We did three analyses in the paper examining the association of Oxo with (1) BDNF and race; (2) BDNF, race, and CRP covarying main effects for IL18, TNFalpha, and RAGE; and, (3) BDNF, race, and pulse pressure
The reviewer suggested (in essence) that menopausal status might confound these relationships.
In these data, there are sufficient number of pre-menopausal and post-menopausal women to add menopausal status to the analyses.
Here's the distribution of menopausal status by CRPgroup:
CRPgroup
Menopause Lo Mid Hi
Pre-menopause 13 10 11
Post-menopause 24 28 24
Although there are significant differences in Oxo associated with menopausal status, the interaction between BDNF and race is still significant. This suggests that although menopausal is a significant effect, it does not confound the relationship between Oxo and BDNF by race.
Simultaneous Tests for General Linear Hypotheses
Fit: lmer(formula = Oxo ~ Menopause + B * R + (CRPgroup | group),
data = oxoMen, control = lmerControl(check.nobs.vs.nRE = "warning",
check.scaleX = "ignore"))
Linear Hypotheses:
Estimate Std. Error z value Pr(>|z|)
(Intercept) == 0 0.1585950076987 0.0087301704448 18.16631 < 2e-16
MenopausePost-menopause == 0 0.0192184179337 0.0077825663682 2.46942 0.013533
B == 0 -0.0000001184363 0.0000002012292 -0.58856 0.556154
RAfrAm == 0 -0.0168803968672 0.0097112628327 -1.73823 0.082171
B:RAfrAm == 0 0.0000006815705 0.0000002710439 2.51461 0.011916
(Univariate p values reported)
Menopause is not associated significantly (though it's close) with Oxo nor does it confound the relationships among BDNF, race, and CRP after adjusting for IL18, TNFalpha, and RAGE.
Simultaneous Tests for General Linear Hypotheses
Fit: lmer(formula = Oxo ~ Menopause + C * B * R + hIL18 + hTNFa +
hRAGE + (CRPgroup | group), data = oxoMen, control = lmerControl(check.nobs.vs.nRE = "warning",
check.scaleX = "ignore"))
Linear Hypotheses:
Estimate Std. Error z value Pr(>|z|)
(Intercept) == 0 0.1478258905934 0.0167218435544 8.84029 < 2e-16
MenopausePost-menopause == 0 0.0174323931632 0.0089544859681 1.94678 0.051561
C == 0 0.0029166300029 0.0051917043940 0.56179 0.574261
B == 0 -0.0000001741240 0.0000003182321 -0.54716 0.584269
RAfrAm == 0 -0.0298829643475 0.0169553808404 -1.76245 0.077994
hIL18 == 0 0.0000356306758 0.0000420459121 0.84742 0.396759
hTNFa == 0 0.0000449985795 0.0000449525285 1.00102 0.316815
hRAGE == 0 -0.0000037286984 0.0000080697631 -0.46206 0.644040
C:B == 0 0.0000001320872 0.0000001401075 0.94276 0.345806
C:RAfrAm == 0 0.0098498287721 0.0074819185698 1.31648 0.188012
B:RAfrAm == 0 0.0000014828502 0.0000004751346 3.12091 0.001803
C:B:RAfrAm == 0 -0.0000004956073 0.0000002027219 -2.44476 0.014495
(Univariate p values reported)
Menopause is not associated significantly with Oxo nor does it confound the relationships among BDNF, race, and pulse pressure.
Simultaneous Tests for General Linear Hypotheses
Fit: lmer(formula = Oxo ~ Menopause + B * R + PP + (CRPgroup | group),
data = oxoMen, control = lmerControl(check.nobs.vs.nRE = "warning",
check.scaleX = "ignore"))
Linear Hypotheses:
Estimate Std. Error z value Pr(>|z|)
(Intercept) == 0 0.1267507159710 0.0129243044504 9.80716 < 2e-16
MenopausePost-menopause == 0 0.0141101439736 0.0077892554325 1.81149 0.070065
B == 0 -0.0000002441967 0.0000001948852 -1.25303 0.210195
RAfrAm == 0 -0.0220885998504 0.0096382019447 -2.29178 0.021919
PP == 0 0.0007848403590 0.0002433661699 3.22494 0.001260
B:RAfrAm == 0 0.0000006411774 0.0000002614462 2.45243 0.014190
(Univariate p values reported)
I think we can respond to this reviewer by saying that we examined the relationships with menopausal status and found no evidence that it confounded any of the results we reported.
If we want to appear especially cooperative we could add menopausal status to the methods (that we examined it), the results (no evidence for confounding), and the discussion (we don't find evidence to support the reviewer's point about estrogen, BDNF, and angiogenesis).