Medicaid in Residential Care. Methods


For the descriptive analyses, we analyzed frequencies for categorical variables and means for continuous variables and tested the statistical differences using chi-square tests for categorical variables and t-tests for continuous variables. The facility-level logistic regression model estimates the effect of various characteristics (such as staffing levels) on facility participation in Medicaid. For the logistic analysis, we calculate odds ratios and corresponding 95 percent confidence intervals for each of the independent variables included in the model. Differences with probability of 0.05 or less are considered statistically significant and are reported in the text.

We follow National Center for Health Statistics' conventions by presenting only those estimates that are statistically reliable and have at least the minimum appropriate sample size. All analyses are conducted in SUDAAN® Software for Statistical Analysis of Correlated Data (Research Triangle Institute [RTI], 2008). The stratification variables of number of beds and census region, in addition to the final sample weights for the facilities and residents and the sampling design method, were incorporated into the SUDAAN procedures to account for the complex sampling design. Only weighted results are presented. With a few exceptions, differences are reported only when there is a statistically significant difference between the Medicaid and non-Medicaid facilities and residents.

We present data for facilities and residents in total as well as for Medicaid and non-Medicaid facilities and residents. We coded facilities as participating in Medicaid if respondents reported at least one resident had some or all of his or her LTC services paid by Medicaid in the 30-day period prior to the survey. Of the 2,302 RCFs in the sample, 998 facilities had at least one Medicaid resident and 1,292 had none.1 We coded residents as Medicaid residents if the facility staff reported that Medicaid paid for any of their LTC services provided at the facility in the 30-day period prior to the survey.2 Of the 8,094 RCF residents, 1,904 had their LTC paid by Medicaid in the 30-day period preceding the survey.3

Data are presented from several perspectives using different units of analysis so as to provide a full understanding of RCFs and their residents. For analyses of resident characteristics, we analyze the resident file and interpretation is straightforward. A more complex approach is required to fully understand facilities because a large number of RCFs are small (4-10 beds), but only a relatively small proportion of residents live in these facilities. More specifically, 50 percent of RCFs are small, but they serve only 10 percent of residents (Park-Lee et al., 2011). Conversely, although only about half of RCFs are larger than ten beds, they account for 90 percent of residents. Thus, a simple analysis of facilities will give disproportionate weight to the small facilities even though they serve only a small proportion of residents. To address this problem, we show facility characteristics from two perspectives. First, we analyze facility characteristics with the facility as the unit of analysis. Second, to present a perspective that more closely aligns with the number of persons served and to represent the perspective of RCF residents, we also analyze the facility characteristics at the resident level. For these analyses, we match residents with the characteristics of the facilities in which they live and present the facility characteristics with the resident as the unit of analysis. For these types of analyses, we refer to "the facilities in which residents live." This type of analysis can be thought of as facility analyses weighted by the number and type of residents.

View full report


"FacBenDif.pdf" (pdf, 1.23Mb)

Note: Documents in PDF format require the Adobe Acrobat Reader®. If you experience problems with PDF documents, please download the latest version of the Reader®