Socio-spatial inequalities in late-stage cancer diagnosis in Illinois: spatiotemporal trends and methodological challenges

Luo, Lan

Permalink

https://hdl.handle.net/2142/26354 Copy

Description

Title

Socio-spatial inequalities in late-stage cancer diagnosis in Illinois: spatiotemporal trends and methodological challenges

Author(s)

Luo, Lan

Issue Date

2011-08-26T15:24:36Z

Director of Research (if dissertation) or Advisor (if thesis)

McLafferty, Sara L.

Doctoral Committee Chair(s)

McLafferty, Sara L.

Committee Member(s)

• Wang, Shaowen
• Ruiz, Marilyn O.
• Grigsby-Toussaint, Diana S.

Department of Study

Geography

Discipline

Geography

Degree Granting Institution

University of Illinois at Urbana-Champaign

Degree Name

Ph.D.

Degree Level

Dissertation

Keyword(s)

• late-stage colorectal cancer
• late-stage breast cancer
• social-demographic-spatial risk factors
• hierarchical logistic regression
• spatial aggregation error
• SaTScan
• Bernoulli-based spatial scan statistic
• ZIP code
• Zone Improvement Plan (ZIP)

Abstract

This dissertation examines the effects of social and spatial inequalities on late-stage diagnosis of colorectal and breast cancer, and it addresses several methodological challenges surrounding the use of ZIP codes as a study unit in analyzing late-stage cancer at diagnosis. Given that my dissertation follows the ‘three-paper’ format, the abstract section is divided into three parts to describe each paper respectively. The first paper entitled “Spatial Distribution of Late-Stage Colorectal Cancer in Illinois from 1988 to 2002: Associations with Social-Spatial Covariates”, examines spatial patterns of late-stage colorectal cancer diagnosis over time in Illinois during a period of increasing screening, and it analyzes the varying associations between social, demographic and spatial risk factors and late-stage colorectal cancer diagnosis within the same period. The Bernoulli-based spatial scan statistic was used to detect clusters of late-to-early stage cancer ratios at the ZIP code level in Illinois during two periods: 1988 to 1992, and 1998 to 2002. Then the whole state was divided into three study region: Chicago city, Chicago suburbs, and other areas. For each region in each time period, hierarchical logistic regression models were estimated to assess the associations between demographic, social and spatial factors and late-stage colorectal cancer risk. ZIP code level risk factors include three indicators of socio-economic status and the shortest travel time to the nearest colonoscopy facility and individual-level factors including age, race, and gender. The socio-economic indicators were created using factor analysis. The results show some changes over time in the spatial distribution of late-stage colorectal cancer and the impacts of risk factors at the ZIP code and individual levels. Specifically, results of the Bernoulli-based spatial scan statistic find statistically significant clusters of late-stage colorectal cancer in the Chicago metropolitan area and rural region in southern Illinois in the period of 1988 to 1992. In the later time period, the cluster outcomes were no longer statistically significant. The change indicates that late-stage risk of colorectal cancer has become more evenly distributed in Illinois over time. In terms of the hierarchical logistic regression results, both individual-level demographic factors and zip-code level covariates present variously important impacts on the risk of the late-stage colorectal cancer diagnosis in different study regions in the two time periods. The risk of late-stage diagnosis is higher among younger colorectal cancer patients. Gender has contradictory impacts on risk in Chicago city and its suburb. The shortest travel time to the nearest cancer screening providers is positively associated with late-stage diagnosis risk outside the Chicago region, suggesting that spatial access to screening services may be an important barrier to early detection in rural areas of the state. One socio-economic status indicator, Minority Disparities, demonstrated a significantly positive relationship with late-stage diagnosis risk outside the Chicago region. Similar to the effects of gender, Factor 3 (Cultural-Language Barriers) also had contradictory effects in Chicago city and suburbs. Overall, the results showed no clear trends over time in the effects of various factors on late-stage risk, and few strong and statistically significant results. The inconsistent findings suggest the need for more detailed and localized information. The second paper is titled “Analyzing Spatial Aggregation Error in Statistical Models of Late-Stage Cancer Risk: A Monte Carlo Simulation Approach”. This paper examines the effect of spatial aggregation error on statistical estimates of the association between spatial access to health care and late-stage cancer. Monte Carlo simulation was used to disaggregate breast cancer cases for two Illinois counties from ZIP codes to census blocks in proportion to the age-race composition of the block population. After the disaggregation, a hierarchical logistic model was estimated examining the relationship between late-stage breast cancer and risk factors including travel distance to mammography, at both the ZIP code and census block levels. Model coefficients were compared between the two levels to assess the impact of spatial aggregation error. Spatial aggregation error is found to influence the coefficients of regression-type models at the ZIP code level, and this impact is highly dependent on the study area. In one study area (Kane County), block-level coefficients were very similar to those estimated on the basis of ZIP code data; whereas in the other study area (Peoria County), the two sets of coefficients differed substantially raising the possibility of drawing inaccurate inferences about the association between distance to mammography and late-stage cancer risk. The paper reveals that spatial aggregation error can significantly affect the coefficient values in statistical models of the association between cancer outcomes and spatial and non-spatial variables and thus affect inferences drawn from these models. Relying on data at the ZIP code level may lead to inaccurate findings on health risk factors, and the effects are likely to vary from one study area to another. The third paper, titled “The Impact of Spatial Aggregation Error on Spatial Scan Analysis: A Case Study of Colorectal Cancer,” aims to examine the effect of spatial aggregation error on results of the spatial scan statistic by geographically and statistically comparing results at the ZIP code level and three reference (census tract, census block group and census block) levels. Data on colorectal cancer cases in Cook County, IL for a 5-year interval (1998-2002) were used. The Monte Carlo simulation approach from the second paper was applied to disaggregate the cancer data from the ZIP code level to each reference level. The Bernoulli-based spatial scan statistic was implemented in SaTScan to detect primary clusters based on cancer data at the four levels. An interactive procedure involving SAS and Java programming, was designed to automatically run SaTScan hundreds of times. Characteristics of clusters at each reference level were compared to those of the ZIP code level cluster to observe differences related to spatial aggregation. The comparison reveals that the ZIP code level spatial scan statistic can generate reliable clusters at the global level in areas with a large number of cases. Nonetheless, the ZIP code analysis sometimes fails to detect clusters in areas with a lower density of cases. Spatial aggregation error is minimized in areas with sizeable numbers of cases. In the absence of cancer data at a lower level, the ZIP code level data can be used effectively to implement the spatial scan statistic and identify large and dominant clusters. However, smaller clusters located in areas with a relatively low density of cases may be missed. Given that this study focused on a highly urbanized and populated area, future research should assess the influence of spatial aggregation error on spatial scan analysis in suburban and rural regions.

Graduation Semester

2011-08

Permalink

http://hdl.handle.net/2142/26354

Copyright and License Information

© 2011 by Lan Luo, All rights reserved

Owning Collections