[show abstract][hide abstract] ABSTRACT: BACKGROUND: To explore contextual effects and to test for interactions, this study examined how breast cancer stage at diagnosis among U.S. women related to individual- and county-level (contextual) variables associated with access to health care and socioeconomic status. METHODS: Individual-level incidence data were obtained from the National Program of Cancer Registries (NPCR) and the Surveillance, Epidemiology and End-Results (SEER) program. The county of residence of women with diagnosed breast cancer (n = 217,299) was used to link NPCR and SEER data with county-level measures of health care access from the 2004 Area Resource File (ARF). In addition to individual-level covariates such as age, race, and Hispanic ethnicity, we examined county-level covariates (residence in a Health Professional Shortage Area, urban/rural residence; race/ethnicity; and number of health centers/clinics, mammography screening centers, primary care physicians, and obstetrician-gynecologists per 100,000 female population or per 1000 square miles) as predictors of stage of breast cancer at diagnosis. RESULTS: Both individual-level and contextual variables are associated with later stage of breast cancer at diagnosis. Black women and women of "other race" had higher odds of receiving a diagnosis of regional or distant stage breast cancer (P <0.0001 and P = 0.02). With adjustment for age, Hispanics were more likely to receive a diagnosis of later stage breast cancer than non-Hispanics (P <0.0.001). Women living in areas with a higher proportion of black women had greater odds of receiving a diagnosis of regional or late stage breast cancer compared with women living in areas with the lowest proportion of black women. The same was noted for women living in areas with intermediate proportions of Hispanic women (age-adjusted odds ratio [OR], 0.94; 95% confidence interval [CI], 0.92-0.97]. Other important contextual variables associated with stage at diagnosis included the percentage of persons living below the poverty level and the number of office-based physicians per 100,000 women. Women living in counties with a higher proportion of persons living below the poverty level or fewer office-based physicians were more likely to receive a diagnosis of later stage breast cancer than those living in other counties (P < 0.001). In multivariable analysis, residence in areas with a higher proportion of non-Hispanic black women modified the associations of age and Hispanic ethnicity with later stage breast cancer (P = 0.0159 and P = 0.0002, respectively). CONCLUSIONS: This study found that county-level contextual variables related to the availability and accessibility of health care providers and health services can affect the timeliness of breast cancer diagnosis. This information could help public health officials develop interventions to reduce the burden of breast cancer among U.S. women.
The Open Health Services and Policy Journal 01/2009; 2:45-46.
[show abstract][hide abstract] ABSTRACT: In the linear mixed model (LMM), several R2 statistics have been proposed for assessing the goodness-of-fit of fixed effects. However, the performance of these statistics has not been fully demonstrated either analytically or through simulations. We report results of simulations to asses the ability of these statistics to select the most parsimonious model. R2 statistics from a full model were compared to other models in which fixed-effect covariates were removed. The full model was also compared to an overfitted model that included additional covariates not linked to the outcome. All models compared involved the same random effects. In this paper, we show that R2 statistics that involve the residuals are unable to adequately discriminate between the correct model and one from which important fixed-effect covariates are omitted if the computation of the predicted values for the residuals included the random effects (referred to as conditional R2 statistics). However, if the random effects are excluded from the computation of the predicted values that lead to the residuals, these R2 statistics (referred to as marginal R2 statistics) are able to select the most parsimonious model. Other R2 statistics that have been proposed by Xu [2003. Measuring explained variation in linear mixed effects models. Statist. Med. 22(22), 3527–3541] performed poorly in that there was little variation in the value of these statistics from a full model to a reduced model.
Computational Statistics & Data Analysis 02/2008; · 1.30 Impact Factor
[show abstract][hide abstract] ABSTRACT: Remediation of hazardous waste sites requires efficient and cost-effective methods to assess the extent of contamination by toxic substances including dioxin-like chemicals. Traditionally, dioxin-like contamination has been assessed by gas chromatography/high-resolution mass spectrometry (GC/MS) analysis for specific polychlorinated dibenzo-p-dioxins, dibenzofurans, and biphenyl congeners. Toxic equivalency factors for these congeners are then used to estimate the overall dioxin toxic equivalency (TEQ) of complex mixtures found in samples. The XDS-CALUX bioassay estimates contamination by dioxin-like chemicals in a sample extract by measuring expression of a sensitive reporter gene in genetically engineered cells. The output of the XDS-CALUX assay is a CALUX-TEQ value, calibrated based on TCDD standards. Soil samples taken from a variety of hazardous waste sites were measured using the XDS-CALUX bioassay and GC/MS. TEQ and CALUX-TEQ from these methods were compared, and a mathematical model was developed describing the relationship between these two data sets: log(TEQ) = 0.654 x log(CALUX-TEQ) + 0.058-(log(CALUX-TEQ))2. Applying this equation to these samples showed that predicted and GC/MS measured TEQ values strongly correlate (R2 = 0.876) and that TEQ values predicted from CALUX-TEQ were on average nearly identical to the GC/MS-TEQ. The ability of XDS-CALUX bioassay data to predict GC/MS-derived TEQ data should make this procedure useful in risk assessment and management decisions.
Environmental Science and Technology 07/2007; 41(12):4354-60. · 5.26 Impact Factor
[show abstract][hide abstract] ABSTRACT: Racial/ethnic differences in influenza vaccination exist among elderly adults despite nearly universal Medicare health insurance coverage. Overall influenza vaccination prevalence in the Veterans Affairs (VA) Healthcare System is higher than in the general population; however, it is not known whether racial/ethnic differences exist among older adults receiving VA healthcare. Racial/ethnic differences in influenza vaccination in VA were assessed, and barriers to and facilitators of influenza vaccination were examined among veteran outpatients aged 50 years and older.
A random sample of 121,738 veterans receiving care at VA outpatient clinics during the 2003-2004 influenza season completed the mailed Survey of Health Experiences of Patients (77% response rate). Multivariate logistic regression was used to examine associations among race/ethnicity and influenza vaccination prevalence, barriers, and facilitators. Analyses were conducted during 2005 and 2006.
Based on unadjusted prevalences, non-Hispanic blacks, Hispanics, and American Indian/Alaskan Natives were significantly less likely to be vaccinated for influenza compared to non-Hispanic whites (71%, 79%, and 74%, respectively, vs 82%). After adjustment for age, gender, marital status, education level, employment, having a primary care provider, confidence and/trust in provider, and health status, only non-Hispanic blacks remained significantly less likely to be vaccinated compared to non-Hispanic whites (75% vs 81%). Influenza vaccination barriers and facilitators varied by race/ethnic group.
Compared to non-Hispanic whites, non-Hispanic blacks were less likely to receive influenza vaccination in the VA healthcare system during the 2003-2004 influenza season. Although these differences were small, results suggest the need for further study and culturally informed interventions.
American Journal of Preventive Medicine 12/2006; 31(5):375-82. · 3.95 Impact Factor
[show abstract][hide abstract] ABSTRACT: The aim of the study is to estimate incidence rates of occupational blood exposure by route of exposure (needlesticks; cuts from sharp objects; mucous membrane exposures to the eyes, nose, or mouth; bites; and blood contact with nonintact skin) in US and California paramedics.
A mail survey was conducted in a national probability sample of certified paramedics.
Proportions of paramedics who reported an exposure in the previous year were 21.6% (95% confidence interval [CI], 17.8-25.3) for the national sample and 14.8% (95% CI, 12.2-17.4) for California. The overall incidence rate was 6.0/10,000 calls (95% CI, 3.9-8.1). These rates represent more than 49,000 total exposures and more than 10,000 needlesticks per year among paramedics in the United States. Rates for mucocutaneous exposures and needlesticks were similar (approximately 1.2/10,000 calls). Rates for California were one third to one half the national rates. Sensitivity analysis showed that potential response bias would have little impact on the policy and intervention implications of the findings.
Paramedics continue to be at substantial risk for blood exposure. More attention should be given to reducing mucocutaneous exposures. The impact of legislation on reducing exposures and the importance of nonintact skin exposures need to be better understood.
Annals of Epidemiology 10/2006; 16(9):720-5. · 2.48 Impact Factor
[show abstract][hide abstract] ABSTRACT: Generalized estimating equation (GEE) models are used in many situations to assess the effect of an independent variable on an outcome in the presence of correlation among the observations. Once an independent variable has been found to be a significant factor, researchers may also be interested in identifying the levels of that independent variable responsible for the variation in the outcome under study. For example, in many toxicology experiments, one of the goals is to determine which doses of a toxic substance are significantly different from a control. Typically this question can be answered by performing a multiple comparison with a control (MCC) test procedure. However, the question of how to perform multiple comparisons, including MCC in GEE models, has not been addressed in the statistical literature. In this paper, we propose an approach for performing MCC in GEE models that can be extended to any models in which the parameter estimates can be assumed to have a normal or an asymptotic normal distribution. This test is simply a generalization of the methods used in traditional general linear models. Simulation results show that this test provides adequate Type I error rates and power.
[show abstract][hide abstract] ABSTRACT: Controlled tabular adjustment preserves confidentiality and tabular structure. Quality-preserving controlled tabular adjustment
in addition preserves parameters of the distribution of the original (unadjusted) data. Both methods are based on mathematical
programming. We introduce a method for preserving the original distribution itself, a fortiori the distributional parameters.
The accuracy of the approximation is measured by minimum discrimination information. MDI is computed using an optimal statistical
algorithm—iterative proportional fitting.
Privacy in Statistical Databases, CENEX-SDC Project International Conference, PSD 2006, Rome, Italy, December 13-15, 2006, Proceedings; 01/2006
[show abstract][hide abstract] ABSTRACT: To construct a single estimate of the number of percutaneous injuries sustained annually by healthcare workers (HCWs) in the United States.
We combined data collected in 1997 and 1998 at 15 National Surveillance System for Health Care Workers (NaSH) hospitals and 45 Exposure Prevention Information Network (EPINet) hospitals. The combined data, taken as a sample of all U.S. hospitals, were adjusted for underreporting. The estimate of the number of percutaneous injuries nationwide was obtained by weighting the number of percutaneous injuries at each hospital by the number of admissions in all U.S. hospitals relative to the number of admissions at that hospital.
The estimated number of percutaneous injuries sustained annually by hospital-based HCWs was 384,325 (95% confidence interval, 311,091 to 463,922). The number of percutaneous injuries sustained by HCWs outside of the hospital setting was not estimated.
Although our estimate is smaller than some previously published estimates of percutaneous injuries among HCWs, its magnitude remains a concern and emphasizes the urgent need to implement prevention strategies. In addition, improved surveillance could be used to monitor injury trends in all healthcare settings and evaluate the impact of prevention interventions.
Infection Control and Hospital Epidemiology 08/2004; 25(7):556-62. · 4.02 Impact Factor
[show abstract][hide abstract] ABSTRACT: To assess the prevalence of HIV antiretroviral resistance among source patients for occupational HIV exposures.
Blood and data (eg, stage of HIV, previous antiretroviral drug therapy, and HIV RNA viral load) were collected from HIV-infected patients who were source patients for occupational exposures.
Seven tertiary-care medical centers in five U.S. cities (San Diego, California; Miami, Florida; Boston, Massachusetts; Albany, New York; and New York, New York [three sites]) during 1998 to 1999.
Sixty-four HIV-infected patients who were source patients for occupational exposures.
Virus from 50 patients was sequenced; virus from 14 patients with an undetectable (ie, < 400 RNA copies/mL) viral load could not be sequenced. Overall, 19 (38%) of the 50 patients had primary genotypic mutations associated with resistance to reverse transcriptase or protease inhibitors. Eighteen of the 19 viruses with primary mutations and 13 wild type viruses were phenotyped by recombinant assays; 19 had phenotypic resistance to at least one antiretroviral agent. Of the 50 source patients studied, 26 had taken antiretroviral agents in the 3 months before the occupational exposure incident. Sixteen (62%) of the 26 drug-treated patients had virus that was phenotypically resistant to at least one drug. Four (17%) of 23 untreated patients had phenotypically resistant virus. No episodes of HIV transmission were observed among the exposed HCWs.
There was a high prevalence of drug-resistant HIV among source patients for occupational HIV exposures. Healthcare providers should use the drug treatment information of source patients when making decisions about post-exposure prophylaxis.
Infection Control and Hospital Epidemiology 10/2003; 24(10):724-30. · 4.02 Impact Factor