MINIMAL CLINICALLY IMPORTANT CHANGE OF MOVEMENT PAIN IN MUSCULOSKELETAL PAIN CONDITIONS

Abstract

Movement pain, which is distinct from resting pain, is frequently reported by individuals with musculoskeletal pain. There is growing interest in measuring movement pain as a primary outcome in clinical trials, but no minimally clinically important change (MCIC) has been established, limiting interpretations. We analyzed data from 315 participants who participated in previous clinical trials (65 with chronic Achilles tendinopathy; 250 with fibromyalgia) to establish an MCIC for movement pain. A composite movement pain score was defined as the average pain (Numeric Rating Scale (NRS): 0 to 10) during two clinically relevant activities. The change in movement pain was calculated as the change in movement pain from pre-intervention to post-intervention. A Global Scale (GS: 1 to 7) was completed after the intervention on perceived change in health status. Participants were dichotomized into non-responders (GS ≥4) and responders (GS <3). Receiver operating characteristic (ROC) curves were calculated to determine threshold values and corresponding sensitivity and specificity. We used the Euclidean method to determine the optimal threshold point of the ROC curve to determine the MCIC. The MCIC for raw change in movement pain was 1.1 (95% CI: 0.9–1.6) with a sensitivity of 0.83 (95% CI: 0.75– 0.92) and specificity of 0.79 (95% CI: 0.72–0.86). For percent change in movement pain the MCIC was 27% (95%CI: 10%- 44%) with a sensitivity of 0.79 (95% CI: 0.70 – 0.88) and a specificity of 0.82 (95% CI: 0.72 – 0.90). Establishing an MCIC for movement pain will improve interpretations in clinical practice and research.

INTRODUCTION

Movement pain is the experience of pain during movement, such as activities of daily living, sports, or work-related physical activity²³. Movement pain is one of the most common symptoms in musculoskeletal conditions¹⁵ and contributes to the high healthcare costs associated with musculoskeletal pain¹⁸. In prospective studies, movement pain is associated with decreased function in individuals with knee osteoarthritis³⁵ and has stronger associations with worse disability than resting pain in individuals with both distal radius fractures²¹ and low back pain²⁷. Further distinguishing movement pain from resting pain, we previously showed that nonpharmacological treatments, including transcutaneous electrical nerve stimulation (TENS) as well as exercise plus education, are more effective at reducing movement pain than resting pain^{9, 14}. Thus, movement pain is a conceptually distinct construct from resting pain and an important component of musculoskeletal pain conditions.

There is a growing interest in measuring movement pain in clinical research and practice. Multiple recent and ongoing clinical trials used movement pain as a primary outcome while studying both nonpharmacologic^{9, 14, 31, 32, 37, 40, 56} and pharmacologic therapies⁵¹ in a variety of musculoskeletal pain conditions. In clinical practice, recommendations to consider movement pain along with other measures of pain have been proposed^{2, 5, 13}. In the current paper, we distinguish movement pain, which is measured in real time, as distinct from pain with activity, which relies on recall.^{11, 23} We also distinguish movement pain, as a raw measure of pain intensity during movement, as distinct from movement-evoked pain, which is often relative to resting pain⁴⁷. To build on this emerging area of movement pain research, there is a need to establish a minimal clinically important change (MCIC) among studies that have used a similar definition of movement pain. We have selected two clinical trials of non-pharmacological treatments for movement pain. One trial used TENS for fibromyalgia and the other used exercise and education for Achilles tendinopathy^{9, 14}. Both conditions are chronic musculoskeletal pain disorders with moderate-to-severe movement pain, a high frequency of elevated kinesiophobia,^{8, 49} and a similar magnitude of impairment in quality of life, despite having different pain locations and mechanisms.⁴⁸ Developing a movement pain specific MCIC will allow a more accurate interpretation of alterations in symptoms for clinicians and researchers.

To date, studies that utilize movement pain as a primary outcome measure have relied on metrics for resting pain to interpret if improvements in movement pain are clinically meaningful^{9, 14}. While interpreting clinically relevant changes for resting pain is well established,³⁴ our understanding of the MCIC for movement pain is limited. Another challenge to interpretation of clinical outcomes, is that there is considerable variation in terminology used when assessing minimal important and clinically important differences. The MCIC differentiates individuals who have a clinically meaningful improvement from non-responders²⁶. In contrast, the minimal clinically important difference (MCID) reflects a between group difference in change over time¹⁶. Another distinct concept is the smallest detectable change (SDC), which reflects the magnitude of measurement error in a patient reported outcome. The purpose of the current study is to establish an MCIC and SDC for movement pain to assist clinicians and researchers in improving the interpretation of changes in movement pain in response to treatments.

METHODS

Participants

This study was a secondary analysis of two previously completed randomized controlled trials (RCT): Fibromyalgia Activity Study with Transcutaneous Electrical Nerve Stimulation (FAST; NCT01888640)³⁹ and Tendinopathy Education on Achilles (TEAch; NCT04059146).³³ Both studies evaluated the efficacy of nonpharmacologic treatments for chronic musculoskeletal pain and utilized movement pain as the primary outcome. A total of 315 participants had complete movement pain and global scale data and were included in the current analysis (FAST, n=250; TEAch, n=65). Demographic characteristics and baseline data for resting pain and movement pain are shown in Table 1. In addition to different types of chronic pain, the samples differed in the distribution of female participants. The FAST study included only female participants while the TEAch study included 54.5% females. The participants with fibromyalgia had clinically meaningful differences in duration of symptoms, higher baseline resting pain, lower physical function, higher pain catastrophizing, and higher symptom severity score than those with Achilles tendinopathy. Otherwise, the study samples were similar in terms of demographic and baseline psychological variables.

Table 1.

Demographic and baseline data for the total population and in each study.

Variable	Combined	Fibromyalgia (FAST)	Achilles Tendinopathy (TEAch)
Sex	Female: 286 (91%) Male: 29 (9%)	Female: 250 (100%)	Female: 36 (54.5%) Male: 29 (43.9%)
Age (Years)	46.71 ± 12.71	47.41 ± 11.98	43.6 ± 15.31
BMI (kg/m 2 )	33.45 ± 8.7	34.3 ± 8.71	30.05 ± 7.60
Race	Caucasian: 289 (91.7%)	Caucasian: 230 (92%)	Caucasian: 59 (89.4%)
	Black or African-American: 17 (5.3%)	Black or Africa-American: 13 (5.2%)	Black or African-American: 4 (6.1%)
	Asian: 4 (1.2%)	Asian: 2 (.8%)	Asian: 2 (3.0%)
	American Indian or Alaskan Native: 5 (1.5%)	American Indian or Alaskan Native: 4 (1.5%)	American Indian or Alaskan Native: 1 (1.5%)
	Native Hawaiian or Other Pacific Islander: 1 (0.03%)	Native Hawaiian or Other Pacific Islander: 1 (0.04%)	Native Hawaiian or Other Pacific Islander: 0 (0.0%)
Ethnicity	Hispanic/Latino origin: 15 (4.6%)	Hispanic/Latino Origin: 12 (4.7%)	Hispanic/Latino Origin: 3 (4.5%)
Duration of symptoms (Years)	5.0 (IQR: 2.0–12.0)	7 (3.0–13)	1.5 (0.6–3)
Resting pain (NRS)	5.0 ± 2.3	5.9 ± 1.6	1.8 ± 1.8
Movement pain (NRS)	Aggregate: 5.7 ± 1.9	Aggregate: 6.0 ± 1.8 6MWT: 6.4 ± 1.9 5STS: 5.6 ± 2.2	Aggregate: 4.6 ± 1.8 HR: 5.0 ± 2.1 DF: 4.2 ± 2.3
Improved, Stable, Worsened	Improved: 89 (28.3%) Stable: 219 (69.5%) Worsened: 7 (2.2%)	Improved: 38 (15.2%) Stable: 205 (82%) Worsened: 7 (2.8%)	Improved: 51 (78.5%) Stable: 14 (21.5%) Worsened: 0 (0%)
Depression (PROMIS Depression)	55.2 ± 8.9	57.1 ± 8.2	47.8 ± 7.6
Anxiety (PROMIS Anxiety)	57.5 ± 8.4	58.4 ± 8.1	54 ± 8.8
Physical Function (PROMIS Physical Function)	39.4 ± 6.9	37.1 ± 4.9	48.3 ± 6.8
Pain Catastrophizing (PCS)	19.5 ±12.6	21.8 ± 12.7	10.9 ± 6.9
Kinesiophobia (TSK-17)	37.1 ±7.7	36.9 ± 8.2	37.5 ± 5.7
Symptom Severity Score (SSS)	8.1 ± 2.8	9.2 ± 1.8	4.4 ± 2.6

Demographic data is shown as the number of participants (% of sample). Baseline data are shown as mean ± standard deviation. That’s except for the duration of symptoms which is shown as median (interquartile range). For movement pain, the composite pain rating is the average NRS (0–10) of the two movement pain tasks in each group. 6MWT- Six-minute walk test, 5STS- 5 times sit-to-stand task, DF- Calf stretching task with knee straight and bent NRS- numeric rating scale, HR- Single leg heel raise task, PCS – Pain Catastrophizing Scale, TSK-17 – Tampa Scale of Kinesiophobia, SSS- Symptoms Severity Scale.

Overview of RCT study designs

The findings from two clinical trials were combined because they were both designed to test the efficacy of a non-pharmacological treatment on movement pain, specified a priori as a primary outcome, and captured a global scale score following treatment. Both studies were approved by local institutional review boards and all participants provided written informed consent prior to participating in the initial trial.

FAST study

The FAST study was a dual-site RCT that established the efficacy of TENS for movement pain and fatigue in women with fibromyalgia. Inclusion criteria included resting pain greater than 4/10 on an NRS. The study was conducted over 8 weeks and included 4 in-person visits. Over the first 4-week period, participants were randomized to receive active TENS, placebo TENS, or no TENS. Participants in the Active and Placebo TENS groups were instructed to use the TENS unit for 2 hours per day during physical activity. At baseline and at 4-week follow-up (primary endpoint), participants used TENS units, according to group assignment, while assessing movement pain¹⁴.

TEAch Study

The TEAch study demonstrated the efficacy of exercise and education on movement pain in adults with chronic Achilles tendinopathy pain. The inclusion criteria included movement pain of greater than 3/10 on a NRS and an increase in pain by 1/10 on an NRS during tendon loading activities (e.g., walking, heel raises, or hopping). Participants completed up to 7 treatment sessions with a physical therapist consisting of progressive Achilles tendon loading exercises and education about their condition and symptom management. movement pain was assessed at baseline and at 8 weeks (primary endpoint)⁹. The protocols and primary outcome data for each trial have been previously published in detail^{9, 14, 33, 39}.

Movement Pain Assessment

Both studies measured movement pain as the participant’s reported pain intensity on a numeric rating scale (NRS: 0 to 10) during two activities. This was a measure of pain intensity reported during the activity and did not account for the change from resting pain. A composite movement pain score was used for the current analysis, consistent with previous literature²⁸. The composite movement pain score was calculated as the average of the pain score during two clinically relevant activities that were chosen specifically to be functionally relevant and typically associated with movement pain in each population.

$$\text{Composite}\text{Movement}\text{Pain}\text{Score}\text{=}\frac{\text{Movement}\text{Pain}\text{Activity}1+\text{Movement}\text{Pain}\text{Activity}2}{2}$$

Composite scores of movement pain have better psychometric properties and greater associations with functional outcomes^{27, 28}. Each study used the two selected activities as primary or secondary aims for their primary analysis no further activities were included in our analysis to maintain this consistency. In the FAST study with individuals who had fibromyalgia, these activities were a 6-minute walk test and a 5-time sit-to-stand task. At 5 minutes into the walk and immediately after completing the 5-time sit-to-stand task, participants were asked to rate their pain during the activity from 0 to 10 where 0 is “no pain” and 10 is the “worse pain imaginable.”³³ In the TEAch study with individuals who had Achilles tendinopathy, these activities included a single limb heel raise endurance task and calf stretching task with the knee straight and bent. At the start of the visit, participants were shown a laminated pain scale and were given the instructions, “Throughout the study I will be asking you to rate your pain using a 0 to 10 scale. A 0 is nothing at all, 1–3 is mild, 4–6 is moderate, and 7 to 10 is severe with 10 being the worst pain imaginable.” Immediately following each task, participants were asked to rate their pain during the task on a scale from 0 to 10 in their Achilles tendon on their more painful side. Due to the SARS-CoV-2 pandemic, in the TEAch study the intervention was provided, and movement pain was assessed both in-person and through telehealth, although this did not impact outcomes³⁸. The change in the composite movement pain score was calculated as the raw change or percent change in movement pain from baseline to the primary endpoint of each trial for each participant.

Responder definition using a Global Scale

Each RCT measured participants’ perceived change in health status using a global scale at the trial’s primary endpoint, and this was used as the anchor measure for the current analysis. A measure of participant overall improvement or condition is recommended by population-specific research consensus statements as a core outcome domain for each population^{52, 53}. The FAST study utilized the Global Impression of Change scale (GIC) which is a 7-point scale ranging from “very much improved” (1) to “very much worse” (7)⁴¹. The TEAch trial used the Global Rating of Change scale (GROC), that utilized a 15-point scale anchored from “a very great deal worse” (−7) to “a very great deal better” (+7)⁴. The GROC was transformed into a 7-point scale by collapsing items using the words minimal, moderate, and greatly worse/improved into a single option representing minimal, moderate, or greatly worse/improved. For example, response options “a very great deal better” or +7 and “a great deal better” or +6 on the GROC were combined into a single score of 7 representing greatly improved on a 7-point scale on the GIC. The GROC score was then flipped to align with the GIC where lower scores represent improvements in symptoms. This combined Global Scale score was used to categorize participants into 3 groups: Improved, Stable, Worsened. The ‘Improved’ group reported at least moderate improvement (Global Scale score 1 to 2). The ‘Stable’ group reported minimal improvement or worsening (Global Scale score 3 to 5). The ‘Worsened’ group reported moderate worsening (Global Scale score 6 to 7). Responders included the ‘Improved’ group. Non-responders including the ‘Stable’ and ‘Worsened’ groups. These cut-offs have previously been used in the calculation of MCIDs and MCICs for resting pain^{20, 44}. One difference between the clinical trials is that the FAST study included a placebo and no-treatment groups, whereas the TEAch study included two active treatments. Therefore, each study individually has an unbalanced ratio of responders to non-responders (Table 1).

Minimal Clinically Important Change (MCIC)

There is considerable variation in terminology and methods when assessing minimal important and clinically important differences.^{16, 26} Distributive-based approaches calculate MCIC/MCID from the outcome’s distribution (e.g. standard deviation), while anchor-based ones compare it to a known anchor measure, such as a global scale⁴². Previously, anchor-based methods were used to establish an MCID for resting and re-call pain in chronic pain conditions²⁰. These studies utilized receiver operating characteristic (ROC) curve methods to estimate an MCID for both raw and percent changes in resting pain.^{19, 24} To facilitate comparison with previous methods while also aligning with current terminology,^{16, 26} we used anchor-based ROC curve in the current manuscript to estimate the MCIC, as further described in the data analysis section³.

Data analysis

Summary statistics are provided for the total sample and by study (FAST, TEAch). QQ-plots were used to assess the normality of continuous measures. The relationships between the Global Scale score and changes in both raw and percent change of movement pain were assessed using a simple linear regression with k-fold cross-validation methods. This method data was split into 6 folds which are used as training groups and one used as a test group. Linear regressions were trained in each of the training groups where the outcome was either raw change or percent change in movement pain. In linear regression models, R² statistics can be used as a goodness-of-fit measure where values closer to 100% represent a better fit of the model. The trained model was applied to the testing group, and the resulting R² values from the testing group were used to calculate the average R² and 95% confidence intervals to quantify the relationship between the Global Scale score and movement pain. A correlation of greater than 0.5 has been recommended to ensure an adequate relationship between the anchor measure and the measure of interest¹⁷. We used a receiver operating characteristic (ROC) curve method to calculate the MCIC when appropriate. ROC curves were calculated using change in movement pain and the binary variable of responder (Yes vs. No) based on the Global Scale score criteria (improved and stable/worsened). The area under the curve (AUC) for the ROC curve was calculated to assess the accuracy of the ROC model. An AUC of 0.5 represents a variable that is no better than a random chance of discriminating between responders and non-responders. An AUC of 1 represents the ability to correctly characterize all participants. AUCs between 0.7–0.8 are acceptable and between 0.8–0.9 are considered excellent²⁵. There can be considerable variability between MCIC calculation methods, therefore it is recommended to assess the MCIC as a point estimate and its variability. The optimal cut-off point of the ROC curve can be used as the MCIC for the change in movement pain. We used the Euclidean method to find the optimal cut-off point which selects the point on the ROC curve closest to the top left corner of the ROC curve area and has been previously used to calculate the MCIC for resting pain³. Bootstrapping methods were used to estimate the 95% confidence interval for the threshold, sensitivity, and specificity⁴³. ROC-based MCIC methods are dependent on the proportion of responders and non-responders. Because of the skewed proportions of responders and non-responders in each individual data, the data from each study were not analyzed separately. The combined data from both clinical samples allows for a more balanced data set and a less biased ROC curve analysis. To evaluate similarities between clinical samples, we utilized the k-fold cross validation to assess the relationship between movement pain and Global Scale score in each sample. The linear regression methods assume that any missing data is missing at random. The ROC analysis required a complete case analysis and did not handle missing data.

The estimate of the SDC used a subset of participants in the FAST study (n=84), who were randomized to receive no treatment during the initial study period. The baseline and 4-week time points were used to assess test-retest reliability. Interclass correlation coefficient (ICC) was calculated using the test re-test analysis. The SDC was calculated using the ICC and pooled standard deviations using established methods^{29, 45}. The SDC was only calculated for raw changes in movement pain as percent change in movement pain is dependent on both baseline and 4-week timepoint and thus we are unable to calculate a test re-test ICC.

$$\text{SDC}=1.96\times \text{SD}\times \sqrt{2(1-\text{ICC})}$$

All analyses were performed using IBM SPSS Statistics (Version 28.0.1.0) and Rstudio (Version:2023.03.1+446).

RESULTS

Relationship between Global Scale scores and movement pain

Both raw and percent change in movement pain improved alongside increased reports of subjective improvement on the Global Scale (Figure 1A and 1B). For the raw changes in movement pain, the average R² across the folds was 0.41 (95%CI: 0.35 – 0.49, Figure 2A). The average R² across the folds for percent change in movement pain was 0.35 (95%CI: 0.23– 0.47, Figure 2B). Separately, the samples with Achilles tendinopathy and fibromyalgia individuals each had about a quarter of the variance of improvement in the Global Scale score accounted for by changes in movement pain. In the Achilles tendinopathy sample, the average R² across the folds was 0.24 (95%CI: 0.02 – 0.46) for the raw change in movement pain and R²=0.23 (95%CI: 0.02 – 0.44) for percent change in movement pain.

Figure 1.

A) Violin plot of raw change in movement pain in each global scale score. B) Violin plot of percent change in movement pain in each global scale score. Median is represented by the bolded line and the IQR is represented by the box for each score category. There is only one data point in the Global Scale Score category of 7 resulting in a mean value.

Figure 2.

A) Scatter plot between raw change in movement pain and global scale score by chronic pain condition Average R² across folds was 0.41 (95%CI: 0.35 – 0.49). B) Scatter plot between percent change in movement pain and global scale score by chronic pain condition. Average R² across folds was 0.35 (95%CI: 0.23– 0.47). Each dot represents individual participant data. (AT- Achilles tendinopathy, FM-Fibromyalgia).

MCIC for raw change in movement pain

The ROC curve for raw change in movement pain had an AUC of 0.87 (95%CI: 0.83–0.92, Figure 3). The MCIC for the raw change in movement pain was 1.1 out of 10 (95% CI: 0.9–1.6, Table 2). This 1.1 threshold had a corresponding sensitivity of 0.83 (95% CI: 0.75– 0.92) and specificity of 0.79 (95% CI: 0.72–0.86).

Figure 3.

Receiver Operating Characteristic (ROC) curves for raw and percent change in movement pain. The Area Under the Curve (AUC) for raw change in movement pain was 0.87 (95%CI: 0.83–0.92). The AUC for percent change in movement pain was 0.86 (95%CI: 0.81–0.91).

The circles represent the optimal cut-off point of 1.1 as the Minimal Clinically Important Change (MCIC) for raw change in movement pain, and 27% as the MCIC for percent change in movement pain.

Table 2.

Minimal clinical important change (MCIC) for raw change and percent change in movement pain and corresponding sensitivity and specificity from each calculation method.

Method	MCIC	Sensitivity	Specificity
Raw Change	1.1 (0.9–1.6)	0.83 (0.75–0.92)	0.79 (0.72–0.86)
Percent Change	27% (10%−44%)	0.79 (0.70–0.88)	0.82(0.72–0.90)

Data are presented as mean (95% confidence interval).

MCIC for percent change in movement pain

The ROC curves for the percent change in movement pain had an AUC is 0.86 (95% CI: 0.81 – 0.91, Figure 3). The MCIC for percent change in movement pain the MCIC is 27% (95%CI: 10%- 44%, Table 2). This threshold had a sensitivity of 0.79 (95% CI: 0.70 – 0.88) and a specificity of 0.82 (95% CI: 0.72 – 0.90).

Smallest Detectable Change

The ICC across the 4-week test re-test time period was 0.77 (95%CI: 0.63 – 0.85). Using this ICC, the SDC was calculated to be 2.5 (95%CI: 1.9 – 3.0).

Missing data

The linear regression analysis to establish the association between movement pain and global scale score assumes these data were missing at random. The MCIC analysis required a complete case analysis at two time points, resulting in exclusion of 15% of the participants due to either missing movement pain or global scale scores. Participants with missing data did not differ from those with complete data (supplemental table 1)

DISCUSSION

The aim of this study was to establish the first movement pain specific MCIC and SDC for chronic musculoskeletal pain conditions. Our analysis found the MCIC for raw change in movement pain is 1.1 (95% CI: 0.9–1.6) on an 11-point NRS and 27% (95%CI: 10%- 44%) for percent change in movement pain. The SDC was 2.5 (95%CI: 1.9 – 3.0) for raw changes in movement pain. These results may assist clinicians and researchers in interpreting improvement in movement pain with non-pharmacological interventions.

The value of the MCIC for movement pain in the current study is consistent with previous research investigating clinically important differences in recalled pain with activity. In individuals with patellofemoral pain syndrome, the MCID for recalled pain with activity from the last week averaged 1.3/10 across 6 different tasks (8/60 on aggregate measure)¹². For patients with chronic tendinopathies both in the upper and lower extremities, the minimal important difference (MID) for recalled pain with activity was between 1.0 and 1.2/10⁶. For individuals with hip and knee osteoarthritis, the minimal clinically important improvement for recalled pain with activity over the last 24 hours was found to be 1/10 for pain during movement³⁶. These findings appear consistent across methodologies, recall times, and clinical populations. It should be noted that the current study was unable to assess MCICs separately for each population due to the skewed proportions of responders and non-responders in each study. We chose to pool the data from both studies to achieve a more balanced ratio of responders and non-responders for the ROC curve analysis to estimate MCICs. The consistency of our findings with the previously investigated MCIDs for recalled pain with activity suggests there may be a similar agreement between recalled pain during activity and movement pain, yet future work is needed to evaluate for potential differences in between populations and types of movement-related pain measures.

Clinical trials using movement pain raw change as an outcome measure have previously used the MCID for resting pain to assess the clinical relevance of treatment effects. A recent systematic review on the raw change in resting pain found the median MCID to be 20/100 (IQR 15–30) when using ROC curve methods for a VAS scale³⁴. For a 0–10 NRS, the MCID ranges from 1⁴⁴ to 2.5³⁰ though there is large heterogeneity in the methods used and the populations studied^{7, 10, 20}. Yet relying on previously established MCIDs for resting pain may lead to an underestimation of the benefits of treatments used in clinical trials targeting movement pain. In individuals with fibromyalgia, two clinical trials assessed the efficacy of TENS on movement pain^{14, 31}. Despite differences in methodology, both studies found statistically significant improvements in movement pain though only one met the proposed MCIC for movement pain¹⁴. Similarly in individuals with temporal mandibular joint (TMJ) pain, movement pain was reduced with the use of TENS. These changes were statistically significant compared to controls and would be considered clinically important with our MCIC for movement pain⁵⁶. Further, clinical trials evaluating pharmaceutical management of movement pain have also demonstrated statistically significant improvements compared to placebo and would be considered clinically meaningful with the proposed MCIC for movement pain⁵¹. This highlights the importance of a movement pain specific MCIC to be used in clinical research.

Our analysis of the MCIC for percent change in movement pain resulted in an MCIC that appears larger than our estimate for the MCIC for raw change in movement pain. While raw and percent change in pain are often reported together, the two measures have inherent differences. A percent change of 20% will result in a greater raw change in individuals with higher baseline values. This is consistent with previous work on MCIC for resting pain which found a larger MCIC for percent change in resting pain, which ranges from 30% to 33%^{20, 44}. Our results may be partially explained by most participants having had moderate levels of pain at baseline which may have resulted in lower raw changes in movement pain resulting in larger percent changes. These differences are likely the reason for differences in MCIC for raw change and percent change in our analysis. This highlights the importance of differentiating interpretations between raw changes and percent changes in movement pain.

The SDC for movement pain was larger than the MCIC in our analysis, indicating that threshold for measurement error may be greater than the meaningful difference. Previous estimates of SDC for resting pain range from 1.3 to 2.6 for resting pain^{1, 55}, which are similar to our estimate of 2.5 for movement pain. The lower estimates were calculated on short follow up periods (<24 hours)¹ than those with longer follow ups (> 48 hours)^{46, 55}. This suggests the 4-week follow up time between test and re-test in our current analysis may partially explain why the SDC is larger than the MCIC. Therefore, the SDC and MCIC values for movement pain should be considered carefully when interpreting the results of clinical trials or practice.

Strengths and Limitations:

A strength of the study is the inclusion of individuals with two different types of chronic musculoskeletal pain with data from two separate studies examining the response to two different non-pharmacological treatments. Yet a limitation is that these findings may not generalize to other types of chronic musculoskeletal pain, men, or individuals seeking pharmacological or invasive treatments. In addition, there is a lot of variability in terminology and methodology for pain, including movement pain, pain with activity, and movement-evoked pain. It is important to consider the differences in how movement pain is measure in clinical research and practice. Our study used movement pain during an activity rather than an evoked pain or change from resting pain to pain with activity. There would likely be differences in the MCIC for movement pain compared to movement-evoked pain. Our MCIC for movement pain will allow researchers to better gauge the effectiveness of interventions in clinical research. Future work is needed to examine the MCIC for other populations, other treatments, and related pain constructs.

A limitation of MCIC calculation, in general, is the large amount of variation between different methods²². This has led to recommendations that urge caution in interpreting a single MCIC⁵⁰. Reporting multiple MCICs or a measure of variance in the MCIC calculation allows for a more nuanced interpretation of clinically important improvements⁵⁴. The current analysis was strengthened by using k-fold cross-validation methods to estimate and assess the relationship between the Global Scale score and changes in movement pain. Further, we were able to use bootstrap methods for the optimal cut-off point which allowed for the calculation of a robust calculation of the MCIC for movement pain and confidence intervals. Reporting the variance of our MCIC estimations will allow for a more appropriate interpretation of the MCIC for movement pain.

Another limitation is that the study samples may differ in how they interpret the global scale score, which covers not only pain but also other symptoms such as fatigue and cognition associated with fibromyalgia or kinesiophobia and activity participation in Achilles tendinopathy. Therefore, a k-cross fold validation was performed in each population, which supports the consistency of the relationship between movement pain and global scale improvement across both populations in this study. Future studies should use a global scale score that is specific to pain rather than general health.

It should also be noted that our choice of cut off balanced the importance of sensitivity and specificity, and this threshold may not be optimal for all uses of the MCIC. In clinical practice, sensitivity may be preferential to minimize a false negative rate while in tightly controlled randomized controlled trials specificity may be emphasized to decrease the false positive rate. In these situations, using different cut-off points would be warranted resulting in a different MCIC. Both studies utilized perceived change in health status, as opposed to directly asking for perceived change in pain which was the outcome of interest. However, our analysis showed a moderate relationship between movement pain and the global score that meets recommendations for MCIC calculation¹⁷. Future studies assessing the MCIC for movement pain could use other methods of calculating MCICs to triangulate the MCIC of movement pain further. Additionally, further research is needed to fully establish the SDC for movement pain to allow for more accurate interpretations of changes in movement pain.

Conclusions:

Our findings are that the MCIC for movement pain was 1.1/10 on an NRS, or an improvement of 27%, with an SDC of 2.5 (95%CI: 1.9 – 3.0). These findings will assist both clinicians and researchers in better understanding the clinical relevance of changes in movement pain in response to non-pharmacologic treatments. Clinically, this will allow clinicians to assess if clinical improvements in movement pain are important and relevant to the patient. In the context of clinical trials, the MCIC for movement pain will allow researchers to assess if treatments for movement pain are clinically effective using a MCIC specific to movement pain instead of relying on previous MCIDs for resting pain which may alter conclusions of effectiveness.

Highlights:

• 310 individuals participated in clinical trials targeting movement pain (0 to 10)

• The minimal clinically important change for raw change in movement pain is 1.1

• For percent change in movement pain the minimal clinically important change is 27%.

• These results will improve the interpretation of changes in movement pain.

PERSPECTIVE.

A minimal clinically important change (MCIC) of 1.1- points (95% CI: 0.9–1.6) for movement pain discriminates between responders and non-responders to rehabilitation. This MCIC provides context for interpreting the meaningfulness of improvement in pain specific to movement tasks.