The diagnosis of diabetes and admission blood glucose concentration are associated with adverse outcome after acute coronary syndromes. We compared the relative association with survival after ST elevation AMI (acute myocardial infarction) of admission blood glucose concentration and of diabetes diagnosis. We carried out a retrospective cohort study in 4702 consecutive patients with STEMI (ST elevation AMI) occurring from 1 April 1993 to 31 December 2005, assessed for mortality at 30 days and 1 year. Patients were classified according to antecedent diabetes and by blood glucose concentration at admission (quartile 1, <7mmol/l; quartile 2, 7–8.2 mmol/l; quartile 3, 8.3–10.9 mmol/l; quartile 4, ≥11 mmol/l). Multivariable models were constructed for determinants of mortality, including year of STEMI and demographic variables, entering blood glucose concentration and antecedent diabetes individually and together. All-cause 30-day and 1-year mortality were 22.8% and 31.3% for patients with antecedent diabetes, compared with 16.3% and 23.0% respectively for those without. For glucose quartiles 1, 2, 3 and 4, crude 30-day mortality was 9.0%, 10.6%, 17.9% and 31.0%. Adjusted 30-day mortality risk was similar in quartile 2, higher by >80% in quartile 3 and by >150% in quartile 4, compared with glucose quartile 1. Antecedent diabetes was associated with an increase in mortality [unadjusted odds ratio (OR) 1.52 (95% CI 1.24, 1.86)]. On multivariable analysis (excluding glucose quartile), this reduced to 1.24 (0.98, 1.58) and changed to a small, statistically non-significant reduction in risk when glucose quartile was added to the analysis [adjusted OR 0.87 (0.67, 1.13)]. Inclusion of antecedent diabetes in multivariable models did not add to the predictive value for mortality of glucose quartile (P=0.368). Similar relationships were observed for 1 year mortality. In patients with STEMI, blood glucose concentration shows graded association with risk of 30-day and 1-year mortality and is of greater prognostic relevance than antecedent diabetes diagnosis. Moderate elevation of blood glucose, below levels previously considered to be clinically relevant, is associated with adverse impact on survival.

INTRODUCTION

Abnormalities of glucose metabolism have powerful association with adverse prognosis for patients with coronary heart disease. In the setting of AMI (acute myocardial infarction), the diagnosis of diabetes is associated with adverse outcome [1,2]. In addition, elevated blood glucose (‘stress hyperglycaemia’) is common among patients hospitalised with AMI, irrespective of diabetes status, and is also associated with adverse outcome [27].

In most studies considering the influence of diabetes on prognosis after AMI, patients have been categorized based upon a history of the condition prior to the index event, an approach likely to underestimate prevalence. Similarly, studies assessing the association with prognosis of hyperglycaemia have, for the most part, considered blood glucose concentration as a dichotomized variable, using varying cut-off values [5,8,9]. Those studies considering blood glucose as a graded variable have been in selected cohorts, considering only patients with prior diabetes diagnosis [3,7,18] or aged >65 years [6], or used fasting glucose [9,10] or glycated haemoglobin [11].

It remains unclear whether it is diabetes per se or the severity of the acute glycaemic response at the time of the event, which impacts on prognosis [4]. Indeed, the relative influence upon outcome after AMI of these factors has not been assessed in routine practice. The aim of the current study was to assess the relative association with survival after STEMI (ST elevation AMI) of antecedent diabetes diagnosis and of admission blood glucose concentration.

MATERIALS AND METHODS

Patients

Data are from consecutive admissions (1 January 1993 to 31 December 2005) to the coronary care unit of a large teaching hospital (Leicester Royal Infirmary), one of two serving the population of Leicestershire, U.K. (approximately 946000 residents in 2004). For all patients, we record routine clinical and demographic data including information on diagnosis [STEMI/NSTEMI (non-STEMI)], ECG site of infarct and details of medical history, coronary heart disease risk factors and medication prescribed prior to and during admission. Mortality is recorded prospectively, as described previously [12]. For patients with multiple admissions, we considered data pertaining only to the first. Predefined outcome measures were the individual and relative strength of association with 30-day and 1-year, all-cause mortality of antecedent diabetes and admission blood glucose concentration.

Patient identification

On a background of changes during the study period in the definition of AMI, we maintained consistent inclusion criteria by restricting the analysis to patients with STEMI. The diagnosis required (i) ECG evidence of dynamic ST segment elevation together with (ii) appropriate symptoms and (iii) an increase in serum levels of CK (creatine kinase) to greater than twice the upper limit of the laboratory reference range (i.e. >400 international units/l). Troponin concentrations were not available for the full study period. Patients were categorized according to (i) a pre-existing diagnosis of diabetes (reported by the patient, or on the basis of prescribed medication), and (ii) by blood glucose concentration at the time of index admission.

The data used in this analysis were gathered during routine care of the patients. Although predating the Myocardial Ischaemia National Audit Project, our database is that used by our hospital for this purpose. In this context, data collection does not require individual patient consent. The project was approved by the local research ethics committee.

Statistical analysis

Baseline differences between groups were examined using independent two-sample Student's t tests for continuous variables and χ2 tests for categorical variables. Data are presented as differences in means and proportions, with 95% CI (confidence intervals). We calculated 30-day mortality proportions for the complete population with follow-up censored at 30 January 2006. One-year mortality proportions were calculated for patients admitted (n=4876) up to 31 December 2004, the most recent date at which complete 1-year follow-up data were available.

The association between year of index AMI and all-cause mortality was assessed using unconditional logistic regression analysis with covariate effects reported as OR (odds ratios) with 95% CI. An interaction term representing year of admission was included to explore the impact of temporal trends in the management of ACS (acute coronary syndrome) [13,14]. As indications for, and proportions of patients receiving (70–77% annually) thrombolysis, were consistent throughout the study period, this variable was included in the model. In all cases, a linear effect for year of admission (log odds scale) provided the best fitting model. Other continuous variables (age, glucose, creatinine and peak CK) showed non-linear relationships with outcome and for initial univariate analyses were categorized to ease interpretation. Glucose, creatinine and peak CK were grouped by quartile, and age was divided into <65, 65–74 and ≥75 years.

The blood glucose concentration used was that first recorded for the index admission, assayed as part of routine investigations. Glucose was divided by quartiles of the range observed over the study period: quartile 1, <7 mmol/l; quartile 2, 7–8.2 mmol/l; quartile 3, 8.3–10.9 mmol/l; quartile 4, ≥11 mmol/l. We initially assessed the unadjusted, univariate association with mortality of antecedent diabetes and of glucose quartile. Differential effects over time were assessed by fitting interactions between these covariates and year of diagnosis. We then fitted models adjusted for age, sex, peak CK, creatinine, previous AMI, smoking status, year of index AMI and administration of thrombolysis. Fractional polynomials were used to model continuous variables to allow for potential non-linearity. To quantify the effects of antecedent diabetes and of glucose quartile, separate models were fitted including these individually, and in combination. Demographic features and outcomes were assessed for the subcohort of patients for whom admission glucose was not recorded. Analyses were performed using Stata 9 (StataCorp. 2005. Stata Statistical Software: Release 9, College Station, TX, U.S.A.).

RESULTS

From 1 January 1993 to 31 December 2004, we recorded 4876 consecutive, index admissions with STEMI. Excluding from analysis 174 (3.6%) individuals not normally residents locally, the study population consisted of 4702 patients (3198, 68.0% male; mean age 66.7, S.D. 12.7, range 21–107 years). Female (72.2±11.4 years) were older than male patients (64.1±12.4, P<0.001). For 683 (14.0%) and two patients respectively details of admission blood glucose concentration and antecedent diabetes status were not available.

Antecedent diabetes

Antecedent diabetes was recorded for 749 (15.9%) patients, increasing from 13.9% in 1993 to 20.3% in 2005. Patients with antecedent diabetes were older by an average of 1.4 years and more often female (Table 1). Mean admission glucose and creatinine concentrations were higher, but peak CK lower, in these patients, who were less likely to receive thrombolysis or to be prescribed aspirin or β-blocker during the index admission, but more likely to receive diuretic therapy, inhibitors of the renin-angiotensin system and insulin.

Table 1
Demographic and in-hospital treatment characteristics of patients with and without antecedent diagnosis of diabetes

Values are means±S.E.M. or numbers (percentage). *Calculated difference in proportions (with 95% CIs); †calculated difference in means (with 95% CIs). 1101 missing values for diabetes and 581 missing values for no diabetes; 2nine missing values for diabetes and 39 missing values for no diabetes; 3nine missing values for diabetes and 39 missing values for no diabetes. ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker.

 Antecedent diabetes  
Characteristic Yes [n=749 (15.9%)] No [n=3951 (84.1%)] Difference (diabetes–no diabetes) 
Male (n460 (61.4%) 2737 (69.3%) −7.9 (−11.6, −4.1)* 
Age (years) 67.9±0.4 66.5±0.2 1.4 (0.5, 2.3)† 
Plasma glucose (mmol/l)1 14.7±0.3 8.9±0.1 5.7 (5.2, 6.3)† 
Creatinine (μmol/l)2 120.7±2.3 109.2±0.7 11.5 (6.8, 16.3)† 
Peak CK (IU/l)3 (normal range <200) 1818±62 2115±31 −296 (−433, −159)† 
History of    
 Smoking    
  Current (n156 (20.8%) 1312 (33.2%) −12.4 (−15.7, −9.1)* 
  Former (n183 (24.4%) 992 (25.1%) −0.7 (−4.0, 2.7)* 
  Never (n410 (54.7%) 1645 (41.6%) 13.1 (9.2, 17.0)* 
  Not known (n 
 Previous AMI (n175 (23.4%) 571 (14.5%) 8.9 (5.7, 12.1)* 
 Angina (n230 (30.7%) 862 (21.8%) 8.9 (5.3, 12.4)* 
 Hypertension (n368 (49.1%) 1296 (32.8%) 16.3 (12.5, 20.2)* 
 Hyperlipidaemia (n151 (20.2%) 451 (11.4%) 8.7 (5.7, 11.8)* 
 Cerebrovascular disease (n83 (11.1%) 191 (4.8%) 6.2 (3.9, 8.6)* 
Treatment    
 Thrombolysis (n468 (62.5%) 2883 (73.0%) −10.5 (−14.2, −6.8)* 
 Diuretic (n457 (61.0%) 1743 (44.1%) 16.9 (13.1, 20.7)* 
 Aspirin (n528 (70.5%) 3099 (78.4%) −7.9 (−11.5, –4.4)* 
 β-Blocker (n282 (37.7%) 1798 (45.5%) −7.9 (−11.7, –4.1)* 
 Insulin (n510 (68.1%) 280 (7.1%) 61.0 (57.6, 64.4)* 
 ACEI/ARB (n385 (51.4%) 1580 (40.0%) 11.4 (7.5, 15.3)* 
 Statin (n196 (26.2%) 1054 (26.7%) −0.5 (−3.9, 2.9)* 
 Antecedent diabetes  
Characteristic Yes [n=749 (15.9%)] No [n=3951 (84.1%)] Difference (diabetes–no diabetes) 
Male (n460 (61.4%) 2737 (69.3%) −7.9 (−11.6, −4.1)* 
Age (years) 67.9±0.4 66.5±0.2 1.4 (0.5, 2.3)† 
Plasma glucose (mmol/l)1 14.7±0.3 8.9±0.1 5.7 (5.2, 6.3)† 
Creatinine (μmol/l)2 120.7±2.3 109.2±0.7 11.5 (6.8, 16.3)† 
Peak CK (IU/l)3 (normal range <200) 1818±62 2115±31 −296 (−433, −159)† 
History of    
 Smoking    
  Current (n156 (20.8%) 1312 (33.2%) −12.4 (−15.7, −9.1)* 
  Former (n183 (24.4%) 992 (25.1%) −0.7 (−4.0, 2.7)* 
  Never (n410 (54.7%) 1645 (41.6%) 13.1 (9.2, 17.0)* 
  Not known (n 
 Previous AMI (n175 (23.4%) 571 (14.5%) 8.9 (5.7, 12.1)* 
 Angina (n230 (30.7%) 862 (21.8%) 8.9 (5.3, 12.4)* 
 Hypertension (n368 (49.1%) 1296 (32.8%) 16.3 (12.5, 20.2)* 
 Hyperlipidaemia (n151 (20.2%) 451 (11.4%) 8.7 (5.7, 11.8)* 
 Cerebrovascular disease (n83 (11.1%) 191 (4.8%) 6.2 (3.9, 8.6)* 
Treatment    
 Thrombolysis (n468 (62.5%) 2883 (73.0%) −10.5 (−14.2, −6.8)* 
 Diuretic (n457 (61.0%) 1743 (44.1%) 16.9 (13.1, 20.7)* 
 Aspirin (n528 (70.5%) 3099 (78.4%) −7.9 (−11.5, –4.4)* 
 β-Blocker (n282 (37.7%) 1798 (45.5%) −7.9 (−11.7, –4.1)* 
 Insulin (n510 (68.1%) 280 (7.1%) 61.0 (57.6, 64.4)* 
 ACEI/ARB (n385 (51.4%) 1580 (40.0%) 11.4 (7.5, 15.3)* 
 Statin (n196 (26.2%) 1054 (26.7%) −0.5 (−3.9, 2.9)* 

Admission blood glucose

Table 2 shows characteristics of patients according to glucose quartile and for the subcohort in whom admission blood glucose was not recorded. Even in quartiles 3 (10.3%) and 4 (42.7%) of blood glucose, a minority of patients had antecedent diabetes.

Table 2
Demographic and in-hospital treatment characteristics of patients according to glucose quartile

†One-year survival was assessed on patients admitted up to 31 December 2004, being the last date for which 1-year follow-up was available. Therefore the total number of patients included in the 1-year analysis (n=4474) is fewer than in the 30-day analysis (n=4702). ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin receptor blocker; EQ, glucose quartile; SBP, systolic blood pressure.

(a) 
  GQ1 <7 GQ2 7–8.3 GQ3 8.3–11 GQ4 ≥11 Missing glucose Overall 
Characteristic  n n n n n n 
Gender Male 713 76.7 776 72.1 609 63.9 643 60.7 457 66.9 3198 68.0 
 Female 217 23.3 300 27.9 344 36.1 417 39.3 226 33.1 1504 32.0 
Diabetes No 882 94.8 1027 95.4 855 89.7 606 57.2 581 85.1 3951 84.0 
 Yes 48 5.2 49 4.6 98 10.3 453 42.7 101 14.8 749 15.9 
 Missing 0.1 0.1 0.04 
Statin No 567 61.0 803 74.6 714 74.9 859 81.0 508 74.4 3451 73.4 
 Yes 363 39.0 273 25.4 239 25.1 201 19.0 175 25.6 1251 26.6 
ACEI or ARB No 499 53.7 655 60.9 547 57.4 608 57.4 427 62.5 2736 58.2 
 Yes 431 46.3 421 39.1 406 42.6 452 42.6 256 37.5 1966 41.8 
Diuretic No 596 64.1 645 59.9 483 50.7 418 39.4 360 52.7 2502 53.2 
 Yes 334 35.9 431 40.1 470 49.3 642 60.6 323 47.3 2200 46.8 
β-Blocker No 373 40.1 578 53.7 534 56.0 733 69.2 403 59.0 2621 55.7 
 Yes 557 59.9 498 46.3 419 44.0 327 30.8 280 41.0 2081 44.3 
Thrombolysis No 272 29.2 248 23.0 193 20.2 408 38.5 229 33.5 1350 28.7 
 Yes 658 70.8 828 77.0 760 79.8 652 61.5 454 66.5 3352 71.3 
Insulin No 895 96.2 1046 97.2 877 92.0 493 46.5 600 87.9 3911 83.2 
 Yes 35 3.8 30 2.8 76 8.0 567 53.5 83 12.2 791 16.8 
(b) 
  GQ1 <7 GQ2 7–8.3 GQ3 8.3–11 GQ4 ≥11 Missing glucose Overall 
 n Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. 
Age 4702 63.8 13.27 65.3 12.88 67.6 12.10 69.4 11.54 67.4 13.06 66.7 12.69 
SBP 4627 141.8 28.06 139.8 28.49 137.6 31.71 135.4 36.48 137.2 31.02 138.4 31.44 
Creatinine (μmol/l) 4654 103.3 47.33 102.6 35.78 109.1 38.73 126.3 56.83 113.8 5.64 111.0 48.19 
(a) 
  GQ1 <7 GQ2 7–8.3 GQ3 8.3–11 GQ4 ≥11 Missing glucose Overall 
Characteristic  n n n n n n 
Gender Male 713 76.7 776 72.1 609 63.9 643 60.7 457 66.9 3198 68.0 
 Female 217 23.3 300 27.9 344 36.1 417 39.3 226 33.1 1504 32.0 
Diabetes No 882 94.8 1027 95.4 855 89.7 606 57.2 581 85.1 3951 84.0 
 Yes 48 5.2 49 4.6 98 10.3 453 42.7 101 14.8 749 15.9 
 Missing 0.1 0.1 0.04 
Statin No 567 61.0 803 74.6 714 74.9 859 81.0 508 74.4 3451 73.4 
 Yes 363 39.0 273 25.4 239 25.1 201 19.0 175 25.6 1251 26.6 
ACEI or ARB No 499 53.7 655 60.9 547 57.4 608 57.4 427 62.5 2736 58.2 
 Yes 431 46.3 421 39.1 406 42.6 452 42.6 256 37.5 1966 41.8 
Diuretic No 596 64.1 645 59.9 483 50.7 418 39.4 360 52.7 2502 53.2 
 Yes 334 35.9 431 40.1 470 49.3 642 60.6 323 47.3 2200 46.8 
β-Blocker No 373 40.1 578 53.7 534 56.0 733 69.2 403 59.0 2621 55.7 
 Yes 557 59.9 498 46.3 419 44.0 327 30.8 280 41.0 2081 44.3 
Thrombolysis No 272 29.2 248 23.0 193 20.2 408 38.5 229 33.5 1350 28.7 
 Yes 658 70.8 828 77.0 760 79.8 652 61.5 454 66.5 3352 71.3 
Insulin No 895 96.2 1046 97.2 877 92.0 493 46.5 600 87.9 3911 83.2 
 Yes 35 3.8 30 2.8 76 8.0 567 53.5 83 12.2 791 16.8 
(b) 
  GQ1 <7 GQ2 7–8.3 GQ3 8.3–11 GQ4 ≥11 Missing glucose Overall 
 n Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. 
Age 4702 63.8 13.27 65.3 12.88 67.6 12.10 69.4 11.54 67.4 13.06 66.7 12.69 
SBP 4627 141.8 28.06 139.8 28.49 137.6 31.71 135.4 36.48 137.2 31.02 138.4 31.44 
Creatinine (μmol/l) 4654 103.3 47.33 102.6 35.78 109.1 38.73 126.3 56.83 113.8 5.64 111.0 48.19 

Patients with glucose in quartile 4 were on average 5.6 years older than those in quartile 1. Higher glucose quartile was associated with higher serum creatinine and lower admission systolic blood pressure. Thrombolysis was administered in fewer patients in quartile 4 compared with quartiles 1–3. In-hospital diuretic prescription increased, and β-blocker prescription fell, as glucose quartile increased. There was little variation among quartiles in the use of inhibitors of the renin–angiotensin system.

In patients for whom admission blood glucose was not recorded, demographic and treatment features were broadly similar to those seen in the population as a whole. In terms of proportions with antecedent diabetes, treatment with insulin and case fatality, these patients most closely resembled those in quartile 3 (Table 2).

Survival

During follow-up, case fatality was 42.2% (1992/4702). By 30 days, 841 (17.9% of the population, 42.2% of all events), and by 1 year, 1136 (24.2% of the population, 57.0% of events) deaths had occurred. The univariate strengths of association with mortality for important clinical and demographic variables are shown in Table 3.

Table 3
Univariate association with 30-day and all-cause mortality
  30-Day mortality 1-Year mortality 
Variable  No. deaths/no. at risk (%) Odds ratio (95% CI) No. deaths/no. at risk (%) Odds ratio (95% CI) 
Antecedent diabetes No 550/3370 (16.3) 743/3225 (23.0) 
 Yes 148/648 (22.8) 1.52 (1.24, 1.86) 192/614 (31.3) 1.52 (1.24, 1.84) 
Gender Female 316/1278 (24.7) 403/1224 (32.9) 
 Male 382/2740 (13.9) 0.49 (0.42, 0.58) 532/2615 (20.3) 0.52 (0.45, 0.61) 
Thrombolysis No 307/1120 (27.4) 405/1065 (38.0) 
 Yes 391/2898 (13.5) 0.41 (0.35, 0.49) 530/2774 (19.1) 0.38 (0.33, 0.45) 
Previous MI No 541/3388 (16.0) 704/3231 (21.7) 
 Yes 157/630 (24.9) 1.75 (1.43, 2.14) 231/608 (38.0) 2.20 (1.83, 2.64) 
Age (years) <65 117/1669 (7.0) 150/1574 (9.5) 
 65–74 187/1153 (16.2) 2.57 (2.01, 3.02) 256/1111 (23.0) 2.84 (2.28, 3.54) 
 ≥75 394/1196 (32.9) 6.52 (5.21, 8.15) 529/1154 (45.8) 8.04 (6.55, 9.86) 
Glucose (mmol/l, quartile) <7.3 84/930 (9.0) 122/871 (14.0) 
 7.3–8.2 114/1076 (10.6) 1.20 (0.89, 1.61) 180/1037 (17.4) 1.29 (1.00, 1.66) 
 8.3–10.9 171/953 (17.9) 2.20 (1.67, 2.91) 228/910 (25.1) 2.05 (1.61, 2.62) 
 ≥11 329/1059 (31.1) 4.54 (3.50, 5.88) 405/1021 (39.7) 4.04 (3.21, 5.07) 
Creatinine (μmol/l, quartile) <87 80/972 (8.2) 114/927 (12.3) 
 87–100.9 97/1032 (9.4) 1.16 (0.85, 1.58) 135/971 (13.9) 1.15 (0.88, 1.50) 
 101–120.9 139/1000 (13.9) 1.80 (1.35, 2.41) 203/961 (21.1) 1.91 (1.49, 2.45) 
 ≥121 381/1009 (37.8) 6.76 (5.21, 8.79) 483/976 (49.5) 6.99 (5.54, 8.82) 
Peak CK (IU/l, quartile) <752 253/997 (25.4) 299/927 (32.3) 
 752–1567 131/989 (13.3) 0.45 (0.36, 0.57) 196/940 (20.9) 0.55 (0.45, 0.68) 
 1568–2808 140/1000 (14.0) 0.48 (0.38, 0.60) 202/971 (20.8) 0.55 (0.45, 0.68) 
 ≥2809 168/1019 (16.5) 0.58 (0.47, 0.72) 232/990 (23.4) 0.64 (0.53, 0.79) 
Year Annual 0.94 (0.92, 0.96) 0.95 (0.93, 0.97)   
  30-Day mortality 1-Year mortality 
Variable  No. deaths/no. at risk (%) Odds ratio (95% CI) No. deaths/no. at risk (%) Odds ratio (95% CI) 
Antecedent diabetes No 550/3370 (16.3) 743/3225 (23.0) 
 Yes 148/648 (22.8) 1.52 (1.24, 1.86) 192/614 (31.3) 1.52 (1.24, 1.84) 
Gender Female 316/1278 (24.7) 403/1224 (32.9) 
 Male 382/2740 (13.9) 0.49 (0.42, 0.58) 532/2615 (20.3) 0.52 (0.45, 0.61) 
Thrombolysis No 307/1120 (27.4) 405/1065 (38.0) 
 Yes 391/2898 (13.5) 0.41 (0.35, 0.49) 530/2774 (19.1) 0.38 (0.33, 0.45) 
Previous MI No 541/3388 (16.0) 704/3231 (21.7) 
 Yes 157/630 (24.9) 1.75 (1.43, 2.14) 231/608 (38.0) 2.20 (1.83, 2.64) 
Age (years) <65 117/1669 (7.0) 150/1574 (9.5) 
 65–74 187/1153 (16.2) 2.57 (2.01, 3.02) 256/1111 (23.0) 2.84 (2.28, 3.54) 
 ≥75 394/1196 (32.9) 6.52 (5.21, 8.15) 529/1154 (45.8) 8.04 (6.55, 9.86) 
Glucose (mmol/l, quartile) <7.3 84/930 (9.0) 122/871 (14.0) 
 7.3–8.2 114/1076 (10.6) 1.20 (0.89, 1.61) 180/1037 (17.4) 1.29 (1.00, 1.66) 
 8.3–10.9 171/953 (17.9) 2.20 (1.67, 2.91) 228/910 (25.1) 2.05 (1.61, 2.62) 
 ≥11 329/1059 (31.1) 4.54 (3.50, 5.88) 405/1021 (39.7) 4.04 (3.21, 5.07) 
Creatinine (μmol/l, quartile) <87 80/972 (8.2) 114/927 (12.3) 
 87–100.9 97/1032 (9.4) 1.16 (0.85, 1.58) 135/971 (13.9) 1.15 (0.88, 1.50) 
 101–120.9 139/1000 (13.9) 1.80 (1.35, 2.41) 203/961 (21.1) 1.91 (1.49, 2.45) 
 ≥121 381/1009 (37.8) 6.76 (5.21, 8.79) 483/976 (49.5) 6.99 (5.54, 8.82) 
Peak CK (IU/l, quartile) <752 253/997 (25.4) 299/927 (32.3) 
 752–1567 131/989 (13.3) 0.45 (0.36, 0.57) 196/940 (20.9) 0.55 (0.45, 0.68) 
 1568–2808 140/1000 (14.0) 0.48 (0.38, 0.60) 202/971 (20.8) 0.55 (0.45, 0.68) 
 ≥2809 168/1019 (16.5) 0.58 (0.47, 0.72) 232/990 (23.4) 0.64 (0.53, 0.79) 
Year Annual 0.94 (0.92, 0.96) 0.95 (0.93, 0.97)   

Antecedent diabetes and survival

Overall 30-day mortality was 22.8%, and 1-year mortality was 31.3% for patients with antecedent diabetes, compared with 16.3% and 23.0% respectively for those without. Antecedent diabetes was associated with a univariate, unadjusted increase in the odds of death of approximately 50%, at both 30 days (OR 1.52, 95% CI 1.24, 1.86) and 1 year (OR 1.52, 95% CI 1.24, 1.84) (Table 3).

Blood glucose and survival

Figure 1 shows the fully adjusted odds of mortality by 30 days and admission blood glucose considered as a continuous variable.

Blood glucose above the median (8.3 mmol/l) was associated with increased mortality (Table 3). For patients with glucose in quartile 4 (≥11 mmol/l), 30-day mortality (31%) was over 3-fold higher than in quartile 1 (9.0%). For 30-day mortality, compared with quartile 1, the OR for quartile 2 was 1.20 (95% CI 0.89, 1.61), for quartile 3 it was 2.20 (95% CI 1.67, 2.91) and for quartile 4 it was 4.54 (95% CI 3.50, 5.88).

Adjusted odds of 30-day mortality according to admission blood glucose concentration

Similar associations were evident for 1-year mortality. Compared with quartile 1, the OR for quartile 2 was 1.29 (95% CI 1.00, 1.66), for quartile 3 it was 2.05 (95% CI 1.61, 2.62) and for quartile 4 it was 4.04 (95% CI 3.21, 5.07).

Glucose concentration and antecedent diabetes: relative impact on prognosis

Using multivariable, unconditional logistic regression models, we assessed the association with prognosis of antecedent diabetes, glucose quartile and year of index AMI (Table 4). Unadjusted analyses for diabetes and glucose quartile were followed by analysis adjusted for age, sex, previous MI, peak CK, creatinine, thrombolysis, smoking status and year of index AMI, considering diabetes or glucose separately in individual analyses. Finally, modelling was carried out adjusting for these covariables and including both diabetes and glucose.

Table 4
Results of modelling proportions surviving to 30 days and 1 year

Adjusted models are adjusted for age at MI, gender, previous AMI, CK, creatinine, thrombolysis, smoking status and year of hospitalization: 1for blood glucose concentration (quartile); 2antecedent diabetes and 3for both blood glucose concentration (quartile) and antecedent diabetes. Year, odds ratio per year 1993–2005 (30-day) or 1993–2004 (1-year); AIC, Akaike's information criterion (lower values indicate a better fitting model); Q, quartile.

(a) Surviving to 30 days 
Parameter  Unadjusted (n=4018) Adjusted1 (excluding diabetes) (n=4001) Adjusted2 (excluding glucose) (n=4001) Adjusted3 (n=4001) 
Glucose Q1 – 
 Q2 1.19 (0.89, 1.61) 1.07 (0.77, 1.49) – 1.07 (0.77, 1.49) 
 Q3 2.20 (1.67, 2.91) 1.70 (1.20, 2.37) – 1.73 (1.27, 2.36) 
 Q4 4.54 (3.50, 5.88) 2.458 (1.80, 3.32) – 2.58 (1.90, 3.52) 
Diabetes No – 
 Yes 1.52 (1.24, 1.87) – 1.24 (0.98, 1.58) 0.87 (0.67, 1.13) 
Year  0.937 (0.92, 0.96) 0.936 (0.90, 0.92) 0.93 (0.90, 0.95) 0.94 (0.91, 0.96) 
AIC   0.716135 0.7286741 0.7163706 
(b) Surviving to 1 year 
Parameter  Unadjusted (n=3839) Adjusted (n=3825) Adjusted (n=3825) Adjusted1 (n=3825) 
Glucose Q1 – 
 Q2 1.29 (1.00, 1.66) 1.24 (0.93, 1.65) – 1.24 (0.93, 1.65) 
 Q3 2.05 (1.61, 2.62) 1.64 (1.24, 2.17) – 1.64 (1.24, 2.17) 
 Q4 4.04 (3.21, 5.07) 2.28 (1.74, 2.99) – 2.34 (1.76, 3.11) 
Diabetes No – 
 Yes 1.52 (1.26, 1.84) – 1.24 (0.99, 1.57) 0.93 (0.73, 1.20) 
Year  0.93 (0.93, 0.97) 0.95 (0.93, 0.98) 0.94 (0.92, 0.97) 0.95 (0.93, 0.98) 
AIC   0.841869 0.8519422 0.8423143 
(a) Surviving to 30 days 
Parameter  Unadjusted (n=4018) Adjusted1 (excluding diabetes) (n=4001) Adjusted2 (excluding glucose) (n=4001) Adjusted3 (n=4001) 
Glucose Q1 – 
 Q2 1.19 (0.89, 1.61) 1.07 (0.77, 1.49) – 1.07 (0.77, 1.49) 
 Q3 2.20 (1.67, 2.91) 1.70 (1.20, 2.37) – 1.73 (1.27, 2.36) 
 Q4 4.54 (3.50, 5.88) 2.458 (1.80, 3.32) – 2.58 (1.90, 3.52) 
Diabetes No – 
 Yes 1.52 (1.24, 1.87) – 1.24 (0.98, 1.58) 0.87 (0.67, 1.13) 
Year  0.937 (0.92, 0.96) 0.936 (0.90, 0.92) 0.93 (0.90, 0.95) 0.94 (0.91, 0.96) 
AIC   0.716135 0.7286741 0.7163706 
(b) Surviving to 1 year 
Parameter  Unadjusted (n=3839) Adjusted (n=3825) Adjusted (n=3825) Adjusted1 (n=3825) 
Glucose Q1 – 
 Q2 1.29 (1.00, 1.66) 1.24 (0.93, 1.65) – 1.24 (0.93, 1.65) 
 Q3 2.05 (1.61, 2.62) 1.64 (1.24, 2.17) – 1.64 (1.24, 2.17) 
 Q4 4.04 (3.21, 5.07) 2.28 (1.74, 2.99) – 2.34 (1.76, 3.11) 
Diabetes No – 
 Yes 1.52 (1.26, 1.84) – 1.24 (0.99, 1.57) 0.93 (0.73, 1.20) 
Year  0.93 (0.93, 0.97) 0.95 (0.93, 0.98) 0.94 (0.92, 0.97) 0.95 (0.93, 0.98) 
AIC   0.841869 0.8519422 0.8423143 

As noted above, unadjusted odds of mortality by 30 days and 1 year was approximately 50% higher for patients with antecedent diabetes compared with patients without this diagnosis.

Adjustment for covariates abolished the estimated early and late mortality risk associated with an antecedent diagnosis of diabetes, and the addition of blood glucose concentration did not contribute further to the model build (Table 4).

Although attenuated by covariables adjustment, higher blood glucose concentrations retained association with mortality. Compared with quartile 1, adjusted risk was similar in quartile 2, and approximately 75% and 150% higher in quartile 3 and 4 respectively. The addition of antecedent diabetes did not materially alter the OR associated with the degree of glycaemia (P=0.368) or the overall model fit (Table 4).

DISCUSSION

Key findings

This is the first study to assess, in a large, unselected cohort of patients with ACS, the relative impact upon prognosis of pre-existing diabetes diagnosis and of admission glucose level. Our study presents two novel, clinically important findings. First, blood glucose concentration at admission to hospital with STEMI had more powerful association with prognosis than did antecedent diabetes. Even minor elevation of blood glucose concentration above the normal range was associated with adverse impact on outcome. Secondly, the adverse impact upon mortality of antecedent diabetes was markedly attenuated when adjusted for covariables and was completely absent when correction included admission blood glucose. In contrast, the prior diagnosis of diabetes had no meaningful impact upon the risk associated with glucose concentration at the time of index admission.

Our data lend indirect support to the recent statement from the American Heart Association that optimum blood glucose levels after AMI have yet to be established and should be the focus of future clinical trials [15]. While we observed little variation in outcome for blood glucose below 8.2 mmol/l, mortality risk was increased above this level. This may be considered in the context of the target, and achieved, concentrations in trials of intensive glycaemic control after AMI. These studies recruited patients with admission glucose >11 mmol/l [16,17], had target levels of 7–10.9 mmol/l [16] or 7–10 mmol/l [17] and achieved mean glucose levels of 9-10 mmol/l [16,17]. In a previous study, mortality was lowest in patients for whom blood glucose at 24 h after AMI was below 7 mmol/l [18]. Taken together, these observations suggest that in patients with STEMI, both the threshold at which active intervention is applied and target blood glucose levels may be lower than previously considered.

The benefit of active lowering of blood glucose after AMI has been demonstrated in a number of studies [810,19], and greater falls in blood glucose in the period following admission with AMI is associated with better survival [16,18,19]. In both DIGAMI (Diabetes Mellitus, Insulin Glucose Infusion in Acute Myocardial Infarction) studies, effective glucose lowering soon after AMI was associated with improved survival [16,17], and recently, prescription of insulin in patients with AMI, not previously known to have diabetes, is associated with improved outcome [20]. It is noteworthy that the only controlled study of glucose lowering after AMI, which achieved lower blood glucose compared with standard management, was the first DIGAMI study [16]; this was also the only such study in which active intervention was associated with improved outcome.

It may be argued that some of our cohort, patients in quartile 4, for example, should be considered as having diabetes. However, in many patients with hyperglycaemia at admission with AMI, the diagnosis of diabetes is not confirmed on formal testing at a later time [21]. Our study was designed as a pragmatic assessment of the impact on prognosis of admission blood glucose values. Moreover, if the degree of dysglycaemia impacts directly on prognosis, it is this, rather than later diagnosis of diabetes, to which the clinician should respond.

Our observation of associations between higher blood glucose and a number of markers of adverse prognosis, e.g. lower blood pressure, higher creatinine and greater age, may support the contention that hyperglycaemia is only a marker of adverse prognosis. However, hyperglycaemia may be directly detrimental to the ischaemic myocardium through a variety of mechanisms, including free radical release and endothelial dysfunction [22], exaggerated reperfusion injury [23], the abolition of ischaemic preconditioning [24] and increased oxygen demand due to reliance on free fatty acid metabolism [25]. Further evidence for direct myocardial toxicity of hyperglycaemia during ischaemia comes from a recent report in a large cohort of patients undergoing CABG (coronary artery bypass grafting), in which suboptimal perioperative blood glucose control (≥11.1 mmol/l) showed graded association with increased risk of adverse outcome, irrespective of diabetes status [26]. Overall, these data are strongly suggestive of at least some direct toxic influence of elevated blood glucose in ischaemic myocardium.

Study limitations

Our database lacks information on some important variables, including objective assessment of left ventricular function and regarding clinical evidence of heart failure. Although information on blood glucose was missing in a proportion of patients, the distribution of demographics and outcomes were similar between those with and without a glucose measure, making it unlikely that the missing information would bias our results. In a proportion of our population, elevated blood glucose undoubtedly represents hitherto unrecognized diabetes, a limitation applying to the vast majority of studies classifying patients based upon antecedent diagnosis. The blood glucose concentrations considered were measured on admission and thus represent non-fasting measurements made at varying times of the day and from the onset of the index event. However, this measurement is the one available to physicians soon after admission and is the only measurement upon which early therapeutic decisions may be based. We did not adjust for prescription of individual secondary prevention therapies, the impact of which was not the focus of this analysis.

Consideration of individual treatment effects may introduce bias for a number of reasons. In the most severely ill patients, early mortality and adverse clinical features in survivors will limit treatment prescription. Our study is limited by assessment of survival up to 1 year, and consideration of the influence of dysglycaemia and diabetes on longer term prognosis would be appropriate. We have no information on therapy or interventions after discharge. However, as the risk of death was greatest in the first 30 days, it is unlikely that changes after discharge impacted on overall outcome in a major way. Finally, our findings are based upon a historical cohort, admitted over a prolonged period during which the management of ACS evolved considerably. While we have corrected for such changes by inclusion of year of AMI in multivariable analysis, this may not have adjusted fully for such changes.

In summary, admission blood glucose concentration is a powerful, readily available marker of adverse outcome after STEMI and is prognostically more informative than consideration of antecedent diabetes status. Minor elevation of blood glucose outwith the normal range is associated with adverse impact upon survival. Admission blood glucose concentration should be considered in the early assessment of prognosis after AMI. Further studies of intensive blood glucose management after AMI are merited.

FUNDING

C. P. N. was funded by a British Heart Foundation Ph.D. studentship (FS/05/080/19415).

Abbreviations

     
  • ACS

    acute coronary syndrome

  •  
  • AMI

    acute myocardial infarction

  •  
  • CI

    confidence intervals

  •  
  • CK

    creatine kinase

  •  
  • OR

    odds ratio

  •  
  • STEMI

    ST elevation AMI

References

References
1
Malmberg
K.
Yusuf
S.
Gerstein
H. C.
Brown
J.
Zhao
F.
Hunt
D.
Piegas
L.
Calvin
J.
Keltai
M.
Budaj
A.
Impact of diabetes on long-term prognosis in patients with unstable angina and non-Q-wave myocardial infarction: results of the OASIS Registry
Circulation
2000
, vol. 
102
 (pg. 
1014
-
1019
)
2
McGuire
D. K.
Emanuelsson
H.
Granger
C. B.
Magnus Ohman
E.
Moliterno
D. J.
White
H. D.
Ardissino
D.
Box
J. W.
Califf
R. M.
Topol
E. J.
Influence of diabetes mellitus on clinical outcomes across the spectrum of acute coronary syndromes. Findings from the GUSTO IIb study
Eur. Heart J.
2000
, vol. 
21
 (pg. 
1750
-
1758
)
3
Svensson
A.-M.
McGuire
D. K.
Abrahamsson
P.
Dellborg
M.
Association between hyper- and hypoglycaemia and 2-year all-cause mortality risk in diabetic patients with acute coronary events
Eur. Heart J.
2005
, vol. 
26
 (pg. 
1255
-
1261
)
4
Capes
S. E.
Hunt
D.
Malmberg
K.
Gerstein
H. C.
Stress hyperglycaemia and increased risk of death after myocardial infarction in patients with and without diabetes: a systematic overview
Lancet
2000
, vol. 
355
 (pg. 
773
-
778
)
5
Wahab
N. N.
Cowden
E. A.
Pearce
N. J.
Gardner
M. J.
Merry
H.
Cox
J. L.
Is blood glucose an independent predictor of mortality in acute myocardial infarction in the thrombolytic era?
J. Am. Coll. Cardiol.
2002
, vol. 
40
 (pg. 
1748
-
1754
)
6
Kosiborod
M.
Rathore
S. S.
Inzucchi
S. E.
Masoudi
F. A.
Wang
Y.
Havranek
E. P.
Krumholz
H. M.
Admission glucose and mortality in elderly patients hospitalised with acute myocardial infarction
Circulation
2005
, vol. 
111
 (pg. 
3078
-
3086
)
7
Cao
J. J.
Hudson
M.
Jankowski
M.
Whitehouse
F.
Weaver
W. D.
Relation of chronic and acute glycaemic control on mortality in acute myocardial infarction with diabetes mellitus
Am. J. Cardiol.
2005
, vol. 
96
 (pg. 
183
-
186
)
8
Lynch
M.
Gammage
D.
Lamb
P.
Nattrass
M.
Pentecost
B. L.
Acute myocardial infarction in diabetic patients in the thrombolytic era
Diabetic Med.
1994
, vol. 
11
 (pg. 
162
-
165
)
9
Schiele
F.
Descotes-Genon
V.
Seronde
M. F.
Blonde
M. C.
Legalery
P.
Meneveau
N.
Ecarnot
F.
Mercier
M.
Penfornis
A.
Thebault
L.
, et al. 
Predictive value of admission hyperglycaemia on mortality in patients with acute myocardial infarction
Diabetic Med.
2006
, vol. 
23
 (pg. 
1370
-
1376
)
10
Suleiman
M.
Hammerman
H.
Boulos
M.
Kapeliovich
M. R.
Suleiman
A.
Agmon
Y.
Markiewicz
W.
Aronson
D.
Fasting glucose is an important independent risk factor for 30 day mortality in patients with acute myocardial infarction
Circulation
2005
, vol. 
111
 (pg. 
754
-
760
)
11
Soler
N. G.
Frank
S.
Value of glycosylated haemoglobin measurements after acute myocardial infarction
JAMA, J. Am. Med. Assoc.
1981
, vol. 
246
 (pg. 
1690
-
1693
)
12
Blackledge
H. M.
Newton
J.
Squire
I. B.
Prognosis for South Asian and white patients newly admitted to hospital with heart failure in the United Kingdom: historical cohort study
Br. Med. J.
2003
, vol. 
327
 (pg. 
526
-
531
)
13
Cubbon
R. M.
Wheatcroft
S. B.
Grant
P. J.
Gale
C. P.
Barth
J. H.
Sapsford
R. J.
Ajjan
R.
Kearney
M. T.
Hall
A. S.
Evaluation of methods and management of acute coronary events investigators
Eur Heart J.
2007
, vol. 
28
 (pg. 
540
-
545
)
14
Fox
K. A.
Steg
P. G.
Eagle
K. A.
Goodman
S. G.
Anderson
F. A.
Jr
Granger
C. B.
Flather
M. D.
Budaj
A.
Quill
A.
Gore
J. M.
Decline in rates of death and heart failure in acute coronary syndromes, 1999–2006
JAMA, J. Am. Med. Assoc.
2007
, vol. 
297
 (pg. 
1892
-
1900
)
15
Deedwania
P.
Kosiborod
M.
Barrett
E.
Ceriello
A.
Isley
W.
Mazzone
T.
Raskin
P.
Hyperglycemia and acute coronary syndrome. A scientific statement from the American Heart Association Diabetes Committee of the Council on Nutrition, Physical Activity, and Metabolism
Circulation
2008
, vol. 
117
 (pg. 
1610
-
1619
)
16
Malmberg
K.
Ryden
L.
Efendic
S.
Herlitz
J.
Nicol
P.
Waldenström
A.
Wedel
H.
Welin
L.
Randomised trial of insulin-glucose infusion followed by subcutaneous insulin treatment in diabetic patients with acute myocardial infarction (DIGAMI study): effects on mortality at 1 year
J. Am. Coll. Cardiol.
1995
, vol. 
26
 (pg. 
57
-
65
)
17
Malmberg
K.
Ryden
L.
Wedel
H.
Birkeland
K.
Bootsma
A.
Dickstein
K.
Efendic
S.
Fisher
M.
Hamsten
A.
Herlitz
J.
, et al. 
Intense metabolic control by means of insulin in patients with diabetes mellitus and acute myocardial infarction (DIGAMI 2): effects on mortality and morbidity
Eur. Heart J.
2005
, vol. 
26
 (pg. 
650
-
661
)
18
Kosiborod
M.
Inzucchi
S. E.
Krumholz
H. M.
Xiao
L.
Jones
P. G.
Fiske
S.
Masoudi
F. A.
Marso
S. P.
Spertus
J. A.
Glucometrics in patients hospitalized with acute myocardial infarction
Circulation
2008
, vol. 
117
 (pg. 
1018
-
1027
)
19
Ghoyal
A.
Mahaffey
K. W.
Garg
J.
Nicolau
J. C.
Hochman
J. S.
Weaver
W. D.
Theroux
P.
Oliveira
G. B.
Todaro
T. G.
Mojcik
C. F.
, et al. 
Prognostic significance of the change in glucose level in the first 24h after acute myocardial infarction: results from the CARDINAL study
Eur. Heart J.
2006
, vol. 
27
 (pg. 
1289
-
1297
)
20
Weston
C.
Walker
L.
Birkhead
J.
Early impact of insulin treatment on mortality for hyperglycaemic patients without known diabetes who present with an acute coronary syndrome
Heart
2007
, vol. 
93
 (pg. 
1542
-
1546
)
21
Tenerz
A.
Lonnberg
I.
Berne
C.
Nilsson
G.
Leppert
J.
Myocardial infarction and prevalence of diabetes mellitus. Is increased casual blood glucose at admission a reliable criterion for the diagnosis of diabetes?
Eur. Heart J.
2001
, vol. 
22
 (pg. 
1102
-
1110
)
22
Sack
M. N.
Yellon
D. M.
Insulin therapy as an adjunct to reperfusion after acute coronary ischaemia; a proposed direct myocardial cell survival effect independent of metabolic modulation
J. Am. Coll. Cardiol.
2003
, vol. 
41
 (pg. 
1404
-
1407
)
23
Verma
S.
Maitland
A.
Weisel
R. D.
Li
S. H.
Fedak
P. W.
Pomroy
N. C.
Mickle
D. A.
Li
R. K.
Ko
L.
Rao
V.
Hyperglycemia exaggerates ischemiare perfusion-induced cardiomyocyte injury: reversal with endothelin antagonism
J. Thorac. Cardiovasc. Surg.
2002
, vol. 
123
 (pg. 
1120
-
1124
)
24
Kersten
J. R.
Toller
W. G.
Gross
E. R.
Pagel
P. S.
Warltier
D. C.
Diabetes abolishes ischemic preconditioning: role of glucose, insulin, and osmolality
Am. J. Physiol. Heart Circ. Physiol.
2000
, vol. 
278
 (pg. 
H1218
-
H1224
)
25
Liu
Q.
Docherty
J. C.
Rendell
J. C.
Clanachan
A. S.
Lopaschuk
G. D.
High levels of fatty acids delay the recovery of intracellular pH and cardiac efficiency in post-ischemic hearts by inhibiting glucose oxidation
J. Am. Coll. Cardiol.
2002
, vol. 
39
 (pg. 
718
-
725
)
26
Ascione
R.
Rogers
C. A.
Rajakaruna
C.
Angelini
G. D.
Inadequate blood glucose control is associated with in-hospital mortality and morbidity in diabetic and non-diabetic patients undergoing cardiac surgery
Circulation
2008
, vol. 
118
 (pg. 
113
-
123
)