In the last post, Andrew Weatherall went over some of the animal literature touching on and all over the subject of neurotoxicity. If you’ve landed here for the first time, you might even like to check that one out here. Well now it’s time to get on with the next bit – the human bit.
So we’re at the point in the story where we have a problem. It’s quite a big problem.
The agents we use for a fairly crucial part of the anaesthetic triad, the anaesthesia bit, look like they’re bad for neurons. Well, maybe bad for neurons when they are part of an animal. Not just the rodent animals. Also the ones that look a bit more like us.
This is enough to make it fairly important to start looking for evidence in humans. The arguments about animal models and times of peak synaptogenesis are not going to be settled quickly. Plus the stuff that matters to patient, parent and anaesthetist alike isn’t what the agents do to Mr. Whiskers or the rest of rodent-kind. It’s what it means for kids.
Back in med student days, every time something came up involving the question ‘what type of study should you choose?’ more often than not you’d figure out a way to steer it to a phrase including ‘randomised controlled trial’ and fling in a ‘gold standard’ somewhere. It just always seems to be the answer.
Except it’s not the answer that most people reach for when trying to nut out this whole toxic drug thing. For starters it’s not quick. Oh, and you’d have to go through that whole pesky ethics thing where you explain how you’ll tell people the anaesthetic might be not OK and then ask if you can randomise their children to one of two groups and then you have to recruit the patients and then you need multiple years of follow-up and … I’m sorry my eyes are watering at the thought of it.
Cohort studies have a bunch of things to recommend them if you want slightly quicker answers covering things where the difference between groups might be pretty small. All you need is vast databases and a willingness to sift through multiple thousands of those individual files and … wait, my eyes are watering again. It just hurts so much.
It’s lucky for the clinicians amongst us that there are people prepared to do that hard work. For a number of years people have kept putting out paper after paper trying to tackle this question and most of them are cohort studies. And if you break it down they pretty much fall into one of a few groups:
1. Yes, there’s things here
Some of the cohort studies indicate that there really is an association to worry about. Even more than that, some of them indicate the sort of dose-response relationship that should deep worry us.
This generally works out as “we got heaps of records and a much smaller subset had anaesthesia, and then we checked results and things”. So for Wilder et al. it was a cohort born between 1976-82. There were 5357 kids in total and 593 had an exposure. And sure most cases involved halothane and nitrous oxide but the targets are all the same, right?
And they found some sort of association. It was multiple exposures, not single procedures, but there was an association worth following up they thought.
The same is true of Hu et al. (who followed up in a similar cohort) in 2017 (instead of 2009 like Wilder and crew). Of course this time the techniques were probably a little more modern when it applied to these under 3 year old patients tracked down through historical records. Multiple exposures were associated with more learning disabilities and ADHD, although it should be mentioned that the overall rates of learning disability were less than in Wilder’s paper.
That ADHD association also comes out in this paper by Sprung et al. where exposure to general anaesthesia before the age of 2 was associated with a higher rate of ADHD. Though it’s worth noting that kids having general anaesthesia were also more likely to be male, have lower birth weight, lower gestational ages, and more comorbidities. But more about confounders in a bit.
Of course one way to manage those confounders is to look at matched siblings. Which is exactly what DiMaggio et al. did when they showed a higher rate of learning disability when exposed to general anaesthesia. That said, they looked at a total of 304 exposed kids.
What, another qualifier?
Yep. It’s qualifiers all over the joint.
2. No, we can’t find anything here. Or at least the confounders are confounding and effect sizes are limited.
It would just be too easy if everyone found the same thing, right? When Hansen et al looked at test results for Danish adolescents and comparing those post-inguinal hernia repair (under age 1) and those free of groin bumps through their life they did find some concern about lower performance on those test scores. Except the influences of sex, birth weight, parental age and education were all more significant. When you did the adjustments there was no difference. Huh. Population medicine just gets more and more important.
Bartels et al did something even more powerful. They looked at twins. I say ‘more powerful’ because look we’re all scared of identical twins, right? Probably mostly because too many of their parents dress them in the same clothes. Identical two of anything is scary when it’s anything other than food. Plus they were Dutch so I bet they’re tall. Scary tall people wearing matching sailor’s outfits I bet.
And look, the twins in the exposed to anaesthesia group can definitely have issues. But in this study any learning difficulties were the same as their twin. Is it nature, or nurture? Well, it seemed to be ‘not anaesthesia’.
Sun et al? Siblings and just the 1 exposure for hernia repair – no associations with anaesthesia.
And then there are the Swedes. With the daunting efficiency of a flat-packing nation of Volvo drivers Glatz et al looked at a cohort of all children born between January 1973 and December 1993. That’s 2,174,083 kids of whom 33514 had anaesthesia before the age of 4.
And look they found some difference in mean school grades at 16 years old and IQ test scores. 0.41% difference in the former and 0.97% in the latter. Multiple exposures only showed the same magnitude difference. And the overall difference was much less than differences related to sex, maternal educational level, or even month of birth.
So maybe there was a slight difference but is it clinically meaningful to have some such a small difference? And maybe no dose-response. Things aren’t getting clearer.
3. Results are crazy. Like, so crazy is this even a thing?
In 2016 a couple of papers came out that seem particularly muddy. Graham et al looked at kids who had GA under the age of 4 and looked at an educational assessment, the EDI. They included 3850 kids with a single exposure and 620 with multiple exposure. The comparison cohort numbered 13586.
For kids under 2 years of age there was no independent association between exposure and EDI results. If there was a single exposure between age 2 and 4, there was an association. But if they had multiple exposures there was no association.
So no association in the younger, presumed to be at most risk, age range. But some association with a single exposure when they’re older? And multiples mean nothing?
The authors took this as a sign that confounders explained the results, implying that any effect must be not able to dwarf those confounders.
O’Leary et al also used the Early Development Instrument. They looked at 28,366 kids who had surgery before EDI completion and matched them to 55,910 unexposed kids.
They expressed their findings in terms of developmental vulnerability. Their results report increased vulnerability in the exposed group as long as they were exposed above the age of 2, with an adjusted odds ratio of 1.05 (the confidence interval was 1.01 to 1.08).
However for this study kids exposed under 2 years of age again showed no increased risk of vulnerability with an odds ratio of 1.04 and confidence intervals of 0.98-1.10.
So why would there be a greater risk if exposed later, unlike all those other studies?
Isn’t the evidence supposed to make things clearer?
When you take a step back it just seems like despite all the numbers, and all the work, the associations just aren’t consistent.
And if they are consistently inconsistent, does this mean what we’re reading is a variety of justifications to try and make statistics seem like something you can trust?
What a pity no one is crazy enough to do a randomised trial, right?
Well actually, they are.
Sleeping and Waking
The GAS study was driven out of Melbourne and randomised infants to either an awake-regional option or general anaesthesia for inguinal hernia repair. The kids had to be under 60 weeks postconceptual age when recruited.
The aim is to then follow-up and report on developmental outcomes with a variety of scores at the age of 5. Except they’ve reported on the Bayley-III assessment at a couple of years old. They obtained results for 292 kids in the regional group and 295 in the GA group.
And they found things were basically equivalent. And while it’s true that the more useful assessment will come at the age of 5, equivalence after exposure to a single, short anaesthetic is still something to hang your hat on. If a formerly flat-packed hat rack isn’t available of course.
Sweet, Sweet Clarity
That’s what we came for, right? The humans were supposed to speak more clearly than the animals.
And clarity is what we absolutely didn’t get.
On the plus side there are a number of studies, both cohort and the one RCT that’s available, suggesting that a single short anaesthetic is not associated with learning disabilities or other cognitive issues that could be described as anything other than subtle like a tortoise with a particularly excellent poker face.
OK, I actually don’t know what a poker face looks like on a tortoise but I think there’s a good chance it’s hard to pick up when they’ve got a great vs humdrum hand.
The other side of it is that there are studies that raise concerns when it comes to multiple exposures. But even in the cohort studies the numbers in the exposed group who have had multiple experiences drops lower and lower and lower. It’s hard for a rube like me to figure out whether the numbers are even enough to be likely to pick up an issue with those repeat sleeps.
And we don’t really have an answer when it comes to duration of exposure. Damn.
And to make it even more complicated, the grading systems and rating scales aren’t always the same, and it’s hard to pick up whether the chosen scales are even the right scales. I mean, why isn’t everyone using the same scale?
So we need more studies, and more answers. Or to get closer to answers.
So what do we do now? We have to give anaesthetics before the next paper, and the paper after that and the one after that and the … you get the idea. And what should the next questions be, anyway? Should we keep looking at the same questions or research different things entirely on the topic of neurotoxicity?
For the clinical anaesthetist there’s are other calculations to make – what’s the best choice to support safety? What do we say to kids, families and colleagues? And how would we adjust our technique, just in case, if we thought that was necessary?
But those are things to tackle in another post, on another day. I sort of want to find a tortoise with a deck of cards right now.
Again, this isn’t an exhaustive review of the literature. There will be others and feel free to throw more in the mix or dispute the mentions of them in the post. But here they are:
Hu D, Flick RP, Zaccariello MJ, et al. Association between exposure of young children to procedures requiring general anaesthesia and learning and behavioural outcomes in a population-based birth cohort. Anesthesiol. 2017;127:227-40.
Graham MR, Brownell M, Chateau DG, et al. Neurodevelopmental assessment in kindergarten in children exposed to general anaesthesia before the age of 4 years: a retrospective matched cohort study. Anesthesiol. 2016;125:667-77.
Davidson AJ, Disma N, de Graaff JC, et al. Neurodevelopmental outcome at 2 years of age after general anaesthesia and awake-regional anaesthesia in infancy (GAS): an international multi centre, randomised controlled trial. 2016;387:239-50.