Hello, and welcome to today's Journal Club Critical Care Medicine webcast. The webcast hosted and supported by the Society of Critical Care Medicine as part of the Journal Club series, Critical Care Medicine. The webcast features two articles that appear in the June 2022 issue of Critical Care Medicine. The webcast is being recorded. The recordings will be available to registrants on demand within five business days. To access, simply log into mysccm.org and navigate to the My Learning tab. My name is Tony Gerlach, and I'm a clinical pharmacist at Ohio State University Medical Center here in Columbus, Ohio, and I will be moderating today's webcast. Thank you for joining us. Just a few housekeeping items before we get started. There'll be a question and answers or Q&A session at the conclusion of both presentations. To submit a question throughout the presentation, type in to the question box located on your control panel. If you have comments to share during the presentations, you may use the question box for that as well. And finally, everyone joining today's webcast will receive a follow-up email that will include an evaluation. Please take the five minutes or so to complete the evaluation as your feedback is greatly appreciated. Please note the disclaimer stating that the content to follow is really for education purposes only, and the views, opinions expressed herein are those of the presenters and do not necessarily reflect the opinion or views of SCCM. SCCM does not recommend or endorse any specific task, position, product, procedure, opinion, or other information that may be mentioned. And now I would like to introduce both presenters for today. Dr. Lucy Madra is an intensive care consultant at the Austin Hospital in Melbourne in Australia, and she's a senior clinical fellow at the University of Melbourne. She's also undertaking a PhD examination examining the sex differences and illness severity outcomes and treatments of adult ICU patients. Next we have Professor Loren Schlapbach, who's the head of the 41-bed multidisciplinary pediatric and neonatal ICU unit at the University Children's Hospital in Zurich, Switzerland. He has published over 200 papers in the field and has received over $18 million in grants for his research. Professor Schlapbach is an internationally recognized expert in sepsis and other life-threatening infections and inflammatory conditions, and he has worked on such groups as the head of the Pediatric Surviving Sepsis Campaign and is co-chairing the International Pediatric Sepsis Definition Task Force and is also part of the Global Sepsis Alliance Executive. I thank you both for joining us. Now I'll turn the presentation over to Lucy. Thank you, Tony, for that introduction, and thank you for the invitation to speak today on our paper on sex differences in the treatment of adult intensive care patients, a systematic review and meta-analysis. So regarding disclosures, my co-author Dr Vipin Aghi and I are both members of the Women in Intensive Care Medicine Network, which is a committee of the Australia and New Zealand Intensive Care Society. This is an unpaid role. And my co-author, Dr Elisa Higgins, is supported by National Health and Medical Research Council Australia Emerging Leader Fellowship. The authors have no other funding to declare and we've got no financial conflicts of interest. And as I begin today, I'd like to acknowledge that I'm presenting from the unceded sovereign land of the Wurundjeri people of the Kulin Nation here in Melbourne. I'd like to pay my respects to their elders and I extend that respect to any First Nations people here today. So I'll begin the research presentation with what appears to be a pretty uncontroversial statement about our practice in ICU. And that is that the treatments we deliver are delivered according to the physiological requirements of the patient, with reference to the likely benefit to the patient, according to the patient's own values and preferences for treatment. However, if we take even a fairly cursory glance at the literature, we can see that there are several studies suggesting that men receive more organ support in the ICU than women. And this is including across the domains of mechanical ventilation, tracheostomy and even ECMO. Of note, this is not a consistent finding. So there are a couple of studies reporting that women receive more mechanical ventilation than men. And therefore, we felt it was important to undertake a systematic review of the subject. So our objective in this paper was to synthesise and evaluate studies of sex differences in ICU treatment and explore sources of heterogeneity in the literature. Our full methodology is available in the paper and therefore I'll just point out some important key aspects today. And regarding the inclusion criteria, we've included observational studies with broad populations of adult ICU patients. That is to say that we specifically excluded studies of individual diagnostic cohorts of critically ill patients, for example, the sepsis or the trauma population. And we also included studies that explicitly examined the association between sex or gender and at least one of illness severity, mortality and ICU treatment as their stated primary or secondary objective. And this review yielded such extensive findings that we've actually published two papers. So the paper I'm discussing today and a separate paper looking at sex differences in illness severity and mortality that was published last year. We assessed risk of bias using the Newcastle Ottawa scale for observational studies and you can see the domains of bias that that scale considers here. And regarding our statistical analysis, we used a random effects model to calculate pooled odd ratios for dichotomous outcomes. So that is for the use of mechanical ventilation and use of renal replacement therapy and to calculate pooled mean differences for the continuous outcomes, which were the length of stay and duration of mechanical ventilation. And I'll just emphasise that we compared women to men throughout this statistical analysis and therefore odds ratios that were greater than one indicate that women were more likely to receive the treatment than men. And conversely, odds ratios less than one indicate that women were less likely to receive treatment than men. And similarly, mean differences of greater than zero indicate that women received a longer duration of therapy than men. So our search of the Medline and EMBASE databases identified over 4,000 unique records, which we've screened at title and abstract. And this yielded a total of 91 articles that we screened in full text. I should add that all of our screening we completed independently and in duplicate. And overall, we found that 21 studies fulfilled our selection criteria. And these studies are summarised here in this Table 1 and they included 11 studies from Europe, seven from the USA, two from Canada and one from Taiwan. Which is to say that these are really all studies from high income countries. Of note, there were none from Australia, which is my neck of the woods, none from Africa or South America. You can see there are some small single centre studies sitting alongside the larger registry based studies and the eight registry based studies accounted for 90% of participants. This one study here by Mahmood and colleagues in the United States, this study contributed nearly half of participants in the meta-analysis. The other thing I draw your attention to is the sex balance across the study populations. So the percentage of female patients within each study population ranged quite widely from 31.3% to 53.1%. So regarding our participants, we had over half a million participants in the pooled study population of whom 42.7% were women. Overall, the women tended to be older than men and there was a slight tendency towards women having higher illness severity scores than men. This slide summarises our risk of bias assessment and risk of bias was most common across three domains that I'll highlight. So first, representativeness. As I just mentioned, there were several single centre studies and many of these were judged to be at high risk of bias or being poorly representative of the average ICU patient. Next comparability. So there were several studies that did not adjust the treatment outcomes for at least two important confounders. And this was often because the studies considered the treatment outcomes as a secondary outcome or secondary objective and rather considered sex differences in mortality as the primary objective of their paper. The final thing I'll point out is this ascertainment of exposure domain, which for this study really looked at the question, were the patients men or women? And actually, for several studies, it was judged unclear how the study defined sex or gender, nor how they ascertained the sex or gender of their patients. And this is notable because these were studies that were primarily about sex or gender. We're examining that as their primary exposure of interest. So I'll move on now to the forest plots or pooled estimates for our study. And just to orient you as I start, in all the forest plots I present, estimates on the left of the reference line indicate that women were less likely to receive the intervention than men. And those on the right indicate that women were more likely to receive the intervention. So for mechanical ventilation, there were 12 studies with a total of over 370,000 participants reporting the use of mechanical ventilation. And the pooled odds ratio for the use of mechanical ventilation was 0.83. That is, women were less likely to receive mechanical ventilation. So we calculated separate pooled estimates that included only results adjusted for important confounders, and we'd predefined that as meaning that they were adjusted for two or more of age, illness severity, comorbidities, and diagnosis. And we can see that in this adjusted analysis, women were less likely to receive mechanical ventilation. And this was across six studies with over 100,000 participants. Regarding duration of mechanical ventilation, there was no significant difference for men and women. And for renal replacement therapy, there were eight studies with nearly 100,000 patients, so 99,700 participants reporting sex differences in renal replacement therapy. And the pooled odds ratio for renal replacement therapy in women compared to men was 0.79. So again, reflecting that women were less likely to receive the intervention. In the adjusted analysis of renal replacement therapy, there were five studies with over 94,000 patients. And women in this pooled analysis, women were still less likely to receive the intervention. This forest plot represents the mean difference in length of stay in the ICU in days. And therefore, the pooled estimate down the bottom indicates that women stayed in the ICU approximately six hours less than men. So a small but significant difference there. However, we found no difference or no sex difference in the hospital length of stay. And briefly, we undertook sensitivity analysis that excluded studies at high risk of bias in any domain. And our key findings regarding renal replacement therapy and mechanical ventilation were unchanged. So in summary, we found 21 studies with over half a million patients in totals from ICUs in predominantly in high-income countries. We found a fairly consistent finding that women received less mechanical ventilation and renal replacement therapy and had shorter length of stay in the ICU compared to men. And these key findings, so there really was substantial heterogeneity and risk of bias across the studies, but our key findings persisted in studies at low risk of bias. So even once we'd excluded those studies that were considered at higher risk of bias. So how do our findings in this paper compare to the results of studies of different populations in medicine? I'd like to consider a couple of examples. So in the field of nephrology, women with chronic kidney disease are less likely to receive dialysis than men. And this is attributed to both physiological and psychosocial factors. So women experience a slower trajectory of decline in renal function than men. And also women who need dialysis are more likely to actually refuse the therapy. And then in cardiology, which is the field of medicine in which sex differences are probably best studied, women are less likely to receive primary reperfusion. So a trip to the cath lab if they present with an ST elevation myocardial infarction or cardiac arrest. They're even actually less likely to receive some simple drug therapies like antiplatelet agents or statins. And for women who have, or in the management of advanced heart failure, women are less likely to receive some of these advanced therapies, including ventricular assist devices. And again, the reasons for this is, you know, are myriad. Reperfusion therapies in the setting of coronary artery disease are relatively less effective in women than men due to differences in the coronary artery anatomy. And clinician bias likely also plays a role with medical students taught about typical chest pain, which I guess some might say could be loosely translated as men with chest pain and atypical chest pain, which might be loosely translated to women with chest pain. And these terms have been significantly challenged in the literature recently. So it seems that our finding of significant sex differences in the treatment of ICU patients is actually part of a broader phenomena in medical treatment, rather than one that's isolated to critical care. So a key question arising from all this is why? Why do women receive less treatment in ICU than men? And I'd like to discuss three possible answers to that question. So first, in intensive care medicine, our treatments are often prescribed based on physiological parameters that are not sex adjusted, you know, or in fact, even age adjusted or weight adjusted. However, there are significant sex differences in the normal ranges of physiological, of many physiological parameters. An example of this is the normal range for blood urea nitrogen, which is higher in men than women. And that relates to the higher muscle mass on average in men than women. And so if, as clinicians, we're prescribing renal replacement therapy for uremia or azithemia, we'd be more likely to prescribe renal replacement therapy for men than women. And so that's quite a convincing explanation, but it doesn't actually adequately explain the other treatment outcomes that we've considered. For example, the sex differences in mechanical ventilation use or length of stay. And as a counterexample, for much of the lifespan, or up to the age of about 70, the average BP, blood pressure for a woman is lower than for a man. And so if patients are staying in ICU for monitoring or management of hypotension, this would likely lead to a longer ICU length of stay for women compared to men, rather than what we observed, which was a shorter length of stay for women compared to men. So a second possible explanation is that there was a higher rate of limitations of medical treatment orders in women than men, also referred to as limitations of life-sustaining treatment. When we looked at the studies in our meta-analysis, only three reported on such treatment limitations and there were no sex differences detected there. So we can't actually answer the question based on data from the included studies. There's certainly a systematic review by McPherson and colleagues that looked at limitation of treatment orders in the critically ill and that found that women were more likely to have treatment limitations than men. However, Rubio and colleagues undertook a study of treatment limitations and they found that most treatment limitations were created by clinicians in response to the patient's prognosis and the perceived benefits of treatment, rather than being created directly to reflect the patient's preferences for treatment. And so therefore, even if we did find that the women in these studies had a higher rate of treatment limitations than men, this wouldn't necessarily reflect a sex difference in the patient's preferred treatment intensity. Instead, it would still lead us back to clinical decision-making, you know, the decisions of clinicians treating patients in the ICU. And that brings me to a third possibility, that systemic bias may be contributing to the differential treatment received by men and women in ICU, and particularly the unconscious cognitive bias of clinicians. And this possibility is always quite difficult to pin down or prove. There's certainly a nice systematic review of unconscious bias among healthcare professionals that found that we have similar implicit biases as the general population, and that these biases impact upon our clinical decision-making. Looking at the potential role of cognitive bias in critical care decision-making, Larson and colleagues have undertaken a very large written survey of over 1,000 critical care doctors from 75 countries, in which critical care doctors were asked to consider matched male and female patients for admission to the ICU. And they found no evidence that gender bias impacted upon the decision whether to admit a hypothetical patient to the ICU. But again, this is actually a recent study in Nature Communications, that's by Sentola and colleagues, that actually asked clinicians to assess hypothetical patients presenting with chest pain that were played by actors. So there was a black female actor and a white male actor who used a match script to describe their symptoms. They actually wore matched clothing or outfits and were sitting even in a sort of identical chair and environment to try and emulate each other as closely as possible. And they found that the black female patient played by an actor was more likely to receive unsafe undertreatment than the white man. And this might highlight that bias responds to visual clues, and therefore is more likely to be detected in situations that more closely approximate real life. So another important question arising from our paper is, what about mortality? If we get less treatment than men, how does this impact upon their relative mortality outcomes? And I have to say their answer is not clear. I'll present here the forest plot of our related meta-analysis that looked at the ICU mortality of women compared to men. And so this is an overlapping but not identical population of studies as that that's included in the CCM paper. And we found that women were more likely to die at two time points, and that is by ICU discharge and at one year. However, at the other two time points reported at hospital discharge and 30 days, there was actually no sex difference in mortality. So I guess as I just said, it's really not clear. It is possible that women are being harmed by undertreatment in the ICU. It's also possible that in some instances, women are relatively benefiting from undertreatment. For example, women may be benefiting from quite a conservative approach to commencing renal replacement therapy. And finally, and importantly, it may be that women and men respond differently to some of these treatments in ways that we've yet to define. I'd like to highlight a few of the important limitations of our meta-analysis. The first is that we've considered patients who have already made it in the door of the ICU. We examined patients who were admitted to the ICU, and this is not necessarily the denominator of all critically unwell patients, for example, admitted to the hospital. And therefore, we cannot speak to the question, you know, do women and men have equitable access to ICU? Death, of course, is a competing risk for length of stay in the ICU. So our finding that there was a shorter ICU length of stay may reflect or be linked to the fact that women had a higher ICU mortality than men. And finally, of all the studies we examined, they all considered a binary definition of sex. And therefore, and there was no study that considered the non-binary population, and we we can't comment on any findings for that population. So in conclusion, I'd like to return to my, you know, initial apparently uncontroversial statement and that treatments in the ICU are provided based on physiological requirements, the likely benefit of treatment and consideration of the patient's preference for treatment. Our study shows that actually patient sex impacts upon the treatment we deliver even after adjusting for illness severity. And this begs the question, what other demographic or social characteristics may be impacting upon the care of the critically ill? And I think this is quite an urgent question for the critical care community to consider and certainly one that's worthy of further research. And I'd like to end just by acknowledging once again, my co-authors on this paper. Thank you very much. So there are my references. Sorry, I'll briefly. And I'll now pass over to my colleague, Dr. Schlapbach who'll be presenting his paper. Thank you very much. Thank you, Lucy, very much. It's been wonderful to hear this exciting and very thought provoking data. We will move to the second paper here. My name is Lorenz Schlapbach, I'm an intensivist and I had the opportunity to work 10 years in Australia during which as well this work here was created. And that work was done on behalf of the Australian New Zealand Intensive Care Society Center of Outcomes Resource Evaluation and has been endorsed as well a bit by the Australian New Zealand Intensive Care Society Pediatric Study Group. So I'm talking on behalf of a large group of colleagues who have helped to make that work possible. The focus of this paper which has been published as well in this month in CCM is to assess educational outcomes of children who were in the ICU before age birth. And I will show you a linkage study which is based on population-based data from the state of Queensland in Australia. In terms of disclosures, the study has been supported by a number of grants including Intensive Care Foundation Australia as well as an educational rise and ground from the Department of Education in Queensland and the Children's Hospital Foundation. And I've been supported during the work as well by NHMRC Petition Fellowship and the Foundation. There's no other conflicts of interest to disclose by any of the co-authors. So in ICU as we all know, in a mortality is essentially one of the major benchmarks. And we have excellent tools to measure outcomes, to benchmark outcomes, to assess progress over time. But the focus on actually what happens after ICU has come more and more into actually our attention. What we know from neonatal studies already from as early as the 80s and 90s, in outcome at five to 10 years of age may be quite different to what it looks like early. But in most of the ICU literature, long-term outcome is essentially something which is considered at six to 12 months after discharge. Whereas most children coming to ICU they're less than five years of age. So currently most of these children will have a theoretical life expectancy of somewhere around 75 years. And so what we're looking at is actually a very long-term trajectory of the critical illness. And I think this is a really key question because it is not certain that the approach that we're increasingly applying to assess long-term outcomes for adult ICU survivors, to what degree they really meet the requirements to assess the true impact of pediatric critical illness. As we know, a large proportion of adult critical illness happens in the last decades of life, life sort of towards the end of life. And what's been shown in a number of studies is that life expectancy after adult critical illness on average actually is in the realm of a number of years. Whereas for the pediatric ICU survivors, as mentioned, we look at decades of life ahead. And which of course has an impact not only on the surviving child, but it has an impact on the parents, on siblings of this child, as well as a potential future offspring. So there is a multiplying effect. And what parents often ask us is, how will my child do later? They often say, we want our child to survive, but as well, we hope that our child one day can go to school or learn a profession. Most parents say, we don't mind if our child is smart and doing well in school, but if we know that our child one day can go to school and learn some profession, that would make a big difference for us. But so the question really is, how do we assess this? And there's actually very few long-term outcome data for ICU survivors, which go beyond the 24 months. You know, Ivashnina's landmark paper, which used large linked data as well, assessing patients that had been in hospital or in ICU with sepsis, showed for the first time in a large group of adults, there's really very impressive substantial long-term cognitive impairment and functional disability, which has a major impact on sepsis survivors. But actually the scarcity of data in pediatric ICU or adult ICU really contrasts with what the neonatologists have been doing. Almost 20 years ago, the study from Neil Marlow already assessed outcomes at early school age. And they've been able to show with these large cohorts of children that have been followed up now into even old groups, they've been shown that early prediction of outcomes, such as using Bayley scores done at six to 12 months after neonatal critical illness, both over and under estimate the outcomes at school age. And the question really is now, how can we assess outcome at school age? And as you know, the Society of Critical Care Medicine, you know, led from Erica Fink, has developed a core outcome set for pediatric critical care. And I think this is a extremely important framework for future research and future benchmark and quality control as well. And it really highlights the importance that we consider outcome on a more comprehensive scale, which includes as well cognitive aspects and different domains of functioning. But what is known is that many units actually do not have follow up projects in place for such. And it is quite expensive and resource intensive. And so a key question really is how should we do follow up for which patients should we do follow up and which are actually outcomes which matter. We did the Delphi study in Australia, New Zealand, and the number one outcome that clinicians said, you know, should be prioritized was survival with good functional neurological outcome. But when we rated the feasibility of doing such, it was actually ranked quite low. And so the question behind this study was this, can we use school data to essentially assess how children which were in ICU before their fifth birthday do once they reach primary school? And can we do this in a population based approach using a linkage of mandatory data set rather than to essentially have a prospective cohort? And to see as well, how do children that have been in ICU differ from matched population controls, and as well to see which are predictors of poor educational outcomes. So Australia has what is called NAPLAN, which stands for a National Assessment Program, which is something which is done across all states in Australia, which was started in 2008. And it is performed in school years three, five, seven, and nine. So essentially when children are about eight, 11, 13, and 15 the children they come on the same day in the whole country, the whole states, they come to do this test. They get scored across five domains, which include grammar, numeracy, both of which are considered the most relevant ones in terms of educational impact, as well as reading, spelling, and writing. And the Australian curriculum has defined cutoffs which define a national minimum standard. And these cutoffs are based on the experience that students scoring below these cutoffs, they will usually really struggle actually to perform even minimally in school. And these are usually students that will struggle as well later on, you know, to learn even a relatively basic profession. As well, there is the option for teachers to say a student has significant disability, which interferes with actually being able to do the test, for example, you know, not being cognitively able to, you know, to write or read. And these are classified. And so what we have done is essentially is we've looked at different databases and looking at what you can see on the top of the graph, we thought this, you know, sort of a lifetime journey starting with birth, a child then develops critical illness, comes to PICU, maybe hospitalised once, maybe even several times. And at school age, then actually we'll have the school test and a proportion of patients, you know, may not survive until the school test. So the core database which we used was the Australian New Zealand Pediatric Intensive Care Registry, which is a mandatory database, which has been held since 97. We link this as well with census data and school data, which includes socioeconomic data, aspects on such as remoteness, as well as, you know, the mentioned NAPLAN data from the education department. And we assessed as well the number of patients that died in using linkage with the death registrations. Then, so the study was limited to patients admitted to the ICU in Queensland, Australia, which has a population of about 5 million. So about a pediatric population of about 1 million across nearly 20 years from 98 to 2016. And we used a linkage by, done by the Queensland Statistical Service Branch based on probabilistic linkage, having the sex, date of birth, postcode, the facility information, date of admission and name. We were then able actually to access as well, data from all children that had not been in ICU, which had NAPLAN tests done. And we were able to match these one to one for controls, looking at the year of test, but as well the school grade they were in, the birth cohort year, their sex classifiers for socioeconomic status, school as well as aboriginal Torres Strait Island status. We were able to define socioeconomic status by having information on both maternal and paternal education and profession, which was then classified into low, middle and high socioeconomic status. And we defined as the primary outcome, children that scored below the national minimum standard on both the numeracy and reading domains when they were in year three of primary school. And we then performed multivariate models using stepwise backward regression to identify relevant independent predictors. So you can see here an overview of the linkage and how the database, which essentially contained 5,017 patients including. So we selected children that had been in ICU before their fifth birthday, which were in the ICU at the year when they were predicted to later on actually have a NAPLAN test, as mentioned NAPLAN testing was only started in 2018. And we excluded all the children which had died before reaching NAPLAN. That's important because our prediction models were not focused on predicting deaths, but they focused on patients that have survived ICU to see actually who will do well or not well in school. And out of that cohort, as you can see here, that essentially we have a drop here, which is the main potential source of bias. So there was one and a half thousand patients, so just about 23% where linkage was not enabled, which is actually quite similar to some other education linkage studies in Australia. Most of this is due to interstage mobility, because the linkage could only be done at statewide level. And when we looked at the patients which had not been able to be linked, they were slightly younger, slightly sicker, received more ICU treatment and higher proportion of syndromes. And so this is important to see that our cohort probably was slightly biased towards a less sick cohort. You can now see actually just a brief description of the cohort. We looked at, because some children were in ICU several times, we both assessed factors known at the first admission as well as factors essentially representing an aggregate of all ICU admissions before the fifth birthday. And what you can see here is that overall out of these 5,000 patients, they had a median age just below one year. They had a predicted mortality of 2.33%. And as you can see, roughly 50% were elective admissions. The proportion of intubated children was just below 50%, and the average length of stay in ICU, sorry, the median length of stay in ICU was one day. So, what you can see now on the next step is that we first just looked at the crude scores that children had, both the PICU sample as well as the matched controls for these five domains which I've mentioned. And what you can see is that although the difference is not very large, you can see consistently across all five domains that PICUs survivors consistently score close to about 10 points below their matched below their matched controls. And what we were able to show then as well, which is shown in the next slide, is that even when you do strata based on socioeconomic status, so low, medium, high socioeconomic status, which you can see on the y-axis, that you have consistently this difference between roughly about 10 points in the scores between PICU survivors and controls. And we observed this, although the paper is mainly focusing on year three, I'm showing you here some data not containing the paper, that we observed that effect as well in year five, school year seven, as well as year nine. So, we have a consistent effect of lower scores in ICU survivors. But these essentially children that can do the test, they are children which would still be expected to do at least moderately well in school. And so, as mentioned, the primary outcome was essentially the proportion of children which did not meet the national minimum standard in any of these two domains. And what you can see here is that on across all domains, again, there was between five and 10% of ICU survivors that were not meeting national minimum standard requirements. Even, as you can see as well, the confidence intervals do not overlap. So, roughly one in six PICU survivors is predicted not to be able actually to meet the minimum requirements at school age, and which is an excess rate, which is almost double that what it is in the managed control population. And you can see that differences as well found in secretary outcomes, which assessed, for example, that just one of the domains or all five domains. In the next step, we then tried to see actually which, in adjusted models, which diseases were more strongly present. I think that's very interesting. So, you can see the managed controls, which is the red line. You can see the PICU average, which is the gray line. And you can see essentially the ranking of diseases. And so, on the right hand, you have syndromes, which as expected have a very high chance actually to be disabled, or chronic major neurologic syndromes. On the left hand, actually have aspects which are essentially protective, for example, asthma. But surprising as well, something like cancer, as we know, children coming to ICU with cancer can be pretty sick. They actually did relatively well, and diseases like bronchiolitis too, whereas a disease like septic shock, for example, had a neurological outcome at school age, which was substantially worse than the average ICU survivor. So, there's a clear disease effect. But when we assessed all predictors, this is the following which we found. And the results were essentially relatively robust, both in univariate regressions, multivariate regressions, and in some sensitivity analysis, which looked at different outcomes, or which were including different ICU admissions. The first key point really is that socioeconomic status is a major predictor of outcome. And of course, this is not new. This has been shown in neonatal studies. It has been shown in many education studies. But it's very striking how important this is for PICU survivors. And I think this really points to designing new intervention strategies. We have to think of how do we support survivors after the ICU? And how do we make sure this information's reaching, particularly as well, families from lower socioeconomic areas? There's a moderate age and weight effect, you know, with younger children being more vulnerable. I mentioned asthma. As we would expect, you know, major chronic diseases and syndromes have increased incidence. And interesting as well, PIM2, so the Pediatric Index of Mortality, emerged as a very robust predictor as well of long-term outcome. So it's interesting that actually an index of severity, which has been calibrated against mortality, and as we know, does very well to predict mortality, predicts well as well for long-term outcome. And probably that is because there's a severity factor that some patients may survive, which have been predicted to die, but they may suffer some damage, for example, brain damage, during the course of their illness. So in summary, the study has a number of strengths and weaknesses, which I'd like to point out. It is, to the best of our knowledge, by far the largest court reporting on school outcomes after ICU stay. However, this court is relatively historic. And so many of the children that, because they had to be old enough to attend these NAPLAN scores, you know, they were actually from the previous decade. And NAPLAN score has a major advantage, that it was very standardized, used across the whole state, but it has not been internationally validated. And so this, we cannot transform it, for example, to Dexler IQ scales or so on. But yet, you know, the median follow-up time in our study was about 70 years. So it is a very large cohort, and it's a very substantial follow-up time. And as I mentioned, we think we've done relatively careful matching. And the matching, interesting, actually showed that the match controls that it's substantially worse than just the average statewide children. So finally, as well, you know, as I mentioned, there is a risk of bias, because some of the sicker children were not, could not be matched. And so it is possible that we're actually underestimating the true effect. And of course, you know, it's hard to comment on causality, because we cannot do pre-post comparisons, as by necessity, we couldn't have pre-ISU tests, because we wanted to look at preschool children. So in summary, about one in six PISU survivors will be unable to meet minimum requirements at school age, which is almost double what the normal population would have. One can say this is a really good result, you know, because this means that most children will actually do well. And I think that's an important message for the parents. But what's important, we've been able to identify both non-health predictors, particularly in socioeconomic status, as well as health predictors, including disease severity, the type of illness, I mentioned sepsis, as well as comorbidities and syndromes. And so hopefully, you know, this approach in the future could allow us actually to identify children at higher risk. And we wanted to consider whether this could actually be used to select children, which should undergo follow-up or specific early intervention studies. So as mentioned in the beginning, this work, you know, has been done on behalf of the ANZICS core, as well as the ANZICS registry in the pediatric study group. And I would like to thank, in particular, Wojciech Tomaszewski, Christina Blaza, Lance Drainey, Catherine Taylor, and Johnny Miller for this work. Thank you so much for your time and interest. So we would now go to two polling questions before we start the discussion. The first question is, in your institution, in your institution, what is your policy on performing six-month follow-up in children discharged from the PSU? So please state, do you perform follow-up in every patient? Or do you routinely perform follow-up in selected patient groups? So, for example, cardiac patients? Or do you just perform follow-up on an individual case-by-case basis? Or do you never perform follow-up? So now is the time to vote. Okay, so as you can see, so there's a major response. So most sites actually have selected patient groups, which get follow-up, and there is some case-by-case discussion. So that's going to take us to the next question. So if you would have access to a tool predicting long-term outcome, for example, based on the prediction which I've shown you, would your follow-up approach not change? Or would it improve the selection of patients that you decide to follow up? Or do you think actually that information should go to the GP who will care for the patient after discharge? Or do you think that information should just go to parents? Again, you know, please use, please vote. All right, so it sounds like there is potential interest in such prediction models, both to improve the selection of patients, but as well to consider how could that inform GPs which are responsible to care for the patients after. With that, I will finish and I hand back to Tony for this discussion. Thank you so much. Well, thank you very much, Loren and Lucy, both for some great presentations today. I personally learned a lot, and for the audience, just a reminder, at any time you guys can write questions in the question box and I'll ask our two presenters. Now, based on that polling question, I just thought of a question for you, Loren, especially on that prediction model for survivors of pediatric critical care illness. What do you think we need to do better as a group of intensivists to work with our general practitioners to ensure the best outcomes for survivals of critical illness? I think what you point towards is the key link which often is missing, is often patients will be discharged from the ICU to the ward, and then the ward may have a contact with GPs. But information on long-term sequelae, in particular, if they're not very manifest, may actually be missing. And you know, sepsis is a classic example. If the child suffers an amputation due to sepsis, that may be quite obvious and follow-up support structures may be there. But the risk that this child may be subject to school problems or may need some support, you know, may not be there. And there's a high chance that, in particular, children from socially disadvantaged families will have less support there. So the question is, how can we design early follow-up structures which are targeted and which try and compensate some of these mechanisms? We have examples which work, you know, for example, in the field of myocardial infarction or stroke in adults in particular, there's a very well-established rehabilitation pathways and links back from the hospital to the community, which have been shown to be extremely effective. So I think this is where future research and practice improvement should head towards. Well, thank you very much. And this question is really for Lucy. And it's kind of based on some of the data that I do in my own clinical practice, being a pharmacist, is one of the things I always want to assess. Are we using the right dose of the drug for the right patient? And I'm ashamed to say in my practice, it was only about eight years ago that I realized that the standard equation I used to estimate kidney function, the Kalkreuth-Gald equation, was really not studied in women. There are only maybe 4% of the people in there. And even though I took it as, you know, a gospel, I realized that a lot of what we take in the gospel doesn't really affect potential differences between men and women, let alone people that don't identify as men or women. So with that being said, what problems do you think your studies showed, or what do we need to address to really get back at this question of how do we improve the best outcomes in all patients? Yeah, great question, Tony. Thank you. So I think there's a couple of things sort of within that question. And the first is just highlighting the fact that, you know, some of the data that we work with, or, you know, research that we draw upon frequently in critical care medicine and beyond, is based on research in which, you know, women weren't conventionally represented in the study populations. And I guess that links to one of the points I made in my discussion, which is that in some ways, we may not really have the data to know how men and women respond to our different therapies, if we're basing that upon studies that aren't representative of the critically ill population as a whole. So I mean, I think that's one side of things. The other side is really, I guess, highlighting the idea that, you know, one of the conclusions I think you can draw from six different research is that there are some very simple steps we can make towards providing a more personalised or tailored style of critical care medicine. And, you know, I mean, I think practice varies on this, but one example is even sort of weight-based dosing. I think this is something that Loren and his colleagues in PICU would do routinely, and it's variably done for adults. But again, you know, that's one key difference between men and women. Yeah, and I think the renal function example is a really nice one too, just because of, I think it highlights differences between men and women, but also highlights other factors that might be hidden within that. There's like simply things like sarcopenia. So, you know, I think overall we, in order to improve outcomes, then providing a tailored or personalised critical care medicine is clearly the way forward. And I think there's a lot of really complex algorithmic-based models that will be, I think, coming our way in the next decade. And yet a fairly simple step in the meantime is considering sex differences, because, you know, and particularly, you know, when we look at the population of men and women, these are really big populations. So if we can slightly improve, for example, the therapies for one sex or the other, then that potentially could lead to quite big impacts. Well, thank you very much. That gives us a lot of food for thought. And this was a question from the audience, and this one's more for Loren. And here, the question is really a statement, but it said there's a lot of current focus on follow-up PICU clinics, and this actually could be for Lucy as well. Do you think in either one of these, do we need to really focus on follow-up post-intensive care unit clinics in the hospital, or should we go and collaborate more with the GP as a primary care physician? So, I guess this can be extremely healthcare service-dependent. So, for example, in geographically very spread populations, like Australia was a good example, it may not be feasible, actually, to get the patient in the hospital. So, options to work closely with GPs may be a better option. And I think as well, the remuneration aspects can be quite relevant to see which system will actually be accepted and paid for such as well. So, that part, I would think that every is quite healthcare system dependent. The second point is most countries actually do not have well-established follow-up systems in place outside certain clinically defined risk groups, such as, you know, certain cardiac patients or so on. And hence, you know, the possibility to do sort of a risk stratification and say, who's at greater risk, you know, potentially could be quite valuable in reducing the amount of patients that get follow-up, but to ensure that those who get follow-up actually get decent follow-up. And the third point is, and this is sort of extends to what Lucy has shown is, we increasingly are discovering that non-health determinants, such as, you know, sex, socioeconomic status, other factors, actually have a huge impact on health outcomes. And so, one of the questions is, well, how can follow-up clinics be designed to help mitigating some of these effects? Thank you, Mary. Please go ahead. I think follow-up clinics is really interesting one. And one of the things that always strikes me is that they have the potential to serve two distinct purposes. And one, of course, is direct clinical care of the ICU survivor. So, i.e. to benefit patients who have, you know, been discharged from the ICU. And the other is to have a system of providing follow-up of our, the outcomes for ICU patients in order to benefit, for us to improve our care and benefit future patients. And the reason I highlight that is that, you know, one question that always strikes me is coming from a unit that doesn't have a follow-up clinic, they're really really not common where I work, is whether I as an intensive care consultant or, you know, critical care physician would be well suited to working in a follow-up clinic. And in some ways, I wonder if, you know, I would defer that expertise to people who are trained in, for example, rehabilitation medicine and also then primary care, like general practice. And, but I'm also aware that that's, you know, something that is a really varied practice across the world. There seem to be some regions that are quite strong when having follow-up clinics and some that don't have them. Yeah. I agree. I think there's a lot more that needs to be done, especially with these follow-up clinics. And it's a very convoluted and complex problem. But I think one of the things we probably all will agree on is that there needs to be a lot more research done, not only for high-income countries, high-income countries, but especially in low and moderate income countries. With that, thank you very much. And that concludes our question and answer session. And yet again, thank you again to our presenters, Lucy and Lauren, for joining us today and presenting. And again, anyone who joined us for today's webcast will receive a follow-up email that will include an evaluation. Please take about five minutes to complete your evaluation as your feedback is greatly appreciated. And then on a final note, please join us for our next Journal Club Critical Care Medicine series on Thursday, July 28th. That concludes today's presentation.

Critical Care Medicine

treatment differences

ICU

mechanical ventilation

renal replacement therapy

long-term educational outcomes

children

socioeconomic status

follow-up