Statistics from Altmetric.com
A 76-year-old woman presented to hospital by ambulance 45 min after developing sudden onset left face and arm weakness. She was taking methotrexate and prednisolone for rheumatoid arthritis. She had been unwell with fever and malaise for 3 days. On examination, her temperature was 39.1°C, pulse rate was 90 per minute and blood pressure 150/90 mm Hg. She was alert with no signs of meningism. There was a left-sided facial droop and left arm weakness. Chest and abdomen examinations were normal.
We assessed her on a treatment pathway for suspected stroke with a view to intravenous thrombolysis; her National Institute of Health stroke score was 5. Following CT scan of the head, we arranged a CT perfusion scan, which showed a small area of ‘mismatch’ in the right hemisphere adjacent to the lateral ventricle, consistent with an ischaemic penumbra (figure 1).
Our initial diagnosis was ischaemic stroke complicating systemic sepsis; the radiology seemed to support this. However, upon review, we felt that meningoencephalitis was more likely, given the prominent septic symptoms and the unusual location of the imaging abnormality. We, therefore, withheld intravenous thrombolysis and started immediate treatment with intravenous ceftriaxone 2 g twice daily and amoxicillin 2 g every 4 h.
Initial laboratory results showed a normal white cell count of 7.7×109/L (4.0–11.0) and a serum C-reactive protein concentration of 26 mg/L (<10). Chest X-ray and urinalysis were normal. Cerebrospinal fluid (CSF) (after antibiotics) was turbid with a white cell count of 960 × 106/L (≤5) (95% neutrophils and 5% lymphocytes); the red cell count was 640 × 106/L (0). CSF protein was raised at 1.1 g/L (0.15–0.45). Unfortunately, no CSF glucose was sent. CSF microscopy found no microorganisms and no growth on culture. An antigen test for Streptococcus pneumoniae in CSF was negative.
Blood cultures grew Listeria monocytogenes serotype 4 and we changed the antibiotics to intravenous piperacillin and tazobactam 4.5 g three times per day.
An MRI scan of the brain with gadolinium contrast 6 days after admission showed features consistent with meningitis and a small abscess in the region of the CT perfusion scan abnormality (figure 2).
Her weakness improved over the first few days after admission. She did not need neurosurgical intervention due to the relatively small size of the abscess and the resolution of her neurological symptoms and signs. She completed a 6-week course of intravenous antibiotics with the ‘Hospital at home’ team using a tunnelled peripheral line and made a complete recovery.
This case highlights the diagnostic challenges of rapid assessment of patients with suspected stroke and the risk of an incorrect diagnosis. There is now an imperative for haste in clinical assessment of suspected stroke, given the improved outcome from its prompt treatment. Consequently, there is less time to consider the differential diagnosis, to gather additional history or to monitor developing symptoms and signs; all which can worsen diagnostic precision. Neurologists traditionally have had the luxury of time and have rightly valued diagnostic accuracy above haste. However, in the context of assessment for thrombolysis, clinicians often take a probabilistic approach with a lower degree of certainty, bearing in mind that stroke has a much higher a priori likelihood than most other neurological diagnoses. The clinician, therefore, weighs the risk of missing the short opportunity for thrombolysis against that of getting the diagnosis wrong.
This case also highlights the potential for neurological input on the ‘stroke pathway’ to identify other less common diagnoses that still require rapid treatment, and hopefully to reduce misdiagnosis without delaying treatment with intravenous thrombolysis where it is needed.
This patient's rapid onset of focal neurological symptoms suggested a vascular cause. Although her prodromal symptoms and fever implicated an infective process, sepsis can occasionally precipitate or copresent with stroke. Conversely, intracranial sepsis with focal symptoms (especially with acute onset) is rare, even in people taking immunosuppressive treatment.
Symptoms may sometimes develop rapidly in conditions that generally have a slower onset, although perhaps with a lingering suspicion of inadequate or incomplete history taking. The explanation may be the sudden interruption of vascular supply by vessel occlusion, by haemorrhage or by unwitnessed or subclinical superimposed seizure activity.
Using agreed protocols and ‘pathways’ of stroke care has enabled rapid specialist assessment and treatment decisions, reducing the time to give intravenous thrombolysis. However, with overstretched emergency services, patients may be directed inappropriately to such pathways, following the ‘referral pathway of least resistance’. Furthermore, once on a particular care pathway, patients with alternative diagnoses risk receiving inadequate reassessment and inappropriate management, as their diagnosis remains a negative (in this case, ‘non-stroke’).
This patient was on our local stroke pathway—based on the National Institute for Health and Care Excellence (NICE) acute stroke pathway—which specifies the entry point as ‘clinically confirmed acute stroke’. A clinical diagnosis made rapidly and without access to all relevant information is necessarily imperfect. Additionally, clinicians may vary in their judgement of diagnostic certainty and risk–benefit regarding thrombolysis. It is tempting to gain additional diagnostic support from further imaging, especially to exclude alternatives. However, this approach may bring its own complications. CT perfusion in particular may help to identify patients with ischaemic stroke who would benefit from reperfusion therapy,1 but it may be challenging to interpret such scans in the context of other pathology. The characteristic ‘mismatch’ between flow and volume on CT perfusion sequences in ischaemic stroke penumbra may also result from variations in cerebrovascular anatomy and result from seizures, cerebral vasospasm, chronic infarctions and central nervous system infections.2 The increasing use of multimodal imaging may lead clinicians to fail to respect the traditional principle that radiology complements rather than replaces clinical judgement.
Clinicians involved in assessing patients with acute stroke must be alert to the challenges of rapid imperfect clinical diagnosis, and the problem of over-reliance on pathways of care. Above all, they must be competent to identify and treat alternative ‘mimic’ diagnoses. Such reassessment and pragmatic ongoing management should be part and parcel of the provision of hyperacute stroke care and requires neurological expertise at, or close to, the ‘front door’.
Clinicians involved in acute stroke assessment should be aware of clinical as well as radiological stroke ‘mimics’, particularly where multimodal imaging is performed.
Conditions mimicking stroke may require similarly urgent treatment; identifying them requires neurological expertise on the ‘stroke pathway’.
Contributors KK drafted the article which was edited and amended by TW. DH and IB edited and made suggestions to the text. MB assisted with preparation of the figures and interpretation of the imaging presented. TW prepared the article for resubmission.
Competing interests None declared.
Patient consent Obtained.
Provenance and peer review Not commissioned; externally peer reviewed. This paper was reviewed by Tom Hughes, Cardiff, UK.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.