Diabetes prevalence and descriptive characteristics of adults with and without diabetes before and after matching
Diabetes prevalence and descriptive statistics for sociodemographic characteristics and comorbid factors of diabetes and non-diabetes individuals are presented in Table 1. In this study, a total of 1.996 subjects reported having diabetes and 17.133 subjects reported not having diabetes, in the year 2014. Thus, the weighted diabetes prevalence in the community for the year 2014 was 8.98%.
Table 1 Baseline characteristics of patients in the diabetes and non-diabetes groups before and after propensity score matching
Socio-demographic characteristics / comorbidities
|
Before matching
|
After Matching
|
Standardized Mean Difference
|
Diabetes
(N=1996)
|
Non-Diabetes (N=17133)
|
P**
Value
|
Diabetes
(N=1996)
|
Non-Diabetes (N=1996)
|
Before
|
After
|
N
|
Wt (%)*
|
N
|
Wt (%)*
|
N
|
Wt (%)*
|
N
|
Wt (%)*
|
Age
|
|
|
|
|
|
|
|
|
|
|
|
<75
|
1788
|
89.87 %
|
16414
|
96.55 %
|
<0.001
|
1788
|
89.87%
|
1782
|
89.96%
|
0.203
|
-0.009
|
≥75
|
208
|
10.12 %
|
719
|
3.44 %
|
<0.001
|
208
|
10.12%
|
214
|
10.03%
|
-0.228
|
0.001
|
Gender
|
|
|
|
|
|
|
|
|
|
|
|
Male
|
714
|
36.97 %
|
8007
|
49.25 %
|
<0.001
|
714
|
36.96%
|
713
|
37.88%
|
0.084
|
0.013
|
Female
|
1282
|
63.03 %
|
9126
|
50.74 %
|
<0.001
|
1282
|
63.03%
|
1283
|
62.11%
|
-0.214
|
0.039
|
Marital Status
|
|
|
|
|
|
|
|
|
|
|
|
Married
|
1535
|
75.83 %
|
11626
|
64.15 %
|
<0.001
|
1535
|
75.83%
|
1568
|
77.03%
|
0.370
|
0.010
|
Not Married (Single&Widowed&Divorced)
|
461
|
24.16 %
|
5507
|
35.84 %
|
<0.001
|
461
|
24.16%
|
428
|
22.96%
|
-0.213
|
0.002
|
Occupational status
|
|
|
|
|
|
|
|
|
|
|
|
Currently employed
|
1039
|
50.84 %
|
9643
|
55.48 %
|
<0.001
|
1039
|
50.84%
|
1052
|
52.67%
|
0.860
|
-0.003
|
Currently not-employed
|
957
|
49.15 %
|
7490
|
44.51 %
|
<0.001
|
957
|
49.15%
|
944
|
47.32%
|
0.126
|
0.058
|
Health Insurance
|
|
|
|
|
|
|
|
|
|
|
|
General Insured
|
1925
|
95.57 %
|
15845
|
91.16 %
|
<0.001
|
1925
|
95.57%
|
1926
|
95.63%
|
0.309
|
0.025
|
Uninsured
|
71
|
4.42 %
|
1288
|
8.83 %
|
<0.001
|
71
|
4.42%
|
70
|
4.36%
|
0.341
|
0.003
|
Comorbidities
|
|
|
|
|
|
|
|
|
|
|
|
Heart Disease
|
459
|
23.52 %
|
1270
|
7.13 %
|
<0.001
|
459
|
23.52%
|
450
|
23.95%
|
0.216
|
0.024
|
Hypertension
|
1131
|
56.67 %
|
2404
|
12.2 %
|
<0.001
|
1131
|
56.67%
|
1134
|
54.31%
|
0.388
|
0.011
|
Stroke
|
54
|
3.27 %
|
111
|
0.66 %
|
<0.001
|
54
|
3.27%
|
35
|
1.86%
|
0.126
|
0.058
|
Arthritis
|
438
|
21.46 %
|
1335
|
6.91 %
|
<0.001
|
438
|
21.46%
|
435
|
21.37%
|
0.341
|
0.003
|
Asthma
|
389
|
19.50 %
|
1239
|
6.82 %
|
<0.001
|
389
|
19.50%
|
369
|
18.60%
|
0.309
|
0.025
|
Depression
|
387
|
19.62 %
|
1855
|
10.69 %
|
<0.001
|
387
|
19.62%
|
368
|
19.02%
|
0.216
|
0.024
|
Insomnia/Sleeping Disorders
|
1072
|
53.08 %
|
5885
|
34.10 %
|
<0.001
|
1072
|
53.08%
|
1061
|
54.79%
|
0.388
|
0.011
|
Comparison of the descriptive statistics of the subjects who reported not having diabetes and those who reported having diabetes in the year 2014 showed the follows aspects: age ≥ 75 years (10.12% vs. 3.44%), female gender (63.03% vs. 50.74%), married (75.83% vs. 64.15%), currently unemployed (49.15% vs. 44.51%), and insured (95.57% vs. 91.16%). All these differences were marginally significant (P < 0.001; Table 1). Comorbidity rates for heart disease (23.52% vs. 7.13%), hypertension (56.67% vs. 12.20%), stroke (3.27% vs. 0.66%), arthritis (21.46% vs. 6.91%), asthma (19.50% vs. 6.82%), depression (19.62% vs. 10.69%), and insomnia/sleeping disorders (53.08% vs. 34.10%) showed marginally significant differences (P < 0.001 for all).
Comparison of the descriptive statistics of the two groups after matching showed the following results: age ≥ 75 years (10.12% vs. 10.03%), female gender (63.03% vs. 62.11%), married (75.83% vs. 77.03%), currently unemployed (49.15% vs. 47.32%), and insured (95.57% vs. 95.63%). All these differences were marginally significant (P < 0.001; Table 1). Comorbidity rates for heart disease (23.52% vs. 23.95%), hypertension (56.67% vs. 54.31%), stroke (3.27% vs. 1.86%), arthritis (21.46% vs. 21.37%), asthma (19.50%vs. 18.60%), depression (19.62% vs. 19.02%) and insomnia/sleeping disorders (53.08% vs. 54.79%) were higher among adults with diabetes. It is seen that, after PSM, diabetes and non-diabetes groups were well balanced, and all standardized mean differences were ˂10% for all variables.
Balanced diagnostics and propensity scores
Figure 1 shows absolute standardized mean differences in baseline characteristics of diabetes individuals compared to non-diabetes participants before and after 1:1 PSM. In this figure “y” axis show baseline characteristics. “x” axis was the percentage of absolute difference “before” and “after” matching. It is seen that, prior to matching there exists large differences in the mean values of hypertension between diabetes and non-diabetes individuals. The figure shows that PSM reduced standardized differences for most observed baseline characteristics below 10% in absolute values, except employment status and stroke. Distribution of propensity scores shows that there are not unmatched treatment units.
Factors associated with diabetes
The odds ratios (OR) and 95% confidence intervals (CIs) from the logistic regression analyses of health behaviors and accessibility of health care services changes with respect to the likelihood of developing diabetes are presented in Table 2. Compared with smokers, non-smokers were less likely to develop diabetes (OR: 0.96; 95% CI: 0.95–0.97). Additionally, individuals engaged in regular physical activity were less likely to have experienced diabetes than individuals didn’t engage in any physical activity (OR: 0.51; 95% CI: 0.41–0.52). Furthermore, reporting of diabetes was more among individuals eating healthy (OR: 1.10; 95% CI: 1.09–1.11).
Table 2. A comparison of health behaviors and accessibility of health services of individuals with diabetes and propensity score matched non-diabetes
Health Behaviors / Accessibility of Health Care Services
|
Diabetes
N
(%)*
|
Non- Diabetes
N
(%)*
|
OR**
|
95% CI***
|
P value
|
Lower
|
Upper
|
Study N
|
1996
|
1996
|
|
|
|
|
Health Behaviors
|
|
|
|
|
|
|
Smoking
|
No
|
518 (24.9%)
|
450 (22.3%)
|
0.96
|
0.95
|
0.97
|
<0.001
|
Yes (Reference)
|
1478 (75.1%)
|
1546 (77.7%)
|
|
|
|
|
Physical Activity
|
Yes
|
48
(2.1%)
|
89 (3.7%)
|
0.51
|
0.41
|
0.52
|
<0.001
|
No (Reference)
|
1948 (97.9%)
|
1907 (96.3%)
|
|
|
|
|
Healthy Eating Behavior
|
Yes
|
367 (17.1%)
|
314 (15.4%)
|
1.10
|
1.09
|
1.11
|
<0.001
|
No (Reference)
|
1629 (82.9%)
|
1682 (85.2%)
|
|
|
|
|
Accessibility of Health Care Services
|
Blood Sugar Check
|
No
|
333 (17.9%)
|
981 (51.1%)
|
0.15
|
0.14
|
0.16
|
<0.001
|
Yes (Reference)
|
1663 (82.1% )
|
1015 (48.9%)
|
|
|
|
|
Blood Cholesterol Check
|
No
|
487 (25.3%)
|
993 (51.6%)
|
1.07
|
1.06
|
1.08
|
<0.001
|
Yes (Reference)
|
1509 (74.7%)
|
1003 (48.4%)
|
|
|
|
|
Blood Pressure Check
|
No
|
375 (18.9%)
|
596 (31.1%)
|
1.59
|
1.58
|
1.60
|
<0.001
|
Yes (Reference)
|
1621 (81.1%)
|
1400 (68.9%)
|
|
|
|
|
Notes: *Weighted percentages **OR: Odds Ratio, ***CI: 95% Confidence Interval. Logistic regression models incorporated sampling weights.
On the other hand, compared with individuals whose blood sugar was checked by health professionals, among those blood sugar was not checked were less likely to have diabetes (OR: 0.15; 95% CI: 0.14–0.16). Reporting of diabetes was more among individuals whose blood cholesterol was not checked by health professionals than among those blood cholesterol was checked with an odds ratio of 1.07. Finally, reporting of diabetes was more among individuals whose blood pressure was not checked by health professionals than among those whose blood pressure was monitored by health professionals with an odds ratio of 1.59 (Table 2).