OUP user menu

Acute central nervous system infections in adults—a retrospective cohort study in the NHS North West region

B.D. Michael, M. Sidhu, D. Stoeter, M. Roberts, N.J. Beeching, A. Bonington, I.J. Hart, R. Kneen, A. Miller, T. Solomon
DOI: http://dx.doi.org/10.1093/qjmed/hcq121 749-758 First published online: 24 July 2010


Background: Over the last 15 years, bacterial meningitis has received considerable attention, including national guidelines, whilst viral central nervous system (CNS) infections have been relatively neglected. A recent pilot study suggested that management of patients with suspected viral encephalitis was often suboptimal.

Aim: To examine the relative incidence, clinical features and management of suspected acute CNS infections in adults across the NHS North West Region.

Design: A multicentre cross-sectional retrospective cohort study at 10 hospitals across the region over 3 months (from September to December 2007). Following a screen of all patients who had cerebrospinal fluid (CSF) analysis or received intravenous aciclovir and/or third-generation cephalosporin, those with clinical features suspicious of a CNS infection were included. Management was compared with the national meningitis and regional encephalitis guidelines.

Results: Three hundred and eighty-five patients were screened; 217 patients had a suspected CNS infection and 44 (20%) had a CNS infection: 18 aseptic meningitis (one herpes simplex virus [HSV]-2), 13 purulent meningitis (four Streptococcus pneumoniae) and 13 encephalitis (three HSV-1). The median (range) time from admission to suspicion of CNS infection and to LP was longer for patients with encephalitis than meningitis [4 (0.3–312) vs. 0.3 (0.1–12) h, P < 0.001, and 23 (4–360) vs. 12 (2–48) h, P = 0.042, respectively]; and the median time to treatment was longer for aciclovir than cephalosporin [7 (0.5–312) vs. 3 (0.3–312) h, P = 0.002].

Discussion: Encephalitis was as common as purulent meningitis, and HSV as common as Streptococcus pneumoniae. However, the management of patients with encephalitis was worse than meningitis. National encephalitis guidelines are needed.


Central nervous system (CNS) infections are a neurological emergency requiring urgent investigation and treatment.1,2 Among the most common CNS infections are meningitis, in which the brain meninges are primarily affected, and encephalitis, when the brain parenchyma is primarily involved; however, the distinction between meningitis and encephalitis is not always clear clinically, as many of the symptoms and signs overlap, and the term meningoencephalitis is often used.1,3 In the UK, as in other developed countries, Streptococcus pneumoniae is the most common cause of bacterial meningitis in adults, whilst herpes simplex virus (HSV) is the most common cause of viral encephalitis. However, the number of patients with suspected and clinically diagnosed CNS infection is much greater than that of proven aetiology.4,5

Over the last 15 years, the management of acute bacterial meningitis in adults has received considerable attention in the UK with detailed epidemiological studies, vaccination programmes and the introduction of National Guidelines.6–8 In comparison, CNS infections caused by viruses have been relatively neglected. A recent pilot study from a teaching and secondary care hospital in our region suggested that the management of patients with suspected viral encephalitis may often be suboptimal, particularly with regard to delays in performing a lumbar puncture (LP) and starting treatment.5

To see if these preliminary findings are reproduced more widely, we therefore conducted a detailed retrospective study of adults with all suspected acute CNS infections at 10 hospitals in the NHS North West region. This allowed us to compare the epidemiology, clinical features and initial management of patients with suspected viral and bacterial CNS infections, focusing on the key issues of clinical predictors of a CNS infection, the role of neuro-imaging and LP and the initiation of treatment.


We undertook a multicentre cross-sectional retrospective cohort study over 3 months (from 1 September to 1 December 2007) of adults (>16 years) with suspected acute CNS infections admitted to 10 hospitals in the NHS North West Region serving a population of more than 2 million; seven were district general hospitals and three teaching hospitals, but none had tertiary infectious diseases or neurology units. Data were collected between 1 January and 1 September 2008. The study was approved by each of the Clinical Research and Audit Departments of the individual NHS Trusts involved.

To identify patients in whom a CNS infection might have been suspected, we screened electronic laboratory records for patients who had a cerebrospinal fluid (CSF) sample taken, and electronic pharmacy records for patients who had received intravenous aciclovir and/or an intravenous third-generation cephalosporin, which are the presumptive treatments for suspected acute viral encephalitis and bacterial meningitis, respectively (Figure 1). The case notes were then examined to determine whether there were clinical features suspicious of a CNS infection. These included fever or prodromal flu-like illness, headache, new onset seizures, focal neurological weakness, confusion/altered behaviour, decreased Glasgow coma score (GCS), signs of meningism (photophobia, neck stiffness, Kernig’s sign) or rash.1,4,5 Details, including whether the admitting team had suspected a CNS infection, were recorded on a standardized form by a member of the study team.

Figure 1.

Study profile: identification of patients with suspected CNS infection, clinical case definitions and microbiological/virological confirmation.

Data were collected at the Trusts where the patient had been identified and entered into an anonymized centralized database. Clinical management was compared with the standard of the British Infection Society (BIS) guidelines for the management of suspected meningitis and with a regional guideline for the management of suspected encephalitis;1,6 these indicate that patients with suspected CNS infections should have an urgent LP, unless they have one of the following contraindications: GCS <13 or declining (>2 points), seizures, focal neurological signs, strong suspicion of meningococcal septicaemia, systemic shock, coagulation disorder, immune compromise or signs of raised intracranial pressure (Cushing’s reflex, papilloedema, altered pupillary response, ocular palsies, decorticate/decerebrate posturing or abnormal respiratory pattern).

Case definitions

Patients were classified into clinical case definitions on the basis of their presenting features, initial CSF and imaging findings (Table 1).1,9 Patients that met the definition of purulent or aseptic meningitis or encephalitis were considered to have a clinically diagnosed CNS infection whereas those with meningism only or encephalopathy only were not. After microbiological analysis was completed patients were further classified as microbiologically or virologically confirmed, if a pathogen was identified.

View this table:
Table 1

Clinical case definitions

Clinical case definitionClinical and initial investigatory criteriaMicrobiological and virological confirmation
Encephalopathy onlyAltered consciousness with no evidence of inflammation in the CNS on imaging or CSF analysis (i.e. CSF white cell count <5/ml)
EncephalitisAltered consciousness with no other cause identified and evidence of CNS inflammation on imaging or CSF analysis (i.e. CSF white cell count ≥5/ml)Defined as ‘virologically confirmed’ if a pathogen was identified by culture or PCR of the CSF
Meningism onlyMeningism (headache with neck stiffness and/or photophobia) without evidence of altered consciousness and no evidence of inflammation on imaging or CSF analysis (i.e. white cell count <5/ml)
Purulent meningitisMeningism (as detailed above) with a CSF white cell count >1000/ml or between 100 and 1000/ml with a predominance of polymorphonuclear cells and a CSF: plasma glucose ratio <0.5 (or an unpaired CSF glucose <5 mmol)Defined as ‘microbioloigcally confirmed’ if a pathogen was identified by culture or PCR of the CSF or blood
Aseptic meningitisMeningism (as detailed above) in a conscious patient, with a normal CSF: plasma glucose ratio (>0.5) and either a CSF white cell count of 5–20 or 20–1000/ml with a lymphocyte predominanceDefined as ‘virologically confirmed’ if a viral pathogen was identified by culture or PCR of the CSF, or ‘microbiologically confirmed’ TB meningitis if Mycobacterium tuberculosis was identified by culture or PCR of the CSF or blood

Where cases met none or more than one of these clinical case definitions the most likely diagnosis as judged clinically was used. Patients who received antibiotics for purulent meningitis or an extra-CNS bacterial infection were considered to have received antibiotics appropriately.

Statistical methods

Data were analyzed with the SPSS programme (SPSS Inc., 2007), using the chi-squared test to compare categorical data, the Mann–Whitney test for non-parametric data and odds ratios with 95% confidence intervals where appropriate, with statistical significance defined as P < 0.05. Methodology and results are presented in accordance with the STrengthening the Reporting of OBservational studies in Epidemiology guidelines (STROBE; www.strobe-statement.org, www.annals.org).10 Epidemiological incidence data were calculated with the assumption that all patients within the catchment areas would attend the corresponding hospital and that people from outside of the catchment area would not attend the hospital.

The number of people within the catchment population was identified from the published literature for each of the 10 NHS Trusts involved and the proportion of adults was defined as 79.6% of the catchment population as per the general UK population.11


Identification and clinical diagnosis

Initially 385 patients were identified by CSF and pharmacy screens, of whom 217 (56%) had a suspected CNS infection (Figure 1). Of these 18 were identified by the pharmacy screen only, 116 by the CSF screen only and 83 by both screens. All patients identified were acute admissions. The median (range) age of these 217 patients was 41 (16–87) years and 99 (46%) were male. Ten of them did not have CNS infection suspected by the clinical team, even though they had relevant clinical features; one of these was ultimately diagnosed with aseptic meningitis, following a LP for presumed sub-arachnoid haemorrhage. After initial investigations, 44 (20%) met the clinical case definitions for CNS infection (18 aseptic meningitis, 13 purulent meningitis, 13 encephalitis); the 173 that did not included 32 (18%) with encephalopathy only, 55 (32%) meningism only and 86 (50%) other (Table 2). Patients with a CNS infection were significantly older than those with other diagnoses [median (range) 44 (17–84) years, vs. 37 (19–87) years, P = 0.016].

View this table:
Table 2

Diagnoses of 173 patients with suspected CNS infection who did not meet the clinical case definition of CNS infection

Diagnostic categoryDiagnosisNumber
Non-CNS infectionUnspecified viral illness42
Respiratory tract infection6
Urinary tract infection4
Otitis media1
Vestibulocochlear Varicella zoster1
VascularCerebrovascular accident6
Transient ischaemic attack2
Subarachnoid haemorrhage2
NeoplasticCNS lymphoma2
Cerebral metastases1
ImmuneGuillain–Barre syndrome2
Graft vs. host disease1
Chronic CNS infectionProgressive multifocal leucoencephalopathy3
MetabolicMetabolic encephalopathy3
PsychiatricAcute psychosis2
ToxicAlcohol withdrawal2
IatrogenicMedication side effect2
IdiopathicNo diagnosis reached34
Headache (unspecified)18
Idiopathic intracranial hypertension1
Transient global amnesia1
Pancreatic pseudocyst1
Epilepsy plus poor drug adherence2
First seizure2

The catchment population was 2 218 000 of whom an estimated 1 765 528 were adults giving an approximate annual incidence for any CNS infection of 9.9 per 100 000; 4.1 per 100 000 for aseptic meningitis, 2.9 per 100 000 for purulent meningitis and 2.9 per 100 000 for encephalitis. On average each hospital admitted ∼1.5 patients with a CNS infection per month.

Microbiological and virological diagnoses

CSF bacterial culture was performed for all 199 patients who had a LP and was positive for three patients (all with Streptococcus pneumoniae). Bacterial PCR was performed on the CSF of one patient which identified S. pneumoniae and the serum of one patient, positive for Neisseria meningitidis; both patients had negative CSF and blood cultures. All patients with microbiologically confirmed acute bacterial meningitis came from the group with a clinical diagnosis of purulent meningitis. Viral PCR was performed for five (31%) of the patients with aseptic meningitis and herpes simplex virus-2 (HSV-2) was identified in one case. CSF tuberculosis culture was performed for two patients with aseptic meningitis and was positive in both cases. The causative pathogen was not identified in 15 (94%) patients with aseptic meningitis. Only 4 (31%) of the encephalitis cases had a viral PCR performed, which identified herpes simplex virus-1 (HSV-1) in three cases. Thus the estimated annual incidences for pneumococcal meningitis, HSV-2 meningitis and HSV-1 encephalitis, per 100 000 adults were 0.9, 0.2, and 0.7, respectively.

Clinical features

For the 207 patients in whom a CNS infection was suspected, the median (range) time to suspicion was 1 (0.3–312) h for patients ultimately diagnosed with a CNS infection and 2 (0.2–720) h for those with other diagnoses (P = 0.259); within the former group, time to suspicion of a CNS infection was significantly longer for patients with encephalitis than meningitis [4 (0.3–312) vs. 0.3 (0.1–12) h, P < 0.001].

In Table 3, the clinical features of the patients are compared according to the diagnostic group. Compared to patients with other diagnoses, patients with CNS infections were more likely to have a history of foreign travel, a history of fever, a documented pyrexia on admission, or confusion or behavioural change on examination. However, the sensitivity and negative predictive values of these parameters were not sufficiently high to be clinically useful as predictors of a CNS infection. Both patients with tuberculous meningitis had recently travelled from Africa (Nigeria and Ghana). Two patients with acute CNS infections were immunocompromised due to human immunodeficiency virus infection. Overall, patients with CNS infections were not more likely to be immunocompromised than those with other diagnoses, nor to have a low GCS.

View this table:
Table 3

The clinical features for 44 patients with a CNS infection and 173 with other diagnosesa

Clinical case definition
Clinical featuresAseptic meningitis (n = 18)Purulent meningitis (n = 13)Encephalitis (n = 13)All CNS Infections (n = 44) (%)Other diagnosis (n = 173) (%)SensitivitySpecificityPPVNPVOR (95% CI)P-value
    Fever108321 (48)42 (24)447532832.85 (1.36–5.98)0.002
    Febrile prodrome87419 (43)63 (36)436423811.33 (0.64–2.74)0.409
    Headache1711533 (75)118 (68)753222831.40 (0.62–3.19)0.382
    Nausea108422 (50)84 (49)505121801.06 (0.52–2.160.864
    Vomiting105419 (43)63 (36)425921791.33 (0.64–2.74)0.409
    Photophobia126018 (41)55 (32)466727831.49 (0.71–3.09)0.253
    Immunocompromised1012 (5)8 (5)59520801.88 (0.26–10.54)0.350
    Travel3306 (14)7 (4)149646813.74 (1.04–13.35)0.028
    Seizures2158 (18)12 (7)109327792.98 (1.02–8.58)0.021
    Speech disturbance1124 (9)15 (9)99121801.05 (0.28–3.65)0.564
    Neck stiffness87116 (36)59 (34)426725811.10 (0.52–2.32)0.778
    Drowsiness32611 (25)25 (14)258631821.97 (0.82–4.71)0.093
    Confusion/Behaviour change041216 (36)33 (19)368133832.42 (1.11–5.29)0.014
    GCS<1505510 (23)30 (17)238325811.40 (0.58–3.35)0.411
    Pyrexia910322 (50)43 (25)507534863.02 (1.45–6.34)0.001
    Rash4116 (14)15 (9)149129811.66 (0.53–4.98)0.320
    Focal neurological signs2226 (14)31 (18)148216790.72 (0.25–1.99)0.5
  • aSensitivity, specificity, PPV, NPV, OR and P-values are given for all cases meeting the clinical case definitions of a CNS infection in comparison with those who did not (PPV: positive predictive value; NPV: negative predictive value, OR: odds ratio).


Data were available for 215 (99%) patients regarding the timing of both initial investigations (imaging or LP) and antibiotic and/or antiviral treatment (Figure 2). Most patients had imaging (n = 121, 56%) or treatment (n = 65, 30%) first; only 29 (13%) had the LP first. Sixteen (7%) did not have a LP at any point. Treatment was delayed for imaging in 52 (29%) of the 181 who had imaging and it was delayed for a LP in 41 (21%) of the 199 who had a LP.

Figure 2.

The order of investigations and treatment in the clinical management of 215 patients with suspected CNS infection.

In total, 183 (84%) patients underwent neuro-imaging; 144 had CT only, 5 had MRI only and 34 had both. One hundred and fifty-three (71%) patients had imaging before the LP, although in 101 (66%) there was no contraindication to an immediate LP. Of the 52 (24%) with clinical contraindications to an immediate LP, these were predominantly focal neurological signs (n = 17), seizures (n = 15) and a high suspicion of meningococcal septicaemia (n = 14).

The CT scan was abnormal in nine (17%) of these patients, but seven still required a LP to make the diagnosis (one with a suspected tuberculoma, two with enhancing cerebellar lesions ultimately diagnosed as progressive multifocal leucoencephalopathy, two suggestive of encephalitis, one hyperdense unilateral parietal lesion and one possible unilateral internal capsule infarct). Two patients had a definitive diagnosis made by the CT scan, and so did not subsequently need a LP (one with cerebral metastases and one with a cerebrovascular accident); In no cases was a radiological contraindication to a LP identified in the absence of clinical contraindications to a LP. One patient with a CT scan suggestive of encephalitis, ultimately proved to have a glioma diagnosed by MRI, whilst the other had pneumococcal meningitis diagnosed by CSF examination. Only one of the three patients with HSV-1 encephalitis had an abnormal CT which was thought to show a possible unilateral internal capsule infarct.

Of the 101 patients in whom there was no contraindication to an immediate LP, the neuro-imaging was normal in 88 (87%). In the 13 cases who had an abnormal CT, this typically showed periventricular ischaemia or atrophy; and in none did the CT scan demonstrate mass effect or brain shift precluding a LP and 12 (92%) patients went on to have a LP, with no complications. The one patient who did not have a LP had progressive multifocal leucoencephalopathy diagnosed by MRI.

Thirty-nine (18%) patients underwent an MRI scan a median (range) 5 days (2 h–23 days) after admission. MRI was abnormal in 20 (51%) patients, diagnosing generalized ischaemia (n = 6), cerebral atrophy (n = 3), progressive multifocal leucoencephalopathy (n = 3), acute cerebrovascular accidents (n = 3), CNS lymphoma (n = 2), cystic metastases (n = 1), glioma (n = 1) and a tuberculoma (n = 1).

The median (range) time from admission to LP was significantly longer for patients who had neuro-imaging first than for those who did not [18.5 (2–384) vs. 6 (1–72) h, P < 0.0001]. There was also a significant delay in performing the LP in those who proved to have encephalitis compared with meningitis [23 (4–360) vs. 12 (2–48) h, P = 0.042].

CSF analysis

The CSF opening pressure was recorded for only 71 (36%) of the 199 patients who had a LP; 41 had a pleocytosis and the white blood cell differential was recorded in 29 (70%); the protein was measured in 180 (90%) and although the CSF glucose was recorded in 176 (88%), a simultaneous plasma glucose was only taken in 74 cases (37%).

Of the 44 patients who met the clinical case definition of a CNS infection, 25 (57%) had antibiotic and/or antiviral treatment commenced before the LP of whom 6 (24%) had CNS infection confirmed microbiologically or virologically. For the 19 patients with a CNS infection who had the LP prior to treatment, 5 (26%) were confirmed microbiologically or virologically. Six (55%) of the 11 patients with microbiologically or virologically confirmed CNS infection had the diagnosis made by PCR of the CSF or serum rather than culture. Two of the three patients who had a positive CSF culture had the LP before treatment, whilst the third had one dose of a third-generation cephalosporin 10 h before the LP.


The median (range) time from admission to starting treatment was longer for aciclovir than cephalosporin [7 (0.5–312) vs. 3 (0.3–312) h, respectively, P = 0.002] (Table 4). However, there was no such delay for aciclovir treatment in patients who had the LP performed before imaging. The median (range) time from suspicion of encephalitis to aciclovir and of suspicion of meningitis to a third-generation cephalosporin was similar [2 (1.5–299) vs. 2 (0.3–48) h, respectively, P = 0.148]. Of the 112 patients who had both a LP and antimicrobial treatment, 68 (61%) had treatment prior to the LP.

View this table:
Table 4

Time from admission [hours (range)] to empirical treatment in patients with suspected CNS infection according to the order of investigations

Time to aciclovir (n = 48)Time to cephalosporin (n = 88)P-value
All patients7 (0.5–312)3 (0.3–312)0.002
Imaging then LP7.5 (0.5–312)3 (0.5–312)0.004
LP then Imaging5 (2–21)5 (0.3–76)0.725

Thirteen patients with purulent meningitis and 10 with non-CNS bacterial infections clearly needed antibiotic treatment; 84 additional patients received intravenous third-generation cephalosporin for a median (range) duration of 3 (0.25–13) days. One hundred and fourteen patients received antimicrobial therapy (a cephalosporin and/or aciclovir) that proved unnecessary, and had information available about the timing of the LP. Antimicrobial treatment was more likely in those who had started treatment before the LP than those who had started treatment after the LP [52 (74%) of 70 vs. 14 (32%) of 44 (P < 0.0001)].


The median (range) in-patient hospital stay was 12.3 (1–163) days for the 44 patients with a CNS infection. Microbiological advice was sought for 18 (38%) of them and neurological advice for 13 (28%). One patient was transferred to the intensive care unit and one to the regional neurological centre; both had HSV1 encephalitis complicated by seizures and reduced consciousness. Both survived, but with sequelae; one had a residual hemiplegia and speech disturbance and the other had epilepsy. A third patient had impaired hearing following pneumococcal meningitis. One patient who presented with suspected encephalitis proved to have a large cerebrovascular accident and subsequently died.


CNS infections remain an important cause of morbidity and mortality in the UK in the 21st century. Hospital Episode Statistics show more than 14 million finished consultant episodes in the UK NHS, using nearly 53 million bed-days.12,13 Although there are few good health economic studies, the burden to the NHS and society can be judged from medical negligence costs, which typically exceed three million pounds for the most serious cases.14

Over the last 15 years acute bacterial meningitis has attracted considerable attention in the UK, with increased awareness among the public and health care workers, national management guidelines, vaccination programmes and improved outcomes.6,7,15 Neurological diseases caused by viruses have received less attention, and are generally considered to be less important, but as far as we are aware, there has been no comparative study, examining the incidence, clinical features and management of acute viral and bacterial CNS infections in a single population.

In our study of adults admitted to 10 hospitals across the NHS North West region, we calculated an annual incidence for purulent meningitis of 2.9 per 100 000, for aseptic meningitis it was 4.1 per 100 000 and for encephalitis it was 2.9 per 100 000. Thus during the study period the incidence of encephalitis was the same as that of purulent meningitis. Our calculated incidence of purulent meningitis was similar to those reported from previous studies of adults in western industrialized settings of 0.6–4 per 100 00016 and the incidence of aseptic meningitis was lower than previous studies from similar settings 5.2–7.6 per 100 000,4,11,17,18 while the incidence of encephalitis was higher than previous studies from similar settings 1.5–2.2 per 100 000.12,17 We conducted our study during the winter months only; if anything this might have led to an over-estimate of the annual incidence of purulent meningitis, which is more common in the winter, and an under-estimate of viral meningitis, which is more common in the summer months; in non-tropical countries the incidence of viral encephalitis has no seasonality.1,16

In terms of aetiology, we found that HSV was equally as common as S. pneumoniae, both accounting for an annual incidence of 0.9 CNS infections per 100 000 adults (0.7 cases of encephalitis and 0.2 of meningitis for HSV). Although our study was limited in scope, the incidences we found were similar to those reported previously: 1.2–2.3 per 100 000 for pneumococcal meningitis in all ages,19 0.2–0.4 per 100 000 for HSV encephalitis in all ages13 and 1.4 per 100 000 adults for HSV meningitis in adults.17 Although the denominator in our study, based on the reported hospital catchment populations, is likely to be an overestimate, the fact that it was the same for bacterial and viral diseases allows for a meaningful comparison. Whereas all patients had bacterial culture of CSF performed, only a handful had CSF PCR for viruses, so the true incidence of HSV infection may be even higher.

There were also important differences in the clinical management between patients with suspected meningitis and suspected encephalitis in our study. The average time from admission to suspicion of CNS infection was significantly longer for encephalitis (4 h) than for purulent or aseptic meningitis (0.3 h). Moreover, there was a significant delay in performing the LP in patients with encephalitis as opposed to meningitis, and patients with suspected meningitis had antimicrobial treatment commenced sooner after admission than those with suspected encephalitis.

There may be a number of possible explanations for these differences including the lack of awareness of the clinical features of encephalitis, and the broad differential diagnosis of patients with confusion or altered behaviour, focal neurological signs or seizures; of note less than half the patients with encephalitis had a reduced GCS, but nearly all had altered behaviour or confusion. In addition the management of patients with suspected meningitis appears to have improved following the publication of British Infection Society guidelines, particularly with regard to time to treatment.4,6

Compared to patients that ultimately had other diagnoses, those with CNS infections were more likely to have a travel history, altered behaviour or confusion, and pyrexia. However none of these features in isolation, or in combination, could identify such patients with sufficient sensitivity or specificity to reliably diagnose a CNS infection. This can only be done with CSF analysis.1,20 The management guidelines and reviews for meningitis and encephalitis emphasize the need for an urgent LP in all patients, unless there are clinical contraindications.1,6,20,21 The LP gives an immediate indication of whether there is a CNS infection, whether this is more likely to be bacterial or viral, and subsequently, what the pathogen is.1,20 This allows rationalization of treatment and avoidance of unnecessary antimicrobials, thus minimizing the chances of resistance developing, and of avoidable iatrogenic events; these include nephrotoxicity with aciclovir, and opportunistic infection with cephalosporins.22–24 In our study many patients were exposed to antibiotics they did not require, which might have been avoided with an earlier LP. There was a significantly higher percentage of patients who received antibiotics that proved to be unnecessary if treatment was started before rather than after the LP.

We found, as have others previously, that the majority of patients underwent unnecessary imaging before the LP that can cause delays in treatment.4,5,25,26 Only 13% of the patients in this study had a LP undertaken before any imaging or treatment was started. Patients with encephalitis that did not have imaging before a LP were started on aciclovir sooner. Although some have postulated that a CT scan first may obviate the need for a LP, in our study only two patients who had a CT first did not subsequently need an LP performing. Encouragingly, the median time from admission to starting aciclovir treatment in this study was 7 h; this is a considerable improvement on the median 48 h, from our previous pilot study, which was conducted 3-years ago at one of the hospitals included in the current study.5

In summary, our study conducted across 10 hospitals in the NHS North West Region showed that in adults clinically diagnosed encephalitis was as common as purulent meningitis, and that HSV was a more common cause of CNS infection than previously thought. The management of patients with suspected encephalitis was worse than that of suspected meningitis in terms of time to suspicion of the diagnosis, time to LP and time to initiation of treatment. National encephalitis guidelines are needed to assist clinicians in the management of these complex patients where early investigation and treatment has a significant effect of morbidity and mortality.


The authors would like to acknowledge the work of all the microbiologists, infectious disease physicians, virologists, neurologists, acute physicians and junior doctors who, as members of the North West Neurological Infections Network (www.neuroinfections.net), have been involved in data collection for this study: Holland Mark, Blanchard Tom, Klapper Paul, Deepak Doddabele, Atherton Steve, Chandy Rajiv, Qazzafi Zaman, Croall John, Silverdale Monty, Khan Rizwan, Odonoghue Maire, Kedia Ravindra, Jones Kevin, Mckee David, Ealing John, Varma Anoop, Pattrick Martin, Dick Jeremy, Sussman Jon, Marinaki Katina, Maher Fergus, Lehane Daragh, Zafar Saman, Moss Nicole, Zakeri Bijaseri, Costa Janet, Tiongo Aisha, Newby Catherine, Bengashir Najla, Reid Fiona, Crook Helena, Eleftheriadou Viktoria, Loizou Constantinos, Parmar Vijal, Menakaya Chinyelu, Morocos Sarah, Francis Natalie, Sheehan Alison, Symons Gemma, Evans Sorcha, Zadeh Reza, Ali Sidrah, Lilleker James, Pearson Scott, Wai Ch'ng Soon, Nemit Ilhaq, Keenan Mark, White Katherine and Warner Jo. T.S. is an MRC Senior Clinical Fellow and is the lead for the national UK Encephalitis Guideline Group and B.M., R.K., I.J.H., N.J.B. are members. We thank the Liverpool NIHR Biomedical Research Centre in Microbial Diseases and the North West Development Agency for support. All authors grant an exclusive license to publish to the Association of Physicians.

Conflict of interest: None declared.


  • *For a full list of collaborators please see acknowledgments section.


View Abstract