Performance Characteristics of Sepsis Screening Tools During Antepartum and Postpartum Admissions.

OBJECTIVE: To evaluate the performance characteristics of existing screening tools for the prediction of sepsis during antepartum and postpartum readmissions. METHODS: This was a case-control study using electronic health record data obtained between 2016 and 2021 from 67 hospitals for antepartum sepsis admissions and 71 hospitals for postpartum readmissions up to 42 days. Patients in the sepsis case group were matched in a 1:4 ratio to a comparison cohort of patients without sepsis admitted antepartum or postpartum. The following screening criteria were evaluated: the CMQCC (California Maternal Quality Care Collaborative) initial sepsis screen, the non-pregnancy-adjusted SIRS (Systemic Inflammatory Response Syndrome), the MEWC (Maternal Early Warning Criteria), UKOSS (United Kingdom Obstetric Surveillance System) obstetric SIRS, and the MEWT (Maternal Early Warning Trigger Tool). Time periods were divided into early pregnancy (less than 20 weeks of gestation), more than 20 weeks of gestation, early postpartum (less than 3 days postpartum), and late postpartum through 42 days. False-positive screening rates, C-statistics, sensitivity, and specificity were reported for each overall screening tool and each individual criterion. RESULTS: We identified 525 patients with sepsis during an antepartum hospitalization and 728 patients with sepsis during a postpartum readmission. For early pregnancy and more than 3 days postpartum, non-pregnancy-adjusted SIRS had the highest C-statistics (0.78 and 0.83, respectively). For more than 20 weeks of gestation and less than 3 days postpartum, the pregnancy-adjusted sepsis screening tools (CMQCC and UKOSS) had the highest C-statistics (0.87-0.94). The MEWC maintained the highest sensitivity rates during all time periods (81.9-94.4%) but also had the highest false-positive rates (30.4-63.9%). The pregnancy-adjusted sepsis screening tools (CMQCC, UKOSS) had the lowest false-positive rates in all time periods (3.9-10.1%). All tools had the lowest C-statistics in the periods of less than 20 weeks of gestation and more than 3 days postpartum. CONCLUSION: For admissions early in pregnancy and more than 3 days postpartum, non-pregnancy-adjusted sepsis screening tools performed better than pregnancy-adjusted tools. From 20 weeks of gestation through up to 3 days postpartum, using a pregnancy-adjusted sepsis screening tool increased sensitivity and minimized false-positive rates. The overall false-positive rate remained high.

Performance Characteristics of Sepsis Screening Tools During Delivery Admissions.

OBJECTIVE: To evaluate the screening performance characteristics of existing tools for the diagnosis of sepsis during delivery admissions. METHODS: This was a case-control study using electronic health record data, including vital signs and laboratory results, for all delivery admissions of patients with sepsis from 59 nationally distributed hospitals. Patients with sepsis were matched by gestational age at delivery in a 1:4 ratio with patients without sepsis to create a comparison group. Patients with chorioamnionitis and sepsis were compared with a complete cohort of patients with chorioamnionitis without sepsis. Multiple screening criteria for sepsis were evaluated: the CMQCC (California Maternal Quality Care Collaborative), SIRS (Systemic Inflammatory Response Syndrome), the MEWC (the Maternal Early Warning Criteria), UKOSS (United Kingdom Obstetric Surveillance System), and the MEWT (Maternal Early Warning Trigger Tool). Sensitivity, false-positive rates, and C-statistics were reported for each screening tool. Analyses were stratified into cohort 1, which excluded patients with chorioamnionitis-endometritis, and cohort 2, which included those patients. RESULTS: Delivery admissions at 59 hospitals were extracted for patients with sepsis. Cohort 1 comprised 647 patients with sepsis, including 228 with end-organ injury, matched with a control group of 2,588 patients without sepsis. Cohort 2 comprised 14,591 patients with chorioamnionitis-endometritis, of whom 1,049 had sepsis and 238 had end-organ injury. In cohort 1, the CMQCC and the UKOSS pregnancy-adjusted criteria had the lowest false-positive rates (6.9% and 9.6%, respectively) and the highest C-statistics (0.92 and 0.91, respectively). Although other screening criteria, such as SIRS and the MEWC, had similar sensitivities, it was at the cost of much higher false-positive rates (21.3% and 38.3%, respectively). In cohort 2, including all patients with chorioamnionitis-endometritis, the highest C-statistics were again for the CMQCC (0.67) and UKOSS (0.64). All screening tools had high false-positive rates, but the false-positive rates for the CMQCC and UKOSS were substantially lower than those for SIRS and the MEWC. CONCLUSION: During delivery admissions, the CMQCC and UKOSS pregnancy-adjusted screening criteria have the lowest false-positive results while maintaining greater than 90% sensitivity rates. Performance of all screening tools was degraded in the setting of chorioamnionitis-endometritis.

Examining paradoxical session attendance and weight loss relationships in a clinic based lifestyle modification intervention.

Objective: Evaluations of lifestyle modification interventions (LMIs), modeled after the Diabetes Prevention Program, have repeatedly shown a dose-response relationship between session attendance and weight loss. Despite this, not all participants had "average" weight loss experiences. Nearly one-third of LMI participants experienced unexpected, paradoxical outcomes (i.e., high attendance with little weight loss, and low attendance with clinically significant weight loss). Paradoxical weight-loss outcomes were characterized based on session attendance among participants in a group-based LMI in a real-world healthcare setting. This group-based LMI was delivered over 1 year to participants with the possibility of attending up to 25 sessions total. Methods: LMI participants identified in 2010-2017 from electronic health records were characterized as having low (<75%) or high (≥75%) session attendance. Weight-loss outcomes were defined as expected (≥5%, high-attendance; <5%, low-attendance) or paradoxical (≥5%, low-attendance; <5%, high-attendance). Paradoxical-outcome-associated characteristics were identified using logistic regression. Results: Among 1813 LMI participants, 1498 (82.6%) had low and 315 (17.4%) high session attendance; 555 (30.6%) had paradoxical outcomes, comprising 415 (74.8%) responders (≥5% weight-loss) and 140 (25.2%) non-responders (<5% weight-loss). Among participants with high session attendance, paradoxical non-responders were more likely to be female (odds ratio [OR]: 2.76; 95% confidence interval [CI]: 1.32, 5.77) and have type 2 diabetes (OR: 3.32; 95% CI: 1.01, 10.95). Among low-attendance participants, paradoxical responders were more likely to be non-Hispanic White and less likely to be non-Hispanic Black (OR: 0.35; 95% CI: 0.18, 0.69), non-Hispanic Asian (OR: 0.40; 95% CI: 0.22, 0.73), or Hispanic (OR: 0.53; 95% CI: 0.35, 0.80). Conclusions: In a healthcare setting, nearly one-third of LMI participants experienced paradoxical outcomes. More research is needed to understand the facilitators and barriers to weight loss above and beyond session attendance.

Changes in Healthcare Utilization After Lifestyle Intervention for Weight Loss.

INTRODUCTION: This study evaluates the real-world impact of a lifestyle change program (LCP) on healthcare utilization in a large health system. METHODS: Using electronic health record data from a large health system in northern California, U.S., LCP participant and propensity-score-matched nonparticipant outcomes were compared in the second year postparticipation: (1) overall healthcare utilization and (2) utilization and medications related to cardiometabolic conditions and obesity. Adult LCP participants between 2010 and 2017 were identified and matched 1:1 with replacement to comparable nonparticipants. Participants without electronic health record activity in the 12-36 months before baseline, or with conditions or procedures associated with substantial weight change, were excluded. Statistical analysis and modeling were performed in 2021-22. RESULTS: Compared to matched nonparticipants, LCP participants in the 12-24 months postbaseline were more likely to have specialty-care visits (+4.7%, 95% CI +1.8%, +7.6%), electronic communications (8.6%, 95% CI +5.6%, +11.7%), and urgent-care visits (+6.5%, 95% CI +3.0%, 10.0%). Participants also had more office visits for cardiometabolic conditions and obesity (+1.72 visits/patient, 95% CI +1.05, +2.39). CONCLUSIONS: Compared with matched nonparticipants, LCP participation was associated with higher utilization of outpatient services postparticipation. Additional research could assess whether this indicates an increase in preventive care that could lead to improved future outcomes.

Racial Disparities in Pulse Oximeter Device Inaccuracy and Estimated Clinical Impact on COVID-19 Treatment Course.

Arterial blood oxygen saturation as measured by pulse oximetry (peripheral oxygen saturation (SpO2)) may be differentially less accurate for people with darker skin pigmentation, which could potentially affect the course of coronavirus disease 2019 (COVID-19) treatment. We analyzed pulse oximeter accuracy and its association with COVID-19 treatment outcomes using electronic health record data from Sutter Health, a large, mixed-payer, integrated health-care delivery system in Northern California. We analyzed 2 cohorts: 1) 43,753 non-Hispanic White (NHW) or non-Hispanic Black/African-American (NHB) adults with concurrent arterial blood gas oxygen saturation/SpO2 measurements taken between January 2020 and February 2021; and 2) 8,735 adults who went to a hospital emergency department with COVID-19 between July 2020 and February 2021. Pulse oximetry systematically overestimated blood oxygenation by 1% more in NHB individuals than in NHW individuals. For people with COVID-19, this was associated with lower admission probability (-3.1 percentage points), dexamethasone treatment (-3.1 percentage points), and supplemental oxygen treatment (-4.5 percentage points), as well as increased time to treatment: 37.2 minutes before dexamethasone initiation and 278.5 minutes before initiation of supplemental oxygen. These results call for additional investigation of pulse oximeters and suggest that current guidelines for development, testing, and calibration of these devices should be revisited, investigated, and revised.

A population-health approach to characterizing migraine by comorbidity: Results from the Mindfulness and Migraine Cohort Study.

BACKGROUND: The heterogeneity of migraine has been reported extensively, with identified subgroups usually based on symptoms. Grouping individuals with migraine and similar comorbidity profiles has been suggested, however such segmentation methods have not been tested using real-world clinical data. OBJECTIVE: To gain insights into natural groupings of patients with migraine using latent class analysis based on electronic health record-determined comorbidities. METHODS: Retrospective electronic health record data analysis of primary-care patients at Sutter Health, a large open healthcare system in Northern California, USA. We identified migraine patients over a five-year time period (2015-2019) and extracted 29 comorbidities. We then applied latent class analysis to identify comorbidity-based natural subgroups. RESULTS: We identified 95,563 patients with migraine and found seven latent classes, summarized by their predominant comorbidities and population share: fewest comorbidities (61.8%), psychiatric (18.3%), some comorbidities (10.0%), most comorbidities - no cardiovascular (3.6%), vascular (3.1%), autoimmune/joint/pain (2.2%), and most comorbidities (1.0%). We found minimal demographic differences across classes. CONCLUSION: Our study found groupings of migraine patients based on comorbidity that have the potential to be used to guide targeted treatment strategies and the development of new therapies.

Hyperlocalized Measures of Air Pollution and Preeclampsia in Oakland, California.

Exposure to nitrogen dioxide (NO2), black carbon (BC), and ultrafine particles (UFPs) during pregnancy may increase the risk of preeclampsia, but previous studies have not assessed hyperlocalized differences in pollutant levels, which may cause exposure misclassification. We used data from Google Street View cars with mobile air monitors that repeatedly sampled NO2, BC, and UFPs every 30 m in Downtown and West Oakland neighborhoods during 2015-2017. Data were linked to electronic health records of pregnant women in the 2014-2016 Sutter Health population, who resided within 120 m of monitoring data (N = 1095), to identify preeclampsia cases. We used G-computation with log-binomial regression to estimate risk differences (RDs) associated with a hypothetical intervention reducing pollutant levels to the 25th percentile observed in our sample on preeclampsia risk, overall and stratified by race/ethnicity. Prevalence of preeclampsia was 6.8%. Median (interquartile range) levels of NO2, BC, and UFPs were 10.8 ppb (9.0, 13.0), 0.34 µg/m3 (0.27, 0.42), and 29.2 # × 103/cm3 (26.6, 32.6), respectively. Changes in the risk of preeclampsia achievable by limiting each pollutant to the 25th percentile were NO2 RD = -1.5 per 100 women (95% confidence interval (CI): -2.5, -0.5); BC RD = -1.0 (95% CI: -2.2, 0.02); and UFP RD = -0.5 (95% CI: -1.8, 0.7). Estimated effects were the largest for non-Latina Black mothers: NO2 RD = -2.8 (95% CI: -5.2, -0.3) and BC RD = -3.0 (95% CI: -6.4, 0.4).

Mind the clinical-analytic gap: Electronic health records and COVID-19 pandemic response.

Data quality is essential to the success of the most simple and the most complex analysis. In the context of the COVID-19 pandemic, large-scale data sharing across the US and around the world has played an important role in public health responses to the pandemic and has been crucial to understanding and predicting its likely course. In California, hospitals have been required to report a large volume of daily data related to COVID-19. In order to meet this need, electronic health records (EHRs) have played an important role, but the challenges of reporting high-quality data in real-time from EHR data sources have not been explored. We describe some of the challenges of utilizing EHR data for this purpose from the perspective of a large, integrated, mixed-payer health system in northern California, US. We emphasize some of the inadequacies inherent to EHR data using several specific examples, and explore the clinical-analytic gap that forms the basis for some of these inadequacies. We highlight the need for data and analytics to be incorporated into the early stages of clinical crisis planning in order to utilize EHR data to full advantage. We further propose that lessons learned from the COVID-19 pandemic can result in the formation of collaborative teams joining clinical operations, informatics, data analytics, and research, ultimately resulting in improved data quality to support effective crisis response.

Recruitment, retention, and adherence in a randomized feasibility trial of mindfulness-based stress reduction for patients with migraine.

OBJECTIVES: Increasing evidence demonstrates effectiveness of Mindfulness-Based Stress Reduction (MBSR) for pain-related and functional disorders. In order to conduct successful and efficient trials of MBSR, evidence regarding the relative performance of strategies to improve recruitment, retention, and adherence is required, but few studies have examined these issues specifically. DESIGN: In preparation for a fully powered trial, we conducted a 2-arm, parallel comparison randomized controlled feasibility trial of MBSR vs. usual-care for 60 patients with migraine headache. SETTING: Two large U.S. health systems in Northern California. INTERVENTION: MBSR is an 8-week classroom-based intervention that combines mindfulness meditation and yoga, with didactic presentations about stress psychology and group process/experiential education. Participants received the intervention at their choice of one of several existing, vetted community-based classes. MAIN OUTCOME MEASURES: Successful recruitment was defined a priori as 18 participants within any 9-week period or 60 participants enrolled within a 36-week period. We considered participants adherent to the intervention if they attended at least 5 of the 8 weekly classes and the day-long retreat. RESULTS: We successfully enrolled 18 participants within a 7-week period, however, we did not attain our second goal of recruiting 60 participants within a 36-week period. Sixty-eight percent of our participants were adherent to the intervention. CONCLUSIONS: We found that close monitoring of recruitment activities, flexibility in protocol modifications, and integration within the delivery system were crucial factors for successful participant recruitment, retention, and adherence in mindfulness research.

Effectiveness of a Group-Based Lifestyle Change Program Versus Usual Care: An Electronic Health Record, Propensity Score-Matched Cohort Study.

INTRODUCTION: Translational lifestyle change programs for community and clinical settings have been available for a decade, yet there are limited data on their comparative effectiveness. This study examines the effectiveness of a Centers for Disease Control and Prevention-aligned lifestyle change program relative to usual care in clinical practice. METHODS: This was an electronic health record-based retrospective cohort study conducted in a community-based healthcare system. Investigators identified adult program participants and usual-care patients in the electronic health record between 2010 and 2018 and defined their index date (baseline) as the first lifestyle change program encounter or a random encounter date, respectively. Participants were matched 1:2 to usual-care patients on baseline demographics and clinical characteristics using propensity-score methods. Changes in body weight and blood pressure were examined from baseline through 24 months. RESULTS: The authors identified 2,833 program participants and 438,432 usual-care patients meeting study eligibility criteria. A total of 2,833 program participants were matched to 4,776 usual-care patients; the average age was 54 years, and 80% of the participants were female. Program participation was associated with a 1.9- and 1.6-fold higher prevalence of clinically meaningful (≥5%) weight loss at 12- and 24-month follow-up than usual care and a higher prevalence of blood pressure control at 12 months but not at 24 months. Patients without type 2 diabetes at baseline had more pronounced outcomes than those with type 2 diabetes. CONCLUSIONS: This study demonstrates the effectiveness of an evidence-based, Centers for Disease Control and Prevention-aligned lifestyle change program in reducing cardiometabolic risk factors compared with usual care in clinical practice, with long-term reductions in weight and transient reductions in blood pressure.

Short-term weight trajectories and long-term weight outcomes from a lifestyle intervention in real-world clinical practice.

Centers for Disease Control and Prevention aligned lifestyle change programs are effective in promoting weight loss among those with elevated cardiometabolic risk; yet, variability in weight outcomes among participants is high. Little is known about heterogeneity of short-term weight changes among participants in real-world clinical practice. We sought to identify short-term weight trajectory clusters among lifestyle change program participants in real-world clinical practice and to examine the relationship between cluster membership and long-term weight outcomes. We identified participants from the electronic health records (2010-2017) with weight measured ≤30 days prior to program initiation (baseline) and in four intervals (3-week segments) in the 12 weeks after baseline. Clustering analysis was performed to identify distinct trajectories in percent weight change over 12 weeks. Cluster-specific differences in weight change at 12 and 52 weeks were assessed. Among 1,148 participants, across 18 clinic sites, three clusters were identified: minimal-to-no weight loss (MWL), delayed-minimal weight loss (DWL), and steady-moderate weight loss (SWL), corresponding to mean weight changes of 0.4%, -2.3%, and -4.8% at 12 weeks follow-up, respectively. Mean weight changes were 0.4%, -1.8%, and -5.1% for MWL, DWL, and SWL clusters, respectively, at 52 weeks follow-up, which correlated in direction and magnitude with short-term weight changes. Clustering analysis reveals heterogeneous, short-term weight trajectories among lifestyle change program participants in real-world clinical practice. Given the relationship between the magnitudes of short- and long-term weight change, individual participant weight trajectories may be useful in identifying potential non-responders in need of adjunctive or alternative therapy.

Early Weight Loss and Treatment Response: Data From a Lifestyle Change Program in Clinical Practice.

INTRODUCTION: The purpose of this study was to develop and validate a predictive model for the early identification of nonresponders to a 12-month lifestyle change program in clinical practice. METHODS: Investigators identified lifestyle change program participants in the electronic health records of a large healthcare delivery system between 2010 and 2017. Nonresponse was defined as weight gain or no weight loss at 12 months from the program initiation (baseline). Logistic regression with percentage weight change at 2-12 weeks from baseline was used as an independent predictor of nonresponse. Baseline demographics and clinical characteristics were also tested as potential predictors. The authors performed ten-fold cross-validation for model assessment and examined model performance with the area under the receiver operating characteristic curve, sensitivity, specificity, and positive and negative predictive values. The analyses were conducted in 2019. RESULTS: Among 947 program participants, 30% were classified as nonresponders at 12 months. The model with the best discrimination of responders from nonresponders included weight change at 12 weeks from baseline as the sole predictor (area under the receiver operating characteristic curve, 0.789). Sensitivity and positive predictive value were maximized at 0.56 (specificity and negative predictive value, 0.81 each). CONCLUSIONS: In a cohort of lifestyle change program participants from clinical practice, percentage weight change at 12 weeks from baseline can serve as a single indicator of nonresponse at the completion of the 12-month program. Clinicians can easily apply this algorithm to identify and assess participants in potential need of adjunctive or alternative therapy to maximize treatment outcomes.

Longitudinal Weight Outcomes From a Behavioral Lifestyle Intervention in Clinical Practice.

PURPOSE: The purpose of this electronic health record (EHR)-based retrospective cohort study was to characterize a population of patients participating in a 12-month, lifestyle change program in a community-based health system and to examine longitudinal weight outcomes. METHODS: Program participants were identified in the EHRs of a health care delivery system across 18 sites between 2010 and 2017. Outcomes were mean weight change and proportion of patients with ≥5% weight loss through 24 months from program initiation. RESULTS: Among 4463 program participants, 3156 met study eligibility criteria, with a mean ± SD age of 53.5 ± 13.1 years; 77.7% were women. Mean baseline weight ± SD was 101.3 ± 23.8 kg. Three main cardiometabolic risk groups were identified: prediabetes/high risk for diabetes (47.3%), overweight/obese in the absence of elevated diabetes risk (27.2%), and existing diabetes (23.9%). Maximal mean weight loss was 3.9% at 6 months from baseline. At 12 and 24 months from baseline, mean weight loss was 3.2% and 2.3%, respectively, with 31% and 29% of participants attaining ≥5% weight loss. Long-term weight outcomes were similar across risk groups. CONCLUSIONS: A lifestyle change program in a clinical practice setting is associated with modest weight loss, sustained through 24 months, among participants with a range of cardiometabolic risk factors. More than one-quarter of participants achieve ≥5% weight loss, regardless of cardiometabolic risk.

Conducting a pilot randomized controlled trial of community-based mindfulness-based stress reduction versus usual care for moderate-to-severe migraine: protocol for the Mindfulness and Migraine Study (M&M).

BACKGROUND: Migraine is one of the most common neurological disorders in clinical practice and is a substantial cause of disability worldwide. Current approaches to therapy are primarily based on medication but are often limited by inadequate effectiveness and common side effects. Newer, more effective medications are expensive. Mindfulness-based stress reduction (MBSR), an 8-week classroom-based meditation intervention, is inexpensive, has no known side effects, and has demonstrated clinically meaningful effectiveness for several chronic-pain syndromes. In addition, MBSR has shown promising results for migraine therapy in a few small case studies and pilot studies. We present here the protocol for a two-arm randomized controlled pilot trial of MBSR for moderate-to-severe episodic migraine, which, if successful, will form the basis for a fully powered clinical trial. METHODS/DESIGN: This study, set in Northern California, is a two-arm parallel-comparison single-blinded randomized controlled pilot trial with the goal of recruiting approximately 60 participants with moderate-to-severe episodic migraine. The feasibility outcomes include ability and time required to recruit, adherence to the MBSR treatment, and ability to measure outcomes using 31-day headache diaries and patient-reported questionnaire data. The active treatment arm consists of an 8-week community-based MBSR class plus usual care, and the wait-list control group is usual care. Recruitment is underway and expected to be complete by the end of 2018. DISCUSSION: To our knowledge, this is the first pragmatic trial in the U.S. of MBSR for migraine using community-based classes, and if it proves viable, we plan to conduct a fully powered trial to determine the effectiveness of the intervention for reducing headache days for moderate-to-severe episodic migraineurs. TRIAL REGISTRATION: Clinicaltrials.gov, NCT02824250 . Registered on 6 July 2016.

Predicting need for advanced illness or palliative care in a primary care population using electronic health record data.

Timely outreach to individuals in an advanced stage of illness offers opportunities to exercise decision control over health care. Predictive models built using Electronic health record (EHR) data are being explored as a way to anticipate such need with enough lead time for patient engagement. Prior studies have focused on hospitalized patients, who typically have more data available for predicting care needs. It is unclear if prediction driven outreach is feasible in the primary care setting. In this study, we apply predictive modeling to the primary care population of a large, regional health system and systematically examine the impact of technical choices, such as requiring a minimum number of health care encounters (data density requirements) and aggregating diagnosis codes using Clinical Classifications Software (CCS) groupings to reduce dimensionality, on model performance in terms of discrimination and positive predictive value. We assembled a cohort of 349,667 primary care patients between 65 and 90 years of age who sought care from Sutter Health between July 1, 2011 and June 30, 2014, of whom 2.1% died during the study period. EHR data comprising demographics, encounters, orders, and diagnoses for each patient from a 12 month observation window prior to the point when a prediction is made were extracted. L1 regularized logistic regression and gradient boosted tree models were fit to training data and tuned by cross validation. Model performance in predicting one year mortality was assessed using held-out test patients. Our experiments systematically varied three factors: model type, diagnosis coding, and data density requirements. We found substantial, consistent benefit from using gradient boosting vs logistic regression (mean AUROC over all other technical choices of 84.8% vs 80.7% respectively). There was no benefit from aggregation of ICD codes into CCS code groups (mean AUROC over all other technical choices of 82.9% vs 82.6% respectively). Likewise increasing data density requirements did not affect discrimination (mean AUROC over other technical choices ranged from 82.5% to 83%). We also examine model performance as a function of lead time, which is the interval between death and when a prediction was made. In subgroup analysis by lead time, mean AUROC over all other choices ranged from 87.9% for patients who died within 0 to 3 months to 83.6% for those who died 9 to 12 months after prediction time.

Correction: Predicting all-cause risk of 30-day hospital readmission using artificial neural networks.

[This corrects the article DOI: 10.1371/journal.pone.0181173.].

Impact of home-based, patient-centered support for people with advanced illness in an open health system: A retrospective claims analysis of health expenditures, utilization, and quality of care at end of life.

BACKGROUND: Home-based care coordination and support programs for people with advanced illness work alongside usual care to promote personal care goals, which usually include a preference for home-based end-of-life care. More research is needed to confirm the efficacy of these programs, especially when disseminated on a large scale. Advanced Illness Management is one such program, implemented within a large open health system in northern California, USA. AIM: To evaluate the impact of Advanced Illness Management on end-of-life resource utilization, cost of care, and care quality, as indicators of program success in supporting patient care goals. DESIGN: A retrospective-matched observational study analyzing medical claims in the final 3 months of life. SETTING/PARTICIPANTS: Medicare fee-for-service 2010-2014 decedents in northern California, USA. RESULTS: Final month total expenditures for Advanced Illness Management enrollees ( N = 1352) were reduced by US$4824 (US$3379, US$6268) and inpatient payments by US$6127 (US$4874, US$7682). Enrollees also experienced 150 fewer hospitalizations/1000 (101, 198) and 1361 fewer hospital days/1000 (998, 1725). The percentage of hospice enrollees increased by 17.9 percentage points (14.7, 21.0), hospital deaths decreased by 8.2 percentage points (5.5, 10.8), and intensive care unit deaths decreased by 7.1 percentage points (5.2, 8.9). End-of-life chemotherapy use and non-inpatient expenditures in months 2 and 3 prior to death did not differ significantly from the control group. CONCLUSION: Advanced Illness Management has a positive impact on inpatient utilization, cost of care, hospice enrollment, and site of death. This suggests that home-based support programs for people with advanced illness can be successful on a large scale in supporting personal end-of-life care choices.

Predicting all-cause risk of 30-day hospital readmission using artificial neural networks.

Avoidable hospital readmissions not only contribute to the high costs of healthcare in the US, but also have an impact on the quality of care for patients. Large scale adoption of Electronic Health Records (EHR) has created the opportunity to proactively identify patients with high risk of hospital readmission, and apply effective interventions to mitigate that risk. To that end, in the past, numerous machine-learning models have been employed to predict the risk of 30-day hospital readmission. However, the need for an accurate and real-time predictive model, suitable for hospital setting applications still exists. Here, using data from more than 300,000 hospital stays in California from Sutter Health's EHR system, we built and tested an artificial neural network (NN) model based on Google's TensorFlow library. Through comparison with other traditional and non-traditional models, we demonstrated that neural networks are great candidates to capture the complexity and interdependency of various data fields in EHRs. LACE, the current industry standard, showed a precision (PPV) of 0.20 in identifying high-risk patients in our database. In contrast, our NN model yielded a PPV of 0.24, which is a 20% improvement over LACE. Additionally, we discussed the predictive power of Social Determinants of Health (SDoH) data, and presented a simple cost analysis to assist hospitalists in implementing helpful and cost-effective post-discharge interventions.