Table 3.
Rate ratio of myocardial infarction and 95% CI by quartile of adrenal hormone concentrations among predominantly postmenopausal women in the Nurses’ Health Study.
Hormone concentration quartile categories |
P value for linear trend |
||||
---|---|---|---|---|---|
1 | 2 | 3 | 4 | ||
DHEA, nmol/L | <9.60 | 9.61–16.55 | 16.56–26.33 | ≥26.34 | |
Cases/controls | 58/114 | 53/112 | 56/110 | 62/113 | |
Model a | 1.0 (referent) | 0.97 (0.77–1.23) | 1.01 (0.79–1.28) | 1.05 (0.83–1.34) | 0.06 |
Model b | 1.0 (referent) | 0.92 (0.71–1.21) | 1.09 (0.82–1.44) | 1.19 (0.90–1.57) | <0.01 |
Model c | 1.0 (referent) | 0.93 (0.68–1.26) | 1.19 (0.87–1.63) | 1.27 (0.92–1.74) | <0.01 |
Model d | 1.0 (referent) | 0.92 (0.66–1.27) | 1.22 (0.88–1.68) | 1.23 (0.89–1.72) | 0.08 |
DHEA-S, nmol/L | <709 | 710–1110 | 1111–1674 | ≥1675 | |
Cases/controls | 43/112 | 61/111 | 61/108 | 57/107 | |
Model a | 1.0 (referent) | 1.23 (0.96–1.57) | 1.23 (0.96–1.58) | 1.26 (0.97–1.64) | 0.52 |
Model b | 1.0 (referent) | 1.38 (1.04–1.82) | 1.46 (1.08–1.96) | 1.52 (1.11–2.06) | 0.14 |
Model c | 1.0 (referent) | 1.44 (1.05–1.96) | 1.56 (1.12–2.17) | 1.58 (1.13–2.21) | 0.06 |
Model d | 1.0 (referent) | 1.43 (1.01–2.02) | 1.60 (1.11–2.31) | 1.58 (1.09–2.30) | 0.14 |
Data are RR (95% CI). Model a, Conditional logistic regression model controlling for matching factors only; model b, conditional logistic regression additionally controlling for age at blood draw (as a linear continuous variable), time of blood collection (1201–0700, 0701–1100, 1101–1200), fasting status at blood collection (fasting for ≥10 h vs <10 h or unknown), menopausal status (postmenopausal, uncertain status vs premenopausal), parents’ history of myocardial infarction, current postmenopausal hormone use, aspirin use, mean alcohol intake, body mass index, and physical activity in MET-hours (the latter 3 variables modeled as continuous variables with natural cubic splines); model c, conditional logistic regression additionally controlling for history of diabetes, history of hypertension, and history of hypercholesterolemia; model d, conditional logistic regression additionally controlling for plasma CRP and total-to-HDL cholesterol ratio (both modeled as continuous variables with natural cubic splines).