Acute Care Events Research Articles

The impact of recreational cannabis legalization on cannabis-related acute care events among adults with schizophrenia.

Patients with schizophrenia have a higher risk of cannabis use disorder and may be uniquely affected by the legalization of recreational cannabis. This study examined whether cannabis legalization led to changes in acute care utilization among patients with schizophrenia. Using linked health administrative data, we included adult patients with schizophrenia in Ontario from October 2015 to May 2021 (n = 121,061). We examined the differences in cannabis, psychosis, and mental health-related emergency department (ED) visits over three periods: pre-legalization, legalization of flowers and herbs (phase 1), and legalization of edibles, extracts, and topicals (phase 2) using interrupted time-series methods. Our study found that phase 1 was associated with decreases in cannabis-related, mental health-related, and cannabis + psychosis-related ED visits among the patients with schizophrenia. Notably, an immediate 25.8% (95% CI 13.8-37.6%) decrease in cannabis-related ED visits was observed in men, and an immediate 18.5% decrease in mental health-related ED visits (95% CI 6.0-31.2%) in women. These decreases were also shown in the comparative ITS models, demonstrating that the changes observed were distinct from trends in the general population. However, phase 2 was not associated with any significant changes. Despite higher baseline rates of acute care utilization among patients with schizophrenia, cannabis legalization was associated with significant reductions, particularly during phase 1. Our findings suggest that regulatory measures accompanying legalization could enhance the quality and safety of cannabis products, potentially leading to fewer adverse health outcomes in vulnerable patient populations. Further research is needed to optimize healthcare responses for this vulnerable population.

The association of oncology telehealth visits with acute care events and follow-up appointments.

384 Background: Use of telehealth visits in cancer care expanded with the pandemic, yet few studies have assessed its impact on subsequent care outcomes. We evaluated differences in acute care events (ACEs) and follow-up visits among patients with telehealth compared to in-person oncology visits. Methods: We identified a cohort of adults with an upcoming oncology visit related to their cancer treatment at two academic medical centers (in central NC and southeast MI) and their community affiliates. Black adults and those with a scheduled telehealth visit were oversampled. Pre-visit surveys administered by telephone collected information on sociodemographic characteristics, prior telehealth use, health status/symptoms and travel distance. We joined survey and electronic health record (EHR) data and used geocoding to assign census-tract-level community characteristics to participants. ACE and follow-up visit use were ascertained via the EHR. Unadjusted and adjusted prevalence differences in ACEs within 30 and 180 days as well as follow-up visits within 7 and 30 days were assessed by visit type. Results: 502 patients were study eligible, 90 telehealth and 412 in-person visits. In unadjusted and adjusted models, the proportion of patients incurring an ACE within 7 and 30 days was not significantly different (6 vs. 3% and 17 vs. 10%, telehealth vs. in-person, respectively). Nor did the proportion of patients having a follow-up visit within 7 or 30 days differ significantly (26 vs. 20% and 65 vs. 54%, telehealth vs. in-person, respectively). Some patient characteristics were similar regardless of visit format. For example, most patients were older (63 mean age SD 13), not currently employed (66%), and reported getting by/living comfortably with current income (88%), having some college education (80%), no transportation troubles (85%), and activated patient portal accounts (89%). Symptom presentation per the PRO-CTCAE was similar between groups as was broadband connectivity. Telehealth patients were significantly less likely to be Black (14 vs. 44%), female (48 vs. 72%), have breast cancer (22 vs. 48%), received chemotherapy (72 vs. 83%), or to live within 60 minutes of care (49 vs. 68%), and more likely to have advanced disease (24 vs. 14%), report LGBTQ sexual orientation (10 vs. 4%), and be married (70 vs. 57%). Conclusions: Similar ACE and follow-up visit use among those with telehealth and in-person visits supports the idea that telehealth could be a viable substitute for some in-person care within oncology. Differences between those using telehealth and in-person visits highlight equity concerns and missed opportunities to enhance patient convenience and other aspects of care access. To better understand telehealth’s impact, not only on care use, but other indicators of quality (including patient outcomes), matching via propensity score analyses are needed and underway.

Counting the days: Exploring healthcare contact days during episodes of cancer care.

233 Background: Cancer diagnosis and treatment brings about not only emotional upheavals but also logistical hurdles of navigating a fragmented healthcare system. Little is known about the frequency of healthcare contact for cancer patients and its implications. Healthcare contact days, defined as the number of days a patient spends in direct contact with healthcare professionals outside of home, measures the patient’s interaction with the healthcare system. Methods: Using historical claims data (Part A institutional and Part B ambulatory claims) from the Enhancing Oncology Model (EOM) for 12 practices in The US Oncology Network covering the period between January 2019 to June 2022, we derived healthcare contact days per episode (6-month care periods initiating with the administration of chemotherapy), counting each unique date once. When patients encountered multiple care settings on a single date, the following hierarchy was used: inpatient (IP) > emergency room (ER) > skilled nursing facility (SNF)> hospice > ambulatory care. We further counted contact days for different services conducted during the ambulatory visits. Results: For the 75,246 episodes, an average of 31.9 days per episode (1 in every 6 days) were spent in healthcare contact. Cancer patients’ healthcare contact was predominantly within ambulatory care, and visits with oncology providers exceeded that with all other specialties. ER and IP visits, generally a representation of acute care events, accounted for <10% of the contact days. Conclusions: During cancer treatment, patients spend approximately one-sixth of their time in healthcare contact. The optimal frequency of contact, pattern of primary care support, care of comorbidities, and the interrelationship between acute care and clinician visits are apt for further exploration. Healthcare contact days can reflect different care delivery approaches between providers and may be a metric to assess both healthcare efficiency and accessibility. Understanding caregiver burden and social disparities associated with high contact days offers an opportunity to improve care coordination and streamline healthcare encounters while balancing patients’ care needs. CMMI Disclaimer: The statements contained in this document are solely those of the authors and do not necessarily reflect the views or policies of CMS. The authors assume responsibility for the accuracy and completeness of the information contained in this document. Healthcare contact days. Mean Percentiles 5 th -95 th Per Episode Healthcare Contact Days 31.9 10-65 Ambulatory Visits* 8-55 Chemotherapy 6.6 0-18 Imaging 2.7 0-7 Procedure 1.8 0-7 Radiation 2.4 0-23 Testing 11.8 3-25 Clinician Oncology 6 1-14 Other Specialties 3.4 0-10 Primary Care 2.8 0-8 Facility Visits ER 0.4 0-2 Hospice (Facility) 0.1 0-0 IP 2.7 0-16 SNF 0.8 0-0 *Ambulatory visits days are not mutually exclusive.

Mental health care use and quality among Medicaid adults with serious mental illness receiving care at Federally Qualified Health Centers vs. other settings

BackgroundFederally Qualified Health Centers (FQHCs) are a critical source of care for medically underserved populations and often serve as medical homes for individuals with serious mental illness (SMI). Many FQHCs provide mental health services and could facilitate access to mental health treatment within and outside of FQHCs. This study compared mental health care utilization and acute care events for adult Medicaid enrollees with SMI who receive care at Federally Qualified Health Centers (FQHCs) vs. other settings.MethodsThis study used the 2015–2016 Massachusetts All-Payer Claims Database to examine outpatient mental health care and acute care events for 32,330 Medicaid adults, ages 18–64 and with major depressive, bipolar, or schizophrenia spectrum disorders (SSD), who resided in FQHC service areas and received care from FQHCs vs. other settings in 2015. Multivariable linear regressions assessed associations between receiving care at FQHCs and outpatient mental health visits, psychotropic medication fills, and acute care events in 2016.ResultsThere were 8,887 (27.5%) adults in the study population (N = 32,330) who had at least one FQHC visit in 2015. Those who received care at FQHCs were more likely to have outpatient mental health visits (73.3% vs. 71.2%) and psychotropic medication fills (73.2% vs. 69.0%, both p < .05), including antidepressants among those with depressive disorders and antipsychotics among those with SSD. They were more likely to have ED visits (74.0% vs. 68.7%), but less likely to be hospitalized (27.8% vs. 31.9%, both p < .05). However, there was no significant difference in the likelihood of having an acute psychiatric hospitalization (9.5% vs. 9.8%, p = .35).ConclusionsAmong Medicaid enrollees with SMIs who had access to care at FQHCs, those receiving care at FQHCs were more likely to have outpatient mental health visits and psychotropic medication fills, with lower rates of hospitalization, suggesting potentially improved quality of outpatient care. Higher ED visit rates among those receiving care at FQHCs warrant additional investigation.

Independent validation of 30-day emergency department visit after initial chemotherapy A-PROACCT model.

e23243 Background: This study assessed model performance and, more importantly, the clinical usefulness of the PROACCT model created by Grant et al. (JAMA, 2019), refined as A-PROACCT by Stein et al. (JOP, 2023) with data from Orlando Health. The predictive performance of models anticipating acute care events within thirty days (ACE30) of initial chemotherapy administration (ICA) were evaluated. Methods: Stein et al. models classified high-risk patients as those whose model output risk score was at or above the 90th percentile. They studied 5153 ICAs between July and November of 2021, of which 559 (10.8%) had an ACE30. Their first model used drug category, age group, ED visits and hospitalizations in the prior year, and cancer type, and had sensitivity (Sen), specificity, (Spec), C-Statistic (C-Stat), and positive predictive value (PPV) of 0.28, 0.93, 0.79 and 0.18 respectively. Their second model added insurance category, number of anti-cancer agents, race, and BMI category and reported Sen, Spec, C-Stat, and PPV of 0.30, 0.94, 0.80 and 0.23 respectively. Of the high-risk patients, 33.6% and 34.1% had an ACE30 respectively. Of the low-risk patients, 8.3% and 8.1% had an ACE30 respectively. Our data were propagated through the two models. Results: Between February of 2012 and April 2021, Orlando Health had 15,483 distinct patient visits with ICAs. Overall, 2046 (13.2%) had an ACE30. Model one had Sen, Spec, C-Stat, and PPV of 0.19, 0.92, 0.65 and 0.27 respectively. Model two had Sen, Spec, C-Stat, and PPV of 0.19, 0.91, 0.65 and 0.25 respectively. Both models had an accuracy of 0.82. This demonstrated our data fit the Stein model. Of the 1460 ICAs identified as high-risk by model one, 387 (26.5%) had an ACE30, and of the 14,023 ICAs identified as low-risk, 1659 (11.8%) had an ACE30. Of the 1549 ICAs identified as high-risk by model two, 393 (25.4%) had an ACE30, and of the 13,934 ICAs identified as low-risk, 1653 (11.9%) had an ACE30. The difference of ACE30 rates between high risk and low risk groups were statistically significant (p &lt; 0.0001) in both models. This demonstrated both models were clinically useful in identifying chemotherapy patients at high risk of acute care events. Conclusions: These models allowed for clinically relevant stratification of patients into meaningful high-risk and low-risk groups (26.5% vs 11.8% with our data) for risk of acute care events within 30 days of initial chemotherapy administration. Nevertheless, only 26.5% of these high-risk patients actually had an acute care event within thirty days of initial chemotherapy administration. Further research is needed to improve the precision of identifying high-risk patients, by potentially adding more clinical variables, using alternative sampling methods, and employing nonlinear and more complex machine learning methodologies.

Refining the 30-day emergency department visit after initial chemotherapy administration A-PROACCT model.

e23253 Background: This study attempted to address limitations in the A-PROACCT model developed by Stein et al. (JOP, 2023) to predict the rate of acute care events within 30 days (ACE30) of initial chemotherapy administration (ICA). The four most important predictors were drug category (immunotherapy), cancer type (gastrointestinal), age group ( &lt; 40), and BMI category (underweight). Only linear relationships were considered. For patients for whom an ACE30 was predicted by their first and second models, only 0.18 and 0.23 actually had an ACE30, a low positive predictive value (PPV). We attempted to improve the PPV with advanced modeling techniques. Methods: Using Orlando Health data between February of 2012 and April 2021, random sampling was used to split the dataset into training (10,838) and validation (4645). For the training data, resampling was employed to obtain a balanced quota, equal observations followed by an ACE30 and not followed by an ACE30. We included only variables used in Stein (drug category, age group, ED visits and hospitalizations in the prior year, cancer type, insurance category, number of anti-cancer agents, race, and BMI category). Logistic regression with L1-penalty, XGBoost (a non-linear nonparametric tree-based algorithm to account for nonlinear relationships), and artificial neural networks were used to develop and evaluate predictive models. Different sampling methods (bootstrap and SMOTE) as well as cuff-off thresholds for high-risk groups were tested. Results: Based on evaluation with validation data set, the best performing approach was an XGBoost model with SMOTE resampling in the training data. The four most important predictors were ED visits in the prior year, payor category (self-insured), cancer type (genitourinary) and hospitalization in past year. This model reported a PPV of 0.27. Many other combinations of methodologies described above were performed, and the PPV varied between 0.19 to 0.27. Of the 282 ICAs identified as high-risk by the best model, 76 (27.0%) had an ACE30. Of the 4346 ICAs identified as low-risk, 533 (12.3%) had an ACE30. The differences between high risk and low risk ACE30 rates were statistically significant (p &lt; 0.0001). In comparison, using Stein et al. models with our data reported a PPV of 0.27 and 0.25 which are basically at the same level of our best trained model. Conclusions: Using the variables included in A-PROACCT and a variety of machine learning models, advanced sampling methodologies and threshold cut-off limits, the best results were similar to those obtained using basic logistic regression. This suggests improvement in predicting acute care events following chemotherapy administration need to incorporate more extensive multimodal data, such as vital signs, performance status, patient reported outcomes, socioeconomic factors, laboratory and radiology results, remote patient monitoring and wearable device data.

Allied health assistant management of people with hip fracture is feasible and may improve patient adherence to hip fracture mobilisation guidelines: a feasibility randomised controlled trial

ObjectivesDetermine the feasibility of allied health assistant (AHA) management of people with hip fracture an acute hospital. DesignAssessor-blind, parallel, feasibility randomised controlled trial with qualitative component. SettingAcute orthopaedic ward. ParticipantsPeople with surgically-managed hip fracture, who walked independently pre-fracture and had no cognitive impairment. InterventionsRehabilitation from an AHA, under the supervision of a physiotherapist, compared with rehabilitation from a physiotherapist. Main outcome measuresFeasibility was evaluated according to focus areas of demand, acceptability, practicality and implementation. Secondary outcomes included estimates of effect of adherence to hip fracture mobilisation guidelines, discharge destination, 30-day readmission, functional activity, and length of stay. ResultsFifty people were allocated to receive rehabilitation from an AHA (n = 25) or physiotherapist (n = 25). AHA rehabilitation had high demand with 60% of eligible participants recruited. Satisfaction with AHA rehabilitation was comparable with physiotherapy rehabilitation (acceptability). The AHA group received an average of 11 min (95% CI 4 to 19) more therapy per day than the physiotherapy group (implementation). The AHA group may have had lower cost of acute care (MD -$3 808 95% CI −7 651 to 35) and adverse events were comparable between groups (practicality). The AHA group may have been 22% (HR 1.22, 95% CI 0.92 to 1.61) more likely to walk on any day and may have had a shorter length of stay (MD −0.8 days, 95% CI −2.3 to 0.7). ConclusionsAHA management of patients with hip fracture was feasible and may improve adherence to mobilisation guidelines and reduce cost of care and length of stay. Clinical Trial Registration NumberACTRN12620000877987. Contribution of the paper•It is feasible for allied health assistants, under the supervision of a physiotherapist, to manage the acute care of people with hip fracture who walked independently pre-fracture and had no cognitive impairment.•Delegating acute management of people with hip fracture to allied health assistants may improve adherence to hip fracture mobilisation guidelines and reduce length of stay and cost of acute care, however, further trials are needed to confirm these effects.

Prescription and Dispensation of QT-Prolonging Medications in Individuals Receiving Hemodialysis

Individuals with dialysis-dependent kidney failure have numerous risk factors for medication-related adverse events, including receipt of care by multiple clinicians and initiation of some QT-prolonging medications with known risk of torsades de pointes (TdP), which is associated with higher risk of sudden cardiac death. Little is known about the prescription and dispensation patterns of QT-prolonging medications among people receiving dialysis, hindering efforts to reduce drug-related harm from these and other medications in this high-risk population. To examine prescription and dispensation patterns of QT-prolonging medications with known TdP risk and selected interacting medications prescribed to individuals receiving hemodialysis. This cross-sectional study included patients 60 years or older who were enrolled in Medicare Parts A, B, and D receiving in-center hemodialysis from January 1 to December 31, 2019. Analyses were conducted from October 20, 2022, to June 16, 2023. New-user prescriptions for the 7 most frequently filled QT-prolonging medications characterized by the timing of the new prescription relative to acute care encounters, the type of prescribing clinician and pharmacy that dispensed the medication, and concomitant use of selected medications known to interact with the 7 most frequently filled QT-prolonging medications with known TdP risk. The main outcomes were the frequencies of the most commonly filled and new-use episodes of QT-prolonging medications; the timing of medication fills relative to acute care events; prescribers and dispensing pharmacy characteristics for new use of medications; and the frequency and types of new-use episodes with concurrent use of potentially interacting medications. Of 20 761 individuals receiving hemodialysis in 2019 (mean [SD] age, 74 [7] years; 51.1% male), 10 992 (52.9%) filled a study drug prescription. Approximately 80% (from 78.6% for odansetron to 93.9% for escitalopram) of study drug new-use prescriptions occurred outside of an acute care event. Between 36.8% and 61.0% of individual prescriptions originated from general medicine clinicians. Between 16.4% and 26.2% of these prescriptions occurred with the use of another QT-prolonging medication. Most potentially interacting drugs were prescribed by different clinicians (46.3%-65.5%). In this cross-sectional study, QT-prolonging medications for individuals with dialysis-dependent kidney failure were commonly prescribed by nonnephrology clinicians and from nonacute settings. Prescriptions for potentially interacting medications often originated from different prescribers. Strategies aimed at minimizing high-risk medication-prescribing practices in the population undergoing dialysis are needed.

Abstract 6377: UNC AYA Cancer Program: Reach and influence on cancer care delivery

Abstract Category: Cancer Disparities and Inequities Research Purpose of the study Adolescent and young adults (AYAs, aged 13-39) with cancer face barriers to care and stagnant outcomes. National groups have called for the development of AYA-specific programs, yet how these programs influence outcomes remains largely unexplored. With a dedicated AYA Cancer Program at the University of North Carolina (UNC) since 2015, we conducted a retrospective study to describe the AYA population and compare outcomes between those receiving program support and those who did not. Methods Linking electronic health records and NC Cancer Registry data, we defined a cohort of AYA patients who received cancer treatment at UNC from 2014-2022. We compared sociodemographic, clinical, utilization, and supportive care measures for those who received AYA Program contact and those who did not. Chi-square and two-group t-tests were used for comparison. Results Of 4044 AYAs receiving care at UNC, 670 engaged with our AYA Program (Table 1). Program-contacted patients were younger at diagnosis, more likely to be Black, have metastatic disease or hematologic malignancies, and more likely to receive systemic therapy. Program patients had increased clinical trial enrollment, more palliative care and fertility counseling. Conclusions Our findings suggest AYAs contacted by our program represent a distinct population with higher clinical and social needs. Despite higher rates of acute care events and no-shows, they were more likely to receive guideline-recommended care - clinical trial enrollment, fertility counseling, and palliative care. Future analyses will compare outcomes using adjusted modeling, and track trends over time. Results will inform next steps for the UNC AYA Cancer Program and other centers aiming to improve care delivery for AYAs. Table 1. Notable results for AYAs with and without program contact AYA Program Contact (n=670) No AYA Program Contact (n=3374) P value Age at diagnosis Mean years 27 31 &amp;lt;0.01 Sex Female 55% 61% 0.02 Male 45% 39% Race White 56% 61% &amp;lt;0.01 Black 26% 20% Other 18% 16% Ethnicity Hispanic 13% 12% 0.29 Non-Hispanic 86% 84% Marital Status Married/Partnered 34% 42% &amp;lt;0.01 Not Married 65% 51% Cancer Type Solid 54% 87% &amp;lt;0.01 Hematologic 46% 13% Stage Metastatic 52% 18% &amp;lt;0.01 Treatment Chemotherapy 85% 37% &amp;lt;0.01 Radiotherapy 6% 12% Immunotherapy 18% 9% Clinical Trial Enrollment 45% 26% &amp;lt;0.01 No-Shows Mean 7 4 &amp;lt;0.01 Supportive Care Palliative care visits 19% 3% &amp;lt;0.01 Advance Care Planning 14% 5% Fertility visit 18% 2% Hospice enrollment 3% 0.7% Complications ED visit 68% 48% &amp;lt;0.01 Unplanned hospitalization 63% 29% Citation Format: Jacob N. Stein, Dawn M. Ritzwoller, Alexis C. Wardell, Emily R. Haines, Bridgette Thom, Lauren Lux, Andrew B. Smitherman. UNC AYA Cancer Program: Reach and influence on cancer care delivery [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 6377.

Medicaid Insurance Predicts Increased Postoperative Care Encounters Among Patients on Long-Term Opioid Therapy.

This study examined the association between insurance type and postoperative unplanned care encounters among patients on long-term opioid therapy prior to surgery. Preoperative long-term opioid therapy is associated with unique risks and poorer outcomes following surgery. To date, the extent to which insurance coverage influences postoperative outcomes in this population remains unclear. Among individuals receiving a supply of greater than 120 total days or at least 10 opioid prescriptions in the year prior to surgery, we examined patients with Medicaid or private insurance who underwent abdominopelvic surgery from 2017 to 2021 across 70 hospitals in the state of Michigan. The primary outcome was unplanned care encounters, defined as an emergency department visit or unplanned readmission within 30 days of discharge from surgery. Multivariable logistic regression was used to assess the likelihood of acute care events with insurance type as the primary covariate of interest. Among 1212 patients on long-term opioid therapy prior to surgery, 45.6% (n = 553) had Medicaid insurance. Overall, one in eight (n=151) patients met criteria for a postoperative unplanned care encounter within 30 days. The probability of an unplanned encounter was 4.5 percentage points higher among patients with Medicaid insurance compared to private insurance (95% CI: 0.5%, 8.4%). Among patients on preoperative long-term opioid therapy, unplanned care encounters were higher among patients with Medicaid when compared to private insurance. While this is likely multifactorial, differences by insurance status may point to disparities in underlying social determinants of health and suggest the need for postoperative care pathways that address these gaps.

Establishing purposeful oncology care disparities reduction strategies.

166 Background: Quality thought leaders have noted the oneness of care quality and equity. Yet, quality improvement (QI) often ignores equity. We implemented a holistic approach to quality specifically designed to facilitate equity. We report the components of the approach and how it is being used to address disparities in unplanned acute care events (ACEs) among chemotherapy patients. Methods: We used the Institute for Health Improvement Whole System Quality framework to integrate care equity in an established quality program. We started by stratifying oncology dashboards by race, ethnicity, and language (REAL) as well as other factors (monitoring). We developed and implemented an Umbrella Action Plan Policy that specified an accountability structure to identify and address disparities (planning) and via goal setting (planning), prioritized, developed, and executed action plans to address disparities once identified (improvement). Results: In February 2023 a system-level senior leadership committee charged with priority setting cancer care quality initiatives approved the Umbrella Action Policy. This led to REAL-stratified, quarterly performance reports being shared with senior leadership who voted for this year’s annual goals to include identification and use of rapid cycle testing to eliminate ACE disparities. The organization-level goal engendered central QI coaching resources, established local entity QI teams within each System-affiliated entity with dedicated oncology program (N=6), and formed a learning collaborative across entities. One of each local QI team chose to focus on reducing race-related, language-related, and payor-related (Medicaid) disparities. The other organizations chose otherwise (e.g., age or gender). Teams are using two approaches to identify ACE root causes among target populations: (1) a structured chart abstraction form to search for, among other things, communication breakdowns, gaps in symptom management or other supportive care, and access barriers; and (2) a semi-structured interview guide to engage patients and caregivers for input and suggestions. In September, teams will work collaboratively to identify and execute small tests of change to alleviate identified drivers contributing to ACE disparities. Data from chart abstractions and patient engagement interviews will be shared along with results from tests of change tried to date. Conclusions: Facilitating the identification of disparities via REAL stratified dashboards, leadership teams setting expectations regarding gap closure, and allocation of resources to support local QI teams has focused the quality planning, monitoring, and, to some extent, improvement activities on care equity among marginalized populations. Prescriptively focusing local organization efforts may be needed for appropriate selection of target populations when wanting to infuse REAL equity in QI programming.

Prospective validation of machine learning-based approaches to predict potentially preventable emergency visits and hospitalizations.

404 Background: Patients undergoing cancer treatment are at risk for unplanned acute care. Early identification of at-risk patients could enable preventative interventions, reducing costs and treatment delays. To address this, the Centers for Medicare &amp; Medicaid Services developed the Chemotherapy Measure (OP-35) to monitor potentially preventable acute care utilization during outpatient treatment. We previously developed machine-learning (ML) models using three approaches: least absolute shrinkage selection operator (LASSO), random forest (RF), and gradient boosted trees (GBT). The models predict risk of an OP-35 qualifying acute care event in the 30-days following a systemic therapy infusion and had good performance in the internal validation cohort, with GBT demonstrating the best predictive ability (receiver operating characteristic area under the curve (ROC-AUC) = 0.805). The aim of this study was to prospectively validate these models in patients undergoing systemic therapy at a single institution. Methods: All three models are being prospectively validated on systemic therapy infusions, including chemo, immuno, biologic, hormone, research and targeted therapies. Assuming a 2% event rate based on our prior data, with an alpha = 0.05 and 84% power, to detect an ROC-AUC of 0.75, validation will run for a total of 8000 infusions, from May 1 to August 21, 2023. We present early findings from May 1 to May 14, 2023 for this prospective validation. Model performance was assessed for calibration based on brier score and predictive ability based on ROC-AUC. Sensitivity and specificity were calculated for the model with the highest ROC-AUC based on a previously determined Youden’s J statistic. Results: This study included 1096 systemic therapy treatments across 957 patients. 21 (1.9%) infusions resulted in an OP-35 qualifying acute care event (3 emergency department visits and 18 hospitalizations). Most events were due to pain, anemia, sepsis, and/or dehydration. All models had good performance, with GBT demonstrating the greatest predictive ability (ROC-AUC = 0.78 [0.67 - 0.88], compared to LASSO (0.76 [0.65-0.85]) and RF (0.70 [0.56-0.81]). GBT also had good calibration (brier score = 0.017 [0.011 - 0.025]) followed closely by LASSO (0.018 [0.011-0.026]) and RF (0.019 [0.012-0.026]). Youden-based cut-off of 0.0249 corresponded to a validation sensitivity of 77.6% and specificity of 61.9%. Conclusions: Early prospective validation of ML models demonstrates accurate predictions of OP-35 qualifying acute care events on a per-infusion basis. The use of computational tools to identify patients at risk for unplanned acute care would enable preventative interventions, reducing costs and treatment disruptions. Prospective validation is ongoing and more comprehensive results will be presented at the meeting.

Machine Learning-Based Early Warning Systems for Acute Care Utilization During Systemic Therapy for Cancer.

Emergency department visits and hospitalizations frequently occur during systemic therapy for cancer. We developed and evaluated a longitudinal warning system for acute care use. Using a retrospective population-based cohort of patients who started intravenous systemic therapy for nonhematologic cancers between July 1, 2014, and June 30, 2020, we randomly separated patients into cohorts for model training, hyperparameter tuning and model selection, and system testing. Predictive features included static features, such as demographics, cancer type, and treatment regimens, and dynamic features, such as patient-reported symptoms and laboratory values. The longitudinal warning system predicted the probability of acute care utilization within 30 days after each treatment session. Machine learning systems were developed in the training and tuning cohorts and evaluated in the testing cohort. Sensitivity analyses considered feature importance, other acute care endpoints, and performance within subgroups. The cohort included 105,129 patients who received 1,216,385 treatment sessions. Acute care followed 182,444 (15.0%) treatments within 30 days. The ensemble model achieved an area under the receiver operating characteristic curve of 0.742 (95% CI, 0.739-0.745) and was well calibrated in the test cohort. Important predictive features included prior acute care use, treatment regimen, and laboratory tests. If the system was set to alarm approximately once every 15 treatments, 25.5% of acute care events would be preceded by an alarm, and 47.4% of patients would experience acute care after an alarm. The system underestimated risk for some treatment regimens and potentially underserved populations such as females and non-English speakers. Machine learning warning systems can detect patients at risk for acute care utilization, which can aid in preventive intervention and facilitate tailored treatment. Future research should address potential biases and prospectively evaluate impact after system deployment.

Developing machine learning algorithms incorporating patient reported outcome measures to predict acute care events among patients with cancer.

1516 Background: Clinical trials have shown that collecting patient-reported outcome measures (PROMs) can improve outcomes among oncology patients. However, there remains uncertainty about how to best collect and use PROMs in routine clinical care. Since 2016 we have collected PROMs for patients receiving cancer treatment within a large health system containing academic and community sites. We sought to examine whether combining PROMs with electronic health record (EHR) data improves the performance of machine learning algorithms predicting patients’ risk of 30-day acute care events. Methods: The sample included patients with solid tumors receiving curative- or palliative-intent systemic therapy between 9/1/20 and 7/31/22 who completed at least one PROMs survey. PROMs were collected via a 12-item questionnaire based on the Common Terminology Criteria for Adverse Events measurement system. Patients completed questionnaires via the EHR’s patient portal or on tablets in waiting rooms. EHR data included 176 variables describing demographics, laboratory values, vital signs, diagnoses, medications, and prior healthcare encounters. A binomial outcome was calculated for each patient encounter to indicate whether the patient had an acute care event (ED visit or hospitalization) in the subsequent 30 days. Data were randomly assigned to training (70%) and test (30%) sets and were used to train and test four models: EHR + PROMs, EHR alone, PROMs alone, and ‘PROMs plus,’ which included PROMs and minimal EHR data such as age, sex, and site of cancer treatment, which served as an accessible proxy for cancer type. For each model, we compared multiple architectures and selected the best-performing model for comparison. Results: Between September 2020 and June 2022, 4,193 patients completed 20,359 PROM surveys during outpatient encounters. There were 2,118 30-day acute care events within the cohort. The area under the receiver operating curves (AUCs) were 0.84 (95% CI 0.81 to 0.86) for the EHR + PROMs model, 0.82 (95% CI 0.80 to 0.84) for the EHR alone model, 0.67 (95% CI 0.65 to 0.69) for the PROMs alone model, and 0.79 (95% CI 0.77 to 0.82) for the PROMs plus model. In the EHR alone and EHR + PROMs models, the most important variables were last albumin, white blood cell count and hemoglobin. In the PROMs plus model, the most important variables were age, treatment site, patient-reported pain, and patient-reported fatigue. Conclusions: Incorporating PROMs into machine learning models to predict risk of acute care events can improve model performance. Notably, the model combining PROMs with a small number of easily accessible EHR variables performed nearly as well as models including all EHR data. These models demonstrate the predictive value of PROMs in oncology patients and they lay the foundation for future interventions aiming to reduce acute care events among high-risk patients.

Predicting Acute Care Events Among Patients Initiating Chemotherapy: A Practice-Based Validation and Adaptation of the PROACCT Model.

Acute care events (ACEs), comprising emergency department visits and hospitalizations, are a priority area for reduction in oncology. Prognostic models are a compelling strategy to identify high-risk patients and target preventive services, but have yet to be broadly implemented, partly because of challenges with electronic health record (EHR) integration. To facilitate EHR integration, we adapted and validated the previously published PRediction Of Acute Care use during Cancer Treatment (PROACCT) model to identify patients at highest risk for ACEs after systemic anticancer treatment. A retrospective cohort of adults with a cancer diagnosis starting systemic therapy at a single center between July and November 2021 was divided into development (70%) and validation (30%) sets. Clinical and demographic variables were extracted, limited to those in structured format in the EHR, including cancer diagnosis, age, drug category, and ACE in prior year. Three logistic regression models of increasing complexity were developed to predict risk of ACEs. Five thousand one hundred fifty-three patients were evaluated (3,603 development and 1,550 validation). Several factors were predictive of ACEs: age (in decades), receipt of cytotoxic chemotherapy or immunotherapy, thoracic, GI or hematologic malignancy, and ACE in the prior year. We defined high-risk as the top 10% of risk scores; this population had 33.6% ACE rate compared with 8.3% for the remaining 90% in the low-risk group. The simplest Adapted PROACCT model had a C-statistic of 0.79, sensitivity of 0.28, and specificity of 0.93. We present three models designed for EHR integration that effectively identify oncology patients at highest risk for ACE after initiation of systemic anticancer treatment. By limiting predictors to structured data fields and including all cancer types, these models offer broad applicability for cancer care organizations and may offer a safety net to identify and target resources to this high risk.

A Model to Improve Care and Enhance the Physician-Patient Relationship for Hospitalized Patients With Opioid Use Disorder.

Opioid-related morbidity and mortality continues to accelerate, with increasing acute care events for opioid-related causes. Most patients do not receive evidence-based treatment for opioid use disorder (OUD) during acute hospitalizations despite this being an invaluable moment to initiate substance use treatment. Inpatient addiction consult services can bridge this gap and improve patient engagement and outcomes, but varying models and approaches are needed to match institutional resources. To improve care for hospitalized patients with OUD, a work group was formed at the University of Chicago Medical Center in October 2019. An OUD consult service, run by generalists, was created as part of a series of process improvement interventions. Important collaborations with pharmacy, informatics, nursing, physicians, and community partners have occurred over the last 3 years. The OUD consult service completes 40-60 new inpatient consults monthly. Between August 2019 and February 2022, the service completed 867 consults from across the institution. Most consult patients were started on medications for opioid use disorder (MOUD), and many received MOUD and naloxone at discharge. Patients who were treated by our consult service experienced lower 30-day and 90-day readmission rates compared with patients who did not receive a consult. Length of stay for patients receiving a consult was not increased. Adaptable models of hospital-based addiction care are needed to improve care for hospitalized patients with OUD. Continued work to reach a higher percentage of hospitalized patients with OUD and to improve linkage to care with community collaborators are important steps to strengthen the care received by individuals with OUD in all clinical departments.

Is There Evidence of an Association Between Acute Health Care Utilization and Function in Older Adults Over Time? A Population-Based Cohort Study.

Acute health care use varies by age, with older adults the highest users of acute health care services. Using data from The Irish Longitudinal Study on Ageing (TILDA), the aim of this study was to investigate the association between acute health care utilization (emergency department [ED] visit with or without hospitalization) at baseline and subjective and objective measures of function measured at 4-year follow-up. This study represents a secondary analysis of a prospective cohort study, where data from Wave 1 (baseline) and Wave 3 of TILDA were analyzed in conjunction with a public and patient involvement group of older adults. Acute health care utilization was defined as an ED visit with or without hospitalization in the previous 12 months. Function was assessed objectively using the Timed Up and Go (TUG) test and a measure of grip strength, and subjectively using self-report limitations in activities of daily living (ADL) and instrumental ADL (IADL). A total of 1 516 participants met the study inclusion criteria. Mean age was 70.9 ± 4.6 years and 48% were male. At baseline, 1 280 participants reported no acute health care use. One hundred and eighteen indicated an ED visit but no hospitalization in the previous 12 months and 118 reported both an ED visit and hospitalization. Adjusting for all covariates, compared to those with no acute health care utilization, those with an ED visit with no hospital admission had poorer TUG performance at follow-up (β = 0.67, 95% confidence interval: 0.34, 1.29, p = .039). This paper supports previous research that acute health care events, specifically ED usage, are associated with reduced function for older adults as assessed by TUG at follow-up. No associations were observed for grip strength, ADL, or IADL. Further research is required in this area, exploring ED visits and the possible benefits of evaluating older adults at this stage.

Immunosuppression Exposure and Risk of Infection-Related Acute Care Events in Patients With Glomerular Disease: An Observational Cohort Study

Infections cause morbidity and mortality in patients with glomerular disease. The relative contributions from immunosuppression exposure and glomerular disease activity to infection risk are not well characterized. To address this unmet need, we characterized the relationship between time-varying combinations of immunosuppressant exposure and infection-related acute care events while controlling for disease activity, among individuals with glomerular disease. Prospective, multicenter, observational cohort study. Adults and children with biopsy-proven minimal change disease, focal segmental glomerulosclerosis, membranous nephropathy, or immunoglobulin A nephropathy/vasculitis were enrolled at 71 clinical sites in North America and Europe. A total of 2,388 Cure Glomerulonephropathy Network participants (36% aged<18 years) had at least 1 follow-up visit and were included in the analysis. Immunosuppression exposure modeled on a weekly basis. Infections leading to an emergency department visit or hospitalization. Marginal structural models were used to estimate the effect of time-varying immunosuppression exposure on hazard of first infection-related acute care event while accounting for baseline sociodemographic and clinical factors, and time-varying disease activity. A total of 2,388 participants were followed for a median of 3.2 years (interquartile range, 1.6-4.6), and 15% experienced at least 1 infection-related emergency department visit or hospitalization. Compared to no immunosuppression exposure, steroid exposure, steroid with any other immunosuppressant, and nonsteroid immunosuppressant exposure were associated with a 2.65-fold (95% CI, 1.83-3.86), 2.68-fold (95% CI, 1.95-3.68), and 1.7-fold (95% CI, 1.29-2.24) higher risk of first infection, respectively. Absence of medication dosing data, lack of a control group, and potential bias in ascertainment of outcome events secondary to the coronavirus 2 pandemic. Corticosteroids with or without concomitant additional immunosuppression significantly increased risk of infection leading to acute care utilization in adults and children with glomerular disease.

Comparing Self-Management Programs with and without Peer Support among Patients with Chronic Obstructive Pulmonary Disease: A Clinical Trial.

Rationale: Self-management support (SMS) is an essential component of care for patients who have chronic obstructive pulmonary disease (COPD), but there is little evidence on how to provide SMS most effectively to these patients. Peer support (i.e., support provided by a person with a similar medical condition) has been successfully used to promote self-management among patients with various chronic conditions, yet no randomized studies have focused on testing its effects for patients with COPD. Objectives: To assess whether adding peer support to healthcare professional (HCP) support to help patients with COPD self-management results in better health-related quality of life (HRQoL) and less acute care use. Methods: A two-arm randomized controlled trial was performed at one academic and one community hospital and their affiliate clinics. The study population included patients aged ⩾40 years who had been diagnosed with COPD by a physician and were currently receiving daily treatment for it. Two self-management support strategies were compared over 6 months. One strategy relied on the HCP for COPD self-management (HCP support); the other used a dual approach involving both HCPs and peer supporters (HCP Plus Peer). The primary outcome was change in HRQoL measured by the St. George's Respiratory Questionnaire at 6 months (range, 0-100, lower is better; four-point meaningful difference). Secondary outcomes included COPD-related and all-cause hospitalizations and emergency department visits. Analysis was conducted under intention to treat. Results: The number of enrolled participants was 292. Mean age was 67.7 (standard deviation, 9.4) years; 70.9% of participants were White, and 61.3% were female. St. George's Respiratory Questionnaire scores were not significantly different between the study arms at 6 months. HCP Plus Peer arm participants had fewer COPD-related acute care events at 3 months (incidence rate ratio, 0.68; 95% confidence interval [CI], 0.50-0.93) and 6 months (incidence rate ratio, 0.84; 95% CI, 0.71-0.99). Conclusions: Adding peer support to HCP support to help patients self-manage COPD did not further improve HRQoL in this study. However, it did result in fewer COPD-related acute care events during the 6-month intervention period. Clinical trial registered with www.clinicaltrials.gov (NCT02891200).

Identification and Validation of Infection-Related Acute Care Events in Patients with Glomerular Disease

Identification and Validation of Infection-Related Acute Care Events in Patients with Glomerular Disease