Assessment of Reliability and Validity of SF-12v2 among a Diabetic Population

• Published in: Value in health 2018-04, Vol.21 (4), p.432-440
• Main Authors: Kathe, Niranjan, Hayes, Corey J., Bhandari, Naleen Raj, Payakachat, Nalin
• Format: Article
• Language: English
• Subjects: Aged; Chronic illnesses; Comorbidity; Confidence intervals; Confirmatory factor analysis; Diabetes; Diabetes mellitus; Diabetes Mellitus - diagnosis; Diabetes Mellitus - psychology; Diabetes Mellitus - therapy; Factor analysis; Female; Glaucoma; Health problems; Health Status; Health Status Indicators; Humans; Longitudinal Studies; Male; Mental Health; Middle Aged; Patient Health Questionnaire; Predictive validity; Predictive Value of Tests; Process Assessment (Health Care); Psychometrics; Quality of Life; Questionnaires; Reproducibility of Results; Retrospective Studies; Risk Factors; SF-12v2; SF-6D; Test-Retest reliability; Treatment Outcome; validation
• ISSN: 1098-3015; 1524-4733
• DOI: 10.1016/j.jval.2017.09.007

To validate the Medical Outcomes Study Short Form version 2 (SF-12v2) in diabetic patients. Adults with self-reported diabetes from the Medical Expenditure Panel Survey (2011–2013) were identified. Reliability (internal consistency and test-retest) and validity (construct, concurrent, criterion, and predictive) of the SF-12v2 were assessed. The SF-12v2 consists of two normalized composite scores: the physical component summary score (PCS12) and the mental component summary score (MCS12). Confirmatory factor analysis was conducted to assess the instrument structure. Concurrent (convergent and discriminant) validity was assessed by a multitrait-multimethod matrix using the Patient Health Questionnaire, the Kessler Scale, and perceived health and mental health questions. The predictive validity was assessed by estimating future limitations. The concurrent validity was tested by comparing the MCS12, PCS12, and utility scores (six-dimensional health state short form) across comorbidity scores. The final sample comprised 2214 diabetic patients with mean normalized (population mean 50; range 0–100) PCS12 and MCS12 scores of 40.81 (standard error 0.33) and 49.82 (standard error 0.26), respectively. The PCS12 and MCS12 scores showed good internal consistency (Cronbach α: PCS12 0.85; MCS12 0.83) and acceptable test-retest reliability (intraclass correlation coefficient: PCS12 0.72; MCS12 0.63) and produced acceptable goodness-of-fit indices (normed fit index 0.95; comparative fit index 0.95; root mean square error of approximation 0.11 [95% confidence interval 0.1017–0.1188]). The PCS12 and MCS12 were moderately correlated with perceived health and perceived mental health. The MCS12 was highly correlated with the Patient Health Questionnaire and the Kessler Scale. Both the PCS12 and the MCS12 could predict the future health limitations. The PCS12, MCS12, and utility scores demonstrated sensitivity to the presence of comorbidity scores. The SF-12v2 is a valid generic instrument for measuring quality of life in diabetic patients.