Table 3—

Multivariate regression analyses for the association of diabetes with subclinical functional limitation

Subclinical functional limitationnModel 1Model 2Model 3Model 4
Any self-reported difficulty versus no difficulty1,2521.69 (1.39–2.06)1.54 (1.26–1.88)1.38 (1.12–1.70)1.40 (1.14–1.73)
Only one difficulty6391.44 (1.13–1.83)1.35 (1.06–1.72)1.29 (1.00–1.65)1.32 (1.02–1.69)
Two difficulties4771.85 (1.43–2.40)1.66 (1.27–2.16)1.44 (1.09–1.89)1.45 (1.10–1.91)
Three difficulties1362.53 (1.69–3.78)2.17 (1.45–3.26)1.71 (1.12–2.61)1.69 (1.10–2.61)
Multivariate linear regression models  β coefficient for diabetes (95% CI)
 Summary physical performance score−0.18* (−0.22 to −0.13)−0.14* (−0.19 to −0.10)−0.11* (−0.15 to −0.06)−0.11* (−0.15 to −0.07)
  • Data are OR (95% CI) of subclinical functional limitation for diabetes. Model 1: Age, sex, race, education, and study site. Model 2: + Body composition measures (total body fat, body height). Model 3: + Comorbidities (peripheral arterial disease, coronary heart disease, congestive heart failure, transient ischemic attack, stroke, hypertension, renal insufficiency, poor eyesight, urinary incontinence, and cognitive function). Model 4: + Potential confounders (osteoarthritis, depression, smoking, physical activity).

  • * P < 0.01;

  • unstandardized β coefficient.