Premium
Reflexive algorithmic approach to clinical decision making: Breast cancer as a model
Author(s) -
Aziz Douglas C.,
Barathur Raj R.
Publication year - 1993
Publication title -
journal of cellular biochemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.028
H-Index - 165
eISSN - 1097-4644
pISSN - 0730-2312
DOI - 10.1002/jcb.240531146
Subject(s) - breast cancer , lymph node , medicine , stage (stratigraphy) , oncology , cathepsin d , algorithm , cancer , computer science , gynecology , biology , paleontology , biochemistry , enzyme
The number of tests available for the prognostication of patients with breast cancer, ( e.g. , estrogen and progesterone receptor, DNA ploidy, % S‐phase analysis, HER‐2/ neu , EGFR, p53, cathepsin D, pS2, PCNA, etc. ) is staggering. Many published studies statistically prove the prognostic significance for each independent test, but the situation becomes confusing and empirical for the clinician making a decision for a particular patient, particularly when test utilization and cost considerations must be weighed into the equation. Other factors such as the pathological stage, histological grade, vascular and lymphatic invasion, and the age and wishes of the patient should all be taken into consideration in arriving at the optimal treatment protocol. We have applied a Bayesian probability approach to published data in order to derive a branched tree algorithm to predict the survival rates for both lymph node‐positive and lymph node‐negative women with breast cancer. Specimen quality and test results suggested which subsequent tests were most clinically useful. The size of the algorithm was reduced to minimize the number of tests requested and thus reduce costs. This type of analysis is necessary to ensure that the most information is obtained at the lowest cost, and serves as a model for other diagnostic situations.