A shorter structure emerges
An exploratory study in people with knee osteoarthritis found that an eight-item version of the Tampa Scale for Kinesiophobia had the strongest relative fit among the questionnaire structures tested. The shorter version divided the measure into two domains, Activity avoidance and Somatic focus. But the researchers did not present it as a finished or confirmed measure. They described it as preliminary and hypothesis-generating, with independent validation still needed.
The researchers compared three candidates: the original 17-item, one-domain questionnaire; an 11-item, one-domain model; and an 11-item model with two correlated domains. Because questionnaire responses were ordered categories, they used ordinal confirmatory factor analysis, a method suited to that kind of data, and several fit measures to see how closely each proposed structure matched the response patterns. All of the a priori structures were described as having borderline fit and weak loadings. The later eight-item model had the highest CFI, at 0.948, and TLI, at 0.924, but also the highest RMSEA, at 0.078. Those mixed results support a relative ranking among models, not a conclusion that the structure had adequate validity.
How the shorter version was built
To produce that version, the team removed items 5, 10 and 13 from the 11-item two-domain model because each had a loading below 0.20. In practical terms, those items were not contributing strongly to the domains they were meant to represent. The retained Activity avoidance items were 1, 2, 15 and 17, while Somatic focus included items 3, 6, 7 and 11. Even then, loadings ranged from 0.24 to 0.78 for Activity avoidance and from 0.32 to 0.68 for Somatic focus. Items 15, 6 and 11 remained relatively weak indicators.
The study was cross-sectional and ran from May 2023 to December 2024 at two physical therapy clinics and two primary care units in São Paulo, Brazil. Of 131 people assessed, 12 were excluded: four had fibromyalgia, two had active inflammatory joint disease and six did not have confirmed knee osteoarthritis. That left 119 participants. The group included 93 women, or 78.2 percent, and had a mean age of 60.93 years. The researchers reported ethics approval, written consent and sequential recruitment.
All baseline data and the 50 retest records were complete. Those 50 participants were reassessed after seven days without treatment. The target sample had been set using a minimum of seven participants per item for the original 17-item scale, which produced a target of 119. The authors cautioned that this rule does not establish adequate precision for ordinal factor analysis, so the findings were treated as preliminary.
The two domains were not equally strong
The two domains did not line up equally well with related measures. Activity avoidance had a strong association with pain intensity, with a Spearman correlation of 0.63, but weak associations with disability (0.25), rumination (0.29), magnification (0.28) and helplessness (0.13). Somatic focus was moderately associated with rumination (0.39), while its associations with pain (0.16), disability (0.22), magnification (0.29) and helplessness (0.11) were weak. The researchers judged the convergent-validity evidence only partly supportive.
Reliability over time was substantial in the 50-person retest group. Somatic focus had an intraclass correlation coefficient, a measure of consistency across repeated assessments, of 0.85, with a 95 percent confidence interval from 0.75 to 0.91. Activity avoidance had an ICC of 0.78, with an interval from 0.62 to 0.88. The standard errors of measurement were 4.71 percent and 5.89 percent, respectively, suggesting relatively little measurement error in this subsample.
Another reliability check produced a less even picture. Composite reliability was adequate for Activity avoidance at 0.71 but inadequate for Somatic focus at 0.56, so the latter domain warrants more cautious interpretation. The study found no floor or ceiling effects. Minimum scores were reached by seven Somatic focus participants (5.9 percent) and three Activity avoidance participants (2.5 percent); maximum scores were reached by two (1.7 percent) and one (0.8 percent), respectively.
Why the finding remains provisional
The main methodological warning is that the shorter model was derived and evaluated in the same dataset. Removing items and then assessing the fit of the resulting model in those same responses can create overfitting and make performance look better than it will in a new sample. The paper also identified missing tests of measurement invariance and responsiveness, along with the lack of a stability anchor at retest. Without that anchor, some of the apparent reliability may reflect real change in symptoms rather than score stability alone.
Taken together, the findings identify a candidate scoring structure, not a settled replacement for the original questionnaire. Within this comparison, the eight-item two-domain version had the most favourable relative fit, but the reliability evidence was stronger for Activity avoidance than for Somatic focus. The authors' message was cautious: the reduced version should be treated as exploratory and hypothesis-generating until independent studies validate it.
Disclosure
The work was partially supported by CAPES under finance code 001. The authors declared no conflicts of interest.
Paper data and sources
Original title: Structural Validity and Measurement Properties of the Tampa Scale for Kinesiophobia in Knee Osteoarthritis: Comparison of Candidate Factor Structures
Authors: Junior J, Carlesso L, Miura B et al.
Journal/Repository: Musculoskeletal care
Status: Peer-reviewed
First online: 2026-10-12
DOI: Not available
Original paper · Full text