Table 2.
Characteristics of the study population and the effects of covariates on BMD
| Reference | Ethnicity/nation | Gender (%women) | Age (y) | Covariates: comparison diabetes and non-diabetes (P value) | Findings |
|---|---|---|---|---|---|
| Barrett-Connor [11] | USA | 61 | 55–88 |
NS: BMI, cigarette smoking, alcohol use (men), regular exercise, diuretic use (women), estrogen use <0.01: alcohol use (women), diuretic use (men) |
No change of statistical significance of mean difference when adjusted for covariates |
| Sosa [47] | Spain | 100 | 61.3/58.8 | <0.05: weight | Analysis of variance (ANOVA) was used to examine the effects of diabetes and weight in bone mass. There were no statistical differences. |
| Tuominen [48] | Finland | 52 | 45–64 |
NS: BMI <0.01: use of loop diuretics |
No change of statistical significance of mean difference when adjusted for covariates |
| Kao [6] | USA | 64 | 30–96 |
NS: diuretics (women), smoking (men), physical activity, calcium intake, estrogen use, menopause status <0.05: diuretics (men), smoking (women), alcohol, BMI |
After adjusted for covariates, the increase of BMD attenuated but the decrease expanded No significant difference between newly diagnosed and previously diagnosed diabetes Positive correlation (hip, forearm): insulin level |
| Dennison [49] | UK | 45 | 59–72 | NA | After adjustment for BMI, all relationship were diminished, even femoral neck and total femur lose significance |
| Bridges [50] | UK | 0 | ≥25 | <0.01: BMI |
Positive correlation: BMI No significant correlation: HbA1C, disease duration, diabetic complication |
| Gerdhem [12] | Sweden | 100 | 75 | <0.001: body weight | Adjustment for body weight, significance remained but the mean difference attenuated |
| de Liefde [14] | Netherlands | 61 | ≥55 | <0.05: BMI, lower limb disability, smoking, baseline use of thiazides, baseline use of loop diuretics | No change of statistical significance of mean difference when adjusted covariates |
| Majima [7] | Japan | 56 | ≥32 |
NS: BMI, Scr <0.01: FPG |
Positive correlation: BMI, insulin level, HbA1C No significant correlation: FPG |
| Schwartz [51] | USA | 50 | 70–79 |
NS: IL-6 (black, white men), current smoker, walking speed (black), statin use, oral estrogen use, renal insufficiency(black), vitamin D supplement use <0.05: weight, weight change, IL-6 (white women), walking speed (white), renal insufficiency (white) |
After adjusting for covariates, white women with DM lost more BMD per year on average than those without DM Adjustment for weight loss resulted in the largest attenuation in the association between DM and bone loss |
| Bonds [45] | USA | 100 | 64.9/63.5 | NA | NA |
| Rakic [52] | Australia | 44 |
Female: 65.5/64.8 Male: 66.0/66.3 |
NA |
Adjustment for BMI, statistical significance of the mean differences was lost at the spine (women) and hip (men) Negative correlation: serum triglycerides, HbA1C |
| Hadzibegovic [53] | Croatia | 100 | 41–84 | NS: BMI, menarche age, alkaline phosphatase |
Positive correlation: BMI, menarche age Negative correlation: alkaline phosphatase |
| Anaforoglu [54] | Turkey | 100 | 61.9/60.1 | <0.05: BMI, calcium intake | Adjustment for BMI and calcium intake, no statistical significant change |
| Yaturu et al. [5] | USA | 0 | 67.5/66.2 | <0.05: BMI, smoking, alcohol | Matched covariates, statistical significance of mean difference at the spine was lost and at the hip was cut down |
BMI body mass index, NS not significant, NA no data, Scr serum creatinine, FBG fasting blood glucose