A Model for Troponin I as a Quantitative Predictor of In-Hospital Mortality
Daniel A. Waxman, MD*, ,*,
Susan Hecht, MD*,
Joseph Schappert, MD ,|| and
Gregg Husk, MD
* Division of Cardiology
Department of Emergency Medicine
Department of Pathology, Beth Israel Medical Center, New York, New York
|| St. Luke's-Roosevelt Hospital, New York, New York
View larger version (30K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 1 In-hospital mortality for derivation and validation sets, for results reported by our laboratory as negative, indeterminate, and positive. Troponin results of 0 are shown separately here, although they were grouped with negatives when reported. Concentrations are given in µg/l.
|
|
View larger version (37K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 2 Mortality versus troponin I concentration. Graphs A and B describe the derivation and validation sets. Graphs C and D describe ACS and non-ACS patients for the combined data. Individual data points represent observed mortality and were constructed by dividing the data into groups of 10 predicted deaths. The size of each marker is proportional to the number of observed deaths for that point. Troponin concentrations are plotted on a log scale, with zero plotted at 0.004 µg/l. The 3 curved lines represent the univariate logistic model predictions (mean and 95% confidence intervals) as a function of log10-troponin. ACS = acute coronary syndrome.
|
|
View larger version (42K):
[in this window]
[in a new window]
[Download PPT slide]
|
Figure 3 Observed versus predicted mortalities. Graphs A, C, and D represent predictions on the basis of troponin alone. Graph B represents predictions on the basis of troponin together with all predictor variables that remained significant in nested models. Points represent division of the data into groups of 10 predicted deaths. All 4 graphs use the combined data from both the derivation and validation sets. ACS = acute coronary syndrome; r/o = rule out.
|
|
|
