The Contribution of First-name Information to the Accuracy of Racial-and-Ethnic Imputations Varies by Sex and Race-and-Ethnicity Among Medicare Beneficiaries.

BACKGROUND: Data on race-and-ethnicity that are needed to measure health equity are often limited or missing. The importance of first name and sex in predicting race-and-ethnicity is not well understood. OBJECTIVE: The objective of this study was to compare the contribution of first-name information to the accuracy of basic and more complex racial-and-ethnic imputations that incorporate surname information. RESEARCH DESIGN: We imputed race-and-ethnicity in a sample of Medicare beneficiaries under 2 scenarios: (1) with only sparse predictors (name, address, sex) and (2) with a rich set (adding limited administrative race-and-ethnicity, demographics, and insurance). SUBJECTS: A total of 284,627 Medicare beneficiaries who completed the 2014 Medicare Consumer Assessment of Healthcare Providers and Systems survey and reported race-and-ethnicity were included. RESULTS: Hispanic, non-Hispanic Asian/Pacific Islander, and non-Hispanic White racial-and-ethnic imputations are more accurate for males than females under both sparse-predictor and rich-predictor scenarios; adding first-name information increases accuracy more for females than males. In contrast, imputations of non-Hispanic Black race-and-ethnicity are similarly accurate for females and males, and first names increase accuracy equally for each sex in both sparse-predictor and rich-predictor scenarios. For all 4 racial-and-ethnic groups, incorporating first-name information improves prediction accuracy more under the sparse-predictor scenario than under the rich-predictor scenario. CONCLUSION: First-name information contributes more to the accuracy of racial-and-ethnic imputations in a sparse-predictor scenario than in a rich-predictor scenario and generally narrows sex gaps in accuracy of imputations.

One In Five Fewer Heart Attacks: Impact, Savings, And Sustainability In San Diego County Collaborative.

Before 2011 rates of hospitalization for heart attacks were about the same in San Diego County as they were in the rest of California. In 2011 a multistakeholder population health collaborative consisting of partners at the federal, state, and local levels launched Be There San Diego. The collaborative's goal was to reduce cardiovascular events through the spread of best practices aimed at improving control of hypertension, lipid levels, and blood sugar and through patient and medical community activation. Using hospital discharge data for the period 2007-16, we compared acute myocardial infarction (AMI) hospitalization rates in San Diego County and the rest of the state before and after the demonstration project started. AMI hospitalization rates decreased by 22 percent in San Diego County versus 8 percent in the rest of the state, with an estimated 3,826 AMI hospitalizations avoided and $86 million in savings in San Diego. Results show that a science-based health collaborative can improve outcomes while lowering costs, and efforts are under way to ensure the collaborative's sustainability.

How Do Gender Differences in Quality of Care Vary Across Medicare Advantage Plans?

BACKGROUND: Healthcare Effectiveness Data and Information Set (HEDIS) quality measures have long been used to compare care across health plans and to study racial/ethnic and socioeconomic disparities among Medicare Advantage (MA) beneficiaries. However, possible gender differences in seniors' quality of care have received less attention. OBJECTIVE: To test for the presence and nature of any gender differences in quality of care across MA Plans, overall and by domain; to identify those most at risk of poor care. DESIGN: Cross-sectional analysis of individual-level HEDIS measure scores from 23.8 million records using binomial mixed-effect models to estimate the effect of gender on performance. For each measure, we assess variation in gender gaps and their correlation with plan performance. PARTICIPANTS: Beneficiaries from 456 MA plans in 2011-2012 HEDIS data. MAIN MEASURES: Performance on 32 of 34 HEDIS measures which were available in both measurement years. The two excluded measures had mean performance scores below 10%. KEY RESULTS: Women experienced better quality of care than men for 22/32 measures, with most pertaining to screening or treatment. Men experienced better quality on nine measures, including four related to cardiovascular disease and three to potentially harmful drug-disease interactions. Plans varied substantially in the magnitude of gender gaps for 21/32 measures; in general, the gender gap in quality of care was least favorable to men in low-performing plans. CONCLUSIONS: Women generally experienced better quality of care than men. However, women experienced poorer care for cardiovascular disease-related intermediate outcomes and potentially harmful drug-disease interactions. Quality improvement may be especially important for men in low-performing plans and for cardiovascular-related care and drug-disease interactions for women. Gender-stratified reporting could reveal gender gaps, identify plans for which care varies by gender, and motivate efforts to address faults and close the gaps in the delivery system.

Mapping the Gaps: Gender Differences in Preventive Cardiovascular Care among Managed Care Members in Four Metropolitan Areas.

BACKGROUND: Prior research documents gender gaps in cardiovascular risk management, with women receiving poorer quality routine care on average, even in managed care systems. Although population health management tools and quality improvement efforts have led to better overall care quality and narrowing of racial/ethnic gaps for a variety of measures, we sought to quantify persistent gender gaps in cardiovascular risk management and to assess the performance of routinely used commercial population health management tools in helping systems narrow gender gaps. METHODS: Using 2013 through 2014 claims and enrollment data from more than 1 million members of a large national health insurance plan, we assessed performance on seven evidence-based quality measures for the management of coronary artery disease and diabetes mellitus, a cardiac risk factor, across and within four metropolitan areas. We used logistic regression to adjust for region, demographics, and risk factors commonly tracked in population health management tools. FINDINGS: Low-density lipoprotein (LDL) cholesterol control (LDL < 100 mg/dL) rates were 5 and 15 percentage points lower for women than men with diabetes mellitus (p < .0001), and coronary artery disease (p < .0001), respectively. Adjusted analyses showed women were more likely to have gaps in LDL control, with an odds ratio of 1.31 (95% confidence interval, 1.27-1.38) in diabetes mellitus and 1.88 (95% confidence interval, 1.65-2.10) in coronary artery disease. CONCLUSIONS: Given our findings that gender gaps persist across both clinical and geographic variation, we identified additional steps health plans can take to reduce disparities. For measures where gaps have been consistently identified, we recommend that gender-stratified quality reporting and analysis be used to complement widely used algorithms to identify individuals with unmet needs for referral to population health and wellness behavior support programs.

Improving Population Health Through an Innovative Collaborative: The Be There San Diego Data for Quality Group.

In 2012, leaders from disparate health care organizations established a data group aligned around a regional goal of preventing heart attacks and strokes in San Diego. The group---now named the Be There San Diego Data for Quality (DFQ) Group---is a safe venue for medical directors and other quality-improvement leaders to share performance data on quality-of-care measures for diabetes, hypertension, and cardiovascular disease, as well as insights, lessons learned, and challenges faced by each organization in treating these conditions. The DFQ Group has focused its efforts on improving the quality of services provided by each participating health care organization, and has placed a strong emphasis on analyzing trends in combined quality data to better understand the health of the entire San Diego population. By fostering collaboration among organizations that collectively serve a large portion of the local population and other key community stakeholders, the DFQ Group has helped form the foundation of a unique, multifaceted, multi-stakeholder, regional effort that is gaining national attention and funding for its community-driven approach.

When Race/Ethnicity Data Are Lacking: Using Advanced Indirect Estimation Methods to Measure Disparities.

A key aim of U.S. health care reforms is to ensure equitable care while improving quality for all Americans. Limited race/ethnicity data in health care records hamper efforts to meet this goal. Despite improvements in access and quality, gaps persist, particularly among persons belonging to racial/ethnic minority and low-income groups. This study describes the use of indirect estimation methods to produce probabilistic estimates of racial/ethnic populations to monitor health care utilization and improvement. One method described, called Bayesian Indirect Surname Geocoding, uses a person's Census surname and the racial/ethnic composition of their neighborhood to produce a set of probabilities that a given person belongs to one of a set of mutually exclusive racial/ethnic groups. Advances in methods for estimating race/ethnicity are enabling health plans and other health care organizations to overcome a long-standing barrier to routine monitoring and actions to reduce disparities in care. Though these new estimation methods are promising, practical knowledge and guidance on how to most effectively apply newly available race/ethnicity data to address disparities can be greatly extended.

Imputing Missing Race/Ethnicity in Pediatric Electronic Health Records: Reducing Bias with Use of U.S. Census Location and Surname Data.

OBJECTIVE: To assess the utility of imputing race/ethnicity using U.S. Census race/ethnicity, residential address, and surname information compared to standard missing data methods in a pediatric cohort. DATA SOURCES/STUDY SETTING: Electronic health record data from 30 pediatric practices with known race/ethnicity. STUDY DESIGN: In a simulation experiment, we constructed dichotomous and continuous outcomes with pre-specified associations with known race/ethnicity. Bias was introduced by nonrandomly setting race/ethnicity to missing. We compared typical methods for handling missing race/ethnicity (multiple imputation alone with clinical factors, complete case analysis, indicator variables) to multiple imputation incorporating surname and address information. PRINCIPAL FINDINGS: Imputation using U.S. Census information reduced bias for both continuous and dichotomous outcomes. CONCLUSIONS: The new method reduces bias when race/ethnicity is partially, nonrandomly missing.

Mapping Gender Differences in Cardiovascular Disease and Diabetes Care: A Pilot Assessment of LDL Cholesterol Testing Rates in a California Health Plan.

Despite improvements over recent decades in care for cardiovascular disease (CVD) and diabetes, which is a major CVD risk factor, evidence suggests that the care women receive-and their health outcomes-continue to lag behind those of men, even for routine care, such as monitoring and control of cholesterol. Awareness of and action to address gender gaps in women's CVD care are limited, in part, because quality of care is not routinely measured and reported by gender. This pilot project describes and maps gender differences in CVD and diabetes care using data from a large health plan. The study aims to identify gender gaps in care in California and, where they are found, to increase awareness of potential gender disparities and begin to inform approaches to address gaps in care. The study assesses gender differences in one key aspect of routine CVD care and related prevention among patients with diabetes: low-density lipoprotein (LDL) cholesterol screening. The results from this study are intended to provide the basis for further work to learn and document what it will take to rapidly spread mapping approaches that prove to be helpful in understanding and addressing disparities in health and health care, particularly with regard to improving women's health outcomes. Key findings include that men had higher rates of LDL screening than did women in both the CVD and diabetes populations. These gaps varied by age and insurance type. In addition, mapping gender gaps revealed larger gaps at local levels than were apparent at the state or regional level.

Bridging the gap between basic science and clinical practice: a role for community clinicians.

BACKGROUND: Translating the extraordinary scientific and technological advances occurring in medical research laboratories into care for patients in communities throughout the country has been a major challenge. One contributing factor has been the relative absence of community practitioners from the US biomedical research enterprise. Identifying and addressing the barriers that prevent their participation in research should help bridge the gap between basic research and practice to improve quality of care for all Americans. METHODS: We interviewed over 200 clinicians and other healthcare stakeholders from 2004 through 2005 to develop a conceptual framework and set of strategies for engaging a stable cadre of community clinicians in a clinical research program. RESULTS: Lack of engagement of community practitioners, lack of necessary infrastructure, and the current misalignment of financial incentives and research participation emerged as the three primary barriers to community clinician research participation. Although every effort was made to learn key motivators for engagement in clinical research from interviewees, we did not observe their behavior and self-report by clinicians does not always track with their behavior. CONCLUSIONS: A paradigm shift involving acknowledgement of the value of clinicians in the context of community research, establishment of a stable infrastructure to support a cohort of clinicians across time and research studies, and realignment of incentives to encourage participation in clinical research is required.

A Prototype Interactive Mapping Tool to Target Low Health Literacy in Missouri.

An estimated 36 percent of American adults have health literacy levels rated at "basic or below," indicating that they have difficulty obtaining, processing, and understanding basic health information and services. To help healthcare decisionmakers in Missouri identify neighborhood-level "hotspots" of suboptimal health or healthcare that may be due to low health literacy, RAND developed a prototype interactive web-based mapping tool. This builds on earlier RAND work to develop a predictive model of health literacy and estimate levels of health literacy in small geographic areas (e.g., census tracts). The interactive mapping tool allows stakeholders to select the level of geography (e.g., census tract, county), obtain information for and map specific regions of interest, select the characteristics to be mapped (i.e., estimates of community-level health literacy, health outcomes and care quality, neighborhood sociodemographic characteristics, and neighborhood health services data), and generate tables and reports on the regions and characteristics of interest. Housed on a dedicated RAND website (http://www.rand.org/health/projects/missouri-health-literacy.html), the mapping tool makes it possible for a range of stakeholders, from health plans to community organizations, to access and use the tool to help address healthcare disparities in their communities.

Developing predictive models of health literacy.

INTRODUCTION: Low health literacy (LHL) remains a formidable barrier to improving health care quality and outcomes. Given the lack of precision of single demographic characteristics to predict health literacy, and the administrative burden and inability of existing health literacy measures to estimate health literacy at a population level, LHL is largely unaddressed in public health and clinical practice. To help overcome these limitations, we developed two models to estimate health literacy. METHODS: We analyzed data from the 2003 National Assessment of Adult Literacy (NAAL), using linear regression to predict mean health literacy scores and probit regression to predict the probability of an individual having 'above basic' proficiency. Predictors included gender, age, race/ethnicity, educational attainment, poverty status, marital status, language spoken in the home, metropolitan statistical area (MSA) and length of time in U.S. RESULTS: All variables except MSA were statistically significant, with lower educational attainment being the strongest predictor. Our linear regression model and the probit model accounted for about 30% and 21% of the variance in health literacy scores, respectively, nearly twice as much as the variance accounted for by either education or poverty alone. CONCLUSIONS: Multivariable models permit a more accurate estimation of health literacy than single predictors. Further, such models can be applied to readily available administrative or census data to produce estimates of average health literacy and identify communities that would benefit most from appropriate, targeted interventions in the clinical setting to address poor quality care and outcomes related to LHL.

Training a patient safety work force: the patient safety improvement corps.

OBJECTIVE: Evaluate short-term effects of the Patient Safety Improvement Corps (PSIC), an Agency for Healthcare Research and Quality­sponsored program to train state teams in patient safety skills/tools, to assess its contribution to building a national infrastructure supporting effective patient safety practices. DATA SOURCE: Self-reported information gathered from (1) group interviews at the end of each year; (2) individual telephone interviews 1 year later; (3) faxed information forms 2 years later. STUDY DESIGN: Program evaluation of immediate and short-term process and impact (use of skills/tools, information sharing, changes in practice). DATA COLLECTION: Semistructured interviews; faxed forms. PRINCIPAL FINDINGS: One year after training, approximately half of Year 1 and 2 state agency representatives reported they had initiated or modified legislation to strengthen safe practices, and modified adverse event oversight procedures. Approximately three-quarters of hospital representatives said training contributed to modifications to adverse event oversight procedures and promotion of patient safety culture. Two years posttraining, approximately three-quarters of Year 1 trainees said they continued to use many skills/tools. CONCLUSIONS: The PSIC contributed to building a national infrastructure supporting effective patient safety practices. Expanded training is needed to reach a larger fraction of the population for which this training is important.

A new method for estimating race/ethnicity and associated disparities where administrative records lack self-reported race/ethnicity.

OBJECTIVE: To efficiently estimate race/ethnicity using administrative records to facilitate health care organizations' efforts to address disparities when self-reported race/ethnicity data are unavailable. DATA SOURCE: Surname, geocoded residential address, and self-reported race/ethnicity from 1,973,362 enrollees of a national health plan. STUDY DESIGN: We compare the accuracy of a Bayesian approach to combining surname and geocoded information to estimate race/ethnicity to two other indirect methods: a non-Bayesian method that combines surname and geocoded information and geocoded information alone. We assess accuracy with respect to estimating (1) individual race/ethnicity and (2) overall racial/ethnic prevalence in a population. PRINCIPAL FINDINGS: The Bayesian approach was 74 percent more efficient than geocoding alone in estimating individual race/ethnicity and 56 percent more efficient in estimating the prevalence of racial/ethnic groups, outperforming the non-Bayesian hybrid on both measures. The non-Bayesian hybrid was more efficient than geocoding alone in estimating individual race/ethnicity but less efficient with respect to prevalence (p<.05 for all differences). CONCLUSIONS: The Bayesian Surname and Geocoding (BSG) method presented here efficiently integrates administrative data, substantially improving upon what is possible with a single source or from other hybrid methods; it offers a powerful tool that can help health care organizations address disparities until self-reported race/ethnicity data are available.

The National Health Plan Collaborative to Reduce Disparities and Improve Quality.

BACKGROUND: Despite numerous reports and initiatives, progress in reducing racial/ethnic disparities in health care has been slow. The National Health Plan Collaborative (NHPC), a novel public-private partnership between nine health plans covering approximately 95 million lives, leading learning and research organizations, the Agency for Healthcare Research and Quality, and the Robert Wood Johnson Foundation, was established in December 2004 to address these disparities. PROGRESS TO DATE: The health plans were able to overcame initial challenges in obtaining information on race/ethnicity of their enrollees and examined their diabetes performance measure to assess disparities in care. By February 2006, the initial nine plans that had joined the NHPC progressed from focusing solely on data collection and management issues and were engaged in outreach activities to members, providers, or community or had completed capacity development for disparities work. Five plans had implemented one or more pilot interventions. Plans also addressed unanticipated challenges, such as sorting through large amounts of data to target disparities. CHALLENGES AND LESSONS LEARNED: Because many of the plans are complex national entities with varying regional and departmental structures, simply achieving coordination of disparities activities across the organization has been a major challenge and, in many cases, a major breakthrough. CONCLUSIONS: The NHPC represents a model of shared learning and innovation through which health plans are tackling racial/ethnic disparities. Now that most of the plans have some data on their enrollees with diabetes and have begun targeting disparities, they want to capitalize on their collective industry strength to influence policy on issues related to disparities.

Challenges to using a business case for addressing health disparities.

The authors consider the challenges to quantifying both the business case and the social case for addressing disparities, which is central to achieving equity in the U.S. health care system. They describe the practical and methodological challenges faced by health plans exploring the business and social cases for undertaking disparity-reducing interventions. Despite these challenges, sound business and quality improvement principles can guide health care organizations seeking to reduce disparities. Place-based interventions may help focus resources and engage health care and community partners who can share in the costs of-and gains from-such efforts.

Differential diffusion of HIV technologies by gender: the case of highly active antiretroviral therapy.

We sought to examine whether diffusion of new HIV technologies differed by gender in the United States, the source of any such disparities, and whether disparities narrow over time as technologies become more established. In particular, we assess how rates of use of highly active antiretroviral therapy (HAART) varied between males and females during the late 1990s, when HAART was rapidly diffusing. We examined data from a prospective cohort study of a national probability sample of 1421 HIV-infected adults in medical care who were enrolled in the HIV Cost and Services Utilization Study (HCSUS) from January 1996 to December 1998. We found that HAART use substantially increased between 1996 and early 1998 for all groups. Women were less likely to receive HAART at all time points, although the diffusion of HAART between 1996 and 1998 reduced gender disparities. Gender disparities in 1998 were only partially explained by women's lower income and educational levels. We conclude that HAART therapy diffused more slowly to HIV-positive females than other groups. Policies that reduce the impact of income and education inequalities on health care may help to narrow gender disparities for new HIV technologies, but other factors may also disadvantage women.

Differences in HIV care between patients with and without severe mental illness.

OBJECTIVE: This study explored how HIV care differs for infected persons with and without severe mental illness. METHODS: Data were obtained through interviews with and chart review of 295 patients with severe mental illness and HIV from public mental health agencies in Los Angeles County and New York City. Data were compared with data from 1,294 HIV patients without severe mental illness from a separate national probability sample. Measures were difficulty obtaining care, whether patients recommend their HIV care provider, hospital problem score, functional health status, and disability days. RESULTS: In Los Angeles, HIV patients with severe mental illness were more likely than those without severe mental illness to have difficulty obtaining care (p<.001); to not recommend their provider (10% versus 5%, p=.007); and to have problematic hospital care (p=.001), poor health status (p=.001), and more disability days (p<.001). In New York City, HIV patients with severe mental illness were more likely than patients without severe mental illness to have difficulty obtaining care (p=.002) and not recommend their provider (p=.02). The relationship between severe mental illness and health status in Los Angeles and access in New York City became insignificant after adjustment for sociodemographic factors, drug use, and CD4 cell count. Further adjustment for higher case management rates among HIV patients with severe mental illness reduced disparities only in the West. CONCLUSIONS: Patients with severe mental illness experienced more problems with HIV care than patients without severe mental illness, although high case management rates for patients with severe mental illness may have offset some problems.

Does quality of care for cardiovascular disease and diabetes differ by gender for enrollees in managed care plans?

PURPOSE: To assess gender differences in the quality of care for cardiovascular disease and diabetes for enrollees in managed care plans. METHODS: We obtained data from 10 commercial and 9 Medicare plans and calculated performance on 6 Health Employer Data and Information Set (HEDIS) measures of quality of care (beta-blocker use after myocardial infarction [MI], low-density lipoprotein cholesterol [LDL-C] check after a cardiac event, and in diabetics, whether glycosylated hemoglobin [HgbA1c], LDL cholesterol, nephropathy, and eyes were checked) and a 7th HEDIS-like measure (angiotensin-converting enzyme [ACE] inhibitor use for congestive heart failure). A smaller number of plans provided HEDIS scores on 4 additional measures that require medical chart abstraction (control of LDL-C after cardiac event, blood pressure control in hypertensive patients, and HgbA1c and LDL-C control in diabetics). We used logistic regression models to adjust for age, race/ethnicity, socioeconomic status, and plan. MAIN FINDINGS: Adjusting for covariates, we found significant gender differences on 5 of 11 measures among Medicare enrollees, with 4 favoring men. Similarly, among commercial enrollees, we found significant gender differences for 8 of 11 measures, with 6 favoring men. The largest disparity was for control of LDL-C among diabetics, where women were 19% less likely to achieve control among Medicare enrollees (relative risk [RR] = 0.81; 95% confidence interval [CI] = 0.64-0.99) and 16% less likely among commercial enrollees (RR = 0.84; 95%CI = 0.73-0.95). CONCLUSION: Gender differences in the quality of cardiovascular and diabetic care were common and sometimes substantial among enrollees in Medicare and commercial health plans. Routine monitoring of such differences is both warranted and feasible.