Background Some patients admitted to acute stroke units are diagnosed as stroke mimics. A minority have a functional neurological disorder (‘functional mimics’).
Aims To determine the incidence of functional stroke mimics admitted to a hyperacute stroke unit (HASU); to compare their clinical characteristics with medical mimics and stroke cases and obtain information about outcomes.
Methods Patients admitted to the King's College Hospital HASU between 2011 and 2012 were analysed. Data were obtained from the Stroke Improvement National Audit Programme (SINAP) database. Expert consensus diagnosis was used to classify functional mimics. Follow-up information was obtained from a retrospective case series in primary care over the year following discharge.
Results 1165 patients were admitted to the HASU; 904 patients with stroke (77.6%), 163 medical mimics (14%) and 98 functional mimics (8.4%). Functional mimics were significantly more likely to be female (63.3%) versus 49.7% medical mimics and 45.5% stroke, and younger (mean age (SD)) 49.1 (18.8) than medical mimic (63.5 years (16.7)) and stroke cases (71 years (15.5)). Weakness and slurred speech were the commonest presentations of functional mimics and diagnostic MRI was used more often. Clinician recorded visual and speech symptoms and neglect were significantly more frequent in patients with stroke than either mimic group. Of the 68 functional mimics on whom follow-up information was obtained, 40 (59%) were referred to another service most often for a psychologically-based intervention.
Conclusions Functional stroke mimics are an important subgroup admitted to acute stroke services and have a distinct demographic and clinical profile. Their outcomes are poorly monitored. Services should be developed to better diagnose and manage these patients.
Statistics from Altmetric.com
More and more people in the UK with suspected acute stroke are admitted to specialised units. For example in London eight hyperacute stroke units (HASUs) have been commissioned to improve access to thrombolytic therapies and standardise multidisciplinary care.1 ,2 Patients with stroke require rapid assessment and triage. While plain CT is excellent at detecting haemorrhagic stroke it has only 40% sensitivity for confirmation of acute ischaemic stroke.3 Thus it is inevitable that a proportion of those who enter such a pathway turn out not to have a stroke—‘stroke mimics’.4
The size of this proportion varies according to setting but has been estimated in one systematic review to be around 26% (95% CIs 17% to 44%).5 Within the group of ‘stroke mimics’ there is a subgroup whose symptoms are otherwise medically unexplained and in whom a psychiatric or psychological cause is suspected. These may be considered to have conversion disorder or a functional neurological syndrome6—and may be termed ‘functional mimics’.
The proportion of functional mimics presenting to HASUs is seldom recorded and where it is, the figures vary markedly. Gibson and Whiteley5 noted that of 813 patients in their review to have a final diagnosis other than stroke or transient ischaemic attack (TIA), 7.4% were functional with a further 5% ‘not specified’. A German study7 found 42 (6.48%) of 648 suspected patients with stroke actually treated with thrombolysis, turned out to be mimics of which 20 (47.6%) had conversion disorder. Reid et al8 in Scotland reported that of 374 presentations to a HASU, 116 (31%) were mimics and of these 15 (13%) were functional with a further 19 (16%) ‘unclear’. In this study we surveyed all patients admitted to a single HASU at King's College Hospital (KCH) from November 2011 to October 2012 who had been entered onto the Stroke Improvement National Audit Programme (SINAP) database. The acute stroke unit at KCH serves a catchment area in SE London of approximately 900 000 people. We aimed to determine the proportion of patients who after initial Stroke team assessment, which included a CT head, were deemed to have had a stroke but after investigation or further review were felt to be stroke mimics. We were particularly interested in the functional mimics so that we could determine their clinical and sociodemographic characteristics. We also carried out a follow-up of this group through their general practitioners (GPs) to gain an impression of continuing difficulties and referral practices.
Before arriving in accident and emergency (A&E), suspected patients with stroke are usually assessed by healthcare professionals such as a GP or the paramedics (ambulance) as being FAST (face arm speech test) positive.9 FAST-positive patients presenting with signs and symptoms suggestive of stroke are reviewed in the A&E department by a Stroke or Neurology Specialist Registrar. Patients in A&E undergo immediate CT imaging. In our unit we do not use MRI in the hyperacute stroke pathway. If the patient is regarded as a stroke case, they are considered for thrombolysis and admitted to the acute stroke ward. If a non-stroke diagnosis is thought likely at this stage the patient is referred to the acute medicine department for further management. Patients with a definite TIA enter a separate care pathway. Thus a proportion of patients with an unclear diagnosis or felt to be stroke mimics are filtered out at the A&E stage. The initial cohort forming the basis of this study comprises those patients that have been admitted to the acute stroke ward with an initial diagnosis of likely stroke.
All patients admitted to the acute stroke unit were reviewed by a stroke consultant within 24 h and managed according to standard protocols. In some a non-stroke diagnosis became evident during their stay on the acute stroke unit due to further brain imaging (eg, second CT head scan or MRI head), additional tests or clinical review. Those given a diagnosis of non-stroke at the time of discharge or transfer to another unit were considered in greater detail.
Data from all cases, stroke and mimics (excluding TIAs (symptoms <24 h; n=135)), were then extracted for further analysis.
Mimics’ case notes were reviewed by the study team. Each set was scrutinised independently by at least two senior clinicians, consultants in neurology, neuropsychology or neuropsychiatry and scored on a 4-point scale based on Carson et al10: 1=‘completely organic’, 2=‘partly functional and largely organic’, 3=‘largely functional and partly organic’ or 4=‘completely functional’.
The stroke database was used to obtain information regarding patients’ gender, and their age at the time of admission to the HASU, as well as the reason for admission (including presenting symptoms, medical history), diagnostic imaging and length of stay on the HASU.
For the functional mimics, contact information for each patient's GP was extracted in order to audit further information about the events during the year following discharge. Of particular interest was whether any referrals were made to other services, mental or physical health. This information was requested initially by fax and followed by up with two telephone calls to each patient's GP surgery.
Data were analysed in STATA using t tests and analysis of variance for parametric and χ2 and Mann-Whitney tests for non-parametric data.
Ethical approval to conduct an audit of patients admitted to the HASU at King's College Hospital was granted via the local National Health Service (NHS) audit monitoring committee.
The criteria for classifying each patient into the four functional-organic categories10 was first piloted and shown to have a good interrater reliability (intraclass correlation coefficient was 0.899; 95% CI 0.818 to 0.951). As an additional check, all cases that received a rating that was neither ‘completely organic’ nor ‘completely functional’ (ie, 2 or 3) were reviewed by a third person from the research team. This additional rating was used as the final classification. Following on from this, patients rated as ‘largely’ or ‘completely functional’ were pooled into a single group: ‘functional mimics’. The remaining cases that is, ‘medical’ mimics were re-reviewed once more by a consultant neurologist and a final diagnosis was given based on the records and results of investigation following the index episode. These were then aggregated into diagnostic categories.
There were 1165 patients altogether; 904 stroke (77.6%), 163 medical mimics (14%) and 98 functional mimics (8.4%; table 1). There was a higher proportion of female patients in the functional mimics group than in the medical mimics and patients with stroke (63.3% vs just under 50%). The functional mimics were younger than the medical mimics and patients with stroke (49 years vs 63.5 and 70.9, respectively).
Limb weakness and/or numbness were recorded in over half of the stroke cases, slightly less in the functional mimics and fewer still medical mimics. Facial weakness or numbness was recorded in over half the stroke cases but in fewer than 30% of both mimic groups. Similarly, less common symptoms putatively relating to posterior circulation including cerebellar syndromes were most common in the medical mimics, while visual or speech and language difficulties occurred in around a third of strokes but under a fifth of the mimic groups. Neglect was noted in a quarter of strokes but in none of the mimics. Altered consciousness (present on examination in hospital) was rare in all except medical mimics in whom it was noted in 15%. Laterality of symptoms was not recorded in the SINAP database but when examined in the functional mimics the figures for left versus right versus not applicable/bilateral were: 46 (49%) versus 16 (17%) versus 32 (34%), respectively (p=0.059).⇓
Medical mimics had a variety of conditions (table 2). The commonest was previous stroke with functional decompensation followed by focal seizures, migraine and vestibular disturbances (labyrinthitis, benign paroxysmal positional vertigo, etc). In comparison with functional mimics they tended to have more cardiovascular and cerebrovascular risk factors but fewer general medical syndromes/symptoms (asthma, migraine, etc, see table 1). MRI was used in 40% of functional and 30% of medical mimics but very seldom in patients with stroke. Functional patients had symptoms for longer before arriving in A&E and had briefer stays on the stroke unit (table 1).
From review of the functional mimics’ case records, isolated weakness was the commonest presentation followed by slurred speech. Alterations in alertness and consciousness were noted as well as a number of miscellaneous presentations such as tremor, confusion and difficulties swallowing. Many presentations occurred in combination and the presence of headache was common as an additional symptom (eg, with speech disturbance, weakness or sensory symptoms).
Information from GPs was returned on 68 (72.3%) of the 98 functional mimics regarding referrals. This group did not differ significantly from the remainder in terms of age and gender: 40 out of 68 (59%) were referred to another service in the year following admission to the HASU (table 3). This was most often for a psychologically based intervention such as the Improving Access to Psychological Treatments service (IAPT). A mental health condition in the year following admission was noted in 32 (47%; table 4) most commonly depression followed by ‘stress’ or a ‘stress-related condition’.
Stroke mimics overall accounted for 22.4% of admissions to a HASU in South London, in line with the published literature.5 The proportion of functional mimics was found to be 8.4% and is considerably greater than Gibson and Whiteley's estimate based on a literature review. However, our study is the first to have classified functional mimics on the basis of operational criteria and expert evaluation of the records by neuropsychiatrists. Reports for other units have been based on those patients actually receiving thrombolysis treatment or on those who have undergone sophisticated brain imaging, hence the proportion of mimics is correspondingly lower. For example, Vroomen et al11 in the Netherlands reported on 669 patients admitted to a specialist stroke unit after expert triage and investigation including CT and MRI. In that cohort, only 32 (5%) were finally classed as mimics but of these 13 had a conversion disorder. Over half of all mimics were under 50 years including six conversion cases. It is interesting that nearly 40% of functional mimics in our study had a non-hyperacute MRI (with obvious cost implications) compared to a minority of the medical mimics and very few stroke cases (see table 1).
Another reason for differences in reported rates of functional mimics may be the corresponding rates of ‘unclear’ or unspecified diagnoses. In our study this accounted for 4% of medical mimics only. Reid et al8 found a roughly similar prevalence of medical mimics but overall only 15/374 (4%) were functional while 5% were ‘unclear’; they also noted the preponderance of females (60%) in their mimics and lower mean age.8 Our data extend and clarify these figures by showing that it is the functional mimics which show the marked deviation towards younger age (as in11) and female preponderance, with the medical mimics similar to stroke cases in terms of gender.
The most common presentation of the functional mimics was weakness—of arm, leg or face—alone or in combination, with a trend towards left lateralisation.12 ,13 This was followed by speech disturbance and then sensory disturbance. Duration of symptoms prior to arrival at A&E was longer and skewed presumably due to less acute onset, or perhaps difficulty pin-pointing the onset in time. Headache was a common accompanying symptom:13 we found that about 18% of functional cases had been given a diagnosis of ‘migraine’ or ‘hemiplegic migraine’. This was in addition to the 13% of cases in the medical mimic group with migraine. However we resisted using the label of hemiplegic migraine in the absence of a prior history of migraine with hemimotor symptoms and/or a family history. The subjective experience of head pain plus transient or persisting sensorimotor symptoms in the absence of neuroimaging findings is clearly a diagnostic challenge.
Cortical stroke presentations, such as speech and language disruption, visual symptoms and neglect were highly significantly more associated with stroke than mimics. In the case of functional mimics this may be due to such presentations not forming part of the lay prototype of a stroke. The FAST mnemonic may change that although it may, inadvertently have led to ‘facial spasm’ being frequently mistaken for stroke.
Another contentious diagnostic issue arose when the patient had a prior diagnosis of stroke and presented with similar symptoms perhaps in the context of an infection, but without progression of clinical signs or further ischaemia after specialist investigations. This led to some cases being classed as medical mimics (ie, ‘decompensation’) although others were diagnosed ‘functional’ where anxiety and fear of recurrence appeared to be the dominant mechanism leading to misattribution of existing deficits to another stroke. A diagnosis of previous stroke was recorded in around a quarter of both mimic groups although this had not been verified.
The medical mimic group had a higher prevalence of vascular risk factors than functional mimics which presumably raised the suspicion of stroke at presentation. It should be noted that they were on average 14 years older than the functional group which nevertheless had more ill-defined medical conditions such as back pain and migraine. The latter might indicate increased symptom reporting or a tendency towards somatisation. Depression, though probably under-recognised in all groups was nevertheless noted in 17% of functional mimics and under 2% of medical mimics.
Our study methodology was novel in the way in which a positive diagnosis was made of functional disorder based on expert consensus. This probably increased the number of cases so diagnosed rather than them being assigned to an unclear category. However, it is acknowledged as a limitation of the study, that there is often genuine doubt about the diagnosis of conversion disorder especially where one is reliant on case note review and a relatively short ‘acute’ assessment period and without psychiatric input. Indeed, such patients are moved on quickly once stroke has been excluded. Factors which inclined raters towards a functional diagnosis included atypical neurological signs or those consistent with a functional disorder (eg, positive Hoover sign), rapid change or variability in symptoms, a history or coded references to psychiatric disorder or ‘stress’ or social circumstances suggesting a serious dilemma for the patient.14 We were however careful not to classify FSMs purely on the basis of a plausible ‘story’. Clinically, acknowledging genuine uncertainty is appropriate especially in the context of ongoing care but we erred on the side of a definite classification for the purposes of the survey. A small number had a history of several admissions with similar symptoms where there was good evidence for somatisation disorder and relatively chronic functional/conversion symptoms.
Another limitation was the little information we had in terms of prognosis after the acute episode. Most patients are transferred very quickly to a non-hyperacute stroke unit, district general hospital, medical ward, nursing home or home. Our request for GP records elicited a reasonable response rate (69.4%)—direct contact with patients would have been preferable but was not part of our protocol. We may be inclined to assume that those in whom information was forwarded (68 cases) but where there was no mention of onward referral (28 cases), the presenting physical problem resolved spontaneously. Furthermore, 13/40 cases had an explicitly mental health-related referral—mostly to an IAPT service suggesting some level of acknowledgement of a psychogenic aetiology. Seven (17.5%) were referred back to neurology plus other referrals such as those for exercise/physiotherapy, rheumatology or pain clinics or the district nurse might be interpreted as suggesting continuing concern about a ‘physical’ diagnosis but could equally represent a pragmatic management plan and means of accessing rehabilitative services that are sometimes denied overtly ‘functional’ patients.
On the other hand, if a patient sought medical help outside of the NHS or obtained it directly from a secondary care setting which failed to notify the GP—this would have remained unrecorded. Similarly if the patient remained disabled yet reluctant to call on their GP this too would have escaped detection. Another alternative is that functional mimic patients may have believed that they did indeed suffer a stroke and that they either did not accept, or understand the alternative diagnosis given to them, or more likely, may not have been given one. Hence some of these patients would have subsequently returned home believing that they ‘probably had a stroke’ and that there was nothing else to be done. It is our experience from tertiary neuropsychiatric clinics that there is a small population of patients with functional disorders living as ‘patients with stroke’ with long-term disability. Multidisciplinary approaches15 and physiotherapy-based treatments16–18 have been developed for functional neurological syndromes and are likely to be effective in FSMs.
In conclusion, approximately 8% of admissions to a hyperacute stroke unit had a functional neurological disorder mimicking stroke. These patients were younger than patients with stroke and more often female. Presenting symptoms were varied but weakness, speech disturbance and sensory symptoms were the most common. They were more likely to have a prior history of psychiatric disorder and GPs referrals during the year post episode tended to be directed towards psychological interventions. The role of headache and the diagnosis of migraine require clarification in this group. Our study may be limited in terms of generalisability. Hence, further prospective study of such patients is needed to improve diagnosis, determine outcomes along a range of dimensions and to put in place a cost-effective care pathway.
The authors acknowledge the work of all the staff on the Friends Stroke Unit, KCH for their contributions.
SG and RW are joint first authors.
Contributors ASD, RW, NK-B and PS conceived and designed the study and with SG, carried out the consensus diagnoses. SG conducted data extraction and data analysis supervised by ASD. SS, CB, VD collected follow-up data. LA extracted information on medical mimics. ASD drafted the paper. All authors contributed to editing the paper and all authors approved the final manuscript. ASD is guarantor. ASD was supported by the National Institute for Health Research (NIHR) Biomedical Research Centre at the South London & Maudsley NHS Trust and Institute of Psychiatry, King's College London.
Competing interests None declared.
Ethics approval This study conformed to local approval standards for audit.
Provenance and peer review Not commissioned; externally peer reviewed.
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.