Confirmatory factor analysis of the diabetes-39 and derivation of a 23-item short form in Mexican adults with type 2 diabetes

Introduction Health-related quality of life (HRQoL) is increasingly recognized as a clinical indicator for patients with type 2 diabetes mellitus (T2DM), yet instruments to measure it remain psychometrically limited. The Diabetes-39 (D-39) has been culturally adapted across several countries, but in Mexico only exploratory factor analysis and convergent validity had previously been reported, and confirmatory evidence was absent. Objective To examine the factor structure of the D-39 through confirmatory factor analysis (CFA) and to derive a short form with initial reliability and validity evidence in Mexican adults with T2DM. Methods Using nonprobabilistic purposive sampling, 201 adults with T2DM (57.2% women; M age =65.12, SD = 11.617) attending a public primary care clinic in Mexico City completed the D-39, the Therapeutic Adherence Scale, and the WHOQOL-BREF. CFA compared the original five-factor 39-item model, the 21 items of a previously reported short form (D-21) specified under the original five-factor structure, and unidimensional models. Results The five-factor models did not fit adequately, and inter-factor correlations (.85–.97) indicated that the domains were not empirically distinguishable. A unidimensional 23-item model (D-23), derived through sequential item removal and including seven residual covariances between items of the same original domain, showed satisfactory fit (χ²[223] = 416.061, p <.001; χ²/ df = 1.866; SRMR = .0238; RMSEA = .066, 90% CI [0.056, 0.076]; TLI = .963; CFI = .968). Standardized loadings ranged from.69 to.92 (AVE = .74). As hypothesized, D-23 scores correlated negatively with the WHOQOL-BREF total score ( r = −.542), with therapeutic adherence ( r = −.237) and were higher among patients with uncontrolled glycemia ( r = .34). Internal consistency was high (α = .985, 95% CI [.982,.988]; ω = .985). Schmid–Leiman analysis of a second-order model supported a total score (explained common variance = .975; ω H = .982). Conclusion The D-23 is a promising candidate brief measure of HRQoL for Mexican adults with T2DM. Once cross-validated, it may serve as a complementary psychological indicator, used alongside glycemic control and medical treatment, to monitor the impact of diabetes on patients’ daily lives in routine care. Cross-validation in independent samples is required before routine clinical use.