The Demographic of Cohort with and without Type 2 Diabetes
As shown in Table 1, finally, 5,329 people met the standard, including 4,639 non-diabetes patients and 690 diabetes patients. Compared with non-diabetes patients, T2DM patients were older(P<0.001). Their BMI and LSBMD were higher (P<0.001). Adults suffering from T2DM had few entertainment activities and low family income, poverty rate, and education level. Among them, differences between the two groups were significant in alkaline phosphatase, total cholesterol, blood urea nitrogen, serum sodium, serum potassium, serum calcium, blood glucose, and HbA1c in the standard serum biochemical test (P<0.001).
Table 1
Weighted Characteristics of Study Sample with and without Type 2 Diabetes
|
Type 2 Diabetes(690)
|
Non-Diabetes(4639)
|
P-value
|
P-value*
|
Age
|
50.76 ± 5.37
|
49.07 ± 5.71
|
<0.001
|
<0.001
|
Gender
|
|
|
0.084
|
-
|
Male
|
360 (52.17%)
|
2257 (48.65%)
|
|
|
Famale
|
330 (47.83%)
|
2382 (51.35%)
|
|
|
Race
|
|
|
<0.001
|
-
|
Mexican American
|
130 (18.84%)
|
637 (13.73%)
|
|
|
Other Hispanic
|
74 (10.72%)
|
494 (10.65%)
|
|
|
Non-Hispanic White
|
174 (25.22%)
|
1638 (35.31%)
|
|
|
Non-Hispanic Black
|
186 (26.96%)
|
1014 (21.86%)
|
|
|
Other Race
|
126 (18.26%)
|
856 (18.45%)
|
|
|
BMI
|
33.37 ± 7.60
|
29.16 ± 6.49
|
<0.001
|
<0.001
|
Duration
|
8.88 ± 8.27
|
/
|
|
|
Education
|
|
|
<0.001
|
-
|
Less than 9th grade
|
79 (11.45%)
|
356 (7.67%)
|
|
|
9-11th grade
|
110 (15.94%)
|
571 (12.31%)
|
|
|
High school graduate
|
152 (22.03%)
|
1025 (22.10%)
|
|
|
Some college or AA degree
|
219 (31.74%)
|
1358 (29.27%)
|
|
|
College graduate or above
|
130 (18.84%)
|
1329 (28.65%)
|
|
|
Ratio of family income to poverty
|
2.36 ± 1.59
|
2.78 ± 1.69
|
<0.001
|
<0.001
|
Vigorous Activities
|
|
|
<0.001
|
-
|
Yes
|
103 (14.93%)
|
1087 (23.43%)
|
|
|
No
|
587 (85.07%)
|
3552 (76.57%)
|
|
|
Smoked
|
|
|
0.268
|
-
|
Yes
|
308 (44.64%)
|
1967 (42.40%)
|
|
|
No
|
382 (55.36%)
|
2672 (57.60%)
|
|
|
Standard Biochemical Examination
|
|
|
|
|
HbA1c
|
7.94 ± 2.12
|
5.58 ± 0.67
|
<0.001
|
<0.001
|
Serum glucose
|
9.35 ± 4.92
|
5.38 ± 1.37
|
<0.001
|
<0.001
|
Alkaline Phosphatase
|
77.02 ± 26.39
|
69.98 ± 23.98
|
<0.001
|
<0.001
|
Blood Urea Nitrogen
|
5.26 ± 2.56
|
4.69 ± 1.64
|
<0.001
|
<0.001
|
Serum Cholesterol
|
4.90 ± 1.28
|
5.24 ± 1.04
|
<0.001
|
<0.001
|
Serum Creatinine
|
84.79 ± 84.90
|
76.09 ± 27.72
|
<0.001
|
0.329
|
Phosphorus
|
1.20 ± 0.19
|
1.19 ± 0.18
|
0.032
|
0.124
|
Uric Acid
|
324.77 ± 89.53
|
318.51 ± 83.43
|
0.069
|
0.155
|
Sodium
|
138.40 ± 2.83
|
139.41 ± 2.28
|
<0.001
|
<0.001
|
Potassium
|
4.04 ± 0.37
|
3.95 ± 0.33
|
<0.001
|
<0.001
|
Total Proten
|
71.79 ± 5.03
|
71.46 ± 4.48
|
0.075
|
0.126
|
Total Calcium
|
2.34 ± 0.09
|
2.33 ± 0.09
|
<0.001
|
<0.001
|
Lumbar Spine BMD
|
1.05 ± 0.17
|
1.02 ± 0.16
|
<0.001
|
<0.001
|
Results in the table: mean + SD / N (%) |
P value *: if it is a continuous variable, it shall be obtained by Kruskal Wallis rank sum test. If the theoretical number of counting variables is less than 10, it shall be obtained by Fisher exact probability test. |
Relationship between the status and duration of T2DM and lumbar spine BMD
As shown in Table 2, we finally built three models. Among them, Model 1: β = 0.039, 95% CI: 0.025-0.054, P<0.00001; Model 2 : β = 0.043, 95% CI: 0.029-0.057, P<0.00001; and Model 3 (full-adjusted model): β = 0.023, 95% CI: 0.004-0.041, P=0.01829. We can know that they were all positively correlated.
Table 2
Associations Between the status of T2DM and Lumbar Spinal Bone Mineral Density (g/cm2)
Exposure
|
Non-adjusted Model
|
Adjust Model I
|
Adjust Model II
|
Non-Diabetes
|
Reference
|
Reference
|
Reference
|
Type 2 Diabetes
|
0.039 (0.025, 0.054) <0.00001
|
0.043 (0.029, 0.057) <0.00001
|
0.023 (0.004, 0.041) 0.01829
|
Male with Non-Diabetes
|
Reference
|
Reference
|
Reference
|
Male with Type 2 Diabetes
|
0.050 (0.030, 0.070) <0.00001
|
0.045 (0.025, 0.065) <0.00001
|
0.015 (-0.013, 0.042) 0.29783
|
Famale with Non-Diabetes
|
Reference
|
Reference
|
Reference
|
Famale with Type 2 Diabetes
|
0.027 (0.007, 0.047) 0.00770
|
0.038 (0.019, 0.058) 0.00009
|
0.032 (0.007, 0.057) 0.01349
|
Data in the table: β ( 95%CI) P-value; Outcome variable: lumbar spine BMD; Exposure variable: Type 2 Diabetes Non-adjusted model adjust for: None; Adjust I model adjust for: Age; Gender; Race ;Adjust II model adjust for: Age; Gender; Race; Education; Ratio; BMI; Duration;Vigorous Activities; Smoking; HbA1c; Serum glucose;Alkaline Phosphatase; Blood Urea Nitrogen; Serum Cholesterol; Serum Creatinine; Phosphorus;Uric Acid; Sodium; Potassium; Total Proten; Total Calcium
|
According to the subgroup stratified by gender, we had the following further findings. For men suffering from diabetes, Model 1: β = 0.050, 95% CI: 0.030-0.070, P<0.00001; Model 2: β = 0.045 95% CI: 0.025-0.065, P<0.00001; and Model 3: β = 0.015 95% CI:-0.013-0.042, P=0.29783. For women suffering from diabetes, Model 1: β = 0.027, 95% CI: 0.007-0.047, P=0.0077; Model 2: β = 0.038, 95% CI: 0.019-0.058, P=0.00009; and Model 3: β = 0.032, 95% CI: 0.007-0.057, P=0.01349. In summary, the positive correlation between the two was relatively stable.
However, as shown in Table 3, there was no significant association between disease duration of T2DM and LSBMD in both genders in all three models (in the Model 3, for males: β = 0.001, 95% CI:-0.0015–0.0035, P=0.433215; for females: β= -0.0018, 95% CI: -0.0037–0.0001,P=0.061801).
Table 3
Associations Between T2DM Duration (Year) and Lumbar Bone Mineral Density (g/cm2)
Model
|
Non-adjusted Model
|
Adjust Model I
|
Adjust Model II
|
Total
|
-0.0003 (-0.0018, 0.0013) 0.742754
|
-0.0004 (-0.0019, 0.0011) 0.613685
|
-0.0007 (-0.0022, 0.0008) 0.363552
|
Male
|
0.0010 (-0.0015, 0.0035) 0.432817
|
0.0003 (-0.0021, 0.0028) 0.784566
|
0.0010 (-0.0015, 0.0035) 0.433215
|
Famale
|
-0.0013 (-0.0032, 0.0006) 0.197111
|
-0.0011 (-0.0029, 0.0007) 0.226405
|
-0.0018 (-0.0037, 0.0001) 0.061801
|
Data in the table: β ( 95%CI) Pvalue; Outcome variable: lumbar spine BMD; Exposure variable: Duration Non-adjusted model adjust for: None; Adjust I model adjust for: Age; Gender; Race ;Adjust II model adjust for: Age; Gender; Race; Education; Ratio; BMI; Vigorous Activities; Smoking; HbA1c; Serum glucose; Alkaline Phosphatase; Blood Urea Nitrogen; Serum Cholesterol; Serum Creatinine; Phosphorus; Uric Acid; Sodium; Potassium; Total Proten; Total Calcium
|
Relationship between serum glucose and lumbar spine BMD
As shown in Table 4, we built 3 models. Among them, Model 1: β = 0.004, 95% CI: 0.002-0.005,P=0.00029; Model 2: β = 0.005, 95%CI: 0.003-0.006,P=<0.00001; and Model 3: β = 0.004, 95%CI: 0.000-0.007,P=0.02742.We found that there seems to be a slightly linear relationship. We further explored the relationship between the quartile of serum glucose and BMD of the lumbar spine. Finally,We found that they did not have a linear correlation(Model 1:P for trend= 0.734, Model 2:P for trend= 0.092, and Model 3:P for trend= 0.909). We further used the smooth curve fitting method to directly find that they had the relationship of a U-shaped curve (Figure 1). According to the saturation effect and threshold effect analysis, the inflection point was 7.77 mmol/L and the log-likelihood ratio P=0.031༜0.05 (Table 5). It indicated that the curvilinear relationship was established. The subgroups of different genders and races were analyzed. According to the smooth curve fitting, the serum glucose of male ,non-Hispanic whites, Spanish, and other races had a line relationship with the BMD of the lumbar spine. The serum glucose of female non-Hispanic black people and Mexican Americans had the relationship of a U-shaped curve with the BMD of the lumbar spine. (Figure 3, Figure 4)
Table 4
Associations Between Serum Glucose (mmol/L) and Lumbar Spinal Bone Mineral Density (g/cm2)
Exposure
|
Non-adjusted
|
Adjust I
|
Adjust II
|
Serum Glucose
|
0.004 (0.002, 0.005) 0.00029
|
0.005 (0.003, 0.006) <0.00001
|
0.004 (0.000, 0.007) 0.02742
|
Serum Glucose
|
|
|
|
Q1
|
Reference
|
Reference
|
Reference
|
Q2
|
0.003 (-0.009, 0.015) 0.61733
|
0.007 (-0.004, 0.019) 0.21283
|
0.011 (-0.001, 0.023) 0.06510
|
Q3
|
-0.005 (-0.017, 0.006) 0.36505
|
0.000 (-0.012, 0.012) 0.99526
|
-0.000 (-0.012, 0.012) 0.98841
|
Q4
|
0.006 (-0.007, 0.018) 0.37293
|
0.014 (0.002, 0.026) 0.02401
|
0.003 (-0.011, 0.017) 0.66512
|
P for trend
|
0.734
|
0.092
|
0.909
|
Data in the table: β ( 95%CI) P-value; Outcome variable: lumbar spine BMD; Exposure variable:Serum Glucose (mmol/L) |
Non-adjusted model adjust for: None; Adjust I model adjust for: Age; Gender; Race ;Adjust II model adjust for: Age; Gender; Race; Education; Ratio; BMI; Duration;Vigorous Activities; Smoking; HbA1c;Serum glucose;Alkaline Phosphatase;Blood Urea Nitrogen;Serum Cholesterol;Serum Creatinine ;Phosphorus;Uric Acid;Sodium;Potassium;Total Proten;Total Calcium |
Table 5
Nonlinearity addressing of Serum glucose(mmol/L)and Lumbar spinal Bone Mineral Density (g/cm2)
Outcome:
|
β 95%CI P value
|
Model 1: Fitting model by standard linear regression
|
0.004 (0.000, 0.007) 0.0274
|
Model2: Fitting model by two-piecewise linear regression Inflection point
|
|
Inflection point
|
7.77
|
< 7.77
|
-0.001 (-0.007, 0.004) 0.6271
|
> 7.77
|
0.007 (0.002, 0.011) 0.0021
|
P for log likelyhood rati o
|
0.031
|
Data in the table: β ( 95%CI) P-value; Outcome variable: lumbar spine BMD; Exposure variable:Serum Glucose (mmol/L) |
Non-adjusted model adjust for: None; Adjust I model adjust for: Age; Gender; Race ;Adjust II model adjust for: Age; Gender; Race; Education; Ratio; BMI; Duration;Vigorous Activities; Smoking; HbA1c;Serum glucose;Alkaline Phosphatase;Blood Urea Nitrogen;Serum Cholesterol;Serum Creatinine ;Phosphorus;Uric Acid;Sodium;Potassium;Total Proten;Total Calcium |
Relationship between HbA1c and lumbar spine BMD
As shown in Table 6, we built 3 models. Among them, Model 1: β = 0.007, 95% CI: 0.004-0.011,P=0.00018; Model 2: β = 0.009, 95% CI: 0.005-0.013,P<0.00001; and Model 3: β = 0.001, 95% CI:-0.007-0.008,P=0.87344. We explored the relationship between the quartile of HbA1c and BMD of the lumbar spine. We found that they did not have a linear correlation((Model 1:P for trend= 0.332, Model 2:P for trend= 0.843, and Model 3:P for trend= 0.016 ). We further used the smooth curve fitting method to directly find that they had the relationship of a U-shaped curve (Figure 2). According to the saturation effect and threshold effect analysis, the inflection point was 5.4% and the log-likelihood ratio P༜0.001 (Table 7). It indicated that the Curve relation was established. We further made the subgroup analysis with gender and race. According to the smooth curve fitting, HbA1c of male,non-Hispanic whites and other Spanish had a line relationship with the BMD of the lumbar spine. The HbA1c of female non-Hispanic black people and Mexican Americans had the relationship of a U-shaped curve with the BMD of the lumbar spine. (Figure 3, Figure 4)
Table 6
Associations Between Glycohemoglobin (%) and Lumbar Spinal Bone Mineral Density (g/cm2)
Exposure
|
Non-adjusted
|
Adjust I
|
Adjust II
|
HbA1c
|
0.007 (0.004, 0.011) 0.00018
|
0.009 (0.005, 0.013) <0.00001
|
0.001 (-0.007, 0.008) 0.87344
|
HbA1c
|
|
|
|
Q1
|
Reference
|
Reference
|
Reference
|
Q2
|
-0.029 (-0.040, -0.018) <0.00001
|
-0.023 (-0.034, -0.011) 0.00008
|
-0.020 (-0.031, -0.008) 0.00081
|
Q3
|
-0.031 (-0.044, -0.019) <0.00001
|
-0.024 (-0.037, -0.012) 0.00013
|
-0.020 (-0.033, -0.007) 0.00328
|
Q4
|
-0.006 (-0.019, 0.007) 0.34641
|
-0.001 (-0.014, 0.012) 0.90267
|
-0.018 (-0.034, -0.002) 0.02599
|
P for trend
|
0.332
|
0.843
|
0.016
|
Data in the table: β ( 95%CI) P-value; Outcome variable: lumbar spine BMD; Exposure variable:Glycohemoglobin (%);Non-adjusted model adjust for: None; Adjust I model adjust for: Age; Gender; Race ;Adjust II model adjust for: Age; Gender; Race; Education; Ratio; BMI; Duration;Vigorous Activities; Smoking; HbA1c;Serum glucose;Alkaline Phosphatase;Blood Urea Nitrogen;Serum Cholesterol;Serum Creatinine ;Phosphorus;Uric Acid;Sodium;Potassium;Total Proten;Total Calcium |
Table 7
Nonlinearity addressing of Glycohemoglobin (%) and Lumbar Spinal Bone Mineral Density (g/cm2)
Outcome:
|
β 95%CI P value
|
Model 1: Fitting model by standard linear regression
|
0.001 (-0.007, 0.008) 0.8734
|
Model2: Fitting model by two-piecewise linear regression Inflection point
|
|
Inflection point
|
5.4
|
< 5.4
|
-0.039 (-0.064, -0.014) 0.0021
|
> 5.4
|
0.006 (-0.002, 0.015) 0.1374
|
P for log likelyhood ratio
|
<0.001
|
Data in the table: β ( 95%CI) P-value; Outcome variable: lumbar spine BMD; Exposure variable:Glycohemoglobin (%); Non-adjusted model adjust for: None; Adjust I model adjust for: Age; Gender; Race ;Adjust II model adjust for: Age; Gender; Race; Education; Ratio; BMI; Duration;Vigorous Activities; Smoking; HbA1c;Serum glucose;Alkaline Phosphatase;Blood Urea Nitrogen;Serum Cholesterol;Serum Creatinine ;Phosphorus;Uric Acid;Sodium;Potassium;Total Proten;Total Calcium |