Melatonin prescribing characteristics and delirium outcomes: A health system database-based study.

Medicare is shifting from payment for postacute care services based on the volume provided to payment based on value as determined by patient characteristics and functional outcomes. Matching therapy time and length of stay (LOS) to patient needs will be critical to optimize functional outcomes and manage costs. To investigate the association among therapy time, LOS, and functional outcomes for patients following hip fracture surgery. This retrospective cohort study analyzed data on patients from 4 inpatient rehabilitation facilities and 7 skilled nursing facilities in the eastern and midwestern United States. Participants were patients aged 65 years or older who received inpatient rehabilitation services for hip fracture and had Medicare fee-for-service as their primary payer. Data were collected from 2005 to 2010. Analysis was conducted from November 2018 to June 2019. Therapy minutes per LOS day. Functional Independence Measure mobility and self-care measures at discharge. Patients were categorized into 9 recovery groups based on low, medium, or high therapy minutes per LOS day and low, medium, or high rate of functional gain per day. A total of 150 patients (101 [67.3%] female; 148 [98.6%] white; mean [SD] age, 82.0 [7.3] years) met inclusion criteria. Participants in all gain and therapy minutes per LOS day trajectories were similar in function at rehabilitation admission (mean [SD] mobility, 16.2 [3.2]; F8,141 = 1.26; P = .27) but differed significantly at discharge (mean [SD] mobility, 23.9 [5.2]; F8,141 = 14.34; P < .001). High-gain patients achieved mobility independence by discharge; low-gain patients needed assistance on nearly all mobility tasks. Medium-gain patients with a mean LOS of 27 days were independent in mobility at discharge; those with a mean LOS less than 21 days needed supervision with toilet transfers and were dependent with stairs. Length of stay and functional gain rate explained much of the variance in mobility and self-care scores at discharge. Although medium- and high-therapy minutes per LOS day groups were statistically significant in the regression model (β = 6.99; P = .001; and β = 11.46; P = .007, respectively), they explained only 1% of the variance in discharge outcome. Marginal means suggest that medium-gain patients with shorter LOS would have achieved mobility independence if LOS had been extended. In this study, rate of recovery and LOS in skilled nursing and inpatient rehabilitation facilities were associated with mobility and self-care outcomes at discharge following hip fracture surgery, particularly for medium-gain patients. Therapy time per day explained only 1% of the variance in discharge outcome. Discharging medium-gain patients before 21 days LOS may transfer burden of care to family and caregivers, home health, and outpatient services.

Background: Recently published data from the CDC has demonstrated that COVID-19 outcomes vary between racial and ethnic groups. Hospital length of stay (LOS) has been used as a surrogate marker for quality of care delivery. Longer duration of hospitalization can lead to increased morbidity, increased risk in hospital acquired infections, and can result in health related productivity loss for patients. The differences in hospital length of stay between various racial/ethnic groups has not been investigated in COVID-19. We explored these differences. Methods: We performed a nationwide registry study of US Hospital in 2020. We extracted the number of COVID-19 patients by race (Asian, African-American(AA), Caucasian) and also the patients who declined to provide race, as well those whose race was unavailable. We collected the case-mix index (CMI) for each hospital, the average length of stay (LOS) and the number of patients in each group. We applied multi-level, mixed-effects generalized linear models (GLM) to evaluate the average length of stay, adjusting for volume in cases and CMI, and a random-effect equation for each individual hospital which was deidentified and coded randomly. We back transformed the results to the original metric of LOS in days. Results: Among 583 US Hospitals, the average COVID-19 LOS was 8.5 days (95% CI 8-8.4). African Americans had an average LOS of 8.5 days (95% CI 8.1-8.8 days), Caucasians 8 days, (95% CI 7.6-8.3), while Asians 8.8 days (95% CI 8.4-9.2). Caucasians had an average of 0.5 days shorter LOS (95% -0.95 to -0.1, p=0.02). Asians had an average non-statistically significant 0.3 days longer LOS (95% CI -0.2 to 0.8, p=0.24). Conclusion: Our large nationwide sample of 583 US hospitals showed that African-American patients have longer hospital length of stay compared to Caucasian patients, after adjusting for the number of COVID-19 cases and the case mix index of each individual hospital. In our study, age, gender, and comorbidities were not adjusted for. There are a number of variables that could contribute to longer LOS for various racial/ethnic groups, including access to healthcare and healthcare utilization, health insurance status, wealth gaps, etc. Further investigation into social determinants of health and their impact on COVID-19 disease burden among various racial/ethnic groups is warranted.

4:12 PM Abstract No. 372 The role of interventional radiology in the contemporary management of splenic trauma: a national trauma data bank study

Good perioperative care is aimed at rapid recovery, without complications or readmissions. Length of stay (LOS) is influenced not only by perioperative care routines but also by patient factors, tumour factors, treatment characteristics and complications. The present study examines variation in LOS between hospitals after minimally invasive lung resections for both complicated and uncomplicated patients to assess whether LOS is a hospital characteristic influenced by local perioperative routines or other factors. Dutch Lung Cancer Audit (surgery) data were used. Median LOS was calculated on hospital level, stratified by the severity of complications. Lowest quartile (short) LOS per hospital, corrected for case-mix factors by multivariable logistic regression, was presented in funnel plots. We correlated short LOS in complicated versus uncomplicated patients to assess whether short LOS clustered in the same hospitals regardless of complications. Data from 6055 patients in 42 hospitals were included. Median LOS in uncomplicated patients varied from 3 to 8 days between hospitals and increased most markedly for patients with major complications. Considerable between-hospital variation persisted after case-mix correction, but more in uncomplicated than complicated patients. Short LOS in uncomplicated and complicated patients were significantly correlated (r = 0.53, P < 0.001). LOS after minimally invasive anatomical lung resections varied between hospitals particularly in uncomplicated patients. The significant correlation between short LOS in uncomplicated and complicated patients suggests that LOS is a hospital characteristic potentially influenced by local processes. Standardizing and optimizing perioperative care could help limit practice variation with improved LOS and complication rates.

Objective: Dysglycemia during an intensive care unit (ICU) stay has been correlated with hospital complications and increased mortality. The objective of this study was to examine how glucose derangements during an ICU admission affect length of stay (LOS) and mortality in patients with and without diabetes mellitus. Methods: Patient-level data for adults (&amp;gt;/=18 years old) admitted to any ICU within our hospital system for at least 24 hours between 5/5/2011 to 11/6/2017 (6.5 years) were pulled from the electronic database. Patients were categorized into four glucose groups: 1) no hyper/hypoglycemia, 2) hyperglycemia only (blood glucose [BG] &amp;gt;180mg/dL), 3) hypoglycemia only (BG &amp;lt;70mg/dL), and 4) both hyper/hypoglycemia. Diabetes status was determined from diagnosis codes prior to the ICU stay. Dates of hospitalization were used to calculate LOS in days. Mortality was examined through the end of the analysis period. Results: A total of 16,345 encounters were included in this analysis (mean age 57.7 years); only a few patients had multiple encounters. Over one-quarter of patients had diabetes (26.1%). A majority of encounters involved some form of dysglycemia: 16.4% hypoglycemia, 42.4% hyperglycemia, and 12.0% with both hyper/hypoglycemia. The overall average LOS was 6.03 days. Patients without diabetes but with any form of dysglycemia had longer LOS and higher mortality compared to patients with diabetes. Patients with both hyper/hypoglycemia had the least favorable outcomes (no diabetes vs. diabetes, LOS mean±SD: 16.2±23.5 vs. 12.9±16.3 days, p &amp;lt;0.01; mortality: 64.9% vs. 43.6%, p &amp;lt;0.01). Conversely, those with diabetes without any dysglycemia had the shortest LOS and lowest mortality (LOS 2.5±3.2; mortality 11.3%). Conclusion: Glucose derangements in the ICU were associated with longer LOS in all patients. However, ICU patients without diabetes but with dysglycemia had longer LOS and higher mortality compared to patients with diabetes. Disclosure A. Zaman: None. M.T. McDermott: Advisory Panel; Self; Novo Nordisk Inc., Savvy Sherpa. V.V. Mansfield: Employee; Self; UnitedHealth Group. C.C. Low Wang: Research Support; Self; CellResearch Corporation. Funding UnitedHealth Group; University of Colorado Health Care Innovation Center

Revisiting Length of Stay in Stroke Rehabilitation in Turkey

Introduction Tracheostomy is indicated in critically injured burn patients when prolonged mechanical ventilation is anticipated. Meta-analyses in non-burn patients have shown early tracheostomy is associated with fewer ventilator days and shorter length of stay (LOS). Limited single institution studies have evaluated early tracheostomy in burn patients and found no benefits. We aimed to use national data to reevaluate the timing of tracheostomy in critically injured burn patients, hypothesizing that early tracheostomy would be associated with reduced LOS and reduced ventilator-associated pneumonia (VAP). Methods Burn encounters undergoing tracheostomy were identified in three separate national databases—the Nationwide Inpatient Sample (NIS), 2012–2021, the US National Trauma Data Bank (NTDB), 2007–2018, and the Burn Care Quality Platform (BCQP), 2013-2022. Encounters were stratified by tracheostomy timing with respect to admission: early: ≤ 10 days vs. late: &amp;gt; 10 days. Propensity score matching was performed to adjust for confounding by indication utilizing 1:1 nearest-neighbor matching without replacement with maximum caliper distance of 0.1. Early tracheostomy patients were matched with late tracheostomy patients on age, sex, and total body surface area (TBSA) burned. Outcomes included overall LOS, ICU LOS, ventilator days, and VAP. Only burn survivors were included. Results Tracheostomy was performed in 10,455 encounters in NIS, 1,852 encounters in NTDB, and 1,599 patients in BCQP. The median day for tracheostomy in NIS was 12.0 (IQR, 6.0, 18.0), 10.4 in NTDB (IQR, 5.7, 16.5), and 13.0 (IQR 8.0, 19.0) in BCQP. After matching, standardized mean differences (SMDs) were ≤ 0.1. Within this matched cohort, early tracheostomy was associated with shorter LOS in all databases, reported as Average Treatment Effect: NIS -5.7 days (95% CI: -10.7– -0.7; p=0.03), NTDB -11.1 days (95% CI: -16.3– -6.0; p&amp;lt; 0.01), BCQP -9.3 days (95% CI: -13.9– -4.7; p&amp;lt; 0.01). Early tracheostomy was additionally associated with shorter ICU LOS in NTDB (-7.7 days; 95% CI: -12.2– -3.3; p&amp;lt; 0.01) and BCQP (-6.2 days; 95% CI: -10.5– -2.0; p&amp;lt; 0.01). In the BCQP, early tracheostomy showed reduced ventilator days by -5.4 days (95% CI: -9.5– -1.2; p=0.01) Early tracheostomy was not associated with reduced VAP in any database. Conclusions In a multi-database analysis with almost 14,000 encounters, early tracheostomy (within 10 days of admission) was associated with decreased ICU LOS, overall LOS, and fewer ventilator days, but did not affect VAP. Applicability of Research to Practice For patients with anticipated prolonged mechanical ventilation, burn practitioners should consider tracheostomy within 10 days of admission. Funding for the Study N/A

11084 Background: Integra Connect previously created in-patient (IP) admission prediction model based on OCM data. One of the challenges for practices in a value-based care (VBC) program is to provide continuous care-coordination during and after an IP admission. Our objective is to show how adopting combination machine learning (ML) models can predict IP length of stay (LOS). Potential benefits include an overall reduction in LOS which could minimize the risk of hospital acquired conditions. Methods: ML models were trained on 5 major cancer types (Lung Cancer, Multiple Myeloma, Lymphoma, Small Intestine / Colorectal Cancer, High-risk Breast Cancer') from 7 OCM practices of PP4-PP9 data, excluding surgery IP admission. The top 5 cancer types accounted for ~50% of total in-patient admissions and ~50% of total LOS in days. The IP admissions were divided into 4 major cohorts in terms of LOS in days (1-3: class 1, 4-8: class 2, 9-15: class 3, and more than 15: class 4). To reduce the overall LOS, we adopted two ML models: 1) To classify LOS, a multi-classification model (Model1, an eXtreme Gradient Boosted Trees Classifier) and 2) To predict the LOS for classes 1-3, a regression-based model (Model2, an eXtreme Gradient Boosted Trees Regressor with Early Stopping). Class 4 cohort had a wide range (16- 90+ days) for LOS and therefore was excluded from Model2. The other three cohorts belonged to Q1, IQR, and Q4, respectively. Model1 was trained on 4,280 randomly selected sample IP admissions to balance each class and Model2 was trained on 19,636 IP admissions, both with 102 features. Results: The models were tested on PP10-PP11 claims excluding 0 days LOS. Model1 predicted 296 (3,945) IP admissions in class 4 of which 88 were true positives. Assuming at least a 10% reduction in associated total LOS translated to lower LOS of 1-3% for 5 (7) practices. The remaining 3,649 IP admissions were used to predict LOS by Model2. The difference between actual LOS and Predicted LOS was found as 18%-24%. To address model prediction errors, we defined the target LOS to be predicted LOS with an upper bound of at least 10%. This led to reductions of 6% -16% across practices. Finally, combining the results of both ML models we determined that the potential to lower LOS was at least 6% and at most 19%. Conclusions: Our ML models identified opportunities to reduce LOS across multiple OCM cancer practices for 5 major cancer types. We also identified a cohort of patients with critical condition (class 4) that is vital for practice transformation initiatives. These identified reduction in LOS for oncology patients provides cost reduction and quality improvement opportunities in VBC programs. In the future, more opportunities to lower LOS will be explored with a re-admission prediction model.

A single-center retrospective study. To identify predictors of length of stay (LOS) days in patients undergoing 1 level minimally invasive (MIS) transforaminal lumbar interbody fusions (TLIF). Recent studies suggest intraoperative fluid administration, and colloid and crystalloid administration among other intraoperative variables may prolong LOS days and increase complications. Therefore, an understanding of which preoperative, intraoperative, and immediate postoperative parameters best predict immediate LOS days will help risk stratify patients and guide decision making. We retrospectively reviewed 104 patients undergoing a MIS TLIF at 1 institution between 2008 and 2010. Two groups were selected on the basis of the time of discharge. Group 1 consisted of patients discharged within 24 hours after surgery and group 2 consisted of patients discharged more than 24 hours after surgery. Multiple regression analysis was performed to determine which preoperative, intraoperative, and postoperative variables were independent predictors of LOS days. Seventy-eight patients (75%) with a LOS greater than 24 hours had significantly higher estimated blood loss, received more crystalloids, had higher total fluids, longer surgical time, lower end of case temperature, lower hemoglobin during hospitalization, and a lower preoperative narcotic use. Multiple regression analysis showed that significant predictors of increased LOS were postoperative creatinine, visual analogue scale score, intraoperative colloids, fluids input at the end of surgical case, crystalloid to colloid ratio, fluid balance, oxycodone (Oxycontin) use, mean percentage of fraction of inspired oxygen, and preoperative hemoglobin. Patients undergoing 1 level MIS TLIF for degenerative conditions can overall expect a short LOS postoperatively. Multiple preoperative, intraoperative, and immediate postoperative factors can prolong the LOS in this group. This information should help the surgical team in optimizing their intraoperative patient management.

IntroductionHospital mortality, readmission and length of stay (LOS) are commonly used measures for quality of care, given availability in administrative data. However, these measures are interrelated. For example, a short...

Length of Stay in the Intensive Care Unit during Index Admission for Left Ventricular Assist Device and Total Artificial Heart Insertion is a Risk Factor for Infection

Preoperative Education for Total Joint Arthroplasty: Does Reimbursement Reduction Threaten Improved Outcomes?

Tracheostomy is indicated in critically ill patients when prolonged mechanical ventilation is anticipated. We leveraged national data to evaluate tracheostomy timing in burn patients, hypothesizing that early tracheostomy would be associated with reduced length of stay (LOS) and ventilator-associated pneumonia (VAP). Surviving burn encounters undergoing tracheostomy in 3 national databases-Nationwide Inpatient Sample (NIS), 2016-2021, National Trauma Data Bank (NTDB), 2007-2014, and Burn Care Quality Platform (BCQP), 2015-2022-were stratified by tracheostomy timing relative to admission: early: ≤10 days versus late: >10 days. Early tracheostomy encounters were propensity-score-matched with late tracheostomy encounters on age, sex, and total body surface area (TBSA) of burns to evaluate the impact of tracheostomy timing on LOS, ICU LOS, ventilator days, VAP, discharge to inpatient rehabilitation, and discharge to long-term acute care (LTAC). In total, 9173 burn encounters underwent tracheostomy (6255 in NIS, 1332 in NTDB, and 1586 in BCQP), of which 51.1% were early. Within matched cohorts, early tracheostomy was associated with shorter LOS, reported as average treatment effect, in days (95% confidence interval): NIS: -22.9 (-32.8 to -13.1), P < .01; NTDB: -12.7 (-18.7 to -6.8), P < .01; BCQP: -7.0 (-12.5 to -1.5), P < .01. Early tracheostomy was associated with shorter ICU LOS and fewer ventilator days in NTDB and BCQP (P < .04). Early tracheostomy was not associated with discharge to inpatient rehabilitation or VAP. Early tracheostomy decreased discharge to LTAC in NTDB and BCQP (P ≤ .02). Our multi-database analysis supports early tracheostomy in critically injured burn patients requiring prolonged mechanical ventilation.

To provide information on live-birth prevalence and hospitalizations, including anxiety and depression, for cleft lip and/or palate (CL/P) in Western Australia (WA), using live-birth data 1980 to 2015. Retrospective data linkage. Tertiary hospital. Cleft cohort consisted of people live-born with CL/P in WA between 1980 and 2015, and a gender and age-matched control group. Live-birth prevalence for CL/P by year. Hospital event counts, event ages, and length of stay (LOS) days by 18 diagnosis groups and 4 birth year categories between the cleft cohort and control group, and between cleft types. Count of events per alive persons per calendar year, and relative risk for proportions of persons in the cleft cohort and control group by diagnosis group. Live-birth prevalence for CL/P was 19.7 per 10 000 (1 in 522). The cleft cohort had significantly higher event counts, lower event ages, and higher LOS days than the control group. Cleft lip and palate had significantly higher event counts, lower event ages, and higher LOS days than cleft lip or cleft palate only. There were 2 significant differences for anxiety or depression between the cleft cohort and control group, lower event ages, and higher LOS days in 1990s birth year category. This study provides a cleft data reference for WA. Live-birth prevalence for all clefts and by cleft type offers an appropriate method for estimating service utilization and provision. Patients with cleft accessed hospital services more frequently, at an earlier age, with higher LOS days than the control group.

Social determinants of health and disparities in spine surgery: a 10-year analysis of 8,565 cases using ensemble machine learning and multilayer perceptron