Rate of relapse among individuals with psychosis is high Reference Robinson, Woerner, Alvir, Bilder, Goldman and Geisler1 and typically necessitates admission to hospital or intensive intervention from a community crisis team. Service cost for individuals who relapse is four times greater than for those who do not. Reference Almond, Knapp, Francois, Toumi and Brugha2 Hence, preventing relapse of psychosis is an urgent challenge for any health service. Severe and victimising childhood trauma may lead to worse long-term outcome of psychotic illness Reference Neria, Bromet, Carlson and Naz3 and, thus, more frequent relapses, although the evidence is equivocal. Reference Greenfield, Strakowski, Tohen, Batson and Kolbrener4 This report summarises available research investigating the influence of childhood trauma on relapse requiring admission to a psychiatric hospital in patients with psychosis. Childhood trauma is defined as including physical, sexual or emotional abuse, or physical or emotional neglect. Relapse has ordinarily been identified as a change in symptom severity or functioning and has been indexed using clinical rating instruments. For the purpose of this report, we have defined relapse as indicating hospital admission only and not changes in symptom severity or functioning. As an outcome measure, hospital admission is a clear, objective and reliable measure that has high face validity, is less vulnerable to bias and comparable across studies, and has implications for utilisation of healthcare resources.
Method
A more detailed description of the methods can be found in online supplement DS1. Relevant studies were identified by searching four electronic databases in February 2015 using search terms derived from those used in previous literature. Reference Varese, Smeets, Drukker, Lieverse, Lataster and Viechtbauer5 Three researchers (N.P., E.F., E.K.) followed a four-phase protocol to identify relevant studies (see online Fig. DS1 and online supplement DS2). Studies were included if they examined outcome as relapse/episode of illness resulting in psychiatric hospital admission and included adults diagnosed with affective or non-affective psychosis, who had experienced childhood trauma. Quality and methodological robustness (see online Table DS1), were examined using an amended quality assessment tool. Reference Beards, Gayer-Anderson, Borges, Dewey, Fisher and Morgan6
Results
A final set of seven studies published between 2005 and 2013 matched the inclusion and exclusion criteria. Reference Alvarez, Roura, Oses, Foguet, Sola and Arrufat7–Reference Brown, McBride, Bauer and Williford13 They reported on a total sample of 946 participants, mostly diagnosed with an affective psychotic disorder with almost half reporting having experienced childhood trauma. Largely, relapse requiring psychiatric hospital admission for psychosis data were gathered by screening clinical records, and childhood trauma data were collected retrospectively using standardised self-report measures (see online Table DS3). Two studies Reference Schenkel, Spaulding, DiLillo and Silverstein9,Reference Brown, McBride, Bauer and Williford13 used observer-rated interviews, which extract fine-grained information on childhood trauma (such as timing/severity) and can therefore offer a more precise measurement of the bearing of stressful life experiences on an individual than self-report measures. The average quality score among the seven included studies was 8.7 out of a possible 16 (median 9; range 4–11). Only one study Reference Larsson, Aas, Klungsoyr, Agartz, Mork and Steen8 scored above the threshold suggested for methodological robustness (⩾70% of total quality score). Studies were limited by modest sample size, lack of valid data collection methods and lack of adjustment for confounders. Additional material relating to the results is available in online supplement DS3.
No consistent pattern demonstrating the influence of childhood trauma on relapse requiring psychiatric hospital admission for psychosis emerged (Table DS3). Two studies Reference Alvarez, Roura, Oses, Foguet, Sola and Arrufat7,Reference Schenkel, Spaulding, DiLillo and Silverstein9 reported a significant difference in relapse requiring psychiatric hospital admission in psychosis between patients with a pre-existing disorder with or without a history of childhood trauma, with the former having more admissions to hospital. Both studies included individuals who had been unwell for at least 15 years on average. However, Alvarez et al Reference Alvarez, Roura, Oses, Foguet, Sola and Arrufat7 reported this effect as significant only in patients with bipolar affective disorder, not in those with schizophrenia and schizoaffective disorder, despite the latter constituting the major proportion (61%) of participants included in the study.
Two studies Reference Larsson, Aas, Klungsoyr, Agartz, Mork and Steen8,Reference Conus, Cotton, Schimmelmann, Berk, Daglas and McGorry11 reported a negative relationship between childhood trauma and relapse requiring psychiatric hospital admission for psychosis. However, Conus et al noted that individuals with a history of childhood sexual or physical abuse were more likely to disengage from treatment (P = 0.017). Reference Conus, Cotton, Schimmelmann, Berk, Daglas and McGorry11 Interestingly, this group were the only sample with early psychosis included in this report. Larsson et al also reported a significant non-linear association between total childhood trauma (d.f. = 8.73, P = 0.038) and sexual abuse (d.f. = 2.8, P = 0.009) and psychotic episodes. However, whether the psychotic episodes described resulted in psychiatric hospital admission was not specified.
Three studies Reference Garno, Goldberg, Ramirez and Ritzler10,Reference Cutajar, Mullen, Ogloff, Thomas, Wells and Spataro12,Reference Brown, McBride, Bauer and Williford13 did not find a significant difference in relapse requiring psychiatric hospital admission between individuals with a history of childhood trauma and those without. Garno and colleagues did note increased past-year rapid cycling in individuals with a history of childhood emotional abuse, physical abuse and emotional neglect in patients with bipolar disorder, but did not specify whether rapid cycling resulted in hospital admission. Reference Garno, Goldberg, Ramirez and Ritzler10 Brown et al found that individuals with a history of any type of childhood abuse were more likely to be admitted to psychiatric hospital involuntarily at index episode (i.e. time of recruitment into the study) compared with those with no abuse (P = 0.029, OR = 2.37, 95% CI 1.10–5.14). Reference Brown, McBride, Bauer and Williford13 However, a significant difference in the number of admissions to hospital or number of days spent in hospital in the past 5 years was not revealed between those with a history of childhood abuse and those without. The sample included in Brown et al overlapped with another study Reference Bauer, Shea, McBride and Gavin14 identified through reference screening, however the latter included a smaller sample size and thus was excluded from this report.
Discussion
These results suggest that evidence of the effect of childhood trauma on risk of relapse of psychotic illness is limited, thus supporting a recent review and meta-analysis investigating the effects of trauma on the persistence of psychotic symptoms. Reference Trotta, Murray and Fisher15 Overall, the evidence does not support a consistent pattern of effect of childhood trauma on risk of admission to hospital, most likely related to heterogeneity in methodology and outcomes of interest in the studies included. Studies that demonstrated a significant association between childhood trauma and relapse requiring psychiatric hospital admission for psychosis included chronic samples, suggesting that stage of illness may have an impact on outcome. Length of follow-up period, for the two studies Reference Conus, Cotton, Schimmelmann, Berk, Daglas and McGorry11,Reference Brown, McBride, Bauer and Williford13 that reported a follow-up period, did not appear to have an effect on the findings (see Table DS3).
All of the studies included in the report scored poorly when assessed for methodological quality. Overall, the main limitations (see also online supplement DS4) of the reviewed studies include: (a) small and heterogeneous samples, which limit the exploration of dose–response relationships, prevent comparisons and the ability to control for potential confounding effects; (b) diverse methodologies, making it difficult to compare and draw conclusions; (c) use of retrospective methods to collect childhood trauma data, which are particularly challenging in patients with psychosis because of the risk of poor event-recall; and (d) lack of adequate consideration of potential confounders, particularly those relevant to relapse, such as cannabis use. Heterogeneity among the studies included in this report precluded any quantitative synthesis and estimation of a pooled effect size of the association between childhood trauma and relapse requiring psychiatric hospital admission.
It is important to note that this report is limited by the restrictive criteria for the outcome of interest, that is, relapse defined as psychiatric hospital admission, as opposed to symptom severity and/or functional decline. Hospital admission figures, which may be influenced by factors other than relapse, may not be representative of the actual number of individuals with psychosis who experience relapse. However, measurement of hospital admissions is perhaps more reliable, objective and comparable across studies in comparison with measurements based upon the recording of change in symptom severity. Exclusion of additional types of childhood trauma (such as bullying/parental loss) may also limit the generalisability of these findings. Despite these weaknesses, the current report suggests that there is little evidence available to suggest a discernible effect of childhood trauma on relapse requiring psychiatric hospital admission for psychosis and highlights the need for more research. This may imply that for some individuals, social or environmental factors have less bearing on the natural path of their illness or may merely reflect that childhood trauma history is not routinely recorded in clinical practice, Reference Read, Hammersley and Rudegeair16 particularly to the extent that other risk factors for poor outcome are. Reference Schoeler, Monk, Sami, Klamerus, Foglia and Brown17
Funding
S.B. has received support from the NIHR (NIHR Clinician Scientist Award; ) and the UK MRC () and from the NIHR Mental Health Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King's College London.
Acknowledgements
We would like to thank Professor Mark Bauer of Harvard Medical School for providing additional study information and clarification.
eLetters
No eLetters have been published for this article.