M. R. MALINOW, J. D. HILL, M. LEES, W. P. MCNULTY AND A. J. OCHSNER III 
107 
!5% dextrose 
EXPERIMENTAL DESIGN 
Aortic flow 
-72 to - 120 tiours 
occlusion 
Figure 1. — Experimental design (discussed in the text). 
for 1 to 2 minutes until an injury current was 
observed in any of the precordial leads in the 
ECG constantly monitored on a 6-channel oscil- 
loscopic recorder (Sanborn, Model 350). The 
following parameters were recorded: (1) 9 
leads ECG; (2) blood flow in the ascending 
aorta (considered to represent cardiac output 
when coronary flow is disregarded) ; (3) left 
atrial pressure; and (4) aortic systolic, dia- 
stolic, and mean pressure. After infarction, re- 
cordings were taken at 5-minute intervals for 
hour, 10-minute intervals for 1 hour, 30-min- 
ute intervals for 2^A hours, and 1-hour inter- 
vals thereafter until 5:00 p.m., and significant 
changes were recorded. External defibrillation 
with D.C. current countershock (American Opt- 
ical Co. ) at 200 watt seconds was applied if se- 
rious ventricular arrhythmias (ventricular 
tachycardia, flutter, or fibrillation) occurred. 
The animals were periodically observed until 
midnight, but no recordings or attempted defi- 
brillations were made after 5:00 p.m. Record- 
ings were then made twice daily until the exper- 
iment was terminated. 
For retrospective prediction, we used dis- 
criminant analysis^^ based on the general for- 
mula : L = K + aW + bX + cY + dZ 
where : 
L = discriminant function 
K = constant 
a = coefficient for aortic systolic blood pres- 
sure W 
b = coefficient for aortic diastolic blood 
pressure X 
c = coefficient for aortic mean blood pres- 
sure Y 
d = coefficient for aortic mean blood flow Z 
At selected periods after ligation, values were 
obtained for systolic, diastolic, and mean blood 
pressure and aortic flow. With these data and a 
program written in Fortran II and executed on 
a Xerox Data System Model 920 digital com- 
puter, the constant and coefficients were deter- 
mined for the "Alive" (La) and "Dead" (Ld) 
groups separately (see results). The pressure 
and aortic flow from individual animals were 
then used with the previously generated con- 
stant and coeflflcients from each group. This 
produced an Li and L2 function for every ani- 
mal. Monkeys whose Li was larger than the L2 
were predicted to survive, and the retrospective 
