CLINICAL STUDIES / ETUDES CLINIQUES
EFFECT OF ADMISSION HYPERGLYCAEMIA ON SHORT-TERM OUTCOME IN ADULT NIGERIANS WITH A FIRST ACUTE ISCHAEMIC STROKE
CONSEQUENCE DE L'HYPERGLYCEMIE A L'ADMISSION DANS LE PRONOSTIC A COURT TERME APRES UN ACCIDENT VASCULAIRE CEREBRAL ISCHEMIQUE
KEY WORDS: ischemic stroke; Nigerians; hyperglycaemia; functional outcome; mortality
Mots clés : Accident vasculaire cérébral ischémique, hyperglycémie, mortalité, morbidité, Nigéria, évolution
An overview of the current epidemiology of stroke indicates that worldwide, stroke remains a significant cause of mortality and morbidity in both developed and developing nations (2, 19, 33). The outcome following stroke is influenced by several factors such as subtype, severity of stroke, the predisposing factor(s), associated factors, presence of complications, access to specialist care, and availability of stroke care facilities (3).
One of the potentially modifiable factors associated with adverse outcome in acute stroke is hyperglycaemia at the time of a stroke. Clinical and experimental studies have demonstrated that admission hyperglycaemia has a deleterious effect on outcome in acute ischaemic stroke though some studies have debated this (6, 7, 22, and 30). Although the exact relationship between hyperglycaemia and stroke outcome (causal or indicative of more severe stroke) remains controversial, putative mechanisms via which hyperglycaemia exerts its deleterious effects have been postulated (16, 22, 29). Elevated blood glucose levels are associated with an increased progression of hypoperfused at-risk tissue to infarction and poor stroke outcome, a mechanism that appears to be promoted by hyperglycaemia-induced increase in lactate production (22).
There is no evidence from a prospective study regarding the effect of admission hyperglycaemia on prognosis in acute ischaemic stroke in Nigerians. Such data are desirable considering the potentially adverse but presumably modifiable nature of hyperglycaemia in ischemic stroke. This study was designed to prospectively determine the prevalence of admission hyperglycaemia in first-ever acute ischaemic stroke and evaluate its impact on short-term outcome.
PATIENTS AND METHODS
The study protocol was approved by the Research and Ethics Committee of the Lagos University Teaching Hospital, Lagos, Nigeria, and informed consent was obtained from all patients or their proxies.
Case definition and clinical assessments
All adult Nigerians consecutively presenting to the Emergency Unit of the Lagos University Teaching Hospital between February 2003 and May 2004 with a first-ever acute ischaemic stroke (according to World Health Organization criteria) (12) were recruited into the study. Ischaemic stroke was defined by brain CT scan (normal brain CT scan or recent infarct in the clinically relevant area on scan performed within 3 days or 72 hours of stroke onset). For cases who could not afford a brain CT scan, combined fulfillment of the WHO criteria and the Siriraj Stroke Score definition of ischaemic stroke was used (4, 14, 23). We excluded patients who presented more than 72 hours after stroke onset and patients with recurrent or haemorrhagic stroke. Baseline demographic data, historical data (for conventional risk factors, previous cerebrovascular disease) and physical examinations were done with a view to documenting risk factors, neurological deficits and ascertaining the presence or absence of complications in each patient.
Stroke severity on admission was assessed using the National Institute of Health Stroke Scale (NIHSS); those with a score of >13 were assessed to have severe stroke while a score of ≤13 was regarded as mild stroke (1, 27). All patients had standard conservative stroke management in accordance with the management guidelines of the neurology unit of the hospital which was adopted from various international management guidelines. The patients had isotonic fluid infusion and regular physiotherapy, with early ambulation where possible. All patients with elevated blood pressure were not given antihypertensives within the first 28 days of stroke onset except there were compelling indications like acute left ventricular failure, myocardial ischemia / infarction, rapid decline in renal function, severe hypertension, or dissecting aortic aneurysm. Unconscious patients were frequently turned in bed to prevent pressure sores while those with dense hemiplegia had prophylactic subcutaneous heparin to prevent deep venous thrombosis. The patients were followed up until 30 days post stroke onset with regular evaluation to document development of neurological and non-neurological complications (including seizures, pneumonia, and sepsis). The outcome measures were 30-day case fatality or 30-day functional outcome using the NIHSS in survivors. A decrease in the NIHSS score by 4 or more points from admission to the 30th day post stroke onset was used to define good neurological improvement, while a decrease of less than 4 points defined poor neurological improvement.
Blood glucose estimation
A non-fasting venous blood sample was collected on admission for random blood glucose (RBG) estimation in each patient. All analyses were conducted in the departmental research laboratory using the glucose oxidase method. Hyperglycaemia was defined as an admission random blood glucose value ≥ 140mg/dL (30). Cases were categorized as hyperglycaemic or normoglycaemic.
Statistical analysis was done using SPSS version 11.0 software (SPSS Inc). Means or proportions were calculated for baseline demographic data and the significance of any intergroup differences was tested using Analysis of Variance (ANOVA) for means and Х2 test for proportions. A multivariate analysis was done to determine the predictors of 30-day case fatality and functional outcome. A p value < 0.05 was considered statistically significant. RESULTS
Baseline clinical and demographic profile of study subjects
Frequency of admission hyperglycaemia
Overall, the frequency of admission hyperglycaemia, based on a cutoff of random blood glucose level ≥140mg/dL was 34% (34/100). The frequency was further characterized in various strata as shown in Table 2, and was not significantly different when compared by gender, age below or above 65 years, or presence of complications during hospitalization (P>0.05 for each stratum). The frequency of hyperglycaemia was however significantly higher in patients with admission NIHSS scores >13 (P=0.0001), and those with a history of diabetes mellitus (P=0.04).
Case fatality rate and functional outcome
Overall, the 30-day case fatality rate was 28% (28/100). However the case fatality rate was significantly higher in hyperglycaemic compared to normoglycaemic patients (14/34 i.e. 41.2% vs. 14/66 i.e. 21.2%; X2=4.39; P=0.04). Table 3 shows a multivariate analysis of various variables with outcome. Factors that were significantly associated with 30-day case fatality are admission stroke severity (42.9% of those with admission NIHSS score >13 died compared to 20% of those with NIHSS score ≤13, p=0.02), admission hyperglycaemia (41.2% of hyperglycaemic patients died compared to 21.2% of normoglycaemics, p=0.04), and presence of complications (54.5% of those with complications died compared to 3.6% of those without, p<0.001). Functional outcome on the NIHSS was categorized as good or poor. On multivariate analysis, none of the variables was significantly associated with the functional outcome (Table 4). Although 60% of the hyperglycaemic compared to 34.6% of normoglycaemic patients had a poor improvement in their NIHSS score, this did not achieve statistical significance (p=0.05). DISCUSSION
The frequency of admission hyperglycaemia (RBG≥140mg/dL) in this prospective study of first-ever acute ischaemic stroke in adult Nigerians was 34%. This is similar to reports from several other studies in which the reported prevalence of hyperglycaemia in stroke has varied from 20% to 50% (10, 26, 28, 31). The variability has been dependent on the subtype of stroke, cut off value for defining hyperglycaemia, and the study design (prospective versus retrospective) (10, 26, 28). In a study conducted in a similar tertiary setting as ours, using the same operative definition, Ogunrin et al documented a lower prevalence rate of 28%, but acknowledged that the retrospective design of that study may have affected their results as only 100 of their 163 cases had blood glucose estimated on admission (21). In our study, about a third (33.3%) of the cases with hyperglycaemia were known diabetics, showing that the larger proportion of hyperglycaemics either have previously unrecognized diabetes, abnormalities of glucose tolerance, stress hyperglycaemia, or an epiphenomenon related to severity of neurologic deficit, as has been suggested in previous studies (8, 13, 18, 25).
Irrespective of the cause however, (and indeed of diabetic status), numerous experimental and clinical studies have implicated hyperglycaemia as a predictor of poor outcome in acute stroke (5, 6, 13, 15, 18, 24, 30). In our study, short-term mortality measured by the 30-day case fatality rate was significantly higher in the hyperglycaemic patients (41%) compared to the normoglycaemic patients (21%). Several studies have found that short and long-term mortality rates and risk of death are higher in stroke patients with admission hyperglycaemia (5, 6, 11, 13, 17, 30, and 32). In a systematic overview of stress hyperglycaemia and stroke outcome, Capes et al reported a pooled two-fold increased risk of short-term mortality (within 1 month) after stroke from three of the studies that reported data for diabetic and non-diabetic patients combined, as we did in this study (6). Williams et al, in a study of 656 hospitalized acute ischaemic stroke patients, found that admission hyperglycaemia (defined as admitting RBG ≥130mg/dL) was present in 40% of cases and independently increased the risk for death at 30 days, 1 year and 6 years after stroke (31). Kiers et al observed that mortality was significantly higher in patients with stress hyperglycaemia, who also tended to have more severe stroke on admission (13). Our findings are similar in this regard, as the frequency of hyperglycaemia was significantly higher in the cases with more severe stroke as measured by the admission NIHSS score, a factor that was also significantly associated with 30-day case fatality in our study. The other factor (apart from hyperglycaemia and stroke severity by NIHSS score) that was associated with increased case fatality was the presence of complications. However, the frequency of complications was similar in normoglycaemic (43.9%) and hyperglycaemic cases (44.1%) and yet mortality was higher in the latter, suggesting that the presence of complications was not the lone mechanistic basis underlying increased mortality in our series.
A multivariate analysis showed that none of the variables, including admission hyperglycaemia was significantly associated with 30-day functional outcome, although hyperglycaemic patients had poor functional outcome compared to normoglycaemic patients. This is similar to the finding of Counsell et al (7) who found no association between admission hyperglycaemia and functional outcome in patients with ischaemic stroke seen within 72 hours of onset. It is however at variance with other studies which have demonstrated that admission hyperglycaemia has an adverse effect on functional outcome in stroke (11, 30).
Although there remains some controversy as to whether acute hyperglycaemia causes a worse outcome following stroke or is a by-product of more severe stroke, Parsons et al, using diffusion-weighted and perfusion-weighted magnetic resonance imaging in stroke patients, have convincingly demonstrated that elevated blood glucose levels in acute stroke are associated with an increased rate of hypoperfused at-risk tissue progressing to infarction and poor stroke outcome (22).
The findings in this study are not novel, but add to the existing evidence of the adverse outcome associated with hyperglycaemia in acute stroke. Our study is however the first prospective study on the effects of admission hyperglycaemia on short-term outcome in Nigerians with ischaemic stroke. We acknowledge that our study had several limitations. We utilized a single measure of RBG in this study, whereas serial measurements may have improved the accuracy of determining the effect of hyperglycaemia on outcome. We did not measure glycated haemoglobin levels and so were unable to distinguish cases with pre-existing hyperglycaemia or diabetes from those with “reactive” hyperglycaemia, as these may represent different pathogenetic entities. However, in our analysis, a prior diagnosis of DM was not significantly correlated with either case fatality or functional outcome. Also, due to limited affordability of brain imaging, we liberalized the case definition of ischaemic stroke by using clinical criteria (combined fulfillment of the WHO criteria and Siriraj stroke score criteria) for cases in which brain imaging was not done. As such, some cases with haemorrhagic stroke may have inadvertently been included. Poungvarin et al (23) and Ogun et al (20) have validated the Siriraj stroke score and WHO criteria respectively for use in the developing world. Poungvarin et al found a sensitivity of 93.2% for the Siriraj stroke score while Ogun et al found the WHO criteria to have a sensitivity of 69% and concluded that these could be useful where brain CT scan is not available. Kolapo et al (14) have also validated the Siriraj stroke score for use in Nigerians and found it to have a predictive accuracy of 80%.
The findings of our study are relevant to stroke care in our environment for several reasons. In the management of acute stroke, attention to modifiable factors that impact negatively on stroke outcome (e.g. hyperglycaemia and complications) may reduce stroke mortality, which is notably high in developing countries. Although the jury is still out on the efficacy and target values of glycaemic control in acute ischaemic stroke, the evidence of a detrimental effect of elevated glucose in stroke provided from this and earlier studies emphasizes the need for well-designed, randomized clinical trials of the effect of blood glucose lowering on stroke outcome. In the interim, incorporating existing guidelines regarding glycaemic control in stroke patients should be an integral part of stroke management protocols.
Table 1. Demographic and clinical characteristics of stroke patients at admission
* Cigarette smoking in 12 months preceding stroke onset
Table 2. Comparison of clinical parameters and outcome in hyperglycaemic and normoglycaemic stroke patients
RR relative risk
Table 3: Relationship of patients’ characteristics to 30-day outcome (death or survival).
Table 4: Relationship of patients’ characteristics to functional outcome in survivors at 30 days post-stroke
*Improvement in NIHSS score categorized as good = ≥4 or poor = <4