Sugammadex and postoperative pulmonary complications after upper gastrointestinal endoscopy: a reply

We thank Pearce et al. [1] for their thoughtful appraisal of our propensity score-matched analysis [2]. They raise three issues: the temporal window of the primary outcome; the pattern across secondary outcomes; and residual confounding from unmeasured neuromuscular monitoring. We agree that residual neuromuscular blockade acts principally during emergence and early recovery, and that attributing a late event to an intra-operative reversal drug is difficult. Our 30-day window was not arbitrary: it follows the StEP-COMPAC consensus definitions for pulmonary endpoints, where the respiratory-failure endpoint is within 30 days of surgery [3]. Pearce et al. propose restricting analysis to the first 24–72 h; yet the definitive randomised trial they cite, SNaPP, measured its primary composite until hospital discharge or postoperative day 7 [4]. A multi-day window is therefore the accepted standard, rather than a flaw. Clinically, tracheal reintubation and postoperative ventilator dependence are generally expected to occur early after anaesthesia, making them more biologically plausible consequences of residual neuromuscular blockade than events occurring weeks later. Crucially, because the cohorts were propensity-matched, late events driven by disease progression, hospital-acquired complications or intercurrent illness – the very mechanisms the correspondents invoke – would be expected to occur similarly in both matched groups and dilute any between-group difference towards the null rather than generate one. The secondary outcomes reflect specificity rather than inconsistency. Although not every respiratory outcome reached statistical significance, the significant outcomes were those linked most plausibly to residual neuromuscular weakness. The benefit was not confined to ventilatory failure: lung atelectasis (6.74% vs. 7.76%, p = 0.014) and unplanned ICU admission (8.27% vs. 9.45%, p = 0.009) were both lower with sugammadex in the matched cohort; and non-invasive ventilation was less frequent with sugammadex in the pre-specified subgroup of patients without chronic pulmonary disease (1.93% vs. 2.70%, p = 0.013). These endpoints share a common mechanism: impaired airway protection and reduced functional residual capacity from residual weakness, to which upper airway dilator muscles are particularly susceptible after cholinesterase-inhibitor reversal. The outcomes highlighted by Pearce et al. as unchanged are either infrequent (30-day mortality) or captured by broad, non-specific codes (dyspnoea or hypoxaemia); their null results therefore reflect limited statistical power rather than positive evidence of no effect. Peri-procedural aspiration is, moreover, often missed by administrative coding (much is clinically silent or captured under pneumonia or hypoxaemia codes) so a null association here is uninformative rather than reassuring. Conversely, outcomes with no mechanistic link to neuromuscular recovery, such as postoperative nausea and vomiting and arrhythmia, did not differ. Were a sicker patient cohort allocated to neostigmine systematically, one might also expect differences in outcomes unrelated to residual neuromuscular recovery; their concordant nullity argues against gross confounding. Propensity score matching balanced a broad panel of measured covariates, including chronic obstructive pulmonary disease, asthma, obesity and age, with most standardised mean differences < 0.10 [2]. Matching nonetheless addresses only measured confounding; as noted by Pearce et al., TriNetX records neither neuromuscular monitoring practice nor train-of-four recovery and, as a federated network without standardised cross-institution codes for quantitative neuromuscular monitoring, could not have captured them. Three considerations temper the residual concern. Because of its higher acquisition cost, sugammadex has often been reserved for patients perceived to be at higher respiratory risk. In precisely such high-risk patients, STIL-STRONGER showed a reduction in pneumonia and respiratory failure with sugammadex [5]; any such preferential allocation (channelling) would therefore bias our estimates towards the null, rendering them conservative rather than inflated. The benefit was, moreover, undiminished among patients without coded chronic pulmonary disease. Most importantly, confounding by indication cannot, by construction, affect a randomised trial and SNaPP found the same direction of effect [4]. We agree that a randomised trial confined to therapeutic upper gastrointestinal endoscopy would be welcome. Until then, our findings – corroborated by SNaPP [4] – support routine quantitative neuromuscular monitoring together with reliable antagonism of neuromuscular blockade, with sugammadex an effective option, irrespective of pre-existing lung disease.