The algorithm for treatment of increased intracranial pressure (ICP) in the neuro-intensive care unit (n = 52). An external ventricular drainage (EVD) was established at operation in 50 patients with initial cerebrospinal fluid (Csf) drainage in 48 of these. A parenchymal ICP- monitor (Codman) was inserted in four cases. Bolus doses of Mannitol were administered in five cases prior to neuro-intervention. The shaded numbers represent different patients. doi:10.1371/journal.pone.0091976.g002 

Contexts in source publication

Context 1

... intracranial hypertension was defined as one or more episodes of ICP, continuously exceeding 20 mmHg for more than 5 minutes. As illustrated in figure 2, CSF-drainage through the EVD was the main treatment of increased ICP. The upper ICP limit triggering drainage was set at 20 mmHg. Additional ICP treatment was applied at the discretion of the treating physicians, guided by clinical, radiological and physiological findings. If signs of interstitial brain edema were observed on the CT-scan, bolus doses of hypertonic saline were given as osmotherapy, allowing S- Na levels of not higher than 160 mmol/L. Intracranial hypertension due to hyperaemia, detected and guided by trans-cranial doppler and/or jugular bulb monitoring, was treated with moderate hyperventilation aiming at a pCO 2 of 4.0–4.5 kPa. In cases with paracetamol resistant hyperthermia, an external cooling blanket (Thermo-wrap) was used, aiming at normothermia. If intracranial hypertension and brain edema persisted, despite previous corticosteroid treatment and above-mentioned ICP- decreasing interventions, 1 g of i.v. methylprednisolone was given. Last resort ICP treatment was a pentothal infusion, inducing a barbituate coma to lower cerebral metabolism, monitored with microdialysis and/or jugular bulb analyses. Clinical, etiological and laboratory findings, mental status (GCS and/or RLS) on admission, and the duration from admission to start of antibiotic and corticosteroid treatment were prospectively registered in the SQRM (Table 1). A CT-scan of the brain was performed in all patients in the intervention group and, based on clinical grounds, in 50 of the 53 cases in the control group. EEG was performed in selected cases to exclude non-convulsive seizures. A follow-up at 2–6 months after discharge included an evaluation of neurological deficits, Glasgow outcome score (GOS) and hearing impairment in all survivors according to the SQRM protocol. Audiometry was performed in selected cases. A power test was performed, based on available data [2,8– 10,12,14–16] suggesting a mortality of 40% with standard ICU treatment, and hypothesizing a 20% mortality when employing ICP-targeted treatment of ABM patients with severely impaired mental status at the NICU. Consequently, at least 50 patients are required in each group to achieve a probability of 80% to detect a 20% difference in mortality with a 95% CI. A two-tailed Fisher’s exact test was used when comparing groups regarding outcomes, demographic, etiological and clinical data. The Student T-test was used for the comparison of laboratory values. Multivariate analyses were performed to adjust for possible differences in case mixes of the control and intervention group. (Table 1). The main characteristics regarding demographic, etiological, clinical, and laboratory data are given in Table 1, not identifying any significant differences between the intervention and control group. A cerebral CT-scan on admission showed one or more of the following; tight sulci, edema, compressed cisterns or hydrocephalus, in 22/52 (42%) patients in the intervention group. However, CT-findings were lacking in the controls as this data was not recorded in the SQRM. The median time from primary admission to insertion of an EVD or a parenchymal ICP-monitor was 8 h:26 min (range 1 h:27 min–96 h) and the median duration of EVD management was 5 days (range 1.5–16 days). Five patients were treated with bolus doses of Mannitol upon admission, due to slit ventricles as seen on the CT-scan (n = 3), or a spinal opening pressure of . 500 mmH 2 O (n = 2). An EVD was applied in 48 cases, and a parenchymal ICP-monitor was inserted in 4 cases (slit ventricles; n = 2, technical problems; n = 1, coagulopathy; n = 1). When the EVD was applied, Csf was momentarily drained in the operating room in 48 cases, and in 38 (79%) of these cases, the neurosurgeon noted markedly elevated ICP during this procedure. In the NICU, 35 of 52 patients (67%) presented significant intracranial hypertension (ICP . 20 mmHg during . 5 min; Figure 3). A parenchymal ICP-monitor but not an EVD was established in three of these 35 cases. Of the remaining 17 patients, 11 exhibited ICP . 20 mmHg, but for shorter episodes than 5 min. In these cases, Csf was also drained through the EVD to maintain an ICP , 20 mmHg. In three additional cases, Csf was drained despite ICP , 20 mmHg (Figure 2). Thus, Csf-drainage was performed in altogether 46 (88%) of the 52 cases. Four of the remaining six patients had parenchymal probes instead of EVD. In aggregate, increased ICP at surgery and/or significant intracranial hypertension during the NICU-care was observed in altogether 43 (83%) of the 52 patients (Table 2). The peak levels of ICP occurred within the first 48 hours of NICU-care in all patients, except in one patient who exhibited the highest level on day three. Altogether, 49 (94%) of the intervened patients were administered specific ICP-decreasing treatment in the NICU (Figure 2, Table 3). Mortality was significantly decreased in the per protocol intervention group (5/52 = 10%) compared to the controls (16/ 53 = 30%; relative risk reduction 68%; p , 0.05; Table 4). Full recovery, defined as GOS 5 with preserved normal hearing, was significantly more frequent in the per protocol group; 28 patients (54%) versus 17 patients (32%) in the control group. Thus, the relative risk reduction for unfavorable outcome was 40% (p , 0.05). Three intervened patients suffered persisting severe neurological deficits, with outcomes of GOS 3, related to complications of ABM; sinus/intracranial venous thrombosis; n = 1, cerebral infarction; n = 1, and ventriculitis and hydrocephalus; n = 1. The remaining 44 survivors reached GOS 4 (n = 12) or 5 (n = 32; Table 2). Of the 28 intervened patients that fully recovered, including normal hearing, 19 (68%) had significant intracranial hypertension in the NICU and a further four had markedly elevated ICP noted only at surgery (Table 2). Significantly more episodes of ICP . 20 mmHg lasting . 5 min were observed among patients that succumbed, versus survivors (p , 0.01). Slit ventricles, not allowing drainage through the EVD, were found in three of the five fatal cases. All deaths occurred within one month after admission. Three of the five fatal cases in the treatment group died within a week, as did 11/16 fatalities among the controls. Two of the five fatalities in the intervention group, and 6 of the 16 controls that died, were immunocompromised (Table 1). S pneumoniae was the etiological agent in all fatal cases of the intervention group, and in 9 of 16 succumbed controls. The causes of mortality were cerebral herniation and/or ischaemia, verified by CT-scan and/or autopsy, in 4/5 (80%) intervened patients and in 13/16 (81%) controls. In the remaining cases, death was caused by septic shock and/or multi-organ failure. Lumbar puncture (LP) was avoided due to contraindication in two (suspected herniation in both) of the five fatal cases in the intervention group, and in four (suspected herniation: n = 2, coagulopathy: n = 2) of the 16 controls that died. Thus, the mortality rate among spinal tapped patients (Table 1) was 3/47 (6%) and 12/44 (27%), in the intervention and control groups, respectively. No patient died within 24 hours after LP and, of the intervened patients with persisting deficit, a significant clinical deterioration shortly after LP was not recorded in any case. Acute bacterial meningitis (ABM) in adults still remains a challenge for the clinician, and in comatose ABM-patients mortality rates of up to 62% have been reported [2,9,10]. ICP- targeting treatment of severe ABM has been reported with promising results [14–16,23] but is not routinely used. [4,22]. The present report is, to our knowledge, the first controlled study of neuro-intensive care and ICP-guided therapy using EVD in adults with ABM and severely impaired consciousness. The mortality was significantly decreased from 30% in the controls to 10% in the treatment group, and the incidence of full recovery was increased from 46% to 68% in the groups, respectively. Significant initial intracranial hypertension was observed in two thirds of the 28 patients that fully recovered, indicating that ICP-guided therapy may result in a satisfactory outcome, even in cases with severe ICP elevation. However, in concordance with earlier findings, extreme ICP levels were observed among the fatal cases [15–17,19,20,28]. The peak ICP occurred early during the course of the disease, as shown earlier [14–16]. In line with previous experience, cerebral herniation and/or infarction dominated as cause of mortality in both groups [2,13– 16,18], and a very aggressive course of ABM with fatal outcome was observed in five intervened cases [20,29,30]. A few case reports suggest that a decompressive craniectomy may be considered early in the course of disease in such cases [31–33]. The pathophysiological mechanisms resulting in increased ICP in ABM are multifactorial. The release of bacterial components in the subarachnoid space leads to an inflammatory response that contributes to increased permeability of the blood-brain barrier causing cerebral extracellular edema, impaired Csf-absorption with increased Csf-volume, a cytotoxic intracellular brain edema, and increased cerebral blood flow (hyperaemia); all adding to elevated ICP [34,35]. Our study suggests that increased ICP is a major contributor to mortality and morbidity in ABM, and that Csf-drainage significantly affects recovery. Both the initial Csf-drainage during EVD insertion and further drainage in the NICU have likely contributed to favorable outcome by decreasing ICP, consistent with earlier studies [14,15]. Additional ways of achieving ICP control may also be effective. Osmotic therapy has been suggested in earlier studies of TBI [36] and ABM [16], and hypertonic saline was administered to 40% of the patients in the present study. The ...

Context 2

... group. Patients in the intervention group were recruited from seven hospitals in the Stockholm region, serving about 2.5 of Sweden’s 9 million inhabitants. The neuro-intensive care unit (NICU) at the Karolinska University Hospital Stockholm was contacted for preliminary inclusion immediately after admission when identifying a patient with a strong suspicion of ABM based on clinical grounds, with or without Csf-analyses, and inclusion criteria 1 and 3 above. Definite inclusion was made after the microbiological analyses. Fifty-nine patients were preliminarily eligible for the intervention group. Two patients were excluded due to non-confirmed ABM diagnosis, and the remaining 57 cases constituted the intention to treat group (Figure 1). Five patients were excluded due to severe coagulopathy (n = 3), lack of beds at the NICU (n = 1), and brain death with computerized tomography (CT)-verified cerebral infarction and herniation on admission (n = 1). Thus, finally 52 patients were included in the per protocol analysis of the intervention group. Control group. The controls were included retrospectively from clinical admission data registered prospectively in SQRM. From this registry, 651 consecutive patients were identified (Figure 1). This constitutes an estimated 75% of all adult cases with community acquired ABM in Sweden, according to data from The National Board of Health and Welfare, during the study period. Of 651 patients 149 (23%) fulfilled the above inclusion criteria. Of these, 57 were included in the intervention group and the remaining 92 cases were eligible as controls. To avoid a possible selection bias the following controls were excluded by independent observers; 1) patients not treated according to national guidelines including initial corticosteroids (dexametha- sone or betamethasone) and adequate antibiotics (cefotaxime or ceftriaxone + / 2 ampicillin, or meropenem) in meningitis dosage (n = 4), and 2) patients not treated at an intensive care unit (ICU) with assisted ventilation and sedation (n = 3; one of these was judged desolate on admission). Moreover, 3) eligible patients from the Stockholm region, where treating physicians had missed to contact NICU for inclusion in the intervention group (n = 12), and 4) patients where ICP-targeted treatment in a NICU outside Stockholm was performed, independently of this study (n = 20), were not included in the final control group. The outcomes of these 20 cases and the 12 eligible patients from Stockholm are presented separately. The remaining 53 patients, admitted to one of 18 Regional or University hospitals outside Stockholm, constituted the final control group. ICP-targeted treatment in a NICU outside Stockholm was not withheld due to poor prognosis in any of these controls according to independent observers. Inclusion or exclusion based on assessing consciousness was performed using GCS in all cases in the intervention group and in 14 controls, whereas 39 control patients were included according to RLS, as this is used in favor of GCS in many Swedish hospitals. All patients were treated in an ICU with adequate antibiotics and corticosteroids in ABM-doses. Standard intensive care management according to current recommendations regarding severe ABM [7] with mechanical ventilation, adequate sedation, and aiming at normal fluid and electrolyte homeostasis, mean arterial pressure $ 65 mmHg, SaO 2 $ 95%, pCO 2 4.5–5.5 kPa, and blood-glucose 5–10 mmol/L, was administered to all controls and initially to the intervention group. The decision to perform ICP-targeted treatment, in addition to standard intensive care, was made on admission in all intervention patients. Thirty-five patients were transported directly to the NICU whereas 17 cases, primarily admitted to one of six other hospitals in the Stockholm region, were treated at local ICUs during 1–36 hours, before being referred to the NICU. Following CT-scanning of the brain, an EVD-catheter for ICP-monitoring was established in 50 patients. In two of these the EVD did not function due to slit ventricles. In these two and in another two cases, where an EVD could not be applied (technical problems: n = 1, moderate coagulopathy: n = 1) a parenchymal ICP-monitor (Codman, Johnson and Johnson Nordic AB, Stockholm) was used. Patients were treated in a 30 u sitting position. The temple was used as zero-point for mean arterial blood pressure and cerebral perfusion pressure. ICP was continuously registered in a computerized patient monitoring system, ICU-pilot ( m -dialysis AB, Solna, Sweden), and the treatment goals were ICP , 20 mmHg and cerebral perfusion pressure . 50 mmHg. Significant intracranial hypertension was defined as one or more episodes of ICP, continuously exceeding 20 mmHg for more than 5 minutes. As illustrated in figure 2, CSF-drainage through the EVD was the main treatment of increased ICP. The upper ICP limit triggering drainage was set at 20 mmHg. Additional ICP treatment was applied at the discretion of the treating physicians, guided by clinical, radiological and physiological findings. If signs of interstitial brain edema were observed on the CT-scan, bolus doses of hypertonic saline were given as osmotherapy, allowing S- Na levels of not higher than 160 mmol/L. Intracranial hypertension due to hyperaemia, detected and guided by trans-cranial doppler and/or jugular bulb monitoring, was treated with moderate hyperventilation aiming at a pCO 2 of 4.0–4.5 kPa. In cases with paracetamol resistant hyperthermia, an external cooling blanket (Thermo-wrap) was used, aiming at normothermia. If intracranial hypertension and brain edema persisted, despite previous corticosteroid treatment and above-mentioned ICP- decreasing interventions, 1 g of i.v. methylprednisolone was given. Last resort ICP treatment was a pentothal infusion, inducing a barbituate coma to lower cerebral metabolism, monitored with microdialysis and/or jugular bulb analyses. Clinical, etiological and laboratory findings, mental status (GCS and/or RLS) on admission, and the duration from admission to start of antibiotic and corticosteroid treatment were prospectively registered in the SQRM (Table 1). A CT-scan of the brain was performed in all patients in the intervention group and, based on clinical grounds, in 50 of the 53 cases in the control group. EEG was performed in selected cases to exclude non-convulsive seizures. A follow-up at 2–6 months after discharge included an evaluation of neurological deficits, Glasgow outcome score (GOS) and hearing impairment in all survivors according to the SQRM protocol. Audiometry was performed in selected cases. A power test was performed, based on available data [2,8– 10,12,14–16] suggesting a mortality of 40% with standard ICU treatment, and hypothesizing a 20% mortality when employing ICP-targeted treatment of ABM patients with severely impaired mental status at the NICU. Consequently, at least 50 patients are required in each group to achieve a probability of 80% to detect a 20% difference in mortality with a 95% CI. A two-tailed Fisher’s exact test was used when comparing groups regarding outcomes, demographic, etiological and clinical data. The Student T-test was used for the comparison of laboratory values. Multivariate analyses were performed to adjust for possible differences in case mixes of the control and intervention group. (Table 1). The main characteristics regarding demographic, etiological, clinical, and laboratory data are given in Table 1, not identifying any significant differences between the intervention and control group. A cerebral CT-scan on admission showed one or more of the following; tight sulci, edema, compressed cisterns or hydrocephalus, in 22/52 (42%) patients in the intervention group. However, CT-findings were lacking in the controls as this data was not recorded in the SQRM. The median time from primary admission to insertion of an EVD or a parenchymal ICP-monitor was 8 h:26 min (range 1 h:27 min–96 h) and the median duration of EVD management was 5 days (range 1.5–16 days). Five patients were treated with bolus doses of Mannitol upon admission, due to slit ventricles as seen on the CT-scan (n = 3), or a spinal opening pressure of . 500 mmH 2 O (n = 2). An EVD was applied in 48 cases, and a parenchymal ICP-monitor was inserted in 4 cases (slit ventricles; n = 2, technical problems; n = 1, coagulopathy; n = 1). When the EVD was applied, Csf was momentarily drained in the operating room in 48 cases, and in 38 (79%) of these cases, the neurosurgeon noted markedly elevated ICP during this procedure. In the NICU, 35 of 52 patients (67%) presented significant intracranial hypertension (ICP . 20 mmHg during . 5 min; Figure 3). A parenchymal ICP-monitor but not an EVD was established in three of these 35 cases. Of the remaining 17 patients, 11 exhibited ICP . 20 mmHg, but for shorter episodes than 5 min. In these cases, Csf was also drained through the EVD to maintain an ICP , 20 mmHg. In three additional cases, Csf was drained despite ICP , 20 mmHg (Figure 2). Thus, Csf-drainage was performed in altogether 46 (88%) of the 52 cases. Four of the remaining six patients had parenchymal probes instead of EVD. In aggregate, increased ICP at surgery and/or significant intracranial hypertension during the NICU-care was observed in altogether 43 (83%) of the 52 patients (Table 2). The peak levels of ICP occurred within the first 48 hours of NICU-care in all patients, except in one patient who exhibited the highest level on day three. Altogether, 49 (94%) of the intervened patients were administered specific ICP-decreasing treatment in the NICU (Figure 2, Table 3). Mortality was significantly decreased in the per protocol intervention group (5/52 = 10%) compared to the controls (16/ 53 = ...

Context 3

... ventricles. In these two and in another two cases, where an EVD could not be applied (technical problems: n = 1, moderate coagulopathy: n = 1) a parenchymal ICP-monitor (Codman, Johnson and Johnson Nordic AB, Stockholm) was used. Patients were treated in a 30 u sitting position. The temple was used as zero-point for mean arterial blood pressure and cerebral perfusion pressure. ICP was continuously registered in a computerized patient monitoring system, ICU-pilot ( m -dialysis AB, Solna, Sweden), and the treatment goals were ICP , 20 mmHg and cerebral perfusion pressure . 50 mmHg. Significant intracranial hypertension was defined as one or more episodes of ICP, continuously exceeding 20 mmHg for more than 5 minutes. As illustrated in figure 2, CSF-drainage through the EVD was the main treatment of increased ICP. The upper ICP limit triggering drainage was set at 20 mmHg. Additional ICP treatment was applied at the discretion of the treating physicians, guided by clinical, radiological and physiological findings. If signs of interstitial brain edema were observed on the CT-scan, bolus doses of hypertonic saline were given as osmotherapy, allowing S- Na levels of not higher than 160 mmol/L. Intracranial hypertension due to hyperaemia, detected and guided by trans-cranial doppler and/or jugular bulb monitoring, was treated with moderate hyperventilation aiming at a pCO 2 of 4.0–4.5 kPa. In cases with paracetamol resistant hyperthermia, an external cooling blanket (Thermo-wrap) was used, aiming at normothermia. If intracranial hypertension and brain edema persisted, despite previous corticosteroid treatment and above-mentioned ICP- decreasing interventions, 1 g of i.v. methylprednisolone was given. Last resort ICP treatment was a pentothal infusion, inducing a barbituate coma to lower cerebral metabolism, monitored with microdialysis and/or jugular bulb analyses. Clinical, etiological and laboratory findings, mental status (GCS and/or RLS) on admission, and the duration from admission to start of antibiotic and corticosteroid treatment were prospectively registered in the SQRM (Table 1). A CT-scan of the brain was performed in all patients in the intervention group and, based on clinical grounds, in 50 of the 53 cases in the control group. EEG was performed in selected cases to exclude non-convulsive seizures. A follow-up at 2–6 months after discharge included an evaluation of neurological deficits, Glasgow outcome score (GOS) and hearing impairment in all survivors according to the SQRM protocol. Audiometry was performed in selected cases. A power test was performed, based on available data [2,8– 10,12,14–16] suggesting a mortality of 40% with standard ICU treatment, and hypothesizing a 20% mortality when employing ICP-targeted treatment of ABM patients with severely impaired mental status at the NICU. Consequently, at least 50 patients are required in each group to achieve a probability of 80% to detect a 20% difference in mortality with a 95% CI. A two-tailed Fisher’s exact test was used when comparing groups regarding outcomes, demographic, etiological and clinical data. The Student T-test was used for the comparison of laboratory values. Multivariate analyses were performed to adjust for possible differences in case mixes of the control and intervention group. (Table 1). The main characteristics regarding demographic, etiological, clinical, and laboratory data are given in Table 1, not identifying any significant differences between the intervention and control group. A cerebral CT-scan on admission showed one or more of the following; tight sulci, edema, compressed cisterns or hydrocephalus, in 22/52 (42%) patients in the intervention group. However, CT-findings were lacking in the controls as this data was not recorded in the SQRM. The median time from primary admission to insertion of an EVD or a parenchymal ICP-monitor was 8 h:26 min (range 1 h:27 min–96 h) and the median duration of EVD management was 5 days (range 1.5–16 days). Five patients were treated with bolus doses of Mannitol upon admission, due to slit ventricles as seen on the CT-scan (n = 3), or a spinal opening pressure of . 500 mmH 2 O (n = 2). An EVD was applied in 48 cases, and a parenchymal ICP-monitor was inserted in 4 cases (slit ventricles; n = 2, technical problems; n = 1, coagulopathy; n = 1). When the EVD was applied, Csf was momentarily drained in the operating room in 48 cases, and in 38 (79%) of these cases, the neurosurgeon noted markedly elevated ICP during this procedure. In the NICU, 35 of 52 patients (67%) presented significant intracranial hypertension (ICP . 20 mmHg during . 5 min; Figure 3). A parenchymal ICP-monitor but not an EVD was established in three of these 35 cases. Of the remaining 17 patients, 11 exhibited ICP . 20 mmHg, but for shorter episodes than 5 min. In these cases, Csf was also drained through the EVD to maintain an ICP , 20 mmHg. In three additional cases, Csf was drained despite ICP , 20 mmHg (Figure 2). Thus, Csf-drainage was performed in altogether 46 (88%) of the 52 cases. Four of the remaining six patients had parenchymal probes instead of EVD. In aggregate, increased ICP at surgery and/or significant intracranial hypertension during the NICU-care was observed in altogether 43 (83%) of the 52 patients (Table 2). The peak levels of ICP occurred within the first 48 hours of NICU-care in all patients, except in one patient who exhibited the highest level on day three. Altogether, 49 (94%) of the intervened patients were administered specific ICP-decreasing treatment in the NICU (Figure 2, Table 3). Mortality was significantly decreased in the per protocol intervention group (5/52 = 10%) compared to the controls (16/ 53 = 30%; relative risk reduction 68%; p , 0.05; Table 4). Full recovery, defined as GOS 5 with preserved normal hearing, was significantly more frequent in the per protocol group; 28 patients (54%) versus 17 patients (32%) in the control group. Thus, the relative risk reduction for unfavorable outcome was 40% (p , 0.05). Three intervened patients suffered persisting severe neurological deficits, with outcomes of GOS 3, related to complications of ABM; sinus/intracranial venous thrombosis; n = 1, cerebral infarction; n = 1, and ventriculitis and hydrocephalus; n = 1. The remaining 44 survivors reached GOS 4 (n = 12) or 5 (n = 32; Table 2). Of the 28 intervened patients that fully recovered, including normal hearing, 19 (68%) had significant intracranial hypertension in the NICU and a further four had markedly elevated ICP noted only at surgery (Table 2). Significantly more episodes of ICP . 20 mmHg lasting . 5 min were observed among patients that succumbed, versus survivors (p , 0.01). Slit ventricles, not allowing drainage through the EVD, were found in three of the five fatal cases. All deaths occurred within one month after admission. Three of the five fatal cases in the treatment group died within a week, as did 11/16 fatalities among the controls. Two of the five fatalities in the intervention group, and 6 of the 16 controls that died, were immunocompromised (Table 1). S pneumoniae was the etiological agent in all fatal cases of the intervention group, and in 9 of 16 succumbed controls. The causes of mortality were cerebral herniation and/or ischaemia, verified by CT-scan and/or autopsy, in 4/5 (80%) intervened patients and in 13/16 (81%) controls. In the remaining cases, death was caused by septic shock and/or multi-organ failure. Lumbar puncture (LP) was avoided due to contraindication in two (suspected herniation in both) of the five fatal cases in the intervention group, and in four (suspected herniation: n = 2, coagulopathy: n = 2) of the 16 controls that died. Thus, the mortality rate among spinal tapped patients (Table 1) was 3/47 (6%) and 12/44 (27%), in the intervention and control groups, respectively. No patient died within 24 hours after LP and, of the intervened patients with persisting deficit, a significant clinical deterioration shortly after LP was not recorded in any case. Acute bacterial meningitis (ABM) in adults still remains a challenge for the clinician, and in comatose ABM-patients mortality rates of up to 62% have been reported [2,9,10]. ICP- targeting treatment of severe ABM has been reported with promising results [14–16,23] but is not routinely used. [4,22]. The present report is, to our knowledge, the first controlled study of neuro-intensive care and ICP-guided therapy using EVD in adults with ABM and severely impaired consciousness. The mortality was significantly decreased from 30% in the controls to 10% in the treatment group, and the incidence of full recovery was increased from 46% to 68% in the groups, respectively. Significant initial intracranial hypertension was observed in two thirds of the 28 patients that fully recovered, indicating that ICP-guided therapy may result in a satisfactory outcome, even in cases with severe ICP elevation. However, in concordance with earlier findings, extreme ICP levels were observed among the fatal cases [15–17,19,20,28]. The peak ICP occurred early during the course of the disease, as shown earlier [14–16]. In line with previous experience, cerebral herniation and/or infarction dominated as cause of mortality in both groups [2,13– 16,18], and a very aggressive course of ABM with fatal outcome was observed in five intervened cases [20,29,30]. A few case reports suggest that a decompressive craniectomy may be considered early in the course of disease in such cases [31–33]. The pathophysiological mechanisms resulting in increased ICP in ABM are multifactorial. The release of bacterial components in the subarachnoid space leads to an inflammatory response that contributes to increased permeability of the blood-brain barrier causing cerebral extracellular edema, impaired Csf-absorption with increased Csf-volume, a cytotoxic intracellular brain edema, and increased cerebral blood flow (hyperaemia); all ...