Our study shows that hyperglycemia is a very common finding in patients presenting with bacterial meningitis. One out of four patients in our cohort had severe hyperglycemia which was related with unfavorable outcome in a univariate analysis (P = 0.008). However, the association between glucose levels and unfavorable outcome did not remain robust in a multivariate analysis.
Many of the factors predictive for unfavorable outcome were also predictive for severe hyperglycemia on admission. High blood glucose levels seem to be an epiphenomenon of disease severity and physical stress, at least partly. Factors predictive for systemic compromise are strong predictors of unfavorable outcome in adults with bacterial meningitis.[8
] This relation is comparable with any other critical illness. However, some of the well known factors predictive for unfavorable outcome in bacterial meningitis, such as low CSF white cell counts, low diastolic blood pressure, and low thrombocyte counts, were not related to high glucose serum levels. On the contrary, these factors had an inverse relation with serum blood glucose levels. This finding could not be explained by differences in causative bacterial agents between groups. Our results therefore emphasize the importance of the insult to the central nervous system, rather than the systemic insult in patients with bacterial meningitis, leading to disturbed blood-glucose regulation mechanisms. This effect has been described previously in patients after stroke and might be caused by effects on central autonomic control sites.[16
Diabetes was by far the strongest predictor for higher glucose levels on admission. The overall proportion of patients with a pre-admission diagnosis of diabetes was 7%. However, this percentage might have been underestimated since many patients had diabetic blood glucose levels on admission. The incidence of diabetes among adults with bacterial meningitis has been reported to be as high as 39% in Taiwan.[18
] The meningitis population in Taiwan differed markedly from our cohort with respect to the causative organisms. In Taiwan, Klebsiella pneumoniae
was the most common causative organism in both diabetic and non-diabetic patients, whereas no infection with S. pneumoniae
occurred in any of the diabetic patients. By contrast, in our cohort S. pneumoniae
was the most common causative microorganism and was significantly more common in patients with hyperglycemia (55%) than in patients without hyperglycemia (42%). Differences in microbial epidemiology, innate immunity and health care might account for these differences.
As in stroke, hyperglycemia could be a measure of previously unrecognized insulin resistance or diabetes.[19
] This idea is supported by the observation that predictors for hyperglycemia were also associated with diabetes before admission: older age, other immunocompromise than diabetes (use of immunosuppressive medication, like corticosteroids), and a distant focus of infection.
The majority of patients with a pre-admission diagnosis of diabetes had pneumococcal meningitis (67%). Diabetes is a well known risk factor for invasive pneumococcal diseases. In the U.S. and some European countries, physicians are advised to vaccinate diabetic patients with the 23-valent pneumococcal polysaccharide vaccine [20
]. The 23-valent pneumococcal polysaccharide vaccine would have provided coverage for 87% of all causative pneumococcal strains in the Dutch Meningitis Cohort Study.[23
] Vaccine efficacy however, is likely less than 60%[24
] and no randomized controlled clinical trials using this vaccine have yet been carried out in diabetic patients to support this recommendation.
This study has several limitations. First, only patients who had a positive CSF culture were included. Negative CSF cultures are estimated to occur in 11 to 30% of patients with bacterial meningitis.[10
] No significant differences in clinical presentation have been reported between culture-positive bacterial meningitis and those with culture-negative bacterial meningitis.[25
] It is therefore unlikely that this factor confounded our results. Second but more importantly, we used a single random blood glucose measurement on admission to define hyperglycemia. As can be seen in stroke[27
] and subarachnoid hemorrhage[3
], persistent hyperglycemia or persistently high levels of fasting blood glucose may be more important than hyperglycemia on admission and might have an independent effect on outcome, which we were not able to assess because follow-up blood glucose concentrations were unavailable. We also did not measure glycosylated hemoglobin (HbA1C) in our patients. This is an important limitation of our study. Finally, we have no data on the adequacy of antihyperglycemic treatment and treatment-related hypoglycemia.
Although the exact nature of the relation between severe hyperglycemia and outcome in bacterial meningitis remains uncertain, data of previous studies including critically ill patients suggest that severe hyperglycemia should be treated, also in adults presenting with bacterial meningitis. Based on these previous studies, we recommend aiming for blood glucose levels below 10.0 mmol/L (180 mg/dL) throughout the clinical course, while trying to avoid hypoglycemia. The increased risk of hypoglycemia has been one of the major concerns put forward with regard to intensive insulin therapy.[28
] A recently published study even found an increased mortality and a higher incidence of hypoglycemia in critically ill patients who where treated with intensive glucose control as compared to those who were treated with conventional glucose control.[29
] Although we were unable to assess the frequency and severity of treatment-related hypoglycemia in our cohort, hypoglycemia can be life threatening and hypoglycemic symptoms are not easily recognized in patients with acute bacterial meningitis. Severe hypoglycemia or prolonged hypoglycemia can lead to convulsions, coma, and irreversible brain damage. In the single-center prospective randomized controlled trials from Leuven, the risk of hypoglycemia (glucose of ≤ 2.2 mmol/L or ≤ 40 mg/dL) increased from 0.8% to 5.1% in the surgical ICU trial and from 3.1% to 18.7% in the medical ICU trial.[7
] In those ICU based trials, insulin adjustments were based on measurements of blood glucose levels in arterial blood, at one- to four-hour intervals, and doses were administered by a team of intensive care nurses, assisted by a study physician. The question arises whether intensive insulin therapy is feasible, safe and effective when implemented outside of the ICU, where most patients with acute bacterial meningitis are managed. Previous studies have shown that control of hyperglycemia in patients admitted to a stroke unit remains difficult.[19
] However, in light of the growing evidence of the detrimental effects of sustained hyperglycemia in other neurological diseases, physicians treating patients with bacterial meningitis should play a more active role in the control of blood glucose levels.