Send to

Choose Destination
Pharmacotherapy. 2005 May;25(5):676-84.

Development and validation of a diabetes mellitus severity index: a risk-adjustment tool for predicting health care resource use and costs.

Author information

Department of Pharmacotherapy, College of Pharmacy, Pharmacotherapy Outcomes Research Center, University of Utah, Salt Lake City, 84108, USA.



To develop and validate a diabetes mellitus-specific risk-adjustment tool--the diabetes severity index (DSI)--to assist in predicting health care costs and resources within populations of patients with diabetes.


Retrospective analysis of clinical and resource use for patients with a diagnosis of diabetes mellitus. Model estimation was conducted with half the sample, and validation analysis was conducted with the other half.


Southern Arizona Veterans Affairs Health Care System.


Seven hundred thirty-four patients with diabetes (710 men, 24 women; mean age 66 yrs).


Clinical measures of diabetes severity (known as the DSI) were used to predict three health care resource outcomes: risk of hospitalization, and total and ambulatory health care costs. Validity of the DSI was assessed by comparing the DSI with the revised chronic disease score (CDS). The DSI weights ranged from -471.5-3081.2 for total health care costs, from -304.3-1582.1 for outpatient costs, and -0.19-0.93 for risk of hospitalization. The DSI explained 6-8% of the variance in total and ambulatory costs and performed significantly (p<0.05) better than demographics alone, but was similar to the CDS. When the DSI was used with the CDS, up to 8% of variability in costs and use were explained.


The DSI was developed to stratify veterans with diabetes according to diabetes severity and comorbidity. Since additional variance in medical care costs were explained by combining the DSI and CDS, the two indexes may explain different dimensions of a patient's severity of diabetes. However, the utility of such an index based on clinical measures to predict short-term use may be limited due to low yet significant variability explained in the outcomes.

[Indexed for MEDLINE]

Supplemental Content

Loading ...
Support Center