Accounting for Planned Postoperative Intubation

We read with interest the article by Bulka et al.¹  regarding the use of intraoperative nondepolarizing muscle relaxants (NDMRs) and their association with postoperative pneumonia. We commend them for increasing knowledge in an area that is exceedingly important. In this article, postoperative pneumonia occurred more frequently in patients who received an NDMR versus propensity-matched patients who were not administered an NDMR. Furthermore, within the NDMR subset, lack of neostigmine administration was associated with a greater than twofold higher incidence of postoperative pneumonia than their propensity-matched counterparts.

Although not explicitly stated in the article, we wonder why these patients were not routinely reversed at the end of their procedure. As described in the accompanying editorial,²  this may have resulted from concerns of paradoxical muscle weakness and/or other side effects of acetylcholinesterase inhibitors. However, another plausible explanation may be that some of these patients were being transported to the intensive care unit for postoperative mechanical ventilation, thus not requiring NDMR reversal. In our experience, the overwhelming reason for nonreversal is predetermined postoperative intubation regardless of patient demographics, attending anesthesiologist, surgeon, or surgical procedure. Because endotracheal intubation and intensive care unit residence are both strongly associated with nosocomial pneumonia,³  there is a high likelihood that the effect of nonreversal on this outcome is confounded by continued postoperative intubation. To determine whether this manner of confounding exists, separate analyses should be performed that only include patients who were extubated at the end of the surgical procedure before leaving the operating room. Although tedious, these additional investigations would strengthen the argument about the importance of NDMR reversal.

There are also separate issues with the propensity match, in particular with the match for the NDMR/no-NMDR analysis. It can be argued that the biggest determinant of NDMR use is the particular surgical procedure itself, and surgeries that are associated with postoperative pneumonia (thoracotomies, laparotomies, etc.)^4–7  are routinely not performed without NMDR. To control for surgical procedure, the authors used Clinical Classifications Software (CCS; Agency for Healthcare Research and Quality; Rockville, Maryland) groupers in the propensity match. Although there are more than 230 single-level CCS procedure categories, there is still too much variability within certain groupings to provide an adequate representation of the surgical procedure variable for the context of the study. As an example, CCS category 96 (fifth most common CCS code in study), “other OR lower gastrointestinal therapeutic procedures,” includes more than 80 Current Procedural Terminology codes with both laparoscopic and open colorectal procedures. Thus, a laparoscopic case may have been paired with a laparotomy despite the dissimilar incidence of postoperative pneumonia attributable to these procedures.⁴  This is also true for a number of other CCS groupers including category 40, “other diagnostic procedures of respiratory tract and mediastinum,” which includes both thoracoscopic surgeries and thoracotomies with differing inherent rates of postoperative pneumonia.⁵  Although the CCS classifier is inadequately broad in this respect, the authors still were unable to produce a propensity match with an unbiased (standardized difference less than 10%)⁸  surgical procedure variable for the NDMR/no-NMDR analysis. To better separate the effects of the surgical procedure from NMDR use with regard to the incidence of postoperative pneumonia, a balanced match with an adequate procedural variable (e.g., hard-matched Current Procedural Terminology code) must be performed. Although the use of a more procedurally specific type of matching would most likely lead to a decrease in statistical power within a given data set, selection bias with regard to surgical procedure cannot be properly controlled for without doing so.

In addition, variables that are known to be correlated with postoperative pneumonia need to be accounted for in the analysis to better elucidate the real impact of NMDR and neostigmine reversal on this outcome. These include patient functional status, smoking history, and presence of chronic obstructive pulmonary disease.^3,6,7  Although these variables were indirectly accounted for in this study through the American Society of Anesthesiologists classification, a previous investigation revealed that each of these aforementioned factors were still associated with postoperative pneumonia even after controlling for American Society of Anesthesiologists class.⁷  Also, this analysis does not account for the beneficial effects of optimum postoperative analgesia, specifically epidural analgesia,⁹  on the occurrence of postoperative pneumonia. Lastly, several references in this article are erroneous. In fact, all four citations in the second paragraph of page 649 do not confirm the ideas expressed in their respective sentences.

The authors declare no competing interests.