, most comparisons showed low heterogeneity (e.g., Resistance vs. Control, I² = 0%), suggesting relatively stable conclusions. Detailed results of heterogeneity tests are shown in Supplementary Material 8.

Overall, although some outcome metrics and comparisons exhibited high heterogeneity, sensitivity analysis results indicated that these differences did not affect the overall conclusions or intervention rankings, validating the robustness of this NMA.

Publication bias

Based on the symmetry criterion and visual inspection, the majority of studies were distributed symmetrically about the center line in the funnel plot, suggesting the robustness of the research findings, with no significant publication bias observed (Fig. 4).

Fig. 4.

Comparison-adjusted funnel plot illustrating potential publication bias across exercise intervention studies

Discussion

Main findings

Exercise therapy has become one of the most commonly used and popular rehabilitation methods for patients with FMS [90]. To our knowledge, this is the first and largest NMA to date, incorporating 65 randomized controlled trials involving 3,764 participants. It systematically compared the effects of 11 different exercise interventions on FMS, covering functional status (FIQ), pain intensity (VAS), tender point count (TP), and aerobic capacity (6MWT). This study used an NMA to simultaneously compare the effects of single and combined exercise interventions and rank multiple intervention measures, thereby providing a more comprehensive reference for clinical practice. Due to the limited number of studies for some interventions and the heterogeneity of the included populations, caution is warranted when interpreting these results. The results showed that in terms of the FIQ metric, stretching combined with balance training was the most effective intervention, followed by resistance combined with stretching, with WBV ranking third. In terms of the VAS metric, combined aerobic, strength, and stretching exercises were the most effective, followed by resistance combined with stretching and aerobic combined with stretching. In terms of the TP metric, AE performed best, with no significant differences observed among other exercise forms. On the 6MWT metric, WBV ranked first, followed by AE and resistance exercise in second and third place, respectively. Overall, composite training and WBV demonstrated a strong advantage in improving multiple core outcomes.

The risk of bias varied across the included studies, with the primary risk stemming from limitations in the implementation of blinding. Sensitivity analyses further supported the robustness of the study’s conclusions.

Effects of different exercise types on health status

Current research indicates that FMS patients often experience a significant decline in health status [91]. Previous meta-analyses suggest that aerobic, resistance, and stretching exercises can significantly improve FMS symptoms (P < 0.05), but no statistically significant differences were observed between different types of exercise [15]. Additionally, neither aerobic nor resistance exercise alone can fully counteract all the negative effects of FMS [92]. Another meta-analysis demonstrated that a combination of exercises significantly improves FMS-related sleep, whereas individual aerobic or strength exercises do not have a positive effect [93]. However, these studies were limited in the types of exercise examined and focused mainly on outcomes such as quality of life, sleep, and mood, failing to cover all types of exercise interventions. In contrast, this study incorporated a more comprehensive range of exercise intervention forms and introduced additional metrics such as TP, VAS, and 6MWT, highlighting the innovative nature of this research. The results showed that all included exercise forms significantly improved the FIQ of FMS patients; however, the ranking results suggested that stretching combined with balance training may be the most effective approach for enhancing the health status of FMS patients, consistent with the positive effects reported in previous studies [55]. Related studies have found that stretching combined with balance training is more advantageous than stretching alone in improving health status. A meta-analysis indicated that stretching exercises can significantly improve FIQ scores [17, 94], while the effects of balance training are considered comparable to those of AE [87]. The underlying mechanisms may include the following: stretching can increase the joint range of motion in FMS patients [95, 96], and by activating the descending pain-inhibitory system, promoting the release of endogenous opioids, and regulating descending pain pathways, it can suppress pain and improve health status [97–99]. Therefore, the combination of stretching and balance training holds potential advantages in FMS intervention.

In addition to widespread pain, loss of balance is also a common complaint in FMS patients, with 45%−68% of FMS cases experiencing balance issues [87]. Current evidence suggests that the severity of FMS symptoms is closely associated with balance disorders, making the incorporation of balance training into rehabilitation programs clinically valuable [100]. Balance training not only improves postural control and stability, reducing the risk of falls, but may also further improve patients’ quality of life by improving motor function and self-efficacy [55]. Future studies may further explore the combined effects of balance training and other forms of exercise to provide more comprehensive treatment strategies for patients with FMS.

Therefore, it is reasonable to assume that combining stretching exercises with balance training may help improve the health status of patients with FMS. However, the evidence supporting this view is currently limited to a single study reporting such benefits, and the underlying physiological mechanisms remain unclear. More high-quality research is needed to confirm these findings and support their clinical application.

Effects of different exercise types on pain intensity and number of painful areas

Our research found that combined aerobic, strength, and stretching exercises may be the most effective way to relieve pain, consistent with the conclusions of Altan et al. [70]. Studies have shown that a combination of AE and yoga can significantly improve pain levels in patients with FMS, with better results than AE alone [101]. Additionally, Kolak et al. noted that a combination of AE with stretching or resistance training was more effective than stretching alone in reducing VAS scores [102]. In contrast, Park et al. found no significant changes in VAS scores in the strength training group, but a noticeable improvement in the stretching group [103]. An increasing number of studies suggest that combined interventions involving multiple forms of exercise are more effective than single-modality exercise in alleviating FMS-related pain [48]. The mechanisms of action may include: AE or strength training can enhance muscle oxygenation, reduce peripheral and central sensitivity, thereby improving clinical pain [104]; combining AE with strength training may alleviate pain by releasing neurotransmitters through the central nervous system and promoting local muscle adaptation; stretching exercises can reduce tender point sensitivity, increase joint range of motion, and alleviate movement fear, thereby further alleviating pain [62, 105].

This study found that AE was most effective in reducing the number of tender points in patients with FMS. Although resistance training and stretching exercises also showed improvement trends, they did not reach statistically significant differences. Previous studies have also suggested that aerobic mind-body exercises such as Ba Duan Jin and Tai Chi are highly effective in improving FMS symptoms [58, 67]. Some studies have reported that yoga can reduce the number of TP [43]. It is worth noting that traditional Chinese medicine exercise therapies (such as Ba Duan Jin and Tai Chi) can effectively reduce the perception of pain in multiple areas by harmonizing the mind and body and improving central nervous system function [20]. Compared with conventional aerobic exercises (e.g., running, swimming, or cycling), traditional mind-body exercises emphasize mind-body integration and deep relaxation, potentially conferring unique benefits in alleviating pain in patients with FMS [106].

Effects of different exercise types on aerobic capacity

Patients with FMS often experience insufficient physical activity, which can exacerbate symptoms and lead to a decline in physical fitness levels [81]. The results of this study suggest that WBV training may be the most effective form of exercise for improving aerobic capacity in FMS patients. WBV has been recommended in recent years as a time-saving, feasible, well-tolerated, and low-risk intervention method [107]. It works by mechanically stimulating muscles and the nervous system through vibration, thereby improving muscle contraction efficiency and blood circulation. This leads to enhanced aerobic capacity [108], reduced visceral fat, optimized body composition [109], increased muscle strength [110], improved balance [111, 112], and enhanced flexibility [113]. These physiological and functional improvements may collectively contribute to enhanced exercise endurance in FMS patients.

Limitations

This study still has some limitations. First, due to the difficulty of implementing blinding for participants and researchers in exercise intervention studies, the majority of included studies have a risk of bias in this dimension. Second, the limited number of studies available for certain outcome measures may lead to small-sample effects and potential publication bias, thereby affecting the robustness of the results. In this study, the results of Egger’s test suggest a possible publication bias for FIQ and VAS, while TP and 6MWT showed no obvious bias. Small-sample studies are more likely to report positive results, potentially leading to an overestimation of intervention effects. As SUCRA rankings are based on effect estimates from network meta-analyses, the presence of publication bias or small-sample effects may affect the stability of rankings for different exercise interventions. For example, in studies with limited sample sizes, fluctuations in effect values may lead to certain exercise forms (such as combined training) being overestimated in SUCRA rankings. Therefore, while SUCRA provides a relative ranking of different interventions, its results should be interpreted cautiously in light of the potential influence of publication bias and small-sample effects. Third, according to the GRADE evidence grading system, this study found that the evidence quality was moderate for FIQ. In contrast, the evidence quality for VAS, TP, and 6MWT was low, indicating that there remains uncertainty regarding the evidence for pain intensity, tender point count, and exercise endurance. Fourth, the NMA itself has methodological limitations. The reliability of NMA depends on the transitivity assumption, which states that differences in participant characteristics, interventions, and outcome measures across studies should be as comparable as possible. However, the included studies exhibited heterogeneity in intervention frequency, intensity, and duration, which may affect the robustness of the results. Although consistency tests suggested overall consistency between direct and indirect evidence, some comparisons relied primarily on indirect evidence and should be interpreted with caution. Additionally, most data for certain outcomes originate from small-sample studies, which increases indirectness and uncertainty. In summary, the conclusions of this study should be applied cautiously in clinical settings. Further large-sample, high-quality, long-term randomized controlled trials are needed to confirm these findings.

Conclusion

This systematic review and NMA indicate that stretching combined with balance training is most effective in improving the overall health status of FMS patients; AE and strength training combined with stretching are significantly effective in alleviating pain intensity; AE performs best in reducing the number of pain points; and WBV training has advantages in enhancing aerobic exercise capacity. Overall, compared to single exercise modalities, combined training may offer unique advantages in improving multidimensional outcomes for patients with FMS.

Based on the current evidence, stretching and balance training can be prioritized as adjunctive therapies for patients with FMS. However, the limited number of randomized controlled trials on combined interventions, coupled with significant variations in intervention forms, duration, and intensity, restricts the comparability and generalizability of results. Future studies should focus on conducting large-scale, high-quality, long-term randomized controlled trials to establish the comparative efficacy of various exercise modalities and guide evidence-based clinical exercise prescriptions.

Abbreviations

FMS

Fibromyalgia Syndrome

WBV

Whole-Body Vibration

AE

Aerobic Exercise

NMA

Network Meta-Analysis

TP

Tender Points

Authors’ contributions

(1) Wen Yuan: Literature retrieval, data extraction, data analysis, quality evaluation, article writing; (2) Pengfei Wan: Literature retrieval, data extraction, data analysis; (3) Hui Wang: data extraction, quality evaluation; (4) MingZhu Suo: data extraction, quality evaluation; (5) Pingping Xie: Writing – review & editing.

Funding

Not applicable.

Data availability

All data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.