Usual care including home exercise with versus without spa therapy for chronic low back pain: protocol for the LOMBATHERM’ study, a multicentric randomised controlled trial

In this prospective, evaluator-blinded, multi-centre, two-parallel-arm, Zelen randomised (1:1) controlled trial, we will compare UCHE alone with spa therapy in addition to UCHE for the treatment of chronic low back pain (Fig. 1).

This study will take place in and around well-established spa centres in Europe (see Table 1). Patients living within a 30-km radius of a participating spa centre and meeting eligibility criteria (Table 2) may participate. Recruitment, enrolment and follow-up visits will occur outside of spa centres (and independent thereof) in participating private offices/clinics, public general hospitals and public university hospitals. The single-payer system in France, where most of the participating spa centres are located, offers medical care to all citizens and residents, and the study population should thus represent a wide range of socioeconomic and rural versus urban backgrounds.

In general, eligibility criteria (Table 2) are designed to select for patients with chronic low back pain with a certain level of minimum pain while ruling out cases of specific low back pain. In the case of suspicion of the latter, patients will be aetiologically assessed for rule-out.

Two interventions are implemented during this trial (Fig. 1): (1) usual care including a home-exercise programme (UCHE) and (2) spa therapy. The UCHE intervention is designed to be a comparator arm that represents standard care, while limiting care heterogeneity. Patients assigned to the control arm of the study will have the UCHE intervention only, and those assigned to the experimental arm will have both the UCHE intervention and the spa therapy intervention.

The UCHE intervention will correspond to current practice (including the continuation of previous treatments, if applicable) as decided by the patient’s general practitioner. They will also be provided with a practical guide on how to manage back pain (the “back book”), including a recommended home exercise programme [15]. The practical guide on back pain and the exercise programme [15] will be presented to the patient by the recruiting investigators during the enrolment visit and prior to randomisation. Use of the “back book” and daily home exercises will be requested of patients and encouraged throughout the study (Table 3); home exercise observance will be queried at each follow-up visit (Table 3) by asking the patient on what days during the previous week he/she performed the required exercise routine. The recruiting investigator will further encourage and insist that all study participants are to keep trial participation a secret from evaluating investigators.

The composition of the spa therapy intervention is based on previous results in the domain. Massage [16‐21], heat therapy [20] and water exercises [22] have all been demonstrated as helpful for low back pain. Water exercises in particular have similar efficacy to regular exercise regimens, and are sometimes better tolerated [22]. For the present study, the spa therapy intervention consists of 18 consecutive days of spa therapy (3 weeks of treatment, excepting Sundays), that must occur within 60 days of randomisation (Table 3). The latter will always combine mineral water with physical and hydrotherapeutic treatments, including massages, heat and water exercises: (1) pool activities will be carried out by qualified and trained physiotherapists. Small patient groups (limited to 8–15 patients at a time) will participate in pool sessions with 10 min of free bathing and 15 min of supervised exercises in water at 35 °C; (2) underwater massages on painful areas will be performed by qualified and trained physiotherapists in 10-min sessions with water jets at a temperature of 38 °C; (3) mud applications at a temperature of 45 °C will be performed in 15-min sessions, and will include all painful areas (even outside the lumbar region if necessary). In addition to the latter three criteria, further treatments may be added by consensus of spa practitioners in each spa centre. Patients randomised to the spa therapy arm will be invited by the recruiting investigator to keep trial participation and spa therapy a secret from evaluating investigators. Their participation in spa therapy (observance) will be recorded by spa centre staff.

If required, treatment-stopping for home-exercises will be decided via collaboration between the patient and his/her evaluating doctor. Treatment-stopping may occur for spa-therapy in case of an adverse event, is decided on a case-by-case basis by the on-site doctors overseeing spa-therapy, and does not result in study exclusion.

Following initial consent covering the UCHE intervention and subsequent baseline assessments, patients will be randomly assigned to either control or experimental groups in a 1:1 allocation ratio (Fig. 1, Table 3). In line with a single Zelen design [23, 24], patients randomised to the experimental group will be informed about the additional spa therapy intervention and invited to sign an additional consent form; furthermore, these patients can refuse spa therapy, but statistical analysis will be performed on an intention-to-treat basis. Patients allocated to the control group have already consented to the UCHE intervention and will proceed without information on spa therapy; this design is necessary to avoid bias associated with “resentful demoralisation” effects [24].

Allocation is performed using a computer-generated randomisation schedule and stratified by the recruiting investigator/spa (the latter are paired, but separate) and by whether or not the patient has a professional activity. Baseline evaluations and the subsequent randomisation are performed by the recruiting investigators through specific modules administered via a password-protected Internet-accessible electronic case report form (eCRF). The randomisation module will become accessible to investigators only after key elements of baseline data have been entered in the eCRF.

To blind the study as much as possible, follow-up assessments are performed during specific follow-up visits (see Table 3) by an “evaluating” physician, who is different from the “recruiting” investigator. Every attempt will be made to keep evaluating investigators blinded. Specifically, study participants will be instructed to not mention study participation or spa therapy to their evaluating physicians (at the end of the final follow-up visit, the study clinical research associate (CRA) will ask the evaluator a guess-the-group question in order to evaluate the success of blinding). As concerns follow-up care, emergency situations requiring specific unblinding procedures are not foreseen in this protocol. Finally, the statisticians will be blinded, i.e. study arms will be referred to as “A” or “B” when the database is presented for analysis. Unveiling of the study arms will occur after completion of planned statistical analyses.

The primary outcome is the presence/absence of a clinically relevant change (improvement of at least 30%) in the visual analogue scale (VAS) score for pain at 6 months after baseline assessments (see Table 4). This choice is justified by the fact that it is relevant to the patient, figures in international recommendations [35‐37], and is qualitative in nature. A qualitative outcome was also preferred because it should minimise the role of placebo effects, which are often significant for quantitative pain variables but not so for qualitative ones [38], and minimise effects associated with a lack of evaluator blinding, as is the case for self-evaluation questionnaires [39]. For the purposes of this study, a clinically relevant change in VAS scores for pain is defined by an improvement of at least 30% [40, 41]; improvements below this value will be considered as treatment failures (absence of clinically relevant change). Cases where patients demonstrate a 30% improvement in VAS scores for pain but nevertheless require hospitalization for back pain during the study follow-up period will also be considered as treatment failures.

Further outcomes are presented in Table 4 and cover changes over time in VAS pain scores, and changes over time assessed by three validated questionnaires: (1) the impact of lower back pain on daily life (the Rolland and Morris Disability Questionnaire (RMDQ) [27, 28]), (2) inappropriate fears and beliefs about lower back pain (the fear, avoidance, belief questionnaire (FABQ) [29]) and (3) a general measure of quality of life (the Euroqol Group 5 dimension, 5 level questionnaire (EQ-5D-5 L questionnaire) [30, 31]). Patients will also be questioned about how they perceive the acceptability of their symptoms, using the Patient Acceptable Symptom State (PASS) [33]. The overall opinion of the patient and the physician on the patient’s state of health (5-point Likert scale) will be recorded. Finally, per-patient health resource use will be quantified by estimating drug consumption, infiltrations, hospitalizations, imaging and sick leave. The latter will be used, in conjunction with the French national cost scale, to estimate the cost of care after 12 months of follow up for each patient (Table 4).

Data from two unpublished, open-label studies on typical patients receiving spa therapy indicate estimated rates of clinically relevant change in VAS pain scores of 47% (8/17) and 38% (10/26). Our hypothesis is that the proportion of patients experiencing improvement at 6 months will be 40% in the experimental arm (UCHE plus spa treatment) versus 25% in the control arm (UCHE only). Considering a type 1 risk level set at 4% (1% will be used for an interim analysis following the inclusion of the first 100 patients) and a type 2 risk level set at 20% (i.e. statistical power of 80%), the minimum sample size required is 322 participants. To maintain power for analysis on observed data and expecting a drop-out rate of approximately 10%, this estimate is increased to a total of 358 paricipants (179 per study arm).

Recruitment will be assisted by advertising (not mentioning spa therapy) in the local press and by posting information in local pharmacies, physicians’ offices or clinics. In the case of recruitment difficulties, local companies known to have high rates of workers on sick leave for low back pain may be contacted. Following the advertising campaign, telephone contact between responding, potential study participants and a clinical research technician (CRT) will result in referral to a recruiting investigator (depending on location), who is responsible for obtaining consent and for enrolment, baseline evaluations and randomisation. The enrolment visit will start with an initial presentation and provision of informed consent for the UCHE intervention (Table 3). Following presentation of the “back book” and baseline evaluations, participants will be randomised and those allocated to the experimental arm will be invited to sign an additional consent form for spa therapy. As no biological collections or ancillary studies are foreseen in this protocol, further consent procedures beyond the latter are not required. Appointments will be made for spa therapy (in the next 60 days (maximum) for patients consenting to spa therapy in the experimental arm) and for follow-up visits with evaluating physicians for all patients in both arms at months 1, 6 and 12. The specific assessments required at baseline and during follow-up visits are detailed in Table 3 and closely follow the study outcomes plus the following: demographic data and disease history at baseline, home exercise observance and any required harms reporting. No specific post-trial care (beyond routine care) is foreseen.

Continued contact between study CRTs and patients is encouraged. Patients are requested to contact the study CRT in the case of visit unavailability or other problems. Conversely, the study CRTs will be contacting patients during the week preceding follow-up visits to provide a visit reminder and/or rescheduling as required.

Few prospective studies have sought to evaluate the adverse effects and complications that may occur during a spa treatment [42‐45]. Due to the absence of control groups, imputability is often not established and few studies clearly describe side effects. Among patients undergoing spa therapy, with or without radon, Franke et al. [39] observed: an increase in pain (7/1), hypertension (2/1), fatigue (2/1) or coloration of the skin and nails (2/0). Gáti et al. [46] observed three cases of respiratory tract infection, one case of cardiac arrhythmia, hypertension and cardiac decompensation, one case of cystitis and one case of gastroenteritis in relation to spa therapy. Another study reports a complete absence of adverse effects [37]. In other publications, adverse effects are not discussed. Based on adverse event reports from the largest studies [47, 48], no serious adverse events are expected. Based on the adverse effects observed during spa treatment for knee osteoarthritis, and in epidemiological studies, we expect to observe benign infections of the upper airways, increased pain or potentially leg erysipelas. A rarer but theoretically possible effect is the occurrence of opportunistic pulmonary infections (legionella) - a few cases and case series were described in the 1980s. These infections have become rare following a change in water disinfection methods in the spa centers [49].

Additional events that may be expected because they are associated with physical exercise in general are respiratory disorders and shortness of breath, cardiovascular disorders (chest pain, myocardial infarction, cardiac arrhythmia, tachycardia, vagal discomfort, hypotension or hypertension, fatigue and injuries (cramps, muscle aches, sprains, strains, tears, fractures)).

Harms reporting will be carried out in compliance with the French regulations in force (or their equivalent in other countries as appropriate) throughout the study (see Table 3). There is no anticipated harm or compensation for trial participation.

Statistical analyses will be performed by the Department of Medical Information at the University Hospitals of Montpellier, Montpellier, France, using the R programming environment [50] or SAS (enterprise guide V.7.1 or higher). An intention-to-treat structure will be implemented. The intention-to-treat population will include all randomised patients and will respect their allocations, regardless of whether or not they started the allocated interventions. Analyses may be adjusted according to clinical or inclusion criteria to take into account population heterogeneity. Though generally presented here, a more detailed statistical analysis plan will be drafted prior to the end of participant inclusion (and made available at https://​osf.​io/​ahpwy/​).

Descriptive statistics (for the total population and for each arm) will be provided as means with standard deviations for quantitative variables with normal distributions according to the Shapiro-Wilks test, or as medians with interquartile variables for other quantitative variables. Numbers and percentages will be provided for categorical variables.

As indicated in Table 4, the primary analysis is the comparison of the proportion of patients with a clinically relevant change in pain at 6 months; this will involve the uncorrected chi square (χ²) test for independent groups, or alternatively Fisher’s exact test if the conditions for the χ² test are not met. We expect to perform an interim analysis of the number of patients in both groups with clinically relevant improvement as soon as there are 100 patients suitable for analysis with 6 months of follow up. The purpose of the latter is to assess futility, and a difference between arms (intervention – control) < 0 will stop the trial.

As concerns secondary analyses, comparisons of qualitative variables will be performed in the same manner as for the primary analysis. The centrality of quantitative variables will be compared between arms using parametric (t test) or non-parametric (Mann Whitney test) analysis for independent groups, as appropriate. For longitudinal data, the preferred method will employ mixed models with patients set as a random effect and study arm as a fixed effect; an appropriate interaction term (group × time) will be used to detect differences between study arms in the rate of change. In addition, a medico-economic analysis comparing the two patient management strategies will be conducted according to international [51] and French recommendations [52]. A cost-utility analysis will be carried out, with utility values evaluated using the EQ-5D-5 L. The following direct medical costs will be considered: hospitalizations and consultations, transport, drugs and laboratory assessments.

Variables that are used for stratification (spa or presence of a professional activity) may be used in sub-group analyses. Covariate-interaction tests are the preferred method for detecting sub-group differences [53]. Statistical analyses will be adjusted on variables found to be imbalanced between arms. Sub-group tests will be considered as exploratory in nature.

The general hypothesis of the study supposes more improvement in the experimental arm as compared to the control arm. A difference will be considered statistically significant when the type I error rate is less than or equal to 0.05. As concerns missing data, patients lost to follow up will not be considered as withdrawn from the study but will be analysed as failures for the primary endpoint. Any deviations from the statistical analysis plan must be authorized by the study methodologist and will be thoroughly documented and justified in the study analysis report.

Individual patient data will be entered in a password-protected, web-accessible eCRF (Ennov Clinical; https://​en.​ennov.​com) by participating investigators or their approved delegates. A paper version of the eCRF will be made available at https://​osf.​io/​ahpwy/​, and can be used to facilitate field requirements for speed or as back up in the case of electronic system unavailability. Data entry in the eCRF will be performed in real time as much as possible in order to take advantage of specifically designed data format, range and coherence rules. The Clinical Research and Epidemiology Unit at the University Hospitals of Montpellier (Montpellier, France) is responsible for randomisation lists, database maintenance and data management procedures. Data in the eCRF will be audited against resource documents by data-monitoring personnel throughout the study, and corrections made via a traceable system of queries and responses.

In the eCRF (and paper forms if required) and subsequent study database, participants will be de-identified and only the following will be used for identification purposes during the study: two initials, year of birth and study number. Any publicly available deliverables will be completely de-identified (individual spa centres and patients are referred to by numbers only when per-patient differentiation is required, and absence of any information that may be used to re-identify any individual).

As explained in the patient information materials (https://​osf.​io/​ahpwy/​), the French Public Health Code allows LOMBATHERM’ study data, including the data of patients who withdraw their consent or who are lost to follow up, to be used specifically for the purposes of this study, and only this study, unless the patient so opposes. Re-use of the data by other teams via data sharing is subject to approval by the Commission nationale de l’informatique et des libertés (National Commission for Data Protection) (France), as stated in the data sharing plan (https://​osf.​io/​ahpwy/​).

The project will be monitored by clinical research associates delegated by the Sponsor (Sponsor CRAs) via a subcontract with the University Hospitals of Montpellier. Monitoring will be performed by regular on-site or remote inspections at investigating centres (enrolment visit, follow up according to the pace of inclusions and a closure visit). All monitoring visits will be the subject of a written report and will cover consent procedures and protocol adherence.

The LOMBATHERM’ steering committee (RF, NM, CMS) will (1) supervise study implementation and execution, (2) determine study continuation following interim analysis and (3) oversee communication activities in accordance with the data sharing plan. Key study documents (participant information materials, statistical analysis plan, analytic code, data sharing plan) will be made available to the public on the study’s Open Science Framework website: https://​osf.​io/​ahpwy/​. The final study database will be available to all study investigators immediately after the database is locked. The final study report will be communicated to the study Sponsor and to French authorities within 12 months of database freezing. Aggregated trial results will be made available to the public on ClinicalTrials.gov and by publication in a peer-reviewed journal. At this time, the use of professional writers who are not recognized authors is not foreseen, and authorship will be attributed by general consensus and according to the criteria stipulated by the International Committee of Medical Journal Editors (http://​www.​icmje.​org). In addition, results will be made available to any study participant upon request, as per French regulations in force. Finally, the study data sharing plan stipulates that requests to the AFRETh (Association Française pour la Recherche Thermale, 1 rue Cels 75,014 Paris, France; http://​www.​afreth.​org) for individual datasets can be made anytime following full publication of results.