Introduction

Throughout the twentieth century until now, researchers have discussed whether certain foods may have an addiction potential and, thus, whether the eating behavior of some people may be addictive [1]. In 2009, the Yale Food Addiction Scale (YFAS) was developed as a self-report measure for the assessment of such an addiction-like eating behavior [2]. Specifically, the questionnaire is based on the diagnostic criteria for substance dependence in the fourth version of the Diagnostic and Statistical Manual of Mental Disorders (DSM-IV) and, accordingly, evaluates the presence of seven “food addiction” symptoms. A dichotomous score can also be calculated for classifying individuals as “food addicted.” The YFAS would turn out to be a popular, widely used instrument and, in fact, the standard measure for capturing addiction-like eating tendencies. Because of this, we provided a comprehensive review about the rationale and scoring of the scale, its different versions and translations, and its psychometric properties and correlates as well as about prevalence of food addiction symptoms and diagnoses, for its fifth anniversary in 2014 [3•].

In 2013, DSM-5 was released, which introduced significant changes to the diagnostic criteria for substance dependence. Specifically, the four symptoms of substance abuse and seven symptoms of substance dependence were merged. Additionally, one symptom—legal troubles because of substance use—was removed from the diagnostic criteria and another symptom—craving—was added as a new diagnostic criterion. Thus, DSM-5 now lists 11 symptoms of—which is now called—substance use disorder [4]. To acknowledge these changes, a revised version of the YFAS—the YFAS 2.0—was published in 2016 [5•]. Similar to our previous work about the YFAS [3•], the current article describes the rationale and scoring of the YFAS 2.0, including a detailed description of the changes made compared with the YFAS. Furthermore, we will review studies that have been conducted with the scale and summarize findings about its psychometric properties, prevalence rates of food addiction, and correlates of YFAS 2.0 scores.

Yale Food Addiction Scale 2.0

Development and Scoring

The YFAS consisted of 25 items with different response categories for capturing seven food addiction symptoms. Two items assessed a clinically significant impairment or distress due to one’s eating behavior. Three items were primer items that were not included in the scoring procedure. For the YFAS 2.0, new items were added, unscored items were removed, and item wordings were reformulated. The YFAS 2.0 consists of 35 items for capturing 11 food addiction symptoms (Table 1). In contrast to the YFAS, each item now has the same response format, ranging from 0 = never to 7 = every day. Further, changes in item wordings include lowering reading difficulty, consistent use of past tense, and rewording to improve clarity.

Table 1 Items and scoring of the original and modified Yale Food Addiction Scale 2.0
Full size table

Similar to the original YFAS, items are preceded by an instruction, which references the consumption of foods high in fat and/or refined carbohydrates, as these foods are most relevant to food cravings and eating binges. To score the scale, all item scores are transformed to a dichotomous format (0 and 1). The cut-off values (i.e., which scores are coded with 0 and which scores are coded with 1) differ across items (Table 1). Each of the 11 symptoms is assessed by two or three items. Two items assess a clinically significant impairment or distress due to one’s eating behavior. The dichotomized responses are summed up for each symptom and for the clinically significant distress/impairment questions. If there is a score of at least 1 within a symptom, then this symptom is met. The number of symptoms can then be added up to a symptom score, which can range between zero and 11 symptoms.

In line with the diagnostic criteria for substance use disorder in DSM-5, there are three different severity levels when calculating the diagnostic score of the scale. When the criterion of a clinically significant impairment or distress is met, two or three symptoms indicate “mild food addiction,” four or five symptoms indicate “moderate food addiction,” and six or more symptoms indicate “severe food addiction.” An SPSS syntax for scoring the scale can be found in Appendix 1.

Psychometric Properties

Factor Structure

As item scores of the YFAS 2.0 are not simply added up to a total score, factor structure has been tested at the symptom rather than the item level. That is, factor analyses were applied on the 11 symptom scores (i.e., symptom met vs. symptom not met) instead of the 35 item scores. The distress/impairment criterion was not included here as it reflects the clinical significance of the full syndrome rather than indicators of individual criteria. Across different samples, the 11 symptoms showed a one-factorial structure, indicating that all symptoms represent food addiction as a single construct [5•, 6,7,8,9,10]. Factor structure has been found to be invariant across different racial groups (Black vs. White US participants), yet measurement invariance was only partially supported in men versus women [11].

Reliability

Internal reliability was good (α = .80–90) or excellent (α > .90) in numerous studies [5•, 6,7,8,9,10, 12,13,14,15,16,17,18]. As researchers often do not indicate how internal reliability was determined exactly in their studies, we again would like to highlight here that it is not advisable to calculate Cronbach’s alpha of raw scores of all items as these are not simply summed up to a total score. Instead, we suggest calculating Kuder-Richardson’s alpha for the dichotomous scores of the 11 food addiction symptoms. Retest-reliability across 3 weeks was high for the Arabic version [15].

Translated and Modified Versions

The YFAS 2.0 has been translated into German [6], French [9], Italian [7], Turkish [19], Spanish [20•], Korean [21], Arabic [15], and Japanese [22]. A modified YFAS 2.0 (mYFAS 2.0) has been developed, which is a short form of the YFAS 2.0 [23]. The mYFAS 2.0 consists of 13 items (one item for each symptom and two items for clinically significant impairment or distress). The specific items of the YFAS 2.0 that form the mYFAS 2.0 are depicted in Table 1. An SPSS syntax for scoring the scale can be found in Appendix 2. Similar to the long version, the mYFAS 2.0 showed a one-factor structure, high internal reliability, and full measurement invariance across racial groups [23,24,25]. However, the scale was only partially invariant for men and women, suggesting that two of the 11 symptoms may be less associated with food addiction for women compared with men [24]. The mYFAS has also been used in Brazilian Portuguese [26, 27] and Italian [28]. Finally, a 16-item version of the YFAS 2.0 for children and adolescents has recently been developed, for which items are scored dimensionally (i.e., are summed up to a total score) [29]. The English versions of the YFAS 2.0, mYFAS 2.0, and YFAS 2.0 for children can be downloaded along with scoring instructions here: https://fastlab.psych.lsa.umich.edu/yale-food-addiction-scale.

Prevalence of YFAS 2.0 Diagnoses

Prevalence rates of food addiction diagnoses from different studies that used the YFAS 2.0 are displayed in Table 2. Figure 1 displays a more schematic depiction of prevalence rates as a function of specific samples studied. In broad samples—some very selective and some more representative for the general population—prevalence rates of YFAS 2.0 diagnoses roughly ranged between 3 and 20%. Using the mYFAS 2.0, food addiction prevalence was 4% in a large sample in Brazil [27], 6% in a sample in Italy [28], and ranged between 13 and 15% in US samples [23, 25].

Fig. 1
figure 1

Schematic depiction of prevalence rates of food addiction diagnoses as assessed with the Yale Food Addiction Scale 2.0. Percentages ranged between approximately 3 and 20% in healthy or unselected samples and between approximately 20 and 50% in samples with obesity. Prevalence rates were approximately 60% in individuals with a subthreshold eating disorder, 70% in individuals with anorexia nervosa, 80% in individuals with binge eating disorder, and 95% in individuals with bulimia nervosa

Full size image

In samples with obesity, prevalence rates of YFAS 2.0 diagnoses ranged between approximately 20 and 50%. In samples with individuals with eating disorders, prevalence rates were higher than 60%, with the highest numbers found in individuals with bulimia nervosa. In fact, about 95% of individuals with bulimia nervosa received a YFAS 2.0 diagnosis across different studies [20•, 30, 31]. In individuals with anorexia nervosa, those with binge/purge anorexia subtype have higher prevalence rates than those with restrictive anorexia. Yet, still more than half of individuals with restrictive anorexia receive a food addiction diagnosis [31].

Interestingly, it appears that severe food addiction diagnoses are more prevalent than moderate or mild food addiction diagnoses [5•, 6,7,8, 10, 14,15,16, 32, 33]. That is, most people that meet the distress/impairment criterion also endorse at least six of the 11 symptoms (or vice versa). Although many individuals may meet several (up to five) symptoms, it is rather uncommon that these also report significant distress or impairment.

Correlates of the YFAS 2.0

Sex

Similar to findings with the YFAS, women tend to have a higher likelihood of meeting the YFAS 2.0 diagnosis than men in unselected or nationally representative samples [5•, 7, 9, 11, 33]. It has also been found that men and women had similar YFAS 2.0 symptom counts and that the sex difference was driven by women endorsing the distress/impairment criterion more often [11]. Moreover, the correlation between YFAS 2.0 scores and body mass index was descriptively higher in men (r = .54) than in women (r = .36) in an unselected sample (yet the difference in the size of these correlations was not statistically tested) [13]. Among samples with obesity, however, those with and without a YFAS 2.0 diagnosis did not differ regarding sex distribution in the majority of studies [6, 8, 34•]. In a study with bariatric surgery candidates, however, a significantly higher percentage of women than men received a YFAS 2.0 diagnosis, but this effect was small [16]. Thus, it might be that sex differences in food addiction diagnoses in non-obese samples can be explained by the fact that individuals with bulimia and anorexia nervosa are mostly women who are under- or normal-weight and these conditions highly overlap with YFAS 2.0 diagnoses.

Body Weight

Body mass index usually shows small, positive associations with YFAS 2.0 scores in unselected samples [5•, 6, 7, 9, 15, 33]. Furthermore, samples with obesity have a higher prevalence of food addiction than non-obese samples (Table 2). However, it has been previously observed that the relationship between food addiction and body mass index appears to be non-linear [40]. Specifically, food addiction positively relates to body mass index, but this slope levels off in higher body weight ranges: within samples with obesity, for example, those with and without a YFAS 2.0 diagnosis usually do not differ in body mass index [6, 8, 12, 16, 39]. As of this writing, there has been only one study that examined the prospective relationship between YFAS 2.0 scores and body weight change. In morbidly obese patients seeking bariatric surgery, those with a food addiction diagnosis lost less weight during a dietary and lifestyle intervention prior to surgery than those without a food addiction diagnosis [34•].

Table 2 Prevalence of food addiction as assessed with the Yale Food Addiction Scale 2.0
Full size table

Eating Behaviors

Scores on the YFAS 2.0 are strongly associated with disinhibited eating behavior, experiences of food cravings, binge eating symptoms, and frequency of binge eating episodes [5•, 6, 7, 9, 14, 32, 33]. Relationships with other eating disorder symptoms (e.g., eating, weight, and shape concern) and eating styles (e.g., emotional eating, grazing) are typically of moderate magnitude [6,7,8,9, 17]. Restrained eating has been linked inconsistently to YFAS 2.0 scores with studies finding either no or only small associations [5•, 6,7,8,9, 14]. Regarding nutritional profiles, one study showed that those with food addiction reported higher intakes of confectionary, fast food, snack foods, hot chips, potato crisps, and soft drinks and lower intakes of core foods like fruits and vegetables and were less likely to eat breakfast every day than those without food addiction [13]. Yet, it has also been found that those with a YFAS 2.0 diagnosis report decreased enjoyment of eating highly processed foods [39].

Co-morbid Mental Disorders

Food addiction diagnoses as assessed with the YFAS 2.0 are associated with a range of mental disorders such as depression, anxiety disorders, posttraumatic stress disorder, and attention deficit hyperactivity disorder [7, 8, 12, 14, 16, 32]. Some studies also reported relations to higher stress, hopelessness, suicidality, non-suicidal self-injury, difficulties in emotion regulation, and lower sleep quality [7, 8, 10, 13, 31]. In obese patients seeking bariatric surgery, no associations were detected between YFAS 2.0 scores and substance-related addictions such as alcohol use disorder or tobacco use disorder [8, 16], but one study suggested that food addiction relates to a higher prevalence of other addictive behaviors such as compulsive buying and Internet use [16].

Executive Functioning and Personality

Obese individuals with a YFAS 2.0 diagnosis (either with or without binge eating disorder) did not differ in executive functioning from obese individuals without a YFAS 2.0 diagnosis (either with or without binge eating disorder) [41]. Another study found weak, inconsistent support for deficits in executive functioning with higher YFAS 2.0 scores in a small sample [35]. Higher YFAS 2.0 scores relate to higher self-reported impulsivity, particularly to attentional impulsivity (e.g., having problems concentrating), motor impulsivity (e.g., acting without thinking), and negative urgency (i.e., acting rashly in response to negative emotions) [6, 35, 42]. A study which differentiated between impulsivity and sensation seeking—two positively related constructs—found that although those with food addiction reported higher impulsivity, they showed lower sensation seeking than those without food addiction [13]. A study using the Spanish version of the YFAS 2.0 found no association with novelty seeking, but a positive association with harm avoidance and a negative association with self-directedness [20•].

Summary and Outlook

YFAS Vs. YFAS 2.0

Although the changes from the YFAS to YFAS 2.0 have been quite substantial, it appears that psychometric properties, food addiction prevalence rates, and correlates of the YFAS 2.0 (and mYFAS 2.0) are largely similar to findings with the YFAS. These include, for example, their good-to-excellent internal reliability, unidimensional structure, very high food addiction prevalence rates in samples with eating disorders, and a positive—but non-linear—relationship with body weight [3•]. Few exceptions include, for example, the child version of the YFAS 2.0. Here, the newly added symptoms received very low endorsement rates, which necessitated refinement of the scale through item selection and a different scoring procedure [29]. Thus, future revisions of the YFAS 2.0 for children may be required so that the scale can be used in children and adolescents as an equivalent to the YFAS 2.0 in adults.

After the changes of substance dependence criteria from DSM-IV to DSM-5, one concern regarding the YFAS 2.0 was that food addiction prevalence may now be substantially higher than with the YFAS as meeting only two symptoms (and the distress/impairment criterion) suffices to receive a food addiction diagnosis [43]. Yet, it appears that this concern was unjustified. As described above, most individuals who endorse only two symptoms on the YFAS 2.0 rarely meet the distress/impairment criterion and those who meet the distress/impairment criterion usually endorse several symptoms. In other words, although many people may experience food cravings and feel that they lost control of their consumption, they do not suffer from it. This suggests that the YFAS 2.0 may have high specificity to detect individuals with an addiction-like eating behavior.

Future Directions

While the YFAS 2.0 shows good psychometric properties, only one study has reported retest-reliability (over a relatively short time period of 3 weeks) [15]. Thus, additional studies using longer follow-up periods are necessary to evaluate the long-term stability of YFAS 2.0 scores appropriately. Similarly, more longitudinal studies are necessary that examine the prospective relationship between YFAS 2.0 scores and outcomes such as weight change. To date, only one study has examined such associations and points towards poorer weight loss in individuals with food addiction during a weight-loss intervention [34•].

An ongoing debate is the distinctiveness of food addiction as measured with the YFAS and YFAS 2.0 with established eating disorder diagnoses [44, 45]. While discriminant validity of the YFAS 2.0 is supported regarding constructs such as dietary restraint, it highly overlaps with conditions such as bulimia nervosa and binge eating disorder. Therefore, future research needs to determine not only the cross-sectional relation between these conditions but also examine other characteristics such as their predictive power regarding eating and weight outcomes, for example after psychotherapy or weight management programs.

Yet, the high degree of overlap is somewhat expected given the overlapping mechanisms implicated in both binge-related eating disorders and addictive disorders (e.g., impulsivity, emotion dysregulation, craving), but there are also mechanistic explanations that are unique to an addiction perspective [46]. Specifically, the addiction perspective proposes that the types of foods commonly consumed during binge eating episodes (e.g., foods high in refined carbohydrates and fat) may be capable of causing reward-related adaptations that drive forward compulsive patterns of behavior [46]. Although there is strong animal evidence in support of this concept [47, 48], additional studies are needed to investigate this hypothesis in humans. Such studies may reveal a clearer picture about the distinctiveness of food addiction and binge-related eating disorders. Further, the relatively high rates of YFAS 2.0 food addiction in patients with restrictive-type anorexia nervosa are unexpected and may require qualitative work to investigate whether the scale may be interpreted differently in this population.

Conclusions

In conclusion, the development of the YFAS 2.0 has provided an updated assessment tool to support continued investigation into the food addiction hypothesis based on the current diagnostic understanding of addiction. The YFAS and YFAS 2.0 exhibit similar psychometric properties and estimate similar prevalence rates of food addiction. Importantly, the YFAS 2.0 is associated with clinically relevant correlates, including obesity, disordered eating, depression, and some measures of executive functioning difficulties. Future longitudinal research, particularly regarding the ability of the YFAS 2.0 to predict treatment outcomes, is needed. As the scientific understanding of the best ways to conceptualize and assess addictive disorders evolves, future iterations of the YFAS will likely be needed to reflect these advances.