I Hate Dialysis Message Board
Dialysis Discussion => Dialysis: News Articles => Topic started by: okarol on August 20, 2011, 11:29:30 AM
-
Five-Factor Model Can Estimate Hemodialysis Mortality Rate
Last Updated: August 12, 2011.
Good calibration, reasonable discrimination in model, including age and four other factors
Probable mortality rate in individuals undergoing hemodialysis can be estimated and stratified using a model consisting of five factors, according to a study published online Aug. 3 in the Journal of Internal Medicine.
FRIDAY, Aug. 12 (HealthDay News) -- Probable mortality rate in individuals undergoing hemodialysis (HD) can be estimated and stratified using a model consisting of five factors, according to a study published online Aug. 3 in the Journal of Internal Medicine.
Ingar Holme, Ph.D., from the Oslo University Hospital in Norway, and colleagues developed a prognostic model of mortality in HD over three to four years using the database of the Assessment of Survival and Cardiovascular Events study. From experience, five factors -- age, albumin, high-sensitivity C-reactive protein (hsCRP), history of cardiovascular disease, and diabetes -- were selected and forced into the regression equation. A 67 percent random try-out sample identified no other factors adding significantly to mortality outcome. Individual probabilities of death were estimated in the try-out and the remaining 33 percent test samples. The prognostic index with regression coefficients from the try-out sample to patients in the test sample was calculated to test the calibration. Receiver operating characteristic (ROC) areas were used to assess discrimination.
The investigators found age to be the strongest prognostic factor in the try-out sample, with minor differences between the other four factors. When the number of deaths was multiplied by a constant of 1.33, the calibration in the test sample was good. In both the try-out and test samples, the five-factor model discriminated reasonably well between deceased and surviving patients, with an ROC area of about 0.73.
"A model consisting of five factors can be used to estimate and stratify the probability of death for individuals. The model is most useful for long-term prognosis in an HD population with survival prospects of more than one year," the author write.
The authors disclosed financial ties with the medical device and pharmaceutical industries, including AstraZeneca, which funded the study.
Abstract
Full Text (subscription or payment may be required)
http://www.doctorslounge.com/index.php/news/pb/22314