Late-Stage Breast Cancer Diagnosis and Health Care Access in Illinois∗

∗Financial support from the National Cancer Institute (NCI), National Institutes of Health, under Grant 1-R21-CA114501-01, is gratefully acknowledged. Points of view or opinions in this article are those of the authors, and do not necessarily represent the official position or policies of NCI. The research was conducted when Wang and Luo were at Northern Illinois University, and Escamilla was at University of Illinois at Urbana-Champaign.

Abstract

The variations of breast cancer mortality rates from place to place reflect both underlying differences in breast cancer prevalence and differences in diagnosis and treatment that affect the risk of death. This article examines the role of access to health care in explaining the variation of late-stage diagnosis of breast cancer. We use cancer registry data for the state of Illinois by zip code to investigate spatial variation in late diagnosis. Geographic information systems and spatial analysis methods are used to create detailed measures of spatial access to health care such as convenience of visiting primary care physicians and travel time from the nearest mammography facility. The effects of spatial access, in combination with the influences of socioeconomic factors, on late-stage breast cancer diagnosis are assessed using statistical methods. The results suggest that for breast cancer, poor geographical access to primary health care significantly increases the risk of late diagnosis for persons living outside the city of Chicago. Disadvantaged population groups including those with low income and racial and ethnic minorities tend to experience high rates of late diagnosis. In Illinois, poor spatial access to primary health care is more strongly associated with late diagnosis than is spatial access to mammography. This suggests the importance of primary care physicians as gatekeepers in early breast cancer detection.

Las variaciones de las tasas de mortalidad a causa del cáncer del seno en diferentes áreas reflejan tanto diferencias subyacentes en la predominancia de esta enfermedad como diferencias en el diagnóstico y tratamiento que afectan el riesgo de fallecimiento. En este artículo se examina el papel del acceso a la atención médica para explicar la variación de los diagnósticos de cáncer del seno en etapa avanzada. Utilizamos datos del registro de cáncer del estado de Illinois por código postal para investigar las variaciones espaciales de diagnóstico tardío. Se utilizaron sistemas de información geográfica y métodos de análisis espacial para crear medidas detalladas del acceso espacial a la atención médica, tales como la conveniencia de visitar a los médicos de atención primaria y el tiempo de viaje a la instalación de mamografías más cercana. Se evaluaron los efectos del acceso espacial, en combinación con la influencia de los factores socioeconómicos en los diagnósticos de cáncer del seno en etapa avanzada, utilizando métodos estadísticos. Los resultados sugieren que en el caso del cáncer del seno, un acceso geográfico inadecuado a la atención médica primaria aumenta significativamente el peligro de diagnóstico tardío para las personas que viven fuera de la ciudad de Chicago. Las poblaciones desfavorecidas, que incluyen aquellas de bajos ingresos y grupos minoritarios raciales y étnicos, tienden a experimentar tasas altas de diagnóstico de la enfermedad en etapa avanzada. En Illinois, el acceso espacial inadecuado a la atención médica primaria se asocia más rotundamente con el diagnóstico en etapas avanzadas que el acceso espacial a instalaciones de mamografías. Esto sugiere la importancia que tienen los médicos de consulta primaria para la detección del cáncer del seno en sus etapas tempranas.

Key Words:

• breast cancer
• late-stage cancer
• mammography
• poisson regression model
• primary care access

关键词:

• 乳腺癌
• 晚期癌症
• 乳房X光造影检查
• 泊松回归模型
• 基层医疗服务

Palabras clave:

• cáncer del seno
• cáncer en etapa avanzada
• mamografía
• modelo de regresión de Poisson
• acceso a la consulta médica primaria

Notes

∗Financial support from the National Cancer Institute (NCI), National Institutes of Health, under Grant 1-R21-CA114501-01, is gratefully acknowledged. Points of view or opinions in this article are those of the authors, and do not necessarily represent the official position or policies of NCI. The research was conducted when Wang and Luo were at Northern Illinois University, and Escamilla was at University of Illinois at Urbana-Champaign.

¹To obtain the travel time from a zip code centroid to a mammography facility, we compute (1) the travel time between a zip code centroid and its nearest node (say, A) on the network, (2) the travel time between a mammography and its nearest node (say, B) on the network, and (3) the travel time between the two nodes A and B that are on the network. The first two segments are minor and are approximated as travel time across straight-line (crows-fly) distances. See Wang (2006, 23–31).

²This variable intends to capture an important population characteristic (white vs. nonwhite), and does not include whites of Hispanic origin. The correlation coefficient between nonwhite and non-Hispanic nonwhite is as high as 0.97 in the study area. Defining the variable as non-Hispanic nonwhite does not significantly alter the results in later FA or regression on breast cancer diagnosis.

³A component with eigenvalues less than one accounts for less than an original variable's variance (= 1 after data standardization), and thus does not serve the purpose of data reduction.