Current international guidelines recommend that a cranial computed tomography (CT) be performed on all HIV-positive patients presenting with new onset seizures, before a lumbar puncture (LP) is performed. In the South African setting, however, this delay could be life threatening. The present study sought to measure the number of cranial CTs that contraindicate an LP and to predict which clinical signs and symptoms are likely to pose an increased risk from LP.
The study was performed at a district level hospital in Western Cape Province. Data were collected retrospectively from October 2013 to October 2014. Associations between categorical variables were analysed using Pearson’s chi-squared test. Generalised linear regression was used to estimate prevalence ratios.
One hundred out of 132 patients were studied. Brain shift contraindicated an LP in 5% of patients. Patients with brain shift presented with decreased level of consciousness, focal signs, headache and neck stiffness. Twenty-five per cent of patients had a space-occupying lesion (SOL) (defined as a discrete lesion that has a measurable volume) or cerebral oedema. Multivariate analysis showed a CD4 count <50 (
HIV-positive patients with seizures have a high prevalence of SOL and cerebral oedema but the majority of them are safe for LP. Doctors can use clinical parameters to determine which patients can undergo immediate LP.
New onset seizures in HIV-positive adults have been reported to have an incidence of around 6%.
Raised intracranial pressure is defined, in the acute setting, as pressure within the cranial vault that exceeds 20 mmHg – 25 mmHg for more than 5 minutes (Roytowski).
Brain shift, however, has been associated with an increased risk of brain herniation from LP, and can be measured by a CT scan. Cranial CT can demonstrate hemispherical shift and gross generalised brain swelling, both contraindications to LP.
A prospective South African study performed at Chris Hani Baragwanath Hospital found that 53.3% of HIV-positive patients presenting with new-onset seizures had a SOL.
In developed countries, where CT scanners are readily available, it is feasible to scan all HIV-positive individuals presenting with seizures soon after presentation. In many South African hospitals, however, arranging a CT scan is time-consuming and is often not done after working hours. A working diagnosis and commencing treatment based on LP results can save time and lives whilst waiting for a cranial CT.
The purpose of the present study was to assess if current international guidelines that recommend a cranial CT before LP on all HIV-positive patients presenting with seizures, is applicable to the HIV-positive population of Western Cape Province, South Africa.
This was a cross-sectional, observational study conducted at Mitchells Plain Hospital (MPH) from October 2013 to October 2014. MPH is a large metro district hospital providing level one (district level) as well as some level two (general specialist) care to approximately 440 000 people from the areas of Mitchells Plain, Phillippi and Crossroads. The study population comprised HIV-positive patients with seizures, presenting for cranial CT. The CT request forms filed in the Radiology Department at MPH were used to identify patients to be included in the study. Initially, a broad manual search for any patient whose cranial CT was requested for seizures, was undertaken. The search was then streamlined to only include HIV-positive patients and seizures. Where there was doubt regarding HIV status, confirmation was sought via the National Health Laboratory Service (NHLS) database. Patients under the age of 18 and patients whose seizures were acutely trauma related, were excluded from the study.
Data were extracted onto a data collection sheet from patient folders, the NHLS database and the Radiology Department database where necessary. Patients’ demographic details, CD4 count, risk factors for seizures, seizure history, seizure type, symptoms and signs of brain shift on presentation, CT information, LP findings (
Lumbar puncture findings.
The data were then captured onto a Microsoft Excel spreadsheet and were analysed using STATA 13.0. Associations between categorical variables were analysed using Pearson's chi-squared test and prevalence ratios (PRs) with 95% confidence intervals (CIs). Variables that were considered as potential risk factors, such as age, gender, CD4, seizure history, type of fit and signs and symptoms of brain shift, were included in the model for ‘generalised linear regression analysis’ to estimate the PR. For all analyses, a
Ethical approval was obtained from the University of Cape Town research ethics committee, and the Western Cape Provincial Health Research Committee granted permission to conduct the study. Patients’ identities were protected as their names were not recorded.
A total of 132 CT request forms, for the period October 2013 – October 2014, identified suitable patients to be included in the study. This figure suggests a minimum of 11 cranial CTs performed monthly at MPH on HIV-positive patients presenting with seizures. The first 100 patients whose folders were accessed were studied.
Baseline characteristics of patients are presented in
Baseline characteristics of patients.
Characteristics | |
---|---|
Age | |
Median | 38 years |
IQR | 32–45 years |
CD4 count | |
<50 | 20 |
50–200 | 23 |
200–350 | 24 |
>350 | 32 |
Unknown | 1 |
Risk factors for seizures | |
Systemic illness | 38 |
Alcohol/substance use | 21 |
History of head injury | 13 |
Abnormal chemistry | 8 |
Seizure history | |
New onset seizure | 81 |
Known epileptic | 19 |
Type of seizure | |
Generalised | 74 |
Focal | 16 |
Undocumented | 10 |
Clinical signs and symptoms suggestive of brain shift | |
Headache | 21 |
Vomiting | 4 |
Visual disturbances | 1 |
GCS < 15 | 38 |
Focal signs | 11 |
Neck stiffness | 10 |
Papilloedema | None documented |
IQR, interquartile range; GCS, Glascow Coma Scale.
Brain shift was seen on CT scan in 5% of the patients. Details of these patients are given in
Description of patients with brain shift.
Patient | CD4 count | Gender | Type of seizure | Symptoms | CT finding | LP done | Adverse effect from LP | Diagnosis | Prognosis |
---|---|---|---|---|---|---|---|---|---|
1 | <50 | Male | Generalised | Focal signs Impaired consciousness GCS 14 | Active space-occupying lesion | No | Not applicable | Toxoplasmosis | Referred tertiary institution |
2 | 50–200 | Female | Focal | Focal signs Headache GCS 15 | Active space-occupying lesion | No | Not applicable | Toxoplasmosis | Recovery and discharge |
3 | 50–200 | Female | Unrecorded | Impaired consciousness GCS 13 | Generalised cerebral oedema | Yes, after second scan | No | Meningitis bacterial/TB | Recovery and discharge |
4 | 200–350 | Male | Generalised | Impaired consciousness Neck stiffness GCS not documented | Active space-occupying lesion | Yes, before scan | No | Tuberculoma/TB meningitis | Recovery and discharge |
5 | 200–350 | Female | Generalised | Impaired consciousness GCS 14 | Localised cerebral oedema | No | Not applicable | Chronic haematoma/empyaema | Referred tertiary institution |
Results of cranial CTs are set out in
Computed tomography scan findings.
Generalised linear regression was used to predict patients with SOL/cerebral oedema (
Univariate and multivariate analyses of potential predictors of patients with active space-occupying lesion and cerebral oedema.
Potential predictors for at-risk patients | Totalnumber of patients | Percentageof patients with active SOL or cerebral oedema | Univariate | Multivariate | ||||
---|---|---|---|---|---|---|---|---|
PR | 95% CI | PR | 95% CI | |||||
Age | ||||||||
<40 | 58 | 32.8 | 2.3 | 1.0–5.2 | 0.0352 | 1.8 | 0.7–4.5 | 0.236 |
CD4 | ||||||||
> 350 | 32 | 3.1 | 1 | - | - | 1 | - | - |
200–350 | 24 | 25.0 | 2.3 | 1.4–3.8 | 0.0143 | 8.3 | 1.0–71.0 | 0.053 |
50–199 | 23 | 34.8 | 2.7 | 1.7–4.4 | 0.0017 | 7.5 | 0.9–62.0 | 0.063 |
<50 | 20 | 50.0 | 3.7 | 2.1–6.6 | 0.0001 | 10.1 | 1.2–85.4 | 0.033 |
Clinical presentation | ||||||||
Asymptomatic | 44 | 6.8 | 0.2 | 0.1–0.5 | 0.0002 | 0.3 | 0.1–1.3 | 0.118 |
Focal signs | 11 | 72.7 | 3.8 | 2.2–6.7 | 0.0001 | 2.0 | 0.8–5.1 | 0.154 |
Neck stiffness | 10 | 50.0 | 2.3 | 1.1–4.7 | 0.0543 | 1.0 | 0.4–2.8 | 0.995 |
Vomiting | 4 | 75.0 | 3.3 | 1.7–6.4 | 0.0184 | 1.5 | 0.4–6.2 | 0.586 |
GCS < 15 | 38 | 42.1 | 2.9 | 1.4–5.9 | 0.0020 | 1.0 | 0.4–2.7 | 0.984 |
PR, prevalence ratio; CI, confidence interval; GCS, Glascow Coma Scale.
LPs were performed on 68% (
The proportion of patients who did not have a clear diagnosis on discharge comprised 23% (
Uptodate,
Despite the expected increased prevalence of an active SOL (12%) and cerebral oedema (13%) in the population studied, the actual number of patients who had a contraindication to LP on CT scan was small (5%). This finding prevented us from drawing any significant statistical conclusion from the descriptive results in
The present study confirmed recommendations that a decreased level of consciousness and focal signs are significant predictors of patients at risk for brain shift. It also found that vomiting and neck stiffness might also be positive predictors of SOL/cerebral oedema, in HIV-positive patients presenting with seizures. Of significance is the finding on multivariate analysis that a CD4 count <50 is associated with increased risk for SOL/cerebral oedema. A CD4 count <50 is a specific predictor and should be given more weight in clinical decision making. Two asymptomatic patients, with CD4 <50, underwent LP and subsequently had an active SOL on cranial CT. It must be stressed, however, that none of the asymptomatic patients was unsafe to LP.
Papilloedema has been described as a contraindication to LP. The patients in our study were almost entirely not examined for papilloedema; this may be owing to the fact that it is often difficult to perform ophthalmoscopy in a bright and busy emergency room, especially when doctors’ experience with ophthalmology is limited. Papilloedema, however, is a late finding of raised intracranial pressure, and guidelines from Queens University recommend performing LP even when the optic discs cannot be visualised,
Extensive research was performed by the Swedish Infectious Disease Society regarding the comparative risk between immediate LP before CT and the risk of delayed LP (and inevitably delayed treatment) in adults with suspected acute bacterial meningitis. Hypothetical calculations of these risks, in different clinical settings with varying probabilities of cerebral mass lesions and acute bacterial meningitis (ABM), were presented. The authors worked on the premise that, although there is little evidence of an association between LP and brain herniation in acute bacterial meningitis, there is sufficient evidence of an association between LP and brain herniation in patients with cerebral mass lesions. The risk of brain herniation, associated with LP, in patients with cerebral mass lesions, however, is small and was assumed to be between 1% and 2%. It was concluded that where a patient had no clinical signs to indicate a SOL, an immediate LP will be advantageous; this would apply even where the probability of ABM is >0.5%. This research has led to the revised Swedish recommendation for early LP in 2009, which removed impaired immunity and new onset seizures as indications for a preceding cranial CT.
An American study by O’Laughlin et al.
Cranial CT provided a diagnosis for seizures in more than 50% of our patients and it should remain an important diagnostic tool in our population. What is questionable is the need to provide our patients with urgent cranial CTs to rule out brain shifts before performing an LP. There are substantial financial and clinical implications that arise from this recommendation. The Hasbun study
Apart from the financial implications of arranging urgent CT scans, there is the clinical consideration. The highest number of deaths in our study was from CCM (3 out of 9 deaths), followed by tuberculous meningitis (TBM) (2 out of 9 deaths). The high case fatality rate in CCM, predicted by the WHO to be between 35% and 65% in sub-Saharan Africa prompted the recommendation that patients with a CD4 less than 100 have early screening for the disease.
The most significant limitation of the present study was the sampling strategy. Ideally, to measure the number of people who suffered immediate death after LP, it would have been necessary to identify all patients who had an LP. By identifying patients from their CT request forms, the study overlooked a possible group of patients who might have suffered immediate cerebral herniation post LP and never survived to have had a CT. Consultation with the Head of the Casualty Department, however, revealed that no patient, to his knowledge, suffered cerebral herniation from LP at MPH. Furthermore, at least 50% of our patients had an LP before CT and reported no complications.
The study design was adequate to report on the number of CT scans where LP was contraindicated. The small study sample and the small percentage of patients with brain shift prevented us from predicting any statistically relevant factors for brain shift. Enough information was available to describe these patients and it is not unreasonable to conclude that any decreased level in consciousness or focal signs, excluding isolated cranial nerve palsies, should contraindicate LP before CT in patients with HIV and seizures. Doctors working in hospitals with no CT scanners may benefit the remaining patients by performing LP before CT.
Another limitation is that our research conclusions are based on doctors’ clinical impressions and not necessarily measurable information. Neck stiffness, for example, is a subjective clinical finding and should not be used in isolation to decide on management steps. Similarly, patient diagnoses were not based on hard facts, owing to the low detection of organisms on CSF microscopy and culture as well as the absence of histology on SOLs. In such cases, doctors used their clinical judgement as well as evidence of disease elsewhere to make a diagnosis and start treatment. CCM and toxoplasmosis were, however, definitive diagnoses, based on positive India ink staining or cryptococcal antigen testing and positive toxoplasmosis serology, respectively.
The present study was a retrospective folder review, with all CT scans reported by the same radiologist, and might have been more reliable if the scans had been seen by two radiologists, to reduce any interpretation bias.
HIV-positive patients with seizures have a high prevalence of SOLs and cerebral oedema but the majority of them are safe for LP. Indicators such as a decreased level of consciousness, focal signs, vomiting, neck stiffness and a CD4 count <50 should alert doctors to the possibility of at-risk patients. All the asymptomatic patients were safe for LP but should still have undergone non-urgent cranial CT owing to the limited occurrence of SOL and cerebral oedema in this group (6.8%). CCM accounted for the highest mortality, and doctors need to be more vigilant in performing serum cryptococcal latex antigen tests (serum CLATs) if LP is delayed. It is imperative that results of these tests be followed up promptly so that patients with CCM can be identified and treated early. There were no adverse events reported after any LPs performed on the patients in the present study.
The authors declare that they have no financial or personal relationships which may have inappropriately influenced them in writing this article.
S.M. (University of Cape Town) contributed the bulk of the data collection, collation and interpretation; the literature review and the writing of the study. A.R. (Mitchells Plain Hospital) contributed to the interpretation of CT reports and general radiological concepts, provided the patients’ CT request forms for source patient identification, and generally played a supportive and advisory role. E.d.V. (Mitchells Plain Hospital) was the official supervisor of the research, recommended the topic, assisted with the study methodology, checked data, assisted with statistical conclusions, and provided a mentor role throughout.