Search

OBJECTIVE: To investigate (a) the magnitude of the independent associations of neighborhood‐level and person‐level social risk factors (SRFs) with quality, (b) whether neighborhood‐level SRF associations may be proxies for person‐level SRF associations, and (c) how the association of person‐level SRFs and quality varies by neighborhood‐level SRFs. DATA SOURCES: 2015–2016 Medicare Advantage HEDIS data, Medicare beneficiary administrative data, and 2016 American Community Survey (ACS). STUDY DESIGN: Mixed effects linear regression models (1) estimated overall inequities by neighborhood‐level and person‐level SRFs, (2) compared neighborhood‐level associations to person‐level associations, and (3) tested the interactions of person‐level SRFs with corresponding neighborhood‐level SRFs. DATA COLLECTION/EXTRACTION METHODS: Beneficiary‐level SES and disability administrative data and five‐year ACS neighborhood‐level SRF information were each linked to HEDIS data. PRINCIPAL FINDINGS: For all or nearly all HEDIS measures, quality was worse in neighborhoods lower in SES and in neighborhoods with higher proportions of residents with a disability. Quality by neighborhood racial and ethnic composition was mixed. Accounting for corresponding person‐level SRFs reduced neighborhood SRF associations by 25% for disability, 43% for SES, and 74%–102% for racial and ethnic groups. Person‐level SRF coefficients were not consistently reduced in models that added neighborhood‐level SRFs. In 19 of 35 instances, there were significant (p

The use of surrogate markers to examine the effectiveness of a treatment has the potential to decrease study length and identify effective treatments more quickly. Most available methods to investigate the usefulness of a surrogate marker involve restrictive parametric assumptions and tend to focus on settings where the surrogate is measured at a single point in time. However, in many clinical settings, the potential surrogate marker is often measured repeatedly over time, and thus, the surrogate marker information is a trajectory of measurements. In addition, it is often difficult in practice to correctly specify the relationship between a treatment, primary outcome, and surrogate marker trajectory. In this paper, we propose a model-free definition for the proportion of the treatment effect on the primary outcome that is explained by the treatment effect on the longitudinal surrogate markers. We propose three novel flexible methods to estimate this proportion, develop the asymptotic properties of our estimators, and investigate the robustness of the estimators under multiple settings via a simulation study. We apply our proposed procedures to an AIDS clinical trial dataset to examine a trajectory of CD4 counts as a potential surrogate.

Objective To describe differences in patient experiences of hospital care by preferred language within racial/ethnic groups. Data source 2014-2015 HCAHPS survey data. Study design We compared six composite measures for seven languages (English, Spanish, Russian, Portuguese, Chinese, Vietnamese, and Other) within applicable subsets of five racial/ethnic groups (Hispanics, Asian/Pacific Islanders, American Indian/Alaska Natives, Blacks, and Whites). We measured patient-mix adjusted overall, between- and within-hospital differences in patient experience by language, using linear regression. Data collection methods Surveys from 5 480 308 patients discharged from 4517 hospitals 2014-2015. Principal findings Within each racial/ethnic group, mean reported experiences for non-English-preferring patients were almost always worse than their English-preferring counterparts. Language differences were largest and most consistent for Care Coordination. Within-hospital differences by language were often larger than between-hospital differences and were largest for Care Coordination. Where between-hospital differences existed, non-English-preferring patients usually attended hospitals whose average patient experience scores for all patients were lower than the average scores for the hospitals of their English-preferring counterparts. Conclusions Efforts should be made to increase access to better hospitals for language minorities and improve care coordination and other facets of patient experience in hospitals with high proportions of non-English-preferring patients, focusing on cultural competence and language-appropriate services.

RESEARCH OBJECTIVE: About one‐tenth of Medicare Advantage (MA) enrollees voluntarily disenroll from their health plans annually. Voluntary disenrollment is related to beneficiaries’ negative experiences with their plan, disrupts continuity of care, and conflicts with goals to reduce Medicare costs. Research on racial and ethnic differences in voluntary disenrollment is important, as it may inform efforts to reduce racial and ethnic disparities in care. The few prior studies of this topic have had significant limitations, including reliance on administrative data on race and ethnicity that are often missing or inaccurate, failure to control for potential confounding variables such as dual eligibility for Medicare and Medicaid, and use of data that is more than 20 years old. Our investigation of racial and ethnic differences in rates of disenrollment from MA plans overcomes these limitations. STUDY DESIGN: Disenrollment data came from Medicare’s enrollment system. Social Security Administration data on race and ethnicity were augmented with surname, address, and other Medicare administrative data to calculate probabilities of membership in different racial and ethnic groups. We summarized differences in rates of disenrollment across racial and ethnic groups (Asian or Pacific Islander [API], black, Hispanic, and white) using four types of logistic regression models: adjusted and unadjusted models estimating overall differences and adjusted and unadjusted models estimating within‐plan differences. Unadjusted overall models included only racial and ethnic group probabilities as predictors. Adjusted overall models added age, gender, dual eligibility, disability, and state of residence as control variables. For each model, we calculated the difference from whites for each racial and ethnic minority group. We also estimated differences in disenrollment rates that are attributable to within‐plan effects. Between‐plan differences were estimated by subtracting within‐plan differences from overall differences. POPULATION STUDIED: 16 770 319 Medicare beneficiaries enrolled in 736 MA plans nationwide in 2015. PRINCIPAL FINDINGS: Adjusted rates of disenrollment were significantly (P

Since 1990 the World Health Organization has recommended HIV surveillance among pregnant women as an essential surveillance activity for countries with generalized HIV epidemics. Despite the widespread availability and potential usefulness of antenatal HIV surveillance analyses of such data present important challenges. Within an individual clinic the HIV status of its attendees may be correlated because of similarities in HIV risk among women close in age. Between-clinic correlation may also arise as women often seek antenatal care at clinics located close to their home and individuals living in nearby communities may share important characteristics or behaviours related to susceptibility. A general estimating equation-based approach for spatially-correlated binary data such as that antenatal HIV surveillance based on a pairwise composite likelihood has been described. We present an extended version of this model that can accommodate penalized spline estimators and apply it to antenatal HIV surveillance data collected in 2011 in Botswana to estimate the effects of proximity to the hotspot of the countrys HIV epidemic and age on HIV prevalence. Finally we compare the results with a logistic regression analysis which ignores potential correlation of responses. Copyright (c) 2015 John Wiley & Sons Ltd.