Conversion disorder, the development of neurological symptoms in the absence of demonstrable pathology, despite a recent revival in interest remains a relatively neglected field of medicine and psychiatry. ^{Reference Bass, Peveler and House1} This neglect may be partly explained by the skepticism towards the diagnosis that followed Slater's influential paper, ^{Reference Slater and Glithero2} which concluded that, given time, most hysterical states evolve into true neurological disease. This view has been successfully debunked ^{Reference Crimlisk, Bhatia, Cope, David, Marsden and Ron3} and conversion disorder is now seen to be a relatively stable psychiatric diagnosis ^{Reference Stone, Smyth, Carson, Lewis, Prescott and Warlow4} with a frequently poor prognosis. ^{Reference Carson, Best, Postma, Stone, Warlow and Sharpe5} Despite the renewed interest there remains a paucity of studies addressing therapy. Treatments including psychoeducation, cognitive–behavioural therapy, hypnosis and long-term analytical psychotherapy have been reported to be beneficial but lack a robust evidence base. ^{Reference Ruddy and House6}

Abreaction, which involves interviewing individuals under the influence of a drug, was once commonly used for conversion disorder. The name ‘abreaction’ is rather a misnomer as it more precisely describes just one of the techniques, ‘emotional catharsis’, that can be employed during the drug interview. The physician can also suggest that symptoms will resolve during the interview (‘suggestion’); explore the psychological and social antecedents to symptom formation (‘exploration’); gradually restore motor, sensory or other body functions (‘rehabilitation’); and recordings of battle sounds have been employed for those with war-induced conversion states. Originally considered useful as both a treatment and diagnostic aid, on the one hand to discriminate between hysteria and malingering and on the other between hysteria and organic conditions, the high number of psychiatric casualties made it popular in the Second World War. ^{Reference Horsley7} Enthusiasm for its use became so widespread that Tilkin speculated, ‘the future of [abreaction] is infinite and the possibilities endless’. ^{Reference Tilkin8} In fact, its use has gradually waned: a consequence, perhaps, of Slater's misgivings.

Given the severity of the condition and lack of evidence for any particular therapy, we consider a synthesis of the research timely. A great majority of existing evidence takes the form of case series and reports. Meta-analysis of case reports and series is an unusual but accepted methodology ^{Reference Vandenbroucke9} that has proven able to generate additional knowledge. ^{Reference Trabert10} Noting the wide variety of names for the procedure (see Method), we will keep to the more theory-neutral phrase ‘drug interview’ for the remainder of the paper.

Method

Results

The search methodology identified 811 studies, 55 of which met the study inclusion criteria. Four studies were open-label trials, ^{Reference Lambert and Rees11–Reference Hafeiz14} one a retrospective review of cases, ^{Reference Carter15} but the majority (n = 51) took the form of case reports or series. ^{Reference Semenov12,Reference Horsley16–Reference Vattakatuchery and Cesterman65} One paper both reported individual cases and compared drug interviews for psychogenic versus organic deafness. ^{Reference Semenov12}

The studies are presented in online Table DS1. A brief summary of the case reports and series can be seen in online Table DS2, which presents information on each participant, the drug and dosage used, number of drug interviews undertaken, techniques used during the interviews and a description of the outcome. The characteristics of the study population and results of statistical analyses are presented in Table 1. The median age of participants was 29 years (range 10–74), and the mean duration of symptoms 49 days (range 3–3650). However, nearly a third of studies failed to report the duration of symptoms. The combination of two drugs for the interview involved administering midazolam (n = 1), ether (n = 4), methylphenidate (n = 7) or insulin (n = 1), along with the intravenous sedative. Follow-up was specified for 32 participants and ranged from 1 to 60 months (mean = 2.17).

Table 1 Characteristics of the study population and statistical results

	Total, n (%)	Recovered, n (%)	Cross-tabulation	Odds ratio	95% CI
Total	116	92 (79)
Gender
Male	79 (70)	62 (78)	0.70	0.63	0.21–1.87
Female	34 (30)	29 (85)	0.70	1.59	0.54–4.73
Symptom type
Motor	38 (33)	31 (82)	0.07	1.14	0.43–3.07
Sensory	15 (13)	13 (87)	FET	1.71	0.36–8.16
Identity/amnesia	24 (21)	21 (88)	FET	1.94	0.53–7.19
Non-epileptic seizure	7 (6)	6 (86)	FET	1.52	0.17–13.3
Mutism	16 (14)	11 (69)	1.52	0.48	0.15–1.56
Mixed	14 (12)	11 (79)	FET	0.89	0.23–3.51
Atypical	8 (7)	4 (50)	FET*	0.21	0.05–0.93
Age, years
≤29	52 (45)	41 (79)	0.07	1.14	0.44–2.94
≥30	46 (40)	36 (78)	0.01	1.05	0.41–2.73
Characteristics
Battle acquired	32 (28)	25 (78)	0.12	0.84	0.31–2.28
Stressor identified	61 (53)	51 (84)	0.95	1.58	0.63–3.96
Acute (<6 months)	49 (42)	40 (82)	0.37	1.41	0.46–4.32
Chronic (≥6 months)	29 (19)	22 (76)	0.37	0.70	0.23–2.16
Psychiatric comorbidity	31 (27)	21 (68)	4.11*	0.38	0.15–0.99
Drug
Barbiturate	104 (90)	84 (81)	0.22	1.40	0.35–5.65
Benzodiazepine	5 (5)	4 (80)	FET	0.99	0.10–9.29
Additional medication	15 (13)	9 (60)	4.41*	0.30	0.10–0.97
Study conditions
Follow-up provided	40 (34)	30 (75)	1.03	0.62	0.24–1.57
Published pre-1970	69 (59)	56 (80)	0.10	1.16	0.46–2.93
Technique used
Exploration	83 (72)	67 (81)	0.06	1.04	0.77–1.40
Suggestion	48 (41)	44 (92)	6.80**	4.27	1.35–13.5
Emotional catharsis	31 (27)	29 (94)	4.76*	4.76	1.04–21.7
Rehabilitation	15 (13)	10 (67)	1.98	0.43	0.13–1.42
Battle sounds	14 (12)	10 (71)	0.77	0.57	0.16–2.02

Discussion

This study is the first to synthesise all evidence published in English on drug interviews in conversion disorder. By performing a meta-analysis of case reports it has also been possible to identify predictors of a dichotomised outcome. As papers reporting unsuccessful cases are less likely either to be written or published, no conclusions regarding effectiveness can be drawn. However, two of the large-scale studies ^{Reference Horsley7,Reference Tilkin8} utilised drug interviews only in those unresponsive to standard therapy and both report very high response rates.

The most striking feature of the evidence is the apparent beneficence of suggestion. Suggestions are verbal communications that a specific non-volitional response will be experienced by the recipient. ^{Reference Kirsch66} These communications are statistically associated with a good outcome in the meta-analysis; employed in all of the large studies; and were a constant in Hafeiz's trial. ^{Reference Hafeiz14} This study randomly assigned participants to one of four experimental treatments (faradic stimulation; Somlec, a machine that induces sleep; intravenous (IV) methylphenidate; or IV amytal), but sadly omitted a control group. Over 80% of participants receiving one of the first three procedures, allied with suggestion, recovered. A much lower response rate (20%) was seen in the amytal group, although this treatment was discontinued after only five participants somewhat undermining the original study design. Given the experimental and frankly dubious treatments employed some will not resist the conclusion that this study merely reveals those liable to conversion disorder are biddable by nature. However, if conversion disorder is amenable to suggestion then it is worth considering why this should be so.

Susceptibility to suggestion has long been identified with conversion disorder. Babinski, a pupil of Charcot's, went so far as to call the condition pithiatisme, that is, suggestibility. ^{Reference Merskey67} Pierre Janet, who believed suggestibility occurs only in hysterics and disappears on recovery, regarded it as the tendency for a simple idea, internally or externally generated, to develop into automatous complex chains of association that exert an effect over the body outwith conscious control. ^{Reference Janet68} In this model, restored neurological function could be suggested and the idea of this would quickly gain purchase in the individual. However, for Janet this would not constitute a cure as the tendency to suggestibility persists.

Suggestions may influence the placebo response. Elaborate medical procedures involving injection are thought to enhance the placebo element of a treatment, ^{Reference Kaptchuk, Goldman, Stone and Stason69} which in any case is known to be high in subjectively distressing conditions, such as pain and depression. ^{Reference Stewart-Williams and Podd70} A major explanatory model for the placebo response is expectancy theory: a placebo treatment produces an expectation of a certain effect, which the expectation itself produces. ^{Reference Stewart-Williams and Podd70} It may produce this by ameliorating anxiety; reducing negative cognitions while strengthening coping cognitions that cause salutary modifications in behaviour and arousal; or by directly affecting physiological pathways specific to the particular expectation. Expectancies are influenced by the information given about the placebo. So a placebo can cause sedation or stimulation depending on the alleged effect. In drug interviews the use of suggestions augmented by the drama of the procedure is likely to raise the expectancy of recovery and challenge dysfunctional health beliefs that cause and maintain the condition. The active drug may supplement the placebo response without directly influencing the conversion symptoms. The altered psychic state may confirm that ‘something is happening’, which lends further credence to the expectancies.

Interviews under the influence of drugs were developed as a simpler alternative to hypnosis, ^{Reference Horsley7} hence its original title ‘narcosis: a narcotic-induced hypnosis’. This state was thought to render individuals hyper-suggestible and hysterics were thought to be uniquely or particularly hypnotisable. ^{Reference Horsley7} A special hypnoid state is no longer thought to exist, rather the experience of hypnosis is also based upon expectancies. ^{Reference Kirsch66} For example, the majority of those who anticipate amnesia for the hypnotic ‘trance’ will experience this while none of those who do not expect such an eventuality do so. However, perhaps the potentiation of gamma-aminobutyric acid function in limbic and cortical regions caused by benzodiazepines and barbiturates renders the person more suggestible. This has been shown to occur with sodium amytal injections, ^{Reference Eysenck and Rees71} although the precise mechanism remains unclear. To summarise, suggestibility may be a feature of conversion disorder, a drug effect or mechanism behind the placebo response.

The participant achieving an ‘emotional catharsis’ during the interview is also positively associated in the meta-analysis with amelioration of conversion symptoms and reported by Semenov to occur during most successful treatments. ^{Reference Semenov12} The curative power of catharsis in conversion disorder was first noted by Bertha Pappenheim, ^{Reference Jones72} better known as Anna O. Breuer's case in Studies in Hysteria. ^{Reference Freud and Breuer73} She found that by recalling the circumstance of a symptom's onset, a reliving of the emotional experience was followed by resolution of the symptom. Breuer and Freud believed the symptoms were caused by increased energy in the neurological system associated with repressed traumatic memories and that catharsis was a means to purge this excess. Freud found he could not reliably induce hypnosis, ^{Reference Freud74} which Breuer held to be necessary for the process, and so moved on to examine transference and the symbolic value of symptoms. This hydraulic conception of the mind/brain has long been discredited, ^{Reference McDougall75} but the observation may still have some validity.

If repressed traumatic memories are genuinely retrieved during the interview then it is likely the participant will be affected by their recollection. It is not a matter of expunging some packet of energy but reintegrating an upsetting experience, which is closer to Janet's conception of hysteria as dissociation. ^{Reference van der Hart and Brown76} Alternatively, in a model similar to one proposed for post-traumatic stress disorder, the eliciting of avoided traumatic memories will induce anxiety, which reduces through habituation to the traumatic memory. However, many of the life events linked with conversion disorder are stressful rather than traumatic in nature. ^{Reference Craig, Halligan, Bass and Marshall77} The finding may of course be an artifact. Many of the clinicians who reported catharsis will have been at least implicitly influenced by Freud and Breuer's model. All emotional displays may have been perceived in this light and the clinicians' expectations will subtly influence their patient's behaviour.

It has traditionally been thought that chronic conversion symptoms are less amenable to drug interview, ^{Reference Naples and Hackett78} and Lal & Sharma ^{Reference Lal and Sharma13} report much lower response rates in those with symptoms lasting 6 weeks or longer. The natural history of the disorder may explain this observation as the majority of people with acute conversion symptoms leave hospital recovered or greatly improved ^{Reference Binzer and Kullgren79} without recourse to drug interviews. Also, chronicity of symptoms is known to carry a poorer prognosis. ^{Reference Ron, Halligan, Bass and Marshall80} As symptom duration of 6 months or greater is not associated with non-response in the meta-analysis and some participants responded despite having symptoms lasting many years, ^{Reference Sullivan19,Reference Shorvon and Sargant28,Reference Stevens43} drug interviews may have a role in people with chronic, treatment-resistant conversion disorder.

Likewise, it has been claimed that drug interviews are of most value in individuals with dissociative identity disorder. ^{Reference Kavirajan81} However, in the meta-analysis only atypical symptoms, such as sneezing and stammering, are associated with poor response.

Comorbid mental illness is found in more than 50% of those with conversion disorder ^{Reference Crimlisk, Bhatia, Cope, David, Marsden and Ron3} but in only 27% of our sample. The most common psychiatric diagnosis found is personality disorder, although this was not recorded in any of the cases (Table DS2), which probably reflects underreporting in the reports. Personality disorder indicates a poor prognosis, ^{Reference Mace and Trimble82} whereas the opposite is true for affective disorder. ^{Reference Crimlisk, Bhatia, Cope, David, Marsden and Ron3} However, the presence of a comorbid psychiatric disorder was associated with a lower chance of responding to the drug interview. It is likely this is because drug interviews are ineffective in the treatment of affective disorders. ^{Reference Freimer, Lu and Chen49}

Also negatively associated with response is the use of two medications during the procedure, although a number of different pharmacological agents were used so this result must be interpreted with caution. The rationale for adding a stimulant, usually methylphenidate or ether, is to induce emotional catharsis through hyperexcitability. ^{Reference Sargant and Slater83} Therefore, it is interesting that the occurrence of emotional catharsis is associated with recovery, whereas the use of additional medication is not.