Table 3.
Association FSH with difference in menstrual cycle lengtha, long menstrual cycles, and short menstrual cycles
| FSH (mIU/mL) | No. cycles | Adjustedb Δ in cycle length days (95% CI) | Adjustedc Δ in cycle length days (95% CI) | |
|---|---|---|---|---|
| Linear FSH, per 1 mIU/mL increase | 1773 | –0.15 (–0.22, –0.08) | –0.15 (–0.23, –0.08) | |
| <10 | 1563 | Reference | ||
| ≥10 | 210 | –1.04 (–1.70, –0.39) | –1.10 (–1.76, –0.44) | |
| Long cycles (>35 days) | aOR (95% CI) b | aOR (95% CI) c | aOR (95% CI) d | |
| <10 | 1435 | Reference | ||
| ≥10 | 158 | 0.31 (0.15, 0.65) | 0.32 (0.15, 0.71) | 0.24 (0.10, 0.60) |
| Short cycles (<25 days) | aOR (95% CI) b | aOR (95% CI) c | aOR (95% CI) d | |
| <10 | 1506 | Reference | ||
| ≥10 | 214 | 1.46 (0.97, 2.19) | 1.52 (1.01, 2.30) | 1.43 (0.88, 2.32) |
Results are displayed as change in cycle length in days for a given FSH category, compared to the reference group.
a To achieve normality of the model residuals, the continuous analysis was limited to cycles between 22 and 36 days long.
b Adjusted for age.
c Adjusted for age, race, education, BMI, time since oral contraceptive use, alcohol, smoking, and caffeine consumption.
d Sensitivity analysis the association of categorical FSH with long and short cycles weighted by the inverse of cycles contributed by each woman.