Skip to main content

Table 7 Differences in odds ratios ( OR EG ) of MMSE score and self-reported health by social/behavioral exposure status(E) and rs1042718 minor allele carrier status(G)

From: Interactions between Social/ behavioral factors and ADRB2 genotypes may be associated with health at advanced ages in China

Dependent variable

MMSE score

Self-reported good health

(Model I-A2,ref. Table 2)

(Model II-A2,ref. Table 2)

G: genotypic status

E: regular exercise

E: regular exercise

No

Yes

% difference

No

Yes

% difference

(E = 0)

(E = 1)

E = 1 vs. 0

(E = 0)

(E = 1)

E = 1 vs. 0

OR E0 of Non-carrier rs1042718 (G = 0)

1.00

1.21

21.0%

1.00

1.05

5.0%

OR E1 of Carrier rs1042718 (G = 1)

0.83

2.27***

173.5%

0.91

2.56***

181.4%

GxE interaction

2.26** (1.10 ~ 4.62)

2.68** (1.23 ~ 5.85)

  1. Notes: (1) The estimates of OR 10 , OR 01 and the “odds ratio of GxE interactions (ORIT)” are taken from models I-A2 and II-A2 of logistic regression analysis presented in Table 2. (2) *p < 0.1, **p < 0.05, ***p < 0.01. (3) Because ORIT = OR 11/(OR 10OR 01) [48, 55], OR 11 = OR 10 OR0 01 ORIT. However, we do not know the statistically significant level (i.e., p value) of the OR 11 estimated based on OR 10 , OR 01 and ORIT normally produced by the statistical software. Thus, we alternatively estimated OR 10 , OR 01 and OR 11 and their p values by setting up three exclusive dummy variables V10, V01 and V11 (without interaction term) in the regression equation: V10 =1 if E = 1 and G = 0; V01 = 1 if E = 0 and G = 1; and V11 = 1 if E = 1 and G = 1, while considering V00 (E = 0 and G = 0) as the reference group. In such alternative regression, the OR 11 , OR 10 , OR 01 and their significant levels are estimated and the ORIT can be calculated based on estimates of the OR 10 , OR 01 and OR 11 . Note that the OR 10 , OR 01 , OR 11 and ORIT estimated in such an alternative regression are exactly the same as those estimated in the normal procedure (with two dummy variables and one interaction term) adopted by the statistical software.