Dissociative disorders are common comorbidities in FND

Fewer studies than expected commented on comorbid dissociative disorders in FND samples. In the nine studies that did, the proportion of participants with FND found to have comorbid dissociative disorder ranged considerably, from 8% to 80%.^{Reference Yayla, Bakım, Tankaya, Ozer, Karamustafalioglu and Ertekin14,Reference Litwin and Cardeña41–Reference Tezcan, Atmaca, Kuloglu, Gecici, Buyukbayram and Tutkun48} The most common dissociative disorder diagnosed was dissociative disorder not otherwise specified (DDNOS) – a subtype previously highlighted as being the most prevalent in epidemiological studies.^{Reference Şar, Akyüz and Doğan121,Reference Spitzer, Liss, Dudeck, Orlob, Gillner and Hamm122} When compared with prevalence rates reported for the general population, dissociative disorder rates are higher in FND groups. A review of dissociative disorder prevalence rates in community and clinical samples suggested an overall lifetime prevalence of 10%.^{Reference Sar123} Similarly, a recent meta-analysis of college student populations (n = 31 905) reported an overall prevalence of 11%.^{Reference Kate, Hopwood and Jamieson124} Prevalence of dissociative disorder might be higher in clinical populations; for example, Foote and colleagues observed an overall prevalence of 29% in 82 consecutive out-patient psychiatric clinic attendees, and Ross et al observed considerably higher lifetime prevalence in in-patients (28–44.5%).^{Reference Foote, Smolin, Kaplan, Legatt and Lipschitz125,Reference Ross, Duffy and Ellason126} A separate study found that general psychiatric in-patients were more likely to have a comorbid dissociative disorder if they scored above 30 on the DES, with patients in this bracket having a prevalence of up to 80%.^{Reference Tutkun, Sar, Yargiç, Ozpulat, Yanik and Kiziltan127} The majority of studies reviewed here reported prevalence rates within the clinical range observed by Ross et al, and in some cases considerably higher. However, two of the included studies^{Reference Marchetti, Kurcgant, Neto, Von Bismark and Fiore42,Reference Moene, Spinhoven, Hoogduin, Sandyck and Roelofs45} had prevalence rates of 10% or lower, commensurate with the prevalence rate of dissociative disorders in the general population.^{Reference Sar123,Reference Kate, Hopwood and Jamieson124}

The prevalence of dissociative disorders seen in the samples with FND presented in this review supports the notion that FND and dissociative disorders might share similar mechanisms and aetiology (e.g. traumatic experiences, chronic/severe stressors), in addition to symptoms, supporting the classification of FND as a dissociative disorder in the ICD.^{2,Reference Brown, Cardeña, Nijenhuis, Sar and van der Hart4} Given that dissociative disorders are prevalent in other psychiatric populations,^{Reference Foote, Smolin, Kaplan, Legatt and Lipschitz125,Reference Ross, Duffy and Ellason126} dissociation as an explanatory mechanism for other mental disorders merits exploration. For FND, this proximity invites questions about treatment options and whether some people with FND would benefit from treatments that target dissociative disorders and symptoms, including those currently being evaluated in FND such as eye movement desensitisation and reprocessing (EMDR)^{Reference Cope128–Reference Cope, Mountford, Smith and Agrawal131} and dynamic psychotherapies.^{Reference Malda Castillo, Beton, Coman, Howell, Burness and Martlew132,Reference Russell, Abbass, Allder, Kisely, Pohlmann-Eden and Town133}

Dissociative symptoms are elevated in FND and vary by subgroup

Dissociative symptoms, as measured through validated scales, ranged widely in the FND samples. However, most studies reported elevated dissociative symptom scores in FND samples that were above clinical cut-off scores and/or the scores reported for comparison groups. Elevated scores were observed for both psychoform and somatoform manifestations of dissociation.

The DES was the most used measure of psychoform dissociation. DES scores have recently been meta-analysed in psychiatric populations.^{Reference Lyssenko, Schmahl, Bockhacker, Vonderlin, Bohus and Kleindienst3} In that meta-analysis, the mean DES score for FND was reported at 25.6, based on 20 studies, and dissociative disorders had a mean score of 38.9;^{Reference Lyssenko, Schmahl, Bockhacker, Vonderlin, Bohus and Kleindienst3} 11 studies from our review reported mean or median scores higher than this, and 4 reported scores above the mean score for dissociative disorders. We found 11 studies that used other psychoform dissociation scales (e.g. CADSS, MDI, CDS) – the majority of FND groups scored higher on these dissociative measures when compared with controls.

In our subgroup meta-analysis investigating psychoform dissociation in FND groups versus different control groups, a large effect towards higher dissociation scores was found in the FND groups compared with healthy controls. This effect was also observed in the comparison with neurological controls, but of a medium effect size. High heterogeneity prevents us from confirming an effect between FND groups and psychiatric controls. The last finding may have been affected by the amalgamation of multiple psychiatric populations into one group. Four out of eight studies included a mixed psychiatric control group, while the remaining four involved somatisation, eating disorder, chronic pain and depression.

The SDQ-20 was the next most used dissociative symptom scale, and the only ‘pure’ measure of somatoform dissociation. Mean SDQ-20 scores in samples with dissociative disorder have been reviewed previously, observing mean scores >30 for FND samples, >43 for DDNOS and DPDR samples and >50 for DID samples.^{Reference Nijenhuis134} Using these as a reference, FND samples (of a total of 25) endorsed mean or median SDQ-20 scores >30, supporting the view that compartmentalisation is a significant feature of FND. Our meta-analysis comparing SDQ scores in people with FND and healthy controls displayed high heterogeneity between studies (Fig. 2); however, the prediction interval shows that research in this area has generally found people with FND to exhibit higher somatoform dissociation than healthy controls. Only the comparison from Demartini et al^{Reference Demartini, Goeta, Barbieri, Ricciardi, Canevini and Turner56} involving participants with FND-seizures compared with healthy controls showed a non-significant effect. Sensitivity analyses suggested that this study was an outlier, and removing it shifted the prediction interval to greater than zero, indicating that future studies are likely to find the same pattern of results. Despite this, large heterogeneity remained (75%), meaning it is likely that the FND samples contained important subgroup populations that would need to be studied before effect size estimates could be interpreted. However, subgroup analysis was not possible as there were fewer than ten studies.

Our results suggested that there might be different dissociative symptom profiles across FND subgroups, specifically elevated psychoform dissociation in FND-seizures and greater somatoform dissociation in FND-motor samples. Guz and colleagues,^{Reference Guz, Doganay, Ozkan, Colak, Tomac and Sarisoy66} for example, reported the highest DES scores in an FND-seizures sample (mean score 27) relative to FND-motor (mean 12) and FND-sensory samples (mean 7). Nearly all of the FND-motor sample DES scores were comparable to the aforementioned mean scores of non-clinical populations, or substantially lower.^{Reference Van IJzendoorn and Schuengel135} Only in Tezcan and colleagues’ study did an FND-motor group exhibit high psychoform dissociation (mean DES = 25).^{Reference Tezcan, Atmaca, Kuloglu, Gecici, Buyukbayram and Tutkun48} Demartini and colleagues directly compared FND-seizures and FND-motor groups and noted an inverse relationship between DES and SDQ-20 scores in the two groups (FND-seizures mean DES = 17.2, mean SDQ-20 = 23.3; FND-motor mean DES = 7.9, mean SDQ-20 = 44.9).^{Reference Demartini, Goeta, Barbieri, Ricciardi, Canevini and Turner56} This observation was not totally consistent; some FND-seizures groups scored highly on the SDQ-20 (median SDQ-20 = 34 in Pick et al^{Reference Pick, Mellers and Goldstein90}; mean SDQ-20 = 38 in Koreki et al^{Reference Koreki, Garfkinel, Mula, Agrawal, Cope and Eilon76}), with no clear methodological differences that could account for this.

Significantly higher SDQ-20 scores in an FND-seizures sample relative to epilepsy controls were reported in a study in which the DES and CDS were unable to differentiate participants with FND from controls.^{Reference Lawton, Baker and Brown136} The authors argued therefore that compartmentalisation symptoms are more characteristic of FND-seizures, whereas detachment symptoms are less prominent. In our review, the highest DIS-Q score was also in an FND-seizures group.^{Reference Ozcetin, Belli, Ertem, Bahcebasi, Ataoglu and Canan85} Perhaps, therefore, these results indicate that people with the FND-seizures subtype have a greater overall tendency to dissociate in general, with respect to both somatoform and psychoform symptoms. The discrepant findings noted here require further examination in additional studies using measures of both somatoform and psychoform dissociation in FND-seizures and FND-motor samples.

A potential explanation of the observed trends is that FND motor symptoms are indeed comparable to so-called somatoform dissociative symptoms, whereas psychoform dissociative symptoms are more common in people with FND-seizures. Indeed, the SDQ-20 contains several items that map directly onto FND motor symptoms, whereas there is only one item regarding attacks resembling seizures. One of the developers of the SDQ-20 has already called for physical ‘conversion’ symptoms to be relabelled as somatoform dissociative symptoms.^{Reference Nijenhuis134}

Several limitations affect the generalisability of the observations outlined above. The majority of studies that addressed FND subtypes focused on FND-seizures; very few studies specified the FND subtype and most involved people with FND endorsing multiple symptoms. There were only three studies presenting FND-sensory or ‘mixed’ (F44.6 and F44.7 respectively) subtype specific scores^{Reference Tezcan, Atmaca, Kuloglu, Gecici, Buyukbayram and Tutkun48,Reference Guz, Doganay, Ozkan, Colak, Tomac and Sarisoy66,Reference Spitzer, Spelsberg, Grabe, Mundt and Freyberger97} and our subgroup meta-analysis indicated that FND-seizures groups exhibited different levels of psychoform dissociation from FND-motor groups. This meta-analysis indicates the need for more FND-motor and healthy control comparisons. Importantly, future studies should examine closely the effect on SDQ-20 total scores of items pertaining to the particular FND symptoms experienced by each respondent. One methodological solution is to rerun analyses excluding items that resemble FND symptoms common in the sample under investigation.

Although the DES and SDQ-20 scales were the most commonly used dissociation scales, most studies did not administer both and thus did not screen for both types of dissociative symptoms. It has been suggested that rather than existing as a general attribute, dissociation represents a diverse and not necessarily connected cluster of symptoms; consequently, scales such as the MDI, which measure specific psychoform dissociative symptoms such as depersonalisation and identity dissociation, might be more valuable for assessing dissociative symptom profiles.^{Reference Briere, Weathers and Runtz137} Reporting bias, and additionally recall bias, are also potential limitations of self-report scales, which may be mitigated by use of scales that are clinician-administered and/or state-based, such as the CADSS.

The large values for measures of dispersion seen across the studies indicate that although some individuals with FND report elevated dissociative tendencies, there are some who appear less affected by dissociation. If specific FND presentations present with different types of dissociative symptoms, as suggested by our results, then omission of measures of psychoform and/or somatoform dissociation might result in falsely low or unrepresentative results. However, another interpretation is that dissociative symptoms, although common, are not a ubiquitous feature of FND. More studies that administer both psychoform and somatoform dissociative scales to people with various FND subtypes will help to establish more concretely whether the observed trends represent the population.

Biological associations of dissociation in FND

Little is known about the biological processes underlying pathological dissociation. Roydeva & Reinders^{Reference Roydeva and Reinders15} recently systematically reviewed studies exploring potential biomarkers associated with pathological dissociation transdiagnostically, including dissociative disorders, FND, and affective, personality and stress-related disorders.^{Reference Roydeva and Reinders15} Increased neural activity was observed in several brain regions in the dissociative groups relative to healthy or clinical control groups, in both resting-state and task-based functional neuroimaging studies. This was consistent in regions of the prefrontal cortex, insula and ACC – findings also reported by Drane et al in a previous review^{Reference Drane, Fani, Hallett, Khalsa, Perez and Roberts138} and in models of dissociation in specific disorders, including PTSD and DID.^{Reference Lanius, Vermetten, Loewenstein, Brand, Schmahl and Bremner139,Reference Reinders, Willemsen, den Boer, Vos, Veltman and Loewenstein140}

Trends in connectivity alterations included increased connectivity from amygdala seed regions to prefrontal cortex (e.g. dorsolateral, medial, orbitofrontal), precuneus and superior parietal cortex. In addition, structural imaging studies found evidence of volumetric reductions in several regions in the dissociative groups, including the hippocampus, thalamus and basal ganglia. Other trends included a negative correlation between tumour necrosis factor alpha and dissociative symptoms, and a positive correlation with interleukin 6, raising questions about an interaction between inflammation and dissociative symptoms;^{Reference Roydeva and Reinders15} however, this process may be mediated by other comorbid disorders, such as depression.^{Reference Osimo, Pillinger, Rodriguez, Khandaker, Pariante and Howes141} Further research is needed to better understand the neurobiological basis of dissociative symptoms and disorders.

Our review found a noticeably smaller number of studies examining the relationships between biological factors and dissociative symptoms in people with FND when compared with the Roydeva & Reinders review.^{Reference Roydeva and Reinders15} This is likely due to differing study inclusion criteria; we specified that our studies describe dissociative symptoms or comorbid disorders with validated measures in people with FND, whereas the aforementioned review accepted diagnosis of FND alone as a marker of dissociation. Despite this disparity, there were some common trends noted: three of the identified studies from our review shared consistent findings with some of the reports of altered neural circuitry described above.^{Reference Perez, Matin, Williams, Tanev, Makris and LaFrance88,Reference van der Kruijs, Jagannathan, Bodde, Besseling, Lazeron and Vonck102,Reference van der Kruijs, Bodde, Vaessen, Lazeron, Vonck and Boon118} Volumetric and functional connectivity changes were observed within several brain regions, including the ACC – a region implicated in action planning, decision-making and empathy-related responses. The ACC is consistently highlighted as an area of interest in FND studies.^{Reference Drane, Fani, Hallett, Khalsa, Perez and Roberts138,Reference Perez, Nicholson, Asadi-Pooya, Bègue, Butler and Carson142,Reference Ospina, Jalilianhasanpour and Perez143} Increased functional connectivity and neural activity within this cortical region, among several other related regions described earlier, has been advocated as a potential biomarker for pathological dissociation and for FND.^{Reference Drane, Fani, Hallett, Khalsa, Perez and Roberts138,Reference Wegrzyk, Kebets, Richiardi, Galli, de Ville and Aybek144} Whether these demonstrable alterations in neural circuitry can be translated into an in vivo biomarker of disorder severity or response to treatment remains to be explored. Studies applying this method have shared promising results.^{Reference Faul, Knight, Espay, Depue and LaFaver145,Reference Espay, Ries, Maloney, Vannest, Neefus and Dwivedi146}

Two relevant studies from Labate and colleagues could not be included in this review as they did not meet our inclusion criteria.^{Reference Labate, Cerasa, Mula, Mumoli, Gioia and Aguglia147,Reference Labate, Martino, Caligiuri, Fortunato, Bruni and Segura-Garcia148} The first reported on cortical thinning in similar brain regions, using voxel-based morphometry and cortical thickness MRI techniques.^{Reference Labate, Cerasa, Mula, Mumoli, Gioia and Aguglia147} Significant loss of grey matter volume in participants with FND-seizures was observed in primary motor and premotor cortices, the supplementary motor area and the ACC.^{Reference Labate, Cerasa, Mula, Mumoli, Gioia and Aguglia147} Thickness alterations in these regions of interest were not associated with dissociation (on DES and SDQ scores); however, SDQ scores were significantly negatively associated with reduced cortical thickness in the left inferior frontal gyrus and the left central sulcus.^{Reference Labate, Cerasa, Mula, Mumoli, Gioia and Aguglia147} The second, more recent, study contrasted similar brain regions of interest in subgroups of healthy participants and participants with depression and FND-seizures.^{Reference Labate, Martino, Caligiuri, Fortunato, Bruni and Segura-Garcia148} Relative to healthy controls, both the depression and FND-seizures subgroups had noticeable thinning of the ACC and lateral orbitofrontal cortext on MRI. Although this was more pronounced in participants with depression, a significant positive relationship between SDQ scores and morphological changes in the right medial orbitofrontal cortext was observed.^{Reference Labate, Martino, Caligiuri, Fortunato, Bruni and Segura-Garcia148}

Other biological findings were more variable. Two studies measured skin conductance levels in people with FND, with inconsistent results.^{Reference Herrero, Tarrada, Haffen, Mignot, Sense and Schwan70,Reference Pick, Rojas-Aguiluz, Butler, Mulrenan, Nicholson and Goldstein89} These inconsistent findings are similar to those previously reported in Drane et al's review.^{Reference Drane, Fani, Hallett, Khalsa, Perez and Roberts138} One possible explanation is that in FND, interoceptive impairments lead to a discordance between subjective and autonomic responses. This is reflected further in other studies that have reported impaired interoceptive accuracy of people with FND.^{Reference Demartini, Goeta, Barbieri, Ricciardi, Canevini and Turner56,Reference Demartini, Goeta, Romito, Anselmetti, Bertelli and D'Agostino57,Reference Koreki, Garfkinel, Mula, Agrawal, Cope and Eilon76,Reference Pick, Rojas-Aguiluz, Butler, Mulrenan, Nicholson and Goldstein89} Owing to the limited number of studies found and the variety of objective measures, a meta-analysis was not possible for the biological associations of dissociation in FND.

Clinical features associated with dissociation in FND

Several studies in this review offer compelling evidence in support of using dissociative tendencies, symptoms or comorbidity as a potential prognostic marker in FND. Many of these studies pertained to the FND-seizures subtype. Severity or frequency of seizure symptoms, and presence of ongoing seizures relative to remission states, were seen in people with FND who reported greater degrees of dissociative symptoms.^{Reference Bodde, Janssen, Theuns, Vanhoutvin, Boon and Aldenkamp51,Reference Kuyk, Siffels, Bakvis and Swinkels78,Reference Martino, Bruni, Labate, Vasta, Cerasa and Borzì80,Reference O'Brien, Fortune, Dicker, O'Hanlon, Cassidy and Delanty87,Reference Reuber, House, Pukrop, Bauer and Elger93,Reference Walther, Volbers, Erdmann, Dogan Onugoren, Gollwitzer and Kasper103} Quality of life in people with FND-seizures was also reportedly affected by degree of dissociation, even after controlling for a number of other factors.^{Reference Mitchell, Ali and Cavanna81,Reference Gagny, Grenevald, El-Hage, Chrusciel, Sanchez and Schwan115} Other studies observed varying severity of presentation with level of dissociation in general FND populations. The concerning link between dissociative symptoms and suicide has been established previously in psychiatric samples.^{Reference Calati, Bensassi and Courtet26,Reference Güleç, Ýnanç, Yanartaþ, Üzer and Güleç68,Reference Kılıç, Coşkun, Bozkurt, Kaya and Zoroğlu149} More longitudinal studies could therefore be helpful to further establish the link between dissociative symptoms and FND severity or quality of life.

Several studies observed that alexithymia and emotion dysregulation were associated with dissociation severity in FND participants.^{Reference Brown, Bouska, Frow, Kirkby, Baker and Kemp53,Reference Del Río-Casanova, González-Vázquez, Justo, Andrade, Páramo and Brenlla55,Reference Kienle, Rockstroh, Bohus, Fiess, Huffziger and Steffen-Klatt75} Number of FND symptoms, earlier onset of symptoms and likelihood of symptom severity requiring in-patient management were also predicted, to a degree, by dissociative symptoms.^{Reference Yayla, Bakım, Tankaya, Ozer, Karamustafalioglu and Ertekin14,Reference Moene, Spinhoven, Hoogduin, Sandyck and Roelofs45,Reference Roelofs, Hoogduin, Keijsers, Näring, Moene and Sandijck46} Other findings included elevated general psychopathology and psychiatric comorbidity in FND participants with greater dissociative tendencies.^{Reference Yayla, Bakım, Tankaya, Ozer, Karamustafalioglu and Ertekin14,Reference van der Hoeven, Broersma, Pijnenborg, Koops, van Laar and Stone101}

Taken together, these findings have important clinical implications. Dissociation scales might provide a novel means of categorising FND severity, prognosis and guiding treatment. Reflecting this notion, some of the studies in this review adopted the SDQ-20 specifically as a marker of FND severity.^{Reference Kienle, Rockstroh, Bohus, Fiess, Huffziger and Steffen-Klatt75,Reference Steffen-Klatt, Fiess, Beckha, Schmidt and Rockstroh99} It might be beneficial to measure psychoform or somatoform dissociative symptoms as an outcome for FND in future studies. In addition, high scores on dissociation measures in people with FND also reporting previous trauma might indicate the presence of post-traumatic symptoms,^{Reference Pick, Mellers and Goldstein90} which would need to be addressed specifically in any effective clinical intervention. The associations between traumatic experiences, dissociative and post-traumatic symptoms in FND are important and understudied; although this unfortunately goes beyond the scope of this paper, it merits further exploration in future studies.

Strengths and limitations of the review

Our review has several strengths. First, we have comprehensively summarised existing data on both dissociative disorders and dissociative symptoms across different FND symptom types, and our subgroup analyses have highlighted potentially important subgroup differences. We also included studies that used a range of validated dissociation measures, ensuring that our review captured as much of the existing data as possible and providing insights into the relative prevalence of different types of dissociative symptoms (e.g. detachment, compartmentalisation) in the FND samples. Our findings have also indicated several important directions for future research on biological mechanisms and clinical implications of dissociation in FND.

We acknowledge some limitations in the methods used. Regarding the search strategy, terms for cognitive FND symptoms were not included because we designed the search strategy in 2019, before these symptoms were formally classified as FND in ICD-11. Overall, the meta-analyses indicated a general problem of large heterogeneity between studies, which limits the certainty with which some conclusions can be drawn. This is especially evident in the neurological and psychiatric control groups, which each subsumed several different clinical disorders. In addition to this, funnel plots all showed asymmetry within the data. There are cases of small studies reporting a large effect size and large standard error, which is a potential indictor of publication bias. However, this could also be reflective of the high heterogeneity and the presence of outliers. We were also unable to explore the influence of associated risk factors such as trauma on dissociation scores in the meta-analysis, which might be of value in future. Generally, there were only a small number of studies that investigated the relationship between dissociation scores and biological measures, with inconsistent findings on several measures. Finally, although we excluded any study that stated that participants with FND had comorbid neurological disorders, not all studies explicitly declared this (or that comorbid neurological disorders were excluded), and therefore there is a possibility that some of the data included may have not been FND-specific.

Future directions and clinical implications

The findings presented here indicate that dissociation is a pervasive and clinically relevant feature in FND, also lending support to the view that dissociation is a possible underlying mechanism. Further research is needed to examine more rigorously the role of dissociative processes in the generation of FND symptoms, with methods designed to examine causal hypotheses. Improved understanding of the role of dissociation in FND will have implications for diagnosis and classification, with potential to facilitate resolution of the discrepancy between the ICD-11 and DSM-5 classifications.

Despite the considerable literature reviewed here, few studies investigated the potentially differing degrees of psychoform and somatoform dissociation in different FND subtypes. Furthermore, few studies explicitly sought to examine the distinction between compartmentalisation and detachment subtypes of dissociation in FND subgroups. More studies should closely examine these different forms of dissociation, their underlying aetiological and neurobiological basis, and how they relate to the occurrence of core FND symptoms such as seizure, motor, cognitive and sensory symptoms.

From a clinical perspective, this review highlighted several possible implications. To accurately capture all of the symptomatology and experiences of patients with FND, clinicians might consider screening for both somatoform and psychoform dissociative symptoms, as they are evidently overexpressed in this population and are associated with adverse clinical features and outcomes. For individuals with FND who report significant dissociation, a dedicated clinical interview to assess possible comorbid dissociative disorders would be another consideration, as would be a more detailed screening for past trauma and possible post-traumatic symptoms.

A key research question is whether level of dissociative symptoms could constitute a prognostic indicator of FND severity. Lastly, there are some appreciable potential biomarkers of dissociation that, if explored further, could aid in understanding how this process occurs and potentially act as guides to monitoring treatment efficacy and/or developing novel treatments for the disorder.