Article Text

Download PDFPDF
Original research
Mortality in patients with psychogenic non-epileptic seizures a population-based cohort study
  1. Le Zhang1,
  2. Ettore Beghi2,
  3. Torbjörn Tomson3,
  4. Massimiliano Beghi4,
  5. Giuseppe Erba5,
  6. Zheng Chang1
  1. 1 Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden
  2. 2 Department of Neuroscience, Istituto di Ricerche Farmacologiche Mario Negri IRCCS, Milano, Italy
  3. 3 Clinical Neuroscience, Karolinska Institutet, Stockholm, Sweden
  4. 4 Department of Mental Health, AUSL della Romagna, Cesena, Italy
  5. 5 Department of Neurology, University of Rochester Medical Center, Rochester, New York, USA
  1. Correspondence to Dr Ettore Beghi, Department of Neuroscience, Istituto di Ricerche Farmacologiche Mario Negri Dipartimento di Ricerca Neuroscienze, 20156 Milano, Lombardia, Italy; ettore.beghi{at}marionegri.it

Abstract

Objectives To compare mortality, comorbidities and causes of death in people with psychogenic non-epileptic seizures (PNES), epilepsy and the general population.

Methods Using linkage of multiple Swedish national registers, we identified individuals with incident diagnosis of PNES, epilepsy and conversion disorder with motor symptoms or deficits, and 10 controls for each. Main outcome was all-cause mortality. Causes of death were categorised into non-natural (eg, suicide, injuries) and natural. Conditional Cox regression models were used to estimate HRs and 95% CIs for mortality. HRs were adjusted for socioeconomic factors and psychiatric comorbidities.

Results Included were 885 individuals with PNES, 50 663 with epilepsy and 1057 with conversion disorder and motor symptoms or deficits. We found 32 (3.6%) deaths among individuals with PNES, compared with 46 (0.5%) deaths in controls, giving an adjusted HR of 5.5 (95% CI 2.8 to 10.8). Patients with epilepsy had a 6.7 times higher risk of death (95% CI 6.4 to 7.0) compared with individuals without epilepsy. The association between conversion disorder with motor symptoms or deficits was non-significant after adjusting for psychiatric comorbidities. PNES carried a higher risk of natural (HR 8.1, 95% CI 4.0 to 16.4) and non-natural causes of death (HR 15.3, 95% CI 3.0 to 78.6). Suicide ranked high in patients with PNES (18.8%) and conversion disorder with motor symptoms and deficits. The association between PNES diagnosis and all-cause mortality varied with age and time since diagnosis.

Conclusion Like epilepsy, PNES carries a higher than expected risk of both natural and non-natural causes of death. The high proportion of suicides requires further investigation.

  • epilepsy

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study. The data used in this study are national register information made under ethical permission. The authors had no special privileges in accessing the data. Dissemination of personal information is regulated by the Swedish Secrecy Act. In accordance with Swedish law, researchers seeking access to individual-level data must apply for permission through a Research Ethics Board (etikprovningsmyndigheten.se) and from the National Board of Health and Welfare (socialstyrelsen.se/en/statistics-and-data/statistics/).

Statistics from Altmetric.com

Request Permissions

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.

Introduction

Psychogenic non-epileptic seizures (PNESs) are paroxysmal events characterised by altered responsiveness, with or without movements and behaviour resembling epileptic seizures but lacking the neurobiological origin of epileptic seizures.1 2 PNESs are considered an expression of conversion disorders and are generally classified in the International Classification of Disease, 10th Version (ICD-10) as ‘Dissociative convulsions’.3 Capturing a typical event during video-EEG monitoring (VEM) is considered the gold standard for diagnosis.4

Previous studies show that individuals with PNES have an increased prevalence of psychiatric illness including depression, personality disorder, epilepsy, substance use disorder and suicidality.5 These psychiatric illnesses are found to be associated with shortened life expectancy.6 For example, it is well established that epilepsy is associated with a 2–3 fold increased risk of death.7 However, whether PNES itself is associated with increased mortality remains unclear. Only recently, reports have emerged on increased mortality among patients with PNES.8–10 In these studies mortality was found to be increased. However, these samples were small,8 based on highly selected patients from epilepsy centres,10 or have no confounding adjustment.9 In addition, information on the specific causes of death were lacking.9 This prompted us to investigate mortality in patients with PNES in a nationwide population-based sample. Using linkage of multiple Swedish national registers, we have previously identified a cohort of 885 patients with an incident diagnosis of PNES between 2001 and 2013 and and for each PNES case 10 matched controls who were PNES free at the time of PNES diagnosis of the corresponding case11 We followed the same cohort of PNES cases and control groups (patients with epilepsy and normal controls) to compare mortality, comorbidities and causes of death. In addition, we verified if the risk of death in people with PNES differed from other conversion disorders. PNES cases and controls were followed in the Swedish cause of death register, and the study aims were: (1) To determine the HR for overall mortality associated with PNES; (2) To examine the risk for natural and non-natural causes of death and (3) To compare the mortality among PNES patients with patients with a different type of conversion disorder and with epilepsy.

Methods

Data sources

Using a unique personal identification number,12 we linked several Swedish nationwide registers, including: (1) The Swedish Total Population Register (TPR),13 which covers demographic information on all country residents since 1968; (2) The National Patient Register (NPR), which has nationwide coverage for inpatient care since 197314 and specialist (non-general practitioner) outpatient care since 2001. Every record has a discharge date, a primary discharge diagnosis and multiple secondary diagnoses assigned by the treating medical doctor according to the ICD codes (ICD-7 before 1969, ICD-8 from 1969 to 1986, ICD-9 from 1987 to 1996, and ICD-10 since 1997); (3) the Cause of Death Register,15 which contains information on all deaths since 1952 (complete coverage since 1961) based on ICD codes with records of underlying and contributing causes of death; (4) the Migration Register, which records all migration flows in and out of Sweden; and (5) the Longitudinal Integrated Database for Health Insurance and Labour Market (LISA), which integrates existing data from the labour market, and educational and social sectors.16 Through the linkage of multiple national registers, we created three nationwide retrospective cohorts for PNES, epilepsy and conversion disorder with motor symptoms or deficits, and compared their mortality in relation to the respective comorbidities and socioeconomic factors.

Study population

The study population included all individuals born in Sweden between 1941 and 2009 and living in Sweden at any time between 2001 and 2013 (N=7,345,710). We identified individuals who received an incident diagnosis of PNES during 2001–2013 recorded in the NPR and were younger than 60 years at the time of their incident diagnosis of PNES. For each case with PNES, we randomly selected 10 controls from the TPR by incidence density sampling,17 who were alive and living in Sweden, individually matched to the cases by sex and year of birth, and PNES-free by the date of the corresponding case’s diagnosis. The date of PNES diagnosis was defined as the index date for both PNES cases and controls.

For comparison, we identified individuals with incident diagnosis of epilepsy and individuals with incident diagnosis of conversion disorder with motor symptoms or deficits during 2001–2013, both without lifetime diagnosis of PNES, and 10 controls for each of them, using the same strategy as PNES cases and controls.

All study participants were followed from the index date until death, emigration from Sweden or 31 December 2013, whichever came first.

Measures

Exposures and outcomes

The main exposure was PNES identified by incident diagnosis from NPR (ICD-10 code F44.5; other versions of ICD codes in online supplemental table S1). Two comparators—epilepsy (ICD-10 code G40) and conversion disorder with motor symptoms or deficits (ICD-10 code F44.4) were assessed from NPR. Thus, diagnostic groups in this study include PNES, epilepsy, conversion disorder with motor symptoms or deficits and matched controls corresponding to each diagnostic group.

Supplemental material

The main outcome was all-cause mortality assessed via the Swedish Cause of Death Register. Causes of death were further categorised into non-natural and natural. Natural causes include all somatic diseases (ICD-10 codes A00-R99) and were further grouped by ICD-10 chapters. Non-natural causes (ICD-10 codes S00-Y98) include suicide and other traumatic events.

Other covariates

Socioeconomic characteristics, including county of birth and educational status, were retrieved from TPR and LISA, as they are potentially associated with both PNES and non-natural and natural cause of death. Education status was defined as the highest educational level (up to secondary school, upper secondary school, undergraduate and graduate school) of an individual before the index date. Any clinical diagnosis of comorbid psychiatric disorder, including attention-deficit/hyperactivity disorder, autism, intellectual disability, conduct disorders, eating disorders, substance use disorders, depressive disorders, bipolar disorders, anxiety disorders, schizophrenia and personality disorder, was also considered. Information on comorbid psychiatric disorders was obtained from the NPR based on ICD-10 codes (online supplemental table S1).

Statistical analysis

Conditional Cox regression models were used to examine the association between PNES and all-cause mortality using time since diagnosis as the underlying time scale.18 HRs were estimated with 95% CIs in three models: In model 1, year of birth and sex were automatically adjusted for in the conditional Cox regression analyses. Model 2 additionally adjusted for socioeconomic factors including the county of birth and education status. In model 3, we further included any comorbid psychiatric disorders as covariates to examine the extent to which associations were influenced by psychiatric comorbidities. We also examined the association with death due to natural and non-natural causes. For comparison, similar analyses were performed for epilepsy and conversion disorder with motor symptoms or deficits.

The all-cause mortality risks among PNES cases were further examined by epilepsy status (PNES with and without epilepsy), and source of PNES diagnosis (PNES diagnosed via neurology or psychiatry care). To test whether the association between PNES and all-cause mortality varies with age and time since diagnosis, we fitted two Cox models allowing the HRs to be different (1) for before and after age 50 and (2) for time since PNES diagnosis <and ≥3 years. In addition, we examined to what extent comorbid psychiatric disorders influenced all-cause mortality among PNES cases, by comparing the risk of death in PNES patients with and without any comorbid psychiatric disorder.

Log-rank tests were performed to examine the difference in survival probability of several diagnostic groups (PNES cases with epilepsy, PNES cases without epilepsy, epilepsy cases without PNES) and matched controls.

Data management was performed using SAS V.9.4 (SAS Institute) and all analyses were performed using R V.4.0.5.19

Results

Cohort description

A total of 885 individuals with incident diagnosis of PNES during 2001–2013 were identified and were matched to 8850 controls. Using the same strategy, we identified 50 663 individuals with epilepsy and 506 630 matched controls, as well as 1057 individuals with conversion disorder with motor symptoms or deficits and 10 570 matched controls.

The median follow-up time of PNES cases was 5.2 (IQR, 2.3–8.8) years. Demographic and clinical characteristics of PNES cases and matched controls are presented in table 1. Compared with controls, PNES cases attained lower levels of education (percentage of primary and lower secondary education: 28.8% vs 14.4%) and presented psychiatric comorbidities at higher rates (any comorbid psychiatric disorder: 76.3% vs 12.2%).

Table 1

Characteristics of PNES cases diagnosed between 2001 and 2013 and matched controls in Sweden

Overall mortality

By the end of follow-up, there were 32 (3.6%) deaths among individuals with PNES, compared with 46 (0.5%) deaths in controls (table 2), which gives a HR of 7.5 (95% CI 4.7 to 11.9; model 1). The association remained statistically significant after adjustment for socioeconomic factors (HR 8.3, 95% CI 4.7 to 14.8; model 2) and psychiatric comorbidities (HR 5.5, 95% CI 2.8 to 10.8; model 3). The risk of all-cause mortality was even higher in individuals with comorbid epilepsy (HR 17.0, 95% CI 4.1 to 70.4). The association was similar for PNES diagnosed by neurologists (HR 7.5, 95% CI 1.3 to 44.2) and psychiatrists (HR 8.4, 95% CI 0.7 to 101.8). Among PNES cases, individuals with any comorbid psychiatric disorder had a higher risk of all-cause mortality compared with individuals without psychiatric comorbidities (HR 3.0, 95% CI 1.0 to 8.9).

Table 2

Association between PNES, epilepsy and conversion disorder with motor symptoms or deficits, with all-cause mortality

The association of epilepsy with all-cause mortality was also statistically significant. Individuals with epilepsy had a 6.7 times higher risk of death compared with individuals without epilepsy (95% CI 6.4 to 7.0). The association between conversion disorder with motor symptoms or deficits and all-cause mortality was weaker compared with PNES, and became non-significant after adjusting for psychiatric comorbidities (HR 1.4, 95% CI 0.9 to 2.3; table 2).

The survival probability of all diagnostic groups was significantly lower than controls (figure 1). In addition, log-rank test indicates lower survival probability among patients with epilepsy without PNES compared with patients with PNES without epilepsy (p<0.01). The differences in survival probability comparing patients with PNES with epilepsy to patients with PNES without epilepsy (p=0.27), and comparing patients with PNES with epilepsy to patients with epilepsy without PNES (p=0.38), were not significant.

Figure 1

Kaplan-Meier curves of survival probability by diagnostic groups. Epilepsy cases without PNES are matched to PNES cases on sex and birth year, with a PNES-to-epilepsy 1:10 ratio. Log-rank test indicates the survival probability of all diagnostic groups are statistically lower than controls. The difference of survival function comparing epilepsy cases without PNES to PNES cases without epilepsy was statistically significant (p<0.01), while non-significant differences were found comparing PNES cases with epilepsy to PNES cases without epilepsy (p=0.27), as well as comparing PNES cases with epilepsy to epilepsy cases without PNES (p=0.38). PNES, psychogenic non-epileptic seizures.

Mortality by natural and non-natural causes

Individuals with PNES were at higher risk of both natural (HR 8.1, 95% CI 4.0 to 16.4; model 2) and non-natural causes of death (HR 15.3, 95% CI 3.0 to 78.6; model 2) after adjusting for demographic and socio-economic factors (table 3). After further adjustment for psychiatric comorbidity, the strength of association with natural death decreased moderately but retained significance (HR 6.8, 95% CI 2.8 to 16.3; model 3), while the strength of association with non-natural death became non-significant (HR 5.0, 95% CI 0.8 to 31.2; model 3). Individuals with epilepsy were also at significantly higher risk of both natural and non-natural causes of death, and the association was stronger with natural (HR 7.5 95% CI 7.2 to 7.9) compared with non-natural death (HR 3.8, 95% CI 3.3 to 4.3). The association between conversion disorder with motor symptoms or deficits and natural death remained significant when psychiatric comorbidities were considered (HR 2.5, 95% CI 1.5 to 4.2 in model 2; HR 1.7, 95% CI 1.0 to 3.1 in model 3), while the association with non-natural death became non-significant (HR 3.0, 95% CI 1.2 to 7.7 in model 2; HR 1.0, 95% CI 0.3 to 3.0 in model 3). Specific causes of death, grouped by ICD chapters among individuals with PNES, epilepsy and conversion disorder with motor symptoms or deficits are shown in online supplemental table S2.

Table 3

Association between PNES, epilepsy and conversion disorder with motor symptoms or deficits, with all-cause mortality, by natural and non-natural causes of death

Association by age and time since diagnosis

The association between PNES diagnosis and all-cause mortality varied with age and time since diagnosis (table 4). The association was stronger for age less than 50 years (HR 8.5, 95% CI 3.5 to 20.7) compared with age ≥50 years (HR 4.0, 95% CI 1.6 to 9.8). We also noted stronger association less than 3 years following PNES diagnosis (HR 18.6, 95% CI 6.5 to 53.6) compared with ≥3 years (HR 4.2, 95% CI 2.0 to 9.0).

Table 4

Association between PNES and all-cause mortality by age and time since diagnosis

Discussion

In this nationwide retrospective cohort study, patients with PNES experienced a sevenfold higher risk of death than the general population. The risk was only partly attributable to socioeconomic factors and to psychiatric comorbidities. Although all-cause mortality in people with PNES was fairly similar to the mortality in people with epilepsy, the risk more than doubled in individuals with comorbid epilepsy. The association between conversion disorder with motor symptoms or deficits and all-cause mortality was weaker compared with PNES, and became non-significant after adjusting for psychiatric comorbidities, suggesting that the higher mortality risk in people with PNES is disease-specific among conversion disorders. Like epilepsy, PNES carried a higher than expected risk of both natural and non-natural causes of death. There was an inverse association between mortality, age and duration of follow-up after diagnosis.

Our findings on excess mortality in individuals with PNES (HR=5.5) partly confirm the results of other reports. In a study of 260 patients seen in a Scottish PNES clinic, death certificate of 17 were examined.8 The authors found a modest increase in mortality among PNES patients compared with the general population. In keeping with our results, some excess mortality was attributed to poor socioeconomic conditions.

In a study using data from the Danish NPR, 1057 individuals with a diagnosis of PNES had higher mortality risk than matched controls (HR: 3.21).9 The study found that patients with PNES showed increased psychiatric comorbidities (HR: 15.45).9 Although that study is population-based and their design are similar to ours, epilepsy was only diagnosed in 8% of patients with PNES—a significantly lower proportion when compared with a pervious study (20.6%)10 and ours (35.1%). As we do not know the proportion of our patients who received the diagnosis of PNES based on VEM, PNES might have been misdiagnosed in some of our cases with both diagnoses.

In a recent retrospective cohort study including 5508 patients evaluated in the VEM units of three tertiary hospitals in Melbourne, Australia, mortality was determined in patients diagnosed with PNES, epilepsy or both conditions by linkage to the Australian National Death Index.10 The 849 patients diagnosed with PNES were found to have an increased mortality risk (standardised mortality ratio of 2.5) compared with the general population. External causes of death were found to contribute 18% of death cases.10 Those proportions are significantly lower than ours (31.3%), (online supplemental table S2). We also found suicide as the leading cause of death in PNES patients (18.8%), (online supplemental table S2) which might be attributable to PNES’s psychological origin such as traumatic experiences and developmental factors.20 The high proportion of deaths due to suicide is also observed in other conversion disorders (26.5% in conversion disorders with motor symptoms and deficits). This is in line with the few studies investigating suicide in individuals with conversion disorder, where the rate of suicide attempts in patients with conversion was found to be high.21 Individuals with suicide attempts or death have a high rate of risky alcohol use, hospitalisations and history of abuse.21 22 The latter is known to be very high in patients with PNES11 23 and this correlation has been found higher in PNES patients than in patients with epilepsy and in other conversion disorders in the Swedish population.11 This indicates that suicide-preventive interventions are not only beneficial for management of epilepsy,24 but could be potentially beneficial for the management of PNES and other conversion diorders.5 Further studies are needed to test if these findings replicate to other settings and investigate the underlying mechanisms.

The authors of the Australian study reported that mortality in patients with PNES (8.2%) did not differ from mortality in patients with epilepsy (9.4%) and in those with both clinical conditions (8.0%).10 Although these differences must be interpreted with caution, due to the small numbers, the findings from the two studies might also reflect the different sources investigated. The mortality of epilepsy in the Australian study might be underestimated because patients with seizures caused by severe medical conditions (eg, stroke, dementia, traumatic brain injury) are less likely to be seen in tertiary epilepsy centres unless the severity of epilepsy requires the intervention of experienced epileptologists. As the cause of death is not available for some patients in that study, selection bias cannot be also excluded.

In addition to the higher than expected mortality in people with PNES independently from psychiatric comorbidities, the coexistence of PNES with epilepsy in moderate proportion of patients is worth noting. The particularly high mortality among patients with the combined diagnoses is of special concern. It is possible that having two conditions that require completely different managements, both acutely and chronically, may lead to suboptimal treatment, which could result in augmentation of mortality.

Strengths and limitations

The study has several strengths. The major strengths are the population base and the large sample size. The population base ensures the representativeness of the study population. In support of this, the incidence of PNES, epilepsy, and conversion disorder with motor symptoms or deficits during the study period was estimated to be 1.2, 68.0 and 1.4/100 000 person-years, and data on PNES and epilepsy are in line with previous studies from Nordic countries.25–27 The large sample size allows us to obtain separate estimates across age groups, years since diagnosis, as well as natural and unnatural causes of death. Our results should also be interpreted in light of several limitations. The main limitation is that cases were identified using register data as we do not have access to detailed documentation of neurological examination, VEM or other neuroimaging to support the diagnosis of PNES, epilepsy or conversion disorder with motor symptoms or deficits. The high degree of comorbidity and symptom overlap among these disorders may make the register-based diagnoses susceptible to misclassification. Although diagnoses from the NPR have shown good to excellent validity for a range of conditions,14 28 including epilepsy (positive predictive value of 90%),29 the diagnosis of PNES and conversion disorder with motor symptoms or deficits has not been formally validated before. Thus, the incident diagnosis of PNES on which this study is based relies principally on the compound impression of the clinician in charge at the time of the initial evaluation and who filed the principal diagnosis on the national registry. In most cases this was likely based more on clinical judgement than on strictly objective data and does not take into consideration refinements or changes that may have taken place later on. It must be noted that a diagnosis based on expert opinion is more readily accepted in areas where VEM facilities are limited, like Sweden. On the other hand, such limitation serves as a stimulus for the clinician to be particularly considerate when formulating the diagnosis. We recognise that all these deviations from the gold-standard diagnosis represent obstacles that preclude definitive conclusions in this particular patient sample. Nonetheless, despite the contextual uncertainty, we believe that the incident diagnosis of PNES entered in the NPR represented the least controversial cases. It is standard of practice that clinicians, when the diagnosis is in doubt, lean towards the diagnosis of epilepsy rather than PNES so that false positive diagnoses of epilepsy are more frequent than vice versa. Furthermore, the cases with epilepsy who are most likely to be misdiagnosed as PNES are those with frontal lobe hyper motor seizures who are quite rare. It is reassuring that even considering the potential margin of error, our data aligned with those of the Australian study, based exclusively on patients screened by VEM.10 Second, we cannot rule out misdiagnosing of PNES as conversion disorders with motor features; if those with more severe PNES have higher mortality risk than those with less severe PNES, and the latter group are more likely to be misdiagnosed as conversion disorders with motor features, then the risk of mortality associated with PNES would have been inflated. Thirdly, we were unable to explore those diagnosed before 2001 as the Swedish NPR only has included outpatient data since 2001, yet the risk profiles of individuals with diagnosis made before 2001 may be different from those with diagnosis made after 2001. Finally, this study only included individuals with incident diagnosis below 60 years of age; thus, the results are not necessarily generalisable to older patients with PNES. However, older subjects are more likely to have severe physical health problems, a possible cause of premature mortality, making it difficult to assess the death risk attributable to PNES.

Conclusions

This study demonstrates that the diagnosis of PNES, irrespective of how established and by whom, is associated with higher mortality risk in comparison to matched controls, similar to what is observed among patients with epilepsy, and that such risk is highest in those affected by both conditions. Of particular concern is that suicide ranks high among the causes of death in patients with PNES and other conversion disorders. This indicates that suicide-preventive interventions are not only beneficial for the management of epilepsy but could be potentially beneficial for the management of PNES and other conversion disorders. Further studies are needed to test if these findings replicate in other settings and to investigate the underlying mechanisms that lead to such findings.

Data availability statement

Data sharing not applicable as no datasets generated and/or analysed for this study. The data used in this study are national register information made under ethical permission. The authors had no special privileges in accessing the data. Dissemination of personal information is regulated by the Swedish Secrecy Act. In accordance with Swedish law, researchers seeking access to individual-level data must apply for permission through a Research Ethics Board (etikprovningsmyndigheten.se) and from the National Board of Health and Welfare (socialstyrelsen.se/en/statistics-and-data/statistics/).

Ethics statements

Patient consent for publication

Ethics approval

The study was approved by the Regional Ethics Review Board in Stockholm, Sweden. According to the Swedish law, informed consent is not required because this is a register-based study using anonymised data.

Acknowledgments

ZC is supported by the Swedish Research Council (2018–02213).

References

Footnotes

  • Contributors ZC and LZ linked the data from the registries, performed statistical analyses and interpreted the data; EB helped with the discussion of the study findings and prepared the final version of the manuscript; MB performed the review of the literature, interpreted the data; LZ and MB wrote the first draft; TT conceived the project and critically appraised the data. GE made an intellectual contribution and helped with the refinement of the manuscript. ZC acts as guarantor.

  • Funding This study was funded by Swedish Research Council (2018-02213).

  • Competing interests EB reports grants from Italian Ministry of Health, grants from Swedish Orphan Biovitrum, personal fees from Arvelle Therapeutics, grants from American ALS Association, outside the submitted work. TT reports grants form EISAI, GSK, UCB, Bial, personal fees from EISAI, Sanofi, Sun Pharma, UCB, Sandoz, grants from EU, grants from Stockholm County Council, grants from CURE, outside the submitted work. The remaining authors have nothing to disclose.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.