This analysis aims to estimate measures of association and assess for confounding and effect modification for the relationship between race/ethnicity and unplanned hospitalizations in 2009 among Illinois Medicare beneficiaries aged 65 and older with multiple chronic conditions. This time, regression models will be used to generate estimates.

1. Using the appropriate crude regression models, estimate crude RRs and RDs with 95% CIs for the relationship between race/ethnicity and unplanned hospitalizations among those with multiple chronic conditions. (Use the four category race/ethnicity variable with white, non-Hispanic as the common referent group). Present your results in a table(s), with interpretive summary text below. In the summary text, do not repeat the same quantitative estimates shown in table(s).

Table 1: Crude Relative Risks and Risk Differences for the Association between Race/Ethnicity and Unplanned Hospitalizations in 2009 among Medicare Beneficiaries with Multiple Chronic Conditions (n = 30,441)

Race/Ethnicity	(95% CI)	Crude RD (95% CI)	(95% CI)	Crude RR (95% CI)
Black, non-Hispanic
Hispanic
Other
White, non-Hispanic		Ref		Ref

2. Restrict the sample to only non-Hispanic Blacks and Whites. Using the appropriate multivariable regression models, assess for additive and multiplicative effect modification by the following variables of the relationship between race/ethnicity and unplanned hospitalizations among those with multiple chronic conditions. Present your results in tables, with interpretive summary text below it (as defined above in Q1).

Sex

Age (in three categories: 65-74, 75-84, 85-99)

Table 3a: Stratified Analysis of the Association between Race/Ethnicity (Black versus White) and Unplanned Hospitalizations in 2009, by Sex and Age Group, among Medicare Beneficiaries (65+) with Multiple Chronic Conditions
	Binomial	Log-binomial
Covariate	(95% CI)	Stratum-Specific RD (95% CI)	(95% CI)	Stratum-Specific RR (95% CI)
Sex
Female
Male
Age Group
65-74
75-84
85-99