Among 941 stroke patients, 846 were successfully followed up with 3 months after discharge, and 743 of these patients remained in contact at 12 months (Fig. 1). Therefore, we performed most analyses with 846 patients, and the 12-month analysis was restricted to 743 patients. Of the 846 stroke patients, 522 (61.7%) were male, and the mean age was 65.7 ± 12.6 years. The mean and median FBG/HbA1c ratios were 0.98 (SD, 0.34) and 0.91 (IQR, 0.81–1.06), respectively. The median of admission NIHSS score was 3 (IQR, 1–7). (Table 1) shows the baseline characteristics of the patients by quartiles of the FBG/HbA1c ratio.
Table 1
| FBG/HbA1c | | | | p Value |
| Q1 (n = 214) | Q2 (n = 212) | Q3 (n = 210) | Q4 (n = 210) |
Patient characteristics |
Female sex, n (%) | 61 (28.5) | 95 (44.8) | 79 (37.6) | 89 (42.4) | 0.003 |
Age, mean (SD), years | 66.3 (11.8) | 66.1 (13.1) | 64.8 (12.8) | 65.5 (12.6) | 0.624 |
NIHSS, median (IQR) | 3 (1–5) | 2 (1–5) | 3 (1–6.5) | 4 (1–9.5) | < 0.001 |
Barthel Index, median (IQR) | 75 (45–95) | 75 (45–95) | 60 (30–92.5) | 45 (20–75) | < 0.001 |
Risk of malnutrition, n (%) | 91 (42.7) | 87 (41.0) | 85 (40.5) | 99 (47.5) | 0.503 |
Nutrition support, n (%) | 11 (5.1) | 11 (5.2) | 26 (12.4) | 49 (23.3) | < 0.001 |
Alcohol abuse, n (%) | 34 (15.9) | 25 (11.8) | 30 (14.3) | 24 (11.4) | 0.482 |
Length of hospital stay, median (IQR), days | 10 (8–13) | 9 (7–13) | 10 (7–14) | 12 (8–17) | < 0.001 |
Cardiovascular risk factors, n (%) |
Atrial fibrillation | 18 (8.4) | 12 (5.7) | 11 (5.2) | 21 (10.0) | 0.187 |
Hypertension | 151 (70.6) | 143 (67.5) | 159 (75.7) | 160 (76.2) | 0.131 |
Diabetes mellitus | 65 (30.4) | 42 (19.8) | 65 (31.0) | 97 (46.2) | < 0.001 |
Coronary heart disease | 14 (6.5) | 6 (2.8) | 8 (3.8) | 9 (4.3) | 0.286 |
Hyperlipemia | 89 (41.6) | 111 (52.4) | 105 (50.0) | 109 (51.9) | 0.092 |
Previous stroke | 58 (27.1) | 39 (18.4) | 38 (18.1) | 33 (15.7) | 0.018 |
History of smoking | 75 (35.0) | 57 (26.9) | 61 (29.0) | 55 (26.2) | 0.172 |
History of drinking | 72 (33.6) | 55 (25.9) | 61 (29.0) | 56 (26.7) | 0.289 |
Blood glucose indicators | | | | | |
FBG, mean (SD), mmol/L | 4.88 (1.1) | 5.34 (1.0) | 6.28 (1.3) | 9.37 (4.0) | < 0.001 |
HbA1C, mean (SD), % | 6.75 (1.7) | 6.18 (1.1) | 6.41 (1.3) | 6.99 (2.0) | < 0.001 |
FBG/HbA1C, mean (SD) | 0.73 (0.09) | 0.86 (0.02) | 0.98 (0.04) | 1.34 (0.49) | < 0.001 |
Notes: Quartiles of FBG/HbA1c ratio, Q1 ≤ 0.81, 0.82 ≤ Q2 < 0.91, 0.92 ≤ Q3 < 1.1.06, Q4 ≥ 1.07. |
Stress Hyperglycemia and Primary Outcomes
(Table 2) and (Fig 2) show the 3- and 12-month all-cause mortality after stroke across quartiles of the FBG/HbA1c ratio. There were 35 (4.1%) all-cause deaths at 3 months and 85 (11.4%) at 12 months. After adjustment for sex, age, NIHSS, atrial fibrillation, hypertension, diabetes, coronary heart disease, hyperlipemia, previous smoking, history of smoking, history of drinking, and infectious complications, multivariable regression showed that patients in the highest quartile of the FBG/HbA1c ratio had an elevated risk of all-cause death at 3 months (adjusted OR: 5.14, 95% CI: 1.06–24.90) and at 12 months (adjusted OR: 2.58, 95% CI: 1.14–5.83).
Table 2 Logistic regression of all-cause death according to FBG/HbA1c quartiles.
Outcomes
|
FBG/HbA1c
|
n
|
Events, n (%)
|
Crude OR (95% CI)
|
Adjusted OR (95% CI)a
|
Primary Outcomes
|
|
|
|
|
|
All-cause death at 3 months
|
Q1 (≤0.81)
|
214
|
2 (0.9)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
212
|
6 (2.8)
|
3.09 (0.62–15.48)
|
2.74 (0.52–14.48)
|
Q3 (0.92–1.06)
|
210
|
10 (4.8)
|
5.30 (1.15–24.49)
|
3.75 (0.77–18.31)
|
Q4 (≥1.07)
|
210
|
17 (8.1)
|
9.34 (2.13–40.94)
|
5.14 (1.06–24.90)
|
All-cause death at 12 months
|
Q1 (≤0.81)
|
183
|
11 (6.0)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
189
|
22 (11.6)
|
2.06 (0.97–4.38)
|
2.47 (1.09–5.58)
|
Q3 (0.92–1.06)
|
190
|
17 (8.9)
|
1.54 (0.70–3.38)
|
1.31 (0.56–3.07)
|
Q4 (≥1.07)
|
181
|
35 (19.3)
|
3.78 (1.84–7.64)
|
2.58 (1.14–5.83)
|
Notes: Quartiles of FBG/HbA1c ratio, Q1≤0.81, 0.82≤Q2<0.91, 0.92≤Q3<1.1.06, Q4≥1.07.
Abbreviations: FBG, fasting blood glucose; HbA1c, glycated hemoglobin.
aLogistic regression is adjusted for sex, age, NIHSS, atrial fibrillation, hypertension, diabetes, coronary heart disease, hyperlipemia, previous stroke, history of smoking, history of drinking, and infectious complications.
Stress Hyperglycemia and Secondary Outcomes
(Table 3) shows the outcomes of hospitalization across quartiles of the FBG/HbA1c ratio in the study patients. There were 140 (16.5%) infectious complications, including 102 (12.1%) instances of pneumonia and 22 (2.6%) UTIs. A total of 165 (19.5%) patients had NIHSS scores ≥6, and 337 (39.8%) had dysfunction (Barthel index value ≤60) after discharge. Compared with patients in the lowest quartile, patients in the highest quartile of the FBG/HbA1c ratio had higher NIHSS scores (P <0.001). Compared with patients in the lowest quartile, patients in the highest quartile of the FBG/HbA1c ratio had a higher risk of infectious complications (adjusted OR 2.52, 95% CI 1.33–4.76) and dysfunction (adjusted OR 1.87, 95% CI 1.13–3.10) after adjusting for covariates. There was also a trend that patients in the highest quartile of the FBG/HbA1c ratio had a higher risk of pneumonia (adjusted OR 1.38, 95% CI 0.69–2.76) and UTI (adjusted OR 2.06, 95% CI 0.46–9.29) and a higher prevalence of NIHSS score ≥6 (adjusted OR 1.75, 95% CI 0.90–3.40) than patients in the lowest quartile.
Table 3 Logistic regression of outcomes of hospitalization according to FBG/HbA1c quartiles.
Outcomes
|
FBG/HbA1c
|
n
|
Events, n (%)
|
Crude OR (95% CI)
|
Adjusted OR (95% CI)
|
Secondary outcomes: of hospitalization
|
Infectious complicationsa
|
Q1 (≤0.81)
|
214
|
21 (9.8)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
212
|
23 (10.8)
|
1.12 (0.60–2.09)
|
1.08 (0.52–2.24)
|
Q3 (0.92–1.06)
|
210
|
29 (13.8)
|
1.47 (0.81–2.68)
|
1.22 (0.61–2.42)
|
Q4 (≥1.07)
|
210
|
67 (31.9)
|
4.31 (2.52–7.36)
|
2.52 (1.33–4.76)
|
Pneumoniaa
|
Q1 (≤0.81)
|
214
|
19 (8.9)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
212
|
14 (6.6)
|
0.73 (0.35–1.49)
|
0.70 (0.31–1.60)
|
Q3 (0.92–1.06)
|
210
|
21 (10.0)
|
1.14 (0.59–2.29)
|
0.81 (0.38–1.73)
|
Q4 (≥1.07)
|
210
|
48 (22.9)
|
3.04 (1.72–5.38)
|
1.38 (0.69–2.76)
|
Urinary tract infectiona
|
Q1 (≤0.81)
|
214
|
3 (1.4)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
212
|
6 (2.8)
|
2.05 (0.51–8.30)
|
1.08 (0.22–5.27)
|
Q3 (0.92–1.06)
|
210
|
5 (2.4)
|
1.72 (0.41–7.27)
|
1.78 (0.39–8.10)
|
Q4 (≥1.07)
|
210
|
8 (3.8)
|
2.78 (0.73–10.65)
|
2.06 (0.46–9.29)
|
NIHSS ≥ 6 post dischargeb
|
Q1 (≤0.81)
|
214
|
29 (14.1)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
212
|
30 (14.7)
|
1.05 (0.61–1.83)
|
1.22 (0.60–2.49)
|
Q3 (0.92–1.06)
|
210
|
39 (19.4)
|
1.47 (0.87–2.49)
|
0.82 (0.41–1.65)
|
Q4 (≥1.07)
|
210
|
67 (34.0)
|
3.15 (1.93–5.14)
|
1.75 (0.90–3.40)
|
Dysfunction post discharge (Bathel≤60)b
|
Q1 (≤0.81)
|
214
|
81 (31.9)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
212
|
88 (41.5)
|
1.11 (0.74–1.66)
|
1.08 (0.66–1.78)
|
Q3 (0.92–1.06)
|
210
|
114 (51.3)
|
1.56 (1.04–2.32)
|
1.23 (0.76–2.02)
|
Q4 (≥1.07)
|
210
|
139 (66.2)
|
2.72 (1.82–4.04)
|
1.87 (1.13–3.10)
|
Notes: Quartiles of FBG/HbA1c ratio, Q1≤0.81, 0.82≤Q2<0.91, 0.92≤Q3<1.1.06, Q4≥1.07.
Abbreviations: FBG, fasting blood glucose; HbA1c, glycated hemoglobin; NIHSS, The National Institutes of Health Stroke Scale.
aAdjusted for sex, age, NIHSS, diabetes, previous stroke, alcohol abuse, risk of malnutrition, length of hospital stay, and nutrition support
bAdjusted for sex, age, NIHSS, atrial fibrillation, hypertension, diabetes, coronary heart disease, hyperlipemia, previous stroke, history of smoking and history of drinking.
(Table 4) shows the 3- and 12-month poor functional outcomes after stroke across quartiles of the FBG/HbA1c ratio. There were 204 (24.1%) poor outcomes at 3 months and 241 (32.9%) at 12 months. There was a trend that patients in the highest quartile of the FBG/HbA1c ratio had a higher risk of poor functional outcomes at 3 months (adjusted OR 1.38, 95% CI 0.78–2.45) and 12 months (adjusted OR 1.68, 95% CI 0.98–3.28).
Table 4 Logistic regression of poor functional outcomes according to FBG/HbA1c quartiles.
Outcomes
|
FBG/HbA1c
|
n
|
Events, n (%)
|
Crude OR (95% CI)
|
Adjusted OR (95% CI)a
|
Secondary Outcomes: after discharge
|
Poor functional outcomes (mRS of 3–6) at 3 months
|
Q1 (≤0.81)
|
214
|
32 (16.4)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
212
|
46 (21.7)
|
1.42 (0.87–2.31)
|
1.51 (0.86–2.66)
|
Q3 (0.92–1.06)
|
210
|
50 (23.8)
|
1.60 (0.99–2.59)
|
1.19 (0.68–2.08)
|
Q4 (≥1.07)
|
210
|
72 (34.3)
|
2.67 (1.68–4.23)
|
1.38 (0.78–2.45)
|
Poor functional outcomes (mRS of 3–6) at 12 months
|
Q1 (≤0.81)
|
183
|
52 (28.4)
|
Ref.
|
Ref.
|
Q2 (0.82–0.91)
|
189
|
49 (25.9)
|
0.88 (0.56–1.39)
|
1.16 (0.68–1.99)
|
Q3 (0.92–1.06)
|
190
|
57 (30.0)
|
1.08 (0.69–1.69)
|
0.94 (0.55–1.59)
|
Q4 (≥1.07)
|
181
|
83 (45.9)
|
2.13 (1.38–3.29)
|
1.68 (0.98–3.28)
|
Notes: Quartiles of FBG/HbA1c ratio, Q1≤0.81, 0.82≤Q2<0.91, 0.92≤Q3<1.1.06, Q4≥1.07.
Abbreviations: FBG, fasting blood glucose; HbA1c, glycated hemoglobin.
aLogistic regression is adjusted for sex, age, NIHSS, atrial fibrillation, hypertension, diabetes, coronary heart disease, hyperlipemia, previous stroke, history of smoking, history of drinking, and infectious complications.
The sensitivity analyses excluding patients with preexisting diabetes yielded similar results (Additional File 1).
The role of infectious complications in the pathway from admission stress hyperglycemia to death
To assess the role of infectious complications in the pathway from stress hyperglycemia to mid- and long-term all-cause death, we also added infectious complications as an interaction variable with stress hyperglycemia quartiles in the logistic regression model. Multivariable regression showed that adding infectious complications (both as individual variables and as an interaction variable with quartiles of FBG/HbA1c ratio) to the models did not modify the relationships of the highest quartile of the FBG/HbA1c ratio with all-cause death at 3 months (adjusted OR 5.63, 95% CI 1.05–30.16) and at 12 months (adjusted OR 3.36, 95% CI 1.28–8.85). No statistical significance was found in the interaction of the FBG/HbA1c ratio with infectious complications in predicting all-cause mortality at 3 and 12 months (P = 0.516 and P = 0.369, respectively).
Subgroup Analysis
In the subgroup analysis of all-cause mortality at 12 months, the highest quartile of the FBG/HbA1c ratio was associated with or showed a trend toward an association with infectious complications in all subgroups (Fig 3). Compared with patients in the lowest quartile, patients in the highest quartile of the FBG/HbA1c ratio had a higher risk of infectious complications in the subgroups of NIHSS score <6 (adjusted OR 3.23, 95% CI 1.01–10.32), age ≥75 (adjusted OR 5.01, 95% CI 1.13–22.25), AIS subjects (adjusted OR 3.66, 95% CI 1.36–9.83), hypertensive subjects (adjusted OR 2.57, 95% CI 1.03–6.37), and noninfectious subjects (adjusted OR 3.45, 95% CI 1.29–9.21).