In Reply

We thank Dr. Pivalizza et al. and Dr. Carette et al. for their letters in response to our research on increased postoperative mortality associated with intraoperative “triple low”: concurrent low mean arterial pressure (less than 75 mmHg), low volatile anesthetic minimum alveolar concentration (MAC; less than 0.8), and low bispectral index (less than 45).¹ 

We are surprised that Pivalizza et al. found our conclusion to be strongly worded when the adjective “weak” is the chief qualifier in our abstract’s conclusion: “There is a weak [emphasis added] independent association between the triple low state and postoperative mortality....”¹  Pivalizza et al. raise four specific points. (1) We did not conduct a power analysis since we used a convenience sample from previously conducted studies. The concern cannot be that our study was underpowered that—contrary to our hypothesis—we found a positive result (i.e., a significant association between “triple low” and postoperative death). The concern could be that our study was “overpowered” in that a very large study can find statistically significant, but clinically irrelevant, results. However, death is an important outcome, and even a small impact on mortality could be clinically important. (2) We are more sanguine than the letter writers about the potential for propensity score matching to balance risk factors between groups in nonrandomized studies. Nonetheless, we agree that prospective clinical trials are better able, in general, to reveal causal links than observational studies, where confounders are more likely. (3) We agree that “triple low” was less potently associated with death than preexisting risk factors (e.g., American Society of Anesthesiologists physical status and chronic obstructive pulmonary disease). But the salient point is that “triple low” is potentially modifiable or even avoidable (during general anesthesia), whereas patient comorbidities are largely immutable. (4) We are baffled by the contention in point 4 that sensible anesthesiologists currently assiduously guard against epochs of “triple low,” since the arguments put forward in the rest of the letter seem to suggest that “triple low” is not likely to be injurious and that our results reflect statistical “tap-dancing” or clinical irrelevance. Moreover, despite peer-reviewed data demonstrating that intraoperative hypotension is common²  and likely harmful,^3,4  there are no peer-reviewed data demonstrating that clinicians routinely avoid hypotension by administering medications or decreasing depth of anesthesia. Nor have any standard protocols for the treatment of hypotension (e.g., decrease depth of anesthesia, administer fluids, infuse vasopressors, institute inotropes) been established in any surgical setting or in any patient population. The intuition of Dr. Pivalizza et al. regarding the conduct of reasonable clinicians is therefore unsupported.

Carette et al. raise the important point that using non–age-adjusted MAC values might have affected our conclusions. They could be right in that “single high” (age) might be much more important than “triple low” and that some of the patients in our study included in the “triple low” group might only have had “double low” (low mean arterial pressure and low bispectral index) when considering age-adjusted MAC. We chose the methodology in our study to approximate the approach that was used by Sessler et al.,⁵  who chose not to use age-adjusted MAC values. But our findings would not have changed substantially had we used age-adjusted MAC. Based on the population in our study, the low MAC cutoff would likely have shifted from the (arbitrary) 0.8 age-unadjusted value to about 0.9 age-adjusted MAC.^6,7  Furthermore, although age was associated with 30- and 90-day mortality in the multivariable analyses, “triple low” remained independently linked to death despite the inclusion of age as a variable in the models. It is also notable that age was one of the variables used in our propensity score matching.

In conclusion, we apologize if the letter writers or readers were alarmed by our study or our conclusions. We wish to clarify that we do not believe that our findings mandate any changes in clinical practice, and we remain skeptical that “triple low” is causally implicated in postoperative death.