Comorbidity assessment using the Index of Coexistent Diseases in a multicenter clinical trial

The Hemodialysis (HEMO) Study is a multicenter trial designed to determine whether hemodialysis dose and membrane flux affect survival. Comorbid conditions are also important determinants of survival, and thus, an accurate and reliable method to assess comorbidity was required. Comorbidity was being assessed at baseline and annually in the HEMO Study using the Index of Coexistent Disease (ICED). We describe the instrument, its implementation in the HEMO Study, and the results of comorbidity assessment in the first 1000 randomized patients in the trial.

The ICED aggregated the presence and severity of 19 medical conditions and 11 physical impairments within two scales: the Index of Disease Severity (IDS) and the Index of Physical Impairment (IPI). The final ICED score was determined by an algorithm combining the peak scores for the IDS and IPI. The range of the ICED was from 0 to 3, reflecting increasing severity.

Study personnel at 15 clinical centers were trained to update and abstract data from the dialysis medical records. Availability of data, measures of construct validity, and measures of reliability were adequate; 99.8% and 60.6% of patients had comorbid conditions in at least one IDS or IPI category, respectively. The distribution of patients by ICED level was 0 (0.2%), 1 (34.9%), 2 (31.2%), and 3 (33.7%). In multivariable analysis, the following factors were significantly associated with more severe comorbidity: older age, diabetes and other causes of renal disease, a lower level of education, employment status (unemployed and retired), longer duration of dialysis, and lower serum creatinine. There was a significant variation in the severity of comorbidity among clinical centers after adjustment for other factors. The R2 of the model was 25.3%, indicating that a substantial proportion of the variation in the ICED was not explained by these factors.

We conclude that comorbidity assessment using the ICED is feasible in multicenter clinical trials of dialysis patients. There is a large burden of comorbidity in dialysis patients, which is not well explained by the cause of renal disease, demographic, and socioeconomic factors and common clinical and laboratory measurements. These variables should not be considered substitutes for comorbid conditions in case-mix adjustment. Comorbidity assessment is useful to describe the sample population, to improve the precision of the treatment effect, and to use possibly as an outcome measurement.