Search

Elective thoracic surgery is safe in well-selected elderly patients. The association of frailty with postoperative morbidity in elective-lobectomy patients is understudied. We examined frailty as defined by abbreviated modified frailty index (mFI-5), mFI-11 in the thoracic surgery population, and the correlation between frailty and postoperative complications.We studied outcomes of patients in two cohorts, 2010-2012 and 2013-2019, from the National Surgical Quality Improvement Program (NSQIP) database and used multivariable logistic regression models to predict all postoperative morbidity, mortality, and major morbidity. The mFI-5 could be calculated for all subjects (both 2010-2012, and 2013-2019); the mFI-11 could only be calculated for the 2010-2012 cohort. Patient frailty was defined as mFI≥3 (with either index). We used odds ratios (ORs) to examine associations of preoperative characteristics with postoperative complications and C-statistics to assess overall predictive power.Complications were less prevalent in the 2013-2019 cohort (17.9% versus 19.5%, P = 0.008). Open lobectomies were more common in the 2010-2012 cohort (53.9% versus 34.6%) and were strongly associated with postoperative morbidity and mortality (ORs1.5) in both cohorts. Each frailty measure was associated with morbidity and mortality (ORs1.4) after adjusting for other significant preoperative factors. Models on the 2010-2012 cohort had nearly identical C-statistics using the mFI-11 versus mFI-5 frailty indices (0.6142 versus 0.6139; P 0.8).Frailty, as captured in the mFI-5, is a significant associated factor of postoperative morbidity and mortality following elective lobectomies. As a modifiable risk factor, frailty should be considered in surgical decision-making and when counseling patients regarding perioperative risks.

About 7 million people, 2.8% of US adults, have bipolar disorder (BD). While second-generation antipsychotics (SGA) are indicated as acute and maintenance treatments for BD, therapeutic success requires medication adherence and reported nonadherence estimates to range as high as 60%. Identifying patient risk factors for nonadherence is important for reducing it.The objective of this study was to quantify the associations of risk factors, including social determinants of health, with SGA nonadherence among patients with BD.In this cross-sectional study of 2015-2017 MassHealth Medicaid data, we examined several definitions of adherence and used logistic regression to identify risk factors for nonadherence (medication possession ratio0.8) among all adults aged 18-64 diagnosed with BD who could be followed for 12 months following SGA initiation.Among 5197 patients, the mean (±SD) age was 37.7 (±11.4) years, and 42.3% were men. Almost half (47.7%) of patients were nonadherent to SGAs when measured by medication possession ratio. The prevalence of nonadherence peaked at middle age for men and younger for women. Nonadherence was less common among Massachusetts' Department of Mental Health clients (odds ratio=0.60, 95% confidence limit: 0.48-0.74) and among those who used other psychotropic medications (odds ratios between 0.45 and 0.81); in contrast, increase in neighborhood socioeconomic stress was associated with increased odds of nonadherence.Adherence to SGA treatment is suboptimal among people with BD. Recognizing risk factors, including those related to social determinants of health, can help target interventions to improve adherence for people at high risk and has implications for adherence-based quality measures.

Background Precision prevention is increasingly important in HIV prevention research to move beyond universal interventions to those tailored for high-risk individuals. The current study was designed to develop machine learning algorithms for predicting adolescent HIV risk behaviours. Methods Comprehensive longitudinal data on adolescent risk behaviours, perceptions, peer and family influence, and neighbourhood risk factors were collected from 2564 grade-10 students at baseline followed for 24 months over 2008-2012. Machine learning techniques [support vector machine (SVM) and random forests] were applied to innovatively leverage longitudinal data for robust HIV risk behaviour prediction. In this study, we focused on two adolescent risk behaviours: had ever had sex and had multiple sex partners. Twenty percent of the data were withheld for model testing. Results The SVM model with cost-sensitive learning achieved the highest sensitivity, at 79.1%, specificity of 75.4% with AUC of 0.86 in predicting multiple sex partners on the training data (10-fold cross-validation), and sensitivity of 79.7%, specificity of 76.5% with AUC of 0.86 on the testing data. The random forest model obtained the best performance in predicting had ever had sex, yielding the sensitivity of 78.5%, specificity of 73.1% with AUC of 0.84 on the training data and sensitivity of 82.7%, specificity of 75.3% with AUC of 0.87 on the testing data. Conclusion Machine learning methods can be used to build effective prediction model(s) to identify adolescents who are likely to engage in HIV risk behaviours. This study builds a foundation for targeted intervention strategies and informs precision prevention efforts in school-setting.

BACKGROUND Secure messaging use is associated with improved diabetes-related outcomes. However, it is less clear how secure messaging supports diabetes management. OBJECTIVE We examined secure message topics between patients and clinical team members in a national sample of veterans with type 2 diabetes to understand use of secure messaging for diabetes management and potential associations with glycemic control. METHODS We surveyed and analyzed the content of secure messages between 448 US Veterans Health Administration patients with type 2 diabetes and their clinical teams. We also explored the relationship between secure messaging content and glycemic control. RESULTS Explicit diabetes-related content was the most frequent topic (72.1% of participants), followed by blood pressure (31.7% of participants). Among diabetes-related conversations, 90.7% of patients discussed medication renewals or refills. More patients with good glycemic control engaged in 1 or more threads about blood pressure compared to those with poor control (37.5% vs 27.2%, P=.02). More patients with good glycemic control engaged in 1 more threads intended to share information with their clinical team about an aspect of their diabetes management compared to those with poor control (23.7% vs 12.4%, P=.009). CONCLUSIONS There were few differences in secure messaging topics between patients in good versus poor glycemic control. Those in good control were more likely to engage in informational messages to their team and send messages related to blood pressure. It may be that the specific topic content of the secure messages may not be that important for glycemic control. Simply making it easier for patients to communicate with their clinical teams may be the driving influence between associations previously reported in the literature between secure messaging and positive clinical outcomes in diabetes.

Background: Causal mediation analysis is conducted in biomedical research with the goal of investigating causal mechanisms that consist of both direct causal pathways between the treatment and outcome variables and intermediate causal pathways through mediators. Recently, this type of analysis has been applied in the context of bioinformatics; however, it encounters the obstacle of high-dimensional and semi-continuous mediators with clumping at zero. Methods: In this article, we develop a methodology to conduct high-dimensional causal mediation analysis with a modeling framework that involves (i) a nonlinear model for the outcome variable, (ii) two-part models for semi-continuous mediators with clumping at zero, and (iii) sophisticated variable-selection techniques using machine learning. We conducted simulations and investigated the performance of the proposed method. It is shown that the proposed method can provide reliable statistical information on the causal effects with high-dimensional mediators. The method is adopted to assess the contribution of the intestinal microbiome to the risk of bacterial pathogen colonization in older adults from US nursing homes. Conclusions: The proposed high-dimensional causal mediation analysis with nonlinear models is an innovative and reliable approach to conduct causal inference with high-dimensional mediators.

Objectives The Massachusetts Medicaid and Children's Health Insurance Program, MassHealth, offers comprehensive Senior Care Options (SCO) plans to its Medicare-eligible members 65 years and older. Historically, MassHealth has paid a fixed per-person capitation rate for any "nursing home-certifiable" SCO member despite considerable heterogeneity of need. Our objective was to develop a model to predict long-term services and supports (LTSS) costs for community-dwelling SCO members. Study design Concurrent predictive modeling. Methods We studied nursing home-certifiable SCO members who were enrolled for at least 183 days during 2016-2017 and used linear models to predict annual cost of community-based LTSS from demographic, medical, social determinants of health, and functional characteristics. We evaluated model performance using predictive performance (R2) and predictive ratios (observed costs divided by predicted costs) for various vulnerable subgroups. Results The modeling population included 35,259 enrollees. Mean (SD) annualized LTSS cost was $14,071 ($13,174). Functional status (ie, activities of daily living [ADLs] and instrumental ADLs) accounted for most of the variability in community LTSS cost (R2 = 18.4%) explainable by available variables. The Massachusetts SCO (MA-SCO) model (R2 = 21.6%) predicts accurately for several high-cost, vulnerable subgroups. Compared with fixed per-member capitation payments for all, the MA-SCO model reduces, for example, the payment to one plan by 28% and increases that to another by 35%. Conclusions Predictive models using administrative data and functional status information can appropriately allocate payments for subgroups of members with LTSS needs that differ substantially from average. Calibrating payment to need mitigates incentives for population skimming and promotes the sustainability of mission-oriented organizations.

BACKGROUND: Diabetes is a complex, chronic disease that requires patients’ effective self-management between clinical visits; this in turn relies on patient self-efficacy. The support of patient autonomy from healthcare providers is associated with better self-management and greater diabetes self-efficacy. Effective provider-patient secure messaging (SM) through patient portals may improve disease self-management and self-efficacy. SM that supports patients’ sense of autonomy may mediate this effect by providing patients ready access to their health information and better communication with their clinical teams. OBJECTIVE: We examined the association between healthcare team–initiated SM and diabetes self-management and self-efficacy, and whether this association was mediated by patients’ perceptions of autonomy support from their healthcare teams. DESIGN: We surveyed and analyzed content of messages sent to a sample of patients living with diabetes who use the SM feature on the VA’s My HealtheVet patient portal. PARTICIPANTS: Four hundred forty-six veterans with type 2 diabetes who were sustained users of SM. MAIN MEASURES: Proactive (healthcare team-initiated) SM (0 or ≥ 1 messages); perceived autonomy support; diabetes self-management; diabetes self-efficacy. KEY RESULTS: Patients who received at least one proactive SM from their clinical team were significantly more likely to engage in better diabetes self-management and report a higher sense of diabetes self-efficacy. This relationship was mediated by the patient’s perception of autonomy support. The majority of proactive SM discussed scheduling, referrals, or other administrative content. Patients’ responses to team-initiated communication promoted patient engagement in diabetes self-management behaviors. CONCLUSIONS: Perceived autonomy support is important for diabetes self-management and self-efficacy. Proactive communication from clinical teams to patients can help to foster a patient’s sense of autonomy and encourage better diabetes self-management and self-efficacy.

Background Rural patients with diabetes have difficulty accessing care and are at higher risk for poor diabetes management. Sustained use of patient portal features such as secure messaging (SM) can provide accessible support for diabetes self-management. Objective This study explored whether rural patients’ self-management and glycemic control was associated with the use of SM. Methods This secondary, cross-sectional, mixed methods analysis of 448 veterans with diabetes used stratified random sampling to recruit a diverse sample from the United States (rural vs urban and good vs poor glycemic control). Administrative, clinical, survey, and interview data were used to determine patients’ rurality, use of SM, diabetes self-management behaviors, and glycemic control. Moderated mediation analyses assessed these relationships. Results The sample was 51% (n=229) rural and 49% (n=219) urban. Mean participant age was 66.4 years (SD 7.7 years). More frequent SM use was associated with better diabetes self-management (P=.007), which was associated with better glycemic control (P Conclusions More frequent SM use is associated with engaging in diabetes self-management, which, in turn, is associated with better diabetes control. Among rural patients with diabetes, SM use is indirectly associated with better diabetes control. Frequent patient-team communication through SM about diabetes-related content may help rural patients with diabetes self-management, resulting in better glycemic control.

Objective The study sought to investigate whether consistent use of the Veterans Health Administration’s My HealtheVet (MHV) online patient portal is associated with improvement in diabetes-related physiological measures among new portal users. Materials and Methods We conducted a retrospective cohort study of new portal users with type 2 diabetes that registered for MHV between 2012 and 2016. We used random-effect linear regression models to examine associations between months of portal use in a year (consistency) and annual means of the physiological measures (hemoglobin A1c [HbA1c], low-density lipoproteins [LDLs], and blood pressure [BP]) in the first 3 years of portal use. Results For patients with uncontrolled HbA1c, LDL, or BP at baseline, more months of portal use in a year was associated with greater improvement. Compared with 1 month of use, using the portal 12 months in a year was associated with annual declines in HbA1c of -0.41% (95% confidence interval [CI], -0.46% to -0.36%) and in LDL of -6.25 (95% CI, -7.15 to -5.36) mg/dL. Twelve months of portal use was associated with minimal improvements in BP: systolic BP of -1.01 (95% CI, -1.33 to -0.68) mm Hg and diastolic BP of -0.67 (95% CI, -0.85 to -0.49) mm Hg. All associations were smaller or not present for patients in control of these measures at baseline. Conclusions We found consistent use of the patient portal among new portal users to be associated with modest improvements in mean HbA1c and LDL for patients at increased risk at baseline. For patients with type 2 diabetes, self-management supported by online patient portals may help control HbA1c, LDL, and BP.

BACKGROUND Rural patients with diabetes have difficulty accessing care and are at higher risk for poor diabetes management. Sustained use of patient portal features such as secure messaging (SM) can provide accessible support for diabetes self-management. OBJECTIVE This study explored whether rural patients’ self-management and glycemic control was associated with the use of SM. METHODS This secondary, cross-sectional, mixed methods analysis of 448 veterans with diabetes used stratified random sampling to recruit a diverse sample from the United States (rural vs urban and good vs poor glycemic control). Administrative, clinical, survey, and interview data were used to determine patients’ rurality, use of SM, diabetes self-management behaviors, and glycemic control. Moderated mediation analyses assessed these relationships. RESULTS The sample was 51% (n=229) rural and 49% (n=219) urban. Mean participant age was 66.4 years (SD 7.7 years). More frequent SM use was associated with better diabetes self-management (P=.007), which was associated with better glycemic control (PP=.01). CONCLUSIONS More frequent SM use is associated with engaging in diabetes self-management, which, in turn, is associated with better diabetes control. Among rural patients with diabetes, SM use is indirectly associated with better diabetes control. Frequent patient-team communication through SM about diabetes-related content may help rural patients with diabetes self-management, resulting in better glycemic control.

Purpose Monoclonal gammopathy of undetermined significance (MGUS) is a prevalent yet largely asymptomatic precursor to multiple myeloma. Patients with MGUS must undergo regular surveillance and testing, with few known predictors of progression. We developed an algorithm to identify MGUS patients in electronic health data to facilitate large-scale, population-based studies of this premalignant condition. Methods We developed a four-step algorithm using electronic health record and health claims data from men and women aged 50 years or older receiving care from a large, multispecialty medical group between 2007 and 2015. The case definition required patients to have at least two MGUS ICD-9 diagnosis codes within 12 months, at least one serum and/or urine protein electrophoresis and one immunofixation test, and at least one in-office hematology/oncology visit. Medical charts for selected cases were abstracted then adjudicated independently by two physicians. We assessed algorithm validity by positive predictive value (PPV). Results We identified 833 people with at least two MGUS diagnosis codes; 429 (52%) met all four algorithm criteria. We randomly selected 252 charts for review, including 206 from patients meeting all four algorithm criteria. The PPV for the 206 algorithm-identified charts was 76% (95% CI, 70%-82%). Among the 49 cases deemed to be false positives (24%), 33 were judged to have multiple myeloma or another lymphoproliferative condition, such as lymphoma. Conclusions We developed a simple algorithm that identified MGUS cases in electronic health data with reasonable accuracy. Inclusion of additional steps to eliminate cases with malignant disease may improve algorithm performance.

BACKGROUND: Long-term trends in the incidence rates (IRs) and hospital case-fatality rates (CFRs) of ventricular tachycardia (VT) and ventricular fibrillation (VF) among patients hospitalized with acute myocardial infarction (AMI) have not been recently examined. METHODS AND RESULTS: We used data from 11,825 patients hospitalized with AMI at all 11 medical centers in central Massachusetts on a biennial basis between 1986 and 2011. Multivariable adjusted logistic regression modeling was used to examine trends in hospital IRs and CFRs of VT and VF complicating AMI. The median age of the study population was 71 years, 57.9% were men, and 94.7% were white. The hospital IRs declined from 14.3% in 1986/1988 to 10.5% in 2009/2011 for VT and from 8.2% to 1.7% for VF. The in-hospital CFRs declined from 27.7% to 6.9% for VT and from 49.6% to 36.0% for VF between 1986/1988 and 2009/2011, respectively. The IRs of both early (

OBJECTIVE: To describe the characteristics of high‐frequency hospital users (four or more hospitalizations in a year) and the consequences of including or excluding their data from a readmission‐based measure. DATA SOURCES: 2015 and 2016 Massachusetts Medicaid data. STUDY DESIGN: We compare demographics, morbidity burden, and social risk factors for high‐ and low‐frequency hospital users, and membership in 17 accountable care organizations. We evaluate how excluding hospitalizations of high‐frequency users from a 30‐day readmission measure (with or without risk adjustment) changes its rate and variability and affects performance rankings of accountable care organizations. The outcome is readmission within 30 days; each live discharge from a hospital contributes one observation. DATA COLLECTION/EXTRACTION METHODS: We studied 74 706 hospitalizations of 42 794 MassHealth members, 18–64 years old, managed‐care‐eligible, and ever hospitalized in 2016. PRINCIPAL FINDINGS: Among adult managed‐care‐eligible MassHealth members with at least one acute hospitalization, 8.7% were high‐frequency hospital users; they contributed 30.2% of hospitalizations and 69.4% of readmissions. High‐frequency users were more often male (77.1% vs. 50.0%; P

Importance Better patient management can reduce emergency department (ED) use. Performance measures should reward plans for reducing utilization by predictably high-use patients, rather than rewarding plans that shun them. Objective The objective of this study was to develop a quality measure for ED use for people diagnosed with serious mental illness or substance use disorder, accounting for both medical and social determinants of health (SDH) risks. Design Regression modeling to predict ED use rates using diagnosis-based and SDH-augmented models, to compare accuracy overall and for vulnerable populations. Setting MassHealth, Massachusetts' Medicaid and Children's Health Insurance Program. Participants MassHealth members ages 18-64, continuously enrolled for the calendar year 2016, with a diagnosis of serious mental illness or substance use disorder. Exposures Diagnosis-based model predictors are diagnoses from medical encounters, age, and sex. Additional SDH predictors describe housing problems, behavioral health issues, disability, and neighborhood-level stress. Main outcome and measures We predicted ED use rates: (1) using age/sex and distinguishing between single or dual diagnoses; (2) adding summarized medical risk (DxCG); and (3) further adding social risk (SDH). Results Among 144,981 study subjects, 57% were women, 25% dually diagnosed, 67% White/non-Hispanic, 18% unstably housed, and 37% disabled. Utilization was higher by 77% for those dually diagnosed, 50% for members with housing problems, and 18% for members living in the highest-stress neighborhoods. SDH modeling predicted best for these high-use populations and was most accurate for plans with complex patients. Conclusion To set appropriate benchmarks for comparing health plans, quality measures for ED visits should be adjusted for both medical and social risks.

Background Studies suggest that exposure to ambient air pollution during pregnancy may be associated with increased risks of birth defects (BDs), but conclusions have been inconsistent. This study describes the ethnic distribution of major BDs and examines the relationship between air pollution and BDs among different ethnic groups in Liuzhou city, China. Methods Surveillance data of infants born in 114 registered hospitals in Liuzhou in 2019 were analyzed to determine the epidemiology of BDs across five major ethnic groups. Concentrations of six air pollutants (PM2.5, PM10, SO2, CO, NO2, O3) were obtained from the Liuzhou Environmental Protection Bureau. Logistic regression was used to examine the associations between ambient air pollution exposure and risk of BDs. Results Among 32,549 infants, 635 infants had BDs, yielding a prevalence of 19.5 per 1000 perinatal infants. Dong ethnic group had the highest prevalence of BDs (2.59%), followed by Yao (2.57%), Miao (2.35%), Zhuang (2.07%), and Han (1.75%). Relative to the Han ethnic group, infants from Zhuang, Miao, Yao and Dong groups had lower risks of congenital heart disease, polydactyly, and hypospadias. The Zhuang ethnic group had higher risks of severe thalassemia, cleft lip and/or palate, and syndactyls. Overall BDs were positively correlated with air pollutants PM10 (aOR =1.14, 95% CI:1.12 ~ 2.43; aOR =1.51, 95% CI:1.13 ~ 2.03 for per 10μg/mg3 increment) and CO (aOR =1.36, 95% CI:1.14 ~ 2.48; aOR =1.75, 95% CI:1.02 ~ 3.61 for every 1 mg /m3 increment) in second and third month of pregnancy. SO2 was also significantly associated with BDs in the second month before the pregnancy (aOR = 1.31; 95% CI: 1.20 ~ 3.22) and third month of pregnancy (aOR =1.75; 95% CI:1.02 ~ 3.61). Congenital heart disease, polydactyl, cleft lip and/or palate were also significantly associated with PM10, SO2 and CO exposures. However, no significant association was found between birth defects and O3, PM2.5 and NO2 exposures (P > 0.05). Conclusion This study provides a comprehensive description of ethnic differences in BDs in Southwest China and broadens the evidence of the association between air pollution exposure during gestation and BDs.

Objective To develop a cardiovascular disease (CVD) risk score that can be used to quantify CVD risk in the Medicare Current Beneficiary Survey (MCBS). Data sources We used 1999-2013 MCBS data. Study design We used a backward stepwise approach and cox proportional hazards regressions to build and validate a new CVD risk score, similar to the Framingham Risk Score (FRS), using only information available in MCBS. To assess its performance, we calculated C statistics and examined calibration plots. Data collection/extraction methods We studied 21 968 community-dwelling Medicare beneficiaries aged 65 years or older without pre-existing CVD. We obtained risk factors from both survey and claims data. We used claims data to derive "CVD event within 3 years" following the FRS definition of CVD. Principal findings About five percent of MCBS participants developed a CVD event over a mean follow-up period of 348 days. Our final MCBS-based model added morbidity burden, reported general health status, and functional limitation to the traditional FRS predictors of CVD. This model had relatively fair discrimination (C statistic = 0.69; 95% confidence interval [CI], 0.67-0.71) and performed well on validation (C = 0.68; CI, 0.66-0.70). More importantly, the plot of observed CVD outcomes versus predicted ones showed that this model had a good calibration. Conclusions Our new CVD risk score can be calculated using MCBS data, thereby extending the survey's ability to quantify CVD risk in the Medicare population and better inform both health policy and health services research.

Current disease risk-adjustment formulas in the US rely on diagnostic classification frameworks that predate theTo develop anPhysician teams mapped allFourteen concurrent outcomes were predicted: overall and plan-paid health care spending (top-coded and not top-coded); enrollee out-of-pocket spending; hospital days and admissions; emergency department visits; and spending for 6 types of services. The primary outcome was annual health care spending top-coded at $250 000.A total of 65 901 460 person-years were split into 90% estimation/10% validation samples (n = 6 604 259). In all, 3223 DXIs were created: 2435 main effects, 772 modifiers, and 16 scaled items. Stepwise regressions predicting annual health care spending (mean [SD], $5821 [$17 653]) selected 76% of the main effect DXIs with no evidence of overfitting. ValidatedIn this diagnostic modeling study, the new DXI classification system showed improved predictions over existing diagnostic classification systems for all spending and utilization outcomes considered.

Trauma centers improve outcomes for young patients with serious injuries. However, most injury-related hospital admissions and deaths occur in older adults, and it is not clear whether trauma center care provides the same benefit in this population.To examine whether 30- and 365-day mortality of injured older adults is associated with the treating hospital's trauma center level.This prospective, population-based cohort study used Medicare claims data from January 1, 2013, to December 31, 2016, for all fee-for-service Medicare beneficiaries 66 years or older with inpatient admission for traumatic injury in 2014 to 2015. Data analysis was performed from January 1 to June 31, 2021. Preinjury health was measured using 2013 claims, and outcomes were measured through 2016. The population was stratified by anatomical injury pattern. Propensity scores for level I trauma center treatment were estimated using the Abbreviated Injury Scale, age, and residential proximity to trauma center and then used to match beneficiaries from each trauma level (I, II, III, and IV/non-trauma centers) by injury type.Admitting hospital's trauma center level.Case fatality rates (CFRs) at 30 and 365 days after injury, estimated in the matched sample using multivariable, hierarchical logistic regression models.A total of 433 169 Medicare beneficiaries (mean [SD] age, 82.9 [8.3] years; 68.4% female; 91.5% White) were included in the analysis. A total of 206 275 (47.6%) were admitted to non-trauma centers and 161 492 (37.3%) to level I or II trauma centers. Patients with isolated extremity fracture had the fewest deaths (365-day CFR ranged from 16.1% [95% CI, 11.2%-22.4%] to 17.4% [95% CI, 11.8%-24.6%] by trauma center status). Patients with both hip fracture and traumatic brain injury had the most deaths (365-day CFRs ranged from 33.4% [95% CI, 25.8%-42.1%] to 35.8% [95% CI, 28.9%-43.5%]).These findings suggest that older adults do not benefit from existing trauma center care, which is designed with younger patients in mind. There is a critical need to improve trauma care practices to address common injury mechanisms and types of injury in older adults.

Background Little is known about the association of hyperglycemia with the development of ventricular tachycardia (VT) in patients hospitalized with acute myocardial infarction (AMI) which we examined in the present study. The objectives of this community-wide observational study were to examine the relation between elevated serum glucose levels at the time of hospital admission for AMI and occurrence of VT, and time of occurrence of VT, during the patient’s acute hospitalization. Methods We used data from a population-based study of patients hospitalized with AMI at all central Massachusetts medical centers between 2001 and 2011. Hyperglycemia was defined as a serum glucose level ≥ 140 mg/dl at the time of hospital admission. The development of VT was identified from physicians notes and electrocardiographic findings by our trained team of data abstractors. Results The average age of the study population was 70 years, 58.0% were men, and 92.7% were non-Hispanic whites. The mean and median serum glucose levels at the time of hospital admission were 171.4 mg/dl and 143.0, respectively. Hyperglycemia was present in 51.9% of patients at the time of hospital admission; VT occurred in 652 patients (15.8%), and two-thirds of these episodes occurred during the first 48 h after hospital admission (early VT). After multivariable adjustment, patients with hyperglycemia were at increased risk for developing VT (adjusted OR = 1.48, 95% CI = 1.23–1.78). The presence of hyperglycemia was significantly associated with early (multivariable adjusted OR = 1.39, 95% CI = 1.11–1.73) but not with late VT. Similar associations were observed in patients with and without diabetes and in patients with and without ST-segment elevation AMI. Conclusions Efforts should be made to closely monitor and treat patients who develop hyperglycemia, especially early after hospital admission, to reduce their risk of VT.

Importance Socioeconomic disadvantage is associated with poor health outcomes. However, whether socioeconomic factors are associated with post–myocardial infarction (MI) outcomes in younger patient populations is unknown. Objective To evaluate the association of neighborhood-level socioeconomic disadvantage with long-term outcomes among patients who experienced an MI at a young age. Design, Setting, and Participants This cohort study analyzed patients in the Mass General Brigham YOUNG-MI Registry (at Brigham and Women’s Hospital and Massachusetts General Hospital in Boston, Massachusetts) who experienced an MI at or before 50 years of age between January 1, 2000, and April 30, 2016. Each patient’s home address was mapped to the Area Deprivation Index (ADI) to capture higher rates of socioeconomic disadvantage. The median follow-up duration was 11.3 years. The dates of analysis were May 1, 2020, to June 30, 2020. Exposures Patients were assigned an ADI ranking according to their home address and then stratified into 3 groups (least disadvantaged group, middle group, and most disadvantaged group). Main Outcomes and Measures The outcomes of interest were all-cause and cardiovascular mortality. Cause of death was adjudicated from national registries and electronic medical records. Cox proportional hazards regression modeling was used to evaluate the association of ADI with all-cause and cardiovascular mortality. Results The cohort consisted of 2097 patients, of whom 2002 (95.5%) with an ADI ranking were included (median [interquartile range] age, 45 [42-48] years; 1607 male individuals [80.3%]). Patients in the most disadvantaged neighborhoods were more likely to be Black or Hispanic, have public insurance or no insurance, and have higher rates of traditional cardiovascular risk factors such as hypertension and diabetes. Among the 1964 patients who survived to hospital discharge, 74 (13.6%) in the most disadvantaged group compared with 88 (12.6%) in the middle group and 41 (5.7%) in the least disadvantaged group died. Even after adjusting for a comprehensive set of clinical covariates, higher neighborhood disadvantage was associated with a 32% higher all-cause mortality (hazard ratio, 1.32; 95% CI, 1.10-1.60;P = .004) and a 57% higher cardiovascular mortality (hazard ratio, 1.57; 95% CI, 1.17-2.10;P = .003). Conclusions and Relevance This study found that, among patients who experienced an MI at or before age 50 years, socioeconomic disadvantage was associated with higher all-cause and cardiovascular mortality even after adjusting for clinical comorbidities. These findings suggest that neighborhood and socioeconomic factors have an important role in long-term post-MI survival.

To explore the influence of contextual factors on health-related quality of life (HRQoL), which is sometimes used as an indicator of quality of care, we examined the association of neighborhood socioeconomic status (NSES) and trajectories of HRQoL after hospitalization for acute coronary syndromes (ACS).We studied 1481 patients hospitalized with acute coronary syndromes in Massachusetts and Georgia querying HRQoL via the mental and physical components of the 36-item short-form health survey (SF-36) (MCS and PCS) and the physical limitations and angina-related HRQoL subscales of the Seattle Angina Questionnaire (SAQ) during hospitalization and at 1-, 3-, and 6-month postdischarge. We categorized participants by tertiles of the neighborhood deprivation index (a residence-census tract-based measure) to examine the association of NSES with trajectories of HRQoL after adjusting for individual socioeconomic status (SES) and clinical characteristics.Participants had mean age 61.3 (SD, 11.4) years; 33% were female; 76%, non-Hispanic white; 11.2% had household income below the federal poverty level. During 6 months postdischarge, living in lower NSES neighborhoods was associated with lower mean PCS scores (1.5 points for intermediate NSES; 1.8 for low) and SAQ scores (2.4 and 4.2 points) versus living in high NSES neighborhoods. NSES was more consequential for patients with lower individual SES. Individuals living below the federal poverty level had lower average MCS and SAQ physical scores (3.7 and 7.7 points, respectively) than those above.Neighborhood deprivation was associated with worse health status. Using HRQoL to assess quality of care without accounting for individual SES and NSES may unfairly penalize safety-net hospitals.

Background Hospital performance measures based on patient mortality and readmission have indicated modest rates of agreement. We examined if combining clinical data on laboratory tests and vital signs with administrative data leads to improved agreement with each other, and with other measures of hospital performance in the nation's largest integrated health care system. Methods We used patient-level administrative and clinical data, and hospital-level data on quality indicators, for 2007–2010 from the Veterans Health Administration (VA). For patients admitted for acute myocardial infarction (AMI), heart failure (HF) and pneumonia we examined changes in hospital performance on 30-d mortality and 30-d readmission rates as a result of adding clinical data to administrative data. We evaluated whether this enhancement yielded improved measures of hospital quality, based on concordance with other hospital quality indicators. Results For 30-d mortality, data enhancement improved model performance, and significantly changed hospital performance profiles; for 30-d readmission, the impact was modest. Concordance between enhanced measures of both outcomes, and with other hospital quality measures – including Joint Commission process measures, VA Surgical Quality Improvement Program (VASQIP) mortality and morbidity, and case volume – remained poor. Conclusions Adding laboratory tests and vital signs to measure hospital performance on mortality and readmission did not improve the poor rates of agreement across hospital quality indicators in the VA. Interpretation Efforts to improve risk adjustment models should continue; however, evidence of validation should precede their use as reliable measures of quality.

Much of emergency department use is avoidable, and high-quality primary care can reduce it, but performance measures related to ED use may be inadequately risk-adjusted. To explore associations between emergency department (ED) use and neighborhood poverty, we conducted a secondary analysis of Massachusetts managed care network data, 2009-2011. For enrollees with commercial insurance (n = 64,623), we predicted any, total, and total primary-care-sensitive (PCS) ED visits using claims/enrollment (age, sex, race, morbidity, prior ED use), network (payor, primary care provider [PCP] type and quality), and census-tract-level characteristics. Overall, 14.6% had any visit; mean visits per 100 persons were 18.8 (±0.2) total and 7.6 (±0.1) PCS. Neighborhood poverty predicted all three outcomes (all Plt; .001). Holding providers accountable for their patients' ED use should avoid penalizing PCPs who care for poor and otherwise vulnerable populations. Expected use targets should account for neighborhood-level variables such as income, as well as other risk factors.

RESEARCH OBJECTIVE: To describe the characteristics of high‐frequency hospital users (with 4 or more hospital visits per year) and to assess the impact of their inclusion or exclusion on 30‐day hospital readmission. STUDY DESIGN: We assigned MassHealth members to 17 accountable care organizations (ACOs) and 2 other groups (managed care organization or primary care clinician plan). We then compared demographics, morbidity burden, and social determinant of health factors for high‐frequency hospital users and low‐frequency users. We also evaluated the extent to which including or excluding hospitalizations of high‐frequency users from a 30‐day readmission measure changes its rate and variance. POPULATION STUDIED: Managed care eligible MassHealth members between 18 and 64 of age with at least one acute hospitalization during 2016. PRINCIPAL FINDINGS: Of the 42,794 eligible members, only 8.7% were high‐frequency hospital users, contributing 30.2% of all hospital visits. High‐frequency hospital users have many medical morbidities and significant psychiatric, substance abuse, and housing problems. Their readmission rate was 50.7% compared to 9.7% for other patients, and they contributed 72.0% of the total variance in 30‐day readmission. CONCLUSIONS: Despite their small numbers, high‐frequency hospital users exert a large influence on 30‐readmission rates. IMPLICATIONS FOR POLICY OR PRACTICE: Thirty‐day readmission is being extended to ACOs as a quality measure. ACOs that disproportionately serve many high‐frequency hospital users may face unfair penalties for high readmission rates. PRIMARY FUNDING SOURCE: The study was funded by the State of Massachusetts.

Explicit bias reflects our perceptions at a conscious level. In contrast, implicit bias is unintentional and operates at a level below our conscious awareness. Implicit stereotypes shaping implicit biases are widely studied in criminal justice, medicine, CEO selection at Fortune 500 companies, etc. However, the problem of unconscious bias remains. E.g., while women constitute an increasing proportion of all STEM undergraduates, they still make up only a small proportion of faculty members at research universities, and they are substantially under-represented in organizational leadership and as recipients of professional awards and prizes. Can we afford to have unintentional perceptions continue to hinder the success and advancement of women and other underrepresented groups? Can we afford to continue to underuse human capital in science? This session at the 2015 Joint Statistical Meetings (JSM) aimed to illuminate what statisticians need to know and do to break the glass ceiling of implicit bias.

We review statistical methods for estimating and interpreting league tables used to infer hospital quality with a primary focus on methods for partitioning variation into two types: (a) that associated with within-hospital variation for a homogeneous group of patients and (b) that produced by between-hospital variation. We discuss the types of covariates included in the model, hierarchical and nonhierarchical logistic regression models for conducting inferences in a low-information context and their associated trade-offs, and the role of hospital volume. We use all-cause mortality rates for US hospitals to illustrate concepts and methods.

Early rehospitalization after discharge for an acute coronary syndrome, including acute myocardial infarction (AMI), is generally considered undesirable. The Centers for Medicare and Medicaid Services (CMS) base hospital financial incentives on risk-adjusted readmission rates after AMI, using claims data in its adjustment models. Little is known about the contribution to readmission risk of factors not captured by claims. For 804 consecutive patients >65 years discharged in 2011 to 2013 from 6 hospitals in Massachusetts and Georgia after an acute coronary syndrome, we compared a CMS-like readmission prediction model with an enhanced model incorporating additional clinical, psychosocial, and sociodemographic characteristics, after principal components analysis. Mean age was 73 years, 38% were women, 25% college educated, and 32% had a previous AMI; all-cause rehospitalization occurred within 30 days for 13%. In the enhanced model, previous coronary intervention (odds ratio [OR] = 2.05, 95% confidence interval [CI] 1.34 to 3.16; chronic kidney disease OR 1.89, 95% CI 1.15 to 3.10; low health literacy OR 1.75, 95% CI 1.14 to 2.69), lower serum sodium levels, and current nonsmoker status were positively associated with readmission. The discriminative ability of the enhanced versus the claims-based model was higher without evidence of overfitting. For example, for patients in the highest deciles of readmission likelihood, observed readmissions occurred in 24% for the claims-based model and 33% for the enhanced model. In conclusion, readmission may be influenced by measurable factors not in CMS' claims-based models and not controllable by hospitals. Incorporating additional factors into risk-adjusted readmission models may improve their accuracy and validity for use as indicators of hospital quality.

Key Points Question Was the transition from International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) to the Tenth Revision (ICD-10-CM) in October 2015 associated with changes in diagnostic category prevalence when diagnoses are grouped by classification system? Findings This interrupted time series analysis and cross-sectional study examined insurance claims for more than 18 million privately insured adults and children in the US from 2010 to 2017 and found instantaneous increases or decreases of 20% or more associated with the ICD-10-CM transition for nearly 1 in 6 (16%) diagnostic categories in 2 of 3 influential diagnostic classification systems. Meaning These findings suggest that diagnostic classification systems developed with ICD-9-CM data may need to be refined for use with ICD-10-CM data for disease surveillance, performance assessment, or risk-adjusted payment., This interrupted time series analysis and cross-sectional study examines changes in the prevalence of diagnostic codes before the implementation of the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) vs after., Importance On October 1, 2015, the US transitioned to the International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) for recording diagnoses, symptoms, and procedures. It is unknown whether this transition was associated with changes in diagnostic category prevalence based on diagnosis classification systems commonly used for payment and quality reporting. Objective To assess changes in diagnostic category prevalence associated with the ICD-10-CM transition. Design, Setting, and Participants This interrupted time series analysis and cross-sectional study examined level and trend changes in diagnostic category prevalence associated with the ICD-10-CM transition and clinically reviewed a subset of diagnostic categories with changes of 20% or more. Data included insurance claim diagnoses from the IBM MarketScan Commercial Database from January 1, 2010, to December 31, 2017, for more than 18 million people aged 0 to 64 years with private insurance. Diagnoses were mapped using 3 common diagnostic classification systems: World Health Organization (WHO) disease chapters, Department of Health and Human Services Hierarchical Condition Categories (HHS-HCCs), and Agency for Healthcare Research and Quality Clinical Classification System (AHRQ-CCS). Data were analyzed from December 1, 2018, to January 21, 2020. Exposures US implementation of ICD-10-CM. Main Outcomes and Measures Monthly rates of individuals with at least 1 diagnosis in a diagnostic classification category per 10 000 eligible members. Results The analytic sample contained information on 2.1 billion enrollee person-months with 3.4 billion clinically assigned diagnoses; the mean (range) monthly sample size was 22.1 (18.4 to 27.1 ) million individuals. While diagnostic category prevalence changed minimally for WHO disease chapters, the ICD-10-CM transition was associated with level changes of 20% or more among 20 of 127 HHS-HCCs (15.7%) and 46 of 282 AHRQ-CCS categories (16.3%) and with trend changes of 20% or more among 12 of 127 of HHS-HCCs (9.4%) and 27 of 282 of AHRQ-CCS categories (9.6%). For HHS-HCCs, monthly rates of individuals with any acute myocardial infarction diagnosis increased 131.5% (95% CI, 124.1% to 138.8%), primarily because HHS added non–ST-segment–elevation myocardial infarction diagnoses to this category. The HHS-HCC for diabetes with chronic complications increased by 92.4% (95% CI, 84.2% to 100.5%), primarily from including new diabetes-related hypoglycemia and hyperglycemia codes, and the rate for completed pregnancy with complications decreased by 54.5% (95% CI, –58.7% to –50.2%) partly due to removing vaginal birth after cesarean delivery as a complication. Conclusions and Relevance These findings suggest that the ICD-10-CM transition was associated with large prevalence changes for many diagnostic categories. Diagnostic classification systems developed using ICD-9-CM may need to be refined using ICD-10-CM data to avoid unintended consequences for disease surveillance, performance assessment, and risk-adjusted payments.

Objective Conceptually, access to primary care (through insurance) should reduce emergency department (ED) visits for primary care sensitive (PCS) conditions. We sought to identify characteristics of insured Massachusetts residents associated with PCS ED use, and compare such use for public versus private insurees. Population and setting People under age 65 in the Massachusetts All-Payer Claims Data, 2011-2012. Study design Retrospective, observational analysis of PCS ED use with nonurgent, urgent/primary care treatable, and urgent/potentially avoidable visits being considered PCS. We predicted utilization in 2012 using multivariable regression models and data available in 2011 administrative records. Principal findings Among 2,269,475 nonelderly Massachusetts residents, 40% had public insurance. Among public insurees, PCS ED use was higher than for private (mean, 36.5 vs. 9.0 per 100 persons; adjusted risk ratio, 2.53; 95% confidence limits, 2.49-2.56), while having any primary care visit was less common (70% vs. 83%), as was having any visit to one's own (attributed) primary care provider (38% vs. 44%). Conclusions Public insurance was associated with less access to primary care and more PCS ED use; statewide labor shortages and low reimbursement rates from public insurance may have provided inadequate access to care that might otherwise have helped reduce PCS ED use.

Marginalized two-part random-effects generalized Gamma models have been proposed for analyzing medical expenditure panel data with excessive zeros. While these models provide marginal inference on expected healthcare expenditures, the usual unilateral specification of heteroscedastic variance on one of the two shape parameters for the generalized Gamma distribution in these models fails to encompass important special cases within the generalized gamma modeling framework. In this article, we construct marginalized two-part random-effects models that employ the log-normal, log-skew-normal, generalized Gamma, Weibull, Gamma, and inverse Gamma distributions to delineate the spectrum of nonzero healthcare expenditures in the second part of the models. These marginalized models supply additional choices for analyzing healthcare expenditure panel data with excessive zeros. We review the concepts of marginal effect and incremental effect, and summarize how these effects are estimated. For studies whose primary goal is to make inference on marginal effect or incremental effect of an independent variable with respect to healthcare expenditures, we advocate empirical mean square error criterion and information criteria to choose among candidate models. Then, we use the proposed models in an empirical analysis to examine the impact of the New Cooperative Medical Scheme on healthcare expenditures among older adults in rural China.

Objective Systematic reviews and meta-analyses (SRMAs) rely upon comprehensive searches into diverse resources that catalog primary studies. However, since what constitutes a comprehensive search is unclear, we examined trends in databases searched from 2005-2016, surrounding the publication of search guidelines in 2013, and associations between resources searched and evidence of publication bias in SRMAs involving human subjects. Study Design To ensure comparability of included SRMAs over the 12 years in the face of a near 100-fold increase of international SRMAs (mainly genetic studies from China) during this period, we focused on USA-affiliated SRMAs, manually reviewing 100 randomly selected SRMAs from those published in each year. After excluding articles (mainly for inadequate detail or out-of-scope methods), we identified factors associated with the databases searched, used network analysis to see which resources were simultaneously searched, and used logistic regression to link information sources searched with a lower chance of finding publication bias. Results Among 817 SRMA articles studied, the common resources used were Medline (95%), EMBASE (44%), and Cochrane (41%). Methods journal SRMAs were most likely to use registries and grey literature resources. We found substantial co-searching of resources with only published materials, and not complemented by searches of registries and the grey literature. The 2013 guideline did not substantially increase searching of registries and grey literature resources to retrieve primary studies for the SRMAs. When used to augment Medline, Scopus (in all SRMAs) and ClinicalTrials.gov (in SRMAs with safety outcomes) were negatively associated with publication bias. Conclusions Even SRMAs that search multiple sources tend to search similar resources. Our study supports searching Scopus and CTG in addition to Medline to reduce the chance of publication bias.

Background— Cognitive function is often impaired during hospitalization, but whether this impairment resolves or persists after discharge is unknown. Methods and Results— We enrolled (April 2011–May 2013) and interviewed during hospitalization and 1-month post-discharge 1521 nondemented acute coronary syndrome survivors enrolled in TRACE (Transitions, Risks and Actions in Coronary Events). Cognitive function was assessed using the Telephone Interview of Cognitive Status (range: 0–41) at both time points. Patients reported demographic and psychosocial characteristics and medical records were abstracted. Using the Telephone Interview of Cognitive Status cut point of ≤28, we defined 4 groups of cognitive change based on cognitive status during hospitalization and 1 month later: consistently impaired, transiently impaired, newly impaired, and consistently nonimpaired. Characteristics associated with cognitive change categories were examined using multinomial logistic regression. Participants were 67% male, 84% non-Hispanic white, with mean age±SD 62±11 years; 16% (n=237) were cognitively impaired during hospitalization, and 11% (n=174) were impaired 1 month after discharge. Overall, 80% were consistently nonimpaired, 9% transiently impaired, 7% consistently impaired, and 4% newly impaired. Lower education level, minority status, low health literacy and numeracy, and higher severity of disease were independently associated with cognitive impairment during and after hospitalization. Male sex was associated with increased risk of cognitive impairment after hospital discharge. Conclusions— Cognitive function changes during the transition from hospital to home after acute coronary syndrome are less favorable for men and those with psychosocial vulnerability. Assessing cognitive status both in hospital and post-discharge is important for detecting patients who could benefit from tailored transitional care including early follow-up and booster discharge instructions.

Importance Managed care payment formulas commonly allocate more money for medically complex populations, but ignore most social determinants of health (SDH). Objective To add SDH variables to a diagnosis-based payment formula that allocates funds to managed care plans and accountable care organizations. Design, Setting, and Participants Using data from MassHealth, the Massachusetts Medicaid and Children's Health Insurance Program, we estimated regression models predicting Medicaid spending using a diagnosis-based and SDH-expanded model, and compared the accuracy of their cost predictions overall and for vulnerable populations. MassHealth members enrolled for at least 6 months in 2013 in fee-for-service (FFS) programs (n = 357 660) or managed care organizations (MCOs) (n = 524 607). Exposures We built cost prediction models from a fee-for-service program. Predictors in the diagnosis-based model are age, sex, and diagnoses from claims. The SDH model adds predictors describing housing instability, behavioral health issues, disability, and neighborhood-level stressors. Main Outcomes and Measures Overall model explanatory power and overpayments and underpayments for subgroups of interest for all Medicaid-reimbursable expenditures excepting long-term support services (mean annual cost = $5590 per member). Results We studied 357 660 people who were FFS participants and 524 607 enrolled in MCOs with a combined 806 889 person-years of experience. The FFS program experience included more men (49.6% vs 43.6%), older patients (mean age of 26.1 years vs 21.6 years), and sicker patients (mean morbidity score of 1.16 vs 0.89) than MCOs. Overall, the SDH model performed well, but only slightly better than the diagnosis-based model, explaining most of the spending variation in the managed care population (validated R 2 = 62.4) and reducing underpayments for several vulnerable populations. For example, raw costs for the quintile of people living in the most stressed neighborhoods were 9.6% ($537 per member per year) higher than average. Since greater medical morbidity accounts for much of this difference, the diagnosis-based model underpredicts costs for the most stressed quintile by about 2.1% ($130 per member per year). The expanded model eliminates the neighborhood-based underpayment, as well as underpayments of 72% for clients of the Department of Mental Health (observed costs of about $30 000 per year) and of 7% for those with serious mental illness (observed costs of about $16 000 per year). Conclusions and Relevance Since October 2016, MassHealth has used an expanded model to allocate payments from a prespecified total budget to managed care organizations according to their enrollees’ social and medical risk. Extra payments for socially vulnerable individuals could fund activities, such as housing assistance, that could improve health equity.

BACKGROUND: Children with multiple chronic conditions (CMCC) are increasingly using hospital care. We assessed how much of US pediatric inpatient care is used by CMCC and which chronic conditions are the key drivers of hospital use. METHODS: A retrospective analysis of all 2.3 million US acute-care hospital discharges in 2012 for children age 0 to 18 years in the Kids’ Inpatient Database. The ∼4.5 million US hospitalizations for pregnancy, childbirth, and newborn and neonatal care were not assessed. We adapted the Agency for Healthcare Research and Quality’s Chronic Condition Indicators to classify hospitalizations for children with no, 1, or multiple chronic conditions, and to determine which specific chronic conditions of CMCC are associated with high hospital resource use. RESULTS: Of all pediatric acute-care hospitalizations, 34.3% were of children with no chronic conditions, 36.5% were of those with 1 condition, and 29.3% were of CMCC. Of the $23.6 billion in total hospital costs, 19.7%, 27.4%, and 53.9% were for children with 0, 1, and multiple conditions, respectively, and similar proportions were observed for hospital days. The three populations accounted for the most hospital days were as follows: children with no chronic condition (20.9%), children with a mental health condition and at least 1 additional chronic condition (20.2%), and children with a mental health condition without an additional chronic condition (13.3%). The most common mental health conditions were substance abuse disorders and depression. CONCLUSIONS: CMCC accounted for over one-fourth of acute-care hospitalizations and one-half of all hospital dollars for US pediatric care in 2012. Substantial CMCC hospital resource use involves children with mental health–related conditions.

Background— Chronic liver disease presents a relative contraindication to warfarin therapy, but some patients with liver disease nevertheless require long-term anticoagulation. The goal is to identify which patients with liver disease might safely receive warfarin. Methods and Results— Among 102 134 patients who received warfarin from the Veterans Affairs from 2007 to 2008, International Classification of Diseases -Ninth Revision codes identified 1763 patients with chronic liver disease. Specific diagnoses and laboratory values (albumin, aspartate aminotransferase, alanine aminotransferase, creatinine, and cholesterol) were examined to identify risk of adverse outcomes, while controlling for available bleeding risk factors. Outcomes included percent time in therapeutic range, a measure of anticoagulation control, and major hemorrhagic events, by International Classification of Diseases -Ninth Revision codes. Patients with liver disease had lower mean time in therapeutic range (53.5%) when compared with patients without (61.7%; P P P =0.59), whereas those with the worst score (4) had poor control (29.4%) and high hazard of hemorrhage (hazard ratio, 8.53; P Conclusions— Patients with liver disease receiving warfarin have poorer anticoagulation control and more hemorrhages. A simple 4-point scoring system using albumin and creatinine identifies those at risk for poor outcomes.