Table 2.
Hazard ratios of severe hypoglycaemia and chronic kidney disease (CKD) for the risk of all-cause death in patients with type 2 diabetes
| Exposures | Number of death (%) | Hazard ratio | 95% CI | P value |
|---|---|---|---|---|
|
Independent Models |
|
|
|
|
| Patients without CKD |
|
|
|
|
| Model 1: Hypoglycemia |
749(9.45%) |
3.48 |
2.68 to 4.52 |
<0.0001 |
| Model 2: Hypoglycemia |
749(9.45%) |
1.81 |
1.38 to 2.38 |
<0.0001 |
| Model 3: Hypoglycemia |
749(9.45%) |
1.72 |
1.31 to 2.26 |
<0.0001 |
| Patients with CKD |
|
|
|
|
| Model 1: Hypoglycemia |
326(36.34%) |
2.87 |
2.16 to 3.82 |
<0.0001 |
| Model 2: Hypoglycemia |
326(36.34%) |
2.70 |
2.00 to 3.64 |
<0.0001 |
| Model 3: Hypoglycemia |
326(36.34%) |
2.63 |
1.95 to 3.55 |
<0.0001 |
|
Interactive Models |
|
|
|
|
| Model 1 |
|
|
|
|
| Hypoglycemia = Yes and CKD = No |
61(22.43%) |
3.47 |
2.67 to 4.52 |
<0.0001 |
| Hypoglycemia = No and CKD = Yes |
267(34.19%) |
5.70 |
4.93 to 6.57 |
<0.0001 |
| Hypoglycemia = Yes and CKD = Yes |
59(50.86%) |
16.21 |
12.38 to 21.23 |
<0.0001 |
| Hypoglycemia = No and CKD = No |
683(8.99%) |
|
Reference |
|
| Model 2 |
|
|
|
|
| Hypoglycemia = Yes and CKD = No |
61(22.43%) |
1.87 |
1.43 to 2.44 |
<0.0001 |
| Hypoglycemia = No and CKD = Yes |
267(34.19%) |
1.75 |
1.49 to 2.07 |
<0.0001 |
| Hypoglycemia = Yes and CKD = Yes |
59(50.86%) |
4.32 |
3.25 to 5.75 |
<0.0001 |
| Hypoglycemia = No and CKD = No |
683(8.99%) |
|
Reference |
|
| Model 3 |
|
|
|
|
| Hypoglycemia = Yes and CKD = No |
61(22.43%) |
1.81 |
1.38 to 2.37 |
<0.0001 |
| Hypoglycemia = No and CKD = Yes |
267(34.19%) |
1.63 |
1.38 to 1.93 |
<0.0001 |
| Hypoglycemia = Yes and CKD = Yes |
59(50.86%) |
3.91 |
2.93 to 5.21 |
<0.0001 |
| Hypoglycemia = No and CKD = No | 683(8.99%) | Reference |
Model 1, not adjusted for other covariables at enrollment;
Model 2, adjusted for age, sex, body mass index (BMI), smoking status, alcohol use, low-density lipoprotein cholesterol (LDL-C), high density-lipoprotein cholesterol, triglyceride, systolic blood pressure (SBP), HBA1c, duration of disease, and Ln (urinary albumin to creatinine ratio [ACR] +1), prior cardiovascular disease and prior cancer. As BMI violated the proportional hazard assumption, Cox models stratified on quartiles of BMI were used to adjust for the confounding effect of BMI;
Model 3, further adjusted for drug use at enrollment, including lipid lowering drugs, renin-angiotensin system inhibitors, oral anti-diabetes drugs and insulin.