Introduction
Emotional disorders (EDs) encompass various disorders that are associated with intense emotional states that cause interference in individuals’ everyday lives over a long period of time, such as anxiety and depression (Barlow et al., 2004), and have been reliably reported as a prevalent and alarming concern. Estimates of the Institute for Health Metrics and Evaluation (IHME, 2018) indicate that across European countries, the most common mental disorder is anxiety. An estimated 25 million people (5.4% of the population) suffer from anxiety, followed by depression, which affects 21 million people (4.5% of the population). By country, the estimated prevalence of mental health disorders is higher in Finland, the Netherlands, France, Ireland and Portugal (with rates of 18.4% or more of the population with at least one disorder). In Portugal, anxiety disorders (16.5%) and mood disorders (7.9%) are the most prevalent mental disorders. Depressive symptoms affect approximately 10% of the Portuguese population (Conselho Nacional de Saúde [CNS], 2019). Worldwide, the estimated global prevalence of major depressive disorders and anxiety disorders in 2020 was relevant (GBD 2019 Mental Disorders Collaborators, 2022) and increased after adjustment for the COVID-19 pandemic (Santomauro et al., 2021).
Empirical research has shown more commonalities than differences among EDs (Bernstein et al., 2010). In fact, in addition to their high prevalence, EDs present high rates of comorbidity, with a lifetime occurrence of up to 75% of anxiety and depressive disorders (Brown & Barlow, 2009). Taxometric studies also suggest that the structure of these disorders is dimensional rather than categorical for a variety of constructs, including depressed mood, worry, social anxiety, and somatic symptoms (Kliem et al., 2014; Olatunji et al., 2010). Taken together, these issues suggest that EDs may be also understood through a classification system that is focused on dimensional processes.
Considering some limitations of the categorical classification system, Brown and Barlow (2009) developed a 10-dimensional transdiagnostic model for EDs. The higher-order model incorporates two key genetically based core dimensions of temperament in the etiology and development of EDs: neurotic temperament, which reflects the tendency to experience a negative effect in response to subjectively threatening triggers (e.g., neuroticism, behavioral inhibition, negative affectivity), and positive temperament, which is the tendency to experience a positive effect in response to goal-oriented social activities (e.g., extraction, behavioral activation, positive affectivity). Neurotic and positive temperaments are part of the classification system because they are associated with the onset, severity, co-occurrence, and course of many EDs and related symptoms (Barnett et al., 2011; Brown et al., 1998).
The hybrid dimensional-categorical profile approach to EDs proposed by Brown and Barlow (2009) also includes eight lower-order transdiagnostic phenotypes to specify the nature of the problem and ease treatment planning. These transdiagnostic phenotypes are defining features of several EDs and include the following: depressed mood and mania dimensions, which capture excessive sadness and positive affect, respectively, that frequently co-occur with other EDs (Brown et al., 2001a, b; Rosellini & Brown, 2019); autonomic arousal, which is characterized by the experience of physiological symptoms due to sympathetic activation of the nervous system (i.e., panic symptoms); somatic anxiety, which is incorporated to reflect anxiety focused on somatic symptoms and associated worries about health; the dimension of social anxiety, which represents fear of negative evaluation in situations of interaction and performance (Rosellini & Brown, 2019); intrusive cognitions, which relate to the experience of uncontrollable thoughts, images and impulses; the dimension of traumatic reexperiencing, which can be defined as experiences of negative effect, dissociation, and flashback centered on past traumatic events; and finally, avoidance, which is defined as behavioral and cognitive strategies to prevent or reduce the intensity of acute states of negative or positive affect. These eight lower-order dimensions were selected because of evidence showing that neuroticism and extraversion alone would not provide adequate information about the foci of ED symptoms (Brown & Barlow, 2009). In fact, the system classification proposed by these authors, by suggesting new ways of understanding, is the only one that has implications for treatment planning and opens the door for new ways of assessment.
To date, several attempts have been made to conduct a preliminary evaluation of the validity of a profile approach to the classification of EDs. However, these proposals involved some limitations, and none of them could be used to assess all the dimensions included in Brown and Barlow’s (2009) approach. For example, the Research Domain Criteria (RDoC) project, launched by the National Institute of Mental Health (Insel et al., 2010), included five dimensions of psychological processes (negative and positive valence, social, cognitive and arousal/regulatory processes; the current version of this framework includes the sensorimotor domain as an additional domain) that contribute to mental health generally rather than EDs specifically. However, as noted by Boettcher et al. (2020), these dimensions and their subfacets were nonspecific and could not be translated into clinical practice. Another example is the approach of the Hierarchical Taxonomy of Psychopathology (HiTOP), proposed by Kotov et al. (2017). This taxonomy classifies all psychopathology in a multilevel framework of symptom dimensions ranging from higher-order spectra to narrow maladaptive processes. While HiTOP offers both specificity and comprehensiveness, Boettcher et al. (2020) suggested that there is no way (or may be exceptionally challenging) to assess the complete framework parsimoniously.
Existing measures also provide a limited assessment of the broad set of EDs dimensions delineated in the profile proposed by Brown and Barlow (2009). For example, the 42-item Beck Depression and Anxiety Inventories (Beck et al., 1996) or the 21-item Depression Anxiety Stress Scale (Lovibond & Lovibond, 1995) can only be used to assess depressed mood and autonomic arousal. In a similar way, many other measures are limited by the use of nonspecific constituent items. Even the 99-item Inventory of Depression and Anxiety Symptoms-II (IDAS-II; Watson et al., 2012), which was developed within the HiTOP paradigm, and the measures of personality and psychopathology developed to detect all major mental health conditions (e.g., the Minnesota Multiphasic Personality Inventory-2-Restructured Form, MMPI-2-RF) cannot be used to assess all dimensions in this profile (Rosellini & Brown, 2019). Accordingly, the only way to assess the wide range of empirically identified EDs traits and lower-order phenotypes would involve the application of numerous self-report questionnaires, which could be burdensome. Therefore, the need for a measure that allows rapid and effective assessment was addressed by the development of the Multidimensional Emotional Disorders Inventory (MEDI; Rosellini & Brown, 2019).
Considering the evidence that the dimensions proposed by Brown and Barlow (2009) are widely recognized by researchers and clinicians and particularly because of the solid support from theory and research (Rosellini & Brown, 2019), the need emerged to develop a measure that included and effectively assessed these transdiagnostic dimensions, thus conferring validity to the classification system proposed by Brown and Barlow (2009). With this in mind, Rosellini et al. (2015) developed a dimensional measure called the Multidimensional Emotional Disorders Inventory in which the authors analyzed EDs not by their symptoms but by the similarity in their psychological processes.
The MEDI is a 49-item measure that covers nine of the 10 dimensions proposed in the profile by Brown and Barlow (2009): neurotic temperament, positive temperament, depressed mood, autonomic arousal, somatic anxiety, social anxiety, intrusive cognitions, traumatic reexperiencing, and avoidance. This inventory is a novel and unique approach because it is intended to provide a brief but valuable assessment of temperament, personality processes and other transdiagnostic phenomena commonly found across EDs (Rosellini et al., 2015). The MEDI has some advantages. First, it was developed to be transdiagnostic in nature, which means that the items emphasize features of each phenotype that intersect multiple DSM disorder categories (Rosellini et al., 2015). Second, the MEDI is practical for psychopathological research and clinical practice. Most of its dimensions are also assessed by other measures but typically use 20 or more items. With this inventory, we can assess these dimensions more parsimoniously and briefly. In clinical settings, the MEDI can also provide the scores of each individual, which would allow the creation of a profile of each patient that indicates the main processes that maintain EDs and the design of more tailored interventions. Regardless of the ED diagnosis, assessing and identifying the transdiagnostic processes that maintain the disorder may lead to a more unique and personalized treatment plan (Barlow et al., 2004; Gallagher, 2017). With these advantages, the MEDI has the ability to provide a more efficient way of assessing the full range of EDs, including the common standards of comorbidity and subthreshold symptoms, thereby providing an optimal process and assessment tool for transdiagnostic treatment approaches such as the unified protocol (Rosellini et al., 2015).
The original study of the MEDI in a clinical sample supported the structure that encompasses the nine dimensions proposed by Brown and Barlow (2009) and showed its ability to assess traits and phenotypes associated with the development, expression, and maintenance of a range of EDs (Rosellini & Brown, 2019). All dimensions had acceptable composite reliability and large factor determinacy, which indicated acceptable validity of the factor/dimension scores. The correlations with other measures demonstrated strong correlations between the MEDI dimensions and their respective convergent validity measures (Rosellini & Brown, 2019). In addition to the original study, this inventory has been validated in Spain in both a community (Osma et al., 2021) and a clinical sample (Osma et al., 2023). The results of the Spanish validations (Osma et al., 2021, 2023) supported the original factor structure and presented adequate reliability indices (Cronbach’s alphas between 0.74 and 0.92 in the community sample and between 0.66 and 0.91 in the clinical sample) as well as strong validity evidence with relevant measures.
To the best of our knowledge, because no other validation studies of the MEDI have been conducted, the present study contributes to the literature by examining the validity and reliability of this measure in different cultures. The importance of validating the European Portuguese version of the MEDI is also justified by the significant prevalence of EDs in Portugal (CNS, 2019). By doing so, we can provide mental health clinicians with a brief and specific questionnaire that can measure the core transdiagnostic processes involved in the maintenance of emotional problems, which ultimately could contribute to more effective planning of treatments and better intervention outcomes.
Method
Participants
The sample consisted of 585 individuals from the community (365 women; M = 25.95 years, SD = 8.07; range: 18–63). Most participants reported being single (56.6%), not having children (89.1%), having university studies (79.1%), currently studying (46.3%) and living in urban areas (65%). The majority of participants did not report the presence of any physical (86.5%) or psychological problems (76.4%). Among the participants who reported the presence of psychological problems, 50 reported current psychological/psychiatric counseling. The detailed characteristics are presented in Table 1.
Table 1
Sociodemographic and clinical characteristics of the sample (N = 585)
n | % | |
|---|---|---|
Gender | ||
Male | 220 | 37.6 |
Female | 365 | 62.4 |
Marital status | ||
Single | 331 | 56.6 |
Married | 52 | 8.9 |
De facto union | 33 | 5.6 |
In a relationship (without living together) | 162 | 27.7 |
Divorced | 7 | 1.2 |
Education | ||
≤ 9 years | 7 | 1.2 |
10-12years | 115 | 19.7 |
University Studies | 463 | 79.1 |
Professional status | ||
Employed | 236 | 40.3 |
Unemployed | 44 | 7.5 |
Student | 271 | 46.3 |
Worker-Student | 33 | 5.6 |
Retired | 1 | 0.2 |
Residence | ||
Rural Area | 205 | 35.0 |
Urban Area | 380 | 65.0 |
Physical health problem | ||
Yes | 73 | 12.5 |
No | 506 | 86.5 |
No response | 6 | 1.0 |
Mental health problem | ||
Yes | 129 | 22.1 |
No | 447 | 76.4 |
No response | 9 | 1.5 |
Currently taking psychiatric medication | ||
Yes | 40 | 6.8 |
No | 544 | 93.0 |
No response | 1 | 0.2 |
Procedures
Participants were invited through the mailing lists of the researchers (e.g., personal contacts) and social networks (e.g., Facebook and Instagram) to participate in an online survey on EDs and the development of the European Portuguese version of the MEDI. Before conducting any study-related process, ethical approval was obtained from the Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Coimbra. After being informed about the main objectives of the study, the inclusion criteria (i.e., being Portuguese and age equal to or above 18 years), the composition of the research team, the role of the researchers and participants, and contacts for additional information, the participants were provided with the option to give their informed consent (i.e., clicking on the “I agree to participate in the study” option). Then, the participants were redirected to a set of questions about their sociodemographic and clinical information and a set of self-report questionnaires. The data collection was conducted between December 2020 and July 2021.
Measures
Sociodemographic and Clinical Data
Sociodemographic and clinical information were collected through a self-report questionnaire developed by the authors that included questions concerning age, gender, education, employment and marital status. Clinical data included questions about the existence of physical health and psychological problems, including if applicable, diagnosis, attendance at psychiatric or psychological appointments and medication.
Multidimensional Emotional Disorder Inventory (MEDI)
The MEDI (Rosellini & Brown, 2019) is a 49-item self-report questionnaire designed to assess the transdiagnostic dimensions included in Brown and Barlow’s (2009) approach to emotional disorder classification. The items are answered on a nine-point response scale ranging from 0 (Not characteristic of me/Does not apply to me) to 8 (Extremely characteristic of me/Applies to me very much). The nine dimensions assessed by the MEDI are neurotic temperament (5 items; e.g., “I get upset by trivial things”), positive temperament (5 items; e.g., “I am an optimistic person”), depressed mood (5 items; e.g., “I feel sad and blue”), autonomic arousal (5 items; e.g., “I have been experiencing rushes of fear that come on very suddenly”), somatic anxiety (5 items; e.g., “I am preoccupied by illnesses and diseases”), social anxiety (5 items; e.g., “I am uncomfortable mingling at social events”), intrusive cognitions (6 items; e.g., “Unpleasant thoughts, images, or memories come into my mind against my will”), traumatic reexperiencing (5 items; e.g., “I have disturbing dreams about awful events that occurred in my past”), and avoidance (8 items; e.g., “I will do almost anything to get rid of unpleasant feelings”). In the original study, all dimensions showed good convergent and discriminant validity with other well-established measures of ED symptoms as well as acceptable reliability (Rosellini & Brown, 2019).
Brief Symptom Inventory (BSI)
The BSI (Derogatis, 1993; Portuguese version by Canavarro, 2007) was used to assess psychopathological symptoms. The 53 items of the BSI are rated on a five-point response scale from 0 (Not at all/Never) to 4 (Extremely/A lot of times). The scores were obtained for nine primary symptom dimensions (somatization, obsession-compulsion, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, psychoticism and paranoid ideation) and three global indices of psychological distress (Global Severity Index [GSI], Positive Symptom Distress Index [PSD], and Positive Symptom Total [PST]). Higher scores denote higher levels of psychopathology. In the present study, Cronbach’s alphas for the dimensions ranged from 0.77 (hostility and psychoticism) to 0.88 (depression).
NEO-Five Factor Inventory (NEO-FFI)
The NEO-Five Factor Inventory (Costa & McCrae, 1992; McCrae & Costa, 2004; Portuguese version by Magalhães et al., 2014) is a reduced version of the Revised NEO Personality Inventory (NEO-PI-R). This instrument provides a concise measure of the five basic personality factors: neuroticism, extraversion, openness, agreeableness and conscientiousness. The NEO-FFI consists of 60 items (12 items for each dimension) and uses a five-point Likert-type format ranging from 0 (Strongly disagree) to 4 (Strongly agree). In the present study, only the neuroticism and extraversion subscales were used. In our sample, the reliability values were 0.76 for extraversion and 0.88 for neuroticism.
Posttraumatic Stress Disorder Checklist– Civilian Version (PCL-C)
The PCL-C (Weathers et al., 1991; 1993; Portuguese version by Marcelino & Gonçalves, 2012) is a 17-item questionnaire that assesses symptoms of PTSD in civilian populations. The items correspond to the criteria of DSM-VI-TR for PTSD: B (reexperience), C (avoidance) and D (hyperactivation). Each item is rated on a five-point response scale ranging from 1 (Not at all) to 5 (Extremely). Scores equal to three or higher indicate the presence of symptoms of PTSD. In our sample, the internal consistency was.94 for the total scale and 0.87 for criteria B, C, and D.
Acceptance and Action Questionnaire-II (AAQ-II)
The AAQ-II (Bond et al., 2011; Portuguese version by Pinto Gouveia et al., 2012) was used to assess experiential avoidance. The AAQ-II consists of 7 items rated on a seven-point response scale ranging from 1 (Never true) to 7 (Always true). In the AAQ-II, higher scores denote higher experiential avoidance. In the present study, Cronbach’s alpha was 0.92.
Data Analyses
Data analyses were performed using the Statistical Package for Social Sciences (SPSS), version 25.0 (IBM Corp., Armonk, NY). Item-level confirmatory factor analysis (CFA) was performed using the JASP software (version 0.19.3), with the weighted least squares mean-variance adjusted (WLSMV) estimator, as the MEDI items are ordinal (Rogers, 2024). To characterize the sample, descriptive analyses were performed. The distributional characteristics of items were examined by calculating the mean values (M) and standard deviations (SD), the percentage of missing values, floor and ceiling effects and skewness and kurtosis in the distributions of scores. The reliability of the MEDI was examined through Cronbach’s alpha and McDonald’s omega. For both indices values above 0.70 were considered acceptable (Dunn et al., 2013). The original nine-factor structure of the MEDI was examined with CFA. The models’ goodness-of-fit was assessed based on the maximum-likelihood χ2 statistic, the ratio χ2/degrees of freedom (df), the comparative fit index (CFI), the root mean square error of approximation (RMSEA) and the standardized root mean squared residual (SRMR). A model was considered to have a good fit when CFI ≥ 0.95, RMSEA ≤ 0.06 (p >.05) and SRMR ≤ 0.08 (Hu & Bentler, 1999). To assess the convergent validity, the nine MEDI dimensions were correlated with other measures that assess the dimensions proposed by Brown and Barlow (2009), such as the BSI and the PCL-C. Pearson correlations were computed to assess the associations between the MEDI dimensions and other relevant measures. Finally, to test whether the MEDI dimensions differentiated individuals at higher risk of developing EDs, based on the PSD index of the BSI, a multivariate analysis of variance (MANOVA) was performed. All tests were two-tailed, and a p value < 0.05 was defined as the cutoff for statistical significance.
Results
Distributional Characteristics of Items
Table 2 presents an overview of the distributional characteristics of the 49 items of the MEDI. A floor effect was detected in most items, with the exception of 12 items, including the five items of the Positive Temperament dimension (i.e., more than 15% of respondents obtained the lowest possible score on the response scale; Terwee et al., 2007). Ceiling effects were generally absent (only four items were equal to or above the criterion of 15%). Overall, no substantial deviations from normality were observed in the distribution of most items, considering a coefficient of absolute skewness > 2 and a coefficient of absolute kurtosis > 7 as reference values for samples with more than 300 participants (West et al., 1995), except for Items 26 and 43, which showed skewness > 2 (2.54 and 2.14, respectively).
Table 2
Descriptive statistics for items of the MEDI
Item (No.) | M | SD | Floor (%) | Ceiling (%) | Skewness | Kurtosis |
|---|---|---|---|---|---|---|
Neurotic temperament | ||||||
Easily Upset (1) | 2.77 | 1.80 | 14.4 | 0.5 | 0.28 | -0.67 |
Always been worrier (10) | 4.37 | 2.39 | 6.3 | 11.8 | -0.11 | -1.06 |
Poor stress coping (16) | 3.41 | 2.33 | 10.3 | 6.8 | 0.40 | -0.83 |
More keyed up than average (32) | 2.95 | 2.37 | 20.5 | 3.9 | 0.42 | -0.90 |
Feelings hurt easily (35) | 3.38 | 2.36 | 14.0 | 6.2 | 0.27 | -0.90 |
Positive temperament | ||||||
Easily laughs (2) | 5.83 | 1.79 | 0.5 | 21.4 | -0.74 | 0.14 |
Optimistic person (17) | 5.01 | 1.94 | 1.2 | 11.5 | -0.34 | -0.51 |
Cheerful and happy person (24) | 5.89 | 1.59 | 0.3 | 15.0 | -0.76 | 0.52 |
Always motivated (33) | 5.12 | 1.90 | 1.2 | 12.6 | -0.38 | -0.36 |
Satisfied when finishing jobs (36) | 6.72 | 1.45 | 0.3 | 39.5 | -1.37 | 2.23 |
Depressed mood | ||||||
Disappointed in self (3) | 2.50 | 2.24 | 25.1 | 1.5 | 0.64 | -0.65 |
Feel sad (11) | 2.49 | 2.25 | 25.3 | 2.6 | 0.66 | -0.58 |
Loss of interest (25) | 2.02 | 2.11 | 34.7 | 1.4 | 0.93 | -0.001 |
Nothing to look forward to (37) | 2.32 | 2.32 | 36.4 | 3.8 | 0.97 | -0.05 |
Life not worth living (43) | 0.97 | 1.79 | 66.0 | 1.2 | 2.14 | 4.06 |
Autonomic arousal | ||||||
Experiencing breathlessness (4) | 1.07 | 1.77 | 60.7 | 0.2 | 1.80 | 2.44 |
Feeling trembling/shaky (13) | 1.08 | 1.72 | 59.5 | 0.5 | 1.75 | 2.41 |
Sudden rushes fear (18) | 1.33 | 1.92 | 54.4 | 0.9 | 1.48 | 1.32 |
Felt dizzy/lightheaded/faint (26) | 0.70 | 1.51 | 74.0 | 0.5 | 2.54 | 6.21 |
High resting heat rate (44) | 2.51 | 2.45 | 32.3 | 3.8 | 0.63 | -0.81 |
Somatic anxiety | ||||||
Fears physical sensations (6) | 2.35 | 2.34 | 30.4 | 3.9 | 0.80 | -0.41 |
Worry about health (19) | 5.69 | 1.87 | 0.5 | 21.9 | -0.62 | -0.15 |
Preoccupied by illnesses (28) | 3.30 | 2.43 | 17.6 | 5.8 | 0.25 | -1.00 |
Closely monitor health (38) | 3.56 | 2.34 | 11.3 | 6.0 | 0.18 | -0.97 |
Believes has undiagnosed illness (45) | 1.69 | 2.30 | 49.2 | 2.6 | 1.29 | 0.47 |
Social anxiety | ||||||
Uncomfortable mingling (7) | 2.05 | 2.17 | 35.9 | 1.4 | 0.88 | -0.26 |
Uncomfortable center of attention (14) | 3.41 | 2.43 | 16.8 | 6.0 | 0.20 | -1.02 |
Anxious with strangers (22) | 2.31 | 2.18 | 28.4 | 1.9 | 0.71 | -0.49 |
Nervous when talking to others (41) | 2.28 | 2.10 | 26.0 | 1.5 | 0.74 | -0.34 |
Nervous in social situations (47) | 2.49 | 2.21 | 24.3 | 2.2 | 0.67 | -0.47 |
Intrusive cognitions | ||||||
Odd thoughts (5) | 2.47 | 2.43 | 28.2 | 3.9 | 0.76 | -0.61 |
Unpleasant thoughts/images (12) | 2.51 | 2.43 | 29.7 | 4.4 | 0.70 | -0.65 |
Inappropriate/nonsensical thoughts (21) | 1.90 | 2.28 | 41.9 | 3.2 | 1.12 | 0.24 |
Actions driven by thoughts (30) | 1.08 | 1.59 | 54.7 | 0.2 | 1.71 | 2.51 |
Unacceptable thoughts/images (40) | 1.19 | 1.85 | 54.2 | 1.5 | 1.85 | 2.93 |
Unrealistic fear of losing control (46) | 1.51 | 2.23 | 55.0 | 2.1 | 1.40 | 0.77 |
Traumatic re-experiencing | ||||||
Thinking about horrific experiences (8) | 1.86 | 2.06 | 35.7 | 1.4 | 1.11 | 0.42 |
Disturbing dreams of past events (20) | 1.32 | 1.98 | 53.3 | 2.2 | 1.70 | 2.20 |
Intrusive images of past trauma (29) | 1.47 | 2.05 | 49.7 | 1.7 | 1.47 | 1.28 |
Feels like reliving trauma (39) | 1.26 | 1.83 | 52.5 | 0.9 | 1.63 | 2.03 |
Distressed by trauma reminders (48) | 2.07 | 2.26 | 34.7 | 3.9 | 1.05 | 0.18 |
Avoidance | ||||||
Distraction coping (9) | 3.04 | 2.52 | 23.4 | 4.1 | 0.31 | -1.17 |
Avoids upsetting places/things (15) | 3.49 | 2.49 | 17.9 | 6.2 | 0.07 | -1.14 |
Carries protective objects (23) | 1.16 | 1.96 | 61.9 | 2.2 | 1.86 | 2.78 |
Gets rid of unpleasant feelings (27) | 3.67 | 2.56 | 15.0 | 9.6 | 0.12 | -1.17 |
Tries to suppress upsetting thoughts (31) | 3.77 | 2.23 | 8.5 | 4.8 | 0.04 | -0.91 |
Avoids feared objects (34) | 2.14 | 2.13 | 31.3 | 1.7 | 0.85 | -0.18 |
Routine actions taken to cope (42) | 2.49 | 2.35 | 29.7 | 2.9 | 0.61 | -0.74 |
Fears prevent day-to-day tasks (49) | 1.59 | 2.03 | 43.4 | 2.2 | 1.41 | 1.32 |
Correlations, Descriptive Statistics and Reliabilities of MEDI Dimensions
All correlations between the MEDI dimensions were in the expected directions (see Table 3). The MEDI dimensions of Neurotic Temperament and Positive Temperament were inversely correlated (r = −.23, p <.001). MEDI-Neurotic Temperament was positively associated with all phenotype dimensions (rs range = 0.57 to 0.63), whereas MEDI-Positive Temperament was inversely associated with all phenotype dimensions (rs range = − 0.08 to − 0.46). All the lower order phenotype dimensions had significant and positive associations (rs range = 0.41 to 0.80). The internal consistency of the MEDI dimensions, either considering the Cronbach’s α or the McDonald’s ω, varied between acceptable and excellent (α = 0.70 to 0.91), except for the dimension of Somatic Anxiety, in which the Cronbach’s α and the McDonald’s ω were slightly below 0.70 (α = 0.69; ω = 0.68).
Table 3
Scale correlations, descriptive statistics and reliabilities
MEDI Factor | NT | PT | DM | AA | SOM | IC | SOC | TRM | AVD |
|---|---|---|---|---|---|---|---|---|---|
Neurotic temperament | - | ||||||||
Positive temperament | − 0.21*** | - | |||||||
Depressed mood | 0.59*** | − 0.46*** | - | ||||||
Autonomic arousal | 0.58*** | − 0.23*** | 0.62*** | - | |||||
Somatic anxiety | 0.49*** | 0.02 | 0.29*** | 0.48*** | - | ||||
Intrusive cognitions | 0.63*** | − 0.25*** | 0.65*** | 0.69*** | 0.49*** | - | |||
Social anxiety | 0.57*** | − 0.31*** | 0.53*** | 0.49*** | 0.31*** | 0.49*** | - | ||
Traumatic re-experiencing | 0.59*** | − 0.22*** | 0.59*** | 0.65*** | 0.46*** | 0.80*** | 0.47*** | - | |
Avoidance | 0.59*** | − 0.08 | 0.52*** | 0.54*** | 0.51*** | 0.62*** | 0.50*** | 0.62*** | - |
Reliability (Cronbach’s α) | 0.79 | 0.70 | 0.85 | 0.81 | 0.69 | 0.87 | 0.91 | 0.89 | 0.74 |
Reliability (McDonald’s ω) | 0.80 | 0.71 | 0.85 | 0.82 | 0.68 | 0.87 | 0.91 | 0.89 | 0.74 |
Scale M (SD) | 16.87 (8.40) | 28.54 (5.88) | 10.09 (8.45) | 6.69 (7.17) | 13.94 (7.81) | 10.67 (10.01) | 12.54 (9.48) | 7.98 (8.46) | 21.35 (10.93) |
Confirmatory Factor Analysis (CFA)
In the CFA, the original 49-item model of the MEDI had an unacceptable fit to the data at the item level, with χ2(1091) = 3488.12.67; p <.001; CFI = 0.912; RMSEA = 0.061; 90% CI for RMSEA = [0.059-0.064]; SRMR = 0.070. The analysis of the modification indices suggested that three pairs of errors may be correlated (Item 27 - Item 31; Item 38– Item 28; Item 38– Item 19). Because these modifications were theoretically plausible, they were added to the model. After this procedure, the fit of the model increased, χ2(1088) = 2948.54; p <.001; CFI = 0.931; RMSEA = 0.054; 90% CI for RMSEA = [0.052-0.054]; SRMR = 0.065.
Convergent Validity
The convergent validity of the MEDI dimensions was assessed by examining the correlations between the nine dimensions and relevant measures of psychopathological symptoms, personality, posttraumatic stress and experiential avoidance. There were strong and significant correlations with the expected dimensions of the BSI (see Table 4), specifically between MEDI-Autonomic Arousal and BSI-Somatization (r =.74); MEDI-Autonomic Arousal and BSI-Anxiety (r =.74); and MEDI-Depressed Mood and BSI-Depression (r =.83). There were also moderate correlations between MEDI-Somatic Anxiety and BSI-Anxiety (r =.41) and between MEDI-Social Anxiety and BSI-Interpersonal Sensitivity (r =.53). Conversely, there were weak to moderate, negative and significant correlations between all dimensions of the BSI and the MEDI-Positive temperament. The two dimensions of personality, symptoms of PTSD and psychological inflexibility were moderate to strongly correlated with their convergent validity measure (see Table 5): MEDI-Neurotic Temperament and NEO-FFI-Neuroticism (r =.74); MEDI-Positive Temperament and NEO-FFI-Extraversion (r =.68); MEDI-Intrusive Cognitions and PCL-Total (r =.68); MEDI-Traumatic Re-experiencing and PCL-Total (r =.73); and MEDI-Avoidance and AAQ-II-Experiential Avoidance (r =.59).
Table 4
Differential associations of the MEDI dimensions with brief symptom inventory dimensions
MEDI Factor | SOM_BSI | OBC BSI | SI BSI | DEP BSI | ANS BSI | HST BSI | FOB BSI | PAR BSI | PST BSI | |
|---|---|---|---|---|---|---|---|---|---|---|
Neurotic temperament | 0.45*** | 0.56*** | 0.54*** | 0.53*** | 0.57*** | 0.51*** | 0.44*** | 0.43*** | 0.50*** | |
Positive temperament | − 0.16*** | − 0.34*** | − 0.34*** | − 0.42*** | − 0.26*** | − 0.24*** | − 0.20*** | − 0.25*** | − 0.35*** | |
Depressed mood | 0.44*** | 0.66*** | 0.66*** | 0.83*** | 0.58*** | 0.49*** | 0.45*** | 0.53*** | 0.67*** | |
Autonomic arousal | 0.74*** | 0.56*** | 0.50*** | 0.55*** | 0.74*** | 0.53*** | 0.58*** | 0.46*** | 0.58*** | |
Somatic anxiety | 0.38*** | 0.28*** | 0.25*** | 0.23*** | 0.41*** | 0.30*** | 0.38*** | 0.32*** | 0.29*** | |
Intrusive cognitions | 0.52*** | 0.55*** | 0.54*** | 0.58*** | 0.61*** | 0.56*** | 0.51*** | 0.47*** | 0.65*** | |
Social anxiety | 0.34*** | 0.51*** | 0.53*** | 0.45*** | 0.45*** | 0.32*** | 0.47*** | 0.35*** | 0.46*** | |
Traumatic re-experiencing | 0.52*** | 0.53*** | 0.54*** | 0.51*** | 0.59*** | 0.54*** | 0.53*** | 0.46*** | 0.58*** | |
Avoidance | 0.39*** | 0.50*** | 0.45*** | 0.43*** | 0.51*** | 0.40*** | 0.49*** | 0.41*** | 0.48*** |
Table 5
Differential associations of the MEDI dimensions with measures of convergent validity
MEDI Factor | NFFI Neuroticism | NFFI Extraversion | PCL-C Reexperience | PCL-C Avoidance | PCL-C Hyperactivation | PCL-C Total | AAQ-II Total |
|---|---|---|---|---|---|---|---|
Neurotic temperament | 0.74*** | − 0.31*** | 0.42*** | 0.52*** | 0.52*** | 0.54*** | 0.66*** |
Positive temperament | − 0.46*** | 0.68*** | − 0.18*** | − 0.33*** | − 0.22*** | − 0.28*** | − 0.38*** |
Depressed mood | 0.73*** | − 0.49*** | 0.42*** | 0.64*** | 0.53*** | 0.60*** | 0.74*** |
Autonomic arousal | 0.56*** | − 0.29*** | 0.55*** | 0.58*** | 0.59*** | 0.63*** | 0.62*** |
Somatic anxiety | 0.35*** | − 0.06 | 0.36*** | 0.37*** | 0.44*** | 0.43*** | 0.37*** |
Intrusive cognitions | 0.62*** | − 0.28*** | 0.58*** | 0.63*** | 0.62*** | 0.68*** | 0.72*** |
Social anxiety | 0.59*** | − 0.53*** | 0.29*** | 0.45*** | 0.40*** | 0.43*** | 0.51*** |
Traumatic re-experiencing | 0.58*** | − 0.26*** | 0.71*** | 0.66*** | 0.64*** | 0.73*** | 0.68*** |
Avoidance | 0.53*** | − 0.23*** | 0.46*** | 0.52*** | 0.50*** | 0.55*** | 0.59*** |
Known-Groups Validity
To assess the known-groups validity, we compared the nine MEDI dimensions considering the PSD index of the BSI (i.e., in the Portuguese version, a score equal to or above 1.7 is an indicator of a higher risk of EDs; Canavarro, 2007). The results indicated a significant multivariate effect, Pillai’s trace = 0.35, F(9, 508) = 29.88, p <.001, ηp2 = 0.35. The subsequent univariate analyses (see Table 6) showed significant differences in all dimensions of the MEDI. Individuals at higher risk of EDs presented significantly higher scores in all dimensions, most notably in Depressed Mood and Neurotic Temperament.
Table 6
Comparison of the MEDI dimensions considering the BSI cut-off for the likelihood of an emotional disorder
PSD < 1.7 (n = 353) | PSD ≥ 1.7 (n = 165) | |||
|---|---|---|---|---|
M (SD) | M (SD) | F(1, 516) | ηp2 | |
Neurotic Temperament | 13.96 (7.05) | 22.48 (7.47) | 158.01*** | 0.23 |
Positive Temperament | 29.44 (5.35) | 26.87 (6.35) | 22.97*** | 0.04 |
Depressed mood | 6.95 (5.84) | 16.12 (9.29) | 186.17*** | 0.27 |
Autonomic arousal | 4.23 (4.51) | 11.02 (8.50) | 141.03*** | 0.22 |
Somatic anxiety | 14.77 (6.48) | 20.10 (8.22) | 63.81*** | 0.11 |
Intrusive cognitions | 7.17 (7.13) | 16.81 (11.06) | 141.98*** | 0.22 |
Social anxiety | 9.98 (7.46) | 17.37 (11.00) | 80.34*** | 0.14 |
Traumatic re-experiencing | 5.20 (5.83) | 12.68 (9.52) | 121.18*** | 0.19 |
Avoidance | 17.99 (9.10) | 27.13 (10.82) | 100.29*** | 0.16 |
Discussion
In this study, we present the first results of the development and validation of the European Portuguese version of the MEDI, a self-report inventory specifically developed to assess the dimensions included in Brown and Barlow’s (2009) profile approach to the classification of EDs. Our main findings indicate the rather satisfactory reliability and validity of the MEDI and attest to its use in research and practice in Portugal. The reliability of the nine factors of the MEDI was generally very acceptable, with all Cronbach alphas above 0.70 as recommended (Nunnally, 1994), except for Somatic Anxiety, which was slightly below the.70 threshold (Cronbach’s α = 0.69; McDonald’s ω = 0.68). Our results are quite similar to other validations of this inventory (Osma et al., 2021, 2023), except for the Avoidance dimension, which in our study reached acceptable reliability (α = 0.74; ω = 0.74) and in the original (Rosellini & Brown, 2019) version was below the recommended value of 0.70. Nevertheless, despite minor differences, the reliability values were generally very acceptable.
The descriptive statistics of the 49 items demonstrate an absence of ceiling effects (percentages higher than 15%; Terwee et al., 2007), except for Item 19 and most items of the positive temperament dimension. However, for several items, there are relevant floor effects; that is, a significant proportion of participants answered in the lower anchor of the response scale. Floor and ceiling effects are population dependent (Hyland, 2003). Therefore, these floor effects are not surprising because our results are based on responses of a community sample, which is less likely to report an ED than a clinical sample. Notably, the ceiling effect in Item 19 (“worry about health”) may be due to the fact that the validation took place during the COVID-19 pandemic, which made people more aware of and vulnerable to health-related issues.
Regarding construct validity, overall, the CFA corroborated the proposed model of the MEDI. Although in our sample the model showed slightly lower results than those recommended in the literature (CFI >.95; RMSEA <.06; Hu & Bentler, 1999), the CFA conducted at the item level provided very promising results pertaining to the nine-factor structure proposed by Rosellini and Brown (2019). In our study, after correlating three pairs of errors that were theoretically plausible, namely, the correlation between Item 27 (“I will do almost anything to get rid of unpleasant feelings”) and Item 31 (“If something upsets me, I try very hard to not think about it”), Item 38 (“I pay close attention to my health because I am afraid of getting sick”) and Item 28 (“I am preoccupied by illnesses and diseases”), and the correlation between Item 38 and Item 19 (“I worry about my health”), the model significantly improved despite the CFI being below the conventional threshold. This is not particularly uncommon, as instruments with a larger number of items are less likely to achieve a suitable model fit by conventional standards (Marsh et al., 2004). However, it is important to note that the remaining indices were acceptable, particularly the lower RMSEA and the acceptable SRMR. These data should be replicated in future studies that may involve larger sample sizes and a clinical sample.
The intercorrelations between the dimensions assessed by the MEDI were of small-to-strong magnitude and were largely consistent with prior research and the original validation study (Rosellini & Brown, 2019) as well as Spanish validations in both community and clinical samples (Osma et al., 2021, 2023). Consistent with various studies (Brown et al., 1998; Watson et al., 1988) and predictably, the dimensions MEDI-Neurotic Temperament and Positive Temperament were inversely correlated. As expected, there were positive correlations between the MEDI-Neurotic Temperament and all seven lower-order phenotype dimensions and negative correlations between the MEDI-Positive Temperament and all lower-order dimensions. This evidence is congruent with the results of hierarchical structural models of traits and symptoms of EDs (Brown et al., 1998; Rosellini & Brown, 2011) and replicates the pattern of associations of previous validation studies.
Supporting the convergent validity, all nine MEDI dimensions were strongly or moderately correlated with their convergent self-report measures (rs > 0.50). One of the strongest correlations was found between MEDI-Depressed Mood and BSI-Depression (r =.83), as expected, since both self-report measures assess exactly the same dimension with similar items (e.g., Item 11 “I feel sad and blue” from the MEDI and Item 17 “feel sad” from the BSI). Support was also obtained for the convergent validity of other dimensions, such as MEDI-Autonomic Arousal and its moderate-to-strong correlations with BSI-Somatization and BSI-Anxiety (r =.74), which may be explained by the presence and similarity of items that assess the occurrence of physical symptoms. The dimensions that focused on anxiety (MEDI-Somatic Anxiety and MEDI-Social Anxiety) showed moderate correlations with anxiety measures (BSI-Anxiety and BSI Interpersonal Sensitivity, respectively). Despite being significant, compared to the other convergent self-report measures, these were the weakest correlations (both r =.53). Overall, this pattern of associations related quite well to the pattern observed in recent Spanish validation studies (Osma et al., 2021, 2023). However, future validation studies of the MEDI may benefit from the selection of more specific measures to correlate with the MEDI-SAS, such as the Social Interaction Anxiety Scale (SIAS), which was developed to reliably assess the construct of social anxiety.
Regarding the known-groups validity and because our sample was recruited in the community, we examined whether the MEDI was able to differentiate between individuals at risk of emotional distress and those who were not at risk (based on an index provided by the BSI). The results demonstrated an important ability of the MEDI to clearly distinguish between individuals with and without an increased risk of emotional distress, reinforcing the conclusions of the European Portuguese validation of the BSI regarding the PSD index (Canavarro, 2007) as well as the usefulness of the MEDI as a brief diagnostic screening tool that could facilitate the early detection of populations at risk for EDs.
Currently, the MEDI is the only self-reported measure that effectively assesses the traits and phenotypes associated with the development, expression and maintenance of several EDs. Despite interest in applying transdiagnostic approaches such as the Unified Protocol (Barlow et al., 2018), the MEDI can bring added value to clinicians who treat patients with EDs (and in some cases with comorbid EDs) as well as researchers who are interested in the nature and development of these disorders (Rosellini & Brown, 2019). Our results may also open new avenues for the study of classification systems, as the MEDI appears promising in balancing the advantages of both categorical and dimensional classification. Indeed, it enables the dimensional evaluation of patients while also allowing classification into categorical profiles based on specific phenotypes and vulnerabilities, as already demonstrated by Rosellini and Brown (2014). The validation of this measure in Portugal is also a relevant contribution to the field and allows the dimensions of EDs developed by Brown and Barlow (2009) to be assessed through a single measure, avoiding the need for participants to answer a long set of questionnaires. It should also be noted that existing questionnaires were not specifically designed to assess the constructs under the transdiagnostic approach, although they can assess constructs that overlap with the dimensions covered by the MEDI (Rosellini & Brown, 2019). Our validation study makes an essential contribution to the widespread use of the MEDI, which to date and to the best of our knowledge has only been published in two cultures (the US and Spain). The fact that it was applied in Portugal with promising evidence of reliability and validity may help to verify the universality of the transdiagnostic constructs in EDs, proving that this measure can be used in different contexts, cultures and populations.
Despite the important contribution of this study, there are some limitations that should be acknowledged. First, this study responds to one of the suggestions by Rosellini and Brown (2019) that validation should be done with a significant sample of participants from the nonclinical community. However, despite its advantage, our sample is also a limitation. The replication of this study in a clinical sample, particularly of patients with a diagnosis of EDs or other mental health problems, would be of particular value. Second, as part of the precision studies, we did not assess the instrument’s test-retest reliability. In future studies, it would be important to evaluate the temporal stability of the MEDI as well as its sensitivity to change in the context of treatment. The use of a convenience (self-selected) sample and online recruitment should also be noted. It is possible that these factors may have contributed to a higher incidence of younger individuals as well as individuals with higher education and internet access, which may have introduced biases that compromise the generalization of these results.
With the results of this study, we can conclude that the European Portuguese validation study of the MEDI provides solid support for this inventory and for its practical use in clinical settings. Our findings demonstrate that this instrument is an efficient and valid assessment of the nine dimensions and phenotypes of EDs proposed by Brown and Barlow (2009). In addition to this efficient assessment, the MEDI is a useful tool for research as well as for more general clinical purposes. For example, the strong support for convergent validity demonstrates that it is possible to use the MEDI to replace long (and multiple) questionnaires to assess the constructs covered by the MEDI. In addition, as noted by Rosellini and Brown (2019), its use would bridge the difficulty associated with decisions that must be made by mental health professionals regarding which self-report instruments to use since a measure such as the MEDI can assess a broad range of dimensions associated with EDs. Thus, the MEDI has important benefits for researchers, clinicians in mental health care and patients because it enhances standardized outcomes and decreases the amount of time spent on clinical assessments. Finally, as also underscored by Rossellini and Brown (2014), the confluence of dimensional indicators and related categorical profiles may also serve as a valuable complement to the categorical diagnosis, which still has strong roots in current healthcare systems.
Declarations
Ethical Approval
This project was approved by the Ethics Committee of the Faculty of Psychology and Educational Sciences of the University of Coimbra.
Conflicts of Interest
The authors have no conflicts of interest to report.
Consent To Participate
Informed consent was obtained from all participants of this study.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
