Impact of serum glucose levels on disease severity and outcome in patients with severe sepsis: an analysis from a multicenter, prospective survey of severe sepsis
Abstract
Aim
To determine whether glycemic abnormality and pre-existing diabetes are associated with disease severity and mortality in patients with severe sepsis.
Methods
Six hundred and nineteen patients with severe sepsis were grouped into four categories according to their blood glucose levels (<100, 100–199, 200–299, and ≥300 mg/dL). We compared disease severity and mortality between glycemic categories. In addition, we examined whether there was any relationship with pre-existing diabetes status.
Results
There were no significant differences in disseminated intravascular coagulation, Sequential Organ Failure Assessment, or Acute Physiology and Chronic Health Evaluation II scores and mortality rates between patients with or without pre-existing diabetes. However, in patients without pre-existing diabetes, those with blood glucose level <100 mg/dL had higher disseminated intravascular coagulation, Sequential Organ Failure Assessment, and Acute Physiology and Chronic Health Evaluation II scores than those with levels of 100–299 mg/dL. In addition, those with level ≥300 mg/dL had a higher hospital mortality rate than those with levels of 100–199 mg/dL (odds ratio = 4.837). Multivariate logistic regression analysis revealed that a blood glucose level ≥300 mg/dL is an independent predictor of hospital mortality in these patients. In contrast, no significant differences among severity scores or mortality were observed in patients with pre-existing diabetes.
Conclusions
In patients with severe sepsis, the impact of glycemic abnormality on disease severity and hospital mortality depends on the pre-existing diabetes status. Specifically, a blood glucose level ≥300 mg/dL may be associated with increased mortality in patients without pre-existing diabetes.
Introduction
As glucose metabolism is essential for normal functioning of the brain, heart, and other organs, stress hyperglycemia can be a useful adaptive response when metabolic demand increases. Therefore, transient hyperglycemia in critically ill patients without a history of diabetes was considered to be harmless or even advantageous.1 However, several studies have indicated that impaired glucose metabolism is associated with increased morbidity and mortality.2, 3
Diabetes mellitus is linked to impaired immunity and considered a predisposing factor for a wide variety of infectious conditions,4, 5 and it was shown that up to 25% of sepsis patients have diabetes.6, 7 Recent studies suggested that pre-existing diabetes or alterations to glucose levels are associated with a prolonged hospital stay or higher mortality secondary to infections.8, 9
Although hyperglycemia may be associated with increased morbidity and mortality, it has been suggested that established diabetes may weaken the relationship between hyperglycemia and death.1 Several studies suggested that patients without pre-existing diabetes who have hyperglycemia may suffer worse consequences than those with pre-existing diabetes.1, 8, 10 Post-hoc analysis from a large clinical trial also suggested that patients with established diabetes have a lower risk of mortality than those without diabetes.11 Therefore, the impact of glycemic derangement in critically ill patients may be different between patients with and without pre-existing diabetes.
Although there are many reports regarding the influence of hyperglycemia in critically ill patients, little is known about the effect of glycemic abnormality on disease severity and outcomes in patients with severe sepsis. Therefore, we aimed to determine whether abnormal blood glucose levels and pre-existing diabetes are associated with disease severity and hospital mortality in patients with severe sepsis.
Methods
This study is a post-hoc analysis of a multicenter prospective study of severe sepsis in Japan by the Japanese Association for Acute Medicine Sepsis Registry Study Group.12 The study protocol was approved by the Japanese Association for Acute Medicine as well as the ethics committees of all participating hospitals. However, the need for written informed consent from patients was waived because data were collected as a part of routine clinical examinations without any medical intervention, and then analyzed anonymously.
Patients
Between June 2010 and May 2011, we enrolled 624 patients with severe sepsis admitted to one of 15 critical care centers in Japan. Of 624 patients, 5 with missing blood glucose record on the day of enrolment were excluded from this analysis. Patients were divided into two groups according to their pre-existing diabetes status. We assessed the diabetes status on the basis of self-reporting or the prescription of glucose-lowering drugs, or a combination of both, and defined the condition as pre-existing diabetes.
Patients were also grouped into four categories based on their glucose levels (<100, 100–199, 200–299, and ≥300 mg/dL) on the day of enrolment (day 1), as previously reported.8, 13, 14 For patients with multiple blood glucose measurements, we used the median value to categorize their levels.
Definitions
Sepsis, severe sepsis, septic shock, and systemic inflammatory response syndrome were defined according to the American College of Chest Physicians/Society of Critical Care Medicine Consensus Conference and subsequent revisions.15, 16 Disease severity was scored using the Acute Physiology and Chronic Health Evaluation (APACHE) II system at the time of enrolment.17 Organ dysfunction was quantified by the Sequential Organ Failure Assessment (SOFA) score,18 and a score of ≥12 was the criterion for Multiple Organ Dysfunction Syndrome.18 Disseminated intravascular coagulation (DIC) was diagnosed by the scoring system of the International Society on Thrombosis and Haemostasis.19 The outcome measures were 28-day and all-cause hospital mortality rates.
Statistical analyses
Data are expressed as medians and interquartile ranges. All statistical analyses were carried out using spss 19.0 for Windows (SPSS, Chicago, IL, USA). Disease severity scores and mortality rates between patients with or without pre-existing diabetes were compared using the Mann–Whitney U-test. Categorical variables were summarized using proportions and compared using Fisher's exact test. Kruskal–Wallis one-way anova and χ2-tests were used to compare results among multiple groups. Odds ratios are reported relative to a reference range of glucose level, as previously reported.20 We defined the reference range here as the glucose level of 100–199 mg/dL. The Kaplan–Meier method was used for survival analysis and the log–rank test was used to compare the survival distributions for patients in each category.
We used a multivariate logistic model to assess the relationships between hospital mortality and various independent variables. Mortality rate was used as the criterion variable (death = 1; survival = 0); age, gender, DIC, SOFA, and APACHE II scores, and blood glucose levels (≥300 mg/dL = 1, <300 mg/dL = 0) were used as explanatory variables. Results are reported as odds ratios, P-values, and 95% confidence intervals.
We defined statistical significance as P < 0.05 for single comparisons and P < 0.0083 for multiple comparisons (after Bonferroni correction).
Results
Baseline characteristics and outcomes
The characteristics of 619 patients are shown in Table 1. The median initial APACHE II and SOFA scores were 23 and 8, respectively. More than half of the patients had dysfunction of three or more organ systems. The 28-day mortality was 23.1% and the hospital mortality was 29.6%.
| Age, years | 72 (61–81) |
| Gender, male/female | 387/232 |
| APACHE II score | 23 (17–29) |
| SOFA score | 8 (6–11) |
| MODS, SOFA score >12 | 144 (23.3%) |
| Septic shock | 280 (45.2%) |
| Admission category | |
| Medical | 514 (83.0%) |
| Trauma | 27 (4.4%) |
| Surgery | 19 (3.1%) |
| Burns | 19 (3.1%) |
| Other | 40 (6.5%) |
| Infection site | |
| Pulmonary | 257 (41.5%) |
| Intra-abdominal | 133 (21.5%) |
| Skin/soft tissue | 78 (12.6%) |
| Urinary | 77 (12.4%) |
| Meningitis | 15 (2.4%) |
| Catheter-related | 11 (1.8%) |
| Bone/joint | 10 (1.6%) |
| Wound | 10 (1.6%) |
| Infective endocarditis | 3 (0.5%) |
| Other | 25 (4.0%) |
| 28-Day mortality rate | 143 (23.1%) |
| Hospital mortality rate | 183 (29.6%) |
- Data are presented as median with interquartile range in parentheses or number with percentage in parentheses. APACHE, Acute Physiology and Chronic Health Evaluation; MODS, Multiple Organ Dysfunction Syndrome; SOFA, Sequential Organ Failure Assessment.
Comparisons of severity scores and mortality between patients with and without pre-existing diabetes
When patients were stratified by pre-existing diabetes status, no significant differences in mortality rates or DIC, SOFA, and APACHE II scores were observed (Table 2).
| Without pre-existing diabetes mellitus (n = 497) | With pre-existing diabetes mellitus (n = 122) | P-value† | |
|---|---|---|---|
| Age, years | 73.0 (61.0–82.0) | 70.5 (62.0–79.0) | 0.243 |
| Male | 303 (61.0%) | 84 (68.9%) | 0.118 |
| Septic shock | 232 (46.7%) | 48 (39.3%) | 0.156 |
| Positive blood culture | 192 (39.1%) | 60 (49.2%) | 0.052 |
| Glucocorticoid administration | 56 (11.3%) | 9 (7.4%) | 0.250 |
| Body temperature, °C | 37.9 (36.1–38.8) | 38.0 (36.8–38.8) | 0.531 |
| Blood glucose, mg/dL | 132 (106–170) | 194 (139–275) | <0.001 |
| SIRS criteria | 3.0 (3.0–4.0) | 3.0 (3.0–4.0) | 0.694 |
| ISTH DIC score | 2.0 (1.0–4.0) | 3.0 (1.0–4.0) | 0.575 |
| ISTH DIC score ≥5 | 94 (18.9%) | 19 (15.6%) | 0.435 |
| SOFA score | 9.0 (6.0–11.0) | 8.0 (5.0–11.0) | 0.168 |
| MODS (SOFA score ≥12) | 122 (24.5%) | 22 (18.0%) | 0.151 |
| APACHE II score | 23.0 (18.0–29.0) | 23.0 (16.0–28.3) | 0.267 |
| 28-Day mortality rate | 118 (23.7%) | 25 (20.5%) | 0.475 |
| Hospital mortality rate | 147 (29.6%) | 36 (29.5%) | 1.000 |
- Data are presented as median with interquartile range in parentheses or number with percentage in parentheses. †Mann–Whitney U-test or Fisher's exact test. APACHE, Acute Physiology and Chronic Health Evaluation; ISTH DIC, International Society on Thrombosis and Haemostasis Disseminated Intravascular Coagulation; MODS, Multiple Organ Dysfunction Syndrome; SIRS, Systemic Inflammatory Response Syndrome; SOFA, Sequential Organ Failure Assessment.
Comparison of blood glucose levels and severity scores in patients with or without pre-existing diabetes
The severity scores for patients with or without pre-existing diabetes in each of the four glucose categories are shown in Tables 3 and 4. In patients without pre-existing diabetes, those with blood glucose level <100 mg/dL had significantly higher DIC, SOFA, and APACHE II scores on days 1 and 4 than those with 100–199 mg/dL. In addition, on day 1, DIC and SOFA scores of patients with a level <100 mg/dL were significantly higher than those with a level of 200–299 mg/dL. The SOFA scores on days 1 and 4 were also significantly higher in patients with a blood glucose level ≥300 mg/dL than in patients with a level of 100–199 mg/dL. However, in patients with pre-existing diabetes, no significant differences in severity scores among patients with all blood glucose categories were observed.
| <100 mg/dL | 100–199 mg/dL | 200–299 mg/dL | ≥300 mg/dL | |
|---|---|---|---|---|
| (n = 89) | (n = 342) | (n = 50) | (n = 16) | |
| Age, years | 72.0 (61.5–81.5) | 73.0 (61.0–82.0) | 73.0 (59.8–80.5) | 71.0 (57.8–85.8) |
| Day 1 | ||||
| Body temperature, °C | 37.2 (35.7–38.4)* | 38.0 (36.6–38.9) | 38.0 (36.8–39.1) | 36.7 (34.8–38.7) |
| SIRS criteria | 3.0 (3.0–4.0) | 3.0 (3.0–4.0) | 3.0 (3.0–4.0) | 4.0 (3.3–4.0) |
| ISTH DIC | 4.0 (2.0–5.0)*† | 2.0 (1.0–4.0) | 2.0 (0.0–3.0) | 3.0 (1.3–4.8) |
| SOFA | 10.0 (7.5–13.5)*† | 8.0 (5.0–11.0) | 8.0 (6.0–10.0) | 10.0 (10.0–12.8)* |
| APACHE II | 25.0 (19.0–30.0)* | 22.0 (17.0–28.0) | 24.0 (20.0–31.0) | 27.0 (23.0–32.5) |
| (n = 75) | (n = 321) | (n = 47) | (n = 16) | |
|---|---|---|---|---|
| Day 4 | ||||
| Body temperature, °C | 37.5 (36.7–38.2) | 37.5 (36.8–38.0) | 37.4 (36.7–38.0) | 37.5 (36.8–38.2) |
| SIRS criteria | 2.0 (1.0–3.0) | 2.0 (1.0–3.0) | 2.0 (1.0–3.0) | 2.0 (1.0–3.0) |
| ISTH DIC | 3.0 (1.0–5.0)* | 2.0 (0.0–4.0) | 2.0 (0.75–4.0) | 4.0 (2.0–5.0) |
| SOFA | 8.0 (4.0–11.0)* | 5.0 (3.0–9.0) | 6.0 (3.0–10.0) | 12.0 (8.0–16.0)* |
- Data are presented as median with interquartile range in parentheses. *P < 0.0083 (Kruskal-Wallis test and Mann-Whitney's U-test with Bonferroni correction) when compared with 100–199 mg/dL category, and †compared with 200–299 mgd L-1 category. APACHE, Acute Physiology and Chronic Health Evaluation; ISTH DIC, International Society on Thrombosis and Haemostasis Disseminated Intravascular Coagulation; SIRS, Systemic Inflammatory Response Syndrome; SOFA, Sequential Organ Failure Assessment.
| <100 mg/dL | 100–199 mg/dL | 200–299 mg/dL | ≥300 mg/dL | |
|---|---|---|---|---|
| (n = 12) | (n = 52) | (n = 32) | (n = 26) | |
| Age, years | 76.0 (64.8–78.0) | 70.0 (64.3–80.8) | 70.5 (51.3–76.8) | 68.0 (51.5–79.3) |
| Day 1 | ||||
| Body temperature, °C | 37.9 (36.3–38.7) | 38.1 (37.5–39.0) | 38.0 (36.7–39.0) | 37.2 (35.4–38.5) |
| SIRS criteria | 3.0 (3.0–4.0) | 3.0 (2.0–4.0) | 3.0 (3.0–4.0) | 3.0 (3.0–4.0) |
| ISTH DIC | 3.0 (2.0–4.75) | 3.0 (1.0–4.0) | 2.0 (1.0–3.0) | 3.5 (1.0–4.0) |
| SOFA | 9.5 (7.25–11.0) | 7.5 (5.25–11.75) | 7.0 (2.25–10.0) | 8.0 (5.0–10.0) |
| APACHE II | 21.5 (14.3–29.0) | 21.0 (16.0–28.0) | 21.5 (15.3–25.8) | 25.5 (15.8–32.3) |
| (n = 11) | (n = 47) | (n = 31) | (n = 25) | |
|---|---|---|---|---|
| Day 4 | ||||
| Body temperature, °C | 37.4 (36.8–37.7) | 37.7 (36.8–38.3) | 37.4 (36.9–38.3) | 37.8 (37.1–38.3) |
| SIRS criteria | 2.0 (1.0–3.0) | 2.0 (1.0–3.0) | 2.0 (1.0–3.0) | 2.0 (1.0–3.0) |
| ISTH DIC | 3.0 (1.0–5.0) | 2.0 (1.0–3.0) | 2.0 (1.0–3.0) | 2.5 (1.0–4.0) |
| SOFA | 7.0 (6.0–13.0) | 6.0 (4.0–9.0) | 6.0 (1.0–9.0) | 6.0 (3.0–8.5) |
- Data are presented as median with interquartile range in parentheses. APACHE, Acute Physiology and Chronic Health Evaluation; ISTH DIC, International Society on Thrombosis and Haemostasis Disseminated Intravascular Coagulation; SIRS, Systemic Inflammatory Response Syndrome; SOFA, Sequential Organ Failure Assessment.
Comparison of blood glucose levels and mortality rates in patients with or without pre-existing diabetes
As shown in Table 5, the status of pre-existing diabetes did not have any significant effect on mortality rates, regardless of patients' glucose level on day 1. However, in patients without pre-existing diabetes, the Kaplan–Meier estimate of the probability of survival at 365 days was lower in patients with a blood glucose level of ≥300 mg/dL than in those with levels <100 mg/dL (P = 0.008), 100–199 mg/dL (P < 0.001), or 200–299 mg/dL (P = 0.012). In contrast, in patients with pre-existing diabetes, no significant differences in the probability of survival were found among the four categories (Fig. 1).
| Blood glucose level (mg/dL) on day 1 | Without pre-existing diabetes mellitus† | With pre-existing diabetes mellitus† | P-value‡ |
|---|---|---|---|
| <100 | 27/89 (30.3%) | 2/12 (16.7%) | 0.501 |
| 100–199 | 94/342 (27.5%) | 14/52 (26.9%) | 1.000 |
| 200–299 | 15/50 (30.0%) | 10/32 (31.3%) | 1.000 |
| ≥300 | 11/16 (68.8%) | 10/26 (38.5%) | 0.111 |
- †Data are presented as proportion with percentage in parentheses. ‡Fisher's exact test.
Kaplan–Meier survival curves for patients with severe sepsis in four categories of blood glucose levels (≤100 mg/dL, blue; 100–199 mg/dL, green; 200–299 mg/dL, beige; and ≥300 mg/dL, purple). (A) In patients without pre-existing diabetes mellitus, the probability of survival at 365 days was significantly lower in patients with a blood glucose level ≥300 mg/dL than in those with a level ≤100 mg/dL (P = 0.008), 100–199 mg/dL (P < 0.0001), or 200–299 mg/dL (P = 0.0122). (B) In patients with pre-existing diabetes mellitus, no statistically significant differences in the probability of survival at 365 days were observed among the four categories of blood glucose levels.
When we analyzed the odds ratios for glucose levels on day 1 (relative to the reference range of 100–199 mg/dL) in patients without pre-existing diabetes, we did not find any relationship between hospital mortality and glucose levels ≤299 mg/dL (Table 6). However, there was a significant relationship between hospital mortality and glucose levels ≥300 mg/dL (odds ratio = 4.837; P = 0.003). Multivariate logistic regression analysis also revealed that a blood glucose level ≥300 mg/dL is an independent prognostic factor for hospital mortality in patients without pre-existing diabetes (Table 7).
| Blood glucose level (mg/dL) on day 1 | Hospital mortality† | Unadjusted odds ratio‡ | 95% confidence interval | P-value |
|---|---|---|---|---|
| <100 | 27/89 (30.3%) | 1.149 | 0.690–1.914 | 0.594 |
| 100–199 | 94/342 (27.5%) | 1.000 | (reference) | |
| 200–299 | 15/50 (30.0%) | 1.131 | 0.590–2.165 | 0.332 |
| ≥300 | 11/16 (68.8%) | 4.837 | 1.740–13.449 | 0.003 |
- †Data are presented as proportion with percentage in parentheses. ‡Odds ratios are relative to the reference blood glucose category (100–199 mg/dL), as previously reported.20
| Prognostic factor | Odds ratioa | P-value | 95% CI |
|---|---|---|---|
| Age | 1.022 | 0.005 | 1.006–1.037 |
| Male gender | 1.155 | 0.526 | 0.740–1.804 |
| ISTH DIC score | 1.085 | 0.198 | 0.958–1.228 |
| SOFA score | 1.128 | 0.002 | 1.046–1.217 |
| APACHE II score | 1.057 | 0.001 | 1.024–1.092 |
| Blood glucose ≥300 mg/dL | 4.365 | 0.013 | 1.364–13.972 |
- a Odds ratios are relative to the reference blood glucose category (100–199 mg/dL), as previously reported.20 APACHE, Acute Physiology and Chronic Health Evaluation; CI, confidence interval; ISTH DIC, International Society of Thrombosis and Haemostasis Disseminated Intravascular Coagulation; SOFA, Sequential Organ Failure Assessment.
Discussion
Our findings suggest that abnormal blood glucose levels may affect the disease severity and mortality of patients with severe sepsis, but only in those without pre-existing diabetes. A blood glucose level <100 mg/dL was associated with increased disease severity, whereas a blood glucose level ≥300 mg/dL was associated with increased mortality in patients who do not have pre-existing diabetes.
Impact of hyperglycemia in patients with or without pre-existing diabetes
In our study, the relationship between a glucose level of ≥300 mg/dL and increased mortality was shown in patients without pre-existing diabetes, but not in patients with pre-existing diabetes. Several studies suggested that patients with stress hyperglycemia and no previous diagnosis of diabetes face worse consequences at a given severity of hyperglycemia than those with pre-existing diabetes. A retrospective review, in which patients were stratified as normoglycemia, pre-existing diabetes, or newly diagnosed hyperglycemia, showed that mortality was 18.3 times higher in patients with newly diagnosed hyperglycemia, but only 2.7 times higher in those with known diabetes as compared with normoglycemia.21 Other studies have reported high mortality rates in hyperglycemic patients without known diabetes and an absence of a relationship between hyperglycemia and mortality in patients with diabetes in different conditions.8, 10 Hyperglycemia may have different clinical implications in patients with or without pre-existing diabetes.
Hypoglycemia in patients without pre-existing diabetes
Sepsis is a common cause of hypoglycemia,22 because inflammatory cytokines not only increase glucose utilization but also inhibit gluconeogenesis.23 Hypoglycemia has been suggested to be associated with an increased mortality in critically ill patients.24-26 There is a J-shaped relationship between glucose levels and mortality, such that hyper- and hypoglycemic patients have higher mortality than normoglycemic patients.27, 28 The NICE-SUGAR Study Investigators reported the relationship between hypoglycemia and risk of death.24 The mortality rates in patients who did not have hypoglycemia, moderate hypoglycemia, or severe hypoglycemia were 23.5%, 28.5%, and 35.4%, respectively. The adjusted hazard ratio for death among patients with severe hypoglycemia was 2.10. In other studies, hypoglycemia has been independently associated with increased mortality.25, 26 Although spontaneous hypoglycemia is strongly associated with mortality,29 a causal relationship may be also plausible because hypoglycemia may increase mortality by autonomic impairment, alteration of blood flow, white cell activation, vasoconstriction, and the release of inflammatory mediators.24
There are three possible limitations of our study. First, we may have underestimated the number of patients with diabetes, because previous studies have shown that diabetes is often unrecognized in hospitalized patients.30 Second, although each institution has a protocol of glucose control for sepsis, we used the median value of blood glucose levels and we did not control for insulin therapy. Finally, the lowest blood glucose category in our study (<100 mg/dL) includes some blood glucose measurements that are greater than the usual definition of hypoglycemia (<70 mg/dL). Although this may affect the comparison of our results with those from other studies, we do not think this is a significant limitation because many studies define hypoglycemia differently.24, 25
Conclusion
In patients with severe sepsis, the impact of abnormal blood glucose levels on disease severity and mortality depends on their pre-existing diabetes status. Specifically, in patients who do not have pre-existing diabetes, a blood glucose level <100 mg/dL is associated with increased disease severity, whereas a blood glucose level ≥300 mg/dL is associated with increased mortality. Thus, blood glucose levels may be a useful biomarker to improve therapeutic management of critically ill patients who have severe sepsis, but no history of diabetes mellitus.
Conflict of Interest
None.
Acknowledgments
This study was funded by the Japanese Association for Acute Medicine.
