Table 3.

Regression estimate of different variables’ impacts on subsequent revision rate, aTSA versus rTSA

	aTSA			rTSA
Variable	Estimatea, %	95% CI, %	p value	Estimatea, %	95% CI, %	p value
Age	-3	(-3 to -2)	< 0.001	-5	(-5 to -4)	< 0.001
Female	41	(28 to 56)	< 0.001	-50	(-54 to -46)	< 0.001
Non-White	-12	(-27 to 7)	0.20	-6	(-20 to 10)	0.43
Left shoulder	-5	(-14 to 4)	0.29	6	(-2 to 15)	0.17
BMI > 40 kg/m2	4	(-14 to 26)	0.69	0	(-16 to 19)	> 0.99
CHF	3	(-18 to 29)	0.82	25	(7 to 46)	< 0.001
COPD	40	(21 to 62)	< 0.001	13	(0 to 27)	0.05
Cancer	-3	(-38 to 52)	0.90	10	(-22 to 55)	0.59
Diabetes	-5	(-16 to 7)	0.36	1	(-8 to 12)	0.78
Kidney disease	6	(-23 to 45)	0.74	26	(0 to 57)	0.05
Liver disease	41	(-15 to 135)	0.18	22	(-20 to 85)	0.36
Surgeon volume group, mid	-14	(-27 to 2)	0.09	4	(-13 to 24)	0.68
Surgeon volume group, high	-27	(-39 to -13)	< 0.001	-11	(-26 to 6)	0.20
Surgeon volume group, very high	-33	(-45 to -18)	< 0.001	-26	(-39 to -9)	< 0.001
Surgeon graduation year	0	(0 to 1)	0.56	0	(-1 to 0)	0.20
Hospital volume group, mid	-12	(-27 to 5)	0.14	-4	(-18 to 14)	0.67
Hospital volume group, high	-7	(-23 to 12)	0.45	8	(-9 to 29)	0.39
Hospital volume group, very high	-27	(-42 to -9)	< 0.001	4	(-16 to 28)	0.73
Urban	15	(-5 to 40)	0.16	12	(-4 to 31)	0.14
Bed size	0	(0 to 0)	0.69	0	(0 to 0)	0.82
Teaching	14	(0 to 31)	0.05	3	(-8 to 16)	0.60

^aEstimates were made from a logistic regression at episode level with standard errors clustered at the hospital level. In addition to the listed variables, year fixed effects and state fixed effects were controlled for. Coefficients were converted as percentage changes in the odds of receiving a subsequent revision. CHF = congestive heart failure; COPD = chronic obstructive pulmonary disease.