Predictors of late-stage cancer diagnoses: multilevel factors and spatial interactions across the entire US cancer population

Problem: Breast cancer is the most common cancer in women, and colorectal cancer is the third most common cancer in both men and women. Currently, population cancer screening rates are suboptimal and over 40% of these cancers are diagnosed at an advanced stage, which results in higher morbidity and mortality than would obtain with optimal cancer screening utilization.

Purpose: To provide information that might reduce the prevalence of late-stage cancer at first diagnosis, we use spatial analysis to answer questions related to Why disparities in late-stage cancer diagnoses are observed.  In our previous study (Mobley et al, 2012) we found that characteristics of state insurance regulatory environments were significantly associated with late-stage cancer diagnosis outcomes in 11 states with SEER cancer registries.  The purpose here is to expand the same analysis to cover all of the US and to examine a more comprehensive set of state insurance regulations, and a vastly more comprehensive set of local community contexts.

Study Sample: To model disparities in the geographic distributions of rates of late-stage BC and CRC cancer diagnoses, we use the cancer population for the entire United States during 2000-2010 included in the National Program of Cancer Registries. We have been granted special permission to conduct this research inside the secure federal NCHS/Census Research Data Center labs.

Methods: We use a multilevel modeling approach that includes person, county, and state level characteristics to examine Why there are geographic disparities in late-stage cancer diagnoses.  We focus on state level characteristics reflecting regulations and competitiveness of state health insurance markets, managed care penetration, prevalence of supplemental insurance (e.g.MediGap), and maturity of cancer control planning efforts.  We use these along with county-level compositional and contextual characteristics to characterize supply and demand conditions, and include cross-level interactions to help model the spatial heterogeneity that is observed from place to place.  We model the county cancer population rates of late-stage BC separately from the rates of late-stage CRC diagnoses.

Conclusions:  There is considerable heterogeneity across states, and across counties within states, in the rates of late-stage cancer diagnosed for these two cancer types.  The health insurance environment varies considerably across states, and variation in health insurance regulation and mandates is statistically and significantly associated with cancer outcomes. 

References

Mobley, L., Kuo, T., Watson, L., and Brown, G., “Geographic disparities in late-stage cancer diagnosis: Multilevel factors and spatial interactions”, Health&Place, v 18 (2012): 978-990.