Massachusetts General Hospital Hairpulling Scale: Validity and Reliability Study of the Turkish Form

Abstract

Objective:

The aim of this study is to analyze the validity and reliability of the Turkish form of Massachusetts General Hospital Hairpulling Scale (MGH-HPS), which is used to measure the severity of Trichotillomania (TTM).

Methods:

Fifty patients diagnosed with TTM according to the DSM-5 diagnostic criteria and fifty healthy controls participated in the study. The participants were asked to complete a sociodemographic questionnaire, the MGH-HPS-TR, the Clinical Global Impression (CGI), the Beck Depression Inventory (BDI), the Beck Anxiety Inventory (BAI) and the Barratt Impulsiveness Scale (BIS-11). The construct validity and the criterion validity of the MGH-HPS-TR were determined by means of exploratory factor analysis (EFA) and confirmatory factor analysis (CFA), respectively. The reliability analysis of the MGH-HPS-TR was assessed by calculating the Cronbach’s α coefficient and the item total correlation coefficient. The values for the area under the curve (AUC), sensitivity and specificity were based on the ROC analysis.

Results:

AFA and CFA results indicated a single factor structure with 7 items explaining 82.5% of the variance. The item/factor loadings were satisfactory with the best fit indeces. Correlations were found between the scores on the MGH-HPS-TR and the other scales used for criterion validity analyses. The internal consistency and the item-total correlation coefficients of the scale were found to be satisfactory. Based on a cut of point of ≥ 9, the scale had high power for discriminating between the patient and the control groups and high sensitivity and specificity.

Conclusion:

This study showed that the MGH-HPS-TR can be used as a valid and reliable psychometric tool in Turkey.

INTRODUCTION

Hair pulling behavior is listed among the body-focused repetitive behaviors (BFRBs) such as nail-biting (onychophagia), cheek chewing and skin picking, with a population prevalence ranging from 3% to 13% (Selles et al. 2015, Siddiqui et al. 2012; Solley and Turner 2018). Hair pulling is considered to be pathological as the behavior becomes progressively repetitive, resulting in shame and apparent physical harm (Bohne et al. 2005).

Trichotillomania (TTM) is a psychiatric disorder characterized by pulling out of one’s hair, which results in hair loss and impairment in functioning, and its prevalence ranges from 0.5% to 2.5% (Grant and Chamberlain, 2016, Grant et al. 2020, Grzesiak et al. 2017).

TTM was not included in diagnostic categories until DSM-III-R, and it falls under the category of impulse control disorder in DSM-IV (APA-1994). The difficulty of resisting and controlling the urge to pull hair, the presence of extreme and involuntary rituals and the relief provided by hair pulling from the tension, restlessness and negative emotions indicate the compulsive nature of hair pulling behaviour and a common phenomenology with obsessive-compulsive disorder (OCD). Observations of freqent presence of OCD in family history and good response to clomipramine therapy has brought about the proposal to include TTM in the OCD spectrum (Ferrão et al. 2009, Grant et al. 2007, Hollander et al. 2005, Stanley et al. 1995, Swedo et al. 1989). Ultimately, TTM was included in the category of ‘OCD and Related Disorders’ in the DSM-5 (APA 2013).

The equivalent prevalence of TTM in females and males in early childhood changes with aging and a higher incidence is observed among females. Distribution of the onset age of TTM is bimodal by presenting at early childhood or in adolescence (Duke et al. 2010). Whereas the symptoms are generally transient with less need for treatment in the early-onset form, the late-onset form is characterized by more severe symptoms and higher rates of treatment resistance and comorbidities (Ricketts et al. 2019). The scalp is the most common hair pulling site, followed by the eyebrows, eyelashes and pubic hair (Grant and Chamberlain 2016). The factors triggering hair pulling behaviour can be sensory such as hair thickness, length, location and associated physical sensations, emotional by involving anxiety, boredom, tension or anger, or cognitive with judgements on the appearance of hair such as inacceptability of whitening (Grant et al. 2012). The experience of pleasurable feelings and elimination of negative emotions associated with pulling hair reinforce the behaviour causing repetitiveness (Duke et al. 2010). Patients tend to conceal the sites of hair loss by putting on make-up, changing hairstyle and wearing a wig or a scarf. Patients usually report feeling of shamefulness, reduced self-esteem, and social isolation by avoiding daily activities, spending time at home, the inability to take vacations or to be intimate with their partners (Flessner et al. 2008).

There are self-report and clinician assissted psychometric tools developed to support the diagnosis of TTM, measure the symptom severity and monitor the treatment outcomes. These assessment instruments include the Psychiatric Institute Trichotillomania Scale (PITS), the National Institute of Mental Health-Trichotillomania Symptom Severity Scale (NIMH-TSS), the National Institute of Mental Health-Trichotillomania Impairment Scale (NIMH-TIS), the Clinical Global Impression Scale (CGI), and the Massachusetts General Hospital Hair Pulling Scale (MGH-HPS), none of which have been adapted to the Turkish language as yet. The MGH-HPS, unlike the others cited, is a self-report scale with the advantage of rapid assessment of TTM severity. Developed by Keuthen et al. (1995) from Yale-Brown Obsessive Compulsive Scale (YBOCS) items and reported to have high validity and reliability, it is the most commonly used scale in reported studies.

TTM is usually overlooked and neglected by the psychiatrists and psychologists at the outpatient clinics with little knowledge of treatment algorithms (Woods et al. 2006). We believe that the use of a self-report scale such as the MGH-HPS would be fast and practical for determining TTM severity, following up the treatment response in outpatient clinics and improving the knowledge for diagnosis and treatment of TTM. In this study, we aimed to analyse the validity and reliability of the MGH-HPS after its adaptation to the Turkish language.

METHODS

Sample

This study included 50 patients who consulted the University of Health Sciences, Şişli Hamidiye Etfal Training and Research Hospital, Psychodermatology Outpatient Clinic between April 2017 and October 2019, and were diagnosed with TTM according to the DSM-5 criteria. Exclusion criteria comprised having been diagnosed with mental retardation, dementia or psychotic disorder according to DSM-IV-TR criteria and illiteracy. The control group consisted of 50 healthy individuals without a history of psychiatric disorder.

Data Acquisiton

The participants were asked to complete a purpose designed sociodemographic questionnaire, the MGH-HPS-TR, the Clinical Global Impression (CGI), the Beck Depression Inventory (BDI), the Beck Anxiety Inventory (BAI) and the Barratt Impulsiveness Scale (BIS).

The Sociodemographic Questionnaire (SDQ): This questionnaire was designed for the study to acquire data on participant age, gender, educational status, employment status, family history of psychiatric disorders, age of disease onset, disease duration, sites of hair pulling, the time spent pulling hair specified as (a) 1-, b) 16-30 , c) 31-60 , d) 1-3 hours or, e) ≥4 hours per day and the frequency of hair pulling specified as (a) 1-3, b) 4-6, c) 7-10, d) 11-20 or e) <20 times per day.

The Massachusetts General Hospital Hair Pulling Scale (MGH-HPS): The original MGH-HPS was developed as an 8-item scale by Keuthen et al. (1995), by replacing the questions measuring the “compulsions” and the “obsessions” in the YBOCS with those on “hair pulling behaviour’’ and “urge/impulse’’, respectively, The Cronbach’s alpha reliability coefficient the scale was 0.89, and the one-factor structure explained 88% of the variance. The authors demonstrated the validity and the reliability of the final 7-item and 5-point Likert-type self-report scale with a single dimensional structure. The scale questions the symptoms of the previous one week in relation to the frequency and intensity of the urge to pull hair, degree of controlling the urge to pull hair, the frequency of pulling behaviour, the degree of resistance to pulling behaviour, the degree of controlling the pulling behaviour and the distress caused by the pulling behaviour. The total score of the scale ranges between 0 and 28.

The test-retest reliability coefficient of the MGH-HPS investigated in 22 TTM patients was 0.97 (df=20, p<0.001), when MGH-HPS was shown to have good convergent validity with the PITS (r=0.63, df=24, p<0.001) and the CGI (r=0.75, df=24, p<0.001) and good divergent validity with the BDI (r=0.30, df=21, p>0.10) and the BAI (r=0.10, df=21 p>0.10) (O’Sullivan et al. (1995). The factor structure of the MGH-HPS was investigated in 990 participants on the internet and a two-factor structure was reported with items 1, 2, 4, and 7 included in the severity subscale as factor 1, and items 3, 5, and 6 in the resistance and control subscale as factor 2. The scales were found to be valid and reliable with a Cronbach’s α coefficient of 0.846 (Keuthen et al. 2007).

The Beck Depression Inventory (BDI): The BDI is a 21-item and 4-point Likert type self-report scale designed to determine the presence and severity of depressive symptoms. The validity and reliability of the BDI adapted to the Turkish langauge was demonstrated by Hisli (1989). The total score of the scale ranges between 0 and 63. Higher scores indicate increased severity of depressive symptoms.

The Beck Anxiety Inventory (BAI): The BAI is a 21- item 4-point Likert-type self-report questionnaire designed to differentiate anxiety symptoms from depression symptoms by questioning the common anxiety symptoms of anger, fear and fear of death. The reliability and validity of the Turkish language version of the BAI was shown by Ulusoy et al. (1998). Total score of the scale ranges between 0 and 63. High scores indicate the increased severity of anxiety symptoms.

The Clinical Global Impression Scale (CGI): The CGI rates disease severity on 7 points with 1=normal, not at all ill; 2=borderline mentally ill; 3=mildly ill; 4=moderately ill; 5=markedly ill; 6=severely ill; 7= being among the most extremely ill patients (Guy 1976). Based on the study conducted by Hougton et al. (2015), illness was rated, and CGI score was determined by considering the factors such as frequency of hair pulling, distress and impaired functionality caused by the illness, and need for social support.

The Barratt Impulsiveness Scale (BIS): This is a 30-item, 4-point Likert-type self-report scale with scores ranging from 30 to 120, developed to measure impulsiveness (Patton et al. 1995). Three subfactors of attentional impulsiveness, motor impulsiveness and non-planning impulsiveness are evaluated. Attentional impulsiveness represents intolerance to cognitive confusion and impatience, motor impulsiveness represents acting suddenly without thinking, and non-planning impulsiveness represents a lack of sense of the future. High scores indicate a high level of impulsiveness. The validity and reliability study of the BIS adapted to the Turkish language was carried out by Güleç et al. (2008).

Procedures

The study was approved by the document numbered ‘No: 1508’ issued by the Ethics Committee of University of Health Sciences, Şişli Hamidiye Etfal Training and Research Hospital. After explaining the purpose and design of the study, the informed written consents of the participants were obtained. For adaptation ıf the MGH-HPS to the Turkish languge, permissions of the developers of the scale were obtained via e-mail and the royalty payment was made to the relevant publisher. The scale was then translated to the Turkish language using the translation-back translation method. The scale items were translated from the English to the Turkish language by 2 psychiatrists with a good command of both languages to form the MGH-HP-TR, which was then translated back to English by a translator proficient in English and compared with the original form of the scale in the English language. After the evaluation of all recommendations, the items were edited, and the final form of the MGH-HPS-TR was decided upon. A pilot study carried out with 5 TTM patients with different clinical characteristics. It was observed that patients who only pulled eyebrows or eyelashes rated with “0” point the queries using the term “hair” which was subsequently replaced by “trichome/hair” to improve the clarity of the questions. This improved version of the MGH-HPS-TR was then completed together with the SDQ and the other psychometric tests by the TTM patients consulting the psychodermatology outpatient clinic.

Statistical Analysis

The SPSS 20.0 software for Windows was used for statistical analysis of the data. Reliability of the MGH-HPS-TR was tested by the item-total correlation and the Cronbach’s α coefficient, which was ≥0.70, indicating a good internal consistency. The factor structure of the scale was determined by confirmatory factor analysis (CFA) and exploratory factor analysis (EFA) which was expected to show factor loading of the items to explain at least 40% of the total variance of the scale (Büyüköztürk 2002). Data fit for factor analysis was analysed by the Kaiser-Meyer-Olkin (KMO) test, which should give a value higher than 0.60, and the Bartlett’s test on the Chi-square value which should show statistical significance. Model fit was evaluated on the AMOS 22 program. The goodness of fit for the model tested by CFA was evaluated with several fit indices including χ² /sd, RMSEA, CFI, NFI, NNFI (TLI), and SRMR with the values of χ² /sd=0-3, 0.0≤RMSEA≤0.05, 0.95≤CFI≤1, 0.95≤NFI≤1, 0.95≤NNFI≤1 0.0≤SRMR≤0.05 indicating perfect fit, and the values of χ²/sd=3-5, 0.05≤RMSEA≤0.1, 0.90≤CFI≤0.95, 0.90≤NFI≤0.95, 0.90≤NNFI≤0.95, 0.05≤SRMR≤0.08 indicating acceptable fit (Schumacker and Lomax, 2004). For validity of the scale, the Pearson correlation analysis was used when the assumptions were met for normal distribution, and the Spearman correlation analysis was used for non-normal distribution. For divergent validity, the mean MGH-HPS-TR scores of the patient and control groups were compared using the student t-test. ROC analysis was carried out to determine the scale’s cut-off point and the specificity and sensitivity in discriminating the patient and control scores. The p value of <0.05 was accepted to indicate statistical significance.

RESULTS

Sociodemographic characteristics of the participants

The mean age of the patient and the control groups were, respectively, 24.26 ±6.58 and 26.96 ±6.92 years, differing with marginal significance (t=2, p=0.048). The patient and the control groups consisted, respectively, of 41 (82%) and 38 (76%) females, and a gender based difference was not determined (χ²=0.542, p=0.461). The mean duration of education was 10.96 ±3.31 years in the patient group and 12 ±2.05 years in the control group. did not differ significantly (t=1.890, p=0.062) There were not significant intergroup differences on marital status (χ²=4.320, p=0.115), with 41 (82%) , 8 (16%) and 1 (2%) patients being, respectively, single, married and divorced in comparison to 34 (68%) single and16 (32%) married participants in the control group. The patient and control groups differed significantly with respect to employment status (χ²=12.522, p=0.006), with 19 (38%), 17(34%), 12 (24%) and 2 (4%) of the patients being, respectively,employed, unemployed, students and housewives, in comparison to, respectively,16 (32%), 5 (10%), 22 (44%) and 7 (14%) participants in the control group (Table 1).

Table 1.

Comparisons of the Sociodemographic Data of Patients Diagnosed with Trichotillomania and the Control Group

	Trichotillomania (n=50)	Control (n=50)	p value
Age	24.26 ±6.58	26.86±6.92	0.048
Gender (Female), n (%)	41 (82)	38 (76)	0.461
Marital status, n (%)			0.115
Single	41 (82)	34 (68)
Married	8 (16)	8 (16)
Divorced	1 (2)	-
Employment status, n (%)			0.006
Employed	19 (38)	16 (32)
Unemployed	17 (34)	5 (10)
Student	12 (24)	22(44)
Housewife	2 (4)	7 (14)
Education (Year)	10.96±3.31	12±2.05	0.062

Clinical characteristics of the participants

The numbers of the sites of hair pulling were one, two and three or more for 34 (68%), 13 (26%), and 3 (6%) of the patients. The sites of hair pulling were the scalp, eyebrows, eyelashes, beard, the genital area, chest, armpits and the legs, respectively, by 42(84%), 9 (18%), 6 (12%), 5(10%), 4 (8%), 2 (4%), 1 (2%) and 1(2%) of the patients. The frequencies of hair pulling per day varied as >20, 1-3, 7-10, 4-6 and 11-20 times/day for, respectively, 13 (26%),12 (24%), 10 (20%), 9 (18%) and 6 (12%) of the patients.

The time spent on hair pulling per day was >4 and 1-3 hours or 16-, 31-60 and 6-15 minutes among respectively 5(10%) and14 (28%) or 13(26%), 12 (24%) and 1(2%) of the patients. TTM onset age was 17.3 ±7.69 years, and the duration of TTM was 5.88± 4.98 years. Mean patient scores were 19.54 ±12.21 on the BDI, 18.4 ±14.2 on the BAI, 4.28 ±1.31 on the CGI, 70.18±13.48 for the BIS total scale and 28.84 ±8.93, 18.6 ±9.47 and 22.74±4.65, respectively, on the BIS attentional impulsiveness, motor impulsiveness and non-planning impulsiveness subscales (Table 2).

Table 2.

Clinical Characteristics and the Psychometry Scores of the Patients with Trichotillomania

	Trichotillomania (n=50)
	Mean	SD
Age of Onset	17.3	7.69
Duration of disorder, year	5.88	4.98
CGI	4.28	1.31
BDI	19.54	12.21
BAI	18.4	14.2
BIS-AI	28.84	8.93
BIS-P	22.74	4.65
BIS-MI	18.6	9.47
BIS-Total	70.18	13.48
	n	%
Which area/s of hair are you pulling?
Scalp	42	84
Eyebrows	9	18
Eyelashes	6	12
Bread	5	10
Genital area	4	8
Chest	2	4
Leg	1	2
Armpit	1	2
How many areas of hair are you pulling?
One	34	68
Two	13	26
Three and above	3	6
Pulling frequencies per day
1-3 times	12	24
4-6 times	9	18
7-10 times	10	20
11-20 times	6	12
>20 times	13	26
Time spent hair pulling per day
1-15 min	6	12
16-30 min	13	26
31-60 min	12	24
1-3 h	14	28
≥4 h	5	10

CGI: Clinical Global Impression (CGI)

BDI: Beck Depression Inventory

BAI: Beck Anxiety Inventory

BIS-Total: Barratt Impulsiveness Scale-11 BIS-AI: BIS-11 attentional impulsiveness

BIS-P: non-planning impulsiveness

BIS-MI: motor impulsiveness

Validity Results

Exploratory Factor Analysis (EFA): The results of the KMO and Bartlett’s tests, used to evaluate the data fit for factor analysis, were, respectively 0.90 and χ² =873.134 (p<0.001), indicating suitability for EFA, which in turn resulted in a one-factor structure that explained 82.5% of the variance with an eigenvalue of 5.774. The analysis showed a factor loading of 0.60 for each item, indicating a high factor loading (Table 3).

Table 3.

Results of Exploratory Factor Analysis on the MGH-HPS-TR

Items	Patients (Mean±SD)	Sample (Mean±SD)	Factor loadings
M1. On an average day, how often did you feel the urge to pull your hair?	2.48±1.23	1.42±1.48	0.930
M2. On an average day, how intense or ‘strong’ were the urges to pull your hair?	2.34±1.08	1.35±1.35	0.947
M3. On an average day, how much control do you have over the urges to pull your hair?	2.40±1.43	1.39±1.42	0.931
M4. On an average day, how often did you actually pull your hair?	2.08±1.23	1.17±1.31	0.908
M5. On an average day, how often did you make an attempt to stop yourself from actually pulling your hair?	2.22±1.17	1.33±1.39	0.844
M6. On an average day, how often were you successful at actually stopping yourself frompulling your hair?	2.80±1.24	1.57±1.61	0.914
M7. During the past week how uncomfortable did your hair pulling make you feel?	2.44±1.40	1.35±1.57	0.851
Eigenvalue	5.774
Variance (%)	82.5

Confirmatory Factor Analysis (CFA): The one-dimensional structure determined by EFA was tested by CFA. The results of the first analysis indicated a poor model fit (χ^{2 /}sd=6.199, CFI=0.917, NFI=0.904, NNFI=0.876, RMSEA=0.229 and SRMR=0.040). Based on the given modification recommendations, associating the errors in items 5 and 6 with each other resulted in values showing perfect fit, except for the result for RMSEA (χ² /sd=3.431, CFI=0.964, RMSEA=0.157, NFI=0.951, NNFI=0.942 and SRMR=0.03) (Table 4). Item loading values of the one-factor structure ranged between 0.762 and 0.971 (Figure 1).

Table 4.

Goodness of fit Values on the MGH-HPS-TR^*

Fit indices	Perfect fit	Acceptable fit	First analysis	Second analysis
χ²/sd	0-3	3-5	6.199	3.431
RMSEA	≤ 0.05	≤ 0.1	0.229	0.157
NNFI (TLI)	≥0.95	≥0.90	0.876	0.942
CFI	≥0.95	≥0.90	0.917	0.964
SRMR	≤ 0.05	≤ 0.08	0.040	0.03
NFI	≥0.95	≥0.90	0.904	0.951

^*Source: Schumacker ve Lomax, 2004

Figure 1.

Confirmatory factor analysis model of MGH-HPS

Criterion-Related Validity: The correlations between the responses given in the SDQ to questions on the “frequency of pulling hair per day” and the “time spent pulling hair per day” and the scores on the CGI, BDI, BAI and the BIS were analysed to evaluate the validity of the MGH-HPS-TR. On convergent validity analysis, the MGH-HPS-TR score showed statistically significant positive correlations with the CGI score (rho=0.580, p<0.001), frequency of pulling hair per day (rho=0.296, p=0.037) and the time spent pulling hair per day (rho=0.569, p=0.001). Weak positive correlations were determined between the MGH-HPS-TR score and the scores on the BDI (r=0.250, p=0.080), the BAI (r=0.225, p=0.115) and the BIS (r=0.116, p=0.422). For divergent validity, comparison of the mean MGH-HPS-TR score of the TTM patients (16.76 ±6.38) and the control group (2.4 ±4.78) indicated a statistically significant intergroup difference (t= -13.496, p<0.001). Based on the ROC analysis, AUC was 0.955 (95% CI=0.938-0.984, p<0.001) (Figure 2). Considering a cut-off point ≥9, the MGH-HPS-TR sensitivity and specificity of TTM diagnosis were, respectively 86% and 86% (Table 5).

Figure 2.

ROC curve of MGH-HPS

Table 5.

Cut-off points for the MGH-HPS-TR and the corresponding psychometric properties

Cut-off points	Sensitivity (%)	Specificity (%)	Odds Ratio
6	96	78	4.4
7	94	82	5.2
8	92	84	5.7
9	86	86	6.2
10	84	88	7
11	82	88	6.8
12	80	90	8
13	74	92	9.2
14	74	94	12.3
15	68	94	11.3
16	60	98	30

Reliability Results

The Cronbach’s α internal consistency coefficient of the MGH-HPS-TR was determined to be 0.961. Item-total correlation coefficients of all question items ranged between 0.794 and 0.924, indicating that the scale was highly reliable (Table 6).

Table 6.

Correlation coefficients of MGH-HP

Items	Item-total correlation	Cronbach’s alpha if item deleted item-total correlation
M-1 (Frequency of urges)	0.899	0.952
M-2 (Intensity of urges)	0.924	0.951
M-3 (Ability to control the urges)	0.902	0.952
M-4 (Frequency of pulling behavior)	0.871	0.955
M-5 (Resistance to pulling behavior)	0.794	0.96
M-6 (Ability to control the pulling behavior)	0.88	0.954
M-7 (Distress caused by the pulling behavior)	0.798	0.961

DISCUSSION

In this study analyses were made to determine the validity and reliability of the MGH-HPS-TR originally developed by Keuthen et al (1995). The single structured scale with seven items was found to be a valid and reliable psychometric tool after translation to the Turkish languge.

Firstly, investigating the factor structure of the scale was by EFA resulted in a one-factor structure which explained 82.5% of the variance with an eigenvalue greater than 1. Given that one-factor scales are expected to explain ≥%30 of the variance (Büyüköztürk, 2002), this result for the explained variance was sufficient as expected (Klein, 2013) with a factor loading of >0.32 for each item. The original analyses on the psychometric properties of the MGH-HPS with 8 items derived from the YBOCS had been carried out with 119 TTM patients. Subsequently, item number 7 on social effect was removed on grounds of having a factor loading of 0.41 with the Cronbach’s α coefficient being lowered to 0.39. EFA carried out with 7 items resulted in 2 subscales with eigenvalues of 7.06 and 1.24 for, respectively, factor 1 and 2; but as the internal consistency of factor 2 was low with a Cronbach’s α coefficient of 0.21, the single factor structure was selected. In agreement with the results of Keuthen et al. (1995), our study demonstrated the validity of the one-factor and seven-item version of the MGH-HPS scale after it was adapted to the Turkish language.

There are reports in the literature on the one-factor and two-factor structures for the MGH-HPS. CFA analysis on the MGH-HP scores of 635 patients consulting psychiatry and other clinics in Iran showed a good fit for the one-factor structure (GFI=0.92, RMSEA=0.060, AGFI=0.90) (Rabiei et al. 2013). Investigation of the psychometric properties of the MGH-HPS by Keuthen et al. (2007) on 990 participants who met the DSM-IV diagnostic criteria for TTM, resulted in a valid and reliable two-factor structure with items 1, 2, 4, and 7 included in factor 1 on the severity of hair pulling and items 3, 5, and 6 included in factor 2 on resistance to and control over hair pulling behaviour. The differences in the results reported by Keuthen in 1995 and 2007 might be attributed to the use of clinically based participants in the original study in comparison to the subsequent population-based study. In our study, the first CFA analysis showed that some of the fit indices were not acceptable for the statistical goodness of fit. On the basis of the relevant recommendations, the CFA was repeated after the errors in items 5 and 6 were associated with each other. This resulted in a perfect fit on the basis of all fit indices except the RMSEA index. The MGH-HP-TR item 5 measures the ability to resist pulling, and item 6 measures the ability to control pulling and it is thought that the participants might have perceived a similarity in the item expressions ‘’resistance to’’ and ‘’control over’’ the behaviour. Also, the individuals who scores low in item 5 on resisting the urge are expected to score low in item 6 on controlling the behaviour as these items are interrelated. Keuthen et al. (2007) had items 5 and 6 grouped under the same subscale of their model with two-factors. Moreover, patients scoring high in items 5 and 6 are believed to benefit more from cognitive behavioral therapy (CBT) as the primary intervention (Francazio et al. 2017). Therefore, the model with the 2-factor structure could be tested in future studies with larger participant groups.

The first-line therapies for TTM include habit reversal training and stimulus control (Jones et al. 2018) in being more effective behavioral interventions for particularly the patients who lack awareness of their hair pulling behaviour. Habit reversal training comprises, amongst others, awareness training, development of a competing response and building motivation. Awareness training can ensure focusing the attention on hair pulling behaviour by self monitoring and recording the frequency, intensity, the trigger and the associated sensation of the behaviour (Francazio et al. 2017). The MGH-HP design thus contributes to the self-monitoring method by allowing hair pulling episodes to be recorded.

Reliability based on Cronbach’s α coefficients of >0.70 and item-total correlations of >0.30 is considered to be statistically sufficient (Karakoç and Dönmez 2014). Our study found that the reliability of the one-factor structure of the MGH-HP-TR was at an acceptable level with Cronbach’s α coefficient determined as 0.961, which was reported by others to vary as 0.89 (Keuthen et al. 1995), 0.80 (Diefenbach et al. 2005), 0.846 (Keuthen et al. 2007) and 0.82 (Rabiei et al. 2013). Also, the test-retest reliability was found to be 0.97 (O’Sullivan et al. 1995).

Validity analyses within the scope of our study showed that MGH-HP-TR scores strongly correlated with the CGI scores and the patient responses to the queries in the SDQ on the “frequency of pulling hair per day” and “time spent pulling hair per day”. Others reported strong correlations of the MGH-HP score with the PITS and the CGI scores and weak correlations with the BDI and the BAI scores (O’Sullivan et al. 1995). Positive correlations were determined between the MGH-HP score and the scores on the Depression, Anxiety and Stress Scale (DASS-21), the Yale-Brown Obsessive Compulsive Scale Modified for Body Dysmorphic Disorder (BDD-YBOCS), and the Obsessive-Compulsive Inventory-Revised (OCI-R) (Rabiei et al. 2013). Comparison of the psychometric properties of the MGH-HPS and the clinician assisted and self-report scales for hair pulling behaviour demonstrated not only positive correlations between the MGH-HPS and the clinician assisted scales but also the higher internal consistency of the MGH-HPS (Diefenbach et al. 2005). We found the mean MGH-HPS-TR score of the TTM patient group to be significantly higher as compared to the control group. Also, the ROC analysis result for AUC was 0.966 (p<0.001), both results indicating a strong power for discriminating TTM patients from healthy individuals (Fan et al. 2006). The cut-off point of the MGH-HPS for differentiating patients and the healthy subjects was not previously reported. The cut-off point of ≥9 used in our study demonstrated 86% sensitivity and 86% specificity of the obtained MGH-HPS-TR scores. Siddiqui et al. (2012) and Solley et al. (2018) determined the prevalence of TTM in non-clinical participants by selecting the cut-off point of ≥17, based on the results of Keuthen et al. (2007) for the total mean MGH-HPS score of 17.25± 5.07 obtained with 990 population based participants. The spread of the general distribution of a parameter is determined by its standard deviation. Therefore, the use of the mean value alone as reference might have led to a methodological error in these studies, resulting in underestimation of TTM prevalence. A 35% reduction or a 7-point change in MGH-HPS total score were reported to be the best indicators of treatment response in clinical practice (Farhat et al. 2019). Decreases by 9 points or more in the MGH-HPS score of TTM patients was considered as improvement of the behaviour (Hougton et al. 2015). The use of MGH-HPS cut-off point can be a rapid and practical method for determining TTM prevalence in population- and clinical-based samples. Furthermore, the use of MGH-HPS may guide us determine whether or not the cases of the individuals consulting outpatient clinics with complaints of hair pulling are pathological, identify the patients needing treatment, select pharmacotherapy or psychotherapy as the initial treatment and follow-up the treatment and the response.

In reference to the limitations of this study, one can firstly cite the restriction of the participants to patients consulting an outpatient clinic. As TTM complaints are scarcely met in outpatient clinics, probabilities of having higher disease severity and comorbidity are increased among the consulting patients which would prevent the generalisation of the observed psychometric properties of the MGH-HPS-TR to the general public. Secondly, as the numbers of the patients were low in this study, the results should be interpreted with caution. Hence, the validity and reliability of the scale needs to be reassessed in a larger population-based sample.

In conclusion, the self-report MGH-HPS-TR with 7 items evaluating TTM symptoms within the period of the previous one week on the basis of the frequency and intensity of urges to pull, control over the urge, frequency of pulling behaviour, resistance to and control over the pulling behaviour and the distress caused by hair pulling behaviour. Our study demonstrated that the one-factor MGH-HPS-TR showed a significant power of differentiating TTM patients from healthy individuals and can be used as a valid and reliable psychometric tool in our country.