SCCM Pod-524 PCCM: Impact of Neighborhood on Pediatric ICU Outcomes

Host Maureen A. Madden, DNP, RN, CPNC-AC, CCRN, FCCM, is joined by Michael C. McCrory, MD, MS, FCCM, to discuss a multicenter retrospective study evaluating the impact of neighborhood, as categorized by the Child Opportunity Index, on pediatric intensive care unit (PICU) outcomes such as mortality, illness severity, and PICU length of stay. The study highlights the disparities in PICU admissions based on socioeconomic factors (McCrory MC, et al. Pediatr Crit Care Med. 2024 Apr;25:323-334). Michael C. McCrory, MD, MS, FCCM, is an associate professor in the departments of Anesthesiology and Pediatrics at Wake Forest University School of Medicine in Winston-Salem, North Carolina, USA.

*If you are unable to play the podcast please click here to download the file.

Category: PCCM Podcast

Transcript:

Dr. Madden: Hello and welcome to the Society of Critical Care Medicine Podcast. I’m your host, Maureen Madden. Today I’ll be speaking with Dr. Michael C. McCrory, MD, MS, FCCM, about the article, “Child Opportunity Index and Pediatric Intensive Care Outcomes: A Multicenter Retrospective Study in the United States,” published in the April 2024 issue of Pediatric Critical Care Medicine. To access the full article, visit pccmjournal.org. Dr. McCrory is an associate professor in the departments of anesthesiology and pediatrics at Wake Forest University School of Medicine in Winston-Salem, North Carolina. Welcome, Dr. McCrory. Before we start, do you have any disclosures to report?

Dr. McCrory: No disclosures, and thanks for having me, Maureen.

Dr. Madden: It’s really my pleasure to have the opportunity to sit and chat with you today. As we were saying earlier, you never know when you’re going to run into people, and this is a great opportunity to talk a little bit more about you and your research interests and about the article. Can you give me a little bit of background about yourself and what is it that brought you to this publication?

Dr. McCrory: Thanks. I did my training up in Baltimore at Johns Hopkins, and it’s certainly a city of neighborhoods and a city of challenges in terms of areas of deprivation, poverty, different challenges that lead to health inequities. As I’ve been an attending here at Wake Forest, I’ve had the opportunity to get involved with the Virtual Pediatric Systems, or VPS, database, doing some research with them, with their data, and in thinking about that, and also being a part of the Pediatric Acute Lung Injury and Sepsis Investigators, PALISI, Social Determinants of Health Subgroup.

The idea for this study really came out through a PALISI SDOH group meeting a couple of years ago. We were just kind of discussing all of these challenges and health equity and sort of this idea that a significant portion of children’s health outcomes is related to social determinants. As much as we provide great clinical care, and that continues to advance in the pediatric ICU setting, a significant portion of kids’ spectrum of illness, whether they come to the PICU in the first place, maybe how they do in the PICU, and then how they do after the PICU, is likely related to a lot of other factors as well. Depending on the study you look at, not specifically in the PICU population, but across the board, say between 20% and 80% of your health outcomes are related to social determinants, depending on how you define that, and healthy behaviors and access to care and those sorts of things.

Anyway, as we were discussing those things, I was thinking about how we could potentially use these VPS data to help look at that. I got into discussions with some of the leaders from the SDOH PALISI subgroup, Manzi Akande, Katherine Slain, and others who have thought about these things a lot, and we started looking into it.

The VPS historically collects some zip code data, optionally, I believe. When we were looking into it, it didn’t seem like that was going to be possible just to collect across, in other words, there wasn’t going to be enough data across the entire database. So we started talking about making a smaller project, trying to put together centers that were interested in looking at address map to indices of social determinants and trying to utilize that high-quality clinical database to look at how clinical outcomes may differ based on children’s neighborhood and their socioeconomic factors.

Dr. Madden: Okay, so that’s what led you to the Child Opportunity Index, I’m assuming.

Dr. McCrory: I didn’t know a whole lot about the Child Opportunity Index when we first started talking about this a couple of years ago. Thanks to Manzi and Katherine, we were looking into these different area-level indices and the Social Vulnerability Index, the Area Deprivation Index. There are quite a few different ways to look at area-level socioeconomic status, percent of households in poverty, and so forth. We settled on the Child Opportunity Index as the most pediatric-specific one, and I can go into more detail.

Dr. Madden: I wanted to ask you to discuss a little bit more about the Child Opportunity Index, because I really don’t think that there are that many people familiar with it. So tell me, if you could describe a little bit more, why you thought this one, more than you’ve already stated specifically, was going to fit well with your study.

Dr. McCrory: Thanks. Well, the Child Opportunity Index is a 29-indicator, area-level indicator of socioeconomic factors. In three domains, it’s educational opportunity, health and environmental opportunity, and social and economic opportunity. It’s housed up at Brandeis University. It’s at diversitydatakids.org if anybody wants to check it out. Again, I have no formal affiliation with them, but they’ve been super-helpful to me throughout this process, and they’ve helped generate a lot of policy research as well as some medical studies using that.

But it has some aspects of it that are more pediatric specific than the other indices that have been commonly used in adults. For example, proximity to early childhood education centers, third grade reading level, and so forth. Some of the other things like social and economic are quite similar to the other indices like foreclosure rates, poverty rates, and also there had been some studies, especially some using FIS and administrative data, some with smaller center studies.

But we hadn’t seen one that had used a larger clinical database like VPS, and that’s part of why we saw this opportunity to potentially put those two together and try to get some interesting answers or at least insights into how that neighborhood may affect PICU outcomes. Like any area-level index, just because they live in a certain spot doesn’t mean they have the same exact challenges as someone next door to them so it’s obviously imperfect, like any of those.

But another advantage of it is that it is indexed to census tract, which is a little bit more specific to the neighborhood than zip code, for example. There are over 70,000 census tracts in the U.S. I believe, for zip codes, there’s maybe 40,000 or so, so it’s a bit more of a specific area that’s hopefully possibly more indicative of that family’s environment. That was another reason that we thought it would be a good one to use.

Dr. Madden: OK, we’re going to dive into the actual study now. Looking at it, you had set up some purpose to evaluate for the association between the neighborhood, as you spoke about, where the child lives and characteristics and outcomes of PICU admission from a large, multicentered, geographically diverse cohort using that census tract designated COI and clinical data. The items you focused on, really the objectives, were to see whether COI was associated with PICU mortality, severity of illness on PICU admission, or PICU length of stay, and admitted to the cohort of U.S. PICUs.

When you looked at it, first of all, interestingly enough, your retrospective study went from January 1, 2019, to December 31, 2020, and included 15 PICUs across the United States. Talk a little bit about your time frame that you included, because we all know the pandemic certainly altered pediatric ICU admissions.

Dr. McCrory: Yes. When we first started putting this together, it was around the spring PALISI meeting in 2022. We wanted to make sure that we had closed cases. In other words, a lot of centers, including us, sometimes have trouble staying up to date with their VPS cases. We try to get them in as soon as we can. But we wanted everybody to have complete data on the time frame. That’s why we went through the end of 2020, because some places may not have had more recent data.

Now, of course, the data seem pretty old and of course there was the pandemic situation. In our appendix we looked at a secondary analysis of just the March-to-December 2020 time frame because we know that the epidemiology of PICU admissions in the world in general was quite a different place at that time. We didn’t see any significant differences from what we looked at.

But still, we realized that there are a number of reasons it would be interesting to look at more recent data, as well as, as an aside, they’ve also just within the past month released a new version of the COI with now 44 indicators instead of 29. It’s called the COI 3.0. We used the 2.0. Things move fast. What if we had a little bit more recent indices and data and stuff? That’s what moves us forward to continue to do more studies.

Dr. Madden: I was going to say, there’s your next publication. But looking at that, you included 15 PICUs in the United States. Can you discuss a little bit how you chose those PICUs or why that specific number?

Dr. McCrory: We actually didn’t, we expected to have about five PICUs. Again, we were discussing this in the Social Determinants of Health PALISI meeting. We had a few centers that were interested that are VPS members or contributing centers, as well as just had PIs who would be interested in doing the work, which involved pulling the admissions and then having to go back to the electronic medical record to get the addresses and map them to the census tract and the COI.

We also wanted to pull insurance payer and so forth, as well as primary language spoken so that we could get some important covariates there. First, we just sort of used that group and reached out to some people in the VPS group that I’d worked with before who I thought might be interested. Then, during the main meeting of PALISI, Drs. Slain and Akande mentioned it to the whole group and we actually ended up with several more centers, luckily, that were interested.

I unfortunately wasn’t there. I was supposed to be attending remotely, but my son had a GI illness and we were actually in the emergency department. Luckily he was fine but it was a bizarre situation because people started texting me like, oh yeah, we’re excited about this. I was like, I can’t talk right now. Anyway, he was fine, but I just remember it clearly because of that challenging situation at the time.

Dr. Madden: One of the things I wanted to ask: When you dive into the publication, it talks about that the data was really pulled from academic children’s hospitals, meaning freestanding children’s hospitals, is that correct?

Dr. McCrory: Yeah, I didn’t realize, until you were mentioning earlier that you were trying to figure it out, I didn’t realize that I could have made that more clear up front. We would have taken whatever variety of children’s hospitals we could get. We wanted to have some geographic variability or diversity because you can see clearly if you look at our, well, it’s on our figure. If you go to diversitydatakids.org, it’s clear that that Child Opportunity Index varies a lot over the country but, of course, also across municipalities.

But in any case, we wanted to make sure we had some from different areas of the country. Generally, it’s academic children’s hospitals that are contributing to VPS. Not always, but we basically just needed places that had the VPS data and were willing to also map the addresses. And I actually didn’t, like I said, I thought it was going to be more like five to seven centers. I did not realize we were going to get 15 centers and that introduced some additional challenges regarding just coordinating things and finding out differences in how people do things or different challenges that we didn’t anticipate. So we were happy to to get that many together and able to contribute.

But, sorry, to answer your question, no, they’re not all freestanding. They’re all pediatric ICUs that contribute to VPS and either tertiary or quaternary pediatric ICUs but not necessarily freestanding children’s hospitals. A weakness of the study that’s come up a lot with the reviewers and others is, is this really representative of who presents to PICUs across the country and other types of places? We don’t know for sure.

Dr. Madden: OK, well, at least I’m aligned with what some of the other people are discussing because I’ve worked in a variety of different PICUs and can recognize the resources that may or may not be present depending on what type of setting you’re in. I thought it certainly might influence some of your findings. But you did list it as a limitation, so you’ve already recognized it, you and the other authors.

But let’s get to the meat of the study, what you really ended up finding. Really what you found, that children without insurance coverage, irrespective of the COI neighborhoods that they were at, had significantly higher odds of PICU mortality. Let’s talk about that. I mean, there’s such a push for every child to have insurance. We know in recent years in the United States that the goal has been to ensure that everybody truly has access to insurance. But here this sample is already showing you that that’s still not the case.

Dr. McCrory: Yeah, that was interesting and not our main focus. We wanted it to be a covariate and ended up being the strongest association with mortality besides severity of illness. As you alluded to, the very low Child Opportunity Index group had an odds ratio of 1.3, but it was not statistically significant. So it looked like there may be increased severity adjusted mortality in that group but it didn’t reach statistical significance.

Whereas the group with none or missing insurance, was the ratio of 3.5. That’s a really eye-popping increased mortality for those patients. It wasn’t a huge part of the study, but it was, I want to say, 5%. It was over 1500 patients.

We don’t know exactly why those patients didn’t have insurance. In our supplement, we looked a little bit at the demographics of those patients and there wasn’t a clear-cut race or ethnicity or origin of admission or other signal as to who or how these patients might be different from the rest of the cohort. But in talking with at least our social worker here about what she observes when families don’t have insurance, sometimes it’s just entered incorrectly. Sometimes the family doesn’t qualify because they’re above the Medicaid threshold. But according to our social worker, in our population, at least, the most common scenario is that the family qualified but hasn’t applied or has let it lapse. So it may just sort of be a signal.

Again, I don’t know. This is sort of speculation but interesting to look into further. It may just be indicative of a family that has particular difficulties in filling out the paperwork, getting the insurance coverage set up or keeping it set up. That probably translates to other health behaviors, preventive care visits, vaccines, and any number of other things that may influence the child’s health and lead to them ending up with worse outcomes.

Dr. Madden: Yeah, it’s a lot of information to sift through to try and figure out the reasons why. Then we also know that the bureaucracy associated sometimes with trying to obtain the insurance pieces really is challenging. As you’re talking about COI and you talked about education and such, it’s the education of the child, but it’s the education of the parents and caregivers as well.

But now I want to talk about another thing, because as both of us are clinicians in the ICU, when one of the things you did want to look at was mortality, the study had found approximately 2% of an overall PICU mortality from this and the mortality difference of 0.8% higher in the moderate and 0.5 to 0.6 higher in the very low and low groups as compared with the very high group. Particularly, it’s not statistically significant, but you had brought out in your paper that it may be clinically significant. We know that PICU mortality overall isn’t really high, particularly in comparison to adults. Let’s talk a little bit more about the clinical significance of the differences.

Dr. McCrory: Yeah, that’s a great point. Over time, the PICU mortality seems to have trended down, which is good. We hope we’re providing better care. We may also be admitting a different overall case mix. But either way, if you’re thinking about something like below 2% in the high and very high groups and well above in the very low through moderate groups, and you’re looking at 15 sites and we had 50,000 overall admissions, only 33,000 that were index admissions that were able to have their COI mapped. But anyway, that’s a significant number of children. And you’re talking about, again, children. We all know, years of life lost and so forth.

Another really interesting thing that I’d like to mention about that is a study came out last year, since we started working on this, by Slopen Adele in Pediatrics, where they looked at death and childhood. By linking the American Community Survey and the National Death Index, they looked at, do children in lower COI neighborhoods, are they more likely to die in childhood, or in the next 11 years, or is it just kind of they’re having years of life lost later in life? They did find a very similar odds ratio of something like 1.3 for the lower COI groups to die within the next 11 years.

The other thing I’ll mention is that the high and the very high groups were very similar. So it seems like maybe these bins, these quintiles that people tend to use with COI of very low, low, moderate, high, very high; some of our authors, especially Adrian Zurca in our author group, are raising this. Just because they’re divided into these quintiles doesn’t mean that the very high group has it made and the next group is totally different and so forth.

Maybe, and this came up with the reviewers, too, maybe that the high and very high groups sort of have the resources they need. Then there’s a step up because we definitely saw that the moderate group had the highest severity of illness and the highest mortality. It is interesting to see there and it kind of lines up with some of that other data that the high and very high group together had the lowest mortality and lowest severity of illness compared with the other three groups.

Dr. Madden: Yeah. One of the things I was wondering about, when I had looked at this and trying to understand the COI and, as we talked about, it’s the neighborhoods, and have your map up and you can look at it and it’s a snapshot with the 15 PICUs, even though there’s geographic diversity there and you had severity of illness in there as well but, at the same time, diving a little bit deeper, were there more prevalent diagnoses? The one in particular, there wasn’t any discussion if these were all just acute episodic illnesses versus there were underlying chronic conditions and how severe that chronicity may be.

Dr. McCrory: Good point. In our Table 2, we talked about primary diagnosis category and we did see that respiratory diagnoses were more common in the very low group. That’s been consistent with what’s described in the prehospital and emergency department setting regarding lower SES groups, that there’s more often respiratory diagnoses. In this paper, we didn’t dive into it but, you know, asthma, bronchiolitis, some of these common ones and specifically have also been implicated as more commonly coming from very low SES or COI groups. So it does seem to be more commonly the unscheduled admissions, the respiratory admissions.

As far as the chronic complex conditions, we felt like we had so much to jam into this article that we didn’t get into that. We do have the diagnoses for these patients. We have coded in the chronic complex conditions and are working on that as a secondary manuscript. Hopefully we’ll have more information on that soon because we know, as you have alluded to, that a significant proportion, like 50% of our patients, have some kind of chronic complex condition. And we need to think about how that may work out when there are limited resources to get healthcare follow-up or therapies or medications or any number of challenges related to those ongoing health conditions.

Dr. Madden: Right. And now we have started to turn our focus, as you’re showing, to social determinants of health. But we’re also trying to look at outcomes and beyond survival and really what are meaningful outcomes and such. I know you didn’t have the opportunity to delve into this and see them, beyond their admission and such, but all of these elements are playing into the children that we see.

Longer-term mortality in those lower COI neighborhoods seem to have a higher risk of death in childhood. But also functional decline, those that came in and now had a complete change. It’d be very interesting to be able to put those together at some point in time. I know it’s a huge dataset, but to see what correlations truly are there.

Dr. McCrory: Yeah, absolutely. I definitely think that’s where we need to go down the road. We’re trying to understand some of the features carefully from what we have so far but, yeah, down the road, we definitely need to know more about those longer-term outcomes.

We are going to try to look at readmissions because some of these patients that came in, these index admissions, may have later been admitted to another PICU, but we can get some idea of, were they admitted to the same PICU during the study period, and we’re looking at that. It’s going to be severely limited again, getting back to our discussion about the COVID times because PICU admissions were thankfully down and they were different.

So it may be very, very different, but we can at least look at what we have and get an idea. It’d be interesting to see it. We only have the two-year follow-up period, but each child, we know the month they were admitted in, so they’ll have certain months at risk for the rest of the study period and we can look at, were they more likely to be readmitted to the PICU, just as a small portion of what you’re talking about.

Dr. Madden: Right. Then the other thing to think about too, we were talking about the makeup of the population and it was discussed that it didn’t really match the U.S. population in terms of White, Hispanic, and Black. How do you think that impacted some of this? I know you brought it out as a limitation.

Dr. McCrory: Yes, that was an observation by the reviewers, which certainly is valid, that our racial and ethnic distribution didn’t exactly match the U.S. population. We did have over 19,000 census tracts and there’s something like 73,000 in the U.S., so it was like 27% or 28% of U.S. census tracts that we mapped, so we were happy with that. But again, some of it may be reflective of, even though we had good geographic representation in some sense, we had east, northeast, west, midwest sites, we were still in major children’s hospitals and cities. So we really in the future need to look at some of these rural and urban differences or nonacademic PICUs and so forth to try to get a broader representation.

Dr. Madden: Yeah, I come from, when I first started, I was out in a relatively rural environment and people had to travel an enormous distance potentially to find pediatric ICUs. So there’s a huge gap in there. That’s partly where my interest stems from, understanding the level of care that’s available as well as what time or distance, etc., that brings them into the ICU. That’s just me, though.

Dr. McCrory: Yeah. I don’t know if you were one of the reviewers, Maureen, but we got some questions about that too and distance to care. Unfortunately we did know if they came from an outside ED or ICU or floor, but we really had no way of knowing how far they had to come to get to, even though we had looked up addresses, oftentimes they hadn’t come straight to that ED or we may not know for sure. I think there’s a lot of literature about distance to care in the literature regarding births or pregnant moms and so forth and trying to get to care. In this population, though, there’s not a whole lot and that may be an interesting avenue.

Dr. Madden: Yeah. I’ll disclose to you, I was not a reviewer.

Dr. McCrory: Okay. Well, you’re hitting on a lot of the points that they also hit on, so they’re excellent points. Thank you.

Dr. Madden: No, you’re welcome. Our time’s about up, but I wanted to give you the opportunity maybe to bring up some lessons learned and then, moving forward, how you can use the information that you found. I’d love you to try and bring it all together.

Dr. McCrory: Yeah. Well, this was my first time trying to organize multiple other sites for a study and it was quite a bit harder than I expected. It’s a little bit frustrating. The IRBs were okay. The DUAs were really challenging and that’s part of the reason it took us so long and then you end up publishing data that are over two years old, so that was frustrating. Even though we were all using the same database, VPS, there’s differences in what statistical software, what kind of support they had to kind of move the data along or clean it up or have it ready, you know, map the addresses and so forth.

The census geocoder is a nice free online resource, but it took some doing to get that to work. I would say that the COI mapping is pretty easy. The team there is really great. I learned a lot of lessons about the FIPS geocoding, the 11-digit census tracks, about how to use the geocoder and the COI data. I think PALISI was super awesome as just a grassroots research network for recruiting centers. If you want to find people who are interested in all kinds of different things and willing to spend their time working on these studies, it was just awesome to get all that help from my collaborators and all these centers.

I would say how to use the data going forward, I think that’s a big challenge. It’s hard for us as clinicians to change the world or change how insurance works or anything like that. But I do think it’s, more and more, over the past decade, we’re realizing how much our clinical care, while very, very important, is a portion of a child’s continuum of health. And it’s so important, if we’re thinking about equity for children, to have the tools they need to not get critically ill or not have to come back to the PICU or sometimes not come back to the PICU frequently.

The more we think about that, the more we advocate for social determinants of health screening and for different programs to try to help folks with needs so that they can have the best health outcomes possible. At our center, I have a great hospitalist that I’m working with, Leila DeWitt, who’s doing a lot of food insecurity screening, and she’s managed to convince our administrators to have a free meal trade program. It’s just really cool to see different individuals interested in advancing this line of research and of need to help these families.

Dr. Madden: It is amazing what people are doing. I’m in New Jersey and I had learned about, more in southern Jersey, the Camden area, there’ve actually been some projects that are looking at the neighborhoods and social determinants of health and taking it from more the wellness and preventative strategy. They’ve embedded more clinics and such to look at the population there and try and focus on, as I said, the preventative medicine. If they’re discharged, they’re having increased visits, they’re having increased access, whether or not they have the insurance piece.

So we work in the ICU level, which is a very different focus, and somehow to bring that grassroots in the neighborhood to impact what we’re doing in the ICU, I think we all love what we get to do and take care of critically ill children and see them overcome what brought them into the ICU. And we’ve seen that progress, as you talked about, as we’ve changed some of it, as we’ve done some more preventative strategies, whether it’s the respiratory and some of the antivirals we have but, you know, seatbelts and helmets and all of those pieces as well. But I think we’ll always have a population that will come and require the PICU.

So there’s a lot more to investigate and try and continue to improve upon. I love the fact that you and all your collaborators have really started to focus on this and trying to delve into maybe where we can make an impact. I love the fact that I’ve had the chance to chat with you and talk about this, and hopefully we can talk about it some more. But before we conclude, are there any final things you want to say?

Dr. McCrory: Well, I just want to thank again my collaborators at all these sites. Like I said, it was kind of a grassroots effort. People had to work with their local VPS folks, get the data and then work with folks to get more data out of the EMR and map it to the census tract and COI and so forth. I just really appreciate all of their hard work. Thanks for having me, Maureen. Thanks to SCCM.

Dr. Madden: It’s been a pleasure. This concludes another episode of the Society of Critical Care Medicine Podcast. If you’re listening on your favorite podcast app and you liked what you heard, consider rating and leaving a review. For the Society of Critical Care Medicine Podcast, I’m Maureen Madden.

Announcer: Maureen A. Madden, DNP, RN, CPNC, AC, CCRN, FCCM, is a professor of pediatrics at Rutgers Robert Wood Johnson Medical School and a pediatric critical care nurse practitioner in the pediatric intensive care unit at Bristol Myers Squibb Children’s Hospital in New Brunswick, New Jersey.

Join or renew your membership with SCCM, the only multiprofessional society dedicated exclusively to the advancement of critical care. Contact a customer service representative at +1 847 827-6888 or visit sccm.org/membership for more information.

The SCCM Podcast is the copyrighted material of the Society of Critical Care Medicine and all rights are reserved. Find more episodes at sccm.org/podcast. This podcast is for educational purposes only. The material presented is intended to represent an approach, view, statement or opinion of the presenter that may be helpful to others. The views and opinions expressed herein are those of the presenters and do not necessarily reflect the opinions or views of SCCM. SCCM does not recommend or endorse any specific test, physician, product, procedure, opinion or other information that may be mentioned.

Some episodes of the SCCM Podcast include a transcript of the episode’s audio. Although the transcription is largely accurate, in some cases it is incomplete or inaccurate due to inaudible passages or transcription errors and should not be treated as an authoritative record.