Musculoskeletal Science and Practice 47 (2020) 102139

Contents lists available at ScienceDirect

Musculoskeletal Science and Practice

journal homepage: www.elsevier.com/locate/msksp

Original article

The role of pain and disability changes after physiotherapy treatment on

global perception of improvement in patients with chronic low back pain
Diogo Andr�e da Fonseca Pires a, b, *, Eduardo Jos�e Carvalho Brazete Cruz b, c,
Helena Cristina de Matos Canha ~o b, d, e, Carla do Rosa
�rio Delgado Nunes a, b
a
NOVA National School of Public Health, Public Health Research Centre, Universidade NOVA de Lisboa, Portugal
b
Comprehensive Health Research Center (CHRC), Portugal
c
Department of Physical Therapy, School of Health Care, Polytechnic Institute of Setúbal, Portugal
d
Cedoc, EpiDoC Unit, NOVA Medical School, Nova University of Lisbon, Lisbon, Portugal
e
Department of Rheumatology, CHULC-Hospital Curry Cabral, Lisbon, Portugal

A R T I C L E I N F O A B S T R A C T

Keywords: Background: The effectiveness of physiotherapy in patients with chronic low back pain is usually measured
Global perception of improvement through changes in pain and disability domains. However, recent research has suggested that these two domains
Chronic low back pain are not sufficient to capture all the physiotherapy benefits when patients’ perspective is considered.
Physiotherapy
Objective: The aim of this study was to investigate the role of pain and disability changes in explaining the global
Pain intensity
Disability
perception of improvement in patients with chronic low back pain undergoing physiotherapy.
Design: Prospective cohort study.
Methods: The study was conducted on183 patients who were referred to physiotherapy treatment due to low back
pain lasting more than 12 weeks. Sociodemographic and clinical characteristics were measured at baseline,
together with pain intensity and disability. Eight (post-intervention) and twelve weeks later, global perception of
improvement was measured together with pain and disability. The Pearson correlation coefficient and linear
regression models were used for analyses.
Results: Of the 183 participants included, 144 completed the 12-weeks follow-up. Significant and moderate
correlation was found between pain and disability changes and the global perception of improvement after
intervention and at the 12-weeks follow-up. Pain and disability changes explained 20.7%–36.3% of the variance
in the global perception of improvement.
Conclusions: Pain and disability changes are related and contributed to explaining a partial proportion of variance
in the global perception of improvement. The findings suggest that these domains are not sufficient to explain
and measure all of the benefits of physiotherapy when patients’ global perception of improvement is considered.

1. Introduction Care Excellence, 2016) and the evidence on this topic has accompanied a
global effort of all health research.
Chronic low back pain (CLBP) is one of the most prevalent health An important part of an effective research process is the selection of
conditions worldwide (Meucci et al., 2015). In addition to the impact at appropriate outcome domains (Williamson et al., 2012). This is critical
the individual level, the large economic and social costs related to CLBP to compare and quantify the benefits (or adverse effects) associated with
are well documented in the literature (Gouveia et al., 2016; March et al, the applied interventions as well as to promote evidence-informed
2014; Parthan et al., 2006). Subsequently, an increased research effort practice (Gargon et al., 2014). Physiotherapy studies on CLBP
has been observed in order to understand the associated factors and the frequently measure pain and disability domains (Pires et al., 2020). A
most effective interventions for this health condition (Wand and recent systematic review identified the outcome domains reported in
O’Connell, 2008). Physiotherapy modalities are recommended for pa­ 195 randomized controlled trials examining physiotherapy in­
tients with CLBP (Foster et al., 2018; National Institute for Health and terventions for patients with CLBP (Pires et al., 2020). This review found

* Corresponding author. Centro de Investigaç~ao em Saúde Pública, Escola Nacional de Saúde Pública, Avenida Padre Cruz, Lisbon, 1600-560, Portugal.
E-mail addresses: piresdiogo.af@gmail.com, da.pires@ensp.unl.pt (D.A.F. Pires).

https://doi.org/10.1016/j.msksp.2020.102139
Received 25 November 2019; Received in revised form 6 February 2020; Accepted 25 February 2020
Available online 28 February 2020
2468-7812/© 2020 Elsevier Ltd. All rights reserved.
D.A.F. Pires et al. Musculoskeletal Science and Practice 47 (2020) 102139

that pain intensity and disability were used in 85% of the included 2.3. Intervention
studies and were the most frequent primary outcomes. Others common
domains such as spine mobility and health-related quality of life were All participants received usual care in physiotherapy, without any
used in less than 30% of the studies (Pires et al., 2020). Therefore, pain restriction from the research team. Although a comprehensive definition
and disability are the only domains used in a large proportion of studies. of physiotherapy treatments was provided (manual therapy; therapeutic
Despite the dominance of pain and disability on outcomes assess­ education; therapeutic exercise; electrotherapy and physical agents), the
ment of patients with CLBP, preliminary evidence from qualitative type of treatments used and the number of sessions were the re­
studies supports that they are not sufficient when patients’ perspective is sponsibility of physiotherapists. Usual care was chosen in order to reflect
considered. Hush et al. (2009) developed a qualitative study with 36 current practice and promote the variability of the treatments applied.
patients that recovered or did not recover from low back pain. The au­ Therefore, the characteristics of the intervention applied were not under
thors found a discrepancy between pain and disability scores and the analysis and it was assumed that the variability of treatments washed
perspective of self-reported recovery by patients suggesting the rele­ out specific treatment modifier effects (Kent et al., 2010). The duration
vance of others domains (Hush et al., 2009). More recently, others of the intervention, the number of participants who did not complete the
studies have reinforced this view supporting a potential discord between intervention and their reasons were recorded.
outcome domains based on the health professional perspective and the
effects of intervention that are meaningful to the patients (Gardner et al., 2.4. Instruments
2015; Sanderson et al., 2010).
Consequently, patients’ global perception of improvement measures Participants were assessed at baseline, 8 weeks after the beginning of
have been progressively introduced into chronic pain research (Dworkin the intervention (or earlier if they were discharged) and at the 12-week
et al., 2005). These measures provide reliable information about pa­ follow-up. At baseline, participants were asked to complete a socio­
tients’ perspective on the intervention’s benefits helping to interpret if demographic and clinical questionnaire (see Table 1) along with the
changes in specific outcome domains such as pain or disability were Numeric Pain Rating Scale (NPRS) and the Quebec Back Pain Disability
meaningful to patients (Dworkin et al., 2008; Kamper et al., 2010b). At Scale (QBPDS-PT) (Cruz et al., 2013). The Global Perceived Effect Scale
this point, there is a lack of quantitative studies analysing the relation­ (GPES-PT) (Freitas et al., 2019) was used to assess the patients’ global
ship between pain and disability changes with patients’ global percep­ perception of improvement. Eight and twelve weeks later, GPES-PT was
tions of improvement. This knowledge may contribute to clarifying the applied together with NPRS and QBPDS-PT. GPES-PT is a transition
extent to which the pain and disability domains are sufficient (or not) to scale ranging from 5 (“vastly worse”) to þ5 (“completely recovered”).
analyse the effectiveness of physiotherapy considering the patients’ This measure was previously translated and cross-culturally adapted in
perspective of improvement. The aim of this study was to analyse the European Portuguese showing adequate convergent validity (r ¼ 0.677),
role of pain and disability changes in explaining global perception of test-retest reliability (ICC ¼ 0.758), and responsiveness (Areas under the
improvement in patients with chronic low back pain undergoing curve values of 0.71 and 0.83) (Freitas et al., 2019). To assess the
physiotherapy. average level of pain intensity, the NPRS was used. The NPRS is an
11-point self-report measure (0–10) with the labels “no pain” and “worst
2. Methods imaginable pain” on the ends that has proven to be valid and reliable in
patients with chronic pain (Farrar et al., 2001). Functional disability was
2.1. Study design and setting assessed using QBPDS-PT. This questionnaire consists of 20 items rep­
resenting functional activities with 6 response categories each (0- “not
A prospective cohort study with a 12-week follow-up was under­ difficult at all” to 5- “unable to do”). The total score is calculated by a
taken in patients seeking physiotherapy treatments for nonspecific summation of the scores for each individual item ranging from 0 (no
CLBP. Between October 2015 and December 2018, potential partici­ disability) to 100 (severe disability). The QBPDS-PT has shown good
pants were identified and recruited consecutively from 20 different validity (ρ ¼ 0.62), test-retest reliability (ICC ¼ 0.696) and internal
outpatient clinics in Portugal. A standardized protocol was followed by
local physicians and/or physiotherapists in the recruitment process. The Table 1
minimum sample size required was previously established using formula Baseline characteristics of the participants.
N > 50 þ 8 m (where m is the number of independent variables) (Green,
Variables Total Sample (n ¼ 183)
1991). In this study, eight clinical and demographic variables were
Agea 48.02 � 10.53
considered in addition to pain and disability changes. A potential loss of
BMI (kg/m2)a 26.18 � 4.28
20% of participants during the study was also considered. Accordingly, a Gender [N (%)]
minimum of 156 participants was required. All participants received Male 36 (19.7%)
oral and written information about the study and provided their Female 147 (80.3%)
informed consent prior to participating. Educational level [n (%)]
Primary/Basic education 74 (40.4%)
High school/
2.2. Participants College 109 (59.6%)
Working status [n (%)]
Inclusion criteria for this study were: patients aged 18 to 65 with Employed 152 (83.1%)
nonspecific low back pain with at least 12 weeks duration (Airaksinen Not Active 31 (16.9%)
Duration of pain [n (%)]
et al., 2006), with or without leg pain, pain intensity �3 (measured by 3–24 months 58 (31.7%)
the Numeric Pain Rating Scale) on the day of the initial evaluation, and >24 months 125 (68.3%)
literate in Portuguese. Patients with clinical signs of serious or specific Pain Irradiation [n (%)]
pathologies (inflammatory disorder, fracture, radicular syndrome) Yes 121 (66.1%)
No 61 (33.3%)
(Smeets et al., 2006), pregnancy, and history of back surgery or con­
Medication [n (%)]
servative treatment in the prior 12 and 3 months, respectively, were Yes 85 (46.4%)
excluded. Eligibility was checked by local physiotherapists. No 98 (53.6%)
Pain Intensity (0–10 NPRS)a 5.86 � 1.88
Disability (0–100 QBPDS)a 36.54 � 17.78
a
(mean � SD).

2
D.A.F. Pires et al. Musculoskeletal Science and Practice 47 (2020) 102139

consistency (Cronbach’s α ¼ 0.95) (Cruz et al., 2013). into the regression equation. Second, the same was performed for the
absolute changes in disability (without the presence of absolute changes
in pain). Third, the absolute changes in pain and disability were entered
2.5. Data analysis together into the regression equation. To quantify the variance in the
dependent variable (GPES) attributable to the pain and disability vari­
Descriptive statistics was used to summarize participants’ charac­ ance, the R2 was recorded in each step of the analysis. In addition, the
teristics at baseline. To compare the characteristics between participants relative importance of predictors was used to understand the contribu­
who completed and did not completed the study, the chi-square test or tion of each independent variable in the regression equation (Toni­
Mann-Whitney U-test were used. All variables were assessed for dandel and LeBreton, 2011). The same analysis was repeated for the
normality and outliers. Absolute change in pain and disability were percentage changes after the intervention and at the 12-week follow-up.
computed by subtracting baseline scores from post-intervention and 12 Variance inflation factors (VIF) were used to check multicollinearity.
weeks follow-up scores. Thus, positive changes indicated improvement, VIFs greater than 10 were considered indicative of serious multi­
while negative changes indicated worsening. Based on absolute changes collinearity problems. Data analysis was performed using SPSS (version
and baseline scores, the percentage change was calculated. Pearson 24.0; IBM, Chicago, IL). A significance level of 0.05 was chosen for this
correlation coefficient was performed to quantify the association be­ study.
tween the GPES scores and pain and disability changes. The correlation
coefficients were interpreted as follows: r < 0.10, no association; r ¼ 3. Results
0.10–0.39, weak; r ¼ 0.40–0.69, moderate; r ¼ 0.70–0.89, strong; and r
¼ 0.90–1.0, very strong (Schober and Schwarte, 2018). 3.1. Participants
Linear regression models were preformed to investigate the associ­
ation of changes in pain and disability (independent variables) in rela­ Of the 235 potential participants identified, 183 participants with
tion to the GPES scores (dependent variable). Previously, a univariate CLBP were considered eligible and accepted to participate in this study.
linear regression was conducted between each baseline variable (clinical Of those, 173 (94.5%) completed the physiotherapy treatment and 144
and sociodemographic characteristics) and the dependent variable. (78.7%) completed the follow-up at 12 weeks. A study flowchart and
Baseline variables with a p value � 0.2 were then entered in all multi­ reasons for dropouts are described in Fig. 1. Table 1 describes the
variate regression models as covariates. After that, multivariate models characteristics of all participants assessed at baseline. Participants who
were performed (method: forward stepwise) according to a predefined failed follow-up evaluations differ from those that did not fail in pain
sequence of steps. First, the absolute changes in pain were entered alone

Potentially eligible participants

n=235

Not meeting inclusion criteria

(n=51)
Declined to participate (n=1)

Participants included at baseline

n=183

Incompatible schedules (n=6)

Temporary illness (n=1)
LBP worsening (n=1)
Others reasons (n=2)
Participants completing the
intervention
n= 173

Unable to contact (n= 39)

12-weeks follow-up
n= 144

Fig. 1. Study flow-chart.

3
D.A.F. Pires et al. Musculoskeletal Science and Practice 47 (2020) 102139

Table 2
Mean scores of GPES and changes from baseline in pain intensity and disability.
GPES Scores Absolute Changes Percentage Changes

Pain Disability Pain Disability

Post-intervention 3.02 � 1.30 2.56 � 2.48 13.63 � 16.90 41.73 � 45.82 36.03 � 38.99
12-Weeks follow-up 2.71 � 1.74 2.04 � 2.58 10.68 � 17.41 33.72 � 46.91 24.87 � 50.69

Table 3 Table 5
Correlation between GPES scores and changes in pain and disability. Linear regression results examining contributions of 12-Weeks follow-up
Absolute Changes Percentage Changes
changes in pain and disability to GPES scores.
Independent Variables Adjust Predictors relative p
Pain Disability Pain Disability
entered in the model R2 importance
Post-intervention GPES scores 0.44 0.51 0.48 0.55
Using Absolute Changes
12-Weeks follow-up GPES scores 0.27 0.41 0.28 0.39
1st Δ Pain 0.144 1.0 0.001
*All correlations were significant (p < 0.01). step Δ Paina 0.145 0.95 0.001
2nd Δ Disability 0.179 1.0 0.001
step Δ Disabilitya 0.179 0.97 0.001
intensity (p ¼ 0.009), irradiated pain (p ¼ 0.042) and medication use (p 3rd Δ Pain 0.207 0.33 0.015
¼ 0.033). Significant differences in other variables evaluated at the step Δ Disability 0.67 0.001
baseline were not found. The mean scores for changes in pain and Δ Pain 0.207 0.33 0.015
disability measures as well as the GPES mean scores after the inter­ Δ Disabilitya 0.67 0.001
Using Percentage Changes
vention and at the 12-week follow up are presented in Table 2. 1st Δ Pain 0.160 1.0 0.001
step Δ Paina 0.160 1.0 0.001
3.2. Correlations analysis 2nd Δ Disability 0.184 1.0 0.001
step Δ Disabilitya 0.184 1.0 0.001
3rd Δ Pain 0.222 0.39 0.006
Pearson correlations between the GPES scores and improvements in step Δ Disability 0.61 0.001
pain and disability are outlined in Table 3. All correlations were statis­ Δ Pain 0.222 0.39 0.006
tically significant (p < 0.001), weak to moderate (r ranging from 0.27 to Δ Disabilitya 0.61 0.001
0.55) and positive. a
Adjusted to baseline variables (Age).

3.3. Linear regression analysis less than 2 in all models indicating an absence of multicollinearity
problems.
The explained variance for the GPES scores in the full regression
model ranged between 14.4% and 37.4% (Tables 4 and 5). A maximal 4. Discussion
explained variance was obtained when percentage changes in pain and
disability were analysed together. Partial regression models with pain or In the present study, the role of pain and disability changes in ac­
disability separately, suggest a greater contribution of disability changes counting for the global perception of improvement in patients with CLBP
to the GPES when compared to pain intensity improvement (Tables 4 undergoing physiotherapy was analysed. The main findings are that the
and 5). The same results were observed for the relative importance pain intensity and disability changes during the intervention demon­
values. Regression models after the intervention and including change strated a modest contribution to the GPES scores. Despite the indepen­
scores in percentage obtained higher values of variance. The VIFs were dent and significant associations, pain intensity and disability changes
accounted for a small proportion of total GPES variance even when
Table 4 considered together. In addition, correlations between pain and
Linear regression results examining contributions of post-intervention changes disability changes with the GPES scores were mostly close to or less than
in pain and disability to GPES scores. 0.5. These data suggest that these domains represent different constructs
Independent Variables Adjust Predictors relative p and greater changes in pain and disability may not necessarily mean
entered in the model R2 importance higher levels of perceived improvement.
Using Absolute Changes Changes in pain and disability explained up to 36.3% of the variance
1st Δ Pain 0.194 1.0 0.001 in the GPES scores in this study. Although a complete accounting of the
step Δ Paina 0.219 0.88 0.001 global perception of improvement is not expected, a large proportion of
2nd Δ Disability 0.253 1.0 0.001 variance remains unexplained. The role of other benefits of in­
step Δ Disabilitya 0.264 0.94 0.001
terventions, not assessed in this study, may help to explain these find­
3rd Δ Pain 0.306 0.33 0.001
step Δ Disability 0.67 0.001 ings. Previous studies that included chronic pain patients have
Δ Pain 0.322 0.33 0.001 demonstrated significant associations of changes in other domains such
Δ Disabilitya 0.57 0.001 as fatigue (Hudson et al., 2009), sleep (Geisser et al., 2010), work
Using Percentage Changes
(Hudson et al., 2009), depression (Geisser et al., 2010; Scott and
1st Δ Pain 0.225 1.0 0.001
step Δ Paina 0.245 0.90 0.001
McCracken, 2015) or social function (Scott and McCracken, 2015) with
2nd Δ Disability 0.293 1.0 0.001 the global perception of improvement. Scott and McCracken (2015)
step Δ Disabilitya 0.301 0.96 0.001 reported that perceived changes in pain, mood and physical, social, and
3rd Δ Pain 0.363 0.34 0.001 work-related activities explained 64% of the variance in Patient Global
step Δ Disability 0.66 0.001
Impression of Change. These values of variance are substantially higher
Δ Pain 0.374 0.34 0.001
Δ Disabilitya 0.59 0.001 than those found in this study. Qualitative studies have also reported
a
similar findings. Hush et al. (2009) reported that perception of recovery
Adjusted to baseline variables (BMI; working status; educational level).

4
D.A.F. Pires et al. Musculoskeletal Science and Practice 47 (2020) 102139

is mediated by patients’ appraisal of their function and pain intensity but different authors have questioned the patients’ ability to accurately
they are not a reliable indicator of recovery. Accordingly, other authors consider their previous health state when they evaluate the change after
have suggested that global perception of improvement may incorporate long periods of time (Kamper et al., 2010b, 2009). Therefore, the GPES
a variety of other domains such as self-efficacy, self-esteem, spontaneity scores at the 12-week follow-up could have been influenced by current
or “feeling positive emotions” (Evans et al., 2014; Walton, 2013). This pain and disability rather than by the changes that have occurred since
set of domains fits in the mental health area and can be particularly the beginning of the intervention. However, this is not the only hy­
important in patients’ adaptation and readjustment to the health con­ pothesis to consider. Over time, patients tend to adapt to the pain con­
dition. This point of view and the way these intervention benefits seem dition and adjust their life to minimize its impact. Consequently, other
to contribute to the perception of improvement have been described in health domains can be valued after the intervention, modifying the
previous studies (Beaton et al., 2001; Walton, 2013). Therefore, there importance attributed by patients to pain and disability changes and
appears to be reason to anticipate that the evaluation of other variables thus their contribution to the GPES scores. This hypothesis has been
potentially modifiable by physiotherapy interventions could contribute argued in other studies (Beaton et al., 2001; Rampakakis et al., 2015).
to a better understanding of the GPES scores. The effectiveness of the interventions is usually analysed and inter­
Pain is the most common symptom of CLBP and an important base­ preted considering the absolute values or absolute changes occurring in
line predictor for the intervention success (Cecchi et al., 2014; Verkerk the various outcome domains. In this study, the analyses using per­
et al., 2013). Interestingly, a greater association and contribution of centage changes showing higher values of correlation coefficients and
disability changes to GPES scores in relation to pain intensity changes variance explained the GPES scores. The fact that the percentage
was consistently identified in this study. The highest relative importance changes take into account baseline scores seems to justify these findings
values for disability changes (ranging from 0.61 to 0.67) compared to (Dworkin et al., 2008; Ostelo et al., 2008). For example, 30 points of
those observed for changes in pain intensity (ranging from 0.33 to 0.39) pain reduction may represent a complete recovery but also an unsatis­
also supports this assumption. Several reasons for these findings can be factory change for patients with high baseline levels of pain. Therefore,
discussed. Firstly, chronic pain patients may not expect a complete the analysis of percentage changes should be considered in clinical trials
resolution of their pain condition (Evans et al., 2014). Thus, the in order to improve the interpretation of the intervention results.
importance attributed by patients to improvements in other variables The use of pain intensity and disability to assess the effectiveness of
such as disability may be greater than improvements in pain. The sec­ physiotherapy in patients with CLBP has prevailed in recent physio­
ondary role of pain in patients with CLBP was also demonstrated in the therapy trials (Pires et al., 2020). These two domains seem to be
study conducted by Kamper et al. (2010a). They found that pain im­ considered by physiotherapists and researchers as the most important
provements accurately identified patients with acute low back pain who and other outcome domains are rarely used (Pires et al., 2020). The data
perceive a complete recovery. However, the odd ratios values repre­ from this study suggest that the perspective of patients with CLBP and
senting this relationship were substantially lower in patients with CLBP researchers may not be completely aligned and some potential benefits
(Kamper et al., 2010a). Secondly, recent studies have demonstrated of physiotherapy beyond pain and disability are not being measured.
early pain changes in the intervention and its predictive value for suc­ This underrepresentation of patient-centred domains can reduce the
cess in others variables after the intervention (Cook et al., 2017; Mansell validity of the outcome measurement process and should be addressed
et al., 2017). Therefore, pain changes may act as facilitators for the in future studies. Understanding the perceived benefits by patients with
disability changes that are more easily perceived as important by the CLBP after physiotherapy can be an important step towards a suitable
patients after the intervention. Finally, the type of instrument used to and valid outcome evaluation.
measure pain (unidimensional scale) and disability (multi-item scale) The knowledge about the importance of pain and disability changes
may not be irrelevant to the observed results. Considering the to patients with CLBP undergoing physiotherapy is limited. This study
complexity of pain experience, limited information can be captured addresses this gap and the findings may have important implications for
using a single and unidimensional measure such as the NPRS. In the way the effectiveness of physiotherapy is measured. Together with
contrast, the QBPDS includes a variety of functional activities more the appropriate sample size and variability of recruitment settings
easily understood by the patients and representatives of their daily re­ (external validity), these were the main strengths of this study. How­
strictions due to pain. This may also help to explain why disability is ever, some limitations need to be considered. First, a significant pro­
better related to global perception of improvement. portion of participants failed to attend the 12-week follow-up
In addition to the reasons described above, the role of the inter­ evaluation. A higher proportion of these participants reported irradiated
vention adopted in this study should not be underestimated. The influ­ pain and medication use as well as higher levels of pain intensity at
ence of the type of intervention and its goals in the domains with the baseline when compared to participants who completed the study. For
greatest contribution to the global perception of improvement scales has this reason, the relationship between unavailability to complete the
been reported. Geisser et al. (2010) found that pain changes were the study and worse outcomes after the intervention cannot be excluded.
main contributor when fibromyalgia patients assessed the perceived Second, more than 80% of the participants were women. Although the
benefits with pharmacological treatment. In contrast, changes in mood, reasons for this difference are unknown, the findings of this study must
acceptance and daily functioning were the most important variables be interpreted in the light of this limitation. Third, physiotherapy out­
after a psychological programme in the study with chronic pain patients comes (or other intervention) are influenced by multiple contextual
developed by Scott and McCracken (2015). In this study, disability factors and other sources of bias (e.g. natural course of LBP; patient’s
changes present more relevance than pain for the GPES scores suggest­ expectations) that cannot be controlled through an observational study
ing that physiotherapy modalities may have a particular impact on design (Testa and Rossettini, 2016). Future studies using more robust
reducing disability perceived by patients. Furthermore, different in­ analyses (e.g. mediation analysis) and experimental designs must be
terventions presented different potential benefits and so the contribu­ conducted to confirm our findings (Mansell et al., 2014). Finally, the
tion of specific domains to the GPES scores can diverge according to the discussion of the results was based on studies using different samples
interventions applied. Looking at the results of this study, they suggest and interventions than those used in this study. The interpretation of
that patients with CLBP undergoing physiotherapy perceive benefits in results should be carefully considered and further investigation
other domains beyond pain and disability. including patients with CLBP undergoing physiotherapy should be
Overall, the strength of the associations and the contribution of the conducted to confirm our findings.
independent variables to the GPES scores decreased at the 12-week
follow-up. Based on previous studies, these results could be due to
recall bias associated to the GPES scale (Kamper et al., 2010b). In fact,

5
D.A.F. Pires et al. Musculoskeletal Science and Practice 47 (2020) 102139

5. Conclusion Green, S.B., 1991. How many subjects does it take to do A regression analysis?
Multivariate Behav. Res. 26 (3), 499–510.
Hudson, J.I., Arnold, L.M., Bradley, L.A., Choy, E.H.S., Mease, P.J., Wang, F., et al., 2009.
This study aimed to clarify the role of pain and disability changes to What makes patients with fibromyalgia feel better? Correlations between patient
patients’ perceptions about their improvements after a physiotherapy global impression of improvement and changes in clinical symptoms and function: a
programme. Changes in these domains were significantly related and pooled analysis of 4 randomized placebo-controlled trials of duloxetine.
J. Rheumatol. 36 (11), 2517–2522.
contributed to explaining a partial proportion of variance in patients’ Hush, J.M., Refshauge, K., Sullivan, G., De Souza, L., Maher, C.G., McAuley, J.H., 2009.
perception of improvement. However, these findings suggested that pain Recovery: what does this mean to patients with low back pain? Arthritis Care Res.
and disability may not be the only potential benefits of physiotherapy 61, 124–131.
Kamper, S.J., Maher, C.G., Herbert, R.D., Hancock, M.J., Hush, J.M., Smeets, R.J., 2010a.
perceived by patients with CLBP. The relevance of assessing other How little pain and disability do patients with low back pain have to experience to
outcome domains was reinforced and should be addressed in future feel that they have recovered? Eur. Spine J. 19, 1495–1501.
studies. Kamper, S.J., Maher, C.G., Mackay, G., 2009. Global rating of change scales: a review of
strengths and weaknesses and considerations for design. J. Man. Manip. Ther. 17,
163–170.
Funders Kamper, S.J., Ostelo, R.W.J.G., Knol, D.L., Maher, C.G., de Vet, H.C.W., Hancock, M.J.,
2010b. Global Perceived Effect scales provided reliable assessments of health
transition in people with musculoskeletal disorders, but ratings are strongly
DP is supported by Fundaça ~o para a Ci^encia e a Tecnologia – the
influenced by current status. J. Clin. Epidemiol. 63.
Portuguese national funding agency for science, research and technol­ Kent, P., Keating, J.L., Leboeuf-Yde, C., 2010. Research methods for subgrouping low
ogy (SFRH/BD/133638/2017). back pain. BMC Med. Res. Methodol. 10 (62).
Mansell, G., Hill, J.C., Kamper, S.J., Kent, P., Main, C., Van Der Windt, D.A., 2014. How
can we design low back pain intervention studies to better explain the effects of
References treatment? Spine (Phila Pa 1976) 39 (5), 305–310.
Mansell, G., Jordan, K.P., Peat, G.M., Dunn, K.M., Lasserson, D., Kuijpers, T., et al., 2017.
Airaksinen, O., Brox, J.I., Cedraschi, C., Hildebrandt, J., Klaber-Moffett, J., Kovacs, F., Brief pain re-assessment provided more accurate prognosis than baseline
et al., 2006. Chapter 4: European guidelines for the management of chronic information for low-back or shoulder pain. BMC Muscoskel. Disord. 18 (139).
nonspecific low back pain. Eur. Spine J. 15. March, L., Smith, E.U.R., Hoy, D.G., Cross, M.J., Sanchez-Riera, L., Blyth, F., et al., 2014.
Beaton, D.E., Tarasuk, V., Katz, J.N., Wright, J.G., Bombardier, C., 2001. “Are you Burden of disability due to musculoskeletal (MSK) disorders. Best Pract. Res. Clin.
better?” A qualitative study of the meaning of recovery. Arthritis Rheum. 45, Rheumatol. 28, 353–366.
270–279. Meucci, R.D., Fassa, A.G., Xavier Faria, N.M., 2015. Prevalence of chronic low back pain:
Cecchi, F., Pasquini, G., Paperini, A., Boni, R., Castagnoli, C., Pistritto, S., et al., 2014. systematic review. Rev. Saude Publica 49.
Predictors of response to exercise therapy for chronic low back pain: result of a National Institute for Health and Care Excellence, 2016. Low Back Pain and Sciatica in
prospective study with one year follow-up. Eur. J. Phys. Rehabil. Med. 50 (2), over 16s: Assessment and Management.
143–151. Ostelo, R.W.J.G., Deyo, R.A., Stratford, P., Waddell, G., Croft, P., Korff, M Von, et al.,
Cook, C., Petersen, S., Donaldson, M., Wilhelm, M., Learman, K., 2017. Does early change 2008. Interpreting change scores for pain and functional status in low back pain.
predict long-term (6 months) improvements in subjects who receive manual therapy Spine (Phila Pa 1976) 33, 90–94.
for low back pain? Physiother. Theor. Pract. 33 (9), 716–724. Parthan, A., Evans, C.J., Le, K., 2006. Chronic low back pain: epidemiology, economic
Cruz, E.B., Fernandes, R., Carnide, F., Vieira, A., Moniz, S., Nunes, F., 2013. Cross- burden and patient-reported outcomes in the USA. Expert Rev. Pharmacoecon.
cultural adaptation and validation of the Quebec back pain disability scale to Outcomes Res. 6, 359–369.
European Portuguese language. Spine (Phila Pa 1976) 38, E1491–E1497. Pires, D., Cruz, E.B., Gomes, L.A., Nunes, C., 2020. How do physical therapists measure
Dworkin, R.H., Turk, D.C., Farrar, J.T., Haythornthwaite, J.A., Jensen, M.P., Katz, N.P., treatment outcomes in adults with chronic low back pain? A systematic review.
et al., 2005. Core outcome measures for chronic pain clinical trials: IMMPACT Phys. Ther. https://doi.org/10.1093/ptj/pzaa030 (accept).
recommendations. Pain 113 (1–2), 9–19. Rampakakis, E., Ste-Marie, P.A., Sampalis, J.S., Karellis, A., Shir, Y., Fitzcharles, M.A.,
Dworkin, R.H., Turk, D.C., Wyrwich, K.W., Beaton, D., Cleeland, C.S., Farrar, J.T., et al., 2015. Real-life assessment of the validity of patient global impression of change in
2008. Interpreting the clinical importance of treatment outcomes in chronic pain fibromyalgia. RMD Open 1, e000146.
clinical trials: IMMPACT recommendations. J. Pain 9, 105–121. Sanderson, T., Morris, M., Calnan, M., Richards, P., Hewlett, S., 2010. What outcomes
Evans, R., Bronfort, G., Maiers, M., Schulz, C., Hartvigsen, J., 2014. “I know it’s from pharmacologic treatments are important to people with rheumatoid arthritis?
changed”: a mixed-methods study of the meaning of Global Perceived Effect in Creating the basis of a patient core set. Arthritis Care Res. 62, 640–646.
chronic neck pain patients. Eur. Spine J. 23, 888–897. Schober, P., Schwarte, L.A., 2018. Correlation coefficients: appropriate use and
Farrar, J.T., Young Jr., J.P., LaMoreaux, L., Werth, J.L., Poole, R.M., 2001. Clinical interpretation. Anesth. Analg. 126 (5), 1763–1768.
importance of changes in chronic pain intensity measured on an 11-point numerical Scott, W., McCracken, L.M., 2015. Patients’ impression of change following treatment for
pain rating scale. Pain 94, 149–158. chronic pain: global, specific, a single dimension, or many? J. Pain 16, 518–526.
Foster, N.E., Anema, J.R., Cherkin, D., Chou, R., Cohen Steven, P., Gross, D.P., et al., Smeets, R.J.E.M., Vlaeyen, J.W.S., Hidding, A., Kester, A.D.M., Van Der Heijden, G.J.M.
2018. Prevention and treatment of low back pain: evidence, challenges, and G., Van Geel, A.C.M., et al., 2006. Active rehabilitation for chronic low back pain:
promising directions. Lancet (London, England) 391 (10137), 2368–2383. cognitive-behavioral, physical, or both? First direct post-treatment results from a
Freitas, P., Pires, D., Nunes, C., Cruz, E., 2019. Cross-cultural adaptation and randomized controlled trial [ISRCTN22714229]. BMC Muscoskel. Disord. 7 (5).
psychometric properties of the European Portuguese version of the global perceived Testa, M., Rossettini, G., 2016. Enhance placebo, avoid nocebo: how contextual factors
effect scale in patients with chronic low back pain. Disabil. Rehabil. 1–7. In press. affect physiotherapy outcomes. Man. Ther. 24, 65–74.
Gardner, T., Refshauge, K., McAuley, J., Goodall, S., Hübscher, M., Smith, L., 2015. Tonidandel, S., LeBreton, J.M., 2011. Relative importance analysis: a useful supplement
Patient led goal setting in chronic low back pain-What goals are important to the to regression analysis. J. Bus. Psychol. 26, 1–9.
patient and are they aligned to what we measure? Patient Educ. Counsel. 98 (8), Verkerk, K., Luijsterburg, P.A.J., Heymans, M.W., Ronchetti, I., Pool-Goudzwaard, A.L.,
1035–1038. Miedema, H.S., et al., 2013. Prognosis and course of disability in patients with
Gargon, E., Gurung, B., Medley, N., Altman, D.G., Blazeby, J.M., Clarke, M., et al., 2014. chronic nonspecific low back pain: a 5- and 12-month follow-up cohort study. Phys.
Choosing important health outcomes for comparative effectiveness research: a Ther. 93 (12), 1603–1614.
systematic review. PloS One 9, 12. Walton, D.M., 2013. What does ‘recovery’ mean to people with neck pain? Results of a
Geisser, M.E., Clauw, D.J., Strand, V., Gendreau, R.M., Palmer, R., Williams, D.A., 2010. descriptive thematic analysis. Open Orthop. J. 7, 420–427.
Contributions of change in clinical status parameters to Patient Global Impression of Wand, B.M., O’Connell, N.E., 2008. Chronic non-specific low back pain - sub-groups or a
Change (PGIC) scores among persons with fibromyalgia treated with milnacipran. single mechanism? BMC Muscoskel. Disord. 9 (11).
Pain 149, 373–378. Williamson, P.R., Altman, D.G., Blazeby, J.M., Clarke, M., Devane, D., Gargon, E., et al.,
Gouveia, N., Rodrigues, A., Eus� ebio, M., Ramiro, S., Machado, P., Canh~ ao, H., et al., 2012. Developing core outcome sets for clinical trials: issues to consider. Trials 13.
2016. Prevalence and social burden of active chronic low back pain in the adult
Portuguese population: results from a national survey. Rheumatol. Int. 36, 183–197.