Validity and reliability of the Short Form-36 in cervical spondylotic myelopathy

Validity (the extent to which a test measures what it is intended to measure) and reliability (the stability and reproducibility of measures of the same concept over time or across methods of gathering data) are important characteristics of any outcomes instrument. Generic outcomes instruments are designed for use in any population; however, their validity and reliability in particular diseases should be verified to ensure their appropriateness for use in that disease. In this study the authors assessed the validity and reliability of the Medical Outcomes Study Short Form-36 (SF-36), a generic outcomes instrument, in a population of patients with cervical spondylotic myelopathy (CSM).

The SF-36 was administered to a cohort of patients with CSM on an outpatient basis. Symptom-related data derived from a structured interview and physical examination findings were used to classify cases according to the myelopathy scales of Nurick, Cooper, Harsh, and a Western modification of the Japanese Orthopaedic Association (JOA). Construct validity was assessed by determining whether SF-36 scores varied in accordance with predefined hypotheses relating to the myelopathy scores by using the Cuzick nonparametric test for trend. The reliability of the SF-36 scores was assessed using Cronbach alpha. Eighty-eight patients with CSM completed the SF-36 and interview. Construct validity was demonstrated by confirming the hypothesized relationship between SF-36 scales and the myelopathy scales of Nurick (p < or = 0.003), Cooper leg subscale (p < or = 0.012, except the general health perceptions domain [p = 0.091]), Harsh (p < or = 0.016), and the motor component of the modified JOA (p < or = 0.006). Reliability was demonstrated for all eight SF-36 domain scales and the physical component and mental component summary scales, in which Cronbach alpha satisfied the Nunnally criterion of 0.7.

The SF-36 provides valid and reliable data on patients with CSM.