Temperaments and orthorexia nervosa: a cross-cultural study between Germany and Lebanon

The aim of this study was to explore the association between affective temperaments and orthorexic eating and whether temperament may explain cross-cultural differences in this behavior while considering the two dimensions of orthorexic eating, healthy (HeOr) and nervosa (OrNe). To accomplish this, 337 and 389 individuals were recruited in Lebanon and Germany, respectively. The brief version of the Temperament Evaluation of Memphis, Pisa, Paris and San Diego explored depressive, hyperthymic, cyclothymic, anxious and irritable temperaments, and the Teruel Orthorexia scale explored orthorexic eating. HeOr appeared comparable between countries but OrNe was higher in Lebanon. In terms of affective temperaments, the higher the depressive, cyclothymic, irritable, and anxious temperaments, the higher were the levels of OrNe. Only the hyperthymic temperament scale was positively associated with HeOr. Three-step regression analysis indicated only gender as a unique predictor for HeOr. By contrast, gender, depressive, hyperthymic, and anxious temperament as well as the two-way interactions country*depressive temperament and country*hyperthymic temperament were significant predictors of OrNe. The positive association between OrNe and depressive temperament was only found for the German sample while the negative association between hyperthymic temperament and OrNe was somewhat stronger in the Lebanese sample. Overall, a higher healthy interest in diet was linked to the hyperthymic temperament. Findings emphasized the role of temperaments in pathological orthorexic eating in general as well as in explaining cross-cultural differences in these behaviors. The assessment of temperaments could help to fit treatments for eating pathologies to individuals from different cultures, focusing interventions more on these aspects.


Introduction
Eating behaviors have always been and are still profoundly influenced by social, environmental and cultural factors the compulsive search for pure and natural foods (Daniele et al., 2016).The concept of Orthorexia Nervosa (OrNe), defined as a persistent fixation on healthy eating (Koven & Abry, 2015), has been developed in this temporal and cultural context.Key elements of Orthorexia Nervosa are "(a) obsessional or pathological preoccupation with healthy nutrition; (b) emotional consequences (e.g.distress, anxieties) of non-adherence to self-imposed nutritional rules; and (c) psychosocial impairments in relevant areas of life as well as malnutrition and weight loss" (Barthels et al., 2019).Although OrNe's distinction from established diagnoses and therefore its classification in the Diagnostic and Statistical Manual of Mental Disorders are still under debate (Moroze et al., 2015), several proposals have been made to consider it as a distinct subtype of eating disorders (Busatta et al., 2021) or to place it in the obsessive-compulsive disorder spectrum (Bundros et al., 2016).However, the current state of the research does not allow the categorization of OrNe as a distinct mental disorder beyond established eating disorders (Cuzzolaro & Donini, 2016;Meule & Voderholzer, 2021), but rather a cultural phenomenon and social trend in societies where healthy eating is emphasized (Neliubina et al., 2020).In addition to the pathological aspects, of which the clinical significance remains unresolved (Strahler & Stark, 2020), there is also a healthy part of orthorexic eating, termed Healthy Orthorexia and defined as a "healthy interest in diet, healthy behavior with regard to diet, and eating healthily as part of one's identity" (Depa et al., 2019).
Despite the increase in research on this subject in recent years (Barnes & Caltabiano, 2017), the literature on OrNe is still largely descriptive (Koven & Abry, 2015).However, epidemiological data of OrNe is broadly influenced by the sensitivity and the specificity of the used screening tool.Here, the instruments and findings differ markedly which also reflects in widely varying prevalence numbers between < 1% and > 85% (Opitz et al., 2020).Due to the psychometric challenges in the assessment of OrNe, it is also not yet possible to conclude whether there are socio-cultural differences in prevalence and phenomenology of orthorexic eating.One of the newest tools, the Teruel Orthorexia Scale (TOS) was developed to overcome the questionable psychometric properties of the former tools (Koven & Abry, 2015) and to take the two sides of orthorexic eating into account (Barrada & Roncero, 2018).In this instrument, orthorexic eating is described on a pathological and a healthy dimension without a cut-off beyond which the behavior is considered pathological.Therefore, the correlates of OrNe and healthy interest in diet can be investigated separately.
Sociodemographic characteristics have been associated with OrNe (McComb & Mills, 2019;Strahler & Stark, 2019).Many factors contribute to the development of pathological eating including physiological, genetic, environmental and behavioral ones (Atiye et al., 2015).Findings concerning age and socioeconomic status (SES) are inconsistent, with studies showing either no significant or small negative correlations between age and OrNe (McComb & Mills, 2019), and studies showing either positive, negative or no relation between objective SES measures and orthorexic eating behaviors (Dell'Osso et al., 2016;Dittfeld et al., 2017;Missbach et al., 2015).In addition, evidence highlighted the significant relationship between marital status and OrNe (Eriksson et al., 2008;Fidan et al., 2010); interpersonal skills, such as discussions and conflict with the marital partner, may be an additional factor in triggering and maintaining the eating disorder (Bussolotti et al., 2002).Whether and how BMI is related to OrNe is also a matter of debate.Currently, a U-shaped relationship seems likely, with a higher risk of OrNe associated with both overweight and underweight (Strahler & Stark, 2019).In terms of gender, tools capturing the healthy interest in diet showed no gender differences, while women seemed to be only slightly more affected by OrNe compared to men (Strahler, 2019).Sociocultural influences may be a significant contributor to differences between men and women with culture-dependent differences in personality being a proposed mechanism (Costa Jr ,et al., 2001).
In terms of personality, there are some recent studies indicating a positive correlation between orthorexic eating and certain personality traits such as neuroticism and conscientiousness (Gleaves et al., 2013) or a negative correlation with agreeableness on the other hand (Strahler et al., 2020).In addition, temperament and character traits have been related to orthorexic tendencies.Although definitions vary in the literature, temperament may be defined as a behavioral style and an emotional expression, which can be observed from infancy onwards (McAdams & Olson, 2010).It refers to fundamental and early-appearing dispositional attentional, emotional and motor processes as well as reactivity (Rothbart, 2001).Furthermore, temperament is considered to be a precursor of personality traits found in adolescents and adults, and is associated with other psychological variables such as attachment, self-regulation and attentional systems (Rothbart & Ahadi, 1994).Although related, temperaments are fundamentally different from personality traits.Personality traits develop through the interaction between the individual's constitutionally based temperament, experiences provided by the environment and socialization (Rothbart & Ahadi, 1994).Hence, temperament may serve as a term for personality traits that affects the "three A's of personality": Affect, Activation, and Attention (Rothbart & Bates, 2006).
To evaluate temperament qualities, several models have been proposed.Arising from different perspectives on personality dimensions, Cloninger proposed the biopsychosocial model of temperament (Cloninger et al., 1993).This seven-dimensional model was constructed of four temperament dimensions and three character dimensions.The Temperament and Character Inventory (TCI) was developed to measure these dimensions.Akiskal et al. (1989) conceptualized temperament traits as dimensional emotional dispositions and reactions, from which affective disorders can arise.Five temperament types were proposed: depressive, cyclothymic, hyperthymic, irritable and anxious temperaments.The cyclothymic temperament is characterized by chronic cycling between mood polarities and unstable self-esteem and energy; the hyperthymic temperament by increased energy and optimism (Akiskal & Akiskal, 1992); the irritable temperament by irritable and angry behavior; the anxious temperament by a tendency to worry; and depressive temperament by low levels of energy, introversion and worrying (Akiskal & Akiskal, 2005).This model was thus based on a clinical and psychiatric tradition in the approach to temperament and resulted in the development of the Temperament Evaluation of Memphis, Pisa, Paris and San Diego (TEMPS-A), an auto-test to be filled by participants with the purpose of evaluating affective temperaments (Akiskal et al., 2005).
In eating disorder research, understanding the connections with temperament is a significant question and various models of personality and temperament have been applied to investigate it (Fassino et al., 2004).In fact, eating disorders and affective disorders occur with increased comorbidity (McDonald et al., 2019;Swift et al., 1986), but there seem to be differences between the types of disordered eating in how they relate to affective temperament.For instance, bingeing in contrast to restrictive disordered eating, seems to be associated with depressive, hyperthymic and cyclothymic, but not with irritable temperament (Ramacciotti et al., 2004;Rybakowski et al., 2014).Employing Cloninger's conceptualization of temperament, a meta-analysis showed that eating disorder pathology was related to the temperament traits persistence, harm avoidance, and/or novelty seeking (Atiye et al., 2015).In light of these results highlighting temperaments' role in eating disorders, the relationship between OrNe and temperament is also examined.So far, only the TCI has been used in OrNe research.While in one study, the risk of OrNe was positively related to low persistence (Gramaglia et al., 2019), orthorexic eating was positively correlated with harm avoidance and transcendence, as well as negatively correlated with self-directedness in another study (Kiss-Leizer & Rigo, 2019).To complement this research and in order to foreground the time-stable, dimensional and possibly maladaptive character of affect within temperament, understanding OrNe in the context of conceptualization by Akiskal et al. is beneficial.This may be of particular relevance, as OrNe is suggested to serve as a coping strategy for negative emotional states and anxiety (Barthels et al., 2021b;Bona et al., 2021).
So far, the association between socio-cultural factors, temperament and orthorexic eating is poorly understood.It seems comprehensible that different social norms, models of parenting and conceptualizations of desirable traits lead to cross-cultural variation in temperament subsequently increasing risk of OrNe, as it has been found for eating disorders.For example, the interaction between children and parents within the family environment was related to the development of pathological eating behaviors previously (Gonçalves et al., 2021).Furthermore, cultural and personal factors such as beliefs, goals and habits either increased or decreased the probability of adaptive healthy eating in a recent study (Goukens & Klesse, 2022).Similarly, perfectionism was found to promote pathological eating behaviors (Pamies-Aubalat et al., 2022).A cross-cultural comparison of Italian, Polish and Spanish university students showed that low/medium persistence was associated with orthorexic eating in the whole sample (Gramaglia et al., 2019).Country-specific associations were not reported in this study.Those, however, appear likely as affective temperaments are differentially distributed within different countries (e.g., Korea, Argentine, Germany, Lebanon, Portugal, and Hungary) (Gramaglia et al., 2019).While temperament research has often been conducted in Western countries, researchers have progressively explored cross-cultural temperament differences to better understand the role of culture in shaping individual differences (Rothbart, 1981).Due to the large influence of culture, intercultural research aims to identify the environmental factors that contribute to temperament variability and to improve understanding of the factors that may limit the generalization of the results obtained in a given population.In addition, the long-time goal of orthorexic eating behavior, i.e., achieving and maintaining health, implies not only consequences for the individual, but also for the society and its health care provision.
Following this, the aim of this study was to advance our knowledge of the association between temperament and orthorexic eating, and whether temperament may explain cross-cultural differences in this behavior while considering both dimensions, healthy and pathological orthorexic eating.In a first step for achieving this aim, we compared a sample from Germany to a sample from Lebanon.This was deemed appropriate due to the cultural differences in religious eating practices but also due to the above-mentioned distribution of temperaments.Specifically, the following questions were addressed: 1. Are there differences in pathological and healthy interest in diet between Lebanon and Germany?from 0 (completely disagree) to 3 (completely agree).Internal consistencies (Cronbach's α) of both subscales were high in the present study (TOS-healthy orthorexia 0.85, TOS-OrNe 0.90).
Temperament Affective temperament traits were assessed using the brief German and Lebanese version of the Temperament Evaluation of Memphis, Pisa, Paris, and San Diego (TEMPS), respectively.This brief version contains 35 instead of 110 items.The German version (TEMPS-M) was previously validated showing adequate to good internal consistencies and test-retest reliability (Erfurth, Gerlach, Hellweg et al., 2005).A version of the TEMPS in the Arabic language was also previously validated for the Lebanese population, with good psychometric properties (Karam et al., 2005).The 35 self-rating items can be assigned to 5 subscales: depressive (i.e.tending towards rigid thinking, selfaccusation, and shyness), cyclothymic (i.e.being moody and changeable, tending towards superficial thinking and intense emotion), hyperthymic (i.e.being strongly extroverted and expansive), irritable (i.e.showing higher energy and anger, but on the other hand a lower level of empathy, and dissatisfaction), and anxious (i.e.tending towards worry, ruminate, and continuous tension).All responses are scored on 6-point Likert scales ranging from 1 (not at all) to 5 (very much).Subscale scores range from 5 to 35, with higher scores denoting higher expressions of the respective temperament.All five subscales achieved very good or good internal consistencies in the present study (depressive 0.88, cyclothymic 0.85, hyperthymic 0.79, irritable 0.87, and anxious 0.84).

Sociodemographics
The survey was complemented by the collection of the participant's age (years), gender (male, female), marital status (married, not married), educational level (illiterate, primary education, secondary education, university studies), and subjective social status (SSS, MacArthur ladder) (Adler et al., 2000).Information on height and weight were collected to examine Body Mass Index (BMI, kg/m 2 ).

Data analysis
Multivariate analyses of co-variance (MANCOVA) were conducted to examine the main effects of country and gender as well as the country x gender interaction effect on the TOS subscales and the TEMPS-M subscales, respectively.These analyses included age, marital status, educational level, and SSS as covariates of no interest.There were no outliers (Box plots), no indication of a pronounced multicolinearity (all r < .50),there was linearity between variables 2. To what extent do depressive, cyclothymic, hyperthymic, irritable and anxious temperaments correlate with orthorexic eating?3. Is there a difference in these correlations between the healthy dimension of orthorexic eating as compared to the pathological dimension of orthorexia nervosa? 4. Do temperaments underlie differences in orthorexic eating between Lebanon and Germany?

Procedure and participants
Lebanese participants were recruited among visitors of seven community pharmacies, randomly chosen from a list of the official pharmacists' association in Lebanon.Visitors were approached and asked to participate in the study.Data was collected through face-to-face interviews and paper questionnaires.In Germany, the study was advertised using the universities mailing list and the survey link was made available on numerous social network sites.Participants were asked to complete the online survey using their own computers or smartphones (Platform: SosciSurvey.com).
Inclusion criteria for study participation were (a) being 17 years of age and older, (b) to clearly assign themselves to the male or female gender, and (c) to complete the data.Subjects participated voluntarily and provided informed consent prior to data collection (in written form in Lebanon, via button click in the online survey in Germany).Participating university students could receive course credit for participation.There was no other form of compensation.
The study was approved at both study sites by the local ethic committees.
Complete data was collected from 337 subjects from Lebanon (188 women, mean age total sample = 35.54years, SD 13.53) and from 389 subjects from Germany (314 women, mean age total sample = 27.17years, SD 10.64).

Measures
The following tools were employed to assess orthorexic eating behaviors, temperament, and sociodemographics.
Orthorexic eating behavior To measure orthorexic eating and to distinguish between the two proposed variants, healthy and nervosa, the 17-item Teruel Orthorexia Scale was employed (TOS).Nine items are summarized to form the TOS-healthy orthorexia subscale, indicating a healthy interest in diet and eating healthily only as part of one's identity.The other eight items form the OrNe subscale, TOS-OrNe.All items are rated on 4-point Likert scales ranging

Results
The final Lebanese sample consisted of 188 women and 149 men ranging in age from 17 to 70 years.The gender distribution among the German sample was unbalanced.Three-hundred-fourteen women and 75 men could be enrolled at the German study site ranging in age from 18 to 70 years.Table 1 shows both samples' descriptive characteristics.In terms of age and BMI, there was a main effect of gender and country (age: F gender [1,722] = 9.81, p = .002,η 2 = 0.013; F country [1,722] = 51.75, p < .001,η 2 = 0.067; BMI: F gender [1,698] = 25.05,p < .001,η 2 = 0.035; F country [1,698] = 10.58,p = .001,η 2 = 0.015) but no country x gender interaction effect (both F < 1, p > .450).The Lebanese sample and the male participants were older and had a slightly higher mean BMI level.In terms of marital status, there was no difference between men and women within countries (both X 2 < 1, p > .350)but more Lebanese subjects as compared to German subjects reported to be married (X 2 = 4.93, p = .026,V Cramer = 0.082).The educational level was higher in the German sample as compared to the Lebanese sample (X 2 = 45.51,p < .001,V Cramer = 0.250).In the Lebanese sample, women reported a higher educational level as compared to men (X 2 = 14.79, p = .002,V Cramer = 0.209).By contrast, men and women's educational level was comparable in the German sample (X 2 < 1, p > .350).Subjective social status was rated higher in the German sample (F[1,722] = 31.86,p < .001,η 2 = 0.042); there was no gender difference and the country x gender interaction appeared not significant (both F < 2.0, p > .160).

Differences among study variables by country and gender
Means and standard deviations of TOS and TEMP-M scores depending on country and gender are shown in Table 2. Results of the 2 × 2 MANCOVA revealed main effects of (scatter plots), but the normal distribution assumption (all Shapiro-Wilk p < .05,except TEMPS-M hyperthymic subscale) and the homogeneity of error variances assumption (both TOS subscales and the cyclothymic and irritable subscale of the TEMPS-M p < .05)were partially violated.Since MANOVA is considered relatively robust to violations of the normal distribution (Finch, 2005) and variance equality (Ates et al., 2019), the analysis proceeded without countermeasures.
Pearson's correlation was used to examine zero-order correlations between orthorexic eating and temperament.Following this, hierarchical linear regression models were run to investigate whether the affective temperaments as well as country and gender accounted for unique variance in orthorexic eating (Model 1: TOS-healthy orthorexia; Model 2: TOS-OrNe), and whether country moderated the association between temperament and orthorexic eating.Due to the non-normality of all TEMPS-M subscales (all Shapiro-Wilk < 1.0, p < .001),z-standardization was used to perform normalization.None of the predictors correlated > 0.7 indicating absence of multicollinearity.This was confirmed by variance inflation factor values below 10 for all predictors.In addition, Cook's distance did not identify significant outliers (0.052 for TOS-healthy orthorexia, 0.063 for TOS-OrNe).Finally, normal Predicted Probability plot indicated that residuals were normally distributed and plotting the predicted values and residuals on a scatterplot indicated homoscedasticity.Thus, all assumptions for linear regression were met.The first step included age, marital status, educational level, and SSS as covariates as well as gender and country as predictors of interest.The second step included the TEMPS-M subscales.The two-way interactions of gender x country and temperament x country were entered in the third step.All analyses were run with SPSS Version 23 (IBM Statistics) and p values < 0.05 were considered significant.among continuous study variables indicated TOS-healthy orthorexia to be hardly related to temperament.High levels on the hyperthymic temperament scale were positively associated with a healthy interest in diet.The other correlations were not significant.The opposite pattern appeared for the TOS-OrNe subscale.Here, all temperament subscales, except the hyperthymic, demonstrated a small to medium positive association; the higher the depressive, cyclothymic, irritable, and anxious temperament the higher the levels of orthorexia nervosa.
After adjusting for the effects of sociodemographic variables, gender (β = 0.13, p = .001),hyperthymic (β = 0.15, p < .001)and irritable temperament (β = -0.11,p = .013),but not country or the other temperament subscales were found to be unique predictors of TOS-healthy orthorexia (Step 2 ΔR² = 0.05, p < .001).The final step of the regression revealed only gender (β = 0.25, p = .036;Step 3 ΔR² = 0.01, p = .320)as a unique predictor of healthy orthorexia without any two-way interaction effects between country and temperament.The final model explained 7.1% of the variance of a healthy interest in diet.

Discussion
The present study aimed to determine the possible role of affective temperaments in orthorexic eating behaviors, and whether their expression contributes to cross-country differences in OrNe and the healthy interest in diet, termed healthy orthorexia.In accordance with previous reports (Strahler et al., 2020), present findings from two convenience samples in Lebanon and Germany suggest comparable healthy orthorexia levels but higher orthorexia nervosa levels in the Lebanese sample.Results further showed that the healthy interest in diet was positively associated with the hyperthymic temperament but none of the other affective temperaments.All other temperaments, in turn, were positively associated with OrNe.Hierarchical regressions quantified the contribution of temperament (together with gender, country, and their respective interactions) to healthy orthorexia with about 7.1% explained variance and to OrNe with about 22.7% explained variance.In terms of healthy orthorexia, however, slope parameters of the temperament subdimensions did not reach significance.For OrNe, slope parameters of depressive, hyperthymic and anxious temperament reached significance.The depressive and the hyperthymic temperament moderated differences in OrNe between the Lebanese and the German sample.Higher depressive temperament determined OrNe in Germany but was unrelated to this pathological interest in healthy diet in Lebanon.The connection between hyperthymic temperament and OrNe, on the other hand, was somewhat stronger in Lebanon.

Associations between affective temperaments and orthorexic eating
In our study, the healthy orthorexia dimension was not significantly related to the single dimensions of affective temperaments, though the regression model significantly improved.Albeit with a small effect, bivariate correlations showed higher levels on the hyperthymic temperament scale to be positively associated with a healthy interest in diet.This fits with the literature as hyperthymic temperament was previously associated with less psychopathology (Karam et al., 2010).At this point, it seems important to consider the interrelationship between temperament and emotions (Strelau, 1987).It has been shown that positive mood and emotions more pronounced in the Lebanese sample (R² = 0.024) as Our results showed that higher depressive, cyclothymic, irritable, and anxious temperaments were associated with higher levels of OrNe.It has been established that depression and anxiety are general risk factors for psychopathology and that they play a role in predisposing individuals to other mental illnesses (Clark, 2005).Depressive symptomatology was found to predict eating disorders among women (O'Brien et al., 2017).Similarly, high levels of anxiety were related to restrictive eating disorders (Martin et al., 2019).In a sample of patients with eating disorders, around 20% were diagnosed with depression, around are more likely to promote interest in health and healthy food choices (Gardner et al., 2014).Additionally, an experimental study showed that participants with a positive affect had a higher tendency towards healthy food choices (Hsu & Forestell, 2021).From this previous research, the assumption of temperaments as automated emotional responses, and present study findings, we can cautiously propose that the hyperthymic temperament, characterized by extremely positive emotionality and the self-characterization of being an optimist, sociable and self-confident, predisposes individuals to healthy interest in diet.A higher prevalence of OrNe was found in the Lebanese sample in our study while the level of interest in a healthy diet appeared comparable between Germany and Lebanon.In terms of healthy orthorexia, country, temperament and their interactions did not provide any explanatory value for the variation in healthy orthorexia scores.Given the missing cross-country difference in TOS-healthy orthorexia, this finding is not surprising.We would like to mention two things that were not considered in this way in the current study.From previous research, it is known that the definition of healthy eating varies between countries (Martinez-Gonzalez et al., 2000).In addition, the knowledge of dietary guidelines differs between countries (Patterson et al., 1995).As both presumable important variables have not been measured in the present study, it remains unclear whether healthy eating means similar things in both countries.Future research needs to acknowledge differences in definitions and conceptions of eating-related constructs.
In contrast to the missing links between temperament and healthy orthorexia, the regression model showed that some temperaments on the OrNe dimension were positively related to the anxious temperament, and depressive and hyperthymic temperament were negatively related to the level of orthorexic eating.In previous research, all affective temperaments were found to be associated with pathological eating behaviors (Marzola et al., 2020).In terms of depressive characteristics, OrNe was not significantly related to depressive symptoms in one previous study examining US-American university students, while it was weakly related to depressive features in a study among Chinese university students (Zhou et al., 2020).However, in a German sample from the general population, depressive symptoms were moderately associated with higher OrNe scores (Greetfeld et al., 2021).Our results fit within these previous studies.In terms of our research question whether affective temperaments underlie differences in orthorexic eating between Lebanon and Germany, the regression model suggested an effect.There was a pronounced positive association between TOS-OrNe and depressive temperament in Germany, but not in the Lebanese sample.By contrast, the negative association between hyperthymic temperament and TOS-OrNe was somewhat stronger in the Lebanese sample.Given these results, higher depressive scores in the German sample and lower hyperthymic scores in the Lebanese sample were associated with increased OrNe symptoms suggesting that different temperaments play a role in increasing the risk for OrNe in different countries and cultures.The hyperthymic individual is over-talkative, vigorous, full of plans, warm, people-seeking and extroverted.Overall, findings may be interpreted as reflecting the intense affective states that individuals attempt to modulate by eating and purging (in contrast with OrNe behaviors) (Ramacciotti et al., 2004).
48% had depressive symptoms and 73% showed irritability (Giovanni et al., 2011).In general, affective temperaments were found to underlie psychopathology (Clark, 2005), including eating psychopathology.In fact, the anxious temperament is often seen in eating disorders and was shown to modulate the severity of anorexic symptoms (Marzola et al., 2020).It was assumed that negative temperaments including depressive, cyclothymic, irritable, and anxious could be risk factors for developing psychopathological eating and our findings extent this to OrNe.

The moderating role of temperament in cultural differences of orthorexic eating
Depressive, cyclothymic, and irritable temperament were higher in the Lebanese sample while the German sample showed higher levels of the anxious temperament.These results reflect the cultural variability of temperaments previously established in cross-cultural studies (Gaias et al., 2012).These studies show that temperament and associated behaviors are subject to environmental influences (Rothbart et al., 2000), subsequently showing different manifestations across cultures.This is partly in line with previous findings as the frequency of all dominant affective temperaments seems to be balanced except for a lower frequency of dominant hyperthymic temperament in Germany, whereas the Lebanese sample similarly shows a high frequency of irritable and cyclothymic, and a low frequency of hyperthymic temperaments (Gonda et al., 2011).
Importantly, gender moderated some of the cross-country differences found in this study.Men from the German sample scored the lowest on cyclothymic and anxious temperaments while men from both the Lebanese and German samples scored higher on the hyperthymic and irritable temperament scale.In addition, being female was associated with lower depressive, lower hyperthymic and higher anxious temperaments in our study.A study investigating temperaments in six different countries, including Lebanon and Germany found the same results regarding men scoring higher on hyperthymic and irritable temperaments than women, who scored higher on cyclothymic and anxious temperaments (Vazquez et al., 2012).The same gender differences in our study were found in an investigation of a different German sample, except for the irritability temperament where results were insignificant (Erfurth, Gerlach, Michael et al., 2005).These consistent findings may be traced back to the correlations between temperaments: cyclothymic and anxious temperaments were closely related according to a previous German study (Blöink et al., 2005).The present data confirmed this assumption with the cyclothymic and anxious dimensions correlating by 0.42 and 0.47 in the Lebanese and German sample, respectively.personality in democratic countries, showing higher levels of extraversion/tendency to hyperthymic temperament but lower levels of neuroticism (tendency to depressive temperament) (Barcelo, 2017), in opposite to results of our study.Hence, future studies are necessary to elucidate this relation in depth.It seems reasonable to assume that temperaments also shape dietary behaviors and the present study provides evidence for this assumption.

Limitations
It is important to note some limitations of the study.First, the study was conducted on two convenience samples from Lebanon and Germany and does not represent the populations of both countries.Furthermore, different recruitment strategies, which were used for reasons of practicality, must be taken into account with the Lebanese sample being recruited among pharmacy visitors and the German sample being an online sample recruited primarily among undergraduates.Thus, the samples in both countries are not representative.These reasons hinder the generalization of results onto samples with other characteristics.Therefore, future studies taking representative samples from both countries are needed to make more solid conclusions.Moreover, country comparisons may be biased.Concerning the wording in the TEMPS scale, emotional semantics show cultural variation, which might cause variation especially in our samples from two different countries (Jackson et al., 2019).The TEMPS scale is a self-reported measure, which might cause the participants to respond in a socially desirable way.Furthermore, the study used a cross-sectional design, which hinders it from determining causality.A final limitation is the nondiversity of sociodemographic data, in other words, the German sample contained more females than males, which may lead to a bias in our results.In more detail, collecting data from individuals with a wide range of characteristics would have increased representativeness and thus confidence in conclusions drawn.Furthermore, while balanced samples are not necessary to test interactions, sufficient power is required to demonstrate the interaction effect.In the present study, the influence of gender was not a main research objective, and a priori sample size calculation was therefore not based on this factor.This must be considered in the interpretation.

Practical implications
Our findings have some important practical implications.First, some affective temperaments were associated with OrNe while others appeared relevant for healthy eating behaviors.Hence, identification of temperament types may be beneficial in high-risk populations to quickly take As for the depressive temperaments, orthorexic tendencies have been related to depressive symptoms (Greetfeld et al., 2021).The dissonance between the socio-cultural ideal and the individual's behavior may underlie this association (Hessler-Kaufmann et al., 2021).It is suggested that individuals are pressured to conform to a certain image portrayed in society, which could lead to pathological eating behaviors including orthorexia nervosa, strict dieting, and obsessions about healthy food, and this in turn later creates feelings of anxiety, isolation, and depression.
In terms of anxious temperament, a positive association was to be expected given previous theoretical discussions and empirical evidence.One of the main diagnostic criteria for OrNe are emotional consequences of non-compliance with self-imposed dietary rules such as anxiety, guilt, or grief.In this context, the correlation between orthorexic eating behaviors and fear of illness is also discussed.Initial studies have shown that individuals with a tendency towards orthorexic eating show various symptoms of illness anxiety (Barthels et al., 2021a;Kiss-Leizer et al., 2019).This association, however, did not explain cross-country differences, the two-way interaction appeared not significant.More research about OrNe and temperament is needed to further explore the relationship as the literature about the topic is scarce.
Although these constructs were not part of this study, we would like to briefly outline another debate.Collectivismindividualism is the most widely invoked distinction used to explain cross-cultural differences (Triandis & Suh, 2002).Since individualism tends to be pronounced in Western societies, collectivism is more pronounced of Eastern societies (Hofstede et al., 2005).In cultures with individualistic tendencies, the emphasis is placed on taking care of oneself and one's immediate family, while collectivist values are oriented towards the well-being of one's wider network.A variety of differences in interaction strategies has been identified between the two cultures.For example, individuals in collectivist cultures express a preference for approaches that enhance harmony, and those in individualist cultures prefer confrontation methods (Leung et al., 1992), extending to interactions between caregivers and young children.Generally speaking, individualism leads parents to favor the independence in their children, while collectivism manifests itself in parenting strategies aimed at producing interdependence (Greenfield et al. 2000).Furthermore, cross-cultural or cross-national comparisons are subject to some limitations (e.g., encoding of traits in the language, educational level, moral and religious values, or stereotypes), but the temperament structure itself appears to be universal (McCrae et al., 2005).Research has provided knowledge of underlying mechanisms.For instance, acculturation effects, political acculturation, in particular, affect the expression of countermeasures and to better understand vulnerability factors.Given the finding that OrNe serves as a dysfunctional coping strategy (Obeid et al., 2021;Strahler et al., 2022), results suggest the need to address temperament and difficulties in emotion regulation in psychiatric anamnesis of OrNe as well as in the development of tailored preventive strategies (e.g., psychoeducation).Lastly, culture-dependent associations indicate a relevance for culturally appropriate strategies in prevention and therapy.To better understand each client, culturally relevant information and themes must be explored (e.g., Strength-based interviews (Hays, 2001)).Gained knowledge supports the development of a culturally informed evaluation and treatment plan.

Conclusions
This cross-country study involving a Lebanese and a German sample aimed to investigate the role of temperament in orthorexia nervosa and healthy orthorexia.Our results showed that the hyperthymic temperament was related to a higher healthy interest in diet.The depressive, anxious, cyclothymic and irritable temperaments were associated with the pathological interest in diet.Some temperaments also moderated differences between the two samples.Our findings emphasize the role of temperaments in pathological and healthy eating behaviors, as well as their role in explaining differences in pathological orthorexic eating between two different populations and cultures.Our study suggests the importance of including measures of temperament in future prospective studies that assess persons at risk for orthorexia nervosa.Furthermore, future research that takes the culinary culture, the types of diet followed, and the social values attributed to eating into account, is merited to better understand the etiology of OrNe to derive therapeutic interventions.Hence, the assessment of eating styles and temperaments could help fit treatments to individuals, and focus interventions more on these aspects, and not only on weight.

Fig. 1
Fig. 1 Interaction between Orthorexia nervosa and the Depressive, Hyperthymic, and Irritable Temperaments separately for the Lebanese and German sample.OrNe, Orthorexia nervosa; TEMPS-M, brief form of the Temperament Evaluation of Memphis, Pisa, Paris, and San Diego; TOS, Teruel Orthorexia Scale

Table 1
Means and standard deviations of sample descriptive characteristics in Lebanon and Germany

Table 3
presents correlations among study variables and the final step of models regressing orthorexic eating on gender, country, and affective temperaments.Pearson correlations

Table 2
Means and standard deviations of variables under study in Lebanon and Germany