Research

Misinterpretation with norm-based scoring of health status in adults with type 1 diabetes

Alison L Supina¹ , David H Feeny^2,3 , Linda J Carroll⁴ and Jeffrey A Johnson^4,5

¹  Centre for Health and Policy Studies, University of Calgary, Calgary, AB, Canada

²  Institute of Health Economics, Department of Economics, University of Alberta, Edmonton, AB, Canada

³  Kaiser Permanente Northwest Center for Health Research, Health Utilities Inc., Dundas, ON, Canada

⁴  Department of Public Health Sciences, Faculty of Medicine, University of Alberta, Edmonton, AB, Canada

⁵  Institute of Health Economics, #1200 10405 Jasper Ave NW, Edmonton, Alberta, T5J 3N4 Canada

author email corresponding author email

Health and Quality of Life Outcomes 2006, 4:15doi:10.1186/1477-7525-4-15

Published:	16 March 2006

Abstract

Background

Interpretations of profile and preference based measure scores can differ. Profile measures often use a norm-based scoring algorithm where each scale is scored to have a standardized mean and standard deviation, relative to the general population scores/norms (i.e., norm-based). Preference-based index measures generate an overall scores on the conventional scale in which 0.00 is assigned to dead and 1.00 is assigned to perfect health. Our objective was to investigate the interpretation of norm-based scoring of generic health status measures in a population of adults with type 1 diabetes by comparing norm-based health status scores and preference-based health-related quality of life (HRQL) scores.

Methods

Data were collected through self-complete questionnaires sent to patients with type 1 diabetes. The RAND-36 and the Health Utilities Index Mark 3 (HUI3) were included.

Results

A total of 216 (61%) questionnaires were returned. The respondent sample was predominantly female (58.8%); had a mean (SD) age of 37.1 (14.3) years and a mean duration of diabetes of 20.9 (12.4) years. Mean (SD) health status scores were: RAND-36 PHC 47.9 (9.4), RAND-36 MHC 47.2 (11.8), and HUI3 0.78 (0.23). Histograms of these scores show substantial left skew. HUI3 scores were similar to those previously reported for diabetes in the general Canadian population. Physical and mental health summary scores of the RAND-36 suggest that this population is as healthy as the general adult population.

Conclusion

In this sample, a preference-based measure indicated poorer health, consistent with clinical evidence, whereas a norm-based measure indicated health similar to the average for the general population. Norm-based scoring measure may provide misleading interpretations in populations when health status is not normally distributed.