RCMAR Measurement Tools
J. Gregory Boyer and Jo Anne L. Earp (1997)
Background and Development:
The Diabetes-39 questionnaire was developed initially from a population of 516 diabetic patients located in Cary, North Carolina who received and answered an initial set of 93 questions. This mostly white (0% Black, 12% Hispanic), suburban population had an average age of 52. This tool uses a visual analogue scale for each question on which the respondent places an ‘X’ to indicate level of agreement to the question. Validity and reliability of the measurement tool was established for the total sample and reliability was also assessed for two subsamples: respondents with less than a high school education, and respondents older than 75 years. Factor analysis of the responses resulted in a 42-item instrument with six categories: Energy and Mobility, Diabetes Control, Anxiety and Worry, Social Burden, Sexual Functioning, and Diabetes Medication. Assessment of validity of the 42-item instrument resulted in correlations for each item and its corrected scale within each category ranging from 0.5 to 0.84. Cronbach’s coefficient alpha values from each of the six scales ranged from 0.81 to 0.92. For the subsamples of respondents with less than a high school education or older than 75, Cronbach’s alpha values were greater than 0.7 except for the category of Diabetes Medication.
Further testing of the validity and reliability of the 42-item questionnaire was conducted in two studies. The “Iowa Study” consisted of 165 largely rural respondents from a general practice in Iowa (100% White, 45% Male), and the “Carolina Study”, which consisted of 262 ethnically diverse respondents from an outpatient clinic in Chapel Hill, North Carolina (42% White, 55% Black, 36% male). Following the same factor analysis procedures used for the initial set of questions yielded correlations for each item and its corrected scale within each category ranging from 0.45 to 0.84. After review of these correlations three questions and one category were dropped resulting in the final 39-item instrument (D-39) measuring five categories: Energy and Mobility, Diabetes Control, Anxiety and Worry, Social Burden, and Sexual Functioning. Reliability of the 39-item instrument as measured by Cronbach’s coefficient alpha ranged from 0.82 to 0.93. In validity measurements for the subgroups with less than a high school education or older than 75 years, Cronbach’s coefficient alpha ranged from 0.76 to 0.95.
Construct validity was compared with the SF-36 Health Status Questionnaire. Negative correlations were expected due to high scores on the D-39 representing maximum impact on quality of life and high scores on the SF-36 indicating minimal impact on quality of life. Strong negative correlations were identified for the D-39 scales of Energy and Mobility, Anxiety and Worry, and Social Burden against comparable scales of the SF-36. The SF-36 contains no scales comparable to the D-39 Diabetes Control or Sexual Functioning scales.
Validation in Elderly Populations:
The Diabetes-39 Questionnaire has been tested and validated in older populations (over age 75) in North Carolina with Cronbach’s coefficient alpha ranging from 0.76 to 0.95 for the five individual categories of the instrument.
Validation in Minority Populations:
As part of its development, the Diabetes-39 Questionnaire was tested and validated in largely rural (Iowa) and minority populations (North Carolina – 41.8% White, 54.4% Black, 0.8% Asian, 0.4% Hispanic, 2.7% American Indian). Reliability of the instrument in these populations, as measured by Cronbach’s coefficient alpha, ranged from 0.82 to 0.93.
No additional research studies employing this measurement tool have been published.
Design Strengths and Weaknesses:
In a review of health outcome measures for diabetes Garratt et al. (2002) state that this instrument has good evidence for reliability, and internal and external construct validity. The tool is self-administered, having been mailed to and completed by all of the respondents. A weakness of this measurement tool may be the analogue response scale. This scale, while very patient friendly, may be difficult to interpret by the researchers conducting the scoring since it requires actual measurement with a ruler as to the location of the respondent’s mark along the scale line.
This measurement tool is available at no cost from:
Boyer, J.G. and J.A.L. Earp (1997) The development of an instrument for assessing the quality of life of people with diabetes. Med. Care, 35(5) 440-453.
Garratt, A.M., L. Schmidt, and R. Fitzpatrick (2002) Patient-assessed health outcome measures for diabetes: a structured review. Diabetic Med., 19, 1-11.
© 2006 RCMAR