An Exploratory Study of Prevalence of Food Addiction in Indian Youth

Background and aims: As the understanding of food addiction increases, there is a need to explore the occurrence of this condition in different population groups. This study aimed to assess the occurrence of food addiction in a sample of respondents from India using a Hindi version of the Yale Food Addiction Scale (YFAS). Methods: The Hindi language version of the scale was developed using the back-translation methodology. Subsequently, an online questionnaire-based study was conducted using convenience sampling which presented the Hindi version of YFAS. Results: From 376 respondents (median age 19 years, 42.8% males), the prevalence of food addiction was 13.3%. Persistent desire or repeated unsuccessful attempts to quit was the most common symptom domain endorsed. The weight (median 67 kg versus 60 kg) and BMI (median 25.89 kg/ m 2 versus 23.04 kg/ m 2 ) were higher in the food addiction group as compared to the non-food addiction group. Conclusions: Despite the limitations of selection bias, the study suggests that food addiction may be present in a proportion of young aged Indians. The association of food addiction with higher weight and BMI suggests propensity to develop metabolic syndrome, and the need to evaluate interventions that could modify phenomenological expression of food addiction.


Introduction And Background:
In the recent decades, food addiction as a construct has gathered scienti c attention (Davis et al., 2011;Ziauddeen and Fletcher, 2013). Addiction to food stuff which is rich in fats, sugars and salts have been conceptualized as a condition akin to addiction to other hedonic psychoactive substances like alcohol and tobacco. The conceptualization of food addiction has been proposed given the scenario of some individuals having cravings towards some food stuff, and excess consumption despite having adverse consequences like obesity. Phenomenologically, some of the criteria for addiction to substances have been found with regards to hyper-palatable varieties of food (Meule and Gearhardt, 2014), moving forward the thought that food addiction as a construct has equivalence with substance use disorders.
Neurobiological research seems to highlight the commonalities in neural pathways involved in food addiction and substance use disorders (Lindgren et al., 2018;Volkow et al., 2012). Food addiction has speci c connotations with respect to obesity which often results from the imbalance of food intake and energy expenditure (Davis et al., 2011;Pedram et al., 2013).
In countries like India, with relative abundance of food in recent years, and reduction in physical exercise, the rates of obesity have been on the rise. Obesity itself has been associated with the occurrence of metabolic syndrome which becomes a risk factor for diabetes mellitus and dyslipidemia. The outcomes of obesity and management of concomitant dyslipidemia and diabetes mellitus becomes challenging with the presence of uncontrolled craving for food and excessive consumption behaviors (Jiménez-Murcia et al., 2019;Nightingale and Cassin, 2019;Sarkar et al., 2019). Hence food addiction as a construct seems to have implications for outcomes of non-communicable diseases (Lerma-Cabrera et al., 2016), which themselves has become grave challenges in the Indian healthcare system.
Food addiction, like any other psychiatric condition, needs some structured and systematic manner of assessment. Various aspects of food addiction can be captured through the application of questionnairebased scales and instruments. Among the questionnaire-based instruments assessing food addiction, the Yale Food Addiction Scale (YFAS) is probably the most commonly used one. The instrument is a 25-item questionnaire. The questionnaire looks at past year eating habits and focuses on sweets, starches, salty snacks, fatty foods, sugary drinks or any other food stuff which the individual considers as problematic. Some of the questions in this instrument have Likert type responses, and some of the questions have dichotomous responses. Based upon the responses, the patient is classi ed into having either food addiction or not. The instrument has been used in general populations, and clinical populations (like those with diabetes mellitus or obesity). The prevalence rates of food addiction has been reported to range from 5.4% to more than 50% across different studies which have used the Yale Food Addiction Scale (Pursey et al., 2014). The prevalence rates, from this systematic review involving close to 200,000 subjects suggests that food addiction is higher in women and those from the clinical population, rather than the general population. There is only one study from India which reported the rate of food addiction to be 32.5%, using the English version of YFAS (Wiedemann et al., 2018). We aimed to nd out the prevalence of food addiction using an Hindi language version of the YFAS to facilitate wider application, eliciting response in the native language of the region.

Methodology:
Hindi translation of YFAS was developed through the back-translation methodology. The YFAS comprises 25 single choice items (16 likert rated, 9 yes/no), one multiple choice item, and one open ended question. The initial phase of the study involved translation of the YFAS into Hindi language by two bilingual doctoral level experts who had previously not seen the instrument. The translation was aimed at semantic translation rather than literal translation. Consensus was achieved from the experts on the words that differed in the instrument to have a consistent instrument. The example food items were modi ed to include the common food products in the region. The Hindi version was then translated into English by two other bilingual experts, and the consensus version was sent to the original developer of the scale to ascertain conceptual validity for the items and the scale.
The application of the Hindi version in the larger sample was done through an online survey, cognizant of the ongoing pandemic. A GoogleForms questionnaire was developed. Age, gender, educational status, employment, current weight and height were enquired into. Body mass index (BMI) was computed based upon the reported height and weight. The questionnaire presented the Hindi items of the YFAS. Adult participants who were willing to provide online consent were recruited to participate in the study. Participants were recruited by sending emails and through WhatsApp to the contacts of the investigators. Participants were free to send the questionnaire further to their contacts. Data collection was done for a period of 5 days in the month of October 2020. A formal sample size estimation was not done, but it was aimed to have more than 200 responses.
The responses from the participants were exported into excel les, and then were imported into SPSS version 22.0 (IBM Corp, Armonk, NY). Descriptive statistics were used to represent the nominal, ordinal and scale data. Normality was assessed using Kolmogorov Smirnov test. Internal consistency of the test was measured using Cronbach alpha. The relationship of the presence of a diagnosis of food addiction with gender, age, weight, height and BMI was assessed using suitable non-parametric tests (chi square and Mann Whitney U test). A p value of less than 0.05 was considered and missing value imputation was not done.
Appropriate parametric and non-parametric tests were applied based on distribution of data. Cohen's Kappa was used to nd the internal consistency of the scale. Factor analysis of the questionnaire was attempted. Correlation coe cients were computed for the test-retest reliability and language equivalence. The relationship of the score on Yale Food Addiction Scale with the BMI was assessed.
The responses to individual questions of the Hindi version of YFAS are presented in Table 1. Among the likert rated questions, eating more than planned was most frequently endorsed. Among the dichotomous, yes-no type rated questions, wanting and trying to cut down certain types of foods were most commonly endorsed. The Cronbach alpha for the scale was 0.897. Responses are in the form of Never/ Once a month / 2-4 times a month/ 2-3 times a week/ 4 or more times or daily (question 1 to 16) or Yes / No (question 17 to 24), or 1 or fewer times/ 2 times/3 times/ 4 times/ 5 or more times (question 25) The domains endorsed by the participants are presented in Table 2. The most common domains of food addiction that were endorsed by the participants were persistent desire or repeated unsuccessful attempts to quit, and tolerance. It was seen that 13.3% of the sample had a diagnosis of food addiction. While those with a diagnosis of food addiction did not differ with those without a diagnosis of food addiction on gender, age and height, signi cant differences emerged on weight and BMI. The weight (median 67 kg versus 60 kg, Mann Whitney U = 5991, p = 0.013) and BMI (median 25.89 kg/ m 2 versus 23.04 kg/ m 2 , Mann Whitney U = 5439.5, p = 0.011) were higher in the food addiction group. Discussion: The present study suggests that the prevalence of food addiction in this predominantly young adult population was around 13.3%. Also, presence of food addiction was seemingly more common in those with higher weight and BMI. The prevalence rate was lesser than the weighted mean prevalence reported by the meta-analysis by Pursey et al. (Pursey et al., 2014). The occurrence of food addiction was also lower than the other study from Indian population (Wiedemann et al., 2018). The occurrence rate is similar to studies from other parts of the world (Eichen et al., 2013;Gearhardt et al., 2009). Even lower prevalence rates have also been reported in some other studies (Mason et al., 2013;Meule et al., 2012).
The present study did not nd a relationship between age or gender and food addiction. This is at a variance from previous other studies which have suggested that greater age and female gender was associated with food addiction (Pursey et al., 2014). There is some literature, including nationally representative samples, which suggest that gender differences may not apply for food addiction (Hauck et al., 2017;Pipová et al., 2020;Schulte and Gearhardt, 2018). It is possible that the culturally determined expectations and eating attitudes may in uence how options are perceived, appraised and responded to. There is some evidence to suggest that women from India and other Asian countries have less drive for thinness than Western countries (Kayano et al., 2008). The study failed to replicate higher rates of food addiction in the younger population, as the sample largely consisted of younger individuals (the interquartile range of age was 18 to 20 years).
The association of food addiction and weight and BMI as suggested in the present sample are in similar lines with the other studies (Pursey et al., 2014;Schiestl et al., 2020). The causality of the association has been explored, and it has been suggested that while food addiction may lead to uncontrolled eating and consequently greater weight gain. Food addiction also has been implicated in lack of e cacy of weight loss interventions (Schiestl et al., 2020). The median age of those who had food addiction in this study fell in the overweight range, suggesting an at-risk group, where gradual increase of BMI was likely to lead to obesity and consequent medical conditions. The overall rates of food addiction seems to be higher in the group of obese individuals (Pursey et al., 2014).
Among the symptoms of food addiction, the most common symptoms endorsed pertained to persistent desire or repeated unsuccessful attempts to quit. Other nationally representative studies have also suggested this item to be most common (Pipová et al., 2020;Schulte and Gearhardt, 2018). However, certain other items like giving up important social, occupational, or recreational activities were endorsed quite less frequently. The implication of this phenomenology is that individuals might not have dysfunction in their different functional domains, though they might nd themselves unable to quit speci c food stuff. This probably highlights that the adverse consequences of food addiction may not be immediate, but in the long-term as individuals are unable to control their food intake.
The ndings of the present study have some implications. The occurrence of this condition in the population as ascertained using a commonly used questionnaire, suggests that food addiction can be delimited, furthering research on the construct. The association with high BMI may help direct query towards food consumption habits during the routine clinical encounter. From a policy perspective, attention towards foodstuffs that are consumed for satisfy craving beyond the point of nutritional relevance may help to make informed choices by the consumers. This may look premature in Indian context given the extent of under-nutrition in the country, but childhood and adult obesity is a rising concern and preventive measures may help to address genesis of several non-communicable diseases.
Some limitations of the present study merit attention. The sampling was through contacts through email and social media of the contacts of investigators, resulting in possible selection biases. Also, the sample consisted mainly of the younger population, and may not represent the ndings from other age groups. Additionally, the study did not manually check for weight and height and relied on the self-reported values. We did not look at temporal stability of food addiction, and did not look at other related determinants like physical activity, body fat composition and calorie or fat intake.

Conclusion:
To conclude, the present study suggests that about 13.3% of Indian young adult population may be having food addiction, and that food addiction is associated with higher weight and BMI. Future studies may look at the relationship of food addiction and trajectory of development of obesity. The occurrence of food addiction in different populations can be ascertained. Research in the neurobiological understanding of appraisal of speci c food products, and methods to control craving would be helpful to devise better treatment strategies.