Table 3.

Logistic Regression Model of Main Effects Predicting Workforce Departure

Variable	Estimated Odds Ratio Coefficient*	Estimated 95% CI
Intercept	1.17	0.28 to 4.91
Age > 60 years	0.33	0.07 to 1.54
Female sex	1.75	0.74 to 4.14
Race (1 = African American/Hispanic, 0 = white/Asian)	2.41	0.54 to 10.69
Unmarried	1.52	0.77 to 3.02
Female sex × unmarried	0.24	0.08 to 0.72
Lung cancer v colorectal cancer	2.83	1.54 to 5.19
Chemotherapy receipt	1.36	0.92 to 2.02
Education (1 = elementary, 2 = high school, 3 = college, 4 = graduate)	0.76	0.62 to 0.95
Public health insurance	1.42	0.76 to 2.66
Stage III v I or II	1.65	1.01 to 2.69
Lung cancer × stage III	2.16	0.97 to 4.82
Lung cancer × age > 60 years	0.49	0.23 to 1.04
Race × age > 60 years	1.28	0.55 to 2.98
Income.low†	0.41	0.23 to 0.70
Income > $40,000	2.12	0.62 to 7.28
Female sex × income > $40,000	0.62	0.22 to 1.75
Unmarried × income > $40,000	0.26	0.07 to 1.00
Female sex × unmarried × income > $40,000	6.55	0.98 to 43.87
Public insurance × income > $40,000	0.82	0.33 to 2.06
Age > 60 years × income.low	2.08	1.15 to 3.74
Age > 60 years × income > $40,000	0.48	0.13 to 1.81
Race × income.low	0.79	0.43 to 1.46
Race × income > $40,000	1.84	0.44 to 7.70

^*Odds ratio coefficient represents the exponentiated value of the regression coefficient in the logistic regression model. Thus, for instance, if a patient with stage III lung cancer and a patient with stage III colorectal cancer have identical values of all other covariates, then the model predicts that the odds of workforce departure of the patient with lung cancer are 2.83 × 2.16 = 6.11 times the odds for the patient with colorectal cancer.

^†Income.low is defined as follows: (1/10,000) × (X–max[0, (X–40,000)]), where X = midpoint of patient's income bracket. Thus, a patient whose income falls into the $15,000 to $20,000 bracket has an income.low value of 1.75. A patient whose income is greater than $40,000, by contrast, has an income.low value of 4.