J.Jpn. Surg. Soc.. 78(12): 1165-1176, 1977

---

Original article

PROBABILISTIC QUANTIFICATION OF RISK ASSOCIATED WITH DISEASE OR TREATMENT WITH SPECIAL REFERENCE TO OPERATIVE RISK

When an accurate decision making is required as to whether a risky treatment should be performed in a patient, the prognosis or the risk of the treatment must be weighed against the probable outcome of the patient as left untreated. Hence, each risk must be quantitatively assessed as the probability of dying or developing complications.
The author described a statistical method by which to quantitatively estimate the risk of a new case from the clinical date of the old cases with known outcome . This method is analogous to probit analysis which has been used in dose-quantal response assays. it is based on the assumption that the critical value of a measurement whose high or low value is associated with a poor prognosis shows a nromal (or log normal) distribution although it varies from case to case. Then, the risk being expressed as its cumulative distribution function, its parameters are estimated by the maximum likelihood method. This analysis may be extended from a single measurement to multiple variables, using a linear model.
This method was applied to assessment of the risk involved in entero-enterostomy in intestinal obstruction, and to evaluation of the prognosis of congenital biliary atresia.
It is emphasized that critical values and risk gradings which have been empirically accepted should be reevaluated by such a statistical method as described here.

---

---

To read the PDF file you will need Adobe Reader installed on your computer.