Childhood sexual abuse predicts treatment outcome in conversion disorder/functional neurological disorder. An observational longitudinal study

Abstract Objective Explore trauma, stress, and other predictive factors for treatment outcome in conversion disorder/functional neurological disorder (CD/FND). Methods Prospective observational design. Clinical cohort study among consecutive outpatients with DSM‐IV CD/FND in a specialized mental health institution for somatic symptom disorders and related disorders (SSRD), presented between 1 February 2010 and 31 December 2017. Patient files were assessed for early childhood trauma, childhood sexual abuse, current stress, and other predictive factors. Patient‐related routine outcome monitoring (PROM) data were evaluated for treatment outcome at physical (Patient Health Questionnaire [PHQ15], Physical Symptoms Questionnaire [PSQ]) level as primary outcome, and depression (Patient Health Questionnaire [PHQ9]), anxiety (General Anxiety Disorder [GAD7]), general functioning (Short Form 36 Health Survey [SF36]), and pain (Brief Pain Inventory [BPI]) as secondary outcome. Results A total of 64 outpatients were included in the study. 70.3% of the sample reported childhood trauma and 64.1% a recent life event. Mean scores of patients proceeding to treatment improved. Sexual abuse in childhood (F(1, 28) = 30.068, β = 0.608 p < .001) was significantly associated with worse physical (PHQ15, PSQ) treatment outcome. 42.2% reported comorbid depression, and this was significantly associated with worse concomitant depressive (PHQ9) (F[1, 39] = 11.526, β = 0.478, p = .002) and anxiety (GAD7) (F[1,34] = 7.950, β = 0.435, p = .008) outcome. Conclusion Childhood sexual abuse is significantly associated with poor treatment outcome in CD/FND. Randomized clinical trials evaluating treatment models addressing childhood sexual abuse in CD are needed.

Family members also feel a burden (Griffith, Polles, & Griffith, 1998) as patients with CD/FND often need long-time intensive help, can become dependent on a wheelchair, and may need adaptations of their house and living arrangements. As treatment may have relatively little to offer, family members often provide care. Patients with CD/FND visit neurological wards for diagnostic procedures (Stone & Vermeulen, 2016;Vermeulen & Willems, 2015), often with unsatisfactory results (Régny & Cathébras, 2016;Stone & Vermeulen, 2016;Sveinson, Stafánsson, & Hjaltason, 2009;Van der Feltz-Cornelis, 2015), and there is a high return rate of patients (Merkler et al., 2016). A 3-year follow-up study of 42 patients with CD/FND showed persistence in abnormal movements in more than 90% of the patients (Feinstein, Stergiopoulos, Fine, & Lang, 2001). Thus, the individual and societal burden of CD/FND is high (Mace & Trimble, 1996).
The work on unraveling the pathogenesis of CD/FND, and the role of recent and early stress as well as psychological and biological vulnerabilities, is ongoing, as is indicated in a recent review (Keynejad, Kanaan, Pariante, Reuber, & Nicholson, 2018). Crosssectional studies exploring cortisol and the stress response in CD/ FND found conflicting results. Apazoglou found a baseline HPA axis, and sympathetic hyperarousal state in motor CD/FND was related to life adversities. During a social stress test, dissociation between stress perceived as such by patients with CD/FND, and their biological stress markers, was observed (Apazoglou, Wegrzyk, Frasca Polara, & Aybek, 2017). However, Maurer, LaFaver, Ameli, Toledo, and Hallett (2015) found that current stress levels were not altered in patients with functional movement disorders and suggested that the insistence on heightened stress levels in these patients is unjustified. In an exploration of immune function in CD/ FND, Tilyeki et al found temporarily decreased serum TNF-α levels in the acute phase of CD/FND, suggesting that stress associated with CD/FND might suppress immune function in the acute phase of CD/FND (Tiyekli, Çalıyurt, & Tiyekli, 2013). An association between stress-related neuroplasticity, CD/FND, and reduced insular volume was identified in an MRI study in patients with FND and childhood adversity (Perez et al., 2017). Childhood physical or sexual abuse has been found in 44% of patients with CD/FND (Roelofs, Spinhoven, Sandijck, Moene, & Hoogduin, 2005) and is associated with higher symptom loads in patients with nonepileptic seizures (Selkirk, Duncan, Oto, & Pelosi, 2008). In a case series study, more than 50% reported a history of exposure to physical violence and 25% reported a history of sexual assault in childhood (Régny & Cathébras, 2016). However, a study exploring the role of trauma and stress in the development of CD/FND could not confirm that these events were always present (Ludwig et al., 2018), and the DSM-5 classification for CD/FND no longer requires the presence of a stressor (Association AP, 2013).

Significant outcomes
• Childhood trauma occurs in 70.3% of patients with conversion disorder/FND.
• Childhood sexual abuse is significantly associated with poor treatment outcome for physical symptoms in patients with conversion disorder/FND.
• Randomized clinical trials evaluating treatment models addressing childhood sexual abuse in conversion disorder/FND are needed.

Limitations
• The findings of this study are generalizable to chronic complex conversion disorder/FND treated in the specialty mental health setting, not to incident cases.
• Although in comparison with other studies in this field, the sample size of this study can be considered large, it is relatively small compared with other clinical epidemiological cohort studies.
However, none of these studies, including the study of Ludwig et al. (2018), explored the effect of such factors on treatment outcome in CD/FND.

| OBJEC TIVE S
1. Describe demographic and clinical characteristics of patients presenting themselves with CD/FND in a clinical centre of excellence for somatic symptom disorders and related disorders (SSRD) in the specialty mental health setting.
2. Explore the association between predictive factors and treatment outcome in terms of physical symptoms as the primary outcome.
3. Explore the same association with secondary outcomes; depression, anxiety, general functioning, and pain.

| Design
Prospective longitudinal observational study in a clinical cohort.
This study is reported following the STROBE statement (Group S, 2007).

| Participants and setting
Consecutive outpatients presenting themselves at the clinical centre of excellence for body mind and health (CLGG), a tertiary specialty mental health setting, with CD/FND between 1 February 2010 and 31 December 2017.

| Eligibility criteria
Patients aged 18 years or older, referred to CLGG with CD/FND after neurological assessment and without clear evidence of an underlying somatic condition explaining their symptoms, were eligible. In each case, CD/FND was established by psychiatric examination (PSE), taking all information from the intake into account and according to DSM-IV criteria (Association AP, 2001) or DSM-5 criteria (Association AP, 2013), and if needed, including MINI interview.. Eligible patients were identified in the data warehouse of the specialized mental health institution (SMHI). Patient files of eligible patients were checked for consent. Patients were excluded if they did not complete any PROM questionnaires during intake and during treatment; and in case of IQ < 80; or substance dependency.

| Variables
This study explores type of CD/FND, comorbid somatic and mental disorders, early childhood trauma and childhood sexual abuse, recent life events, duration of symptoms and of earlier treatment before referral to the clinic, psychosocial factors, and family history as possible factors affecting treatment outcome.

| Data sources
Patient files were assessed based on a checklist of potential predictive factors. Data were taken from the files according to a checklist that was put together beforehand, based on a review of the literature for possibly relevant predictive factors. A search was performed in PubMed with the MESH terms 'Conversion Disorder' and 'Prognosis'. This yielded 410 hits, of which, apart from the articles as described in the introduction, three more articles were identified to be of relevance (Jalilianhasanpour et al., 2018;Krishknakumar, Sumesh, & Mathews, 2006;Plioplys et al., 2014;Roelofs et al., 2005). This resulted in the checklist laid down in Table 1.
The standard intake procedure at the CLGG consists of questionnaire assessment during intake (referred to as baseline measurement), medical history assessment, physical assessment including neurological examination, psychiatric evaluation, and psychodiagnostic assessment. Throughout treatment, patient's progress was evaluated using a digital PROM (Van der Feltz-Cornelis et al., 2014).
For this study, we used PROM data with regard to physical symptoms, depression and anxiety scores, and pain and general functioning scores. Patient file: intake or registration form Or Psychodiagnostic examination: INTERMED (Huyse et al., 1999) or LCU (Holmes & Rahe, 1967) Death of a loved one shortly before the onset of conversion disorder (<6 months)

| Data sources/measurements
The variables used and the data sources are shown in the table

| Treatment
Treatment at CLGG was standardized and multimodal, following  (Black, 1996) and (b)

| Bias
Bias was avoided by checking all patient files systematically for the predictive factors according to the checklist which was derived from a review of the literature. In a pilot, the data feeding the checklist were extracted in duplicate (RH and JvE), and the feasibility of this approach was indicated after assessment in 14 files. The data extraction was therefore continued based on this checklist.

| Study size
Study size was determined by the number of patients with CD/FND (N = 64). Based on this number, we estimated that it would be possible to establish an association between treatment outcome and 6 predictive factors with sufficient power (Nunnaly & Bernstein, 1994).

| Statistical methods
Frequencies were explored to establish prevalence of the prognostic factors in the sample. Linear regression analyses using enter method were performed to explore associations of predictors with treatment outcome in CD/FND. Analyses were performed using listwise deletion to address cases of missing data. For the secondary outcomes, hierarchical multiple regression analyses were performed.

| Sensitivity analyses
One sensitivity analysis explored the associations while controlling for baseline scores of secondary outcomes. Another sensitivity analysis explored if patients following up the intake with treatment at the CLGG differed significantly from patients who did not proceed with treatment in terms of baseline characteristics.  Table 3 shows the mean scores for the primary outcomes (PSQ) and  In Table 4, differences in outcomes are shown between patients who experienced childhood sexual abuse and those who did not.

| Treatment outcome
Independent samples t tests show that at the end of treatment those who suffered sexual abuse in childhood reported significantly higher scores on the PSQ (physical symptoms) and on the GAD7 (anxiety) than those who did not report sexual abuse in childhood.

| Predictors of treatment outcome
Significant associations between predictors and primary outcomes and between predictors and secondary outcomes are shown in Table 5.
Another sensitivity analysis explored if patients following up the intake with treatment at the CLGG differed significantly from patients who did not proceed with treatment in terms of baseline characteristics. The only significant difference was found for anxiety symptoms, with mean baseline GAD7 score of 6.56 (SD = 5.45) in the no-treatment group versus 11.05 (SD = 5.46) in the treatment group (X 2 = 6.396 (1), p = .011). with chronic CD/FND referred to a tertiary care specialized mental

| Summary of principal findings
Health Institution for SSRD.
The percentage of patients reporting at least one ACE in this study, eighty percent, is higher than the 24%-50% reported in earlier research in CD/FND (Régny & Cathébras, 2016;Roelofs et al., 2005;Selkirk et al., 2008) and higher than the 64% reported in the original ACE field study among 17,000 people visiting a medical evaluation center in the USA (Centers for Disease Control & Prevention, 2016). This is also higher than the 77.2% of patients reporting at least one ACE in a study in outpatients with depressive or anxiety disorders in another department of the same specialty mental health institution, conducted by this research group (Feltz-Cornelis et al., 2019). The percentage of patients reporting more than one ACE (73%), the mean ACE score of more than 4, and the percentage reporting a score of 4 or more (43%) are higher as well. An ACE score of 4 or more has been found to be associated with depressive disorders, suicide attempts, and alcohol abuse (Anda & Felitti, 1998;Dube, Felitti, Edwards, & Croft, 2002;Edwards, Felitti, & Anda, 2003). Therefore, based on the findings of this study, indications are that CD/FND can join this list of mental disorders associated with high levels of adverse childhood experiences.
Recent life events are highly prevalent as well; however, although the burden of general ACE and recent stressful events in this study is high, they are not associated with treatment outcomes. The primary and several secondary treatment outcomes improved significantly after treatment in the whole group and the only predictor significantly associated with worse physical treatment outcome was childhood sexual abuse. A separate analysis shows that not only worse physical outcomes but also higher anxiety levels at end of treatment in those who suffered sexual abuse in childhood.
Comorbid depressive disorder has a significant negative association with secondary treatment outcome in terms of depressive Note: Missing data (n) for the 43 patients who completed at least one follow-up PROM was PSQ = 1 (2.3%); PHQ15 = 1 (2.3%);GAD7 = 2 (4.7%); SF36 = 9 (20.9%) at baseline; and PSQ = 13 (30.2%); PHQ15 = 15(34.9%); PHQ9 = 2 (4.7%); GAD7 = 7 (16.3%); SF36 = 14 (32.6%); and BPI = 5 (11.6%) at follow-up. Pain seems to be a physical symptom of particular importance in patients with CD/FND as clinically significant levels of pain occur in 87.1% of the sample, and pain requiring medication occurs in more than 55% of the sample. Moreover, the mean BPI score is over 5, which is higher than the mean score reported in a study on cancer patients with bone metastases (Zeng et al., 2011). Stone and Sharpe find a high prevalence of pain in functional weakness (Stone, Warlow, & Sharpe, 2010), and possible interpretations of the association between chronic regional pain syndrome and CD/ FND have been discussed (Popkirov, Hoeritzauer, Colvin, Carson, & Stone, 2019). However, so far, pain has not received much attention in CD/FND research, classification, and guidelines. In this study, pain did not improve in the group as a whole, but did improve significantly in case of comorbid developmental disorders such as adult ADHD or autism spectrum disorder. This influence on treatment outcomes disappeared after correction for baseline pain levels, indicating that patients with adult ADHD or autism spectrum disorder had higher pain levels at intake, and benefitted from the treatment for pain outcomes. This patient group benefitted relatively better from the treatment intervention, at least for the pain component of their condition, than the group as a whole.
The sensitivity analysis comparing the patients that continued to treatment with the patients that did not showed a significant difference in anxiety, with level of anxiety amounting to the level of anxiety disorder (GAD score 10 or more) in the treatment group versus far below this threshold in the no-treatment group. From a clinical point of view, this makes sense as it can be interpreted as those with higher levels of anxiety have a higher motivation for treatment and could also reflect levels of health anxiety related to the symptoms for which treatment was sought.

| Strengths and weaknesses of the study
A strength of this study is the longitudinal, prognostic design, and the exploration of predictors of treatment outcome, including childhood sexual abuse and psychiatric comorbidity, which is a first in the literature. This way we could establish for the first time that sexual abuse in childhood is associated with worse physical treatment outcome; that comorbid depression has a negative impact on concomitant depressive and anxiety symptoms in CD/FND; and that patients with comorbid CD/FND, pain, and comorbid developmental disorder benefit more from treatment. The fact that this study is conducted in a specialty mental health setting can also be seen as a strength of the study, as so far, studies have been performed in neurology clinics that in general see first presentations, whereas here we have the opportunity to explore a cohort with chronic CD/FND. Another strength of the study is that we were able to perform sensitivity analyses showing our findings to be robust and indicating possibly higher health anxiety levels in the group proceeding to treatment at CLGG.
Limitations of the study are that due to the study setting and patient population, the findings of this study are generalizable to chronic complex CD/FND treated in the specialty mental health setting, but not to patients with CD/FND that present themselves for the first time in neurology clinic. Another limitation is that although in comparison with other studies in this field, the sample size of this study can be considered large, it is relatively small compared to other clinical epidemiological cohort studies.

| Implications
This study shows that treatment involving a parallel track combina- TA B L E 5 Linear regression analyses showing significant associations between predictors and primary (PSQ) and secondary outcomes psychological treatment outcomes. As this was an observational study, and in view of the lack of effective evidence-based treatments for this condition, a randomized clinical trial exploring treatment effect of the described intervention is warranted.
The finding that childhood sexual abuse is associated with significantly worse treatment outcome warrants the development and evaluation of treatment interventions specifically targeting childhood sexual abuse in CD/FND alongside regular treatment.
Furthermore, the high prevalence of pain in this study, and the finding that pain did not improve in the group as a whole, warrants development of treatment models focusing at pain in CD/FND. The susceptibility of pain to improvement in patients with comorbid adult developmental disorders warrants development and evaluation of medical and psychotherapeutic treatment models specifically targeting pain in CD/FND with comorbid adult ADHD/autism spectrum disorder, as this may be a subgroup of patients with potential for more benefit of treatment.

ACK N OWLED G M ENTS
Mrs. Dawn Golder, Executive Director FND Hope UK, provided comments from a patient and public involvement perspective. Mr.
Robin Heijnis developed the checklist and wrote a report on part of the patients. Dr. Iman Elfeddali contributed to early-stage data analysis. Mrs. Dilana Ozgül updated and checked the dataset. This study was supported by GGz Breburg, Tilburg, the Netherlands.

AUTH O R CO NTR I B UTI O N S
CFC designed the study, facilitated the conduct of the study, oversaw the analysis, and wrote the paper. SFA codesigned the study, designed the analysis, contributed to interpretation of the analysis, and coauthored the paper. JvEvdS Codesigned the study, oversaw data extraction, contributed to interpretation of the analysis, and coauthored the paper. All authors approved the final version of the manuscript.

E TH I C A L A PPROVA L
This study follows the principles of the World Medical Association's Declaration of Helsinki. No informed consent was required, as for the present research we used data that were collected for administrative purposes and monitoring of treatment outcome by treatment providers. According to Dutch law, in accordance with the Helsinki Declaration, and according to the Dutch Central Medical Ethical Committee, no explicit informed consent is required for the use of clinical or administrative data, collected in the context of treatment provision. At intake, patients were informed that patient-reported outcome measures (PROM) and medical data obtained during intake and treatment could be used for research evaluation on an anonymous basis, unless they indicated their dissent. In such cases, this was notified in the patient file.
Patient files of dissenting patients were excluded from the study.
Data were coded in order to create an anonymous dataset. The research protocol was approved by the IRB scientific committee of GGz Breburg (2017-03).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data are owned by a third party, GGz Breburg, that does not publicly share data. However, interested parties will be able to obtain data upon request as follows. Researchers can submit a research plan, which describes the background and methods of a proposed research question, and a request for specific data of the database used for this study to answer the research question.