Amaral 2019

1
Examination of a subgroup of patients with chronic low back pain likely to

benefit more from Pilates-based exercises compared to an educational
booklet.
Diego Diulgeroglo Vicco Amaral, PT1

Gisela Cristiane Miyamoto, PhD1, 2
Katherinne Ferro Moura Franco, Ms1
Yuri Rafael dos Santos Franco, Ms1
Naiane Teixeira Bastos de Oliveira, Ms1
Mark Jonathan Hancock, PhD3
Maurits W. van Tulder, PhD2
Cristina Maria Nunes Cabral, PhD1
1Master’sand Doctoral Program in Physical Therapy, Universidade Cidade de
São Paulo, São Paulo, Brazil
2Department of Health Sciences, Faculty of Science, Vrije Universiteit
Amsterdam, Amsterdam Movement Sciences, The Netherlands

3Department of Health Professions, Faculty of Medicine and Health Sciences,
Macquarie University, Sydney, Australia
The protocol of this study was approved by the Research Ethics

Committee of Universidade Cidade de São Paulo and the study was
prospectively registered at ClinicalTrials.gov (NCT02241538). The Sao Paulo
Research Foundation (FAPESP) provided a scholarship to the second author of
Downloaded from www.jospt.org by Univ Canberra on 08/25/19. For personal use only.
this study, and also provided a scholarship to the principal investigator of this
study.
Address correspondence to Gisela Cristiane Miyamoto, Rua Cesário Galero,

448/475, Tatuapé, São Paulo - SP, Brazil, zipcode: 03071-000, e-mail:
gfisio_miyamoto@hotmail.com
Word count: 3,282

J Orthop Sports Phys Ther
2
Examination of a subgroup of patients with chronic low back pain likely to

benefit more from Pilates-based exercises compared to an educational
booklet.
The Sao Paulo Research Foundation (FAPESP) provided a scholarship to

the second author of this study, and also provided a scholarship to the principal
investigator of this study.
The second author of this study declares conflict of interest because she
was an instructor of NeoPilates courses at the time of data collection. NeoPilates
is a type of exercise that associates the principles of Pilates with characteristics
of functional training and circus. Although NeoPilates has similar name to Pilates
method, the exercises are performed with different approach and in different
equipment. Other authors have no conflict of interest to declare.
3
ACKNOWLEDGEMENTS
Sao Paulo Research Foundation (FAPESP) (process number:
2013/26321-8, 2015/18974-7 and 2016/07915-2).

4
1 ABSTRACT
2 Study design: Secondary analysis of a randomised controlled trial.
3 Objectives: To investigate whether two previously published classification
4 approaches: updated treatment-based classification system (TBCS) and Pilates
5 subgroup defined by a preliminary clinical prediction rule (CPR) can identify
6 patients with chronic low back pain (LBP) who benefit more from Pilates
7 compared to an educational booklet.
8 Background: Pilates is recommended for the treatment of chronic LBP.
9 Currently, there are potential subgroups of patients with LBP that could potentially
10 benefit more from Pilates. However, these subgroups of patients have not been
11 tested as effect modifiers in a randomized controlled trial.
12 Methods: 222 patients received advice and were randomly allocated to an
13
educational booklet group who did not receive additional treatment (n=74) or a
14 Pilates group (n=148) who received treatment two or three times a week. At
15 baseline patients were classified using a TBCS into having good prognosis
16 (positive movement control) or not. Similarly using the Pilates CPR patients were
17 classified as having a good (positive) or poor prognosis (negative). The analysis
18 was conducted using interaction terms (interaction between characteristics of the
19 subgroups and effect size of treatment) in linear regression models, considering

20 the changes in pain and disability from baseline to six-week after randomization
21 as dependent variables.
22 Results: None of the interaction terms for pain and disability were statistically
23 significant. Thus, the effect of treatment (Pilates versus an educational booklet)
24 was similar in all subgroups.

5
25 Conclusion: The TBCS and the Pilates CPR were unable to identify subgroups
26 of patients with chronic LBP who are likely to benefit more from Pilates compared
27 to an educational booklet.
28 Key words: Low back pain, Pilates, Subgroup, Rehabilitation
29
30 INTRODUCTION
31 Low back pain (LBP) is a common condition with a lifetime prevalence of
32 39%.23, 29 LBP is the primary cause of years lived with disability and absenteeism
33 in the world, and results in very high socioeconomic costs.5, 6, 23, 30, 47 About 80%
34 of patients have non-specific LBP, when a known specific pathology is absent.32,
35 35 Clinical practice guidelines and systematic reviews recommend a range of
36 interventions including general exercise, tai-chi, yoga, Pilates and motor control
37 exercise to improve pain and disability in patients with chronic LBP.1, 18, 45, 56
38 However, there is good evidence that the average benefit of different types of
39 exercise is similar for patients with chronic LBP.1, 26, 48 Furthermore, the
40 magnitude of the treatment effects is typically small to moderate.45
41 Patients with non-specific LBP present a diversity of characteristics
42 (psychological, physical, clinical and demographic) and a variable clinical
43 course.18, 31 Thus, it is unlikely that a standardized intervention for this

44 heterogeneous condition will be effective for all patients.18, 31 Classification and
45 identification of patients with non-specific chronic LBP into subgroups who
46 respond best to specific interventions is important to optimize the effect size of
47 existing interventions.31 The development of clinical prediction rules (CPRs) has
48 been used to identify patients who are likely to benefit from a specific
49 intervention.11, 20, 46 Recent studies have investigated if specific exercise

6
50 programs, such as the McKenzie method, motor control exercises and the
51 Pilates, may be more effective in specific subgroups of patients with LBP.22, 46, 50
52 Pilates has been recommended for the treatment of patients with chronic
53 LBP, though the effects are small to moderate.56 Stolze et al50 developed a
54 preliminary CPR to identify a subgroup of patients with LBP who benefit from
55 Pilates. This CPR suggested that patients with three or more variables, including
56 left or right hip internal or external rotation range of motion of 25 o or more, total
57 trunk flexion range of motion of 70o or less, body mass index of 25 kg/m2 or more,
58 no leg symptoms in the last week, and duration of current symptoms of six months
59 or less, have a 54% to 93% probability of improvement of symptoms after Pilates-
60 based treatment. However, in a cohort study is not possible to distinguish whether
61 the CPR identified is prognostic or if it is an effect modifier.49 The study by Stolze
62 et al50 did not include a control group, and it is essential to test this CPR in a
63 randomized controlled trial to determine if the rule is an effect modifier for Pilates.
64 Effect modifiers are patient characteristics indicative of subgroups of patients who
65 respond differently to the same treatment.25
66 Another classification system that could identify patients who respond best
67 to the Pilates is the updated treatment-based classification system (TBCS).2, 18
68 The subgroups defined by this system include 1) a symptom modulation group

69 with the recommendation of directional preference exercises,
70 mobilization/manipulation, traction and active rest; 2) a movement control group
71 with the recommendation of sensorimotor, stabilization and flexibility exercises;
72 and 3) a functional optimization group with the recommendation of work- or sport-
73 specific tasks, strengthening, conditioning, aerobic and general fitness
74 exercises.2 The main aim of Pilates exercises is to improve muscle control, core
7
75 stability, flexibility, strength and posture.53 Thus, the movement control subgroup
76 is expected to benefit most from Pilates. As Pilates is also focused on
77 strengthening exercises, it could potentially provide benefit for the functional
78 optimization group. However, the approach of this subgroup is focused on work-
79 and sport-specific tasks, conditioning, aerobic and general fitness exercises,
80 which Pilates is not.
81 However, we are not aware of studies that have evaluated the reliability of
82 the TBCS. Furthermore, whether these subgroups act as effect modifiers for
83 these interventions has yet to be tested in a randomized controlled trial. Thus,
84 the aim of this study was to investigate whether two previously published
85 classification approaches (CPR and TBCS) can identify patients with non-specific
86 chronic LBP who are likely to benefit more from Pilates-based exercises
87
compared to an educational booklet.
88
89 METHODS
90 Study design and setting
91 This study is a secondary analysis using data from a randomised
92 controlled trial.37 Details of the study design have been described elsewhere.37,
93 38 The study was conducted at a physical therapy clinic and a Pilates clinic in Sao
94 Paulo, Brazil. The protocol of this study was approved by the Research Ethics
95 Committee of Universidade Cidade de São Paulo and the study was
96 prospectively registered at ClinicalTrials.gov (NCT02241538).
97 The randomized controlled trial assessed the effectiveness and cost-
98 effectiveness of the addition of different doses of Pilates to advice in the treatment
99 of patients with non-specific chronic LBP.37 Two hundred and ninety-six patients,
8
100 recruited from the community, were randomized to one of four groups (n=74 per
101 group): educational booklet group, Pilates group 1 (one session per week),
102 Pilates group 2 (two sessions per week) and Pilates group 3 (three sessions per
103 week). The main results of this randomized controlled trial showed that all Pilates
104 groups were more effective than the educational booklet group for pain and
105 disability at six weeks. However, only Pilates groups 2 and 3 were considered to
106 have clinically important effect sizes compared to the educational booklet group
107 for pain and disability at six weeks. At the 6-month assessment only Pilates group
108 2 was more effective than the educational booklet group for pain and disability,
109 however, the effect was small. At the 12-month assessment none of the Pilates
110 groups provided additional effects compared to the educational booklet group.
111 For this secondary analysis, we therefore prospectively decided to only include
112
the educational booklet group (n=74), and Pilates groups 2 and 3 combined
113 (renamed as Pilates group, n=148) and to analyse only the six-week follow-up
114 data where main effects were larger.
115
116 Patients
117 Two hundred and twenty-two patients aged between 18 and 80 years with
118 non-specific chronic LBP lasting more than 12 weeks32, 52 were included in this
119 study. LBP was defined as discomfort or pain localized below the costal margin
120 and above the inferior gluteal folds, with or without referred lower extremity pain.1
121 Patients with serious spinal pathologies (e.g. tumours, fractures and inflammatory
122 diseases), previous or scheduled spinal surgery, nerve root compromise,
123 pregnancy, Pilates treatment for LBP in the previous three months, and any
124 contraindication to physical exercise (assessed by the Physical Activity

9
125 Readiness Questionnaire)10 were excluded. All patients signed informed consent
126 prior to their participation.
127
128 Assessment
129 The baseline assessment included demographic information, clinical
130 characteristics of pain and physical examination findings. This assessment
131 provided all the data required for the subgroup classifications investigated in this
132 study. The physical examination included the positive prone instability test, and
133 measurement of total trunk range of motion, hip flexion, hip internal and external
134 rotation range of motion using a goniometer.21, 28, 34, 36
135
136 Primary outcomes
137
Patients completed the assessment of pain and disability at six weeks after
138 randomisation. Pain intensity was assessed using the 11-point Pain Numerical
139 Rating Scale (0 representing “no pain” and 10 representing “pain as bad as could
140 be”).14 Patients were asked to rate their average pain during the last seven days.
141 The Pain Numerical Rating Scale has good levels of reproducibility (intraclass
142 correlation coefficient [ICC]: 0.85, 95% confidence interval [CI]: 0.77 to 0.90),
143 responsiveness (standardized effect size: 1.16) and construct validity.14 Disability
144 was assessed using the Roland Morris Disability Questionnaire ranging from 0 to
145 24 points, with scores close to 24 indicating greater limitation.14, 15, 41 The Roland
146 Morris Disability Questionnaire has good levels of reproducibility (ICC: 0.94, 95%
147 CI: 0.91 to 0.96), responsiveness (standardized effect size: 0.70), internal
148 consistency (Cronbach’s alpha: 0.90) and construct validity.14
149
10
150 Subgroup classification
151 Patients were classified into subgroups of the TBCS,2 and Pilates
152 subgroup defined by the CPR50 by two independent assessors based on the
153 baseline data collected. Disagreements were solved first by discussion and then
154 by arbitration of a third assessor, if disagreements persisted. A customized sheet
155 was used to extract relevant information for each subgroup classification.
156 In the TBCS, patients were classified into one of the three subgroups:
157 symptom modulation, movement control and functional optimization.2 Criteria for
158 classification was based on pain intensity (high to moderate, moderate to low,
159 and low to absent), disability level (high, moderate and low) and clinical status
160 (volatile, stable and well-controlled). Details of the classification are presented in
161 TABLE 1. We hypothesised that the movement control subgroup would respond
162
best to Pilates compared with symptom modulation and functional optimization
163 subgroups. Pilates is considered a mind-body exercise with focus on breathing,
164 posture, muscle control, core stability, strength and flexibility,53 presenting some
165 of the characteristics recommended in the intervention prescribed for the
166 movement control subgroup.2 Thus, we combined the symptom modulation and
167 functional optimization subgroups into one subgroup, called negative movement
168 control subgroup (poor prognosis for response to Pilates), and compared it with
169 the positive movement control subgroup (good prognosis for response to Pilates).
170 Using the Pilates CPR for patients with LBP, which consists of five
171 predictors (TABLE 2),50 patients were also classified into a positive Pilates
172 subgroup (good prognosis for response to Pilates) and a negative Pilates
173 subgroup (poor prognosis for response to Pilates). The criteria for the positive
11
174 Pilates subgroup were the presence of three or more of the five predictor
175 variables.
176
177 Intervention
178 Details of the intervention have been published elsewhere.19, 37 All patients
179 received an educational booklet containing information about LBP, anatomy of
180 the spine and recommendations related to activities of daily living and posture.8
181 The booklet group did not receive additional treatment. The Pilates group
182 received an individual Pilates-based exercise program two or three times a week,
183 over six weeks.
184 In the first session of the Pilates group, patients received instructions on
185 the Pilates principles and training for the activation of the deep abdominal
186 muscles while exhaling, during all exercises.39, 40 Pilates-based exercise program
187 consisted of five minutes of warm-up (breathing and mobility exercises), 50
188 minutes of Pilates-based exercises (stretching and strengthening exercises for
189 muscles of the trunk, lower and upper limbs), and five minutes of cool down
190 (relaxation exercises and massage with ball). The Pilates-based exercises were
191 performed in single series, with the number of repetitions varying from eight to
192 12, and at three levels of difficulty (basic, intermediate, and advanced). The
193 progression of the exercises was individualized considering the physical
194 conditioning, comfort, and individual postural compensations.3, 4
195
196 Physical therapists
197 The treatment of the patients was performed by five physical therapists
198 certified in Pilates. These physical therapists had a minimum of three years and
12
199 a maximum of eight years of experience in the treatment of patients with LBP
200 with Pilates. As the physical therapists were certified at different Pilates schools,
201 they received specific training on the Pilates-based exercise program used in this
202 study.
203
204 Statistical analysis
205 A subgroup analysis was conducted using linear regression models
206 considering the change in pain intensity and disability from baseline to six weeks
207 after randomization as dependent variables. The TBCS and the CPR were
208 investigated in separate models. Each model included terms for treatment group,
209 subgroup and the interaction term (group x subgroup). As this is an exploratory
210 secondary analysis and likely underpowered, we assessed both the statistical
211
significance (p<0.05) and the point estimates of the interaction term (the
212 interaction between characteristics of the subgroup and the effect size of
213 treatment). We considered an interaction term of greater than 1-point on the Pain
214 Numerical Rating Scale and 3-point on the Roland Morris Disability Questionnaire
215 as potentially clinically important at six weeks follow-up. Interaction terms
216 represent how much more effective a treatment is, compared to control, in the
217 patients in the subgroup compared to patients who are not in the subgroup. There
218 is no consensus on what constitutes a clinically important interaction for pain or
219 disability as it depends on the main effect size and the costs and harms of the
220 interventions.25 We selected the values for the interaction term after considering
221 these factors. The assumptions of normality, multicollinearity and linearity were
222 not violated in both models and were considered present in the occurrence of
13
223 tolerance lower than 0.10. A test of normal distribution of the linear regression
224 models were conducted by plotting both residuals and normal distribution.
225
226 RESULTS
227 TABLE 3 describes the participants’ characteristics. Most patients were
228 women, married, overweight, with tertiary education, and non-smokers. From the
229 222 patients assessed, one patient was excluded due to being diagnosed with
230 cancer during study, 13 patients did not answer the assessment of pain intensity
231 and disability at six weeks follow-up (five patients in educational booklet group
232 and eight patients in Pilates group), and two patients did not present sufficient
233 information for the classification of the Pilates CPR subgroup. Thus, 208 patients
234 were analyzed in the TBCS analysis, and 206 patients were analyzed in the
235
Pilates CPR subgroup analysis. Considering the TBCS, 64.9% of patients in the
236 educational booklet group and 58.1% in the Pilates group were classified into the
237 negative movement control subgroup. According to the Pilates CPR, 58.1% of
238 patients in the educational booklet group and 54.1% in the Pilates group were in
239 the positive Pilates subgroup.
240 TABLE 4 and TABLE 5 present the results of subgroup analyses for pain
241 intensity and disability, respectively. None of the interaction terms (positive
242 movement control subgroup or positive Pilates CPR subgroup) for pain intensity
243 and disability were statistically significant and point estimates did not exceed the
244 threshold determined for clinical importance. FIGURE 1A and FIGURE 2A show
245 means for pain intensity and disability respectively, at baseline and six weeks
246 follow-up for the Pilates and booklet groups separated by positive movement
247 control subgroup and negative movement control subgroup. The figures
14
248 demonstrate that the effect of treatment (Pilates group versus educational booklet
249 group) was similar in the positive movement control subgroup and negative
250 movement control subgroup participants. FIGURE 1B and FIGURE 2B show
251 means for pain intensity and disability respectively, at baseline and follow-up after
252 six weeks for the Pilates and booklet groups separated by positive Pilates
253 subgroup and negative Pilates subgroup. The figures demonstrate that the effect
254 of treatment (Pilates group versus booklet group) was similar in the positive
255 Pilates subgroup and negative Pilates subgroup participants.
256
257 DISCUSSION
258 Principal findings
259 The purpose of this secondary analysis of a randomized controlled trial
260
was to investigate whether the movement control subgroup of the TBCS or the
261 Pilates subgroup defined by a CPR can identify patients with non-specific chronic
262 LBP who benefit more from Pilates-based exercises compared to an educational
263 booklet. Based on our results, neither of the subgroups investigated are effect
264 modifiers for response to Pilates. The results were consistent for the two
265 assessed outcomes (change in pain intensity and change in disability). While the
266 confidence intervals for the interactions are somewhat wide, two of the four
267 interaction terms were in the opposite direction to the hypothesis. Moreover, the
268 limits of the confidence intervals in the direction of the hypotheses were relatively
269 small (<1.3 for pain and <3.4 for disability) suggesting we did not miss important
270 moderation effects.
271
272
15
273 Strengths and weaknesses of the study
274 A strength of this study was that the data were derived from a randomized
275 controlled trial.37 Furthermore, this study was designed before the beginning of
276 the randomized controlled trial. Thus, the variables and clinical outcomes were
277 collected prospectively with the purpose of investigating effect modification.51
278 This secondary analysis is the first study to investigate whether Pilates exercises
279 promote more benefits for a specific subgroup of the TBCS for patients with non-
280 specific chronic LBP. Although theoretically Pilates may be a good approach for
281 patients in the movement control subgroup of the TBCS, the TBCS was not
282 specifically developed to identify patients who are likely to respond best with
283 Pilates. In addition, this is the first hypothesis-testing study to validate the Pilates
284 CPR in a randomized controlled trial. However, the randomized controlled trial
285
was powered for the evaluation of differences in effect between the intervention
286 groups. Consequently, a limitation of this study was the lack of statistical power
287 for the subgroup analysis.7 In secondary analysis with interaction tests, a
288 randomized controlled trial with 80% power for overall effect has only 29% power
289 to detect an interaction effect of the same magnitude.7 However, the relatively
290 tight confidence intervals suggested we did not miss an important interaction
291 effect.
292
293 Comparison with other studies
294 The main criteria used for the TBCS2 are pain intensity and disability
295 levels. A prospective cohort study43 found that higher pain intensity and disability
296 at baseline are associated with greater clinical improvement in patients with
297 chronic LBP after four weeks of treatment, regardless of the intervention.43 As the
298 symptom modulation subgroup of the TBCS2 is defined by high levels of pain
16
299 intensity and disability, it may not be surprising that effect of Pilates was greater
300 in the negative movement control subgroup participants than in the positive
301 movement control subgroup. Furthermore, in our study, most patients presented
302 significant symptoms and were classified into the symptom modulation subgroup
303 (55.4% of patients in booklet group and 49.3% in Pilates group).
304 In our study the Pilates subgroup based on a CPR did not identify the
305 patients who responded best to Pilates. This Pilates subgroup was developed in
306 a cohort study50 without a control group so it is not surprising the CPR did not
307 identify those who responded best to Pilates when tested in a randomized
308 controlled trial. CPRs developed in cohort studies cannot distinguish whether
309 predictors are simply prognostic factors regardless of treatment or if they are
310 effect modifiers.49 Previous studies have shown that lack of leg symptoms24, 42
311 and shorter duration of current symptoms16, 17

are general prognostic factors
312 regardless of treatment. Other CPRs developed in cohort studies 9, 12, 27 have
313 failed to validate as effect modifiers when tested in randomized controlled trials.11,
314 13, 46 This likely occurs because CPRs developed in cohort studies are prone to
315 identifying prognostic factors rather than effect modifiers. Furthermore, patients
316 of this hypothesis-testing study had similar characteristics compared to CPR
317 study (age, body mass index, moderate pain and disability at baseline). There
318 were some differences in the patients included in the Stolze study50 (patients with
319 acute, subacute and chronic low back pain) and ours (patients with chronic low
320 back pain) and we cannot rule out that these may have contributed to the
321 differences. However, although the CPR study included patients with acute,
322 subacute and chronic low back pain, most part of patients presented symptoms
323 for more than six months.

17
324 Meaning of the study and future research
325 Although the randomized controlled trial showed that Pilates is more
326 effective than an educational booklet for patients with chronic low back pain,37
327 the present study was unable to identify effect modifiers for Pilates exercises.
328 Pilates is an individualized exercise program that aims to be adapted to individual
329 patient characteristics.33, 39, 40, 44, 54, 55 It is possible that Pilates approach had
330 relatively consistent effects across the included population and no important
331 subgroups exist. Given the current evidence26, 48, 56 that there is no specific
332 exercise that produces greater effects than other forms of exercise, and the
333 inability to identify clear effect modifiers for different types of exercise, the choice
334 of exercise approach should be made based on patient preference and clinician
335 expertise. Future research can be conducted to investigate other potential effect
336
modifiers for patients with non-specific chronic LBP who are most likely to benefit
337 from Pilates exercises. This could be conducted in a randomized controlled trial
338 to identify new variables with stronger biological rationale that have not been
339 tested as effect modifiers (e.g. hip flexion range of motion, positive prone
340 instability test, aberrant movements).
341
342 CONCLUSION
343 The results of this study suggest that movement control subgroup of the
344 TBCS nor the Pilates subgroup were not considered as a treatment effect
345 modifier for patients with non-specific chronic LBP. Therefore, specific exercises
346 did not produce greater effects than other types of exercise, thus the choice of
347 exercise approach can be made based on patient preference and clinician
348 expertise.
18
349 KEY POINTS
350 Findings
351 The results of this study show the inability to identify clear effect modifiers
352 for Pilates exercises.
353
354 Implications
355 The choice of exercise approach should be made based on patient
356 preference and clinician expertise.
357
358 Caution
359 A limitation of this study was the lack of statistical power for the subgroup
360 analysis. However, the relatively tight confidence intervals suggested we did not
361
miss an important interaction effect.
362 REFERENCES
363 1. Airaksinen O, Brox JI, Cedraschi C, et al. Chapter 4. European guidelines
364 for the management of chronic nonspecific low back pain. Eur Spine J. 2006;15
365 Suppl 2:S192-300.
366 2. Alrwaily M, Timko M, Schneider M, et al. Treatment-Based Classification
367 System for Low Back Pain: Revision and Update. Phys Ther. 2016;96:1057-
368 1066.
369 3. American College of Sports M. American College of Sports Medicine
370 position stand. Progression models in resistance training for healthy adults. Med
371 Sci Sports Exerc. 2009;41:687-708.
372 4. Armstrong L BG, Berry MJ, et al. Diretrizes do ACSM para os Testes de
373 Esforço e sua Prescrição. Rio de Janeiro, Brazil: Guanabara Koogan; 2007.
19
374 5. Bevan S. Economic impact of musculoskeletal disorders (MSDs) on work
375 in Europe. Best Pract Res Clin Rheumatol. 2015;29:356-373.
376 6. Briggs AM, Cross MJ, Hoy DG, et al. Musculoskeletal Health Conditions
377 Represent a Global Threat to Healthy Aging: A Report for the 2015 World Health
378 Organization World Report on Ageing and Health. Gerontologist. 2016;56 Suppl
379 2:S243-255.
380 7. Brookes ST, Whitley E, Peters TJ, Mulheran PA, Egger M, Davey Smith
381 G. Subgroup analyses in randomised controlled trials: quantifying the risks of
382 false-positives and false-negatives. Health Technol Assess. 2001;5:1-56.
383 8. Burton AK, Balague F, Cardon G, et al. Chapter 2. European guidelines
384 for prevention in low back pain: November 2004. Eur Spine J. 2006;15 Suppl
385 2:S136-168.
386
9. Cai C, Pua YH, Lim KC. A clinical prediction rule for classifying patients
387 with low back pain who demonstrate short-term improvement with mechanical
388 lumbar traction. Eur Spine J. 2009;18:554-561.
389 10. Carvalho T NA, Lazzoli JK, Magni JRT, Rezende L, Drummond FA,
390 Oliveira MAB, Rose EH, Araújo CGS, Teixeira JAC. Posição oficial da Sociedade
391 Brasileira de Medicina do Esporte: atividade física e saúde. Rev Bras Med
392 Esporte. 1996;2:215-220.

393 11. Childs JD, Fritz JM, Flynn TW, et al. A clinical prediction rule to identify
394 patients with low back pain most likely to benefit from spinal manipulation: a
395 validation study. Ann Intern Med. 2004;141:920-928.
396 12. Cleland JA, Childs JD, Fritz JM, Whitman JM, Eberhart SL. Development
397 of a clinical prediction rule for guiding treatment of a subgroup of patients with
20
398 neck pain: use of thoracic spine manipulation, exercise, and patient education.
399 Phys Ther. 2007;87:9-23.
400 13. Cleland JA, Mintken PE, Carpenter K, et al. Examination of a clinical
401 prediction rule to identify patients with neck pain likely to benefit from thoracic
402 spine thrust manipulation and a general cervical range of motion exercise: multi-
403 center randomized clinical trial. Phys Ther. 2010;90:1239-1250.
404 14. Costa LO, Maher CG, Latimer J, et al. Clinimetric testing of three self-
405 report outcome measures for low back pain patients in Brazil: which one is the
406 best? Spine (Phila Pa 1976). 2008;33:2459-2463.
407 15. Costa LO, Maher CG, Latimer J, Ferreira PH, Pozzi GC, Ribeiro RN.
408 Psychometric characteristics of the Brazilian-Portuguese versions of the
409 Functional Rating Index and the Roland Morris Disability Questionnaire. Spine
410
(Phila Pa 1976). 2007;32:1902-1907.
411 16. Costa LCM, Maher CG, Hancock MJ, McAuley JH, Herbert RD, Costa LO.
412 The prognosis of acute and persistent low-back pain: a meta-analysis. CMAJ.
413 2012;184:E613-624.
414 17. da Silva T, Macaskill P, Mills K, et al. Predicting recovery in patients with
415 acute low back pain: A Clinical Prediction Model. Eur J Pain. 2017;21:716-726.
416 18. Delitto A, George SZ, Van Dillen LR, et al. Low back pain. J Orthop Sports
417 Phys Ther. 2012;42:A1-57.
418 19. Franco YR, Liebano RE, Moura KF, et al. Efficacy of the addition of
419 interferential current to Pilates method in patients with low back pain: a protocol
420 of a randomized controlled trial. BMC Musculoskelet Disord. 2014;15:420.
421 20. Fritz JM, Lindsay W, Matheson JW, et al. Is there a subgroup of patients
422 with low back pain likely to benefit from mechanical traction? Results of a
21
423 randomized clinical trial and subgrouping analysis. Spine (Phila Pa 1976).
424 2007;32:E793-800.
425 21. Ganzalez GZ, Costa LdCM, Garcia AN, Shiwa SR, Amorim CF, Costa
426 LOP. Reproducibility and construct validity of three non-invasive instruments for
427 assessing the trunk range of motion in patients with low back pain. Fisioter
428 Pesqui. 2014;21:365-371.
429 22. Garcia AN, Costa Lda C, Hancock M, Costa LO. Identifying Patients With
430 Chronic Low Back Pain Who Respond Best to Mechanical Diagnosis and
431 Therapy: Secondary Analysis of a Randomized Controlled Trial. Phys Ther.
432 2016;96:623-630.
433 23. GBD 2015 Disease and Injury Incidence and Prevalence Collaborators.
434 Global, regional, and national incidence, prevalence, and years lived with
435
disability for 310 diseases and injuries, 1990-2015: a systematic analysis for the
436 Global Burden of Disease Study 2015. Lancet. 2016;388:1545-1602.
437 24. Gurcay E, Bal A, Eksioglu E, Hasturk AE, Gurcay AG, Cakci A. Acute low
438 back pain: clinical course and prognostic factors. Disabil Rehabil. 2009;31:840-
439 845.
440 25. Hancock MJ, Kjaer P, Korsholm L, Kent P. Interpretation of subgroup
441 effects in published trials. Phys Ther. 2013;93:852-859.

442 26. Hayden JA, van Tulder MW, Malmivaara A, Koes BW. Exercise therapy
443 for treatment of non-specific low back pain. Cochrane Database Syst Rev.
444 2005;CD000335.
445 27. Hicks GE, Fritz JM, Delitto A, McGill SM. Preliminary development of a
446 clinical prediction rule for determining which patients with low back pain will
22
447 respond to a stabilization exercise program. Arch Phys Med Rehabil.
448 2005;86:1753-1762.
449 28. Holla JF, van der Leeden M, Roorda LD, et al. Diagnostic accuracy of
450 range of motion measurements in early symptomatic hip and/or knee
451 osteoarthritis. Arthritis Care Res (Hoboken). 2012;64:59-65.
452 29. Hoy D, Bain C, Williams G, et al. A systematic review of the global
453 prevalence of low back pain. Arthritis Rheum. 2012;64:2028-2037.
454 30. Hoy D, March L, Woolf A, et al. The global burden of neck pain: estimates
455 from the global burden of disease 2010 study. Ann Rheum Dis. 2014;73:1309-
456 1315.
457 31. Kent P, Keating J. Do primary-care clinicians think that nonspecific low
458 back pain is one condition? Spine (Phila Pa 1976). 2004;29:1022-1031.
459
32. Koes BW, van Tulder MW, Thomas S. Diagnosis and treatment of low back
460 pain. BMJ. 2006;332:1430-1434.
461 33. Latey P. The Pilates method: history and philosophy. J Bodyw Mov Ther.
462 2001;5:275-282.
463 34. Magee D, J. Avaliação musculoesquelética. 4 Barueri: Manole; 2002.
464 35. Maher C, Underwood M, Buchbinder R. Non-specific low back pain.
465 Lancet. 2017;389:736-747.

466 36. Marques A, P. Manual de Goniometria. 2. Barueri: Manole; 2003.
467 37. Miyamoto GC, Franco KFM, van Dongen JM, et al. Different doses of
468 Pilates-based exercise therapy for chronic low back pain: a randomised
469 controlled trial with economic evaluation. Br J Sports Med. 2018;

23
470 38. Miyamoto GC, Moura KF, Franco YR, et al. Effectiveness and Cost-
471 Effectiveness of Different Weekly Frequencies of Pilates for Chronic Low Back
472 Pain: Randomized Controlled Trial. Phys Ther. 2016;96:382-389.
473 39. Muscolino JE, Cipriani S. Pilates and the “powerhouse”—I. J Bodyw Mov
474 Ther. 2004;8:15-24.
475 40. Muscolino JE, Cipriani S. Pilates and the “powerhouse”—II. J Bodyw Mov
476 Ther. 2004;8:122-130.
477 41. Nusbaum L, Natour J, Ferraz MB, Goldenberg J. Translation, adaptation
478 and validation of the Roland-Morris questionnaire-Brazil Roland-Morris. Braz J of
479 Med Biol Res. 2001;34:203-210.
480 42. Nykvist F, Hurme M, Alaranta H, Kaitsaari M. Severe sciatica: a 13-year
481 follow-up of 342 patients. Eur Spine J. 1995;4:335-338.
482
43. Oliveira IS, Costa LOP, Garcia AN, Miyamoto GC, Cabral CMN, Costa L.
483 Can demographic and anthropometric characteristics predict clinical
484 improvement in patients with chronic non-specific low back pain? Braz J Phys
485 Ther. 2018;22:328-335.
486 44. Penelope L. Updating the principles of the Pilates method: Part 2. J Bodyw
487 Mov Ther. 2002;6:94-101.
488 45. Qaseem A, Wilt TJ, McLean RM, Forciea MA, Clinical Guidelines
489 Committee of the American College of P. Noninvasive Treatments for Acute,
490 Subacute, and Chronic Low Back Pain: A Clinical Practice Guideline From the
491 American College of Physicians. Ann Intern Med. 2017;166:514-530.
492 46. Rabin A, Shashua A, Pizem K, Dickstein R, Dar G. A clinical prediction
493 rule to identify patients with low back pain who are likely to experience short-term
24
494 success following lumbar stabilization exercises: a randomized controlled
495 validation study. J Orthop Sports Phys Ther. 2014;44:6-B13.
496 47. Reid KJ, Harker J, Bala MM, et al. Epidemiology of chronic non-cancer
497 pain in Europe: narrative review of prevalence, pain treatments and pain impact.
498 Curr Med Res Opin. 2011;27:449-462.
499 48. Saragiotto BT, Maher CG, Yamato TP, et al. Motor control exercise for
500 chronic non-specific low-back pain. Cochrane Database Syst Rev.
501 2016;CD012004.
502 49. Stanton TR, Hancock MJ, Maher CG, Koes BW. Critical appraisal of
503 clinical prediction rules that aim to optimize treatment selection for
504 musculoskeletal conditions. Phys Ther. 2010;90:843-854.
505 50. Stolze LR, Allison SC, Childs JD. Derivation of a preliminary clinical
506
prediction rule for identifying a subgroup of patients with low back pain likely to
507 benefit from Pilates-based exercise. J Orthop Sports Phys Ther. 2012;42:425-
508 436.
509 51. Sun X, Briel M, Walter SD, Guyatt GH. Is a subgroup effect believable?
510 Updating criteria to evaluate the credibility of subgroup analyses. BMJ.
511 2010;340:c117.
512 52. Waddell G. The Back Pain Revolution. Second. Churchill Livingston; 2004.
513 53. Wells C, Kolt GS, Bialocerkowski A. Defining Pilates exercise: a
514 systematic review. Complement Ther Med. 2012;20:253-262.
515 54. Wells C, Kolt GS, Marshall P, Bialocerkowski A. The definition and
516 application of Pilates exercise to treat people with chronic low back pain: a Delphi
517 survey of Australian physical therapists. Phys Ther. 2014;94:792-805.

25
518 55. Wells C, Kolt GS, Marshall P, Bialocerkowski A. Indications, benefits, and
519 risks of Pilates exercise for people with chronic low back pain: a Delphi survey of
520 Pilates-trained physical therapists. Phys Ther. 2014;94:806-817.
521 56. Yamato TP, Maher CG, Saragiotto BT, et al. Pilates for low back pain.
522 Cochrane Database Syst Rev. 2015;CD010265.
523
524
525
26
526 TABLE 1. Subgroups of the treatment-based classification system

Outcomes Symptom modulation Movement control Functional optimization
Pain High to moderate (7 to 10 points) Moderate to low (4 to 6 points) Low to absent (0 to 3 points)
intensity
Disability High (14 to 24 points) Moderate (6 to 13 points) Low (1 to 5 points)
Clinical Volatile (symptoms predominate) Stable (movement impairments Well-controlled (performance

status predominate) deficits predominate)
Means: patient’s clinical status can
easily be aggravated, the patient is Means: patient’s clinical status can Means: patient’s clinical status is
highly irritable (i.e. minor lumbar increase with certain movements, asymptomatic most of the time but
spine movements easily provoke postures, or tests but returns to can be aggravated when
pain), and occasionally the baseline level relatively quickly performance demands are
patient’s presentation does not increased
permit physical examination Examination: pain is worst during
sudden movement, active movement Examination: without flexibility

Examination: patient avoids is complete but can be deficits (hip flexion more than 90o),
specific postures (flexion or abnormal/aberrant, flexibility deficit negative prone instability test, low
extension of the spine), range of (hip flexion less than 70o), positive functional limitation (difficulty to
motion is limited, spine movement prone instability test, moderate activities of great physical demand
is painful, lower limb pain, serious functional limitation (difficulty to and long duration, such as handling
functional limitation (difficulty to housework, mowing grass, or lifting heavy materials, participating in
stay more than 15 minutes heavy objects) sports or doing heavy housework)
standing, more than 30 minutes
seated, or walking more than 250
meters)
528
527
Adapted with permission from Alrwaily et al.2
27
28
529 TABLE 2. Five criteria of the clinical prediction rule for Pilates50
Criteria Definition of positive
Leg symptoms Not having symptoms in the last week
Body mass index ≥ 25 kg/m2
Total trunk flexion range of motion ≤ 70o
Hip rotation range of motion 1 hip with ≥ 25o of internal or external
rotation
Duration of symptoms ≤ 6 months
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
29
555 TABLE 3. Baseline characteristics of the patients*

Variable Booklet Pilates group
group (n=148)
(n=73)
Age (years) 48.6 (15.8) 47.9 (15.5)
Gender
Male 18 (24.3) 38 (25.7)
Female 56 (75.7) 110 (74.3)
Weight (Kg) 71.3 (15.1) 71.6 (14.2)
Height (m) 1.6 (0.1) 1.6 (0.1)
Body mass index (Kg/m²) 26.9 (5.3) 26.4 (4.5)
Family income (USD/month) 2.413 (1.700) 2.261 (1.731)
Duration of symptoms (months)a 48.0 (3 to 372) 48.0 (3 to 480)
Marital status
Single 23 (31.1) 40 (27.0)
Married 35 (47.3) 81 (54.7)
Divorced 12 (16.2) 17 (115)
Widower 4 (5.4) 10 (6.8)
Academic level
Primary education 17 (23.0) 30 (20.3)
Secondary education 24 (32.4) 45 (30.4)
Tertiary education 33 (44.6) 73 (49.3)
Smoking
No 70 (94.6) 137 (92.6)
Yes 4 (5.4) 11 (7.4)
Psychosocial statusb
Negative 45 (60.8) 106 (71.6)

Positive 29 (39.2) 42 (28.4)
Disability at baseline (0 to 24 points) 12.3 (5.5) 11.7 (4.8)
Treatment-based classification system
Negative movement control subgroup 48 (64.9) 86 (58.1)
Symptom modulation 41 (55.4) 73 (49.3)
Functional optimization 7 (9.5) 13 (8.8)
Positive movement control subgroup 26 (35.1) 62 (41.9)
Pilates subgroup
Negative 31 (41.9) 66 (44.6)
Positive 43 (58.1) 80 (54.1)
556 *Categorical variables are expressed as number (%); continuous variables are expressed as mean
557 (SD).
558
aDuration of symptoms is expressed as median (minimum to maximum).
559 bPatient was classified into positive psychological status when presented two or more of three
560 psychological characteristics (49 points or more in the Tampa Scale for Kinesiophobia [17 to 64
561 points] and/or 30 points or more in the Pain Catastrophizing Scale [0 to 52 points] and felt depressed
562 during last month [yes or no])
563
30
564 TABLE 4. Results of linear regression model for pain intensity at six weeks follow-up
Variables Beta p 95% CI
coefficient
Movement control
Treatment 2.3 <0.001 1.3 to 3.2
Positive movement control -0.9 0.16 -2.2 to 0.4
Interaction: treatment versus movement -0.3 0.73 -1.9 to 1.3
control
Pilates subgroup
CPR
Treatment 1.7 <0.001 0.5 to 2.9
Positive Pilates subgroup -0.1 0.82 -1.4 to 1.1
Interaction: treatment versus Pilates 0.6 0.19 -0.9 to 2.2
subgroup
565 *Interaction terms provide the critical information for assessing whether effect modification exists.
566 Negative interactions mean the effect was in the opposite direction to that hypothesized. Positive
567 interactions mean that the direction of the effect was in favor of the hypothesis.
568 CI: Confidence Interval
569
570 TABLE 5. Results of linear regression model for disability at six weeks follow-up
Variables Beta p 95% CI
coefficient
Movement control
Treatment 5.0 <0.001 3.0 to 6.9
Positive movement control -0.5 0.69 -3.2 to 2.1
Interaction: treatment versus movement -2.4 0.15 -5.6 to 0.9
control
Pilates subgroup
CPR
Treatment 3.7 <0.001 1.2 to 6.1
Positive Pilates subgroup 0.3 0.80 -2.6 to 2.9
Interaction: treatment versus Pilates 0.3 0.87 -2.9 to 3.4

subgroup
571 *Interaction terms provide the critical information for assessing whether effect modification exists.
572 Negative interactions mean the effect was in the opposite direction to that hypothesized. Positive
573 interactions mean that the direction of the effect was in favor of the hypothesis.
574 CI: Confidence Interval
575
576
31
577
578 FIGURE 1. Means for pain intensity at baseline and follow-up after six weeks for A)
579 movement control of the TBCS and B) Pilates subgroup defined by a CPR.
580
581
32
582
583 FIGURE 2. Means for disability at baseline and follow-up after six weeks for A)
584 movement control of the TBCS and B) Pilates subgroup defined by a CPR.
585
586

Amaral 2019

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Amaral 2019

Uploaded by

Copyright:

Available Formats

1

Examination of a subgroup of patients with chronic low back pain likely to

Diego Diulgeroglo Vicco Amaral, PT1

Amsterdam, Amsterdam Movement Sciences, The Netherlands

Macquarie University, Sydney, Australia

The protocol of this study was approved by the Research Ethics

Address correspondence to Gisela Cristiane Miyamoto, Rua Cesário Galero,

Word count: 3,282

Examination of a subgroup of patients with chronic low back pain likely to

The Sao Paulo Research Foundation (FAPESP) provided a scholarship to

Sao Paulo Research Foundation (FAPESP) (process number:

2013/26321-8, 2015/18974-7 and 2016/07915-2).

2 Study design: Secondary analysis of a randomised controlled trial.

3 Objectives: To investigate whether two previously published classification

4 approaches: updated treatment-based classification system (TBCS) and Pilates

5 subgroup defined by a preliminary clinical prediction rule (CPR) can identify

7 compared to an educational booklet.

8 Background: Pilates is recommended for the treatment of chronic LBP.

11 tested as effect modifiers in a randomized controlled trial.

12 Methods: 222 patients received advice and were randomly allocated to an

17 classified as having a good (positive) or poor prognosis (negative). The analysis

18 was conducted using interaction terms (interaction between characteristics of the

19 subgroups and effect size of treatment) in linear regression models, considering

23 significant. Thus, the effect of treatment (Pilates versus an educational booklet)

24 was similar in all subgroups.

28 Key words: Low back pain, Pilates, Subgroup, Rehabilitation

31 Low back pain (LBP) is a common condition with a lifetime prevalence of

34 of patients have non-specific LBP, when a known specific pathology is absent.32,

35 35 Clinical practice guidelines and systematic reviews recommend a range of

40 magnitude of the treatment effects is typically small to moderate.45

41 Patients with non-specific LBP present a diversity of characteristics

42 (psychological, physical, clinical and demographic) and a variable clinical

43 course.18, 31 Thus, it is unlikely that a standardized intervention for this

44 heterogeneous condition will be effective for all patients.18, 31 Classification and

45 identification of patients with non-specific chronic LBP into subgroups who

46 respond best to specific interventions is important to optimize the effect size of

47 existing interventions.31 The development of clinical prediction rules (CPRs) has

49 intervention.11, 20, 46 Recent studies have investigated if specific exercise

59 or less, have a 54% to 93% probability of improvement of symptoms after Pilates-

60 based treatment. However, in a cohort study is not possible to distinguish whether

61 the CPR identified is prognostic or if it is an effect modifier.49 The study by Stolze

64 Effect modifiers are patient characteristics indicative of subgroups of patients who

65 respond differently to the same treatment.25

67 to the Pilates is the updated treatment-based classification system (TBCS).2, 18

68 The subgroups defined by this system include 1) a symptom modulation group

69 with the recommendation of directional preference exercises,

70 mobilization/manipulation, traction and active rest; 2) a movement control group

71 with the recommendation of sensorimotor, stabilization and flexibility exercises;

72 and 3) a functional optimization group with the recommendation of work- or sport-

73 specific tasks, strengthening, conditioning, aerobic and general fitness

76 is expected to benefit most from Pilates. As Pilates is also focused on

77 strengthening exercises, it could potentially provide benefit for the functional

78 optimization group. However, the approach of this subgroup is focused on work-

79 and sport-specific tasks, conditioning, aerobic and general fitness exercises,

80 which Pilates is not.

83 these interventions has yet to be tested in a randomized controlled trial. Thus,

compared to an educational booklet.

90 Study design and setting

91 This study is a secondary analysis using data from a randomised

95 Committee of Universidade Cidade de São Paulo and the study was

96 prospectively registered at ClinicalTrials.gov (NCT02241538).

97 The randomized controlled trial assessed the effectiveness and cost-

98 effectiveness of the addition of different doses of Pilates to advice in the treatment