Hospital Discharge Data Research Articles

Abstract P224: Rural Hospital Closures Are Associated With Increased Mortality From Ischemic Heart Disease: A County-Level Analysis

Objective: Ischemic heart disease remains the leading cause of death in the United States; yet mortality from heart disease is geographically heterogeneously distributed. From 2011 - 2020, several hospitals across the US have closed. In this study we will investigate the impact of hospital closure on county level disease-specific mortality data. We hypothesize that rural hospital closures will be associated with increased risk of mortality from ischemic heart disease. Methods: The CDC-Wonder database, an online outcomes reporting system was queried for county level, ischemic heart disease specific mortality in Texas - the state with the largest number of hospital closures. Demographic, socioeconomic, and comorbidity information was collated from the County Health Ranking project. Hospital closure data was obtained from the Sheps Center for Health Services Research. Counties were grouped by Health Service Area (HSA) and were considered affected if any county in the HSA was affected by a hospital closure. A difference in difference analysis with unbiased staggered adoption was performed with a multivariate linear model. 4 counties with multiple hospital closures in the time frame were excluded from analysis. Results: Over a 10-years (2011 - 2020) 247,712 deaths were recorded as related to ischemic heart disease in Texas. Difference-in-difference analysis with all counties did not demonstrate a significant treatment effect of hospital closure (ATT = - 0.01, p = 0.71). Restricting comparisons to rural hospitals demonstrated a significant effect of rural hospital closures as compared to other rural counties (ARR = 0.119, p = 0.04). There was so significant effect in comparison of non-rural counties with hospital closures. Conclusions: These results demonstrate the disproportionate impact of hospital closure on rural health outcomes. As compared to patient-level analysis relying on hospital discharge data, this country level analysis captures non-hospital deaths and elucidates the true human cost of hospital closure.

Initial renal replacement therapy (RRT) modality associates with 90-day postdischarge RRT dependence in critically ill AKI survivors

PurposeReal-world comparison of RRT modality on RRT dependence at 90 days postdischarge among ICU patients discharged alive after RRT for acute kidney injury (AKI). MethodsUsing claims-linked to US hospital discharge data (Premier PINC AI Healthcare Database [PHD]), we compared continuous renal replacement therapy (CRRT) vs. intermittent hemodialysis (IHD) for AKI in adult ICU patients discharged alive from January 1, 2018 to June 30, 2021. RRT dependence at 90 days postdischarge was defined as ≥2 RRT treatments in the last 8 days. Between-group differences were balanced using inverse probability treatment weighting (IPTW). ResultsOf 34,804 patients, 3804 patients (from 382 hospitals) had claims coverage for days 83–90 postdischarge. Compared to IHD-treated patients (n = 2740), CRRT-treated patients (n = 1064) were younger; had more admission to large teaching hospitals, surgery, sepsis, shock, mechanical ventilation, but lower prevalence of comorbidities (p < 0.05 for all). Compared to IHD-treated patients, CRRT-treated patients had lower RRT dependence at hospital discharge (26.5% vs. 29.8%, p = 0.04) and lower RRT dependence at 90 days postdischarge (4.9% vs. 7.4% p = 0.006) with weighted adjusted OR (95% CI): 0.68 (0.47–0.97), p = 0.03. Results persisted in sensitivity analyses including patients who died during days 1–90 postdischarge (n = 112) or excluding patients from hospitals with IHD patients only (n = 335), or when excluding patients who switched RRT modalities (n = 451). ConclusionsAdjusted for potential confounders, the odds of RRT dependence at 90 days postdischarge among survivors of RRT for AKI was 30% lower for those treated first with CRRT vs. IHD, overall and in several sensitivity analyses. SummaryCritically ill patients in intensive care units (ICU) may develop acute kidney injury (AKI) that requires renal replacement therapy (RRT) to temporarily replace the injured kidney function of cleaning the blood. Two main types of RRT in the ICU are called continuous renal replacement therapy (CRRT), which is performed almost continuously, i.e., for >18 h per day, and intermittent hemodialysis (IHD), which is a more rapid RRT that is usually completed in a little bit over 6 h, several times per week. The slower CRRT may be gentler on the kidneys and is more likely to be used in the sickest patients, who may not be able to tolerate IHD. We conducted a data-analysis study to evaluate whether long-term effects on kidney function (assessed by ongoing need for RRT, i.e., RRT dependence) differ depending on use of CRRT vs. IHD. In a very large US linked hospital-discharge/claims database we found that among ICU patients discharge alive after RRT for AKI, fewer CRRT-treated patients had RRT dependence at hospital discharge (26.5% vs. 29.8%, p = 0.04) and at 90 days after discharge (4.9% vs. 7.4% p = 0.006). In adjusted models, RRT dependence at 90 days postdischarge was >30% lower for CRRT than IHD-treated patients. These results from a non-randomized study suggest that among survivors of RRT for AKI, CRRT may result in less RRT dependence 90 days after hospital discharge.

Treatment of Anorectal Malformations in German Hospitals: Analysis of National Hospital Discharge Data from 2016 to 2021.

Anorectal malformations (ARMs) are complex congenital anomalies. The corrective operation is demanding and schedulable. Based on complete national data, patterns of care have not been analyzed in Germany yet. All cases with ARM were analyzed (1) at the time of birth and (2) during the hospital stay for the corrective operation, based on the national hospital discharge data (DRG statistics). Patient's comorbidities, treatment characteristics, hospital structures, and the outcome of corrective operations were analyzed with respect to the hospitals' caseload. From 2016 to 2021, 1,726 newborns with ARM were treated at the time of birth in 388 hospitals. Of these hospitals, 19% had neither a pediatric nor a pediatric surgical department. At least one additional congenital anomaly was present in 49% of cases and 7% of the newborns had a birthweight below 1,500 g.In all, 2,060 corrective operations for ARM were performed in 113 hospitals in the same time period. In 24.5% of cases, at least one major complication was documented. One-third of the operations were performed in 56 hospitals, one-third in 20 hospitals, and one-third in 10 hospitals with median annual case numbers of 2, 5, and 10, respectively.Hospitals with the highest caseload operated cloacal defects more often than hospitals with the lowest caseload (7 vs. 2%) and had more early complications than hospitals with the lowest caseload (30 vs. 21%). This difference was not statistically significant after risk adjustment. Children with ARM are multimorbid. Early complications after corrective surgery are common. Considering the large number of hospitals with a very low caseload, centralization of care for the complex and elective corrective surgery for ARM remains a key issue for quality of care.

Short-term effects of positive expiratory pressure mask on ventilation inhomogeneity in children with cystic fibrosis: A randomized, sham-controlled crossover study.

Can physiotherapy with a positive expiratory pressure (PEP) mask improve peripheral ventilation inhomogeneity, a typical feature of children with cystic fibrosis (cwCF)? To answer this question, we used the nitrogen multiple-breath washout (N2MBW) test to measure diffusion-convection-dependent inhomogeneity arising within the intracinar compartment (Sacin*VT). For this randomized, sham-controlled crossover trial, two N2MBW tests were performed near the hospital discharge date: one before and the other after PEP mask therapy (1 min of breathing through a flow-dependent PEP device attached to a face mask, followed by three huffs and one cough repeated 10 times) by either a standard (10-15 cmH20) or a sham (<5 cmH20) procedure on two consecutive mornings. Deception entailed misinforming the subjects about the nature of the study; also the N2MBW operators were blinded to treatment allocation. Study outcomes were assessed with mixed-effect models. The study sample was 19 cwCF (ten girls), aged 11.4 (2.7) years. The adjusted Sacin*VT mean difference between the standard and the sham procedure was -0.015 (90% confidence interval [CI]: -∞ to 0.025) L-1. There was no statistically significant difference in Scond*VT and lung clearance index between the two procedures: -0.005 (95% CI: -0.019 to 0.01) L-1 and 0.49 (95% CI: -0.05 to 1.03) turnovers, respectively. Our findings do not support evidence for an immediate effect of PEP mask physiotherapy on Sacin*VT with pressure range 10-15 cmH20. Measurement with the N2MBW and the crossover design were found to be time-consuming and unsuitable for a short-term study of airway clearance techniques.

Identifying Newborns with Hypoxic-Ischemic Encephalopathy in Hospital Discharge Data: A Validation Study

Hospital discharge databases (HDD) are increasingly used for research on health of newborns. Linkage between a French population-based cohort of newborns with hypoxic ischemic encephalopathy (HIE) and national HDD showed that the HIE ICD-10 code was not accurately reported. Our results suggest that HDD should not be used for research on neonatal HIE. We also caution against using HIE ICD-10 codes in HDD without validation.

Cost of Cardiovascular Disease Event and Cardiovascular Disease Treatment-Related Complication Hospitalizations in the United States.

Cardiovascular disease (CVD) is among the costliest conditions in the United States, and cost-effectiveness analyses can be used to assess economic impact and prioritize CVD treatments. We aimed to develop standardized, nationally representative CVD events and selected possible CVD treatment-related complication hospitalization costs for use in cost-effectiveness analyses. Nationally representative costs were derived using publicly available inpatient hospital discharge data from the 2012-2018 National Inpatient Sample. Events were identified using the principal International Classification of Diseases, Ninth Revision and International Classification of Diseases, Tenth Revision codes. Facility charges were converted to costs using charge-to-cost ratios, and total costs were estimated by applying a published professional fee ratio. All costs are reported in 2021 US dollars. Mean costs were estimated for events overall and stratified by age, sex, and survival status at discharge. Annual costs to the US health care system were estimated by multiplying the mean annual number of events by the mean total cost per discharge. The annual mean number of hospital discharges among CVD events was the highest for heart failure (1 087 000 per year) and cerebrovascular disease (800 600 per year). The mean cost per hospital discharge was the highest for peripheral vascular disease ($33 700 [95% CI, $33 300-$34 000]) and ventricular tachycardia/ventricular fibrillation ($32 500 [95% CI, $32 100-$32 900]). Hospitalizations contributing the most to annual US health care costs were heart failure ($19 500 [95% CI, $19 300-$19 800] million) and acute myocardial infarction ($18 300, [95% CI, $18 200-$18 500] million). Acute kidney injury was the most frequent possible treatment complication (515 000 per year), and bradycardia had the highest mean hospitalization costs ($17 400 [95% CI, $17 200-$17 500]). The hospitalization cost estimates and statistical code reported in the current study have the potential to increase transparency and comparability of cost-effectiveness analyses for CVD in the United States.

Abstract TP163: Associations Between Prior Oral Anticoagulant Use and Stroke Outcomes in a National Stroke Cohort: A Scottish Database Study

Background and aims: Recent studies suggest that oral anticoagulant (OAC) use prior to stroke is associated with better functional outcomes after ischaemic stroke, but worse outcomes after non-traumatic intracerebral haemorrhage (ICH). We aimed to assess the impact of prior OAC use on stroke subtype and outcomes (mortality and discharge to home) in a Scottish stroke cohort. Methods: Data on first-ever strokes between January 1, 2011, and December 31, 2018, were gathered from the Scottish Stroke Care Audit. The Prescribing Information System was used to identify patients with a prescription for OAC within 90-days before stroke, and further linked to hospital admission and discharge data in the Scottish Morbidity Record 01, and mortality data in the National Records of Scotland. Multivariable logistic regression models were adjusted for sociodemographic factors, atrial fibrillation, stroke severity, acute management and Charlson Comorbidity scores. Results: Among 52 988 stroke patients (50.1% women; mean age 72.8 [±13.3] years), 88.9% were ischaemic and 18.3% had a history of atrial fibrillation. OACs (warfarin [73%] and direct oral anticoagulants [28%]) were prescribed to 3533 (7.5%) patients with ischaemic stroke and 847 (14.4%) patients with ICH within 90 days prior to first-ever stroke. After model adjustments, prior OAC use was associated with a three-fold increase in the odds of experiencing an ICH (adjusted Odds Ratio [aOR], 2.92; 95% CI, 2.61 - 3.26]). In patients with ischaemic stroke or ICH, prior OAC use was not associated with better or worse outcomes in 30-day discharge to usual place of residence (aOR, 0.99; 95% CI, 0.89 - 1.09 and aOR, 0.86; 95% CI, 0.66 - 1.12 respectively) or 30-day mortality (aOR, 1.06; 95% CI, 0.94 - 1.21 and aOR, 1.20, 95% CI, 0.93 - 1.56 respectively) in comparison to those with no prior OAC use. Additionally, no statistically significant differences were observed for discharge or mortality outcomes within 90 days after stroke. Conclusions: Prior OAC use was higher in patients with ICH, which may partly reflect ischaemic stroke reduction in patients taking OAC. However, prior OAC use was not associated with better or worse outcomes in terms of mortality and discharge to home in either ischaemic or ICH populations.

Is the long-term poor prognosis of acute myocardial infarction in patients with mental illness mediated through their poor adherence with recommended healthcare?

Compared with patients without evidence of psychiatric symptoms, those with mental disorders experience reduced adherence with recommended healthcare and poorer clinical outcomes. This study aimed to evaluate whether the worse prognosis of patients with mental disorders after experiencing acute myocardial infarction could be fully or partially mediated by their reduced adherence to recommended healthcare. In this retrospective cohort population-based study, 103 389 residents in the Italian Lombardy Region who experienced acute myocardial infarction in 2007-19 were identified. Among them, 1549 patients with severe mental illness (SMI) were matched with five cohort members without evidence of mental disorders (references). Recommended healthcare (cardiac medicaments and selected outpatient services) was evaluated in the year after the date of index hospital discharge. The first occurrences of cardiovascular (CV) hospital admissions and any-cause-death were considered as endpoints. Mediation analysis was performed to investigate whether post-discharge use of recommended healthcare may be considered a mediator of the relationship between healthcare exposure and endpoints occurrence. Compared with references, patients with SMI had lower adherence with recommended healthcare and adjusted risk excesses of 39% and 73% for CV hospitalizations and all-cause mortality. Mediation analysis showed that 4.1% and 11.3% of, respectively, CV hospitalizations and deaths occurred among psychiatric patients was mediated by their worse adherence to specific healthcare. The reduced use of recommended outpatient healthcare by patients with SMI had only a marginal effect on their worse prognosis. Other key factors mediating the prognostic gap between patients with and without mental disorders should be investigated.

Trends in maternal opioid use disorder and neonatal abstinence syndrome in Maine, 2016-2022.

To estimate trends in maternal opioid use disorder (OUD) and neonatal abstinence syndrome (NAS) in Maine using the most recent data available. We used hospital discharge data to estimate the annual prevalence of maternal OUD and NAS between 2016 and 2022. In addition, we used birth certificate-linked Medicaid data to estimate related trends among Medicaid enrollees. From 2016 to 2022, the prevalence of maternal OUD decreased from 35.3 to 18.8 per 1000 deliveries and the prevalence of NAS decreased from 33.2 to 14.0 per 1000 newborns (linear trend p values <0.01). Decreasing trends were also found among Medicaid enrollees. In Maine between 2016 and 2022, there was a decrease in maternal OUD and NAS diagnoses recorded in administrative datasets. These findings should be interpreted with caution due to changes in how OUD and NAS diagnoses are recorded and COVID-related changes in healthcare utilization.

Surgical and per-oral endoscopic myotomy (POEM) for the treatment of primary esophageal motility disorders: A systematic analysis of current trends in Germany between 2011 and 2019.

While surgery remains a standard treatment for primary esophageal motility disorders (PEMDs), per-oral endoscopic myotomy (POEM) has recently evolved as an alternative. Systematic data on current trends of invasive procedures for PEMDs in Germany are missing. Hospital discharge data were used to evaluate trends and mortality of invasive treatment options for PEMDs in Germany between 2011 and 2019. 4543 cases of PEMDs (achalasia: n = 4349, dyskinesia of the esophagus: n = 194) receiving open surgery (n = 200), minimal invasive surgery (n = 2366), or POEM (n = 1977) were identified. The relative proportion of POEM significantly increased from 10.9% (2011) to 65.7% (2019). Hospital mortality was 0.2%. The median duration of mechanical ventilation was significantly lower in POEM patients (29.4 hours) compared to open (274.0 hours) or minimal invasive (91.9 hours) surgery. The duration of hospitalization was lowest among POEM patients (5.7 days) compared to surgical procedures (13.7 and 7.7 days). While the low in-hospital mortality of all procedures combined confirms the solid safety profile of invasive procedures in general, our findings show that POEM has the lowest duration of mechanical ventilation and hospitalization compared to invasive surgical options.

The Effect of Limb Salvage Efforts in Different Age Groups: Relationships Between Major and Minor Amputations

“Limb salvage” efforts, such as performing minor amputations before infections spread proximally from the foot to decrease major lower extremity amputation, are an important part of healthcare today. It is unclear whether these efforts are preventing the number of major amputations and improving patients’ quality of life and the cost-effectiveness of the U.S. healthcare system. Rates of non-traumatic lower extremity amputation (NLEA) among patients with diabetes decreased in the early 2000s but rebounded in the 2010s. We analyzed the proportion of major amputations and differences in amputation rates between age groups in Texas. Patient data was extracted from the Texas Hospital Discharge Data Public Use Data File. Population estimates were obtained from the Texas Population Estimates Program from 2011 to 2015 and from intercensal estimates provided by the U.S. Census Bureau from 2006 to 2010. Raw numbers of minor, major, and all NLEA surgeries and the ratio of major amputations to total amputations per year were reported for each age group. Poisson regression and Joinpoint analyses were performed to capture these changes in trends. Rates of amputations increased, with significant decreasing relative prevalence of major amputations. Patients aged 45 to 64 with diabetes are likely driving these increases. Rates of lower extremity amputation in patients with diabetes increased from 2009 to 2015. This holds for all and minor amputations. In contrast, the ratio of major to all amputations decreased from 2010. Utilization of major and minor amputation differs between age groups, remaining stable in the youngest subjects, with minor amputation rates increasing in those aged 45 to 64.

Streptococcus pneumoniae Infection in Patients with Asplenia: A Spanish Perspective over a 25-Year Period.

Anatomical or functional asplenia constitutes a risk factor for Streptococcus pneumoniae (SP) infection, being more frequent in children and the elderly and in people with multiple comorbidities. We aimed to describe the impact of invasive pneumococcal disease (IPD) on the clinical features and outcomes of patients hospitalized for asplenia in Spain. Discharge reports from the Spanish Minimum Basic Data Set were used to retrospectively analyze hospital discharge data with a diagnosis of asplenia from 1997 to 2021. A total of 132,257 patients with asplenia (splenectomized/non-splenectomized) were identified from the Spanish database. Among the cases, 177 (37.5%) patients with splenectomy and 295 (62.5%) patients without splenectomy developed IPD. The clinical presentations (non-infection vs. infection) did not significantly differ between the two reference groups, except for patients with COPD, rheumatoid disease, AIDS, other neurological disorders, metastatic cancer, and drug abuse. The risk factors for IPD were also more frequently reported in patients without splenectomy (p < 0.001) and with comorbidities (p = 0.005). The study of patients with asplenia provides relevant information about the state of SP infection. This epidemiological tracking can serve to better understand the comorbidities that affect them, the risk factors for the disease, the prediction of antibiotic use, and vaccination in public health, among other factors.

Early impacts of the Pennsylvania Rural Health Model on potentially avoidable utilization.

The Pennsylvania Rural Health Model (PARHM) is a novel alternative payment model for rural hospitals that aims to test whether hospital-based global budgets, coupled with delivery transformation plans, improve the quality of health care and health outcomes in rural communities. Eighteen hospitals joined PARHM in 3 cohorts between 2019 and 2021. This study assessed PARHM's impact on changes in potentially avoidable utilization (PAU)-a measure of admission rates policymakers explicitly targeted for improvement in PARHM. Using a difference-in-differences analysis and all-payer hospital discharge data for Pennsylvania hospitals from 2016 through 2022, we found no significant overall reduction in community-level PAU rates up to 4 years post-PARHM implementation, relative to changes in rural Pennsylvania communities whose hospitals did not join PARHM. However, heterogeneous treatment effects were observed across cohorts that joined PARHM in different years, and between critical access vs prospective payment system hospitals. These findings offer insight into how alternative payment models in rural health care settings may have heterogeneous impacts based on contextual factors and highlight the importance of accounting for these factors in proposed expansions of alternative payment models for rural health systems.

Incidence and Time Course of Symptomatic Thromboembolic Outcomes After Lower Extremity Arthroscopic Surgery, Ankle Fracture Surgery, and Achilles Tendon Repair.

The incidence and time course of acute venous thromboembolism (VTE) after ambulatory surgery for lower extremity orthopaedic conditions is not well-defined. The purpose of this study was to analyze the incidence, the time course, and risk factors associated with clinically diagnosed acute deep vein thrombosis or pulmonary embolism within 3 months of surgery in patients undergoing specific operations for lower extremity injuries. Patients undergoing arthroscopic procedures of the knee, ankle fracture surgery, Achilles tendon repair, and ankle arthroscopy from January 1, 2005, to December 31, 2010, were identified in the California Ambulatory Surgery database with linkage to hospital discharge data, emergency department data, and a death registry. Outcomes were acute VTE and death within 90 days. Time courses were compared using Kaplan-Meier analysis, and risk factors were analyzed using proportional hazard modeling. Analysis of data from 468,699 surgeries showed that the cumulative incidence of acute VTE was significantly higher after Achilles tendon repair (0.72%, P < 0.001) than ankle fracture surgery (0.33%), knee arthroscopy procedures (range, 0.29% to 0.41%), or ankle arthroscopy (0.24%). The time course of diagnosis of VTE was similar for all arthroscopic procedures (median postoperative day for diagnosis = 9 to 10; 80% by 22 to 36 days), whereas for Achilles tendon surgery, the time course was protracted (median postoperative day for diagnosis = 29 days; 80% by 51 days). Ninety-day mortality was low (<0.06%) after all procedures except ankle fracture (0.12%). Predictors of pulmonary embolism included age older than 60 years (HR, 3.1; 95% CI; 2.0 to 4.8, versus younger than 30 years), Achilles tendon repair (HR, 3.8; 95% CI; 2.8 to 5.3), and ankle fracture surgery (Hazard Ratio [HR], 2.1; 95% Confidence Interval [CI]; 1.5 to 2.8); Asian/Pacific Islander (HR, 0.3; 95% CI; 0.1 to 0.6) and Hispanic patients (HR, 0.5; 95% CI; 0.4 to 0.7) had significantly lower risk. The incidence and time course of onset of acute VTE after lower extremity orthopaedic surgeries varies significantly depending on the surgical procedure. These findings have implications regarding the use and duration of pharmacologic thromboprophylaxis.

Validity of Birth Certificate Data Compared With Hospital Discharge Data in Reporting Maternal Morbidity and Disparities.

A growing number of studies are using birth certificate data, despite data-quality concerns, to study maternal morbidity and associated disparities. We examined whether conclusions about the incidence of maternal morbidity, including Black-White disparities, differ between birth certificate data and hospitalization data. Using linked birth certificate and hospitalization data from California and Michigan for 2018 (N=543,469), we found that maternal morbidity measures using birth certificate data alone are substantially underreported and have poor validity. Furthermore, the degree of underreporting in birth certificate data differs between Black and White individuals and results in erroneous inferences about disparities. Overall, Black-White disparities were more modest in the birth certificate data compared with the hospitalization data. Birth certificate data alone are inadequate for studies of maternal morbidity and associated racial disparities.

Dwelling proximity to wildfire and spina bifida risk in offspring

Objective To evaluate the association between wildfire exposure in pregnancy and spina bifida risk. Methods This retrospective cohort study used the California Office of Statewide Health Planning and Development Linked Birth File with hospital discharge data between 2007 and 2010. The Birth File data were merged with the California Department of Forestry and Fire Protection data of the same year. Spina bifida was identified by its corresponding ICD-9 code listed on the hospital discharge of the newborn. Wildfire exposure was determined based on the zip code of the woman’s home address. Pregnancy was considered exposed to wildfire if the mother lived within 15 miles of a wildfire during the pregnancy or within 30 days prior to pregnancy. Results There were 2,093,185 births and 659 cases of spina bifida between 2007 and 2010. The births were analyzed using multivariable logistic regression models and adjusted for potential confounders. Exposure to wildfire in the first trimester was associated with higher odds of spina bifida (aOR= 1.43 [1.11–1.84], p-value = 0.01). Wildfire exposure 30 days before the last menstrual period and during the second and third trimesters were not associated with higher spina bifida risk. Conclusion Wildfire exposure has shown an increased risk of spina bifida during the early stages of pregnancy.

HIP fracture Supplemental Therapy to Enhance Recovery (HIPSTER): a protocol for a randomised controlled trial

IntroductionHip fractures result in substantial health impacts for patients and costs to health systems. Many patients require prolonged hospital stays and up to 60% do not regain their prefracture level...

Understanding Factors Influencing Geographic Variation in Healthcare Expenditures: A Small Areas Analysis Study.

Dramatic geographic variations in healthcare expenditures were documented by developed countries, but little is known about such variations under China's context, and what causes such variations. This study aims to examine variations of healthcare expenditures among small areas and to determine the associations between demand-, supply-factors, and per capita inpatient expenditures. This cross-sectional study utilized hospital discharge data aggregated within delineated hospital service areas (HSAs) using the small-area analysis approach. Linear multivariate regression modeling with robust standard errors was used to estimate the sources of variation of per capita inpatient expenditures across HSAs covering the years 2017 to 2019; the Shapley value decomposition method was used to measure the respective contributions of demand-, supply-side to such variations. Among 149 HSAs, demand factors explained most of the (87.4%) overall geographic variation among HSAs. With each 1% increase in GDP per capita and urbanization rate was associated with 0.099% and 0.9% increase in inpatient expenditure per capita, respectively, while each 1% increase in the share of females and the unemployment rate was associated with a 0.7% and 0.4% reduction in the per capita inpatient expenditures, respectively. In supply-side, for every 1 increase in hospital beds per 1000 population, the per capita inpatient expenditures rose by 2.9%, while with every 1% increase in the share of private hospitals, the per capita inpatient expenditures would decrease by 0.4%. With Herfindahl-Hirschman Index decrease 10%, the per capita inpatient expenditures would increase 1.06%. This study suggests demand-side factors are associated with large geographic variation in per capita inpatient expenditures among HSAs, while supply-side factors played an important role. The evaluation of geographic variations in per capita inpatient expenditures as well as its associated factors have great potential to provide an indirect approach to identify possibly existing underutilized or overutilized healthcare procedures.

Unsupervised clustering analysis of trauma/non-trauma centers using hospital features including surgical care.

Injuries are a leading cause of death in the United States. Trauma systems aim to ensure all injured patients receive appropriate care. Hospitals that participate in a trauma system, trauma centers (TCs), are designated with different levels according to guidelines that dictate access to medical and research resources but not specific surgical care. This study aimed to identify patterns of injury care that distinguish different TCs and hospitals without trauma designation, non-trauma centers (non-TCs). We extracted hospital-level features from the state inpatient hospital discharge data in Washington state, including all TCs and non-TCs, in 2016. We provided summary statistics and tested the differences of each feature across the TC/non-TC levels. We then conducted 3 sets of unsupervised clustering analyses using the Partition Around Medoids method to determine which hospitals had similar features. Set 1 and 2 included hospital surgical care (volume or distribution) features and other features (e.g., the average age of patients, payer mix, etc.). Set 3 explored surgical care without additional features. The clusters only partially aligned with the TC designations. Set 1 found the volume and variation of surgical care distinguished the hospitals, while in Set 2 orthopedic procedures and other features such as age, social vulnerability indices, and payer types drove the clusters. Set 3 results showed that procedure volume rather than the relative proportions of procedures aligned more, though not completely, with TC designation. Unsupervised machine learning identified surgical care delivery patterns that explained variation beyond level designation. This research provides insights into how systems leaders could optimize the level allocation for TCs/non-TCs in a mature trauma system by better understanding the distribution of care in the system.

A Discrete Bayesian Network for Analyzing Hospital Discharge Data

A Discrete Bayesian Network for Analyzing Hospital Discharge Data