Background
Perfectionism is a personality feature characterized by the setting of extremely high and demanding performance standards, which a perfectionist individual strives for and bases their self-evaluation upon (Frost, Marten, Lahart and Rosenblate, Reference Frost, Marten, Lahart and Rosenblate1990). Despite ongoing debate regarding the most appropriate conceptualization, there is general consensus that perfectionism is best understood as a multi-dimensional construct (e.g. Frost et al., Reference Frost, Marten, Lahart and Rosenblate1990; Hewitt and Flett, Reference Hewitt and Flett1991; Dunkley, Zuroff and Blankstein, Reference Dunkley, Zuroff and Blankstein2003).
In line with this, self-report multi-dimensional measures of perfectionism were developed by Frost et al. (Reference Frost, Marten, Lahart and Rosenblate1990) and Hewitt and Flett (Reference Hewitt and Flett1991). The Frost Multi-dimensional Perfectionism Scale (FMPS) assesses the following aspects of perfectionism: concern over mistakes, personal standards, doubts about actions, parental expectation, parental criticism and order. The Hewitt and Flett Multi-dimensional Perfectionism Scale (HMPS) measures self-oriented perfectionism, socially prescribed perfectionism, and other oriented perfectionism. Factor analyses of these widely used measures have identified two underlying factors: “maladaptive evaluative concerns” and “positive achievement striving” (Frost, Heimberg, Holt, Mattia and Neubauer, Reference Frost, Heimberg, Holt, Mattia and Neubauer1993; Cox, Enns and Clara, Reference Cox, Enns and Clara2002). Maladaptive perfectionism is characterized by evaluative concerns, worry and socially prescribed aspects of perfectionism, whilst adaptive perfectionism concerns positive striving, high personal standards and self-oriented perfectionism (Enns and Cox, Reference Enns and Cox1999; Bieling, Israeli, Smith and Antony, Reference Bieling, Israeli, Smith and Antony2003; Rice, Ashby and Slaney, Reference Rice, Ashby and Slaney1998).
Whilst there is some evidence that achievement striving aspects of perfectionism are associated with positive outcomes (Stoeber and Otto, Reference Stoeber and Otto2006 for review), numerous studies have found a robust link between both positive achievement striving and maladaptive evaluative concerns perfectionism and eating disorders (EDs; Egan, Wade and Shafran, Reference Egan, Wade and Shafran2011) and both self-oriented perfectionism and personal standards have been found to be associated with negative outcomes and symptoms of depression and anxiety (Egan et al., Reference Egan, Wade and Shafran2011). In line with this, an alternative definition of perfectionism was proposed by Shafran, Cooper and Fairburn (Reference Shafran, Cooper and Fairburn2002). Within their cognitive behavioural model of perfectionism, “clinical perfectionism” is defined as “the overdependence of self-evaluation on the determined pursuit of personally demanding self-imposed standards in at least one highly salient domain despite adverse consequences.” Within this model, personal standards are regarded as maladaptive when an individual's self-evaluation is based upon meeting these standards.
Perfectionism has important implications for clinical practice, having been implicated in the development and maintenance of a range of disorders including depression, anxiety, EDs and body dysmorphic disorder (BDD) (Shafran and Mansell, Reference Shafran and Mansell2001; Egan et al., Reference Egan, Wade and Shafran2011). It is associated with suicidality (Jacobs et al., Reference Jacobs, Silva, Reinecke, Curry, Ginsburg and Kratochvil2009); self-harm (O’Connor, Rasmussen and Hawton, Reference O’Connor, Rasmussen and Hawton2010); insomnia (Vincent and Walker, Reference Vincent and Walker2000), social phobia (Juster et al., Reference Juster, Heimberg, Frost, Holt, Mattia and Faccenda1996) and Obsessive Compulsive Disorder (OCD; Frost and Steketee, Reference Frost and Steketee1997) and is proposed as an explanation for comorbidity across disorders (Bieling, Israeli and Antony, Reference Bieling, Israeli and Antony2004; Egan et al., Reference Egan, Wade and Shafran2011). Perfectionism is also known to impede successful treatment of depressive disorder (Blatt, Quinlan, Pilkonis and Shea, Reference Blatt, Quinlan, Pilkonis and Shea1995), anxiety disorders (Chik, Whittal and O’Neil, Reference Chik, Whittal and O’Neill2008) and EDs (Sutandar-Pinnock, Woodside, Carter, Olmsted and Kaplan, Reference Sutandar-Pinnock, Woodside, Carter, Olmsted and Kaplan2003). It has been suggested that targeting perfectionism may result in symptom reduction across a range of disorders (Bieling et al., Reference Bieling, Israeli and Antony2004; Shafran et al., Reference Shafran, Cooper and Fairburn2002).
Aims
The aim of this review was to determine the efficacy of interventions explicitly targeting perfectionism in reducing levels of perfectionism in adults or children with a primary problem of a psychiatric disorder and/or clinically significant perfectionism. Given the multi-dimensional conceptualization of perfectionism and limited knowledge on the extent to which perfectionism can be reduced through interventions, a further aim was to investigate which aspects of perfectionism are amenable to change. Due to the purported transdiagnostic nature of perfectionism and relevance to a range of disorders, this review also aimed to investigate change in other symptoms (e.g. anxiety, depression, ED symptoms).
Method
This systematic review was reported according to the “PRISMA statement” (Moher, Liberati, Tetzlaff and Altman, Reference Moher, Liberati, Tetzlaff and Altman2009).
Eligibility criteria for selecting studies
Studies of interventions targeting perfectionism published in peer reviewed journals in English were included, in order that it was possible to access and fully review whole papers. Studies were eligible if they included an intervention explicitly targeting perfectionism. Studies involving interventions of either individual or group format and of any treatment modality were eligible. Studies of interventions involving multiple sessions were included. Studies including participants with any psychiatric disorder, and/or with clinically relevant levels of perfectionism, as defined by scores on an established perfectionism measure (e.g. semi-structured interview, score above a defined cut-off or in relation to a control sample) were eligible. See Table 1 for method of defining clinically relevant perfectionism for each study. Studies without a control group were included – as the focus was upon change between pre and postintervention – as were studies with any type of control group. Case series were included if group means and standard deviations were presented or sufficient data for these to be calculated. Studies were excluded if interventions did not explicitly target perfectionism. Studies were also excluded if participants did not have either a clinical disorder or elevated perfectionism.
Table 1. Study characteristics for eligible studies
Notes: APSR: Almost Perfect Scale Revised, D: Discrepancy subscale; HS: High Standards; CPQ: Clinical Perfectionism Scale; HMPS:Hewitt & Flett Multidimensional Perfectionism Scale. SOP: Self Orientated Perfectionism, SPP: Socially Prescribed Perfectionism; PCI: Perfectionism Cognitions Inventory; FMPS: Frost Multidimensional Perfectionism Scale. CM: Concern over Mistakes subscale, PS: Personal Standards subscale; BAI: Beck Anxiety Inventory; BDI: Beck Depression Inventory; CESD: Centre for Epidemiologic Studies-Depressed Mood Scale; DASS: Depression Anxiety Stress Scale; BDD: Body Dysmorphic Disorder; CFS: Chronic Fatigue Syndrome; GAD: Generalized Anxiety Disorder
Information sources and search strategy
Electronic databases of PsychInfo, PubMed, Medline, EMBASE, SCOPUS and Web of Knowledge were searched up to February 2014 using the following keywords: (“intervention” or “treatment” or “therapy”) and “perfectionism”. The search was repeated combining the names of established perfectionism measures with intervention terms.
Study selection
The search identified 1183 studies, with 32 studies identified as potentially relevant based upon title and abstract. Full papers were selected for 14 studies. Four studies were excluded as participants did not have either a psychiatric disorder or elevated perfectionism, or this was not specified (Wilksch and Wade, Reference Wilksch and Wade2013; Wilksch, Durbridge and Wade, Reference Wilksch, Durbridge and Wade2008; Kutlesa and Arthur, Reference Kutlesa and Arthur2008; Kearns, Forbes and Gardiner, Reference Kearns, Forbes and Gardiner2007). One study was excluded due to not including an intervention of multiple sessions (Aldea, Rice, Gormley and Rojas, Reference Aldea, Rice, Gormley and Rojas2010) and one study because it did not investigate change at group level (Ferguson and Rodway, Reference Ferguson and Rodway1994). Of eight eligible studies, several included multiple measures of perfectionism and were therefore included in more than one meta-analysis. A quality review of studies was conducted (see Table 2). Figure 1 shows a study flow diagram in line with PRISMA.
Table 2. Assessment of quality table for all included studies
Figure 1. Study flow diagram in line with PRISMA
Summary of outcome measures: Perfectionism
Frost Multi-Dimensional Perfectionism Scale (FMPS; Frost et al., Reference Frost, Marten, Lahart and Rosenblate1990) assesses multiple dimensions of perfectionism. A mean clinical cut-off of 24.7 on the CM subscale was established based upon the mean of anxiety disorder samples in a review by Egan et al. (Reference Egan, Wade and Shafran2011).
Hewitt and Flett Multi-Dimensional Perfectionism Scale (HMPS; Hewitt and Flett, Reference Hewitt and Flett1991) assesses three dimensions of perfectionism. Whilst no established clinical cut-offs exist, Hewitt, Flett, Turnbull-Donovan and Mikail (Reference Hewitt, Flett, Turnbull-Donovan and Mikail1991) found mean scores of 70 and 59 for the SOP and SPP scales in a clinical sample.
Clinical Perfectionism Scale (CPQ; Fairburn, Cooper and Shafran, unpublished, cited in Riley, Lee, Cooper, Fairburn and Shafran, Reference Riley, Lee, Cooper, Fairburn and Shafran2007). Assesses cognitive, behavioural and affective aspects of perfectionist goal setting, striving and consequences for self-evaluation.
Almost Perfect Scale Revised (APS-R; Slaney, Rice, Mobley, Trippi and Ashby, Reference Slaney, Rice, Mobley, Trippi and Ashby2001) assesses adaptive (High Standards) and maladaptive (Discrepancy) dimensions of perfectionism. Scores of 42 or above on discrepancy indicate clinical levels of perfectionism, with a cut-off of 37 established for HS (Rice and Ashby, Reference Rice and Ashby2007).
Perfectionism Cognitions Inventory (PCI; Flett, Hewitt, Blankstein and Gray, Reference Flett, Hewitt, Blankstein and Gray1998) is a measure of the frequency of “automatic perfectionistic thoughts”; scores above 66 indicate clinical levels.
Symptom measures
Depression and anxiety
The following measures of anxiety and depression were included: Depression Anxiety Stress Scale (DASS; Lovibond and Lovibond, Reference Lovibond and Lovibond1995); Center for Epidemiological Studies Depression Scale (CESD; Radloff, Reference Radloff1977); Beck Depression Inventory-II (BDI-II; Beck, Steer and Brown, Reference Beck, Steer and Brown1996) and the Beck Anxiety Inventory (BAI; Beck, Epstein, Brown and Steer, Reference Beck, Epstein, Brown and Steer1988). Increased scores on measures indicate higher symptom severity.
ED symptoms
The Eating Disorder Examination (EDE; Fairburn and Beglin, Reference Fairburn and Beglin1994) is a semi-structured interview assessing frequency of behavioural symptoms and ED related cognitions. It consists of four subscales assessing restraint, eating concern, shape concern, and weight concern. Higher scores indicate increased severity.
Obsessive compulsive symptoms
The following measures of obsessive compulsive symptoms were included in studies within the review: the Maudsley Obsessional-Compulsive Inventory (MOCI; Hodgson and Rachman, Reference Hodgson and Rachman1977); Padua Inventory-Washington State University Revision (PI-WSUR, Burns, Keortge, Formea and Sternberger, Reference Burns, Keortge, Formea and Sternberg1996) and the Responsibility Attitude Scale (RAS; Salkovskis et al., Reference Salkovskis, Wroe, Gledhill, Morrison, Forrester and Richards2000). Higher scores indicate greater symptomatology.
Synthesis of data
Means and standard deviations for pre and postintervention scores were taken from each paper. Where standard errors (SE) were reported, standard deviation (SD) was calculated. Mean differences between pre and postintervention scores were standardized by the pooled standard deviation to calculate Hedges’ g (difference between pre and postintervention scores divided by pooled SD; Hedges, Reference Hedges1981). The following values correspond to relative effect sizes: small (g = 0.2), medium (g = 0.5) and large (g = 0.8). For studies that were commented on only, where effect sizes were reported as Cohen's d, Hedges’ g was calculated to allow comparison between studies. Effect sizes for studies are shown in Tables 3 and 4.
Table 3. Perfectionism measures: effect sizesa of each study included in the meta-analyses (FMPS CM, FMPS PS, HMPS SOP, HMPS-SPP)
aStandardized effect sizes according to weighting of studies in meta-analyses
FMPS CM: Frost Multidimensional Perfectionism Scale: Concern over Mistakes subscale; FMPS PS: Frost Multidimensional Perfectionism Scale: Personal Standards subscale; HMPS SOP: Hewitt and Flett Multidimensional Perfectionism Scale: Self Orientated Perfectionism; HMPS SPP: Hewitt and Flett Multidimensional Perfectionism Scale: Socially Prescribed Perfectionism
Table 4. Anxiety and depression measures: effect sizesa of each study included in the meta-analyses
aStandardized effect sizes according to weighting of studies in meta-analyses
Analyses were conducted using STATA 11 (Stata Corp, College Station, TX, USA) using the metan command (Bradburn, Deeks and Altman, Reference Bradburn, Deeks and Altman1998; Harris et al., Reference Harris, Bradburn, Deeks, Harbord, Altman and Sterne2008). Scales were included in the analyses where internal reliability of the measure was acceptable or above in the included studies (Cronbach's alpha = ≥ 0.70; Nunnally and Bernstein, Reference Nunnally and Bernstein1994). Analysis for perfectionism measures was conducted separately for each subscale as we were interested in which aspects of perfectionism were found to change in response to interventions. For perfectionism outcomes, meta-analyses were carried out for PS and CM subscales of the FMPS and SOP and SPP subscales of the HMPS by pooling the standard effect sizes using a random effects model. Meta-analyses were also carried out to investigate change in depressive and anxiety symptoms across different measures.
Results
Study characteristics
All studies used a cognitive behavioural based intervention (for specific details see Table 1). Four studies involved individual interventions, two guided self-help, one web-based intervention and one group format CBT. Interventions varied in the number of sessions delivered, ranging between 8 BDD and 14 sessions. All studies involved adults. The samples of four studies included participants with clinical disorders, including depression, anxiety, chronic fatigue syndrome, OCD, BDD, bulimia nervosa and Eating Disorder Not Otherwise Specified (EDNOS). One study included a mixed sample of patients with clinical disorders and participants with elevated perfectionism. Three studies included participants with elevated perfectionism.
Heterogeneity
I2 (Higgins, Thompson, Deeks and Altman, Reference Higgins, Thompson, Deeks and Altman2003) was calculated as a measure of heterogeneity between studies due to small sample sizes, based on Cochran's Q test: measure of heterogeneity, I2 = 100% x (Q-df)/Q. I2 ranges between 0% indicative of no inconsistency and 100% indicative of high heterogeneity. There was no evidence of heterogeneity between studies using the FMPS CM subscale, HMPS SPP subscale, or those reporting anxiety and depression outcomes (all I2 = 0.0%). There was evidence of heterogeneity for the FMPS PS subscale (I2 = 43.8%) and HMPS SOP subscale (I2 = 61.3%).
Publication bias
Egger tests (Egger, Smith, Schneider and Minder, Reference Egger, Smith, Schneider and Minder1997) were carried out using the user contributed STATA command meta bias and funnel plots generated to investigate the presence of publication bias for perfectionism outcomes. No evidence was found for FMPS CM and PS subscales (p = .22; 0.939 respectively), nor HMPS SOP (p = .25). For the HMPS SPP a trend was detected (p = .07). These results should be interpreted with caution due to the small number of studies.
Synthesis of results
Perfectionism: FMPS
The FMPS was the most widely used measure (seven studies). Three studies reported all subscales. Two studies reported the CM and PS subscales only, one study reported total score plus these two subscales combined and one study reported CM only. Meta-analyses were carried out separately on the five studies reporting PS and six studies reporting CM. Riley et al. (Reference Riley, Lee, Cooper, Fairburn and Shafran2007) was excluded from the meta-analysis as the two subscales were not reported separately. Two studies included the DA subscale and are commented upon only.
Personal Standards (PS)
Five studies used the PS subscale (N = 71), with the meta-analysis showing a pooled standardized mean difference between pre and postintervention of g = 0.79, a large effect size (CIs = 0.44 – 1.12). Figure 2 shows the pooled and individual effect sizes for relevant studies.
Figure 2. Forest plot for FMPS PS subscale: standardized effect sizes for change between pre and postintervention
Concern over Mistakes (CM)
Six studies used this subscale of the FMPS (N = 100). The meta-analysis showed a pooled standardized mean difference between pre and postintervention of g = 1.32, a very large effect size (CIs = 1.02 – 1.64). Figure 3 shows the effect size of studies using the FMPS CM subscale and the pooled estimate.
Figure 3. Forest plot for FMPS CM subscale: standardized effect sizes for change between pre and postintervention
Doubts about Actions (DA)
Two studies used the FMPS DA subscale: Pleva and Wade (Reference Pleva and Wade2007) found significant change during treatment, p < .01, g = 0.61, a medium effect size, whilst Radhu et al. (Reference Radhu, Daskalakis, Guglietti, Farzan, Barr and Arpin-Cribbie2012) did not observe any change over treatment (p not reported).
HMPS: Self Oriented Perfectionism (SOP)
Four studies used the SOP subscale of the HMPS (N = 55). The meta-analysis showed a pooled standardized mean difference between pre and postintervention, g = 0.81 (CIs = 0.41 – 1.20), a large effect size. Figure 4 shows the effect size for each study and the pooled estimate.
Figure 4. Forest plot for the HMPS SOP subscale: standardized effect sizes for change between pre and postintervention
H-MPS: Socially Prescribed Perfectionism
Four studies used the SPP subscale of the HMPS (N = 55). The meta-analysis showed a pooled standardized mean difference between pre and postintervention of g = 0.52 (0.13 – 0.90), a medium effect size. See Figure 5 for individual effect sizes and the pooled estimate.
Figure 5. Forest plot for the HMPS SPP subscale: standardized effect sizes for change between pre and postintervention
CPQ
The CPQ was used in three studies. Glover, Brown, Fairburn and Shafran (Reference Glover, Brown, Fairburn and Shafran2007) and Riley et al. (Reference Riley, Lee, Cooper, Fairburn and Shafran2007) reported significant differences in scores between pre and posttreatment (p = .01, p < .01 respectively), very large effect sizes (g = 1.13, 1.24). Steele et al. (Reference Steele, Waite, Egan, Finnigan, Handley and Wade2013) also reported significant change (p = < .05), a large effect size (g = 0.90).
PCI
Two studies (Arpin-Cribbie, Irvine and Ritvo, Reference Arpin-Cribbie, Irvine and Ritvo2012; Radhu et al., Reference Radhu, Daskalakis, Guglietti, Farzan, Barr and Arpin-Cribbie2012) used this measure, with significant change observed for both studies (p = < .01, <.05 respectively), large (g = 1.01) and very large (g = 1.41) effect sizes.
APS-R: discrepancy
Two studies used this scale. Arpin-Cribbie et al. (Reference Arpin-Cribbie, Irvine and Ritvo2012) reported significant change between pre and posttreatment (p = <.01), a medium effect size (g = 0.72). However Radhu et al. (Reference Radhu, Daskalakis, Guglietti, Farzan, Barr and Arpin-Cribbie2012) found no significant change.
APS-R: High Standards
One study (Radhu et al., Reference Radhu, Daskalakis, Guglietti, Farzan, Barr and Arpin-Cribbie2012) used this subscale and reported significant change between pre and posttreatment (p = < .05), a very large effect size (g = 1.3).
Anxiety
Seven studies reported changes in symptoms of anxiety (N = 98). The meta-analysis showed a pooled standardized mean difference between pre and postintervention of g = 0.52 (CIs = 0.23 – 0.81), a medium effect size. Figure 6 shows the effect size for each study and the pooled estimate.
Figure 6. Forest plot for anxiety measures: standardized effect sizes for change between pre and postintervention
Depression
Seven studies reported changes in symptoms of depression (N = 98). The meta-analysis showed a pooled standardized mean difference between pre and postintervention of g = 0.64 (CIs = 0.35 – 0.92), a medium effect size. Figure 7 shows the effect size for each study and the pooled estimate.
Figure 7. Forest plot for depression measures: standardized effect sizes for change between pre and postintervention
One study (Steele et al., Reference Steele, Waite, Egan, Finnigan, Handley and Wade2013) reported negative affect overall (anxiety, depression and stress measured by the DASS total score) and found significant change (p = < .05) between pre and postintervention, a large effect size (d = 0.98). Table 3 shows the effect sizes for each study reporting anxiety and depression scores.
Eating disorder symptoms
One study investigated changes in ED symptoms (Steele and Wade, Reference Steele and Wade2008) and found significant changes at p < .05 for objectively reported episodes of bingeing (g = 0.31, small effect size), vomiting (g = 0.48, medium effect size) and concerns with shape and weight (g = 3.83, very large effect size). No difference was found between pre and postintervention for subjectively measured bingeing, laxative use or excessive exercise.
Obsessive compulsive symptoms
One study (Pleva and Wade, Reference Pleva and Wade2007) reported significant change (p < .001) of very large effect size on the MOCI (g = 1.73), PI-WSUR (g = 1.90) and RAS (g = 1.77).
Discussion
The purpose of this systematic review and meta-analysis was to assess research evidence for interventions targeting perfectionism. There is support that it is possible to significantly reduce aspects of perfectionism using a cognitive behavioural approach with short interventions in adults with perfectionism as a primary problem or in addition to a psychiatric diagnosis. Meta-analyses demonstrated large pooled effect sizes for change between pre and postintervention on Personal Standards and Concern Over Mistakes subscales of the FMPS. Meta-analyses also found a large pooled effect size for the Self Oriented Perfectionism subscale of the HMPS and medium effect size for the Socially Prescribed Perfectionism subscale. Medium pooled effect sizes were found for changes in symptoms of anxiety and depression. Individual studies not included in the meta-analyses also reported significant change of medium effect size for eating disorder related measures and very large effect sizes for obsessive compulsive symptoms.
These findings are promising as perfectionism is found to impede treatment across a range of disorders. The efficacy of cognitive behavioural interventions in reducing perfectionism is in line with theory implicating biased cognitive processes in the development and maintenance of perfectionism. This review included studies involving participants with a range of psychiatric diagnoses, with evidence not only for reductions in perfectionism but also symptoms of anxiety, depression and EDs. These findings build upon evidence concerning the transdiagnostic nature of perfectionism (Egan et al., Reference Egan, Wade and Shafran2011) and support theory suggesting that targeting perfectionism may be effective in reducing symptoms across a range of disorders (Bieling et al., Reference Bieling, Israeli and Antony2004; Shafran et al., Reference Shafran, Cooper and Fairburn2002).
Observed changes in Self Oriented Perfectionism and Personal Standards reflect changes in aspects of perfectionism considered by some theorists as benign or adaptive. However, within a model of clinical perfectionism (Shafran et al., Reference Shafran, Cooper and Fairburn2002) it is argued that positive achievement striving is toxic when combined with self-evaluation based upon the meeting of these standards. Therefore, in order to sustain clinically relevant changes in perfectionism, arguably both maladaptive evaluative concerns and positive achievement striving aspects of perfectionism need to be modified. This is particularly the case with individuals with ED symptoms, depression and anxiety where robust links with both aspects of perfectionism have been established.
Further research is needed to investigate the most effective format of perfectionism interventions, optimal dosage, and into specific disorders where there is currently a lack of evidence. Only one study included an ED sample and there are currently no published studies investigating interventions targeting perfectionism in Anorexia Nervosa (AN). Research in this area is needed given the implication of perfectionism in AN and its presence at elevated levels relative to other disorders (e.g. Egan et al., Reference Egan, Wade and Shafran2011). There may be important differences that need to be addressed in the treatment of perfectionism in EDs compared with other disorders.
Future research would benefit from inclusion of outcome measures assessing disability, handicap and distress associated with perfectionism. This would prove particularly useful in studies involving participants with different disorders, in order to allow comparison across disorders.
There are a number of limitations to this review. Differences between measures make it difficult to draw conclusions based upon available literature. There were also design features that were heterogeneous across studies and may account for differences in outcomes. Included studies varied in intervention format and dosage and although all studies involved a CBT based intervention, content varied. It is difficult to speculate why studies were associated with different effect sizes. For example, whilst it might be hypothesized that more intensive interventions – e.g. individual therapy versus self-help, or an increased number of sessions – would be associated with larger effect sizes, this was not the case. Differences in participants between studies are another potential confounding variable and it is possible that some disorders may be more amenable to change in perfectionism. Unfortunately, some studies with mixed samples gave limited details of numbers of participants with each diagnosis, making it difficult to draw conclusions regarding the generalizability of interventions. A further limitation is that some studies included participants with elevated perfectionism who were not patients and therefore self-selected to participate. This is a potential source of bias and confounding factor between those choosing to take part and those referred to studies. Those self-selecting may be hypothesized to have higher motivation to change. Motivation to change will be important to address in future studies and has been highlighted as being relevant to perfectionism (Egan, Piek, Dyck, Rees and Hagger, Reference Egan, Piek, Dyck, Rees and Hagger2013).
Caution must be taken especially when interpreting the findings of the FMPS PS subscale and the HMPS SOP subscale given indications of heterogeneity between studies. The small sample sizes meant it was not possible to control for study differences using a random effects model, or to explicitly explore factors affecting outcome. The small number of studies included in the meta-analyses is a limitation of the study. This is a threat to validity and results should therefore be interpreted with caution.
However, this review provides initial evidence that a cognitive behavioural approach may be effective in reducing perfectionism in individuals with a psychiatric diagnosis or elevated of perfectionism. Given existing research demonstrating an association between perfectionism and poorer prognosis in several disorders, these findings have clear clinical implications. However, further research is needed into specific disorders not yet investigated, including AN and BDD.
Acknowledgements
Samantha Lloyd is supported by an Institute of Psychiatry/Medical Research Council PhD studentship. Kate Tchanturia would like to acknowledge support from Maudsley Charity Health in Mind.
Comments
No Comments have been published for this article.