Poverty and Cancer Disparities in Ohio

John Kollman, MS; Holly L. Sobotka, MS


Prev Chronic Dis. 2018;15(12):E152 

In This Article


We studied populations in the 12 poorest and the 10 most affluent counties in Ohio. These counties were identified by the percentage of the county population living in poverty in the 2011–2015 American Community Survey.[6] The US Census Bureau uses a set of income thresholds that vary by family size and composition to determine who is living in poverty. In 2015, for example, the minimum family income threshold for poverty was $12,331 for a person younger than 65 and $24,036 for a family of 4 (2 adults and 2 children aged <18 y).[7] Twelve counties in Ohio had poverty rates of 20% or more in 2011–2015 (Adams, Ashtabula, Athens, Gallia, Highland, Jackson, Lucas, Meigs, Morgan, Pike, Scioto, and Vinton) and were defined as the poorest counties.[8] All 12 counties, except Lucas County, are in the Appalachian region. The population in this group of counties was 904,834 in 2010, 7.8% of the Ohio population, and approximately 82.6% non-Hispanic white and 10.2% non-Hispanic black. Ten counties had poverty rates of less than 10% in 2011–2015 (Auglaize, Delaware, Geauga, Lake, Madison, Medina, Mercer, Putnam, Union, and Warren) and were defined as the most affluent counties.[8] Most of these counties are adjacent to metropolitan areas. The population in this group of counties was 1,099,666 in 2010, 9.5% of the Ohio population, and approximately 91.8% were non-Hispanic white and 2.5% were non-Hispanic black.

Data Sources

We obtained data on cancer incidence, stage at diagnosis, insurance status, and summary treatment status from the Ohio Cancer Incidence Surveillance System (OCISS), the central cancer registry for Ohio.[9] We coded cancer cases to the International Classification of Diseases for Oncology, Third Edition (ICD-O-3), and we categorized data on 23 sites and types of cancer according to the conventions of the National Cancer Institute Surveillance, Epidemiology, and End Results (SEER) Program.[10,11] We tabulated data on incidence from cancer cases diagnosed from January 1, 2011, through December 31, 2015, and accessed through OCISS in January 2018. We calculated cancer incidence rates by counting only invasive cancer cases; we excluded in situ tumors except for cases of in situ bladder cancer. We selected cancers for which the mortality rate in the poorest counties was significantly higher than the mortality rate in the most affluent counties; for these cancers, we calculated the percentage of cases diagnosed at a late stage in the poorest counties and in the most affluent counties. We classified stage at diagnosis by using SEER Summary Stage 2000 and the following categories: early stage (in situ and local), late stage (regional and distant), and unstaged/missing (insufficient information was available to determine the stage of disease at the time of diagnosis or the case was reported without information on stage).[12] Insurance status was based on data for primary payer at diagnosis ("Primary Payer at DX"), and treatment status was based on a summary measure of all treatment modalities ("Rx Summ-Treatment Status"), categorized into the following groups: no treatment given, treatment given, active surveillance (watchful waiting), and unknown.

We obtained cancer mortality data from the Ohio Bureau of Vital Statistics at the Ohio Department of Health.[13] These data, which indicate underlying cause of death, were coded by using the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10)[14] and tabulated for 23 cancers according to methods outlined in the SEER Program's Cause of Death Recode.[15]

We used data from the Ohio Behavioral Risk Factor Surveillance System (BRFSS) for 2011–2015 to analyze 5 risk factors:[16] current smoking (smoking ≥100 cigarettes in lifetime and currently smoke cigarettes every day or some days), obesity (a body mass index of ≥30.0 [weight in kilograms divided by height in meters squared]), physical inactivity (no physical activity or exercise other than regular job during the past 30 days), heavy drinking (men having >2 drinks per day and women having >1 drink per day), and a Papanicolaou (Pap) test in the previous 3 years among women aged 21 to 65 (data on this variable not collected in 2013 BRFSS). We determined the prevalence of each risk factor in the poorest counties and in the most affluent counties. The BRFSS is an annual telephone survey conducted by the Ohio Department of Health and supported by the Centers for Disease Control and Prevention (CDC) and is the primary source of health information on Ohio residents aged 18 years or older.

Statistical Analyses

We tabulated data on incidence and mortality rates per 100,000 people and age-adjusted these data to the 2000 US standard population by using 19 five-year age groups (<1 y, 1–4 y, 5–9 y, … ≥85 y).[17] We calculated rate ratios (rate among the poorest group divided by the rate among the most affluent group) for each cancer site, along with the 95% confidence intervals of the rate ratios, which were used to determine significant differences. If the 95% confidence interval did not contain 1.0, we concluded that a significant difference between the 2 groups of counties existed at the .05 significance level. We analyzed data by using the Ohio Public Health Data Warehouse and SAS version 9.4 for Windows (SAS Institute Inc). To test for differences in stage at diagnosis, insurance status, treatment status, and risk factors, we used the 2-proportion z test at the .05 significance level.