Effectiveness and Cost-Effectiveness of Different Weekly Frequencies of Pilates for Chronic Low Back Pain: Randomized Controlled Trial

Study Design

This study is a randomized controlled trial with blinded assessor.

Study Setting

The study will be conducted at a physical therapy clinic in São Paulo, Brazil.

Sample Size

The sample of 74 participants per group was determined by a sample calculation designed to detect a clinically significant difference of 1 point in pain intensity on the Pain Numerical Rating Scale (estimate for standard deviation=1.84 points) after intervention and 4 points on the Roland-Morris Disability Questionnaire (estimate for standard deviation=4.9 points) after intervention. The calculation was done considering the primary objective of this study. The following specifications were considered: α=.05, statistical power of 80%, and follow-up loss of 15%.

Eligibility Criteria

The study will include 296 patients from the community of both sexes with chronic nonspecific low back pain for at least 3 months and age between 18 and 80 years. Advertisements will be placed in a regional newspaper and on the university website. The exclusion criteria will be any contraindication to exercise (assessed with the Physical Activity Readiness Questionnaire),⁴⁷ Pilates treatment for low back pain in the last 3 months, pregnancy, serious spinal pathologies, and previous or scheduled spinal decompression surgery.¹

Assessment

Initially, the participants will sign an informed consent form to take part in the study. A previously trained blinded assessor will conduct the assessment to confirm the eligibility criteria and collect demographic and anthropometric data from the participants, as well as information on type of medication, previous physical therapy treatment, or other treatment (if any) for low back pain. After assessing participant eligibility, the assessor will evaluate the outcomes. This initial evaluation will be done before the patients are randomized to the treatment groups. Due to the nature of the interventions, it will not be possible to blind the patients and therapists involved in the study. The assessor will be aware of the allocation only after data analysis.

All of the scales and questionnaires to be used have already been translated and adapted to Brazilian-Portuguese and have adequate measurement properties.^48–53 The primary outcomes will be pain intensity and disability assessed 6 weeks after randomization. The secondary outcomes will be quality-adjusted life-years, specific functional disability, global impression of recovery, pain catastrophizing, and kinesiophobia 6 weeks, 6 months, and 12 months after randomization. Other secondary outcomes will be pain intensity and disability 6 and 12 months after randomization.

The Pain Numerical Rating Scale assesses pain intensity on an 11-point numerical scale (0–10 points), with 0 being “no pain” and 10 being “pain as bad as could be.” The patients will be asked to rate their average pain in the last 7 days.⁴⁸ The Roland-Morris Disability Questionnaire evaluates general functional disability caused by the physical limitations resulting from referred pain to the low back in the last 24 hours. It consists of 24 yes/no questions related to normal daily activities, with affirmative answers worth 1 point. The score is the sum of the points, and the lower the score is, the better the results. Scores close to 24 indicate greater limitation. Scores above 14 points represent severe spinal compromise.^48,49,54

The Global Perceived Effect Scale evaluates global impression of recovery, comparing the onset of symptoms with the last few days. It is an 11-point numerical scale (−5 to 5) that varies from “vastly worse” to “completely recovered.” A higher score means better recovery from the condition.⁴⁸ In the Patient-Specific Functional Scale, participants identify 3 important activities that they are having difficulty doing or that they are unable to perform due to chronic low back pain at the time of the assessment. The participants will mark on an 11-point scale (0–10 points) how capable they feel of performing these activities, with 0 meaning “unable to perform the activity” and 10 meaning “able to perform the activity at preinjury level.” The average of the scores of the 3 activities is calculated; the higher the score is, the greater the functional ability.⁴⁸

The Pain Catastrophizing Scale has 13 items regarding thoughts and feelings when patients experience pain. This instrument has 3 subscales: rumination, magnification, and helplessness. The scores vary from 0 (“not at all”) to 4 (“always”) points, and the maximum score is 52 points; higher scores indicate higher pain catastrophizing.^55,56 The Tampa Scale for Kinesiophobia consists of 17 questions related to pain and intensity of symptoms. The scores vary from 0 to 4 points, with 1 point for the answer “strongly disagree,” 2 points for “partially disagree,” 3 points for “partially agree,” and 4 points for “strongly agree.” To calculate the final total score, it is necessary to invert the scores of questions 4, 8, 12, and 16. The total score ranges from 17 to 68 points, and the higher the score is, the more severe the kinesiophobia.^50,51 The Short-Form 6 Dimensions questionnaire (SF-6D) measures health-related quality of life,^57,58 taking the following items into consideration: physical functioning, role–limitation, social functioning, pain, mental health, and vitality.⁵⁷ The SF-6D score varies on a scale of 0 to 1, with 0 representing the worst health condition and 1 representing the best health condition.^52,53,57,58 These scales and questionnaires will be applied at baseline and at the 6-week, 6-month, and 12-month follow-ups in all groups. The 6-week, 6-month, and 12-month follow-ups will be done over the phone to minimize the loss of participants. Recruitment began in September 2014 and has the target date of December 2016 for completion. Table 1 shows the time line of the study.

Randomization and Interventions

The process of randomization will be performed on Microsoft Excel for Windows (Microsoft Corp, Redmond, Washington) by a researcher who was not involved in participant recruitment. Allocation will be concealed in sequentially numbered, sealed opaque envelopes. After assessment, the eligible participants will be referred to the physical therapist responsible for the treatments, who will conduct the randomized allocation to 1 of 4 treatment groups: booklet group, Pilates group 1, Pilates group 2, and Pilates group 3.

All groups will receive an educational booklet containing information on the anatomy of the spine and pelvis, low back pain, and recommendations related to posture and movements of activities of daily living.⁴ The booklet group will not receive additional exercises and will be advised not to receive treatment elsewhere during the first 6 weeks of participation in the study. After the 12-month follow-up, the intervention based on the Pilates method will be offered to this group.

Pilates groups 1, 2, and 3 will receive an exercise program based on mat Pilates (performed on the ground with or without accessories) and equipment-based Pilates (performed on the Cadillac, Reformer, Barrel, and Chair machines) for 6 weeks, and the sessions will last 1 hour. In Pilates group 1, the weekly frequency of treatment will be once a week (6 treatment sessions). In Pilates group 2, the weekly frequency will be twice a week (12 treatment sessions). In Pilates group 3, the weekly frequency will be 3 times a week (18 treatment sessions).

In the first session, all participants of the Pilates groups will receive instructions on the Pilates method and training for the activation of the powerhouse, which consists mainly of isometric contraction of the deep abdominal muscles (transversus abdominis, gluteus maximus, and pelvic floor) while exhaling.^19,20,59 The powerhouse is activated during exercise throughout the session. The main muscles that will be trained are: biceps brachii, deltoid, pectorals, serratus anterior, rhomboid, latissimus dorsi, transversus abdominis, external oblique, internal oblique, rectus abdominis, multifidus, gluteus maximus, gluteus medius, hamstring, hip adductor, iliopsoas, rectus femoris, sartorius, tensor fasciae latae, and pelvic-floor muscles. Exercises with concentric and eccentric contraction will be done in all planes of movement.^19,20 A single series of each exercise will be performed, and the number of repetitions will vary from 8 to 12 (corresponding to approximately 60%–70% of 1 maximum repetition as assessed with the Borg scale) with a 2-minute interval between exercises.^60,61

The exercises will be done at 3 levels of difficulty: basic, intermediate, and advanced. Some basic exercises can still be adapted to the conditions of each patient by reducing or increasing resistance. In some exercises, it is possible to lower the resistance of the spring to make the movement easier. For example, a roll-up exercise using the tower bar on the Cadillac machine, which involves trunk flexion movement articulating the spine in the supine position, can be done with the spring in the low position to make the movement more difficult or with the spring in the high position to make the movement easier. The level of difficulty of the exercises will be defined individually, and the evolution of the exercises will depend on individual postural compensations and comfort, with increases of 1 or 2 repetitions in relation to the desired number (2%–10% load increase)^60,61 or with modifications to the exercise according to the level of difficulty.^33,62,63 The strategy to prevent bias is the individual monitoring of the participant by a trained physical therapist and the control of the level of exercise difficulty presented by the participant; thus, exercises can be adapted if the participant's symptoms worsen. Participants who may need additional interventions will be referred to the outpatient physical therapy clinic from the Universidade Cidade de São Paulo. During the study, the participants will be allowed to use their usual medication, and this information will be monitored during the reassessments at 6 weeks, 6 months, and 12 months. The participants will be able to make up any missed sessions as long as the total intervention period, including make-up sessions, does not exceed 8 weeks. Adverse events will be monitored by the intensity of pain during execution of exercises.

Economic Assessment

The economic assessment of the intervention will consist of cost-effectiveness and cost-utility analyses in terms of incremental cost per quality-adjusted life-years if Pilates is shown to be more effective than minimal intervention. The costs will be measured by the estimate of direct costs to the public health system, costs to the private health system, and out-of-pocket expenses at 6 weeks, 6 months, and 12 months. These data will be collected using a questionnaire at follow-up every 6 weeks and will be analyzed in the 6- and 12-month follow-ups. Based on this estimate, the costs will be measured based on the participants' use of the resources. In this study, intervention costs and out-of-pocket expenses incurred by the participants will be identified, measured, and validated. Table 2 shows the types of resources that will be captured, sources of data, and proposed assessment methods. The costs of absenteeism from work will be estimated by the number of days away from work multiplied by the average wage rate. For the cost-effectiveness analyses, we will use the Pain Numerical Rating Scale and the Roland-Morris Disability Questionnaire as measures of effectiveness. For the cost-utility analyses, we will use the SF-6D as a measure of utility. The sensitivity analysis will test uncertainty in key parameters, such as the selection of cost weights and statistical variation in quality of life scores.

Statistical Analysis

Data monitoring will be performed by one author from the study who is not involved with data collection and has no conflict of interest. Auditing of the randomization process of the participants in the intervention groups will be held monthly. Statistical analysis will be performed by a researcher who will receive the data encoded. All data will be entered twice into the database. The statistician will receive the coded data and will be blind to the participants' allocation to intervention groups. The mean effects of the interventions and the group differences for all outcomes will be calculated using linear mixed models that incorporate terms for the treatment groups, time (follow-ups), and interaction terms “treatment groups” versus “time.” The term “time” will be coded as a categorical variable (ie, 4 variables will be created for the categories baseline, 6-week follow-up, 6-month follow-up, and 12-month follow-up). The coefficients of treatment × time interactions will be equivalent to the estimates for the group differences. No interim analysis will be performed. The analyses will follow the intention-to-treat principle. If a participant drops out of treatment, no additional outcome will be collected, and the missing data will not be replaced. We will use IBM SPSS for Windows software (IBM Corp, Armonk, New York) for all statistical analyses, and the level of significance will be set at 5%.