Kinesiophobia and physical activity: A systematic review and meta-analysis

OBJECTIVE. Physical activity is known to reduce the risk of disability, disease, and mortality. However, in some patients, an excessive, irrational, and debilitating fear of movement (i.e., kinesiophobia) is thought to induce avoidance behaviors, contributing to decreased engagement in physical activity. The aim of this study was to examine whether kinesiophobia is negatively associated with physical activity and what factors influence this relationship. METHODS. Three databases were searched for articles including both a measure of kinesiophobia and physical activity. Two reviewers screened articles for inclusion, assessed risk of bias, and extracted data from each study. Pearson product-moment correlations were pooled from eligible studies using the generic inverse pooling and random effects method to examine the relationship between kinesiophobia and physical activity. RESULTS. Forty-nine articles were included in the systematic review and 41 studies (n = 4,848) in the meta-analysis. Results showed a moderate negative correlation between kinesiophobia and physical activity (r = -0.31; 95% CI: -0.42 to -0.20; I2 = 1.8%; p < 0.0001). Subgroup meta-analyses revealed that the correlation was statistically significant only in patients with a cardiovascular or arthritis condition and in studies using a self-reported measure of physical activity. There was no evidence of an effect of age, gender, or pain. CONCLUSIONS. Higher levels of kinesiophobia were moderately associated with lower levels of physical activity. However, between-study heterogeneity was considerable, and results showed no evidence of this association when physical activity was assessed with accelerometers or pedometers. Additional studies using device-based measures of physical activity are required to confirm these results and to understand the factors and mechanisms influencing this potential relationship. IMPACT. Our results suggest that kinesiophobia could be considered as a limiting factor when developing physical activity promotion strategies for inactive patients.


INTRODUCTION
Seven decades ago, the seminal work of Morris et al. (1953)  1 showed that conductors on London double-decker buses, who were responsible for checking tickets, assisting passengers with luggage, and supervising the loading and unloading of passengers, had a lower incidence and less severe coronary heart disease than bus drivers.Since then, the scientific literature demonstrating the health benefits of physical activity has grown exponentially and expanded to include multiple health conditions 2 .These benefits include reduced risk of disability, disease, and mortality 2,3 .Specifically, higher levels of physical activity have been shown to contribute to a reduced risk of cardiovascular disease 4 , obesity 5 , depression 6 , hypertension 7 , cancer 8 , and dementia 9 .Yet, one in four adults worldwide does not meet the recommendations for physical activity 10 .Physical activity also plays an important role in secondary and tertiary prevention by reducing the impact, slowing the progression, and preventing the recurrence of multiple health conditions, including cardiovascular disease 11,12 , osteoarthritis 13 , stroke 14,15 , and cancer 16 .Several factors may explain physical inactivity 17 , including environmental, interpersonal, and intrapersonal factors 18 .Environmental factors include lack of access, weather conditions, and safety concerns 19 .Interpersonal factors include family responsibilities, lack of support, and lack of a gym partner 20 .Intrapersonal factors include gender 21 , age 22 , cognitive function 23,24 , and socioeconomic circumstances 25 .Another intrapersonal factor of interest is kinesiophobia, which can be defined as an excessive, irrational, and debilitating fear of movement and activity resulting from a sense of vulnerability to pain, injury, or a medical condition 26 .Kinesiophobia is typically measured using self-administered questionnaires, such as the Tampa Scale of Kinesiophobia (TSK) 27,28 , which assesses an individual's belief that physical activity can lead to injury or pain and that the severity of their medical condition is underestimated.While kinesiophobia is often observed in the context of pain or a clinical condition, its presence in otherwise healthy adults is also possible 29 due to the irrational nature of this phobic condition.The irrational fear that characterizes kinesiophobia is likely to influence the desires and impulses for movement and rest 30 , as well as affective determinants of physical activity in general 31 .The relationship between kinesiophobia and physical activity can be explained by theories suggesting that the perception of a cue related to physical activity automatically activates the concept of physical activity as well as the unpleasant (or pleasant) affective memories associated with this concept [32][33][34][35] .This activation results in an impulse that favors the tendency to avoid (or approach) physical activity 36 .Thus, negative affective associations are likely to hinder physical activity.Accordingly, an aversive fear of pain, injury, or aggravation of a medical condition that has been associated with the concept of movement may result in the development of automatic avoidance behaviors that contribute to the maintenance and exacerbation of this fear, and ultimately lead to a phobic state (i.e., kinesiophobia) that diminishes the ability to engage in regular physical activity.Previous systematic syntheses of the literature on this topic include a meta-analysis 37 and two systematic reviews 38,39 .The main results of these reviews suggest that exercise interventions may reduce kinesiophobia in individuals with back pain.While back pain is one condition that may contribute to kinesiophobia, it is not the only one.The relationship between physical activity and kinesiophobia should be investigated in other conditions such as cardiac, neurological, and rheumatologic conditions.
The main objective of this study was to systematically review and meta-analyze the direct relationship between kinesiophobia and physical activity.We hypothesized that levels of kinesiophobia would be negatively associated with levels of physical activity.In addition, we examined the moderating effect of health status, physical activity measurement instruments (i.e., accelerometers, pedometers, questionnaires), physical activity outcomes (e.g., total physical activity, moderate or vigorous physical activity, steps per day), and kinesiophobia measurement instruments.Finally, because kinesiophobia and physical activity can vary with age, sex, and pain 40,41 , we explored the influence of these factors on the association between kinesiophobia and physical activity.

Search Strategy
This review was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines 42 .Potential studies were identified by searching the MEDLINE (via PubMed), PsychInfo, CINAHL, EMBASE, and SPORTDiscus databases.In October 2023, two reviewers (MG and AF) searched for all available records using the following combination of keywords in the title or abstract of the article: ("kinesiophobia" OR "fear avoidance" OR "fear of movement" OR "movement phobia" OR "movement fear") AND ("physical activity" OR "exercise" OR "walking").In PsychInfo the limits "clinical trial", "quantitative study", "peer-reviewed journal", "English", and "human" were used.In PubMed the limits "clinical trial", "observational study", "RCT", "English" were used.In SPORTDiscus the limits "peer-reviewed", "English", "academic journal", and "article" were used.In CINAHL the limits "peer-reviewed", "English", "research article", "journal article", and "humans" were used.To reduce literature bias 43,44 , this systematic review was pre-registered in PROSPERO 45 .

Eligibility Criteria and Study Selection Inclusion Criteria
To be included in this systematic review, articles had to be published in a peer-reviewed journal, be written in English, report original data collected from human participants, include at least one self-reported measure of kinesiophobia and one measure of physical activity, and formally test the association between these two variables, be it a univariate or multivariate test.The physical activity measure could be derived from a self-reported measure of the level of physical activity or from a device (e.g., accelerometer, pedometer) worn while participants are engaged in their normal daily activities.

Exclusion Criteria
Studies were excluded if they were published as a book chapter, study protocol, conference abstract, or were based on laboratory-based measures of physical fitness (e.g., maximal muscle force, V ̇O₂ max) and not on a measure of physical activity.

Study Selection
Article screening was performed in Covidence systematic review software (Veritas Health Innovation, Melbourne, Australia; www.covidence.org), a web-based collaborative software platform that streamlines the production of systematic reviews.After removing duplicates, titles and abstracts were independently reviewed by two reviewers (MG, AF) according to the inclusion and exclusion criteria using a systematic 5-step process.If there was any doubt at any step, the full text was further reviewed.Step 1: Articles not written in English were excluded.Step 2: Articles that did not report original empirical data were excluded (e.g., reviews, meta-analyses, commentaries, technical reports, case studies).Step 3: Articles that did not involve human participants were excluded.Step 4: Articles that did not assess both kinesiophobia and physical activity were excluded.Step 5: Articles that did not formally test the association between kinesiophobia and physical activity were excluded.In addition, we performed reference screening and forward citation tracking on the articles remaining after step 5. Any disagreements between the two reviewers were resolved by consensus among three reviewers (MG, AF, MPB).

Data Extraction
Data extracted from selected articles included first author's name, article title, publication year, digital object identifier (DOI), number of participants, number of men and women, age range, mean age, mean weight, mean height, mean body mass index, health status, mean pain intensity, type of kinesiophobia measure, level of kinesiophobia, type of physical activity measure, type of physical activity outcome, level of physical activity (continuous or categorical), as well as statistical estimates and significance of the association between kinesiophobia and physical activity.

Methodological Quality and Risk of Bias Assessment
The risk of bias of the studies included in the systematic review was estimated using the National Institutes of Health (NIH) Quality Assessment Tool for Observational Cohort and Cross-Sectional Studies 46 , the Transparent Reporting of Evaluations with Non-Randomized Designs (TREND) reporting checklist 47 , and the Consolidated Standards of Reporting Trials (CONSORT) reporting checklist for randomized trials 48 .All scores were normalized to a 0-10 scale to make them comparable across assessment instruments (Table 1).

Main Meta-Analysis
We pooled Pearson product-moment correlations from eligible studies to examine the relationship between kinesiophobia and physical activity.Correlations were pooled using the generic inverse pooling method via the 'metacor' function in the {meta} R package 50 .This function automatically performs a necessary Fisher's z-transformation on the original, untransformed correlations prior to pooling.The 'metacor' function also reconverts the pooled association back to its original form for ease of interpretation.Correlation estimates were nested within studies using the 'cluster' argument to account for the dependencies between these estimates, resulting in a three-level metaanalysis (level 1: participants, level 2: correlation estimates, level 3: studies).The distribution of variance across levels was assessed using the multilevel version of I 2 54 .The performance of the 2level and 3-level meta-analyses was assessed and compared using the {metafor} R package 51,52 .We anticipated considerable between-study heterogeneity, and therefore used a random-effects model to pool correlations.The restricted maximum likelihood (RML) estimator 55 was used to calculate the heterogeneity variance Tau 2 .In addition to Tau 2 , to quantify between-study heterogeneity, we report the I 2 statistic, which provides the percentage of variability in the correlations that is not caused by sampling error 56 .The I 2 statistic was interpreted as follows: 0-40%, may not be important; 30-60%, may represent moderate heterogeneity; 50-90%, may represent substantial heterogeneity; and 75-100%, may represent considerable heterogeneity.To reduce the risk of false positives, we used a Knapp-Hartung adjustment 57 to calculate the confidence interval around the pooled association.We also report the prediction interval, which provides a range within which we can expect the associations of future studies to fall based on the current evidence.The pooled correlation was interpreted using Cohen's conventions 58 : r ≈ -0.10, small negative correlation; r ≈ -0.30, moderate negative correlation; r ≈ -0.50, large negative correlation.

Publication Bias Assessment
Publication bias was assessed using a funnel plot, which is a scatter plot of the studies' effect size expressed as the Fisher's z transformed correlation on the x-axis against a measure of their standard error (which is indicative of precision of the study's effect size) on the y-axis.When there is no publication bias, the data points in a funnel plot should form a roughly symmetrical, upside-down funnel.Studies in the top part of the plot, which have lower standard errors, are expected to lie closely together, and not far away from the pooled effect size.In the lower part of the plot, studies have higher standard errors, the funnel "opens up", and effect sizes are expected to scatter more heavily to the left and right of the pooled effect.Egger's regression 59 can be used to formally test funnel plot's asymmetry.However, since there is no direct function to conduct Egger's test for 3level models, we calculated it by using the standard errors of the effect size estimates as a predictor in the meta-regression 60 .P-curve analysis 61 was conducted to assess whether the distribution of the statistically significant results was consistent with what would be expected if only true effects were present.When the null hypothesis is true (i.e., there is no true effect), p-values are assumed to follow a uniform distribution: highly significant effects (e.g., p = 0.01) are as likely as barely significant effects (e.g., p = 0.049).However, when the null hypothesis is false (i.e., there is a true effect in our data), p-values are assumed to follow a right-skewed distribution: highly significant effects are more likely than barely significant effects.A left-skewed distribution would suggest that some studies used statistical tests to find significant results in ways that may not be reproducible or generalizable (i.e., p-hacking).

Secondary Meta-Analysis
A secondary meta-analysis was conducted using the same approach, but based on Spearman's rho values, to further test the relationship between kinesiophobia and physical activity.

Subgroup analyses and meta-regressions
Subgroup analyses were conducted to examine the differences in correlations between studies including participants with different health conditions and using different types of physical activity measures (i.e., device-based versus self-reported), physical activity measurement instruments (i.e., type of questionnaires, type of devices), physical activity outcomes, and kinesiophobia measures.Exploratory meta-regressions were conducted to examine if the average age of participants, the proportion of women, and pain in a study predicted the reported correlation between kinesiophobia and physical activity.Pain was normalized to a 0-100 scale to make the data comparable across pain scales.A sensitivity analysis was conducted to examine whether the quality of the studies affected the results.

Association Between Physical Activity and Kinesiophobia
Among the 74 articles included in the systematic review, 42 reported correlation coefficients of the association between physical activity and kinesiophobia.Specifically, 32 articles reported at least one Pearson's r correlation coefficient and 12 articles reported at least one Spearman's rho 87,89,91,113,116,124,127,132,135,143,149,151 .When a correlation coefficient was not reported, but the exact p-value (or t value) and sample size were available and it was possible to know the sign of the correlation, which was the case for 7 studies 83,111,115,120,125,133,144 , the Pearson's r estimate was computed using an ad-hoc R code (Supplemental Code 1A).For the studies that reported a relative p-value < 0.001 instead of an exact p-value, we used a p-value of 0.0009 to estimate an approximate r value 82 .Through email correspondence with the authors, we obtained 23 additional Pearson's r estimates 29,[62][63][64][65][66][67][68][69][70][71][72][73][74][75][76][77][78][79][80][81]104,107 . In tota, 83 Pearson's r estimates from 63 studies and 21 Spearman's rho estimates from 12 studies were used in the meta-analysis (Table 1).The remaining study did not report a correlation coefficient and was therefore not included in the meta-analysis 117 .This study reported a non-statistically significant positive association between physical activity and kinesiophobia based on a standardized beta coefficient.

Meta-Analysis Main Meta-Analysis
Our main meta-analysis of 63 studies, 83 Pearson's r correlation estimates, and 12278 participants revealed a statistically significant small-to-moderate negative correlation between kinesiophobia and physical activity (r = -0.19;95% confidence interval [95CI]: -0.26 to -0.13; p < 0.0001) (Table 2; Figure 2).However, we observed substantial-to-considerable between-study statistical heterogeneity (Tau 2 = 0.06, 95CI: 0.02 to 0.09; I 2 = 85.5%, 95CI: 82.6 to 87.9%), and the prediction interval ranged from r = -0.605 to 0.300, indicating that a moderate positive correlation cannot be ruled out for future studies.The sampling error variance on level 1 and the value of I 2 on level 2, i.e., the amount of heterogeneity variance within studies, were small (10.3% and 8.2%, respectively).The largest share of heterogeneity variance was from level 3, with between-study heterogeneity making up 81.5% of the total variation in our data (Supplemental Figure 1).Overall, this indicates that there is considerable between-study heterogeneity, and less than one tenth of the variance can be explained by differences within studies.The 3-level model showed a better fit than the 2-level model with lower Akaike's information criterion (AIC) (28.4 vs. 39.0) and Bayesian information criterion (BIC) (35.6 vs. 43.8),indicating better performance.These lower AIC and BIC are consistent with the significant likelihood ratio test (LRT) comparing the two models (χ 2 = 12.67, p = 0.0004).Therefore, although the 3-level model introduces an additional parameter, this added complexity has improved our estimate of the pooled effect.

Publication bias assessment
Egger's regression test using the standard errors of the effect size estimates as a predictor in the meta-regression showed that the coefficient of the standard error was significant (b = -1.497,95CI: -2.618 to -0.3754, p = 0.0095), suggesting that the data in the funnel plot was asymmetrical (Figure 3A).This asymmetry may be explained by publication bias, but also by other potential causes, such as different study procedures and between-study heterogeneity 184 , which was substantial-toconsiderable here.The 83 Pearson's r correlation estimates were provided to the p-curve analysis.The observed pcurve included 35 statistically significant results (p < 0.05), 27 of which were highly significant (p < 0.025), and was visually right-skewed (Figure 3B).The other results were excluded because they had a p > 0.05.The p-value of the right-skewness test was < 0.001 for both the half curve (curve of p-values ≤ 0.025) and the full curve (curve of p-values < 0.05), confirming that the pcurve was right-skewed and suggesting that the effect of our meta-analysis is true, i.e., that the effect we estimated is not an artifact caused by selective reporting (e.g., p-hacking) in the literature 185 .In addition, the statistical power of the studies that were included in the p-curve analysis was 97% (90CI: 93 to 98%), suggesting that approximately 90% of the significant results are expected to be replicable.

Secondary Meta-Analyses
Results of the secondary meta-analysis of 12 studies, 21 Spearman's rho correlation estimates, and 2084 participants was consistent with the main meta-analysis as it showed a statistically significant small-to-moderate negative correlation between kinesiophobia and physical activity (r = -0.20;95CI: -0.38 to -0.01; p = 0.049) (Table 2; Supplemental Figure 2).However, we observed substantial-to-considerable between-study statistical heterogeneity (Tau 2 = 0.10, 95CI: 0.04 to 0.28; I 2 = 86.3%)and the prediction interval ranged from r = -0.710 to 0.445, indicating that a moderate positive correlation cannot be ruled out for future studies.

Meta-Regressions
Age did not statistically influence the correlation estimates of the meta-analysis studies (k = 72; p = 0.349).Similarly, the proportion of women (k = 72; p = 0.555) and the mean level of pain in the studies (k = 49; p = 0.481) did not influence correlation estimates.

Sensitivity Analysis
The meta-regression by quality score showed that a study's quality did not influence correlation estimates (k = 83; p = 0.373).

DISCUSSION
The main objective of this study was to systematically review and meta-analyze the direct relationship between kinesiophobia and physical activity.In addition, we examined the influence of potential moderators, such as health status.To our knowledge, this is the first review of its kind on this research topic.

Kinesiophobia and Physical Activity
.
Both the main meta-analysis based on Pearson's r correlation estimates and the secondary metaanalysis based on Spearman's rho correlation estimates showed a small-to-moderate negative correlation between kinesiophobia and physical activity.Importantly, this correlation was observed bot in studies using both self-report (e.g., IPAQ) and device-based measures (i.e., accelerometers or pedometers).These results are consistent with our hypothesis and the dual models of physical activity [32][33][34][35] .According to these theoretical models, our findings suggest that the fear of movement characteristic of kinesiophobia triggers an impulse to avoid physical activity behaviors, which contributes to the maintenance or exacerbation of the initial fear.Accordingly, kinesiophobia and physical inactivity would be self-perpetuating or even self-reinforcing.

Health Status
Our results suggest that patients with a cardiac, neurologic, arthritic, or pulmonary condition, as well as older adults, may be at greater risk for this negative relationship between kinesiophobia and physical activity than those with other conditions.In individuals with a cardiac condition, kinesiophobia and its impact on physical activity may be explained by a fear of inducing a new cardiac event 186 or "causing more damage to the heart" 187 , but also by breathlessness (i.e., dyspnea), which reduces the ability to be physically active and damages confidence, leading to persistent anticipation of negative outcomes from physical activity 188 .Dyspnea is also a major barrier to physical activity in people with a pulmonary condition, such as chronic obstructive pulmonary disease (COPD) 147 .Patients with asthma may have additional disease-related barriers to physical activity, such as the fear of provoking respiratory symptoms and exacerbations 189 .Regarding neurologic conditions, chest tightness reported in patients with Parkinson's disease as a barrier to exercise may be a factor contributing to the association between kinesiophobia and physical activity 190 .Another potential factor in these patients 190 , as well as in stroke survivors 191 and healthy older adults 192 , is fear of falling.In patients with osteoarthritis, the belief that physical activity will "damage the joints" 193 and the perceived fragility of their physical status 194 may be factors contributing to the relationship between kinesiophobia and physical activity.Although our results showed no evidence of an association between kinesiophobia and physical activity in other health conditions such as cancer, post-surgery, post-partum, or obstructive sleep apnea, these effects cannot be fully ruled out, as the lack of statistical significance could be explained by a lack of statistical power in these subgroup meta-analyses including fewer estimates (k = 1 to 5).

Pain versus Fear
Our results showed no evidence of an association between kinesiophobia and physical activity in people with fibromyalgia, acute pain, or chronic pain.This finding was surprising because fear of pain is a key component of kinesiophobia, appearing in 10 of the 17 items on the TSK-17 and TSK-Heart scales, and reinforces the importance of considering the multidimensional nature of kinesiophobia, which not only relates to pain but also reflects fear of injury and fear of worsening a health condition.In addition, contrary to our expectations, we found no statistical evidence showing that pain intensity at rest influenced the effect of kinesiophobia on physical activity, despite the substantial number of estimates included in this analysis (k = 23).This result is consistent with the weak relationship that has been shown between kinesiophobia and pain 195 , further suggesting that it is not the actual pain that prevents physical activity, but the fear of triggering pain, injury, or aggravating an underlying condition.However, this absence of evidence might be related to the methods used to assess pain, which may be better assessed by pain history (e.g., pain duration in months) or pain intensity during exercise.

Limitations
The results of this systematic review and meta-analysis should be interpreted with consideration of several limitations.(1) We report considerable heterogeneity across the included studies, which may be explained by the diversity of the methods used to assess physical activity (questionnaires vs. accelerometers vs. pedometers), the instruments used in these methods (14 different questionnaires, 14 different accelerometers and pedometers), and the physical activity outcomes (n = 6), but also by the different questionnaires used to assess kinesiophobia (n = 11).This heterogeneity suggests that the measures of kinesiophobia and physical activity used in the literature reflect different dimensions of these two constructs.For example, self-reported measures of physical activity do not accurately reflect actual levels of physical activity 196 .(2) Because kinesiophobia is a state, i.e., a dynamic psychological variable, the time difference between the physical activity and kinesiophobia assessments, as well as the context of assessment, may have influenced the results.(3) While a subgroup meta-analysis showed no evidence of an effect of the type of TSK scale, inconsistencies have been noted in the purported dimensions assessed by different TSK scales or across populations 197 , which may have influenced our results.(4) Only 21 of the 98 authors we contacted (21%) shared their estimates (n = 13) or raw data (n = 8) with us, which is more than reported in previous literature 198 .Including these missing data may have affected the results.

Conclusion
Higher levels of kinesiophobia were associated with lower levels of physical activity, especially in people with a cardiac, neurologic, arthritic, and pulmonary condition.According to theoretical models, this relationship between kinesiophobia and physical activity results from automatic processes that may be self-reinforcing and should therefore not be overlooked.However, heterogeneity between studies was substantial-to-considerable for some results, and the evidence for publication bias calls for cautious conclusions about this potential relationship.More evidence is required to determine the impact kinesiophobia should have on therapeutic decisions when aiming to maintain or increase physical activity.Particularly, prospective studies are needed to better understand the factors and mechanisms that influence the relationship between kinesiophobia and physical activity.

ARTICLE INFORMATION Funding
Matthieu P. Boisgontier is supported by the Natural Sciences and Engineering Research Council of Canada (NSERC; RGPIN-2021-03153), the Canada Foundation for Innovation (CFI), Mitacs, and the Banting Research Foundation.The funders had no role in the data collection, management, analysis and interpretation, writing of the report, or the decision to submit the report for publication.Ata Farajzadeh is supported by an Admission Scholarship, a Doctoral International Scholarship, and a Special Merit Scholarship from the University of Ottawa.

Data and Code Sharing
According to good research practices 44 , the dataset, R Markdown script, and supplemental material are freely available in Zenodo 199 .A preprint version of this manuscript is publicly available online 200 and has been recommended by Peer Community In Health & Movement Sciences 201 .

Figure 2 .
Figure 2. Main meta-analysis: Correlation between kinesiophobia and physical activity

Figure 3 . 57 # 2 # 25 # 2 #Supplemental Figure 5 .Supplemental Figure 7 .
Figure 3. Publication bias assessment.Contour-enhanced funnel plot of the main meta-analysis (A).The vertical dashed line represents the average effect size.The two other dashed lines represent the idealized funnel-shape that studies are expected to follow.P-curve analysis (B).The blue line indicates the distribution of the analyzed p-values.The red dotted line illustrates a uniform distribution of the p-values, indicating the absence of a true effect.

Table 1 .
Sample characteristics of studies included in the systematic review.