Figure. 1 shows the flowchart of patient enrollment. After screening 200,859 patients, 134,429 (38,739 with DM and 95,690 without DM) patients were included in the final analyses. Table 1 shows the baseline characteristics of all included patients. Patients with DM were older (68[59–77] years vs. 65[53–77] years, p < 0.001), more obese (30.22 [25.74–36.14] kg/m2 vs. 26.68 [22.97–31.45] kg/m2, p < 0.001) compared to non-DM. Patients with DM in comparison with nondiabetic patients also had significantly higher white blood cell, platelet, serum creatinine, blood urea nitrogen, glucose, and potassium levels, and significantly lower hemoglobin, serum albumin and sodium concentrations (all p < 0.001). DM patients in the ICU were admitted more frequently for sepsis (16.9% vs. 12.9%, p < 0.001) and AKI (2.3% vs. 1.1%, p < 0.001), whereas no statistically significant was found in AMI, CVA or pneumonia than non-DM. In the treatment, DM patients had higher need for mechanical ventilation (33.1% vs. 29.0%, p < 0.001) and vasopressor (19.1% vs. 16.5%, p < 0.001) when compared with non-DM.
Table 1
Characteristics of the studied cohorts
Characteristics
|
Total,
n = 134429
|
DM,
n = 38739
|
Non-DM,
n = 95690
|
P-Value
|
Missing value
|
Age,year
|
66(54–77)
|
68(59–77)
|
65(53–77)
|
< 0.001
|
0(0)
|
Male (%)
|
73075(54.4)
|
21204(54.7)
|
51871(54.2)
|
0.088
|
43(0)
|
BMI, kg/m2
|
27.59(23.61–32.88)
|
30.22(25.74–36.14)
|
26.68(22.97–31.45)
|
< 0.001
|
2213(1.6)
|
Racial, n (%)
|
|
|
|
< 0.001
|
|
Caucasian
|
103431(76.9)
|
28176(72.7)
|
75255(78.6)
|
|
0(0)
|
African American
|
14643(10.9)
|
5039(13.0)
|
9604(10.0)
|
|
0(0)
|
Hispanic
|
5281(3.9)
|
1806(4.7)
|
3475(3.6)
|
|
0(0)
|
Asian
|
2328(1.7)
|
792(2.0)
|
1536(1.6)
|
|
0(0)
|
Native American
|
920(0.7)
|
393(1.0)
|
527(0.6)
|
|
0(0)
|
Other/Unknown
|
7826(5.8)
|
2533(6.5)
|
5293(5.5)
|
|
0(0)
|
Hospital discharge year, n (%)
|
|
|
|
0.289
|
|
2014
|
64523(48.0)
|
18506(47.8)
|
46017(48.1)
|
|
0(0)
|
2015
|
69906(52.0)
|
20233(52.2)
|
49673(51.9)
|
|
0(0)
|
Admission Type
|
|
|
|
< 0.001
|
|
Medical-Surgical ICU
|
72550(54.0)
|
21271(54.9)
|
51279(53.6)
|
|
0(0)
|
Coronary Care Unit/Cardiothoracic ICU
|
11598(8.6)
|
3574(9.2)
|
8024(8.4)
|
|
0(0)
|
Neurological ICU
|
11102(8.3)
|
2376(6.1)
|
8726(9.1)
|
|
0(0)
|
Medical ICU
|
11069(8.2)
|
3342(8.6)
|
7727(8.1)
|
|
0(0)
|
Cardiac ICU
|
9478(7.1)
|
2977(7.7)
|
6501(6.8)
|
|
0(0)
|
Surgical ICU
|
8831(6.6)
|
2169(5.6)
|
6662(7.0)
|
|
0(0)
|
Cardiac Surgery ICU
|
5137(3.8)
|
1605(4.1)
|
3532(3.7)
|
|
0(0)
|
Cardiothoracic ICU
|
4664(3.5)
|
1425(3.7)
|
3239(3.4)
|
|
0(0)
|
Source of admission
|
|
|
|
< 0.001
|
|
Emergency Department
|
68306(50.8)
|
20008(51.6)
|
48298(50.5)
|
|
0(0)
|
Operating Room
|
21147(15.7)
|
5666(14.6)
|
15481(16.2)
|
|
0(0)
|
Floor
|
16573(12.3)
|
5262(13.6)
|
11311(11.8)
|
|
0(0)
|
Direct Admit
|
10706(8.0)
|
3102(8.0)
|
7604(7.9)
|
|
0(0)
|
Recovery Room
|
6966(5.2)
|
1588(4.1)
|
5378(5.6)
|
|
0(0)
|
Step-Down Unit (SDU)
|
2005(1.5)
|
617(1.6)
|
1388(1.5)
|
|
0(0)
|
Others
|
8726(6.5)
|
2496(6.4)
|
6230(6.5)
|
|
0(0)
|
Laboratory tests
|
|
Lactate, mmol/L
|
1.7(1.1–2.8)
|
1.7(1.1–2.8)
|
1.7(1.1–2.8)
|
0.132
|
103298(76.8)
|
White blood cell, ×103/uL
|
10.4(7.6–14.4)
|
10.5(7.7–14.4)
|
10.4(7.5–14.4)
|
< 0.001
|
19526(14.5)
|
Hemoglobin, g/dL
|
11.1(9.4–12.7)
|
10.7(9.1–12.2)
|
11.3(9.6–12.9)
|
< 0.001
|
18160(13.5)
|
Platelet, ×103/uL
|
191(142–247)
|
192(144–250)
|
190(142–246)
|
< 0.001
|
20815(15.5)
|
Creatinine, mg/dL
|
0.99(0.74–1.50)
|
1.20(0.84–2.06)
|
0.91(0.70–1.30)
|
< 0.001
|
15746(11.7)
|
Blood urea nitrogen, mg/dL
|
19(13–31)
|
24(15–40)
|
17(12–27)
|
< 0.001
|
16232(12.1)
|
First BGL in ICU admission day, mg/dL
|
127(103–163)
|
158(119–216)
|
119(100–145)
|
< 0.001
|
10250(7.6)
|
Albumin, g/dL
|
2.9(2.4–3.4)
|
2.9(2.4–3.3)
|
2.9(2.4–3.4)
|
< 0.001
|
76666(57.0)
|
Potassium, mmol/L
|
4.0(3.7–4.4)
|
4.1(3.7–4.6)
|
4.0(3.7–4.4)
|
< 0.001
|
14894(11.1)
|
Sodium, mmol/L
|
138(136–141)
|
138(135–141)
|
139(136–141)
|
< 0.001
|
15743(11.7)
|
Admission APACHE IV diagnosis category, n (%)
|
|
Sepsis
|
18893(14.1)
|
6532(16.9)
|
12361(12.9)
|
< 0.001
|
0(0)
|
AMI
|
6488(4.8)
|
1883(4.9)
|
4605(4.8)
|
0.708
|
0(0)
|
CVA
|
5535(4.1)
|
1553(4.0)
|
3982(4.2)
|
0.203
|
0(0)
|
Pneumonia
|
4870(3.6)
|
1432(3.7)
|
3438(3.6)
|
0.357
|
0(0)
|
AKI
|
1932(1.4)
|
884(2.3)
|
1048(1.1)
|
< 0.001
|
0(0)
|
Scores
|
|
|
APACHE IV
|
50(37–67)
|
55(42–71)
|
48(35–65)
|
< 0.001
|
18960(14.1)
|
OASIS
|
22(15–30)
|
23(16–30)
|
22(15–29)
|
< 0.001
|
0(0)
|
SOFA
|
2(1–5)
|
3(1–5)
|
2(1–5)
|
< 0.001
|
0(0)
|
CCI
|
0(0–2)
|
1(0–2)
|
0(0–2)
|
< 0.001
|
0(0)
|
Congestive heart failure
|
20724(15.4)
|
9265(23.9)
|
11459(12.0)
|
< 0.001
|
0(0)
|
Chronic obstructive pulmonary diseases
|
20365(15.1)
|
6928(17.9)
|
13437(14.0)
|
< 0.001
|
0(0)
|
Any tumor
|
20152(15.0)
|
5276(13.6)
|
14876(15.5)
|
< 0.001
|
0(0)
|
Myocardial infarct
|
12999(9.7)
|
5369(13.9)
|
7630(8.0)
|
< 0.001
|
0(0)
|
Renal disease
|
8082(6.0)
|
4248(11.0)
|
3834(4.0)
|
< 0.001
|
0(0)
|
Peripheral vascular disease
|
6581(4.9)
|
3008(7.8)
|
3573(3.7)
|
< 0.001
|
0(0)
|
Dementia
|
4423(3.3)
|
1419(3.7)
|
3004(3.1)
|
< 0.001
|
0(0)
|
Peptic ulcer disease
|
3510(2.6)
|
1053(2.7)
|
2457(2.6)
|
0.117
|
0(0)
|
AIDS
|
179(0.1)
|
26(0.1)
|
153(0.2)
|
< 0.001
|
0(0)
|
Treatment
|
|
|
|
|
|
Ventilation (%)
|
40557(30.2)
|
12815(33.1)
|
27742(29.0)
|
< 0.001
|
5057(3.8)
|
Vasopressor (%)
|
23210(17.3)
|
7401(19.1)
|
15809(16.5)
|
< 0.001
|
0(0)
|
Outcomes
|
|
|
|
|
|
ICU LOS, day
|
1.76(0.96–3.13)
|
1.84(1.00-3.31)
|
1.72(0.95–3.07)
|
< 0.001
|
0(0)
|
Hospital LOS, day
|
5.33(2.94–9.32)
|
5.86(3.24–9.98)
|
5.16(2.84–9.09)
|
< 0.001
|
0(0)
|
Predicted ICU mortalitya, %
|
2.5(1.0-6.7)
|
3.1(1.3–7.9)
|
2.2(0.9–6.2)
|
< 0.001
|
20826(15.5)
|
ICU mortality, n (%)
|
6096(4.5)
|
1816(4.7)
|
4280(4.5)
|
0.086
|
0(0)
|
Hospital mortality, n (%)
|
10392(7.7)
|
3101(8.0)
|
7291(7.6)
|
0.016
|
1093(0.8)
|
Data expressed as median (IQR) or N (%) |
Abbreviations: DM, diabetes mellitus; BMI, body mass index; BGL, blood glucose level; AMI, acute myocardial infarction; CVA, cerebrovascular accident/stroke; AKI, acute kidney injury; APACHE-IV, Acute Physiology, and Chronic Health Evaluation IV; OASIS, Oxford Acute Severity of Illness Score; SOFA, Sequential Organ Failure Assessment; CCI: Charlson comorbidity index; ICU, intensive care unit; LOS, length of stay. |
P values indicate differences between DM and non-DM. P < 0.05 was considered as significant. |
As expected, diabetic patients had higher APACHE IV (55 vs. 48, p < 0.001), OASIS (23 vs. 22, p < 0.001), SOFA (3 vs. 2, p < 0.001) and CCI (1 vs. 0, p < 0.001) scores when admitted to ICU compared with nondiabetic individuals. Among all elements in APACHE IV score, the top eight significant difference between DM and non-DM were BGL, blood urea nitrogen, creatinine, age, mean artery pressure, heart rate, sodium and hematocrit (Figure. 3).
Patients with DM had similar ICU (4.7% vs. 4.5%, p = 0.086), but a higher hospital (8.0% vs. 7.6%, p = 0.016) mortality rates when compared with non-diabetic patients (Table 1). We have calculated the expected mortality based on the APACHE IV score and added them to Table 1. The expected mortality of the entire cohort was 2.5% (1.0%-6.7%), and the observed mortality was 4.5%. Regarding DM patients, the expected and observed mortality rates were 3.1% (1.3%-7.9%) and 4.7%, respectively. In nondiabetics, the expected and observed mortality rates were 2.2% (0.9%-6.2%) and 4.5%, respectively. All observed mortalities were within the IQR of expected mortalities. The mortality index (observed/expected mortality) for the entire cohort, diabetics, and nondiabetics were 1.80, 1.52, and 2.05, respectively. In multivariate analysis, we found that pre-existing DM (OR: 0.846, 95%CI: 0.791–0.905) and higher BMI (OR: 0.994, 95%CI: 0.991–0.998) were associated with decreased ICU mortality, but older age (OR: 1.006, 95%CI: 1.004–1.008), higher APACHE IV score (OR: 1.039, 95%CI: 1.037–1.040), higher SOFA score (OR: 1.109, 95%CI: 1.095–1.122) and higher CCI (OR: 1.062, 95%CI: 1.039–1.086) at admission were associated with increased ICU mortality (Table 2). Similar results were still found in adjusted hospital mortality except for male gender was associated with decreased hospital mortality (Table 2).
Table 2
Univariate and multivariate analysis of risk factors associated with mortality for all included patients
Variable
|
ICU mortality
|
Hospital mortality
|
Univariable
|
Multivariable
|
Univariable
|
Univariable
|
OR (95% CI)
|
P-Value
|
OR (95% CI)
|
P-Value
|
OR (95% CI)
|
P-Value
|
OR (95% CI)
|
P-Value
|
DM
|
1.050(0.993–1.111)
|
0.086
|
0.846(0.791–0.905)
|
< 0.001
|
1.055(1.010–1.102)
|
0.016
|
0.847(0.803–0.893)
|
< 0.001
|
Age, year
|
1.021(1.019–1.023)
|
< 0.001
|
1.006(1.004–1.008)
|
< 0.001
|
1.028(1.027–1.030)
|
< 0.001
|
1.013(1.011–1.014)
|
< 0.001
|
Gender (male vs. female)
|
0.976(0.927–1.027)
|
0.346
|
|
|
0.944(0.907–0.982)
|
0.005
|
0.939(0.896–0.985)
|
0.010
|
BMI, kg/m2
|
0.986(0.983–0.989)
|
< 0.001
|
0.994(0.991–0.998)
|
0.003
|
0.981(0.978–0.983)
|
< 0.001
|
0.990(0.986–0.993)
|
< 0.001
|
Racial, n (%)
|
|
|
|
|
|
|
|
|
Caucasian
|
Reference
|
Reference
|
Reference
|
Reference
|
Reference
|
Reference
|
Reference
|
Reference
|
African American
|
0.859(0.788–0.938)
|
0.001
|
0.757(0.683–0.839)
|
< 0.001
|
0.879(0.822–0.941)
|
< 0.001
|
0.830(0.766-0.900)
|
< 0.001
|
Hispanic
|
0.789(0.682–0.913)
|
0.001
|
0.643(0.540–0.765)
|
< 0.001
|
1.013(0.914–1.121)
|
0.809
|
0.917(0.811–1.036)
|
0.162
|
Asian
|
0.996(0.819–1.211)
|
0.967
|
1.068(0.839–1.360)
|
0.592
|
1.028(0.883–1.195)
|
0.725
|
1.047(0.863–1.270)
|
0.643
|
Native American
|
0.858(0.617–1.193)
|
0.363
|
0.614(0.416–0.905)
|
0.014
|
0.960(0.751–1.226)
|
0.743
|
0.754(0.556–1.022)
|
0.069
|
Other/Unknown
|
0.953(0.852–1.065)
|
0.394
|
1.005(0.882–1.146)
|
0.941
|
0.882(0.806–0.965)
|
0.006
|
0.921(0.827–1.026)
|
0.137
|
APACHE IV score
|
1.049(1.048–1.050)
|
< 0.001
|
1.039(1.037–1.040)
|
< 0.001
|
1.046(1.045–1.047)
|
< 0.001
|
1.036(1.035–1.038)
|
< 0.001
|
SOFA score
|
1.364(1.355–1.374)
|
< 0.001
|
1.109(1.095–1.122)
|
< 0.001
|
1.323(1.315–1.331)
|
< 0.001
|
1.095(1.084–1.106)
|
< 0.001
|
CCI
|
1.190(1.169–1.213)
|
< 0.001
|
1.062(1.039–1.086)
|
< 0.001
|
1.241(1.224–1.259)
|
< 0.001
|
1.106(1.087–1.125)
|
< 0.001
|
Abbreviations: DM, diabetes mellitus; BMI, body mass index; APACHE-IV, Acute Physiology and Chronic Health Evaluation IV; SOFA, Sequential Organ Failure Assessment; CCI, Charlson comorbidity index; OR, odd ratio; CI: confidence interval. |
P < 0.05 was considered as significant. |
When the first BGL on ICU admission day was stratified to quartiles (i.e., < 103; 103–127; 128–163; and > 163 mg/dL), there was a U-shaped relationship between BGL and ICU mortality in both diabetics and nondiabetics. In comparison with nondiabetics, among diabetics, ICU mortality was significantly higher when their blood glucose level was < 103 mg/dL, and was significantly lower ICU when blood glucose level was > 163 mg/dL (Figure. 2A). Using logistic regression models following adjustments for confounding variables, patients with DM had lower ICU mortality when they were in the third and fourth quartiles of BGL (adjusted OR: 0.860, 95%CI: 0.752–0.984, p = 0.029; adjusted OR: 0.716, 95%CI: 0.647–0.793, p < 0.001; respectively) (Table 3). Multivariable Cox survival analysis showed that pre-existing DM was associated with decreased odds of ICU mortality in severe hyperglycemia (> 163 mg/dl) when compared with non-DM (adjusted HR: 0.806, 95%CI: 0.730–0.890, p < 0.001) (Table 4). When mean BGL at second 24 hours (mBGL24) of ICU admission were divided into quartiles (< 107 mg/dl; 107–127 mg/dl; 128–155 mg/dl; and > 155 mg/dl), U-shaped relationship between mBGL24 and ICU mortality was found in diabetic patients, but there was a dose-response relationship between mBGL24 and ICU mortality in nondiabetic patients (Figure. 2B). Diabetic patients were associated with lower ICU mortality when the mBGL24 was at higher range (128–155 mg/dl and > 155 mg/dl) in adjusted model (adjusted OR: 0.804, 95%CI: 0.679–0.951, p = 0.011; adjusted OR: 0.731, 95%CI: 0.835–0.842, p < 0.001; respectively) (Table 3). Cox regression models, adjusted for confounders, showed that DM was associated with a decreased risk of ICU mortality in the fourth quartile of mBGL24 compared with non-DM (adjusted HR: 0.875, 95%CI: 0.767–0.999, p = 0.048) (Table 4). In the quartiles of mean BGL (mBGL) during ICU stay (< 107 mg/dl; 107–123 mg/dl; 124–151 mg/dl; and > 151 mg/dl), patients with DM had higher ICU mortality in the first quartile of mBGL (6.28% vs. 2.74%, p < 0.001), but lower ICU mortality in the third and fourth quartiles of mBGL compared with non-DM (3.94% vs. 5.23%, p < 0.001; 5.07% vs. 11.09%, p < 0.001; respectively) (Figure. 2C). Similar results were shown in the logistic regression model (Table 3). By means of multivariate Cox proportional hazard analysis, patients with DM were associated with higher ICU mortality compared with non-DM in the first and second quartiles of mBGL (adjusted HR: 1.708, 95%CI: 1.432–2.039, p < 0.001; adjusted HR: 1.383, 95%CI: 1.129–1.693, p = 0.002; respectively), with lower ICU mortality in the third and fourth quartiles of mBGL (adjusted HR: 0.828, 95%CI: 0.731–0.938, p = 0.003; adjusted HR: 0.528, 95%CI: 0.483–0.577, p < 0.001; respectively) (Table 4).
Table 3
Multivariable analysis on the effect of DM on ICU mortality in the subgroup analyses at ICU admission
Variable
|
Category
|
No. of patients
|
Model 1
|
Model 2
|
OR (95% CI)
|
P-value
|
OR (95% CI)
|
P-value
|
First BGLa within 24 hours of ICU admission, mg/dL (n = 124179)
|
< 103
|
30340
|
1.049(0.892–1.232)
|
0.564
|
1.030(0.875–1.211)
|
0.725
|
103–127
|
31503
|
0.873(0.740–1.030)
|
0.107
|
0.880(0.737–1.051)
|
0.157
|
128–163
|
31127
|
0.905(0.795–1.031)
|
0.133
|
0.860(0.752–0.984)
|
0.029
|
> 163
|
31209
|
0.649(0.589–0.715)
|
< 0.001
|
0.716(0.647–0.793)
|
< 0.001
|
Mean BGLa in second ICU day, mg/dL (n = 53381)
|
< 107
|
13244
|
0.971(0.757–1.245)
|
0.815
|
0.947(0.738–1.216)
|
0.670
|
107–127
|
13721
|
0.948(0.765–1.176)
|
0.629
|
0.945(0.762–1.172)
|
0.604
|
128–155
|
13053
|
0.807(0.682–0.955)
|
0.013
|
0.804(0.679–0.951)
|
0.011
|
> 155
|
13363
|
0.727(0.632–0.837)
|
< 0.001
|
0.731(0.835–0.842)
|
< 0.001
|
Mean BGLa during ICU stay, mg/dL (n = 128085)
|
< 107
|
32136
|
1.764(1.428–2.179)
|
< 0.001
|
1.708(1.379–2.116)
|
< 0.001
|
107–123
|
31055
|
1.019(0.818–1.269)
|
0.867
|
1.000(0.802–1.246)
|
0.997
|
124–151
|
32961
|
0.696(0.609–0.795)
|
< 0.001
|
0.693(0.606–0.792)
|
< 0.001
|
> 151
|
31933
|
0.443(0.400–0.490)
|
< 0.001
|
0.447(0.404–0.495)
|
< 0.001
|
APACHE IVb (n = 115469)
|
< 37
|
27408
|
1.268(0.826–1.947)
|
0.278
|
1.181(0.766–1.819)
|
0.451
|
37–50
|
30648
|
0.824(0.650–1.044)
|
0.108
|
0.803(0.633–1.019)
|
0.071
|
51–67
|
29580
|
0.765(0.656–0.891)
|
0.001
|
0.767(0.658–0.984)
|
0.001
|
> 67
|
27833
|
0.829(0.770–0.893)
|
< 0.001
|
0.858(0.795–0.926)
|
< 0.001
|
Admission APACHE IV diagnosis categoryc (n = 37718)
|
Sepsis
|
18893
|
0.778(0.690–0.878)
|
< 0.001
|
0.786(0.696–0.887)
|
< 0.001
|
AMI
|
6488
|
0.877(0.596–1.291)
|
0.507
|
0.909(0.614–1.343)
|
0.631
|
CVA
|
5535
|
0.934(0.678–1.286)
|
0.674
|
0.944(0.684–1.301)
|
0.723
|
Pneumonia
|
4870
|
0.820(0.626–1.072)
|
0.147
|
0.832(0.636–1.090)
|
0.182
|
|
AKI
|
1932
|
0.734(0.436–1.233)
|
0.242
|
1.045(0.938–1.165)
|
0.424
|
Aged, years (n = 134429)
|
< 45
|
16248
|
0.643(0.445–0.929)
|
0.019
|
0.632(0.437–0.914)
|
0.015
|
45–59
|
30947
|
0.758(0.641–0.896)
|
0.001
|
0.773(0.653–0.915)
|
0.003
|
60–75
|
49798
|
0.824(0.742–0.916)
|
< 0.001
|
0.831(0.748–0.923)
|
0.001
|
> 75
|
37436
|
0.924(0.830–1.028)
|
0.146
|
0.934(0.839–1.040)
|
0.212
|
BMIe, kg/m2 (n = 132216)
|
< 18.5
|
5733
|
0.786(0.557–1.108)
|
0.169
|
0.789(0.559–1.114)
|
0.178
|
18.5–24.9
|
38599
|
0.845(0.745–0.958)
|
0.009
|
0.844(0.744–0.957)
|
0.008
|
25-29.9
|
38854
|
0.906(0.800-1.027)
|
0.123
|
0.919(0.811–1.041)
|
0.184
|
30-34.9
|
24621
|
0.925(0.789–1.085)
|
0.337
|
0.939(0.800-1.101)
|
0.438
|
> 35
|
24409
|
0.767(0.662–0.889)
|
< 0.001
|
0.772(0.666–0.896)
|
0.001
|
a. Model 1 adjusted for age, gender, racial, BMI, APACHE IV score (exclude blood glucose) and CCI; Model 2 adjusted for model 1 plus SOFA score |
b. Model 1 adjusted for age, gender, racial, BMI and CCI; Model 2 adjusted for model 1 plus SOFA score |
c. Model 1 adjusted for age, gender, racial, BMI, APACHE IV score and CCI; Model 2 adjusted for model 1 plus SOFA score |
d. Model 1 adjusted for gender, racial, BMI, APACHE IV score (exclude age) and CCI; Model 2 adjusted for model 1 plus SOFA score |
e. Model 1 adjusted for age, gender, racial, APACHE IV score and CCI; Model 2 adjusted for model 1 plus SOFA score |
Abbreviations: DM, diabetes mellitus; ICU, intensive care unit; BGL, blood glucose level; BMI, body mass index; APACHE IV, Acute Physiology and Chronic Health Evaluation IV; CCI, Charlson comorbidity index; SOFA, Sequential Organ Failure Assessment; AMI, acute myocardial infarction; CVA, cerebrovascular accident/stroke; AKI, acute kidney injury; HR, Hazard ratio; CI: confidence interval. |
Table 4
Multivariate analysis by Cox Proportional Hazards Model on the effect of DM on ICU mortality in the subgroup analyses at ICU admission
Variable
|
Category
|
No. of patients
|
Cox Proportional Hazards Model
|
HR (95% CI)
|
P-value
|
First BGL a within 24 hours of ICU admission, mg/dL
|
< 103
|
30340
|
1.010(0.877–1.163)
|
0.891
|
103–127
|
31503
|
0.890(0.756–1.048)
|
0.163
|
128–163
|
31127
|
0.988(0.865–1.128)
|
0.861
|
> 163
|
31209
|
0.806(0.730–0.890)
|
< 0.001
|
Mean BGLa during ICU stay, mg/dL
|
< 107
|
32136
|
1.708(1.432–2.039)
|
< 0.001
|
107–123
|
31055
|
1.383(1.129–1.693)
|
0.002
|
124–151
|
32961
|
0.828(0.731–0.938)
|
0.003
|
|
> 151
|
31933
|
0.528(0.483–0.577)
|
< 0.001
|
|
< 37
|
27408
|
1.204(0.783–1.851)
|
0.398
|
APACHE IV scoreb
|
37–50
|
30648
|
0.861(0.681–1.089)
|
0.213
|
|
51–67
|
29580
|
0.864(0.743–1.005)
|
0.057
|
|
> 67
|
27833
|
0.882(0.823–0.945)
|
< 0.001
|
Admission APACHE IV diagnosis categoryc
|
Sepsis
|
18893
|
0.825(0.743–0.916)
|
< 0.001
|
AMI
|
6488
|
0.798(0.564–1.129)
|
0.202
|
CVA
|
5535
|
0.911(0.686–1.208)
|
0.516
|
|
Pneumonia
|
4870
|
0.950(0.741–1.217)
|
0.684
|
|
AKI
|
1932
|
0.665(0.413–1.072)
|
0.094
|
|
< 45
|
16248
|
0.739(0.533–1.024)
|
0.069
|
Aged, years
|
45–59
|
30947
|
0.812(0.698–0.944)
|
0.007
|
|
60–75
|
49798
|
0.859(0.782–0.945)
|
< 0.001
|
|
> 75
|
37436
|
0.962(0.873–1.060)
|
0.429
|
|
< 18.5
|
5733
|
0.875(0.647–1.181)
|
0.382
|
|
18.5–24.9
|
38599
|
0.894(0.799-1.000)
|
0.050
|
BMIe, kg/m2
|
25-29.9
|
38854
|
0.913(0.816–1.023)
|
0.116
|
|
30-34.9
|
24621
|
1.016(0.879–1.174)
|
0.830
|
|
> 35
|
24409
|
0.764(0.669–0.872)
|
< 0.001
|
a. adjusted for age, gender, racial, BMI, APACHE IV score (exclude BGL), SOFA score and CCI. |
b. adjusted for age, gender, racial, BMI, SOFA score and CCI. |
c. adjusted for age, gender, racial, BMI, APACHE IV score, SOFA score and CCI. |
d. adjusted for gender, racial, BMI, APACHE IV score (exclude age), SOFA score and CCI. |
e. adjusted for age, gender, racial, APACHE IV score, SOFA score and CCI. |
Abbreviations: DM, diabetes mellitus; ICU, intensive care unit; BGL, blood glucose level; BMI, body mass index; APACHE IV, Acute Physiology and Chronic Health Evaluation IV; CCI, Charlson comorbidity index; SOFA, Sequential Organ Failure Assessment; AMI, acute myocardial infarction; CVA, cerebrovascular accident/stroke; AKI, acute kidney injury; HR, Hazard ratio; CI: confidence interval. |
Abbreviations: eICU-CRD, eICU Collaborative Research Database; APACHE-IV, Acute Physiology and Chronic Health Evaluation IV; DKA, Diabetic ketoacidosis; HHNC, diabetic hyperglycemic hyperosmolar nonketotic coma; ICU, intensive care unit. |
Abbreviations: DM, diabetes mellitus; BGL, blood glucose level; ICU intensive care unit. APACHE IV, Acute Physiology and Chronic Health Evaluation IV; AMI, acute myocardial infarction; CVA, cerebrovascular accident/stroke; AKI, acute kidney injury. |
Abbreviations: DM, diabetes mellitus; APACHE-IV, Acute Physiology and Chronic Health Evaluation IV. |
When patients were divided based on APACHE IV score quartiles (i.e.,<37; 37–50; 51–67; and > 67) at ICU admission, diabetic patients had lower mortality in the third and fourth quartiles than those in non-diabetics (2.55% vs. 3.36%, p < 0.001; 12.94% vs. 15.63%, p < 0.001; respectively) (Figure. 2D). After adjusting by age, gender, racial, BMI, CCI and SOFA score, DM was still associated with lower mortality in the third and fourth quartiles of APACHE IV score compared with non-DM (adjusted OR: 0.767, 95%CI: 0.658–0.984, p = 0.001; adjusted OR: 0.858, 95%CI: 0.795–0.926, p < 0.001; respectively) (Table 3). Cox hazards modeling revealed that DM was associated with better survival than non-DM in the fourth quartile of APACHE IV score (adjusted HR: 0.882, 95%CI: 0.823–0.945, p < 0.001) (Table 4).
In unadjusted analysis, there were no significant differences in ICU mortality between diabetic patients and non-diabetic individuals with congestive heart failure (4.90% vs. 4.85%, p = 0.926), cerebral vascular diseases (5.36% vs. 4.42%, p = 0.141), pneumonia (7.08% vs. 7.97%, p = 0.291) and acute kidney injury (3.37% vs. 4.85%, p = 0.107), but mortality was lower in diabetes patients with sepsis (9.25% vs. 10.72%, p = 0.002) and higher in diabetes patients with AMI than non-DM (3.77% vs. 2.32%, p = 0.001) (Figure. 2E). After adjustments, the presence of DM was an independent variable associated with lower ICU mortality among patients with sepsis (adjusted OR: 0.786, 95%CI: 0.696–0.887, p < 0.001) when compared with nondiabetic patients (Table 3). Similar results were found in the Cox proportional hazard model (adjusted HR: 0.825, 95%CI: 0.743–0.916, p < 0.001) (Table 4).
We divided patients based on age according to the World Health Organization guidelines into four groups (< 45 years; 45–59 years; 60–75 years; and > 75 years). Patients with DM had a similar mortality rate when compared with nondiabetic patients in each age group (Figure. 2F). After adjustment for gender, racial, BMI, APACHE IV score, CCI and SOFA score, presence of DM was associated with a lower risk of ICU mortality in patients aged < 45 years, between 45–59 years and from 60–75 years (adjusted OR: 0.632, 95%CI: 0.437–0.914, p = 0.015; adjusted OR: 0.773, 95%CI: 0.653–0.915, p = 0.003; adjusted OR: 0.831, 95%CI: 0.748–0.923, p = 0.001; respectively) compare to non-DM (Table 3). Cox regression analysis showed pre-existing DM was associated with decreased odds of ICU mortality in between 45–59 years and from 60–75 years compared with non-DM (adjusted HR: 0.812, 95%CI: 0.698–0.944, p = 0.007; adjusted HR: 0.859, 95%CI: 0.782–0.945, p < 0.001) (Table 4).
Diabetic patients with BMI range of 18.5–24.9, 25-29.9 and 30-34.9 had significantly higher ICU mortality (5.93% vs. 5.26%, p = 0.023; 4.69% vs. 4.06%, p = 0.007; 4.01% vs. 3.47%; p = 0.034; respectively) when compared with non-diabetics (Figure. 2G). In the adjusted model, patients with BMI range of 18.5–24.9 and > 35 kg/m2, had lower ICU mortality when they were diabetic (adjusted OR:0.844, 95%CI:0.744–0.957, p = 0.008; adjusted OR:0.772, 95%CI: 0.666–0.896, p = 0.001; respectively) compare with non-DM (Table 3). Cox hazards regression analysis showed an association of DM (adjusted HR: 0.764, 95%CI: 0.669–0.872, p < 0.001) with decreased of ICU mortality in patients with BMI > 35 kg/m2 (Table 4).