Skip to main content
. 2023 Sep 12;115(10):1204–1212. doi: 10.1093/jnci/djad150

Table 4.

Effects of potential confounders on the racial disparity in COVID-19–associated hospitalizationa in Louisiana

Variables Indirect effect (95% CI) Relative effect, % (95% CI)
Mediation analysis with number of chronic diseasesb
 Age, yc -0.065 (-0.089 to -0.032) -8.5 (-12.4 to -3.74)
 Sex -0.006 (-0.020 to 0.000) -0.7 (-2.65 to 0.0)
 Insuranced 0.012 (-0.027 to 0.050) 1.5 (-3.4 to 6.5)
 Povertye -0.033 (-0.088 to 0.021) -4.3 (-11.7 to 2.6)
 Cancer type -0.010 (-0.035 to 0.006) -1.3 (-4.6 to 0.8)
 Obesity 0.023 (0.002 to 0.041) 3.0 (0.3 to 5.4)
 Number of chronic diseases 0.291 (0.229 to 0.375) 37.8 (29.1 to 48.3)
 Direct effect 0.583 (0.411 to 0.736) 75.6 (58.7 to 86.6)
 Total effect 0.771 (0.618 to 0.961)
Mediation analysis with individual chronic diseasesb
 Age, yc -0.066 (-0.088 to -0.030) -8.9 (-13.3 to -3.9)
 Sex -0.008 (-0.019 to 0.000) -1.1 (-2.8 to 0.00)
 Insuranced 0.004 (-0.037 to 0.045) 0.6 (-5.2 to 6.1)
 Povertye -0.032 (-0.089 to 0.021) -4.4 (-12.7 to 2.8)
 Cancer type -0.008 (-0.031 to 0.010) -1.1 (-4.4 to 1.4)
 Obesity 0.016 (-0.004 to 0.036) 2.1 (-0.5 to 5.2)
 Hypertension 0.071 (0.044 to 0.100) 9.6 (5.9. 14.3)
 Diabetes 0.066 (0.044 to 0.100) 8.9 (6.0 to 14.6)
 Chronic renal disease 0.054 (0.036 to 0.083) 7.3 (4.9 to 11.7)
 Chronic heart failure 0.015 (0.005 to 0.028) 2.0 (0.6 to 4.0)
 Chronic liver disease 0.004 (0.000 to 0.014) 0.5 (0.0 to 2.0)
 Hyperlipidemia 0.009 (-0.006 to 0.025) 1.2 (-1.0 to 3.4)
 Myocardial infarction 0.003 (-0.001 to 0.009) 0.4 (-0.2 to 1.3)
 Peripheral vascular disease 0.000 (-0.003 to 0.003) 0.0 (-0.4 to 0.4)
 Chronic pulmonary disease 0.007 (-0.005 to 0.016) 1.0 (-0.7 to 2.2)
 Other chronic diseasesf 0.011 (0.001 to 0.020) 1.4 (0.1 to 2.8)
 Direct effect 0.585 (0.407 to 0.740) 79.1 (64.1 to 93.0)
 Total effect 0.740 (0.560 to 0.902)
a

Please see the Method section for the definition of COVID-19–associated hospitalization. Statistically significant values are in bold font. CI = confidence interval.

b

All listed variables are included in the multivariable mediation models.

c

Age at the first positive COVID-19 test.

d

Health insurance at cancer diagnosis and around the first course of treatment.

e

Poverty at the residential census tract at the time of the first COVID-19 positive.

f

Included dementia, paralysis, ulcers, rheum, AIDS, and cerebrovascular disease.