Utilization Metrics Research Articles

Development of a Simple Index to Measure Overuse of Diagnostic Testing at the Hospital Level Using Administrative Data.

We developed a diagnostic overuse index that identifies hospitals with high levels of diagnostic intensity by comparing negative diagnostic testing rates for common diagnoses. We prospectively identified candidate overuse metrics, each defined by the percentage of patients with a particular diagnosis who underwent a potentially unnecessary diagnostic test. We used data from seven states participating in the State Inpatient Databases. Candidate metrics were tested for temporal stability and internal consistency. Using mixed-effects ordinal regression and adjusting for regional and hospital characteristics, we compared results of our index with three Dartmouth health service area-level utilization metrics and three Medicare county-level cost metrics. The index was comprised of five metrics with good temporal stability and internal consistency. It correlated with five of the six prespecified overuse measures. Among the Dartmouth metrics, our index correlated most closely with physician reimbursement, with an odds ratio of 2.02 (95% CI, 1.11-3.66) of being in a higher tertile of the overuse index when comparing tertiles 3 and 1 of this Dartmouth metric. Among the Medicare county-level metrics, our index correlated most closely with standardized costs of procedures per capita, with an odds ratio of 2.03 (95% CI, 1.21-3.39) of being in a higher overuse index tertile when comparing tertiles 3 and 1 of this metric. We developed a novel overuse index that is preliminary in nature. This index is derived from readily available administrative data and shows some promise for measuring overuse of diagnostic testing at the hospital level.

Impact of Digital Health Monitoring in the Management of Inflammatory Bowel Disease.

Technological advances now permit self-management strategies using mobile applications which could greatly benefit patient care. The purpose of this study was to investigate whether the use of the inflammatory bowel disease (IBD) digital health monitoring platform, HealthPROMISE, leads to better quality of care and improved health outcomes in IBD patients. IBD patients were recruited in gastroenterology clinics and asked to install the HealthPROMISE application onto their smartphones. Patient satisfaction, quality of care, quality of life, patient symptoms, and resource utilization metrics were collected throughout the study and sent directly to their healthcare teams. Patients with abnormal symptom/SIBDQ scores were flagged for their physicians to follow up. After one-year, patient outcome metrics were compared to baseline values. Overall, out of 59 patients enrolled in the study, 32 patients (54%) logged into the application at least once during the study period. The number of IBD-related ER visits/hospitalizations in the year of use compared to the prior year demonstrated a significant decrease from 25% of patients (8/32) to 3% (1/32) (p = 0.03). Patients also reported an increase in their understanding of the nature/causes of their condition after using the application (p = 0.026). No significant changes were observed in the number of quality indicators met (p = 0.67) or in SIBDQ scores (p = 0.48). Given the significant burden of IBD, there is a need to develop effective management strategies. This study demonstrated that digital health monitoring platforms may aid in reducing the number of ER visits and hospitalizations in IBD patients.Supplementary InformationThe online version contains supplementary material available at 10.1007/s10916-021-01706-x.

Antibiotic use in Brazilian hospitals in the 21st century: a systematic review.

INTRODUCTION This systematic review aimed to assess antibiotic use in Brazilian hospitals in the 21st century, as well as to understand the different drug utilization metrics adopted to assess the consumption of these drugs. METHODS We systematically reviewed five databases (MEDLINE [Medical Literature Analysis and Retrieval System Online], CENTRAL [The Cochrane Central Register of Controlled Trials], EMBASE® [Excerpta Medica Database], Scopus [Elsevier’s abstract and citation database], and LILACS [Literatura Latino-Americana e do Caribe em Ciências da Saúde]) for observational or experimental studies that assessed antibiotic utilization in Brazilian hospitals. The main outcomes were the drug utilization metrics and the consumption of antibiotics. RESULTS We included 23 studies, of which 43.5% were carried out in adult and pediatric care units, 39.1% in adult units, and 17.4% in pediatric units. Regarding the complexity of healthcare, 26.1% of the studies were performed in intensive care units. Two drug utilization metrics were used in these studies: the defined daily dose (DDD) and the percentage of antibiotic prescriptions. The most commonly used antibiotic classes were third-generation cephalosporins, carbapenems, fluoroquinolones, and combinations of penicillins when the DDD was the adopted drug utilization metric. CONCLUSIONS Although few studies have been conducted, existing data indicate a high use of broad-spectrum antibiotics. We found that the lack of standardized antibiotic utilization metrics impaired the mapping of drug consumption at the national level.

Understanding Scanner Utilization With Real-Time DICOM Metadata Extraction.

Understanding system performance metrics ensures better utilization of the radiology resources with more targeted interventions. The images produced by radiology scanners typically follow the DICOM (Digital Imaging and Communications in Medicine) standard format. The DICOM images consist of textual metadata that can be used to calculate key timing parameters, such as the exact study durations and scanner utilization. However, hospital networks lack the resources and capabilities to extract the metadata from the images quickly and automatically compute the scanner utilization properties. Thus, they resort to using data records from the Radiology Information Systems (RIS). However, data acquired from RIS are prone to human errors, rendering many derived key performance metrics inadequate and inaccurate. Hence, there is motivation to establish a real-time image transfer from the Picture Archiving and Communication Systems (PACS) to receive the DICOM images from the scanners to research clusters to conduct such metadata processing to evaluate scanner utilization metrics efficiently and quickly. This paper analyzes the scanners’ utilization by developing a real-time monitoring framework that retrieves radiology images into a research cluster using the DICOM networking protocol and then extracts and processes the metadata from the images. Our proposed approach facilitates a better understanding of scanner utilization across a vast healthcare network by observing properties such as study duration, the interval between the encounters, and the series count of studies. Benchmarks against using the RIS data indicate that our proposed framework based on real-time PACS data estimates the scanner utilization more accurately. Furthermore, our framework has been running stable and performing its computation for more than two years on our extensive healthcare network in pseudo real-time.

Utility-Driven k-Anonymization of Public Transport User Data

In this article, we propose a k-anonymity approach that prioritizes the generalization of attributes based on their utility. We focus on transport data, which we consider a special case in which many or all attributes are quasi-identifiers (e.g., origin, destination, ride start time), as they allow correlation with easily observable auxiliary data. The novelty in our approach lies in introducing normalization techniques as well as distance and utility metrics that allow the consideration of not only numerical attributes but also categorical attributes by representing them in tree or graph form. The prioritization of the attributes in the generalization process is based on the attributes' utility and can further be influenced by either automatically or manually assigned attribute weights. We evaluate and compare different options for all components of our mechanism as well as present an extensive performance evaluation of our approach using real-world data. Lastly, we show in which cases suppression of records can counter-intuitively lead to higher data utility.

Variations in Secondary Health Resource Utilization Post Clozapine Initiation

This study looked at secondary care utilization metrics as an effectiveness indicator of Clozapine initiation in 77 patients over the following year compared to the year prior to initiation. It was found that there was a marginal drop in psychiatric bed utilization, alongside reduced crisis team use. Medical and psychiatric outpatient use increased, as did physical health interventions. Overall, compared with costs over the year preceding Clozapine initiation, an estimated saving per patient of £3,867 ($5,065) was achieved.

Abstract 13218: Machine Learning Can Identify Cardiology Patients With High Future Healthcare Utilization in a Large Regional Health System

Introduction: Healthcare cost has increased drastically in the last decade, and over 50% of the cost can be attributed to a small portion (5-10%) of the population. Certain clinical programs, such as home-based care, aim to reduce this utilization but need methods to identify the most appropriate patients to enroll. We hypothesized that machine learning can predict patients with high future utilization with good accuracy. Methods: 683,160 cardiology patients (defined broadly as those with an ECG, echocardiogram or cardiology visit) with ~17 million clinical episodes since 2004 were identified from Geisinger’s electronic health records. Utilization was estimated as total cost of care for outpatient, inpatient and emergency department visits. Patients with the highest 10% utilization in a given year were defined as high utilizers. Machine learning models were used to predict high utilization over the next 3, 6 and 12 months. Input variables (n=191) included age, sex, smoking, 5 vital signs, 21 labs, 18 medications (current and past), 40 ECG and 44 echocardiographic measurements, 43 comorbidities, 7 time / cyclical features, 6 past utilization metrics and 4 social metrics (e.g. distance to healthcare facilities). Results: XGBoost achieved the best performance with areas under the ROC curve (AUC) of 0.82, 0.81 and 0.78 for 3, 6, 12-month models, and average precision scores (AP) of 0.31, 0.36 and 0.37, respectively, while the commonly used Charlson Comorbidity Index had poor performance with AUCs of 0.63 - 0.64 and APs of 0.1 - 0.17. Past utilization was the best predictor of future utilization. Targeting patients with the top 5 and 10% highest risk for utilization achieved sensitivities of 26 and 40% and positive predictive values of 50 and 38% (12-month model, Figure). Conclusions: Machine learning can be used to predict which patients will have high future healthcare utilization. This may help target populations for intervention programs aimed at reducing utilization.

Optimizing the synthesis of clinical trial data using sequential trees.

ObjectiveWith the growing demand for sharing clinical trial data, scalable methods to enable privacy protective access to high-utility data are needed. Data synthesis is one such method. Sequential trees are commonly used to synthesize health data. It is hypothesized that the utility of the generated data is dependent on the variable order. No assessments of the impact of variable order on synthesized clinical trial data have been performed thus far. Through simulation, we aim to evaluate the variability in the utility of synthetic clinical trial data as variable order is randomly shuffled and implement an optimization algorithm to find a good order if variability is too high.Materials and MethodsSix oncology clinical trial datasets were evaluated in a simulation. Three utility metrics were computed comparing real and synthetic data: univariate similarity, similarity in multivariate prediction accuracy, and a distinguishability metric. Particle swarm was implemented to optimize variable order, and was compared with a curriculum learning approach to ordering variables.ResultsAs the number of variables in a clinical trial dataset increases, there is a pattern of a marked increase in variability of data utility with order. Particle swarm with a distinguishability hinge loss ensured adequate utility across all 6 datasets. The hinge threshold was selected to avoid overfitting which can create a privacy problem. This was superior to curriculum learning in terms of utility.ConclusionsThe optimization approach presented in this study gives a reliable way to synthesize high-utility clinical trial datasets.

Synthetic Microdata for Establishment Surveys Under Informative Sampling

AbstractMany agencies are investigating whether releasing synthetic microdata could be a viable dissemination strategy for highly sensitive data, such as business data, for which disclosure avoidance regulations otherwise prohibit the release of public use microdata. However, existing methods assume that the original data either cover the entire population or comprise a simple random sample, which limits the application of these methods in the context of survey data with unequal weights. This paper discusses synthetic data generation under informative sampling. To utilise design information in survey weights, we rely on the pseudo likelihood approach when building a hierarchical Bayesian model to estimate the distribution of the finite population. Then, synthetic populations are randomly drawn from the estimated finite population density. We present the full conditional distributions of the Markov chain Monte Carlo algorithm for posterior inference with the pseudo likelihood function. Using simulation studies, we show that the suggested synthetic data approach offers high utility for design-based and model-based analyses while offering a high level of disclosure protection. We apply the proposed method to a subset of the 2012 U.S. Economic Census and evaluate results with utility metrics and disclosure avoidance metrics under data attacker scenarios commonly used for business data.

Burden of Illness and Outcomes in the 2nd Line Treatment of Large B-Cell Lymphoma: A Real-World Comparison of Medicare Beneficiaries with and without Stem Cell Transplants

Introduction: Standard first line therapy (LOT) for patients (pts) with large B-cell lymphoma (LBCL) involves chemotherapy with rituximab, cyclophosphamide, doxorubicin, vincristine and prednisone (R-CHOP) or a similar regimen, e.g. R-CHOP with the addition of etoposide (R-EPOCH). While most pts respond, 30-40% of pts relapse or fail to achieve remission with 1st LOT. For those pts, the goal of curative 2nd LOT is typically platinum-based salvage chemotherapy in preparation for autologous stem cell transplant (ASCT). However, many pts (50% or more) do not qualify for intensive therapy, either due to age or comorbidities, or because they do not respond to or cannot tolerate salvage chemotherapy. For those further treatment options are limited. Objectives: In a sample of U.S. LBCL pts 65 and older, who have progressed to 2nd LOT and are eligible for ASCT, to compare the patient characteristics, treatments, costs and overall survival between pts who received SCT with those who did not. Methods: This retrospective cohort study used 100% Medicare Fee-for-Service Parts A/B claims data and Part D Prescription Drug Event data to identify pts ≥ 65 years of age, newly diagnosed with LBCL in 2012-2017, and who had ≥ 6 months continuous enrollment pre-diagnosis (baseline) and ≥ 12 months post-diagnosis, or who died after initiation of 2nd LOT. All were treated with a CHOP-like regimen 1st LOT and progressed to 2nd LOT. Eligible pts who received a 2nd LOT with a SCT-preparative regimen per NCCN guidelines were stratified into the "ASCT-intended" cohort, and the remainder into the "ASCT-not-intended" cohort. The ASCT-intended group was further subset into those who received a SCT and those who did not (Non-SCT). These 2 subsets formed the primary comparison groups. Baseline characteristics were age, gender, race, dual eligibility status (i.e. Medicare plus Medicaid), and Charlson-Deyo Comorbidity Index (CCI). Measures of utilization during follow-up (from initiation of 2nd LOT to loss to follow-up) were hospitalizations, outpatient and emergency department (ED) visits, and post-acute care. Costs during follow-up reflected standardized, inflation-adjusted, all-cause paid amounts. Utilization and cost metrics were standardized to per-patient-per-month (PPPM). Overall survival (OS) was measured from the start of 2nd LOT until death, disenrollment or the end of the study period. Results: 4,758 pts met all inclusion criteria. 3,713 (78%) did not receive a SCT-preparative regimen after 1st LOT. 1,045 pts (22%) received a SCT-preparative regimen and constituted the ASCT-intended group. Of these, 244 (23.3%) received a SCT and 801 (76.7%) did not. The remainder of this report focuses on a comparison of the SCT and non-SCT cohorts. Baseline characteristics, treatments, utilization and cost are summarized in Table 1. Both groups were predominantly white, but SCT pts were, on average, 5 years younger, slightly more likely to be male, and had slightly lower mean CCI scores. 13.1% of non-SCT pts were dual eligible vs. 4.5% of SCT (all p < .01 except Race). 93.9% of SCT procedures were autologous, and no pts had more than 1 SCT during follow-up. Non-SCT pts had higher all-cause utilization for all categories except for acute hospitalizations. The higher inpatient utilization for the SCT group may reflect the SCT procedure itself (94% of SCT were inpatient). A similar pattern was observed in the cost data, where Non-SCT pts incurred higher costs except for acute hospitalizations. Results for OS are summarized in Figure 1. Non-SCT pts had significantly lower rates of survival than SCT pts. 94.7% of SCT pts were still alive 6 months after initiation of 2nd LOT, whereas only 66.9% of non-SCT pts were. At two years post-2nd LOT, survival rates were 69.7% vs. 36.3%. Conclusions: In this study, only 23.3% of SCT-intended pts ultimately received it. The remaining 76.7% were SCT-intended but never received a transplant. Compared to the SCT pts, the non-SCT pts were older, had more comorbidities, had higher rates of healthcare utilization and costs post-2nd line therapy in all categories (except acute inpatient), and lower OS. 38.7% of non-SCT pts proceeded to additional lines of lymphoma-directed therapies beyond 2nd line, which may suggest that their clinicians felt that there was potential benefit to further treatment. These data suggest there may be unmet need for elderly pts with LBCL who are eligible for ASCT, but do not ultimately receive it. Disclosures Kilgore: Kite, A Gilead Company: Research Funding. Wong:Kite, A Gilead Company: Research Funding. Thornton Snider:Precision Medicine Group: Ended employment in the past 24 months, Research Funding; Gilead Sciences: Current equity holder in publicly-traded company, Research Funding; Kite, A Gilead Company: Current Employment. Cheng:Kite, A Gilead Company: Current Employment; Gilead Sciences: Current equity holder in publicly-traded company. Schroeder:Kite, A Gilead Company: Research Funding. Mohammadi:Kite, A Gilead Company: Research Funding.

Highlighting and overcoming data barriers: creating open data for retrospective analysis of US electric power systems by consolidating publicly available sources

Studies of power system operation commonly draw from two key databases produced by the US Environmental Protection Agency: the Acid Rain Program’s Continuous Emission Monitoring System (CEMS) data, and the Emissions and Generation Resource Integrated Database (eGRID). Separate reporting requirements and heterogeneity in data aggregation between these two databases creates a barrier to systematic spatial and temporal retrospective power system analysis. This work describes the inherent challenges to this undertaking and documents a method for reconciling the two seemingly disparate data sources. While fundamental differences in data reporting and aggregation prevent us from achieving full coverage, this work represents an important initial step to aligning these two repositories of US power system data. We demonstrate the value of this linkage by computing relative unit-level, hourly utilization metrics for most thermal power plants in the US. Analysis of these metrics across time illustrates thermal generator cycling trends in California between 2011 and 2017. These unit-level results indicate that combined cycle units within California increased their part-load generation by 15% and resultant CO2 emissions by 17% and decreased their start/stop frequency over time by 8.5% and resultant emissions by 47%. Open cycle gas turbines overall increased their generation- number of start/stop cycles by 97% and resultant emissions by 85%, part-load generation by 120% and resultant emissions by 100%, and full load generation by 40% and resultant emissions 18%. We also observe a temporal shift in thermal generation from morning hours to evening in California.

De-identification of electronic health record using neural network

According to a recent study, around 99% of hospitals across the US now use electronic health record systems (EHRs). One of the most common types of EHR is the unstructured textual data, and unlocking hidden details from this data is critical for improving current medical practices and research endeavors. However, these textual data contain sensitive information, which could compromise our privacy. Therefore, medical textual data cannot be released publicly without undergoing any privacy-protective measures. De-identification is a process of detecting and removing all sensitive information present in EHRs, and it is a necessary step towards privacy-preserving EHR data sharing. Over the last decade, there have been several proposals to de-identify textual data using manual, rule-based, and machine learning methods. In this article, we propose new methods to de-identify textual data based on the self-attention mechanism and stacked Recurrent Neural Network. To the best of our knowledge, we are the first to employ these techniques. Experimental results on three different datasets show that our model performs better than all state-of-the-art mechanism irrespective of the dataset. Additionally, our proposed method is significantly faster than the existing techniques. Finally, we introduced three utility metrics to judge the quality of the de-identified data.

307 Descriptive and Retrospective Analysis of a Triggering Tool Used to Identify Emergency Department Patients With Unmet Palliative Care Needs

Patients with serious or advanced illness are likely to find themselves in an emergency department (ED) at some point along their trajectory of illness. Identification of appropriate patients as they present to the ED can be a challenge and has been extensively studied. It has been shown that > 50% of inpatient palliative care (PC) consults are generated by ED admitted patients however < 7% consultations originated while in ED . Early PC consults are shown to decrease inpatient length of stay and demonstrate a trend in optimized utilization metrics. Our aim using a novel triggering tool was to understand: 1) volume of ED based palliative medicine consultations, 2) why palliative patients are utilizing the ED, and 3) if this tool could predict patients who subsequently obtained supportive services, or had hospital readmissions, multiple ED visits, critical care admissions, or contributed to hospital mortality over a 1yr period. In June 2016, we trialed a simple triggering tool (fig1) completed by ED physicians to help identify patients presenting to the ED with advanced illness and unmet PC needs. A retrospective chart review of the patients identified during the month of June 2016 was then performed to describe health care utilization metrics in a subsequent 1 year period ending June 2017. Specifically, number of ED visits, hospital readmissions, total hospitalized days, critical care admissions, readmission risk score, palliative care consultations, hospice enrollment, and mortality in a subsequent 1-year period ending June 2017 was extracted. Continuous variables were summarized with median and range while categorical variables were summarized with frequency and percent. All tests are two-sided and p-values less than 0.05 are considered statistically significant. In June 2016, 65 encounters were generated with 58 unique patients; 5 patients had 2 or more ED visits during the month. Patient characteristics at their sentinel ED visit are summarized: median age was 73.5y (Range: 45, 101). The most frequent diagnoses were metastatic cancer (43%), advanced CNS disease (21%) and CHF (19%). 66% of patients were admitted and 22% to ICU/PCU. 15.8% had a prior palliative consult and NONE of these were admitted to the ICU. 80% of all had more than one noted unmet PC need (Step 2). ED consultants suggested 5 (9%) could avoid the ICU. Of patients’ eligible, according to Step 1, 28% (16/58) had more than one life-limiting diagnosis (range1-3). Fifty-two (90%) eligible patients had more than one Step 2 palliative modifier (median 3;Range: 0, 6). The most common were predicted death within 12 months (59%), functional decline (52%), and uncontrolled symptoms (45%). Further investigation revealed that patients who had more life limiting diagnoses also had more palliative modifiers (P=0.031). No life-limiting diagnoses (Step 1) or palliative modifiers (Step 2) were indicative that the patient would be readmitted within 30 days, readmitted within 6 months, and/or die within 1 year of their sentinel visit. Finally, a higher proportion of patients who may have benefited from an inpatient consult died within one year of their sentinel ED visit (P=0.007). Patients presenting to the ED with life-limiting diagnoses often have unmet palliative care needs and would likely benefit from PC services provided in the ED. We have begun a prospective study using meaningful Step1/Step 2 triggers and hope to suggest more meaningful indications for the ED-based PC consult.

Knowledge representation in goal-motivated decision making for intelligent active sonar

We consider knowledge representation in a goal-driven model for intelligent, or cognitive, active sonar designed to improve system performance and reduce sonar operator burden. The intelligent sonar uses observations of the environment (broadly defined) and predictions of future behavior to make decisions that leverage system resources to address a set of goals. These goals are created and managed by the system itself with possible input from the sonar operator. Decisions take the form of action selections, where actions may include transmit waveform, ping interval, illuminated region, etc. Each action is assigned a utility score, or metric, based on how well it addresses each system goal, with weighting according to how simultaneous (possibly competing) goals are prioritized by the system and/or operator. Metrics for evaluating candidate actions rely on signal processing models for localization and tracking in active sonar. To illustrate the impact of the structure of utility metrics on system behavior, we study the performance and decision-making behavior of the intelligent sonar system in simulated scenarios in which the system is tasked with addressing multiple search and track goals simultaneously. The effect of the chosen metric structure on the impact of goal prioritization and operator input is explored and discussed.

Randomized controlled trial evaluating a collaborative model of care for transitioning children with medical complexity from hospital to home healthcare: Study protocol

This article describes the study protocol for an evaluation of an innovative model of care that supports home health nurses (HHN) who serve children with medical complexity (CMC). CMC constitute a small proportion of children, but have very high need for health services, are hospitalized frequently, and account for significant proportion of pediatric healthcare expenditures. High-quality home health nursing services are important for CMC, but models of care of home healthcare, after discharge of CMC from the hospital, have not been tested. Our project aims are to develop, implement, and test a model of care, called ICollab, to improve home healthcare delivery for CMC. The ICollab model consists of collaboration between HHN, primary-care physicians and clinicians of the complex care program of a tertiary-care children's hospital in the care of CMC. In this randomized clinical trial, we will recruit 110 CMC discharged home on home health nursing services. The intervention group (n = 55) will receive the ICollab intervention for 6 months post-discharge from the hospital, in addition to usual care. Children in the control group (n = 55) will receive only usual care. Outcome measures will include healthcare utilization metrics (hospitalization rates, emergency room visit rates, and days to readmission), caregiver burden and caregiver satisfaction with home healthcare, HHN retention, and HHN collaboration with other healthcare providers. We hypothesize that ICollab will reduce healthcare utilization and caregiver burden, and improve caregiver satisfaction with home healthcare, increase HHN retention, and increase HHN collaboration with other healthcare providers. Results of this study have the potential to provide a critically needed evidence-base for interventions to improve the quality of healthcare delivery for CMC. This study is registered on clinicaltrials.gov (NCT03978468) and is ongoing.

Health Care Utilization in Patients Undergoing Repeat Stereotactic Radiosurgery for Vestibular Schwannoma with 5-Year Follow-up: A National Database Analysis.

Background Stereotactic radiosurgery (SRS) has been used as an alternative to microsurgery in patients with small vestibular schwannoma (VS). We compare health care utilization metrics in patients undergoing repeat-SRS (re-SRS) and no repeat SRS (nr-SRS) at long-term follow-up. Materials and Methods We queried the MarketScan database using International Classification of Diseases, Ninth Revision and Current Procedural Terminology 4, from 2000 to 2016. We included adult patients who had diagnosis of VS and treatment with SRS with at least 5 years of continuous enrollment after the procedure. Outcomes were hospital admissions, outpatient services, and medication refills. Results Of 1,047 patients, 5.1% ( n = 53) had repeat SRS. Majority of re-SRS (74%, n = 39) were done within 2 years of index procedure and 51% were within 1-year of initial procedure. Patients who required re-SRS incurred higher hospital readmission rate, outpatient services, and had higher payments compared with those who did not require re-SRS at 6 months, 1 year, and 2 years following the initial procedures. Re-SRS received 3.0- and 3.1-times higher payments at 1 and 2 years compared with nr-SRS cohort. At 5 years following SRS, median combined payments for re-SRS cohort was $105286 (interquartile range [IQR] $70999, $156569) compared with $44172 (IQR $22956, $84840) for nr-SRS cohort. Conclusion More than half of the re-SRS procedures were noted within first year of initial SRS for VS. Overall payments at 5 years for repeat SRS was more than double that for nr-SRS. Repeat SRS was also associated with more re-admissions and outpatient services at annual follow-up time points.

The impact of comprehensive air purification on patient duration of stay, discharge outcomes, and health care economics: A retrospective cohort study

BackgroundInfectious airborne and surface pathogens constitute a substantial and poorly explored source of patient subclinical illness and infections. With that in mind, a system of advanced air purification technology was designed to destroy the DNA and RNA of all bacteria, fungi, and viruses. This study compares the effects of advanced air purification technology versus high efficiency particulate air filtration with respect to certain metrics of health care economics and resource utilization at a large, community-based, urban hospital. Our hypothesis was that the use of the advanced air purification technology would decrease health care durations of stay, lead to fewer nonhome discharges, and decrease hospital charges. MethodsAfter the installation of advanced air purification technology, 3 resultant air purification “zones” were established: zone C, a control floor with high efficiency particulate air filtration; zone B, a mixed high efficiency particulate air and advanced air purification technology floor; and zone A, a comprehensive advanced air purification technology remediation. This study included nonbariatric surgical patients admitted to any zone between December 2017 and December 2018, with reported case mix index on discharge. We analyzed hospital duration of stays, discharge destination, and hospital charges with adjustment for severity of illness using the case mix index. The likelihood of mortality, health care-associated infection, and readmission for each study zone was examined using logistic regression adjusting for case mix index, age, sex, and source of admission. ResultsThe study included 1,002 patients across the 3 zones, with mean age of 55.8 years (53.7% female), average case mix index of 1.98, and mortality of 1.7%. Compared with zone C, patients in zones A and B demonstrated decreased hospital stays, a greater percentage of home discharges (86.5–87.8% vs 64.7%), and less hospital charges. In addition, logistic regression modeling performed on 999 study patients showed that the likelihood of mortality, hospital-acquired infections, and readmissions did not differ among the 3 zones. A trend toward a lesser incidence of hospital-acquired infections was noted in zones A and B (0.40% and 0.48%, respectively) when compared with zone C (0.63%). ConclusionPatients in the advanced air purification technology zones demonstrated statistically significant improvements in durations of stay, discharge to home, and costs after adjusting for case mix index. In addition, a trend toward fewer hospital-acquired infections in advanced air purification technology zones was noted. These findings suggest that environmental factors may affect key clinical and economic outcomes, supporting further research in this important and largely unexplored area.

Geographic and Racial/Ethnic Variation in Glycemic Control and Treatment in a National Sample of Veterans With Diabetes.

Geographic and racial/ethnic disparities related to diabetes control and treatment have not previously been examined at the national level. A retrospective cohort study was conducted in a national cohort of 1,140,634 veterans with diabetes, defined as two or more diabetes ICD-9 codes (250.xx) across inpatient and outpatient records. Main exposures of interest included 125 Veterans Administration Medical Center (VAMC) catchment areas as well as racial/ethnic group. The main outcome measure was HbA1c level dichotomized at ≥8.0% (≥64 mmol/mol). After adjustment for age, sex, racial/ethnic group, service-connected disability, marital status, and the van Walraven Elixhauser comorbidity score, the prevalence of uncontrolled diabetes varied by VAMC catchment area, with values ranging from 19.1% to 29.2%. Moreover, these differences largely persisted after further adjusting for medication use and adherence as well as utilization and access metrics. Racial/ethnic differences in diabetes control were also noted. In our final models, compared with non-Hispanic Whites, non-Hispanic Blacks (odds ratio 1.11 [95% credible interval 1.09-1.14]) and Hispanics (1.36 [1.09-1.14]) had a higher odds of uncontrolled HBA1c level. In a national cohort of veterans with diabetes, we found geographic as well as racial/ethnic differences in diabetes control rates that were not explained by adjustment for demographics, comorbidity burden, use or type of diabetes medication, health care utilization, access metrics, or medication adherence. Moreover, disparities in suboptimal control appeared consistent across most, but not all, VAMC catchment areas, with non-Hispanic Black and Hispanic veterans having a higher odds of suboptimal diabetes control than non-Hispanic White veterans.

Costs and Utilization for Low Income Minority Patients with Depression in a Collaborative Care Model Implemented in a Community‐Based Academic Health System

Research ObjectiveThe collaborative care model (CoCM) is effective in treating patients with common behavioral health conditions in primary care settings and improving clinical outcomes. As health care reform efforts move toward model sustainability and scalability, more information is needed to assess its potential impact on costs and utilization. In a community‐based academic health system, serving primarily Medicare and Medicaid recipients with significant racial and ethnic diversity, we examined cost and major health care utilization metrics for depressed patients enrolled in CoCM compared to a group that did not receive CoCM treatment.Study DesignThe community‐based academic health system has 23 primary care practices in an urban setting offering colocated behavioral health care access to a psychiatrist and social worker as part of “usual care.” Seven of these sites implemented the CoCM to enhance care quality through care manager support, psychiatric consultation, measurement‐informed care, and a patient tracking registry. Two types of analyses were conducted to examine cost and utilization metrics; a site‐level intention to treat (ITT) model and an analysis comparing CoCM patients to a matched cohort at all usual care sites using propensity score matching. Two timeframes were considered; the first and second year following an initial positive PHQ‐9 (&gt;10 indicating clinically significant depressive symptoms).Population StudiedIndividuals were eligible for inclusion if they sought care at one of the 23 primary care sites, scored positive on the PHQ‐9, and were enrolled in Medicaid for at least nine months out of the year for both the baseline and analysis year. The outcomes of interest were cost and selected health care utilization metrics including emergency room, admissions, and office visits.Principal FindingsIn the first year, the CoCM patients (n = 4,315) demonstrated a statistically significant decrease in emergency room visits (OR = .95; OR = .91) and non‐BH specialty office visits (OR =. 92; OR = .93), and an increase in PCP visits (OR = 1.03; OR = 1.03) and behavioral health (BH) office visits (OR = 1.03; OR = 1.08) compared to the usual care patients (n = 3,061) for both the ITT and cohort match analyses (n = 1,873 in both groups). The trend in year two continued, with statistically significant decreased emergency room visits (OR = .89; OR = .88) and non‐BH specialty office visits (OR = .87; OR = .92) for CoCM patients (n = 2,623) compared to those in usual care (n = 1,838) for the ITT and cohort match analyses (n = 1,193). Total costs for both years were not significantly different between groups.ConclusionsFindings demonstrate CoCM treatment is associated with potentially positive outcomes, including decreased emergency room visits and improved continuity of primary care. Results indicate these outcomes are sustained for at least 24 months.Implications for Policy or PracticeWhen comparing CoCM treatment of depression with colocation care in a 2‐year analysis, more effective health care utilization was found for decreasing emergency room utilization and other medical specialty utilization, and increasing primary care continuity for patients receiving CoCM. As organizations continue to seek new opportunities to integrate behavioral health into primary care, consideration of health care utilization and costs will be helpful in determining model selection for integration.Primary Funding SourceCenters for Medicare and Medicaid Services.

Mobile Technology Care Coordination of Long-Term Services and Support: Cluster Randomized Clinical Trial.

The purpose of the study was to assess whether the effects of a mobile predictive intervention used by Service Coordinators (SCs) reduce hospital utilization in a Medicaid Long-Term Services and Supports (LTSS) population in Baltimore city during a 5-month intervention. SC participants (n = 11) were recruited to treatment or control groups. LTSS clients (n = 420) followed their SC randomization assignment. Utilization data were obtained from the Maryland Chesapeake Regional Information System for our Patients (CRISP) Health Information Exchange (HIE) system and linked to service coordination records. Study groups were similar in age, gender, race, and years receiving LTSS. SCs' satisfaction with use of the mobile tool was surveyed. SC perceptions were neutral (mean scores ranged from 2.3 to 3.3 on a 5-point scale). No significant differences between groups were observed for all utilization metrics. The mobile technology software system used in this study did not improve health care utilization for a LTSS population needing ongoing clinical and social services coordinated care.