A blinded, randomized, controlled trial of three doses of high-dose insulin in poison-induced cardiogenic shock, J. B. Cole, S. J. Stellpflug, H. Ellsworth, C. P. Anderson, A. B. Adams, K. M. Engebretsen and J. S. Holger, Volume 51, Issue 4, pp. 201–207
The authors would like to replace the following:
Heart rate
At no single points were average heart rates for any two groups statistically different from one another. However, as with CO, a statistically significant dose by time interaction effect upon heart rate was observed (P < 0.0001). On average, heart rate increased by 0.0028 bpm/min each minute for every 1 U/kg/h of insulin (Fig. 3). Thus, the difference in projected averages between the HDI-10 and the HDI-1 arm at the end of a 6-h resuscitation is (0.0028 bpm/min × 360 × 10)−(0.0028 bpm/min × 360 × 1) = 10.08 − 1.008 = 9 bpm.
MAP
Similarly, MAP did not exhibit significant differences between group averages at any point of measurement over the duration of the study, but the trend over time was significantly different (P < 0.001). As is somewhat visible from the projected plot of MAP versus time, the HDI-10 arm maintained an average MAP that was more or less stable, while the other three groups experienced a monotone downward trend (Fig. 4). Thus, assuming a linear dose by time interaction trend, the effect of 1 U/kg/h of insulin is to raise MAP by 0.0024 mmHg every minute, which has given the observed degree of uncertainty, is statistically nonzero (<P = 0.001). The predicted effect is such that MAP would be 8 mmHg higher in the HDI-10 arm than in the HDI-1 arm at the end of the 360 min.
SVR
In keeping with the pattern, SVR was not significantly different by dose at most of the measured time intervals, but a linear mixed effects model indicated that SVR decreased faster over time in groups receiving higher doses of insulin (P < 0.001). This was expected as HDI is not a vasopressor, rather it is a vasodilator. 22 This interaction effect was estimated to be a decrease of approximately 0.067 dyne-s/cm5 per minute, for every 1 U/kg/h of insulin. Thus, SVR would be predicted to decrease by 217 dyne × s/cm5 over the entire 360-min study interval in the HDI-10 arm compared with the HDI-1 arm (Fig. 5).