Thank you. Thank you, everyone, for coming. My name is Deirdre Picetti. I'm a Critical Care Fellow at Boston Children's Hospital. And thanks for coming to my presentation titled, The Epidemiology of Invasive and Non-Invasive Mechanical Ventilation Use for Children at the End of Life. I have nothing to disclose. Start with a brief overview of my research. So we know from prior studies that most children die in the hospital and they undergo intensive interventions and technology use near the end of life, including mechanical ventilation. Now, non-invasive mechanical ventilation is a less intensive potential alternative to invasive mechanical ventilation. But its impact on patients near the end of life is not well studied. Recent research has shown that use of non-invasive ventilation is increasing for adults at the end of life without a decrease in use of invasive mechanical ventilation. The prevalence of invasive and non-invasive ventilation for children at the end of life hasn't been investigated in a current national study. So my primary aim of this research was to determine the prevalence of invasive and non-invasive mechanical ventilation use for children at the end of life and to assess for any changes happening over time. My secondary aim was to identify characteristics that were associated with mode of ventilation and to assess for inter-hospital variability in its use. So this was a multicenter retrospective cohort study done using the PHIS database. For those who don't know, the PHIS database includes clinical and billing data from 48 children's hospitals across the US. I included children between the ages of 0 and 21 years who died anywhere in a hospital between 2010 and 2019. So data was abstracted from the terminal hospitalization only, which is that last hospitalization in which the patient died. And mode of ventilation was identified using a combination of ICD-9 and ICD-10 procedure and diagnosis codes, as well as CTC codes, which indicate that there's been daily billing for ventilatory support services. So I'll give a brief overview of the key results from this study. So initial inclusion criteria resulted in a population of over 48,000 patients from the 48 hospitals. My main exclusion criteria included hospitals that didn't report data for the full study period and individual patients that didn't have billing data available for them, after which I was left with a population of 41,091 patients from 37 hospitals. This study cohort had a median age of 0.4 years with an interquartile range of 0 to 6.7 years, 45% were female, 42% were white, 93% of patients had health insurance, and 95% had one or more complex chronic conditions. I found that about half of patients had a hospital length of stay less than or equal to 7 days, whereas 78% had a length of stay less than or equal to 30 days. Now my primary aim again was to determine the prevalence of ventilatory support for children who died in the hospital. So in order to address this question, we determined the percentage of children who were exposed to each mode of ventilation at some point during that terminal hospitalization. And we found that 69% of patients received IMV only during that hospitalization, 19% were exposed to both IMV and NIV, and only 3% received non-invasive only. Now this meant that in all, 91% of patients who died in the hospital received some form of ventilatory support during that hospitalization. When you combine categories, we can see that 88% of patients received any IMV, and 22% of patients received any NIV during that hospitalization. So next we evaluated for change over the 10-year study period in exposure to each mode of ventilation. And we found that the percentage of patients who were exposed to any IMV did not change over the study period, remained stable around 88.5%. Whereas the percentage of patients who were exposed to any NIV during that hospitalization increased by 7%, from 19% to 26%. And looking more closely at that increase in NIV, we find that the percentage of patients who were exposed to NIV only, only increased by 1%. Whereas the percentage of patients who were exposed to both IMV and NIV increased by 6% over that time period, which accounts for the majority of the overall increase in exposure to non-invasive ventilation. So while we've already reviewed the fact that the overall prevalence of exposure to any invasive ventilation didn't change over the study period, because there is that significant increase in the number of patients that are exposed to both, we see that there's a reciprocal decrease in the percentage of patients that are exposed to only IMV during their hospitalization. So next, we assess for the interhospital variability in use of NIV for children who died in the hospital. So in this figure, you'll find the percentage of patients from each hospital that were exposed to any NIV, shown for both 2010 versus 2019 in light and dark blue, respectively. And you can see that there's wide variability in non-invasive use between hospitals, with a range of 0% to 59% of patients per hospital per year billed for non-invasive at some point during their terminal hospitalization. We also find that some hospitals have a disproportionate increase in non-invasive use from 2010 to 2019. Now note that here I've divided each hospital into quartiles, with Quartile 1 representing the lowest use of NIV, and Quartile 4 representing the highest use of NIV. So these same NIV use quartiles are used in this analysis, which aim to assess if the interhospital variability seen in use of NIV correlates with variability in use of IMV. And here, the hospital NIV use quartiles are featured on the y-axis. And the x-axis is the adjusted odds that a patient from a hospital in that quartile is exposed to ventilatory support. And looking first to the adjusted odds of any NIV in dark blue, we see that patients in hospitals in higher NIV use quartiles were significantly more likely to be exposed to any NIV. In fact, patients in Quartile 4 were 6.2 times as likely to receive non-invasive during their hospitalization compared to those in Quartile 1. Turning our attention now to the odds of IMV, we find that that variability that we see in use of NIV does not correlate with variability in use of IMV. Next we assess for characteristics that were associated with mode of ventilation. So everything here compares all modes of ventilation are compared to no ventilation in a multivariable logistic regression. So characteristics that were associated with a higher likelihood of IMV only included younger age and health insurance. Characteristics that were associated with a higher likelihood of both IMV and NIV again included younger age and health insurance with the addition of elective admissions compared to urgent or emergent. We found that Hispanic ethnicity was associated with a lower likelihood of IMV only and Asian ethnicity was associated with a lower likelihood of receiving both IMV and NIV, both compared to non-Hispanic white ethnicity. Degree of chronic medical complexity was also associated with ventilatory support. And here you can see the median and the interquartile range for the number of complex chronic conditions stratified by mode of ventilation. And we found that the number of complex chronic conditions was highest for patients that received non-invasive ventilation only or both. Characteristics that were not associated with mode of ventilatory support included sex, overall child opportunity score, predicted median household income, and census region. So in conclusion, we found that for children who died in the hospital in the past decade, the vast majority of children received some type of ventilatory support. And use of non-invasive ventilation has increased without a reciprocal decrease in use of invasive mechanical ventilation. And the majority of this increase in use of non-invasive ventilation is being experienced as an increase in the number of patients that are exposed to both invasive and non-invasive ventilation during their hospitalization. And this suggests that there could be a potential additive effect of this technology. So further study is needed to determine, you know, in which situations there is benefit to the addition of non-invasive ventilation for children nearing the end of life. We also found that there's significant interhospital variability in the use of non-invasive ventilation. And that this variability does not correlate with variability in use of invasive mechanical ventilation. We also found that children who are younger age or have health insurance are more likely to receive invasive mechanical ventilation. Now differences in age could be due to more time with chronic illness for older patients, more time for advanced care planning, and maybe time for incorporation of preferences for older patients. The differences seen between patients who have health insurance and self-pay definitely needs further study, especially since other markers of socioeconomic status weren't found to be significant predictors of ventilatory support. We also found that children with increased chronic medical complexity are more likely to receive non-invasive ventilation. And it's not clear to what extent that reflects a pre-existing use of non-invasive ventilation versus new initiation of ventilatory support for these children with increased chronic medical complexity. We also found that additional differences do exist based on race and ethnicity. And that deserves further study as well. So the next step in this PHIS Research Study is to evaluate the association between mode of mechanical ventilation and hospital resource utilization for children at the end of life. And if you're interested in that topic, you can please come see my follow-up presentation tomorrow in the Research Snapshot Theater. So this study was limited by its retrospective design and also the limitations that are inherent with the clinical and billing data that are available on PHIS. Further study is needed to understand the goals and the decision-making process behind choices related to ventilation and also the impact of choice of mode of ventilation on outcomes such as patient-important outcomes and family satisfaction and bereavement. Finally, just want to thank Dr. Burns, who's my faculty mentor, and also Stephen Staffa, who is our statistician, as well as the PHIS analyst team and fellowship program leadership for their guidance. Thank you.

prevalence

invasive ventilation

non-invasive ventilation

children

end of life