Skip to main content

Table 3 Statistical association of dual eligibility and health care utilization

From: Health care utilization among Medicare-Medicaid dual eligibles: a count data analysis

  Office-based physician visits (1) Hospital inpatient nights (2) Home health provider days (3) Total dental visits (4)
Demographics     
Age 0.997 (0.001)*** 0.997 (0.003) 1.003 (0.003) 0.991 (0.002)***
Male 0.993 (0.022) 1.380 (0.110)*** 1.097 (0.105) 0.867 (0.044)***
Married 1.063 (0.023)** 0.768 (0.062)*** 0.616 (0.066)*** 1.178 (0.060)***
Less than High School 0.891 (0.020)*** 0.979 (0.076) 0.969 (0.094) 0.778 (0.054)***
Rural 0.970 (0.023) 0.984 (0.077) 1.439 (0.133)*** 0.759 (0.043)***
Region     
Northeast 1.083 (0.031)*** 0.961 (0.100) 1.242 (0.150)* 1.357 (0.088)***
Midwest 0.962 (0.027) 0.974 (0.091) 0.807 (0.094)* 1.149 (0.074)**
West 0.901 (0.025)*** 0.715 (0.070)*** 0.799 (0.100)* 1.162 (0.075)**
Medical condition     
Poor Health 1.168 (0.026)*** 1.292 (0.093)*** 1.424 (0.125)*** 0.757 (0.042)***
Co-morbidities 1.140 (0.003)*** 1.060 (0.009)*** 1.037 (0.010)*** 1.081 (0.007)***
Chronic Disease 1.147 (0.023)*** 1.120 (0.079) 0.850 (0.070)** 0.909 (0.043)**
Socioeconomic status     
Poverty 1.004 (0.025) 0.897 (0.074) 1.020 (0.092) 0.894 (0.075)
Still Work 0.908 (0.032)*** 0.562 (0.079)*** 1.602 (0.403)* 0.833 (0.062)**
Race & dual eligibility     
Asian 0.886 (0.087) 0.899 (0.458) 0.044 (0.034)*** 0.510 (0.117)***
Hispanic 0.997 (0.036) 1.009 (0.132) 1.790 (0.313)*** 1.082 (0.107)
Afro-American 0.959 (0.033) 1.063 (0.130) 2.216 (0.346)*** 0.622 (0.072)***
Dual Eligibility 1.016 (0.037) 1.003 (0.112) 1.958 (0.234)*** 0.708 (0.085)***
Dual*Asian 1.235 (0.184) 0.334 (0.238) 16.879 (15.54)*** 1.621 (0.564)*
Dual*Hispanic 1.106 (0.067)* 1.193 (0.237) 0.943 (0.216) 1.253 (0.211)
Dual*Afro-American 1.170 (0.072)*** 1.132 (0.226) 0.643 (0.132)** 1.926 (0.329)***
Inflation model = logit
Log likelihood -22573.77 -8603.889 -7757.978 -8696.873
Constant -4.012 (0.912)*** 1.480 (0.138)*** 3.071 (0.201)*** -3.211 (0.840)***
Cancer -22.16 (16771.5) -0.500 (0.088)*** -0.418 (0.088)*** -0.235 (0.134)*
Diabetes -20.73 (7065.7) -0.518 (0.075)*** -0.456 (0.078)*** 0.485 (0.102)***
Hypertension -11.54 (54.42) -0.158 (0.063)*** -0.029 (0.072) 0.043 (0.092)
Stroke -1.228 (0.942) -1.226 (0.129)*** -1.037 (0.115)*** 0.404 (0.165)**
Depression -1.314 (0.590)** -0.269 (0.072)*** -0.352 (0.074)*** -0.136 (0.100)
Less than High School 0.769 (0.235)*** -0.094 (0.072) -0.133 (0.081)* 1.084 (0.138)***
Poverty -0.260 (0.266) -0.314 (0.096)*** -0.167 (0.101)* 0.037 (0.146)
Still Work 0.404 (0.321) 0.084 (0.126) 0.943 (0.191)*** -0.742 (0.314)**
Asian 0.144 (0.945) 1.007 (0.340)*** 1.130 (0.364)*** -0.391 (0.503)
Hispanic 0.130 (0.318) 0.152 (0.096) 0.274 (0.106)*** 0.160 (0.148)
Afro-American 0.917 (0.255)*** 0.321 (0.093)*** -0.026 (0.096) 0.030 (0.207)
Dual eligibility -1.487 (0.761)* -0.285 (0.082)*** -1.066 (0.082)*** 0.241 (0.145)*
Income0 1.298 (0.987) 0.300 (0.189) 0.325 (0.272) 2.725 (0.813)***
Income1 1.614 (0.924)* 0.159 (0.146) -0.511 (0.214)** 2.420 (0.795)***
Income2 1.194 (0.910) -0.035 (0.136) -0.620 (0.206)*** 2.280 (0.790)***
Incoem3 0.861 (0.949) 0.211 (0.150) -0.300 (0.225) 1.724 (0.780)**
LR Chi2 (20) 2,766.11 136.25 199.10 333.40
Prob> Chi2 (20) 0.000 0.000 0.000 0.000
Number of obs. 8,260 8,255 8,185 8,260
Nonzero obs. 7,094 1,501 942 2,395
Zero obs. 1,166 6,754 7,243 5,865
alpha 0.572 (0.014) 1.906 (0.170) 1.522 (0.090) 1.377 (0.119)
Likelihood-ratio (LR) test of alpha = 0: Chibar2(01) = 1.7e+04 Prob>=Chibar2 = 0.000 Chibar2(01) = 1.3e+04; Prob>=Chibar2 = 0.000 Chibar2(01) = 8.3e+04; Prob>=Chibar2 = 0.000 Chibar2(01) = 1457.18; Prob>=Chibar2 = 0.000
Vuong test of ZINB vs. standard NB: Z = 5.40; Prob>z = 0.000 Z = 7.76; Prob>z = 0.000 Z = 7.95; Prob>z = 0.000 Z = 8.21; Prob>z = 0.000
  1. Notes: Zero-Inflated Negative Binomial (ZINB) regressions are performed. Incidence rate ratios and correlation coefficient estimates are presented for negative binomial (NB) and logit regressions, respectively. Corresponding exponential-scale and logit-scale standard errors reported in brackets. Categorized income variables are constructed by dividing family income by the applicable poverty line (based on family size and composition), with the resulting percentages grouped into 5 categories: negative or poor ("Income0", less than 100%), near poor ("Income1", 100% to less than 125%), low income ("Income2", 125% to less than 200%), middle income ("Income3", 200% to less than 400%), and high income ("Income4", greater than or equal to 400%). Two-tailed t-tests and one-tailed chi-square tests are applied to the estimated coefficients and model specification statistics (i.e., LR and Vuong). Asterisks *, **, and *** indicate statistical significance at 10%, 5%, and 1% levels, respectively.