Understanding Reversed Race/Ethnic Associations with Cancer Screening Tests

了解种族/民族与癌症筛查测试的反向关联

• 批准号: 7238472
• 负责人: WILLIAM RAKOWSKI
• 金额: $ 16.85万
• 依托单位: BROWN UNIVERSITY
• 依托单位国家: 美国
• 财政年份: 2007
• 资助国家: 美国
• 起止时间: 2007-04-12 至 2009-03-31
• 项目状态: 已结题
• 来源: https://reporter.nih.gov/project-details/7238472
• 关键词: Address; Adoption; Age; Attention; Behavioral Risk Factor Surveillance System; Breast; Clinical; Colorectal; Data; Data Set; Databases; Decision Making; Diffusion; Disease regression; Ethnic Origin; Ethnic group; Factor Analysis; Future; Generic Drugs; Health Information National Trends Survey; Individual; Insurance Coverage; Intervention; Logistic Regressions; Malignant Neoplasms; Mammography; Medical Surveillance; Morbidity - disease rate; Multivariate Analysis; National Health Interview Survey; Not Hispanic or Latino; Numbers; Pap smear; Persons; Play; Population; Population Group; Procedures; Process; Prostate; Purpose; Race; Rate; Recommendation; Recording of previous events; Relative (related person); Research; Research Project Grants; Resource Allocation; Risk; Role; Screening for cancer; Screening procedure; Statistically Significant; Surveys; Testing; Time; Translating; Weight; Woman; base; cancer prevention; male; men; mortality; racial and ethnic; trend

DESCRIPTION (provided by applicant): Evidence for disparities in the utilization of screening tests plays a central role in the nation's efforts at cancer prevention and control. Persons in traditionally underserved racial and ethnic groups have had lower screening rates relative to non-Hispanic Whites. In addition, they often have disproportionately higher morbidity and mortality. The process of making decisions about groups in the population who are "at-risk" of lower screening test utilization typically starts with bivariate analyses of the associations between a set of potential covariates and use of a screening test. The next step is to conduct multivariable analysis most often using logistic regression, to identify the final set of variables that achieve statistical significance. This final set of statistically significant variables is then translated into population groups who need targeted interventions to reduce disparities. This proposed research addresses a situation we, and others, have observed in several analyses of cancer screening in population-based data sets. That is, analyses done with different survey databases have found several instances where bivariate associations between race/ethnicity and cancer screening have been "reversed" in multivariable analyses. That is, a group with lower screening in bivariate ("crude rate") analysis shows higher screening based on the "adjusted" results from the multivariable analysis. These reversals may have implications not only for intervention planning, but also for the way that analyses are conducted to identify correlates of a screening procedure. The Specific Aims of this proposed research are: 1.) To investigate the extent of reversals of associations from bivariate to multivariable analyses of the correlates of cancer screening, by examining results several population-level surveys, by using multiple years of each survey, and by expanding the scope of screening tests to include both women and men. 2.) To investigate variables that may explain reversals of association when comparing the results of bivariate and multivariate analyses of factors associated with cancer screening tests. This two-year project will conduct secondary analyses with multiple years of data from three national- level surveys: The National Health Interview Survey, The Behavioral Risk Factor Surveillance System, and the Health Information National Trends Survey. Five screening tests will be examined: mammography, clinical breast exam, Pap testing, prostate exam, and colorectal examination. This research will extend research to date in several ways. First, it greatly expands the number of survey databases being examined. Second, it expands the number of screening tests being reviewed, and gives added attention to male screening. Third, it uses definitions of screening that are as consistent as possible across the individual surveys. Fourth, it uses a consistent set of independent variables across the surveys. Fifth, it will be possible to look for trends over time in the presence of reversals of association for repeat screening.

描述(由申请人提供)：在国家癌症预防和控制的努力中，筛查测试使用差异的证据发挥着核心作用。与非西班牙裔白人相比，传统上服务不足的种族和民族群体的筛查率较低。此外，他们往往有不成比例的更高的发病率和死亡率。对人群中处于较低筛查测试利用率“风险”的群体做出决定的过程通常始于对一组潜在协变量与筛查测试使用之间的关联进行双变量分析。下一步是进行多变量分析，最常用的是Logistic回归，以确定最终达到统计意义的变量集。这最后一组具有统计意义的变量随后被转化为需要有针对性的干预措施以减少差异的人群。这项拟议的研究解决了我们和其他人在基于人群的数据集中对癌症筛查进行的几项分析中观察到的情况。也就是说，用不同的调查数据库进行的分析发现，在多变量分析中，种族/民族和癌症筛查之间的双变量关联被“颠倒”了。也就是说，根据多变量分析的“调整”结果，在双变量(“粗率”)分析中筛查较低的组显示较高的筛查。这些逆转可能不仅对干预计划有影响，而且对进行分析以确定筛查程序的相关性的方式也有影响。这项拟议研究的具体目的是：1.)为了调查癌症筛查相关因素从双变量分析到多变量分析的逆转程度，通过检查几次人口层面调查的结果，使用每项调查的多年数据，并扩大筛查测试的范围，将女性和男性都包括在内。2.)在比较癌症筛查试验相关因素的双变量和多变量分析结果时，调查可能解释关联逆转的变量。这个为期两年的项目将利用三项国家级调查的多年数据进行二次分析：全国健康访谈调查、行为风险因素监测系统和卫生信息全国趋势调查。将检查五项筛查试验：乳房X光摄影、临床乳房检查、巴氏试验、前列腺检查和结直肠检查。这项研究将从几个方面扩展迄今的研究。首先，它极大地扩大了正在审查的调查数据库的数量。其次，它扩大了正在审查的筛查测试的数量，并增加了对男性筛查的关注。第三，它使用了在各个调查中尽可能一致的筛查定义。第四，它在所有调查中使用了一组一致的自变量。第五，在存在重复筛查的相关性逆转的情况下，将有可能寻找随时间推移的趋势。