Food addiction (FA) is characterised by excessive consumption of highly energetic, hyperpalatable and processed foods, whose characteristics are similar to addictive substances(Reference Gearhardt and Schulte1). The prevalence of FA is about 14 % in non-clinical samples in general and low-income countries(Reference Praxedes, Silva-Júnior and Macena2,Reference de Melo Barros, da Silva Júnior and Praxedes3) . Horsager et al.(Reference Horsager, Bruun and Færk4) and Silva Júnior and Bueno(Reference Silva Júnior and Bueno5) reported for the first time the association between FA and type 2 diabetes mellitus (DM2) in the general population, with mental disorders and individuals undergoing bariatric surgery, respectively. DM2 is a chronic non-communicable disease that has rapidly grown among young adults worldwide(Reference Ahmad, Lim and Lamptey6). Poverty is an important barrier to prevention or adequate treatment, making vulnerable populations more exposed to developing this grievance(Reference Bekele, Asefa and Getachew7).
However, the relationship between FA and DM2 still needs to be better explored in different social contexts and population groups. Thus, this work presents a novel setting to assess the association between FA and DM2 in women living in poverty. Therefore, this study aims to evaluate whether the diagnosis of DM2 is associated with FA in women living in poverty.
Methods
Study design, sample and location
A population-based cross-sectional study was conducted between October 2020 and May 2021. This is a secondary analysis of a larger study; more information on sample size calculation can be seen in a previous publication(Reference Silva-Neto, Bueno and Santos8). We included adult women of reproductive age (20–44 years) living in subnormal settlements in Maceió, the state’s capital in the Northeast of Brazil, with the lowest Human Development Index (0·721) among the region’s capitals. Subnormal settlements are spaces with a low supply of essential public services and where the population lives in an unfavourable socio-economic and demographic situation. Pregnant women and women with some disability that would compromise their food intake or make it impossible to carry out the interview or understand the research questionnaires were not included. Participants who did not answer the modified Yale Food Addiction Scale (mYFAS) 2.0 were excluded.
Sociodemographic and clinical variables
Age (years), race/skin colour and per capita monthly family income data were collected. Participants with per capita monthly family income less than US$ 95·59 (values converted from reais to US dollars considering the exchange rate on 31 May 2021 – R$ 5·22) were in poverty(9). We adopted the method proposed by the Instituto Brasileiro de Geografia e Estatística (IBGE) to collect race/skin colour data, in which the individual must identify themselves as White (Caucasian), Black (Afro-descendant), Brown (mixed race/skin colour), Yellow (Asian) and Indigenous(10). For this study, we grouped Blacks, Browns, Yellows and Indigenous people in the ‘non-White’ category. The participants were asked whether they had been diagnosed with DM2 by a physician at some point. Information on medication use for treating DM2 was also collected when the answer was positive. The participant was considered diabetic when she reported having a DM2 diagnosis and reported using antidiabetic medication. This population living in poverty is served by Brazil’s Unified Health System (in Portuguese: Sistema Único de Saúde), which provides the following drugs to treat DM2: biguanide (metformin), sulphonylureas (glibenclamide and gliclazide), insulin and SGLT2 inhibitors (dapagliflozin)(Reference Bahia and Almeida-Pititto11).
Physical activity
The validated version of the International Physical Activity Questionnaire – short form (IPAQ-short form) was used for the Brazilian population(Reference Matsudo, Araujo and Matsudo12). The IPAQ-short form consists of eight questions related to the frequency and duration of physical activity in the last week, for a minimum of ten continuous minutes, at moderate, vigorous, and walking intensities, and includes activities done at work, at leisure for sport, for exercise and even household activities. Using the IPAQ-short form, it was possible to classify the participants as sedentary, insufficiently active A, insufficiently active B, active and very active.
Anthropometry
Weight (in kg) was measured with a digital scale and height (in metres) with a portable stadiometer. The participants were barefoot and wearing light clothes during the anthropometric evaluation. The BMI was calculated and classified as underweight (BMI < 18·5 kg/m2), normal weight (18·5–24·9 kg/m2), overweight (25·0–29·9 kg/m2) and obesity (≥ 30·0 kg/m2).
Food addiction
The Brazilian Portuguese translated and validated version of the mYFAS 2.0(Reference Nunes-Neto, Köhler and Schuch13) was used to determine FA. The scale comprises thirteen questions, of which eleven represent symptoms of eating behaviour and two refer to clinical distress/impairment. Each question is answered according to the frequency with which it occurs, ranging from ‘never’ to ‘every day’, each of which has a limit for the criterion of the symptom to be met. In the end, the eleven symptoms are added together to create a symptom count score option. Positive cases of FA were considered those who reached the threshold for two or more symptoms and also for at least one of the items that assess clinical distress. The severity of FA was also considered, classifying it as mild, moderate or severe (mild = 2–3 symptoms plus impairment or distress, moderate = 4–5 symptoms plus impairment or distress, and severe = 6 or more symptoms plus impairment or distress)(Reference Silva Júnior and Bueno14).
Statistical analyses
Continuous variables were summarised as mean and standard deviation, and categorical variables as percentages. To compare the mean FA symptoms between those with and without a DM2 diagnosis, a t test for independent samples was performed. Poisson regression with robust variance estimation was used to estimate the association between DM2 medical diagnosis (dependent) and FA diagnosis and symptoms (independent). The multivariable analysis included age (in years), poverty situation, race/skin colour, physical activity and BMI (kg/m2) as adjustment variables. All analyses were conducted using the statistical program R v. 3.6.1 (R Foundation for Statistical Computing).
Results
This is a secondary analysis of a larger study, to which 2356 women were invited, 118 did not agree to participate and 324 did not meet the established age criteria. Thus, 1914 women were initially included, of which thirty-six were excluded because they had not answered the mYFAS 2.0. Hence, the final sample included in this study was 1878 women with a mean age of 30 ± 7 years. 1608 (85·6 %) self-reported a non-White race/skin colour, and 1424 (75·8 %) lived in poverty. A medical diagnosis of DM2 was reported by sixty (3·2 %), and all women who reported a DM2 diagnosis were taking antidiabetic medication. The mean BMI was 28·6 ± 6·5 kg/m2. Regarding FA, 283 (15·1 %) met the diagnostic criteria. Considering the severity of FA, the prevalence of mild, moderate and severe FA was 141 (7·5 %), 67 (3·6 %) and 75 (4·0 %), respectively. Table 1 presents the characteristics of the sample.
Table 1. Characteristics of sample
mYFAS, modified Yale Food Addiction Scale.
* Food addiction determined by mYFAS 2.0.
Participants with DM2 had more FA symptoms than those without the diagnosis (1·53 ± 2·39 v 1·04 ± 1·79; p = 0·04). In the univariable analysis, the medical diagnosis of DM2 was associated with FA (prevalence ratio (PR): 2·04; 95 % CI (1·17, 3·58)). This association remained significant in the multivariable analysis, even after adjustment for age, poverty, race/skin colour, physical activity and BMI (PR: 2·18; 95 % CI: (1·26, 3·76)). The DM2 was also identified to be associated with role interference (PR: 1·93; 95 % CI (1·01, 3·67)) symptom of FA in the multivariable analysis (Table 2).
Table 2. PR for diabetes mellitus type 2 by food addiction symptoms in Brazilian women living in poverty situation
PR, prevalence ratio by Poisson regression with robust adjustment of variance.
* PR for the diabetes mellitus type 2 by age (in years), poverty situation, race/skin colour, physical activity and BMI.
Discussion
In our study, we observed a positive association between the diagnosis of FA and DM2 in women living in poverty situations. To our knowledge, this is the first study to evaluate such an association.
The association identified in the present study draws attention since the prevalence of FA is higher among women(Reference de Melo Barros, da Silva Júnior and Praxedes3). This situation may contribute to the trend of increasing cases of DM2 worldwide. Brazil is the fifth country in the world in the number of people living with this condition, demanding greater caution(Reference Saeedi, Petersohn and Salpea15). Nevertheless, we observed a prevalence of DM2 of 3·2 % in our sample, which is a relatively low rate but in accordance with a national survey in Brazil, in which the prevalence of DM2 in women aged between 18 and 44 years ranged from 0·9 % to 5·7 %(16).
Our findings are in accordance with the previous findings on different populations. Horsager et al.(Reference Silva Júnior and Bueno5) identified that the association between FA and DM2 remained positive when adjusted for age, income and BMI, demonstrating that these may be important variables that should be considered. Silva Júnior and Bueno(Reference Ahmad, Lim and Lamptey6) also observed an association between FA and DM2 in individuals after bariatric surgery, even after adjusting for age, sex and BMI class. Hence, the association between FA and DM2 seems to exist despite the different living conditions of the studied populations.
The scenario identified here can be further aggravated when it is observed that the population, regardless of their financial situation, is increasingly exposed to ultra-processed foods. This situation may favour the increase in their consumption(Reference Atanasova, Kusuma and Pineda17), leading to a higher prevalence of both FA and DM2. In this direction, if food price projections in Brazil come true, the observed panorama may be even more impacted since there is a trend that, as of 2026, ultra-processed foods will become cheaper than fresh foods, making their access even easier(Reference Maia, Passos and Levy18).
There are some ways to explain the possible relationship between FA and DM2. First, the diagnosis of FA is more frequent in individuals with higher BMI, a factor known to increase the risk of DM2(Reference de Melo Barros, da Silva Júnior and Praxedes3,Reference Klein, Gastaldelli and Yki-Järvinen19) . In addition, individuals who meet the diagnostic criteria for FA appear to have an unhealthy eating pattern, often associated with higher consumption of confectionery, baked sweet products, fast food and ultra-processed foods, including in a Brazilian sample(Reference Pursey, Skinner and Leary20,Reference Silva Júnior, Gearhardt and Bueno21) . This eating pattern is also considered a risk factor for DM2. In addition, evidence shows that a diagnosis of FA among individuals with DM2 is associated with poorer glycemic control and microvascular complications(Reference Nicolau, Romerosa and Rodríguez22). Thus, FA may be a factor that increases the difficulty of adhering to healthy eating habits and, consequently, poorer control of DM2.
Our study has strengths and limitations. Our strengths include the unprecedented association between FA and DM2 in women living in poverty situations. As limitations, we point out the self-report of the medical diagnosis of DM2 by the participants included, especially because of the low level of education. In addition, our sample comprises women in the poverty situation, which can limit access to healthcare and the medical diagnosis of DM2. These factors may have influenced our sample’s relatively low prevalence of DM2. However, a study that evaluated individuals with a medical diagnosis of diabetes found that 75 % of these individuals correctly self-reported their diagnosis(Reference Shah and Manuel23). Furthermore, using mYFAS 2.0 may lead to an underestimation of FA prevalence and, consequently, the association between FA and DM2. However, the mYFAS 2.0 has similar psychometric properties to the YFAS 2.0 and has a translated and validated version for Brazilian Portuguese(Reference Nunes-Neto, Köhler and Schuch13,Reference Silva Júnior and Bueno14,Reference Schulte and Gearhardt24) . Finally, the study’s cross-sectional design allows new hypotheses to be generated but does not allow causality to be inferred. In this sense, prospective studies should be carried out to investigate the possible causality between FA and DM2.
The positive association between FA and DM2 in women living in poverty situations identified in the present study, added to the current evidence of the association between these two conditions, points to a new line of integrated research and care. This line should further investigate the interconnections between FA and DM2 and the possible clinical impacts of this relationship on developing and treating these conditions.
Acknowledgements
The authors thank all the women who participated in this study. The authors thank the research team who worked hard to make this study possible. The authors thank the Nutritional Recovery and Education Center (Centro de Recuperação e Educação Nutricional – CREN) of Alagoas for all logistical support.
L. G. R. S. N. was supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brazil (CAPES) research fellowships (grant number: 88882.330651/2019-01). A. E. S. J. is supported by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brazil (CAPES) research fellowships (grant number: 88887.805029/2023-00). N. B. B. is supported by a research grant from the National Council for Scientific and Technological Development – CNPq (grant number: 311401/2022-8).
L. G. R. S. N. contributed to data collection, results interpretation and writing of the manuscript. A. E. S. J. contributed to the conception and design of the study, statistical analysis, interpretation of results and writing of the manuscript. N. B. B. and T. M. M. T. F. contributed to data interpretation, manuscript writing and critical revision of the intellectual content. All authors reviewed and approved the final version of the work.
There are no conflicts of interest.
This study was conducted according to the guidelines in the Declaration of Helsinki, and all procedures involving human subjects/patients were approved by the Research Ethics Committee of the Universidade Federal de Alagoas (protocol number: 4.836.765). Written informed consent was obtained from all subjects/patients.
