HOTFy: randomised clinical trial for hyperbaric oxygen therapy in fibromyalgia

Abstract

Background

Fibromyalgia is a polysymptomatic central sensitisation disorder characterised by widespread pain, fatigue, sleep disturbances and neuropsychiatric features. Hyperbaric oxygen therapy modulates neuroinflammation, mitochondrial function and neuroplasticity, thereby yielding analgesic and functional benefits.

Objective

Evaluate the efficacy and optimal timing of hyperbaric oxygen therapy as an adjunct to standard care for fibromyalgia.

Design, setting and participants

This single-centre, randomised, cross-over group, assessor-blinded clinical trial was conducted in the Department of Rheumatology at the University Hospital of the Federal University of Juiz de Fora, Juiz de Fora, Brazil, and adhered to Consolidated Standards of Reporting Trials (CONSORT) guidelines. Women (18–70 years) with a diagnosis of fibromyalgia for ≥2 years were randomised 1:1 to early hyperbaric oxygen therapy plus standard care or standard care alone (delayed group). Intention-to-treat (ITT) analysis was conducted with all 56 participants (mean age: 51.0±9.8 years; mean body mass index: 30.5±5.1 kg/m²).

Interventions

Standardised care (education, exercise and pharmacotherapy) plus hyperbaric oxygen therapy was delivered at 2.3 atmospheres absolute for 90 min, five times per week, over 8 weeks (total 32–40 sessions). The early group received hyperbaric oxygen therapy during weeks 0–8, while the delayed group received it during weeks 8–16, following the same protocol.

Outcomes

Primary endpoints included the Fibromyalgia Impact Questionnaire-Brazilian Portuguese (FIQR-Br), the pain visual analogue scale (VAS) and the Symptoms Assessment Scale-40 (EAS-40) for psychopathology. Secondary endpoints included the 12-Item Short-Form Health Survey (SF-12) physical and mental components and adverse effects. Assessments were conducted at baseline, 8 weeks and 16 weeks, and analysed using a mixed-design 2×3 analysis of variance (group: early vs delayed; time: baseline, 8 weeks and 16 weeks) with Greenhouse-Geisser corrections as needed, followed by Bonferroni post hoc tests. Missing data were assessed using Little’s missing completely at random (MCAR), and considering the ITT analysis, the means imputed for missing data were estimated through expectation maximisation. Effect sizes were reported as partial η² and Cohen’s d with α=0.05.

Results

44 participants completed the study, and the overall withdrawal rate was 21.4% with no baseline between-group differences. Significant time effects were observed for all primary outcomes and the SF-12 outcome (p<0.001; η²=0.23–0.60). Group×time interactions were significant for FIQR-Br, VAS, EAS-40 and SF-12 physical and mental (p≤0.02; interaction η² up to 0.23), indicating improvements during active hyperbaric oxygen therapy exposure. Compared with standard care alone over 8 weeks, combined treatment achieved greater gains: FIQR-Br, −31.1% vs −14.4%; VAS, −54.0% vs −33.5%; EAS-40, −28.4% vs −3.7%; SF-12 physical, +39.1% vs +14.8%; SF-12 mental, +57.4% vs +31.9%. Large within-group effect sizes were observed (eg, VAS d=2.5–2.7; FIQR-Br d=1.4–1.7). Efficacy was equivalent regardless of time started, and the benefits converged by the end of each hyperbaric oxygen therapy phase. After stopping hyperbaric oxygen therapy, the FIQR-Br and SF-12 mental component scores regressed towards standard care levels, whereas residual improvements persisted for up to 8 weeks in VAS, EAS-40 and SF-12 physical component scores. Adverse events were infrequent; one case of otalgia required extended management. Withdrawals were primarily due to non-compliance or intolerance to chamber confinement. No serious or unexpected safety concerns were reported.

Conclusions

Hyperbaric oxygen therapy, delivered under a standardised protocol, is an effective and well-tolerated adjunct to multimodal fibromyalgia care. Timing can be individualised: early initiation for rapid relief or stepped introduction after optimised usual care, with comparable overall efficacy. The durability of the benefit appears to be exposure dependent, and maintenance or booster schedules merit further evaluation.

Trial registration number

RBR-6prps8g.

STRENGTHS AND LIMITATIONS OF THIS STUDY.

• Assessment-blinded randomised clinical trial design.

• Intention-to-treat analysis.

• Sample size was previously calculated.

• The protocol was previously published, minimising publication bias.

• Dropout rate of 21.4%, attributable to the lengthy treatment (40 sessions).

Introduction

Among the various causes of chronic pain, fibromyalgia (FM) is a polysymptomatic syndrome characterised by widespread pain, fatigue, sleep disturbance, autonomic dysfunction, cognitive impairment, hypersensitivity to stimuli, somatic symptoms and psychiatric disorders.¹ The prevalence of FM in Brazil is estimated to be approximately 2.5% of the population; however, global rates vary, with a higher prevalence in individuals with specific comorbidities.^{2 3} Notably, FM affects between 0.2% and 13% of the world population and leads to work disabilities in nearly half of those diagnosed with FM.^{4 5}

The pathophysiology of FM is complex and remains unclear. Current evidence suggests alterations in both functional and chemical connectivity within the pain-processing pathways of the central nervous system, positioning FM as a central pain-processing disorder or central sensitisation syndrome.⁶ Additionally, emotional factors can serve as triggers or exacerbators of symptoms. However, the precise relationship between neuroinflammation, stress, anxiety, central sensitisation and the dysregulation of innate and adaptive immune responses remains unknown.7,9

Hyperbaric oxygen therapy (HBOT) has emerged as a promising treatment modality that is capable of modulating key cellular mechanisms implicated in neurological and psychiatric disorders.¹⁰ HBOT involves exposing patients to nearly 100% oxygen within a pressurised chamber, typically at two to three times the sea-level atmospheric pressure. Preclinical studies have shown that HBOT has significant analgesic effects in models of nociceptive, inflammatory and neuropathic pain. Furthermore, HBOT has been associated with clinical benefits across a range of conditions, including complex regional pain syndrome—where reductions in pain intensity and improvements in functional outcomes have been reported—as well as stroke sequelae, traumatic brain injury, spinal cord injury and autism.11,15

Mechanistically, HBOT modulates inflammatory responses following tissue injury, reducing nociceptive activity by approximately 80–95% for up to 90 min postexposure.¹⁶ Interestingly, the antinociceptive effects observed in preclinical studies appear to be independent of oxidative stress.¹⁶ Randomised clinical trials have demonstrated that HBOT significantly reduces pain intensity, decreases the number of tender points and improves functional and neuropsychiatric outcomes, thereby enhancing the overall quality of life in patients with FM.¹⁷ These neuromodulatory effects are primarily attributed to improvements in mitochondrial function, neurogenesis, synaptogenesis and anti-inflammatory processes. HBOT upregulates the pathways involved in cellular repair, including mitochondrial biogenesis, ATP production and neuroplasticity.¹⁰

Despite these promising findings, critical knowledge gaps remain regarding the standardisation of HBOT protocols for FM treatment. Variability in treatment parameters, including pressure settings, session frequency and optimal timing of therapy initiation, has led to inconsistencies in reported outcomes.¹⁸ This study, Hyperbaric Oxygen Therapy for Fibromyalgia (HOTFy), is a randomised clinical trial evaluating the efficacy of HBOT in patients with FM and determining the optimal way to integrate this therapy into multimodal treatment approaches.

Materials and methods

Study design and settings

This randomised, controlled, superiority, cross-over group study, conducted between November 2022 and May 2024, was performed in accordance with the Consolidated Standards of Reporting Trials (CONSORT) guidelines.¹⁹ A primary protocol (online supplemental file 1)¹⁸ was used to perform the clinical trial, comparing HBOT with standardised care at a single research centre in the Department of Rheumatology at the University Hospital of the Federal University of Juiz de Fora, Juiz de Fora, Brazil. There were no major deviations from the protocol. We adopted an early versus delayed treatment comparison because currently available HBOT sham protocols are not physiologically inert. Even modest increases in chamber pressure with normoxic air can increase tissue oxygenation and induce biological effects, limiting their validity as true placebos.¹⁷ Therefore, a delayed treatment design was chosen to better control for non-specific effects while maintaining methodological rigour.

Recruitment

All participants were referred to the rheumatology department after primary care and were enrolled in the study according to the inclusion and exclusion criteria identified during rheumatology appointments (figure 1). After being considered eligible, patients were verbally informed about the study and its objectives. Those who consented to participate were provided with a written consent form (online supplemental file 2). They were then asked to sign the form, and a registration number was assigned to each participant.

Figure 1. Enrolment flow chart. EAS-40, 40-Item Symptoms Assessment Scale; FIQR-Br, Fibromyalgia Impact Questionnaire-Brazilian Portuguese; HOT, hyperbaric oxygen therapy; SF-12, 12-Item Short-Form Health Survey; VAS, visual analogue scale.

Inclusion and exclusion criteria

Eligible participants were women aged 18–70 years. Inclusion required a diagnosis of FM made at least 2 years before enrolment, established according to conventional criteria, namely bilateral, generalised pain occurring above and below the waist for at least 3 months in the absence of another somatic disorder explaining the symptoms and/or the presence of ≥11 of the 18 tender points, while also taking into account the 2016 American College of Rheumatology (ACR) revisions to the classificatory criteria. Participants were recruited between November 2022 and May 2024.20,22 The exclusion criteria were as follows: HOT contraindications (pregnancy; use of bleomycin, cisplatin, disulfiram and doxorubicin; middle ear surgery; untreated pneumothorax or pneumomediastinum; and claustrophobia)²³; associated autoimmune rheumatologic disease (rheumatoid arthritis, systemic lupus erythematosus, scleroderma and others); and inability to sign the consent form.

Withdrawal from the study

Participants who withdrew from the study before completing 32 sessions and/or had interrupted treatment for more than five consecutive sessions were permitted to continue their standard care without interruption at the rheumatology outpatient clinic, provided this did not prejudice the usual recommended treatment. This process was outlined and detailed to the patients in full before obtaining their consent.

Randomisation

All patients who provided written consent to participate and met the eligibility criteria, as assessed by a rheumatologist, were randomised in a 1:1 ratio between combined treatment (HBOT+standard treatment—early group) and standard group (delayed group). Each participant was given a sequential number. The randomisation sequence was generated using computer software (randomizer.org). Participants were allocated to the intervention with equal probability and randomised into a single, fixed, random block. The allocation list was generated by an independent staff member from the university hospital’s Musculoskeletal Unit who had no role in the study design, conduct, data collection or analysis and was responsible solely for the randomisation draw. This individual prepared a sequence of opaque envelopes, identified by the participant’s registration number, containing only one intervention to be performed according to the computer-generated sequence. A request from the researcher triggered allocation; the independent individual then opened the sealed envelope and disclosed the assignment to the study team.

Blinding

Clinical data—including baseline characteristics and follow-up outcome measures (pain visual analogue scale (VAS),²⁴ functional evaluation using the Fibromyalgia Impact Questionnaire-Brazilian Portuguese (FIQR-Br),²⁵ the 40-Item Symptoms Assessment Scale (EAS-40) for psychopathology,²⁶ and the 12-Item Short-Form Health Survey (SF-12) physical and mental components, as well as the quality of life assessment²⁷)—were assessed by a rheumatologist who was blinded to treatment allocation. Given the nature of the intervention, evaluators, data collectors and care providers were blinded to the study hypotheses.

Intervention

The participants underwent daily HBOT sessions five times per week, for a total of 32–40 sessions. Each treatment session consisted of 90 min of oxygen therapy with an inspired fraction of medicinal oxygen (FiO₂) (purity >99%)²⁸ at 2.3 absolute atmospheres (ATA) of pressure in monoplace hyperbaric chambers registered according to ECO BAR 800 (serial numbers: E4-034 and E4-033, manufactured in April 2015 and November 2014, respectively). Each chamber underwent annual inspection and maintenance in accordance with the manufacturer’s recommendations.¹⁸

A rheumatologist provided the standard treatment, which consisted of simultaneous patient education, physical activity, physiotherapy and pharmacological treatment (including antidepressants, anticonvulsants, analgesics and myorelaxants).¹ Both groups completed a baseline symptom questionnaire and a VAS, FIQR-Br, EAS-40 and SF-12 questionnaires immediately after randomisation. Subsequently, they received the same HBOT protocol at different time points. The early group underwent 40 HBOT sessions over an 8-week period and was evaluated by the same rheumatologist using baseline questionnaires. After crossover, the delayed group underwent the identical 8-week HBOT protocol during weeks 8–16. At week 16, participants were evaluated by the same rheumatologist with the same questionnaires (figure 1).

Follow-up

The enrolled patients were evaluated at three predefined time points: baseline (prior to randomisation), week 8 (end of the first 8-week study period) and week 16 (end of the second study period) (figure 1). At each time point, all assessments were conducted during a single study visit and included a clinical evaluation by a blinded rheumatologist, as well as patient-reported outcomes comprising pain intensity (VAS), functional status (FIQR-Br), psychopathological symptoms (EAS-40) and health-related quality of life (SF-12).24,27

Risks

The risks of the intervention were those reported in the literature for HBOT.29,34 Listed in decreasing order of frequency, these comprised: hypoglycaemia in patients with diabetes, barotrauma, central nervous system oxygen toxicity (convulsive seizures), pulmonary toxicity associated with prolonged oxygen exposure, temporary changes in ocular refraction and acceleration of lens opacification.

Harms assessment

Harms were monitored through active/targeted surveillance using prespecified definitions and measurements, including capillary blood glucose levels <60 mg/dL, otalgia during the first 12 min of the session, auras and/or tonic-clonic seizures, diffuse pulmonary infiltrates on chest radiography and improvement or worsening of visual acuity. Outcomes were assessed at prespecified time points: after the first HBOT session and after the final session. Assessments were conducted by qualified hyperbaric medicine professionals (medical and nursing staff) who were not blinded to treatment allocation; risk and complication data were recorded on a prespecified form.

In addition, passive surveillance captured any unsolicited adverse effects reported by participants or observed during routine care; the collection procedures, assessment time points and overall recording period were aligned with the study schedule.

Outcomes

The primary outcomes were changes over time in the VAS, FIQR-Br and EAS-40 scores. All assessments were conducted by the same blinded rheumatologist who performed the initial evaluation at different time points.

Pain was assessed using a VAS ranging from 0 to 10, anchored at 0=no pain and 10=worst pain imaginable.²⁴ The primary analysis metric was change from baseline, evaluated at weeks 8 and 16 in relation to the HBOT intervention. The VAS was patient reported; completeness and outcome assessments were conducted by the same rheumatologist, who was blinded to treatment allocation and had administered the baseline questionnaire.

The FIQR-Br was administered to obtain the total score (0–100) and subdomain scores for function, overall impact and symptoms, with higher values indicating greater disease burden.²⁵ The primary analysis metric was change from baseline for both total and subdomain scores; end-of-period values were analysed as complementary outcomes. At the group level, results were summarised as the mean (SD) according to the distributional assumptions, and between-group differences were estimated for change scores. Assessments were conducted at baseline and postintervention at the study milestones (weeks 8 and 16), aligned with each group’s schedule. The questionnaire was self-completed by participants, and trained study staff verified completeness without influencing responses.

The EAS-40 was used to derive a total score, with higher values indicating greater symptom burden.²⁶ The primary metric was the change from baseline in the total score; endpoint values at each assessment were reported as supportive outcomes. Group summaries were calculated as mean (SD), conditional on distributional assumptions, and between-group contrasts were based on change scores. Assessments were undertaken at baseline and at the prespecified postintervention milestones (weeks 8 and 16), aligned with the early/late group schedules. The instrument was self-completed by participants, and trained study staff verified completeness without influencing responses.

Secondary outcomes comprised health-related quality of life and adverse effects. Quality of life was assessed using the SF-12, yielding physical and mental component scores (PCS/MCS) (higher values indicate better status); tender points were enumerated on clinical examination; adverse effects were defined and recorded according to the prespecified harms-monitoring procedures. The primary participant-level metric was change from baseline for SF-12 PCS/MCS and for the tender point count, with adverse effects analysed as occurrence (yes/no) and counts per participant. Group-level summaries were reported as mean (SD) for continuous outcomes, according to distribution, and as proportions (and counts) for adverse effects. Outcomes were assessed at baseline, week 8 and week 16, aligned with the early/late group schedules. The SF-12 was self-completed by participants, with completeness checks conducted by trained staff. Tender point examinations were performed by a rheumatologist blinded to allocation, and adverse effects were documented by clinical staff according to predefined surveillance procedures.

Data collection and management

The patient demographic and medical history data and questionnaires were collected through study forms (onlinesupplemental files 35) and stored on the REDCap platform, which served as the study repository. The study was conducted from November 2022 to July 2025. The principal investigator filed and securely stored the original study forms. Patient files were maintained in numerical order in a secure, access-controlled location, with a retention period of 5 years after study completion (until July 2030). The principal investigator supervised the completion of the electronic spreadsheet and ensured its reliability and accuracy. Incorrect or missing data were reviewed by the principal investigator and corrected as necessary. During the study, a committee consisting of the main researcher, coresearcher, cosupervisor and principal supervisor monitored the data. One of the researchers verified the completion of the patient questionnaires as a strategy to mitigate data loss. A loss to follow-up (‘dropout’) rate of up to 20% of the sample size was estimated¹⁸; data from participants lost to follow-up were securely stored and included in the analyses under an intention-to-treat framework.

Confidentiality

On enrolment, each participant was issued a unique study ID; personal identifiers were stored separately from research data. All study data were housed in the REDCap repository, with access restricted to the principal investigator (JdMN) and authorised data management staff, and audit trails were enabled. For statistical analysis, datasets were deidentified to ensure sufficient anonymisation; no direct identifiers were retained, and no more than two indirect identifiers were included where necessary. Findings were disseminated only in aggregate form; no potentially identifying information was presented in publications or conference materials. All data used in this research were exclusively obtained and not shared with external institutions without prior ethics approval and, where applicable, participant consent.

Data access and dissemination

Study data were collected solely for academic, non-commercial purposes. Patients were informed in the consent materials that they could request access to their personal data; such requests were honoured via the study’s data management procedures. Access for the research team was role based. Investigators received deidentified summary datasets for analysis, with direct identifiers removed and indirect identifiers minimised to protect confidentiality. Any data shared within the project team for operational purposes were masked to prevent reidentification. External sharing of data did not occur without prior ethics approval and, where applicable, participant consent.

Patient and public involvement

Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

Sample size

The G*Power V.3.1³⁵ software was used to calculate the sample size. The study by Efrati et al¹⁷ guided the calculation when considering the hypothesis for a clinical improvement of the somatic and neuropsychiatric symptoms of FM, associated with a moderate effect size (f=0.25). The correlation between measurements (r=0.30), correction for non-sphericity (ε=1.0), 80% power and 95% CI were also included in the calculation. Ultimately, a sample size of 46 patients was determined. Considering a 20% loss to follow-up (‘dropout’) rate, 56 patients were considered an ideal recruitment size, with 28 patients in each group.¹⁸

Statistical analysis

Descriptive statistics were calculated using means and SDs for numerical variables and frequencies for categorical variables. Missing data were assessed using Little’s missing completely at random (MCAR) test. Following the intention-to-treat principle, missing values were imputed using an expectation-maximisation algorithm. Normality of the data distribution was assessed using the Kolmogorov-Smirnov test. The assumption of homogeneity of variance was verified using Levene’s test.

To investigate potential baseline differences between the early and delayed HBOT groups regarding possible confounding variables, the independent samples Student’s t-test was used for numerical variables. The χ² test or Fisher’s exact test, when appropriate, was used for categorical variables.

A mixed-design 2×3 analysis of variance (ANOVA) was conducted to examine the interaction between the treatment group (between-subjects factor: early vs delayed HBOT) and follow-up time (within-subjects factor: baseline, 8 weeks and 16 weeks), using the following dependent variables: FIQR-Br, VAS, EAS-40 and SF-12. The assumption of sphericity was tested using Mauchly’s test; when violated, df were adjusted using the Greenhouse-Geisser correction. Significant main effects and interactions were explored through multiple comparisons using Bonferroni’s post hoc procedure. ANOVA effect sizes were calculated using partial eta squared (η²), interpreted as small (<0.06), moderate (0.06–0.14) and large (≥0.15), according to Cohen.³³ For paired comparisons, Cohen’s d was used to quantify effect sizes, categorised as small (<0.50), moderate (0.50–0.79) and large (≥0.80).³⁶

All statistical analyses were performed using the SPSS software (V.21.0; IBM SPSS). Statistical significance was set at p<0.05.

Results

Missing data analysis

Of the 56 patients included in this study, 44 completed it. The dropout rate was 21.4% (n=12). Five participants in the early HBOT group (17.2%) and seven participants in the delayed HBOT group (25.9%) did not complete the study. Losses in the early HBOT group were owing to the inability to remain in the hyperbaric chamber for the entire protocol duration (four participants). Additionally, there was one case of otalgia, and treatment of this adverse effect exceeded the limit of five consecutive sessions, constituting a withdrawal criterion. In the delayed HBOT group, all seven participants were excluded because of non-compliance with the treatment protocol within the first week of the HBOT sessions. According to Little’s MCAR test, the data were missing at random (X²=27.499; df=22; p=0.19). Therefore, we chose to analyse the data by intention to treat, retaining all 56 participants.

Sample characteristics

Table 1 shows the characteristics of the study patients. Except for the body mass index (BMI), for which the delayed HBOT group had a higher mean than the early HBOT group, there was no statistically significant difference between the groups for the other confounding variables. The mean age of the patients was 51.0±9.8 years, and most were overweight, with a mean BMI of 30.5±5.1 kg/m². The time since diagnosis of FM ranged from 2 years to 21 years, with a mean of 8.9±5.9 years. Most of the participants (76.8%) had at least one comorbidity. The total number of symptoms reported by the patients ranged from 9 to 15, with all patients (100%) reporting widespread pain, anxiety and sleep disturbances. Among the study cohort, 42.8% of the patients had a family history of FM.

Table 1. Characteristics of patients included in the study.

Variables	Early HBOT (n=29)*	Delayed HBOT (n=27)*	P value
Age (years)	51.1±9.4	50.1±10.4	0.96
BMI (m/kg2)	29.0±4.0	32.2±5.7	0.02†
Fibromyalgia diagnosis (years)	9.9±5.7	7.8±5.9	0.20
Comorbidities (n)	1.0±1.0	2.0±1.0	0.58
Symptoms (n)	13.0±1.3	12.9±1.4	0.76
Tender points (n)	17.5±1.2	17.5±1.1	0.85
Fibromyalgia family history (yes, %)	11 (37.9)	16 (59.2)	0.11
Physical activity (yes, %)	12 (41.4)	6 (22.2)	0.12

^*Mean±SD and frequency (percentage) of patient characteristics.

^†Statistically significant difference according to Student’s t-test, p<0.05.

BMI, body mass index; HBOT, hyperbaric oxygen therapy.

Interventions and analysis of change

Clinical questionnaire scores were analysed with a between-subjects factor (early vs delayed HBOT) and a repeated measures factor (baseline, 8 weeks and 16 weeks). There was no statistical difference between the groups (early vs delayed HBOT) at baseline in the FIQR-Br, VAS, EAS-40 and SF-12 clinical questionnaire scores (table 2).

Table 2. Questionnaire scores (mean±SD) by group: early and delayed hyperbaric oxygen therapy.

Variables/groups	Period 1		P2	Mean difference	Period 2	P3	Mean difference	P4
	Baseline	8 weeks			16 weeks
FIQR-Br
Early HBOT	76.7±16.2	52.0±18.7	<0.001*	−24.7	66.6±18.5	0.003*	14.6	0.14
Delayed HBOT	76.5±14.7	64.3±16.6	0.02*	−12.2	49.3±16.0	0.003*	−15.0	1.00
P1	1.00	0.18			0.007*
VAS
Early HBOT	8.5±1.5	3.8±1.9	<0.001*	−4.7	6.2±2.1	<0.001	2.4	<0.001*
Delayed HBOT	8.2±1.6	5.2±2.2	<0.001*	−3.0	4.2±1.6	0.61	−1.0	1.00
P1	1.00	0.25			0.004*
EAS-40
Early HBOT	1.2±0.4	0.8±0.4	<0.001*	−0.4	0.9±0.4	1.00	0.1	0.04*
Delayed HBOT	1.1±0.4	0.9±0.4	1.00	−0.2	0.8±0.4	0.39	−0.1	1.00
P1	1.00	1.00			1.00
SF-12 PCS
Early HBOT	29.7±7.4	40.2±7.8	<0.001*	10.5	36.1±8.1	0.03*	−4.1	0.003*
Delayed HBOT	29.9±6.6	33.2±6.1	0.24	3.3	35.8±6.1	0.94	2.6	0.36
P1	1.00	0.007*			1.00
SF-12 MCS
Early HBOT	30.4±9.8	44.3±10.4	<0.001*	13.9	35.7±11.3	0.006*	−8.6	0.57
Delayed HBOT	31.3±10.9	37.8±9.3	0.06	6.5	43.9±10.7	0.17	6.1	1.00
P1	1.00	0.26			0.11

P1: p value for the between-group comparison at baseline and at 8 and 16 weeks after treatment—early HBOT (with HBOT) versus delayed HBOT (without HBOT).

P2: p value for the within-group comparison, baseline versus 8 weeks.

P3: p value for the within-group comparison, 8 weeks versus 16 weeks.

P4: p value for the within-group comparison (early HBOT group: 16 weeks vs baseline) and for the comparison of delayed HBOT (16 weeks) versus early HBOT (8 weeks).

Early group n=29; delayed group n=27.

Period 1=weeks 0–8; period 2=weeks 8–16.

^*Statistical differences assessed using a mixed-design 2×3 ANOVA, followed by Bonferroni post hoc tests.

ANOVA, analysis of variance; EAS-40, 40-Item Symptoms Assessment Scale; FIQR-Br, Fibromyalgia Impact Questionnaire-Brazilian Portuguese; HBOT, hyperbaric oxygen therapy; MCS, mental component score; PCS, physical component score; SF-12, 12-Item Short-Form Health Survey; VAS, visual analogue scale.

There was a significant interaction effect (group vs time) for the FIQR-Br scores (p<0.001; η²=0.23), VAS (p<0.001; η²=0.20), EAS-40 (p=0.02; η²=0.07), SF-12 PCS (p=0.001; η²=0.13) and SF-12 MCS (p<0.001; η²=0.16), as visualised in figures26. The variation observed in the clinical questionnaire scores differed over 16 weeks between the early and delayed HBOT groups. The interaction effect size was large except for that of EAS-40. The effect size η² indicates what percentage of the variation in the outcome is due to the group allocation on average across all evaluation points. For example, η²=0.01 means that the group allocation accounts for approximately 1% of the variation in the outcome on average across all evaluation points. The delayed HBOT group showed improved clinical scores over 16 weeks, whereas the early HBOT group experienced worsening clinical scores after discontinuation of HBOT (table 2).

Figure 2. Mean±95% CI of the Fibromyalgia Impact Questionnaire-Brazilian Portuguese (FIQR-Br) questionnaire scores at baseline, after 8 weeks and after 16 weeks of follow-up in women with fibromyalgia who underwent early (n=29) and delayed (n=27) hyperbaric oxygen therapy (HOT).

Figure 6. Mean±95% CI of the 12-Item Short-Form Health Survey (SF-12) mental component scores at baseline, after 8 weeks and after 16 weeks of follow-up in women with fibromyalgia who underwent early (n=29) and delayed (n=27) hyperbaric oxygen therapy (HOT).

Figure 3. Mean±95% CI of the visual analogue scale (VAS) scores at baseline, after 8 weeks and after 16 weeks of follow-up in women with fibromyalgia who underwent early (n=29) and delayed (n=27) hyperbaric oxygen therapy (HOT).

Figure 4. Mean±95% CI of the 40-Item Symptoms Assessment Scale (EAS-40) questionnaire scores at baseline, after 8 weeks and after 16 weeks of follow-up in women with fibromyalgia who underwent early (n=29) and delayed (n=27) hyperbaric oxygen therapy (HOT).

Figure 5. Mean±95% CI of the 12-Item Short-Form Health Survey (SF-12) physical component scores at baseline, after 8 weeks and after 16 weeks of follow-up in women with fibromyalgia who underwent early (n=29) and delayed (n=27) hyperbaric oxygen therapy (HOT).

The combined treatment (standard treatment+HBOT) provided greater benefits to patients than the standard therapy alone, as observed in the clinical questionnaires FIQR-Br, VAS, EAS-40 and SF-12 (table 2). From baseline to 8 weeks, early HBOT group compared with standard therapy resulted in a mean reduction of 31.1% vs 14.4% in FIQR-Br, 54.0% vs 33.5% in VAS and 28.4% vs 3.7% in EAS-40, respectively. Furthermore, the early HBOT group, when compared with standard therapy for 8 weeks, promoted a mean increase of 39.1% vs 14.8% in SF-12 PCS and 57.4% vs 31.9% in SF-12 MCS, respectively. From 8 weeks to 16 weeks, the delayed HBOT group showed a 35.0% reduction in the FIQR-Br, 47.3% in the VAS and 18.5% in the EAS-40 compared with baseline values, as well as a 23.9% increase in the SF-12 PCS and 54.6% in the SF-12 MCS. These results were like those observed in the early HBOT group at 8 weeks.

There were no significant differences in the scores of the early HBOT group after 8 weeks and delayed HBOT group after 16 weeks (table 2). The effect sizes observed in the groups were similar. From a clinical point of view, the effect of standard care combined with HBOT treatment was significant in both the early and delayed HBOT groups for FIQR-Br (d=1.39 vs 1.74), VAS (d=2.71 vs 2.46), EAS-40 (d=0.99 vs 0.74), SF-12 PCS (d=1.36 vs 0.91) and SF-12 MCS (d=1.36 vs 1.15) scores, respectively. This suggests that the benefits of the combined treatment were the same regardless of whether HBOT is initiated early or delayed.

However, the withdrawal of HBOT after 8 weeks of combined treatment resulted in the loss of the additional benefit provided by HBOT, with a return of the FIQR-Br and SF-12 MCS scores to the levels achieved with standard therapy.

Discussion

Our cross-over group, randomised trial, which compared standardised care plus early HBOT with standardised care during the first 8 weeks, after which the delayed arm also received HBOT, showed exposure-dependent improvements across pain and global impact measures and a group-by-time interaction consistent with gains accruing when HBOT was administered. Accordingly, given the chronic nature of FM and the requirement for an established diagnosis with ongoing standard care, baseline was defined as the clinical state at study entry under usual management, rather than assuming a return to a washout or pretreatment condition. A similar exposure-response signal was evident in Efrati et al’s cross-over clinical trial; 40 sessions at 2.0 ATA (90 min, 5 days/week) produced significant amelioration of FM symptoms, and single-photon emission CT demonstrated normalisation of activity in pain-processing regions. In contrast, no improvement occurred during the control period (no treatment).^{14 17} Our findings are also concordant with the delayed treatment group of Curtis et al, in which participants randomised to immediate HBOT (2.0 ATA, 8 weeks) showed improvements in pain, psychological and sleep outcomes. Those initially allocated to the waiting group showed benefits only after receiving HBOT, and these improvements persisted at the 3-month follow-up.³⁷

The persistence of clinical improvement following HBOT is likely to reflect sustained biological adaptations rather than short-lived non-specific effects. Repeated hyperbaric exposure may improve tissue oxygenation and mitochondrial efficiency, potentially mitigating chronic muscular and cerebral hypoxia described in FM and supporting restoration of cellular energy balance. In addition, hyperbaric oxygen has been shown to promote angiogenic and microvascular responses, potentially leading to longer term improvements in tissue perfusion.³⁷ Anti-inflammatory effects and modulation of nitric oxide-related pathways have also been reported and could contribute to prolonged pain reduction.³⁸ Importantly, neuroimaging studies suggest that HBOT may induce neuroplastic changes and partially normalise abnormal activity within pain-related brain networks, possibly mediated by effects on glial function and cerebral metabolism.¹⁷ These mechanisms provide a biologically plausible explanation for the residual effects observed at follow-up and support the use of a delayed treatment comparison design, in which sustained physiological changes are expected to persist beyond the active treatment period.

The domain-specific pattern we observed, characterised by the marked sensitivity of the FIQR/VAS to comparatively minor shifts in broader quality-of-life indices, mirrors prospective observational data. In a study by Atzeni et al on FM treatment, pain and anxiety improved after 10 and 20 sessions at 2.5 ATA, while fatigue improved only after 20 sessions. Sleep quality remained unchanged. The SF-36 physical component improved after 20 sessions, and the mental component summary showed significant improvement only at the later time point.¹⁴ This hierarchy of responsiveness was further supported by a randomised trial by Yildiz et al, in which 15 sessions (2.4 ATA) yielded significant between-group reductions in the number of tender points and increases in pain threshold, with VAS pain showing a pronounced reduction by the 15th session.³⁸ Together, these studies suggest that pain-centric indices change early during HBOT exposure. By contrast, broader health status measures often require longer doses to manifest clinically meaningful differences, which closely parallels our time effect and interaction results.

Finally, our dropout rate (due to intolerance of the chamber and one case with middle ear symptoms) aligns with the safety experience reported in the literature. In the study by Atzeni et al, four of 32 patients discontinued treatment (mild, reversible middle ear barotrauma in two; claustrophobia; dizziness), and middle ear events resolved within days.¹⁴ Curtis et al reported high feasibility (17/18 completers), and the most frequent adverse events were mild middle ear barotrauma (three patients) and new-onset myopia (four patients). Overall improvements were attributed to HBOT rather than waiting.³⁷ This convergence supports both the clinical plausibility of the observed beneficial therapy and the expected generally manageable ocular adverse event spectrum associated with HBOT.

The strengths of this study include its randomised clinical trial design and sample intention-to-treat analysis. The protocol has been previously published to minimise publication bias, and the sample size was statistically calculated for the primary and secondary outcomes. This study reinforces the effects of HBOT on neuroplasticity and the modulation of the inflammatory process, as well as the modulation of nociplastic pain.³⁹ Importantly, this is the first study to specifically examine HBOT as a treatment for FM in the Brazilian population. Our findings confirm the benefits of HBOT as a therapeutic option for FM with minimal side effects.

Regarding study limitations, due to the return of FIQR-Br and SF-12 MCS scores in the early group to levels achieved with standardised treatment alone in the delayed group when compared between the groups, there is likely a time course for the effect of the combined treatment or even a dose–response relationship to the combined treatment for these measurements, which should be investigated in future dose–response studies. Other limitations observed in the study, as foreseen in the protocol,¹⁸ included a dropout rate of 21.4% due to the long-term nature of the treatment (40 sessions) and a moderate power of the sample size ratio for the primary outcome. The absence of a placebo control group is a recognised limitation in studies involving HBOT and was also present in our study. Furthermore, the lack of a classical washout period between study phases cannot be excluded as a factor that may have influenced outcomes during the second phase of the trial.

In this clinical trial, HBOT added to guideline-based care demonstrated clinical and statistical efficacy in FM, with consistent improvements in pain, function, psychopathology and quality of life compared with standard care alone. The observed effect sizes support the incorporation of HBOT as a component of multimodal approaches. Regarding the timing of initiation, gains were equivalent whether HBOT was introduced early or after an initial period of optimisation of standard care, enabling individualised strategies such as early initiation for patients with a high symptom burden or a stepped approach without loss of effectiveness. Following the cessation of HBOT, the effects were partially attenuated, with residual benefits persisting for weeks, suggesting the need to evaluate maintenance schedules or booster cycles to sustain multidomain gains. HBOT is generally safe and well tolerated, with rare adverse events and no unexpected safety signals; however, adherence may be affected by non-pharmacological factors (eg, chamber discomfort).

Data availability statement

Data are available upon reasonable request.