New Chinese scale measures how well diabetes patients feel they know exercise

Researchers in Hangzhou, China, have developed and given a preliminary psychometric evaluation to a modified Chinese-language instrument designed to capture how confidently patients with type 2 diabetes believe they have mastered exercise-related knowledge. The instrument, called the DEKS-C, is not a direct translation of the Turkish Diabetes Exercise Knowledge Scale from which it derives, but a substantially modified version with fewer items, added safety content and a simplified response format. In a study of 438 patients conducted at a single tertiary hospital between February and July 2025, the 17-item scale showed very high internal consistency and a broadly stable three-factor grouping, although the authors themselves caution that the structure remains provisional and requires independent confirmation.

The motivation lies in a measurement gap. According to the 11th edition of the International Diabetes Federation Diabetes Atlas, an estimated 589 million adults aged 20 to 79 years were living with diabetes globally in 2024, a prevalence of 11.1 percent, projected to reach 853 million by 2050, with China bearing the world’s largest burden at more than 140 million affected individuals. Physical exercise is a cornerstone of lifestyle modification, lowering diabetes incidence, reducing the risk of diabetes-related complications and improving quality of life, yet existing tools assess general diabetes knowledge or general physical activity rather than diabetes-specific exercise knowledge. Broadly validated questionnaires such as the Diabetes Knowledge Questionnaire and Diabetes Knowledge Test contain only a small number of exercise-focused items, while instruments like the International Physical Activity Questionnaire measure behavior rather than knowledge, and Chinese exercise instruments are predominantly knowledge-attitude-practice composites with few dedicated exercise-knowledge items.

A narrative literature search across PubMed, Web of Science, CNKI and Wanfang covering publications up to December 2025 identified no dedicated, validated scale for assessing self-rated mastery of exercise knowledge among Chinese patients with type 2 diabetes. The authors are careful to note that this was a narrative rather than a formal systematic review, so the result reflects only their targeted search and does not constitute a formally established evidence gap. The original DEKS, developed by Ercan in 2022 for Turkish populations, evaluates exercise-related knowledge across four dimensions and 25 items, but several of its items showed semantic overlap and its five-point agreement scale posed comprehension difficulties for older participants with lower educational attainment.

The adaptation followed a sequential exploratory mixed-methods design. Two independent bilingual translators produced forward translations, which were reconciled and then back-translated by two further bilingual researchers blind to the original instrument. An expert committee compared the back-translations with the source scale, and substantial modifications were approved in writing by the original developer. Content validation by a 10-member expert panel, whose members had at least five years of experience in diabetes, endocrinology or metabolic disorders, consolidated nine semantically overlapping items into three and added three new items, producing a 22-item pre-interview prototype. Item-level content validity indices reached 1.00 for 23 items and 0.90 for two items, with a scale-level average of 0.99, although these indices apply only to the 25-item translated stage and were not re-quantified for the final 17-item version.

Three successive rounds of cognitive interviews with 23 patients, whose mean age was 52.17 years and of whom 60.87 percent were men, then refined the prototype. Interviews lasted 20 to 60 minutes, were audio-recorded and transcribed, and explored how participants interpreted the instructions, item wording and response options. Revisions added explanatory text and contextual examples for participants with low educational attainment, and four further items were consolidated or relocated, with one merged into the foot-care item, reducing the pool to the final 17 items covering health benefits, exercise program, foot care during exercise and general precautions. A key change concerned the response format: because the items state facts about exercise in diabetes rather than opinions, agreement anchors did not fit, and participants struggled to map partial knowledge onto intermediate numeric values. The final format is a four-point labeled self-rated mastery scale running from completely unaware to completely clear and already practicing, with the neutral midpoint removed.

In the quantitative phase, all 438 patients completed the 17-item DEKS-C electronically during hospital admission, with no missing item responses and no exclusions. Cronbach’s alpha was 0.967, with corrected item-total correlations between 0.668 and 0.837, the Kaiser-Meyer-Olkin measure was 0.965, and Bartlett’s test of sphericity was significant. Horn’s parallel analysis, based on principal axis factoring with oblimin rotation, retained three factors in the full sample and in all four predefined sensitivity subsamples, corresponding to exercise benefits cognition, exercise implementation and exercise safety, and explaining 71.23 percent of total variance in the full sample. An equivalent structure emerged when the analysis was repeated on a polychoric correlation matrix, the more appropriate basis for four-point ordinal items.

The structural findings come with substantial caveats that the authors state explicitly. The three factors were substantially intercorrelated, with absolute correlations of 0.51 to 0.79 across analyses, and the first factor dominated the common variance, accounting for 63.43 percent in the one-factor solution. One pattern coefficient exceeded unity in the oblique solution, item 6 met the salience criterion on no factor, and items 5 and 16 loaded on factors other than their theoretical domains. The authors therefore regard the multidimensional model as unverified, note that a strong general dimension remains plausible, and suggest that the total score may be the more defensible summary. Independent multicenter and multiregional confirmatory factor analysis, comparing three-factor, one-factor and the original four-factor models, is the stated priority.

Response-pattern analyses added a further layer of caution. Floor and ceiling effects of 13.47 and 8.90 percent arose entirely from 59 participants who selected the lowest option on every item and 39 who selected the highest on every item, and both proportions dropped to zero when uniform responders were excluded. The all-minimum subgroup completed the questionnaire far faster than the 335 non-uniform responders, with a median of 89 seconds versus 232 seconds, and was over-represented among older respondents and those with primary-school education or less. The all-maximum subgroup showed no such timing signal, but 48.7 percent reported taking no exercise at all, discordant with the option they endorsed stating they were already practicing every item’s content. Because the scale contains no reverse-scored items, uniform extreme responding and floor or ceiling scores coincide by design, and neither end of the scale can be read as a purely substantive result. Uniform responding also inflates the reliability estimate; sensitivity analyses excluding such respondents yielded Cronbach’s alpha values of 0.929 to 0.949, which the authors treat as the more conservative basis for interpretation.

The instrument’s intended use is deliberately narrow. It captures self-rated mastery of the knowledge content described in each item, not objectively verified factual knowledge, and the highest response option combines knowing with practicing, so knowledge and behavior cannot be separated at the top of the scale. No diagnostic, screening or risk cut-off is proposed; the score is intended for descriptive and educational use, helping identify content areas about which patients feel least certain and which may become targets for tailored patient education. The study evaluated measurement properties rather than intervention effectiveness, and no claim is made that using the DEKS-C improves outcomes.

Limitations are extensive and acknowledged in detail: convenience sampling from a single tertiary hospital limits generalizability; test-retest reliability, measurement invariance, criterion validity, hypothesis-based construct validity and responsiveness remain unexamined; no attention-check items were included, so inattentive responding could not be statistically disentangled from low knowledge; and content validity for the final 17-item version was not re-quantified by an independent panel. Future work should confirm the factor structure in independent samples, test measurement invariance, examine convergent validity against an objective exercise-knowledge measure, and assess a revised response format that separates knowing from practicing. Until then, the DEKS-C stands as a promising but preliminary tool whose dimensional architecture, the authors conclude, cannot yet be regarded as firmly established.

Subject of Research: Development and preliminary psychometric evaluation of the DEKS-C, a modified Chinese instrument for self-rated mastery of exercise knowledge in type 2 diabetes

Article Title: Development and preliminary psychometric evaluation of a modified Chinese instrument for self-rated mastery of diabetes exercise knowledge (DEKS-C)

Article References: Mei, L., Pan, H., Zhu, K., Yin, D., Yao, H., Zhu, M., Zhou, Y., Lv, L., & Shao, J. (2026). Development and preliminary psychometric evaluation of a modified Chinese instrument for self-rated mastery of diabetes exercise knowledge (DEKS-C). Frontiers in Public Health, 14, Article 1902317. https://doi.org/10.3389/fpubh.2026.1902317

Image Credits: AI Generated

DOI: 10.3389/fpubh.2026.1902317

Keywords: type 2 diabetes, exercise knowledge, scale development, cross-cultural adaptation, cognitive interviewing, psychometrics, factor analysis, DEKS-C, self-rated mastery, China, patient-reported outcome, nursing assessment