More Reader Responses
Myron Yaster MD
From Myron Yaster MD on Inhaled isoflurane for sedation of mechanically ventilated children in intensive care: update
In a recent PAAD (September 10, 2025), we reviewed an article by Miatello et al. [1] which investigated and supported the use of isoflurane for PICU sedation. (download here). Our reviewers (and many of you) had lots of problems with this article, particularly our belief that isoflurane is a general anesthetic and not a sedative, that the comparator midazolam, has fallen out of favor, and prolonged use of vapor anesthetics may be neurotoxic. Several letters to the editor of the journal challenged the results of this article and its implications.[2, 3] The authors replied to these criticisms as well.[4] I’d urge all of you to read this correspondence.
References
1. Miatello J, Palacios-Cuesta A, Radell P, Oberthuer A, Playfor S, Amores-Hernández I, Barreault S, Biedermann R, Charlo Molina MT, Encarnación Martínez J et al: Inhaled isoflurane for sedation of mechanically ventilated children in intensive care (IsoCOMFORT): a multicentre, randomised, active-control, assessor-masked, non-inferiority phase 3 trial. Lancet Respir Med 2025.
2. Bragg EA, Nadkarni AS, Barnes SS, Fackler JC, Kudchadkar SR: Rethinking isoflurane for paediatric ICU sedation. Lancet Respir Med 2026, 14(1):e3.
3. Engel J, Fideler F, Nordmeyer J, Drexler B, Neunhoeffer F: Rethinking isoflurane for paediatric ICU sedation. Lancet Respir Med 2026, 14(1):e2.
4. Miatello J, Palacios-Cuesta A, Radell P, Trieschmann U, Sackey P, Tissieres P: Rethinking isoflurane for paediatric ICU sedation - Authors’ reply. Lancet Respir Med 2026, 14(1):e4–e5.
From Alan Jay Schwartz MD MSEd
In my PAAD “Remembering the Classics: Emotional Trauma in Pediatric Patients Undergoing Anesthesia -A Thing of the Past Still Present” (01/12/2026) here I encouraged everyone to share their “special sauce” for reducing the emotional trauma children may experience during anesthesia care. I’d like to provide 2 recipes for effective “special sauce”.
Lynne Maxwell always sings the early 1900s children’s song, “The Teddy Bears’ Picnic”. Children seem to magically focus not on the mask induction of the anesthetic but rather on fun described when hearing the lyrics:
“If you go down in the woods today
You’re sure of a big surprise
If you go down in the woods today
You’d better go in disguise!
For every bear that ever there was will gather there for certain
Because today’s the day the Teddy Bears have their picnic”
My own special diversion during inhalation induction was to recite Ludwig Bemelmans beloved “Madeline” (my granddaughter’s name) (characterized by the New York Times as “Perhaps the best bedtime book ever written):
“In an old house in Paris
That was covered with vines
Lived twelve little girls
In two straight lines.
The smallest one was Madeline.”
I’m certain, as Myron pointed about when applauding Chris Abajian’s magic tricks, that many of us have exceptional wiles to ease our patients journeys into anesthesia. Please share your successes for others to adopt and send to Myron (myasterster@gmail.com) and he will post in a Reader Response.
From John Dexter, CRNA
I read your article on pediatric adenotonsillectomy here with interest. I was initially reminded of how different anesthetic techniques are, depending upon locales and desired objectives. The three anesthetic protocols evaluated, while sound, are quite different from the induction and maintenance I have used for some years with a very low incidence of sequelae. I say this not as criticism of your excellent article, but rather as a query: Has enough study been done on what IS the optimal approach to adenotonsillectomy??
In Defense of Randomization
Benjamin Y. Andrew, MD, MHS, Assistant Professor, Department of Anesthesiology
Division of Pediatric Anesthesia, Duke University School of Medicine
I read the recent PAAD review of Shen and colleagues’ manuscript outlining their randomized trial estimating the effect of intravenous, inhalational, and a combined approach on perioperative respiratory adverse events (PRAEs) in adenotonsillectomy1, as well as the more recent reader response from Dr. Martin outlining concerns about the generalizability of this study. I was struck, in particular, by one of the concluding remarks from this response that this trial is “representative of the artificial world of research and is not easily translatable to the real-world of our clinical practices.”
The concept of inadequate sample representativeness to ensure transportability of effect estimates to a “target population” is an often-cited criticism of randomized trials. On the surface, this argument seems quite reasonable – a study conducted in the exact population that its results are meant to be applied to would seem to be the optimal approach. In practice, however, concern over trial generalizability is often overstated and should not serve to nullify or diminish the results. This is a topic that could fill textbooks, but I hope to summarize the main concepts here -- in defense of the importance of randomization and rigorously conducted trials.2,3
Simplifying the cited study, we can say that this trial was aimed at estimating the effect of treatment A vs. B vs. C on outcome Y. Here, Y is a binary variable (PRAE vs. no PRAE), and the primary outcome model used standard logistic regression to estimate the relative effect of each treatment in the form of an odds ratio. The question of transportability of this relative effect estimate hinges on the presence or absence of clinically and statistically important heterogeneity of treatment effects (HTE).2,4,5 Statistically, HTE represents an interaction between the treatment and another variable of interest (e.g., patient characteristic). If there is not a strong a priori belief that such interactions exist, then, by definition, this effect estimate is transportable to other populations, even those with minimal to no representation in the study sample. If there is no interaction, the relative effect estimate is constant, regardless of other factors. Importantly, this does not mean that the treatment is expected to have the same absolute effect in a new population. In fact, heterogeneity of absolute effects (alternatively termed risk magnification) is an expected mathematical result of applying a constant relative effect estimate to patients with variable baseline probabilities of the outcome.4,6 In any population (original study, new clinical sample, etc.) there will be an underlying distribution of these baseline risks for the outcome, and when the relative effect estimate is applied to these probabilities we generate a distribution of absolute risk reduction (ARR) values for that given population (each patient with an individualized ARR). It is much more likely that clinically important variability exists, and can be effectively modeled, for these baseline differences in outcome risk, than for true heterogeneity of relative treatment effects.7 In the case that interaction between treatment and patient variables is strongly suspected, the statistical power to detect such interactions is substantially lower than for the main effect, requiring a nearly 4x increase in sample size to sufficiently identify.
Put more elegantly by statistician Dr. Frank Harrell:
“RCTs of even drastically different patients can provide estimates of relative treatment benefit on odds or hazard ratio scales that are highly transportable. This is most readily seen in subgroup analyses provided by the trials themselves - so called forest plots that demonstrate remarkable constancy of relative treatment benefit. When an effect ratio is applied to a population with a much different risk profile, that relative effect can still fully apply. It is only likely that the absolute treatment benefit will change, and it is easy to estimate the absolute benefit (e.g., risk difference) for a patient given the relative benefit and the absolute baseline risk for the subject.”
Ultimately, randomization is the gold standard for estimating causal effects. Only randomization can truly ensure exchangeability between groups. This allows us to safely assume counterfactual equivalence – that the observed outcome in each group is an accurate estimate of what would have happened if treatment assignments were swapped. While advanced statistical techniques for causal inference (e.g., propensity score methods, natural experiments, instrumental variables, interrupted time series analyses, etc.) represent improvements over basic observational comparisons, they are each inherently limited in their ability to prove a causal association. One approach rapidly gaining popularity is the “pragmatic” randomized trial, and such a methodology may be helpful with respect to transportability concerns in several ways. Pragmatic trials retain the use of randomization and prospective assessment of interventions and outcomes but allow for more rapid patient recruitment and more cost-effective completion. This may serve to increase sample size to the level required to identify key interactions of interest (and thus true HTE, albeit a rare phenomenon). However, the wider net that such trials cast with respect to patient recruitment and representativeness does not inherently change the transportability of the estimates they produce.
In summary, I strongly believe that randomized trials should remain the gold standard for estimating causal effects of interventions in perioperative medicine. Such trials are the only way to definitively assess causality and to robustly estimate the objective effects of treatments. Estimates of relative efficacy from RCTs are inherently transportable to other populations and clinical practice settings, and, with additional modeling of baseline risk we can use these values to generate informative, patient-specific, estimates of absolute effect to guide our clinical practice.
References
1. Shen F, Zhang L, Wang X, et al. Effect of Intravenous, Inhalational, or Combined Anesthesia Maintenance on Postoperative Respiratory Adverse Events in Children Undergoing Adenotonsillectomy (AmPRAEC): A Multicenter Randomized Clinical Trial. Anesthesiology. Oct 14 2025;143(6):1484-96. doi:10.1097/ALN.0000000000005707
2. Harrell F. Randomized Clinical Trials Do Not Mimic Clinical Practice, Thank Goodness. 2023. https://www.fharrell.com/post/rct-mimic/
3. Rothman KJ, Gallacher JE, Hatch EE. Why representativeness should be avoided. Int J Epidemiol. Aug 2013;42(4):1012-4. doi:10.1093/ije/dys223
4. Harrell F. Assessing Heterogeneity of Treatment Effect, Estimating Patient-Specific Efficacy, and Studying Variation in Odds ratios, Risk Ratios, and Risk Differences. 2019. https://www.fharrell.com/post/varyor/
5. Harrell F. Implications of Interactions in Treatment Comparisons. 2020. https://www.fharrell.com/post/ia/
6. Harrell F. Avoiding One-Number Summaries of Treatment Effects for RCTs with Binary Outcomes. 2021. https://www.fharrell.com/post/rdist/
7. Bradburn MJ, Lee EC, White DA, et al. Treatment effects may remain the same even when trial participants differed from the target population. J Clin Epidemiol. Aug 2020;124:126-138. doi:10.1016/j.jclinepi.2020.05.001