Reliability, Validity and Factor Structure of the Polish Version of the Massachusetts General Hospital Hair Pulling Scale

Trichotillomania (TTM, hair-pulling disorder) is a chronic psychiatric condition whose TTM diagnostic criteria (American Psychiatric Association, 2013) are recurrent pulling out of hair resulting in noticeable hair loss, severe distress and impairment in day-to-day functioning and repeated attempts to decrease or stop hair-pulling. Hair-pulling cannot be attributed to or better explained by another mental or medical condition. Trichotillomania is currently classified within the broader category of Obsessive–Compulsive and Related Disorders (APA, 2013) due to the biological (Monzani et al., 2014) and phenomenological (Lochner et al., 2005) similarities between TTM and OCD.

Trichotillomania usually starts in adolescence and follows a chronic course of waxing and waning symptom severity (Bloch, 2009). Recent metanalyses suggested that the prevalence of TTM is 1.4% in the general population; subclinical hair-pulling is also common, with prevalence rates as high as 8.84% (Thomson et al., 2022). It is plausible, however, that the actual presence of TTM in the community is even higher, since many sufferers deny pulling behaviors due to shame and low social acceptance of the disorder. Given that hair-pulling contributes to wide-ranging and long-lasting impairments in not only physical health but also several psychological domains, such as self-esteem, body image, emotions and social functioning (Diefenbach et al., 2005), the accurate assessment of TTM, including its severity and consequences, seems to be crucial.

Investigating the prevalence of TTM as well as its severity and impact on daily functioning requires psychometrically validated tools. Currently, the assessment of TTM relies primarily on clinical interview and a limited range of paper-and-pencil measurements (Diefenbach et al., 2005), including clinician and patient rating scales. Self-reports are widely used due to the numerous benefits they provide: ease of administration; ability to capture a wide range of hair-pulling characteristics (symptom severity, interference and distress); control over the time and duration of assessment (Diefenbach et al., 2005); and the possibility of multiple administration, even in large groups, which facilitates TTM research. Of these, the MGH-HPS (Keuthen et al., 1995) is the most commonly used self-report rating scale to measure hair-pulling in adults, in both everyday clinical practice and scientific research.

The MGH-HPS includes the relevant items from the Yale-Brown Obsessive–Compulsive Scale (Y-BOCS), modified to refer to trichotillomania symptoms (Keuthen et al., 1995). In particular, the Y-BOCS items that refer to obsessions were replaced with items concerning hair-pulling urges (the frequency of pulling urges; the intensity of pulling urges; the perceived ability to control hair-pulling urges). Also, the subsequent Y-BOCS items concerning compulsions were modified to rate hair-pulling frequency, attempts to resist hair-pulling, and actual control over this behavior. Additionally, to improve the internal consistency of the scale, the item related to the social impact of hair-pulling was dropped due to its poor correlation with the total MGH-HPS score and limited response variability (Keuthen et al., 1995). The final version of the MGH-HPS contains several items assessing frequency, intensity, and control of hair-pulling urges; frequency, resistance and control of hair-pulling behaviors, and distress associated with hair-pulling.

The final 7-item version of the MGH-HPS was found to have very good internal consistency (with coefficient alpha of 0.89) in the sample of chronic hair-pulling patients (Keuthen et al., 1995). Moreover, the data from an independent sample yielded acceptable test–retest reliability for use in clinical assessment, and acceptable convergent and divergent validity (O’Sullivan, et al., 1995). The MGH-HPS was also confirmed to be a useful tool for monitoring the course of this disorder due to its sensitivity to changes in symptoms (O’Sullivan, et al., 1995).

The one-factor structure of the MGH-HPS was obtained in the original validation study that was conducted on a group of 119 hair-pullers during the development of the scale (Keuthen et al., 1995), suggesting that it is a homogeneous tool that measures the severity of TTM symptoms in a clinical sample. However, in a subsequent internet-based survey (Keuthen et al., 2007), two-factors referring to hair-pulling, namely “Severity” and “Resistance and Control”, were identified in a large group (n = 990) of self-reported hair-pullers who confirmed meeting all DSM IV diagnostic criteria. In the same study, this two-factor solution was confirmed on a larger (n = 1697) mixed sample of individuals who satisfied both strict and more liberal DSM IV criteria (Keuthen et al., 2007). A more recent validation of the Persian and Turkish version of the MGH-HPS (Aydin et al., 2023; Rabiei et al., 2014) on a large clinical sample (n = 635) of individuals attending medical and psychology clinics (Rabiei et al., 2014) and patients (n = 50) who were being treated for trichotillomania (Aydin et al., 2023) yielded a unidimensional factor solution that corresponded with the structure that emerged from Keuthen et al.’s (1995) original study.

The aim of the current survey was to develop the Polish version of the MGH-HPS (Keuthen et al., 1995) and evaluate its psychometric properties in a community sample. Specifically, we examined the factor structure, reliability, convergent and divergent validity, and diagnostic accuracy of the MGH-HPS in a group of adults with self-reported clinical and subclinical hair-pulling.

The aim of the study was to examine the reliability and validity of the Polish version of the Massachusetts General Hospital Hair-pulling Scale, originally developed by Keuthen et al. (1995). Also, the factor structure of the scale was assessed using confirmatory factor analyses. The results indicate that the Polish version of the MGH-HPS has acceptable psychometric properties and may be applied to measure hair-pulling symptoms in adult Polish samples.

To investigate the factor structure of the Polish version of the MGH-HPS, we firstly tested the two-factor model endorsed in Keuthen et al.’s (2007) internet-based survey. This model showed good fit to the data; however, the correlation between factors turned out to be very high (r = 0.9), suggesting that the one-factor model may describe the structure of the scale better than the two-factor model. Further analysis showed that the unidimensional, more parsimonious model fitted data very well and did not perform worse than the two-factor model. Hence, one total score reflecting the severity of actual hair-pulling should preferably be calculated for the Polish version of the MGH-HPS by summing up the seven individual item scores. This finding is in line with the results reported by Keuthen et al.’s (1995) original study on the development of the scale and with more recent studies on the Persian (Rabiei et al., 2014) and the Turkish (Aydin et al., 2023) MGH-HPS translations. Aydin et al. (2023) suggested that the differences in factor structure reported in previous studies on the MGH-HPS may be attributed to the use of different study samples. While the use of clinical samples resulted in a unidimensional structure (Aydin et al., 2023; Keuthen et al., 1995; Rabiei et al., 2014), the online survey involving participants recruited from the general population resulted in a two-factor model. However, it is worth noting that these studies actually did not formally compare the unidimensional and two-factor models: they either performed EFA to determine the number of factors (Aydin et al., 2023; Keuthen et al., 2007), or they assumed the unidimensional factor structure of the scale and performed CFA but did not test the alternative model (Rabiei et al., 2014). Our study suggests that both models may perform similarly well in samples recruited from the general population, but the strong correlation between factors suggests that the one-factor model may describe the structure of the scale more accurately.

The Polish version of the MGH-HPS revealed good internal consistency for both the one-factor model (Cronbach’s alpha coefficient as high as 0.89) and the two-dimensional structure (Cronbach’s alphas for “Severity” and “Resistance and Control” subscales 0.85 and 0.84, respectively). The reliability (internal consistency) of the unidimensional structure is similar to the Cronbach’s alpha value obtained in Keuthen et al., and and’s (1995, 2007) previous validation studies (i.e., α = 0.88 and 0.84) and to that yielded from the validation study of the Persian translation of the scale (α = 0.91; Rabiei et al., 2014); however, this reliability is lower than the value calculated for the Turkish language revised version (α = 0.96; Aydin et al., 2023). The comparison of reliability coefficients for the MGH-HPS total score may support Aydin et al.’s (2023) recent observation that replacing the term “hair” with the term “trichome/hair” (as was done for the Turkish version) may improve the reliability of the scale, since subjects who only pulled eyebrows or eyelashes may rate some of the items inaccurately. Also, the internal consistency calculated in our study for the “Severity” and “Resistance and Control” factors (0.85 and 0.84, respectively) was comparable to the reliability identified by Keuthen et al. (2007) for these factors (0.83 and 0.81).

The current study also confirms the ability of the Polish version of the MGH-HPS to distinguish clinical and non-clinical hair-pullers. The findings clearly demonstrated that individuals with subclinical symptoms scored significantly lower on the MGH-HPS than those who endorsed DSM-5 criteria for hair-pulling disorder. A significant difference was yielded not only for the overall score of the scale, but also for the “Severity” and “Resistance and Control” subscales. Moreover, the ROC analyses indicated that the total MGH-HPS score cut-off value of 13 can with 100% sensitivity and 65% specificity distinguish participants who are likely to suffer from hair-pulling disorder from non-clinical hair-pullers. It should be stressed, however, that the cut-off value obtained in our study refers to the difference between the clinical group and the group of individuals who also confirmed hair-pulling behaviors but did not meet DSM 5 criteria for hair-pulling disorder. Therefore, it is not surprising that the cut-off value is higher than the value of ≥ 9 that was adopted by Aydin et al. (2023) in a recent study using a clinical sample and a non-clinical control group.

Our findings also indicated that women and men do not differ according to the MGH-HPS total score; however, symptoms reported by women seem to be more severe, as indicated by a significantly higher Severity subscale score. This finding is in line with a recent meta-analysis (Thomson et al., 2022) which showed that although women are not more likely than men to engage in any hair-pulling behaviors, they are more inclined to report noticeable hair loss due to pulling, which may reflect greater severity of symptoms.

Due to the absence of other valid Polish-language tools for trichotillomania measurement, the convergent validity was examined by calculating the Spearman’s correlations between MGH-HPS score and the number of DSM-5 hair-pulling disorder criteria (without the question concerning the presence of a dermatological issue or mental illness which may better explain the symptoms was not taken into account). As expected, a strong, positive relationship between these measurements was yielded, which indicates that individuals who score high on the MGH-HPS are also more likely to confirm meeting the trichotillomania criteria. Divergent validity was assessed by calculating Spearman’s correlations between the MHG-HPS score and the DASS-21 score, which is a self-reported scale measuring current symptoms of depression, anxiety, and stress. In our study we did not observe a significant relationship between the MHG-HPS scores and the DASS-21 dimensions (except a weak correlation between the MGH-HPS “Severity” subscale and the DASS-21 stress subscale). This finding supports the discriminant validity of the MGH-HPS; however, it also indicates that in the study sample hair-pulling was not associated with emotional distress. This observation is inconsistent with previous studies showing a high co-occurrence of TTM, depression and anxiety (e.g., Grant, et al., 2017; Houghton et al., 2016; Keuthen et al., 2016; Woods et al., 2006); however, it is in line with a recent meta-analysis showing the relatively limited association between trichotillomania and negative affectivity (Snorrason et al., 2023). It is plausible that the lack of correlational relationship yielded in the current study may be at least partially a consequence of the use of a small, non-clinical community sample. While the high comorbidity of anxiety and mood disorders are detected predominantly in help-seeking individuals with severe psychiatric conditions (Caspi & Moffitt, 2018; Lahey et al., 2021), subclinical hair-pulling is less likely to be accompanied by depression and anxiety, given that it does not lead to noticeable distress or functional impairment. The weak, positive correlation between the severity of hair-pulling and the DASS-21 stress subscale suggests, however, that the more severe the TTM symptoms, the higher the stress experienced by individuals. The relationship between these two variables may be bidimensional, as agitation could be both a cause and effect of hair-pulling (Mansueto, et al., 1997).

Validity analyses also showed the lack of a significant relationship between the MGH-HPS and any of the OCI-R subscales. Although this finding confirms the divergent validity of the MGH-HPS, it is inconsistent with the results of previous studies showing the link between trichotillomania and obsessive–compulsive symptoms, as well as with theoretical assumptions that TTM and OCD appear to be strongly comorbid and genetically related (Houghton et al., 2016; Keuthen et al., 2014; Monzani et al., 2014). It is plausible, however, that the relationship between hair-pulling behaviors and various OCD symptoms is less visible in a sample recruited from the general population which consists mainly of non-clinical hair-pullers who may not share the genetic burden of OCD. Although additional correlational analysis conducted on a subsample of individuals who meet the DSM-5 criteria for TTM also did not show a significant relationship between the MGH-HPS and OCI-R scores (see supplementary material for details), this analysis likely did not have enough power to detect such effects due to the small sample size (n = 23).

While the current study provides evidence for the validity of the Polish version of MGH-HPS, some methodological limitations should be mentioned. Firstly, the study sample involved mainly university students with varying severity of hair-pulling behaviors. Although some of the participants met the DSM-5 criteria for hair-pulling disorder, this subgroup was small and the diagnosis was based on self-reports rather than being confirmed in a clinical interview and medical examination of hair loss. As, we were not able to examine the psychometric properties of the Polish version of the MGH-HPS in the clinical group, the results of the ROC analysis need to be treated with some caution. Secondly, the participants were recruited and assessed via the internet. Applying this method of data collection limits control over whether the questionnaires have been completed reliably and in accordance with the instructions. Also using self-reported data forms to assess history of dermatological and psychiatric illness may not reflect the actual occurrence of co-morbid illness. The sample of individuals engaging in hair-pulling behaviors that was used to evaluate the construct validity of the MGH-HPS was relatively small. The appropriate sample size for CFA is a widely debated issue and many different rules have been formulated over the years (see: Kyriazos, 2018 for review). For example, ratios of cases to free parameters of 5:1, 10:1, and 20:1 have previously been suggested as optimal for running CFA (Bentler & Chou, 1987; Bollen, 1989; Jackson, 2003; Kline, 2016; Schumacker & Lomax, 2015). Monte Carlo studies (e.g., Wolf et al., 2013) have shown that – depending on the number of factors and indicators, indicator loadings, and strength of correlation between factors—the sample sizes required to achieve minimal bias, adequate statistical power, and the solution propriety of a given model in CFA may range from as small as 30 to almost 500. Given the non-normal distribution of the data, the results of the CFA (especially these concerning the two-factor solution) should be interpreted cautiously until replicated in a larger sample. Nevertheless, it should be noted that the minimum of 5 participants per estimated parameter (Bentler & Chou, 1987) ratio suggested by the literature as being sufficient for running CFA was exceeded in this study; moreover, on the basis of the previous stimulation study, N = 90 should be sufficient to test a one-factor model with more than 6–8 indicators and factor loadings > 0.50 using CFA (Wolf et al., 2013). Further studies on the Polish version of the MGH-HPS should also examine the test–retest reliability of the tool. Additionally, the scale’s responsiveness to changes in symptoms during the course of disease and its treatment should be investigated in a further longitudinal study. Also, it is worth examining whether the pen-paper and online versions of the scale yield similar results.

The Polish version of the Massachusetts General Hospital Hairpulling Scale (MGH-HPS; Keuthen et al., 1995) is a reliable and easily administered tool which may be used to measure hair-pulling severity, as well as resistance to and control over hair-pulling behaviors in adults with TTM symptoms. Although further studies on the scale’s stability and reliability in larger groups and clinical groups are still needed, the scale may be useful in the diagnosis of TTM and treatment planning, especially as there are no other validated tools for measuring TTM in the Polish language. Hopefully, an additional language version of the scale will also facilitate cross-cultural and epidemiological research on TTM.