Date of Award

Fall 12-11-2017

Degree Type


Degree Name

Master of Public Health (MPH)


Public Health

First Advisor

Lee Rivers Mobley

Second Advisor

Terry Pechacek


INTRODUCTION: Several studies have looked at the various factors that could explain the disparity in cancer diagnosis outcome including that for colorectal cancer between minorities and Whites. Studies have also shown that when it comes to insurance status Blacks and Hispanics are more likely to have higher uninsured rates. 3,1 With the implementation of the Affordable Care Act (ACA) in 2010 there has been a decline in the uninsured rate, with the rate of decline differing by state.

AIM: This thesis examines whether the observed decline in CRC mortality rates observed nationally was comparable for minorities and whites in the 25 states with data for both groups. The questions that will guide the thesis:

1)What association can be found between the expanded health care coverage for colorectal cancer screening following the ACA (2010) and colorectal cancer mortality? Does the disparity pattern in CRC mortality for minorities as compared to whites change after 2010?

METHODS: This thesis uses state-level secondary data on colorectal cancer mortality from the period of 1990 to 2013 for 25 states in the United States which had data for White, Black and Hispanic populations. Data were provided each year by the National Vital Statistics System at the National Center for Health Statistics of the Centers for Disease Control and Prevention and the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) Program. Access to care as measured by the uninsured rate for the 25 states derives from various sources and was not available for each year. Two time periods were identified (early and later period) and mean mortality and insurance rates were calculated over these two intervals. Regression analysis was performed on the two-time intervals of data stacked over two periods (early and later periods), with a binary time indicator variable to differentiate the early and late periods. The regression included the CRC mortality rate for whites and the CRC mortality rates for Blacks and Hispanics for each state in each period (4 observations per state) as the dependent variable, a total of 100 observations. Regressors included the uninsured rate for all populations in each time interval for each state (2 unique observations per state), an indicator of minority versus white race/ethnicity (50 observations=1, rest=0), and a time indicator (later time=1, earlier=0). The time and minority indicators were interacted to test whether minority CRC mortality rates fell after 2010 as compared to Whites, holding constant access to care (insurance status), the overall trend in CRC mortality, and the overall minority effect associated with CRC. Defining the time intervals as periods pre- and post- 2010, the main hypothesis was to assess whether these CRC mortality outcomes changed significantly pre-post the 2010 implementation of the ACA, and whether there were differential effects over time for minorities versus whites.

RESULTS: Over both periods together, a higher state uninsured rate was associated with a higher state CRC mortality rate, but this was not statistically significant (p = 0.294). Over both periods together, the CRC mortality rate for Hispanics and Blacks was higher than that for whites, by 7.94 deaths per 100,000, and this was statistically significant. There was a reduction in the average CRC mortality rate by 4.73 deaths per 100,000 over the time periods for everybody, and this was statistically significant (p = 0.000). However, in the later period, the Black and Hispanic CRC mortality rate fell by an additional 2.073 deaths per 100,000 as compared to the overall decline and this was significant (p = 0.035). Thus, there was a larger decline in the CRC mortality rate for Blacks and Hispanics compared to whites over the two periods, which suggests that the passage of the ACA may have reduced disparities in CRC mortality among minorities as compared to whites.

DISCUSSION: Over the twenty-four-year period from 1990 to 2013, there was a steady decline in the CRC mortality rate for Blacks, Hispanics, and Whites. This thesis shows that the decline in mortality rate is weakly associated with a decline in uninsured rate, when comparing the average rates for both in 2009 and 2013, and not adjusting statistically for other factors. However, the association between the uninsured rate and CRC mortality rate by race/ethnicity group remains unknown, because data on uninsured by each group was not available for all 25 states in the two-time periods pre- and post- 2010. Also, not known is whether there was a change over time in the association between a state’s uninsured rates and the CRC mortality rates, which we could have ascertained using another time interaction with uninsured rates had these been available by group. Holding overall uninsured rate constant statistically in the model was necessary to disentangle the reduction in the minority CRC mortality rate as compared to Whites by the later period. Failure to hold uninsured rates constant statistically could have resulted in omitted variables bias on the minority effects, assuming that minority uninsured rates differed from whites. Future analyses should attempt to determine what if any insurance effects were present for the different groups. A three-way interaction between uninsured by group and time could also determine whether the insurance effects were stronger for one group than for another, which would have interesting policy implications.