Table 3.
Sex differential exposure: impacting the sex difference in gait speed in a multivariable association model
| Beta “sex” [95% CI] b | Percentage change c | |
|---|---|---|
| Full model a | − 0.029 [-0.051 ― − 0.008] | Reference |
| Socio-demographic | ||
| Education (high/middle versus low) | − 0.032 [-0.054 ― − 0.011] | - 10.5 % |
| Lifestyle | ||
| Alcohol consumption (vs. never) | − 0.030 [-0.051 ― − 0.008] | - 0.7 % |
| Smoking status (vs. never) | − 0.028 [-0.048 ― − 0.007] | + 6.3 % |
| Physical activity (MET hours / week) | − 0.027 [-0.048 ― − 0.006] | + 8.4 % |
| Sleep problems (range 0–9) | − 0.028 [-0.050 ― − 0.007] | + 4.5 % |
| Sleep duration (short/long vs. normal) | − 0.030 [-0.051 ― − 0.008] | - 0.9 % |
| BMI (kg/m2) | − 0.030 [-0.052 ― − 0.009] | - 3.4 % |
| Social | ||
| Personal network size (n) | − 0.026 [-0.048 ― − 0.005] | + 10.8 % |
| Living situation (alone vs. not alone) | − 0.033 [-0.054 ― − 0.012] | - 12.9 % |
| Social participation formal (vs. few a year) | − 0.029 [-0.050 ― − 0.007] | + 2.1 % |
| Social participation informal (range 0–42) | − 0.028 [-0.050 ― − 0.007] | + 3.0 % |
| Loneliness (Gierveld scale, range 0–11) | − 0.029 [-0.051 ― − 0.008] | + 1.1 % |
| Health | ||
| Chronic diseases (vs. none) | − 0.031 [-0.053 ― − 0.010] | - 6.4 % |
| Pain (vs. no) | − 0.037 [-0.059 ― − 0.015] | - 25.4 % |
| Depressive symptoms (CES-D, range 0–60) | − 0.033 [-0.055 ― − 0.012] | - 10.0 % |
Note: Each row represents the model excluding the indicated determinant
Note: Bold = significant (beta p<0.05, percentage change >5.0 %)
Note: A negative percentage change represents a contributing determinant (the sex difference decreases when adjusted for this determinant) and a positive percentage change represents a suppressing determinant (the sex difference increases)
Note: To illustrate: Adjusting for pain in a multivariable model causes a decrease of the sex difference (from -.0037 to -0.029). So, pain contributes (25.4 %) to the observed sex difference in gait speed.
a Full model includes birth cohort, age, baseline height and all determinants (column one)
b Represents the regression coefficient for the association between sex (women vs men) and gait speed
c Percentage change is 1 - (Beta “sex” model excluding the indicated determinant) / Beta “sex” full model, representing the change in the sex difference in gait speed when adjusting for the indicated determinant