Early Medicaid Expansion and Cancer Mortality

Abstract

Background: Although Medicaid expansion is associated with decreased uninsured rates and earlier cancer diagnoses, no study has demonstrated an association between Medicaid expansion and cancer mortality. Our primary objective was to quantify the relationship between early Medicaid expansion and changes in cancer mortality rates.

Methods: We obtained county-level data from the National Center for Health Statistics for adults aged 20-64 years who died from cancer from 2007 to 2009 (preexpansion) and 2012 to 2016 (postexpansion). We compared changes in cancer mortality rates in early Medicaid expansion states (CA, CT, DC, MN, NJ, and WA) vs nonexpansion states through a difference-in-differences analysis using hierarchical Bayesian regression. An exploratory analysis of cancer mortality changes associated with the larger-scale 2014 Medicaid expansions was also performed.

Results: In adjusted difference-in-differences analyses, we observed a statistically significant decrease of 3.07 (95% credible interval = 2.19 to 3.95) cancer deaths per 100 000 in early expansion vs nonexpansion states, which translates to an estimated decrease of 5276 cancer deaths in the early expansion states during the study period. Expansion-associated decreases in cancer mortality were observed for pancreatic cancer. Exploratory analyses of the 2014 Medicaid expansions showed a decrease in pancreatic cancer mortality (-0.18 deaths per 100 000, 95% confidence interval = -0.32 to -0.05) in states that expanded Medicaid by 2014 compared with nonexpansion states.

Conclusions: Early Medicaid expansion was associated with reduced cancer mortality rates, especially for pancreatic cancer, a cancer with short median survival where changes in prognosis would be most visible with limited follow-up.

Figure 1.

Temporal trends in overall (A) and pancreatic (B) cancer mortality rates by state early Medicaid expansion status. The year-to-year change in overall and pancreatic cancer mortality rates was similar in early Medicaid expansion and nonexpansion states from 2002-2009 (with the exception of 2005 in the overall cancer mortality rate analysis). However, the trends diverged beginning in 2010, when the first early Medicaid expansions occurred. The dashed line for “trend comparison,” for easier visual comparison of temporal trends, is equal to the trends of the nonexpansion states translated down such that the comparison mortality rates at the end of the preexpansion study period (2009) are equal to the rate in the early expansion group. Mortality rate is the age-adjusted deaths per 100 000 population. Error bars denote 95% confidence intervals.

Figure 2.

Temporal trends in the uninsured rate by state Medicaid expansion status. There was generally a slowly decreasing uninsured rate in early Medicaid expansion states until 2013, with a smaller gap of the uninsured rate between early and not early expansion states in 2013 compared with the gap in 2006. The elevated uninsured rates in 2009 and 2010 may be secondary to the great recession. Beginning in 2014, with the initiation of the larger scale Medicaid expansions, early Medicaid expansion states had a larger decrease in the uninsured rate than the other groups of states, including the group of all states that expanded in 2014.

Figure 3.

Temporal trends in overall (A) and pancreatic (B) cancer mortality rates by 2014 state Medicaid expansion status. The trends in cancer mortality rates for all cancers combined changed similarly over time for Medicaid expansion and nonexpansion states, with the exception of 2009. The trends in pancreatic cancer mortality rates were similar in Medicaid expansion and nonexpansion states until around 2013, when the trends began to diverge, consistent with the timing of the 2014 Medicaid expansions. The dashed line for “trend comparison,” for easier visual comparison of temporal trends, is equal to the trends of the nonexpansion states translated up or down such that the comparison mortality rates at the end of the preexpansion study period (2013) are equal to the rate in the 2014 expansion group. Mortality rate is the age-adjusted deaths per 100 000 population. Error bars denote 95% confidence intervals.