We used data as reported in Hospital Cost Reports for 2011 and 2012 for our analysis. These reports are submitted to CMS annually by all acute-care and critical access hospitals (CAHs). Data on UCC are reported in Worksheet S-10 of Form CMS-2552-10, which was first used beginning in May 2010.55 We defined UCC as the sum of two reported items: (1) the cost of charity care provided to uninsured patients (line 23 column 1); and (2) the cost of non-Medicare bad-debt expense (line 29).56
Table 1 reports the sum of UCC over all reporting hospitals as well as its two components (charity care and bad debt) for the years 2011 and 2012 for the 25 states plus Washington DC that have expanded Medicaid and initiated enrollment as of July 2014, the 2 states that have expanded Medicaid and are now initiating enrollment (PA and NH), and the 23 non-expansion states (note that PA and NH are listed separately because, though they are expanding Medicaid, their enrollment will begin during or after the time period covered in our savings projections; CMS similarly separates out these two states on their most recent Medicaid enrollment report).57 Also note that hospital UCC in expansion states was slightly higher than in non-expansion states in 2011 and 2012:
Table 1: Amounts and Sources of Hospital Uncompensated Care as Reported in CMS Hospital Cost Reports, 2011-2012 (billions of current $)
Source: ASPE calculations from CMS Hospital Cost Reports available publicly from CMS.gov. Figures are summed over all reporting hospitals. Expansion states are defined to include AR, AZ, CA, CO, CT, DE, DC, HI, IL, IA, KY, MD, MA, MI, MN, NV, NJ, NM, NY, ND, OH, OR, RI, VT, WA, and WV. Non-expansion states are defined as all other states, with the exception of NH and PA (see next).
*: Note that NH and PA are implementing the Medicaid expansion, but were not included among actively enrolling states for this analysis because enrollment in these states is scheduled to begin either in mid-2014 or at the beginning of 2015, and thus coverage gains would not be expected to fully accrue during the time frame included in our projection (FY 2014). These states were therefore excluded from the analyses.
The numbers of individuals who were uninsured and who were covered by Medicaid in each state and in each year in 2011 and 2012 were obtained from estimates made by the U.S. Bureau of the Census.58 These estimates are based on survey data from the Current Population Survey's Annual Social and Economic Supplement. These numbers were used to model the association between numbers of uninsured and Medicaid-covered individuals in each state and the amount of UCC provided in 2011 and 2012 (see Appendix B for detailed methodology and model outputs).
We then used this model to project 2014 UCC by using projected numbers of individuals who are uninsured and covered by Medicaid at the state-level. The most recent estimate of the reduction in uninsured was published in the New England Journal of Medicine, using ASPE analyses based on the Gallup-Healthways WBI poll, and suggests that 10.3 million fewer people were uninsured as of June 2014.59 Based on Medicaid enrollment reports,60 we assumed that the number of individuals covered by Medicaid would be 7.9 million higher than it otherwise would have been as a result of coverage expansion, of which 6.9 million is in expansion states, and 1.0 million in non-expansion states. We also estimated that roughly two-thirds of the decline in the number of uninsured persons (which is composed of both new enrollment in Medicaid and new enrollment in private insurance programs, through the Marketplace or through employers) would come from states that expanded Medicaid.
There are limitations to these projections; for example, if consumers cannot pay the cost-sharing amounts required under their coverage, it is possible that hospitals will still be left with a degree of uncompensated care for these individuals. The mix of rates and plans offered and selected in any given state will impact hospital reimbursement; our models reflect overall patterns. Additionally, there is concern that the Cost Report data, because it is self-reported by hospitals, may not be of high enough fidelity to use in estimating hospital UCC. For this reason, Medicare has continued to use each hospital’s number of Medicaid days and Medicare Supplemental Security Income days as their proxy for uncompensated care rather than the Cost Report information. However, MedPAC and others have recommended that the Cost Report data be used because they believe it provides a better estimate of uncompensated care.61 We chose to use Cost Report data because it allows us to examine multiple components of uncompensated care, but we recognize the limitations inherent in hospital-reported data.
55 Please note: the Department has raised concerns regarding the accuracy of the information reported on Worksheet S-10, including that it is not yet subject to audit. These concerns have led the agency not to rely on these data for the purposes of making Medicare uncompensated care payments under section 3133 of the Affordable Care Act. We use them for the purposes of this analysis acknowledging these limitations.