Introduction

Erosive tooth wear (ETW) is an irreversible multifactorial condition associated with behavioural, chemical and biological factors, characterised by progressive loss of tooth structure without bacteria involvement [1, 2]. It is the third most frequently observed oral condition after dental caries and periodontal disease [3]. There is consensus that bulimic behaviour contributes to ETW due to compensatory methods, such as self-induced vomiting practices [4]. However, there is less clear evidence for a direct link between bulimia nervosa (BN) and acidic food choices, which could also contribute to ETW.

Extant studies have typically been conducted with patients already diagnosed with the eating disorder (ED) [4], thereby limiting preventive strategies. We believe that assessing high-risk groups for bulimia nervosa is essential for the development and tailoring of effective strategies to prevent progress towards a full-blown ED and its comorbidities. Evaluating and understanding possible differences in dietary preferences, compensatory practices and dental implications among groups with bulimic symptomatology are both theoretically and clinically relevant.

Previous studies have observed a higher prevalence of ETW among adolescents with severe bulimic behaviour when compared with adolescents without the behaviour [5, 6]. However, they have not focussed in evaluating and understanding the differences in dietary and compensatory practices among groups with varying severities of bulimic symptomology. In fact, the frequency of bingeing and purging practices, which are supposed to be associated with ETW, can be measured by severity scales. This stratification is important, once it allows the analysis of each bulimic pattern and it must be considered during the screening, planning and approach of risk groups [7].

It is known that BN affects more young women than men [8]. Therefore, the overall aim of the present explanatory study was to evaluate dietary habits, purging practices and the presence of ETW among female adolescents with varying severities of bulimic symptomology. Considering the existing scientific evidence pointing self-induced vomiting as a major risk factor for ETW among patients with bulimia [4, 9, 10], our hypothesis is that intrinsic acids (purging practices) are the major cause for this dental implication when compared to extrinsic causes (dietary habits) among this risk group.

Methods

Ethics

The study was approved by the Human Research Ethics Committee of the Universidade Federal de Minas Gerais, Brazil. Informed consents were obtained.

Study design and sample size

The present cross-sectional study was carried out in the city of Belo Horizonte, Southeastern Brazil. The sample size was calculated to give a standard error of 1%, using a confidence interval level of 95% and an 1.8% prevalence of BN [11]. The minimal sample size needed was estimated at 1072 individuals. An additional 20% were added to compensate potential refusals, comprising 1287 female adolescents [5]. Adolescents were randomly selected according to the distribution of female adolescents regularly enrolled in public and private high schools.

From this sample, 72 adolescents, aged 15 to 18 years, were identified with bulimic symptomatology and included in the present study. The Bulimic Investigatory Test of Edinburgh (BITE) was used to identify the symptoms and severity of the bulimic symptomatology. A comprehensive description of the BITE is provided in the ‘Instruments and Settings’ section. Adolescents were divided into three groups, according to the severity of bulimic symptomatology: Group 1 (mild symptomatology), Group 2 (moderate symptomatology) and Group 3 (severe symptomatology) (Fig. 1).

Fig. 1
figure 1

Sample selection criteria

Full size image

Data collection

Data collection occurred at public and private schools, in separate rooms, at previously scheduled times and dates.

Instruments and settings

Risk assessment for bulimic symptomatology

Bulimic symptomatology was evaluated using the validated Brazilian version of BITE [7, 12]. Adolescents completed the instrument at schools. The BITE is a self-report questionnaire, comprising 33 items, used to identify the presence and severity of bulimic symptomatology. It has been used as a screening instrument in epidemiological research, but cannot be used for diagnostic purposes [13]. Participants are asked to answer the questionnaire based on their feelings and behaviour over the past three months.

The instrument consists of two different subscales: a 30-item Symptom Subscale and a 3-item Severity Subscale. The scores for symptoms are indicative of: (1) normal eating pattern; (2) unusual eating habits (which can be divided into unusual eating pattern or a subclinical group); (3) a highly disordered eating pattern and the presence of binge eating (bulimic symptomatology) [7, 13].

The Severity Subscale measures the severity of bingeing and purging behaviour, as defined by its frequency. It is divided into three categories: (1) symptoms not clinically significant (mild); (2) clinically significant bulimic symptoms (moderate); (3) a high degree of severity (severe). A score of 5 or more on this scale is considered clinically significant (Fig. 1) [7].

Assessment of dietary habits and vomiting practices

Dietary information related to acidic diet consumption was assessed by a self-report questionnaire based on a previously validated instrument addressing the frequency of consumption of acidic foods and drinks [14]. To measure adolescents’ daily servings, the sample was asked to use a four-point scale (1 = never, 2 = once or twice per week, 3 = three or four times per week, 4 = five or more times per week). To calculate each adolescent estimated daily servings, the 4-point ratings for each food were converted to daily servings (1 = 0/7 days = 0.000, 2 = 1/7 days = 0.143, 3 = 3/7 days = 0.429, 4 = 5/7 days = 0.714), then the mean daily servings was calculated for each food.

The collection of data related to vomiting practices involved one item on the validated Brazilian version of BITE. The item ‘Do you make yourself vomit to help you lose weight?’ has demonstrated satisfactory internal consistency and reliability [7, 12]. For statistical and theoretical purposes, this item was dichotomised as absent (corresponding to the response options: ‘never’ or ‘rarely’) or present (corresponding to the remaining response options: ‘once a week’, ‘2–3 times a week’, ‘daily’, 2–3 times a day’, ‘5+ times a day’). According to the Diagnostic and Statistical Manual of Mental Disorders 5 (DSM–5) one of the criteria for diagnosing BN is recurrent inappropriate compensatory behaviour to prevent weight gain, such as self-induced vomiting, occurring at least once a week for three months [15]. Therefore, the options ‘never’ and ‘rarely’ were not considered frequencies potentially related to BN symptomology.

Erosive tooth wear examination

ETW examinations were conducted by a single calibrated examiner for the diagnosis of ETW, based on the clinical criteria proposed by O’Sullivan [16]. This index is especially designed for epidemiologic surveys and it records the distribution (site of erosion on each tooth), grades of severity and the area of the tooth surface affected by erosive lesions. Agreement analysis used the Cohen’s Kappa coefficient based on the codes recorded on a tooth-by-tooth basis. Cohen’s Kappa values for intra and inter-examiner agreement were 0.90 and 0.88.

All teeth, except the third molars, were examined with sterilised periodontal probes (WHO-621 Trinity, Campo Mourão, PR, Brazil) and mouth mirrors (PRISMA®, São Paulo, SP, Brazil). During dental examinations, the examiner was blind to the eating behaviour status of the participants.

Data analysis

The data were analysed using the Statistical Package for the Social Sciences (SPSS for Windows, version 24.0, SPSS Inc., Chicago, USA). Kruskal–Wallis and Mann–Whitney tests were used to detect differences in the mean daily consumption of acidic foods and drinks among the three groups with bulimic symptomatology (mild, moderate and severe). A bivariate analysis using the exact versions of the nonparametric Pearson’s chi-squared and Fisher’s tests and Poisson regression models were developed to evaluate the association between erosive tooth wear, vomiting practices and dietary habits. The independent variables were included in the same model in order to control each other, along with the other covariates. Variables with p value ≤ 0.20 in the bivariate analysis or with clinical and epidemiological relevance were included. The level of statistical significance was set at 5%.

Results

The final population consisted of 62 adolescents with different bulimic symptomatology (Group 1, n = 17; Group 2, n = 25; Group 3, n = 20). Ten adolescents (13.9%) had left school after responding the questionnaire and could not be located for the dental examinations. The adolescents were 15–18 years old (30.0% were 15 years old, 24.8% were 16 years old, 34.8% were 17 years old, 10.4% were 18 years old).

The prevalence of ETW differed among adolescents with mild, moderate and severe bulimic pattern (p = 0.001), corresponding to 5.9%, 8.0% and 45.0%, respectively. Pairwise differences were detected in Group 3 (severe bulimic symptomatology): Group 1vs2 (p = 0.795), Group 1vs3 (p = 0.005) and Group 2vs3 (p = 0.007).

With respect to the sites and severity of erosive lesions the most frequently affected surface was the labial and incisal/occlusal edge (26.6%) and the lingual or palatal surfaces were the least affected (6.1%). Regarding the severity of ETW, most affected tooth surfaces exhibited loss of enamel without dentine exposure (55.7%) and 62.5% of the tooth surfaces with ETW presented more than half of the surface affected by the erosive wear (Table 1).

Table 1 Percentage of affected tooth surfaces, severity and area of the surfaces affected by erosive tooth wear
Full size table

In the present sample, only the central incisors, lateral incisors and first molars in the upper and lower jaws were affected by erosive lesions. It was detected that the labial surfaces of upper incisors and occlusal surfaces of maxillary and mandibular first molars were the sites most affected by ETW (Fig. 2).

Fig. 2
figure 2

Percentage of sites, severity and area with erosive lesions according to the tooth affected

Full size image

Regarding vomiting practices, the prevalence of self-induced vomiting was 20.0%, 4.0% and null among adolescents with severe, moderate and mild bulimic symptomatology, respectively (p = 0.060). Significant differences were found among the groups in relation to some dietary habits, such as daily consumption of citric fruits (p < 0.005), diet soda (p < 0.009) and ketchup (p = 0.004) (Table 2).

Table 2 Daily consumptionof acidic foods/drinks by bulimic symptomatology of adolescents
Full size table

The association of self-induced vomiting practices and dietary habits with ETW is shown in Table 3. Self-induced vomiting practice was associated with ETW (p < 0.001). Significant associations were also found between ETW and higher frequency of consumption of citric fruit (p < 0.001), diet soda (p < 0.001) and ketchup (p = 0.021).

Table 3 Bivariate analyses related to the association between erosive tooth wear, vomiting practices and dietary habits (n = 62)
Full size table

In Table 4, an adjusted Poisson regression model showed that adolescents who self-induced vomit at least once a week (PR = 2.42, 95% CI = 1.00–5.86, p = 0.05) and who had higher frequencies of consumption of citric fruits (PR = 7.96, 95% CI = 1.50–42.11, p = 0.015) and diet soda (PR = 2.32, 95% CI = 1.09–4.91, p = 0.029) had higher prevalence of ETW.

Table 4 Unadjusted and adjusted Poisson regression models for associations between erosive tooth wear and vomiting practices and dietary habits (n = 62)
Full size table

Discussion

In the present sample, adolescents with bulimic symptomatology differed with respect to the occurrence of ETW and dietary habits. The prevalence of ETW increased along with the severity for the BN symptomatology. Strong scientific evidence points that the primary factor leading to ETW among patients with EDs is frequent self-induced vomiting [4]. In the present study, this habit was associated with ETW, but no difference was detected between groups in relation to the self-reported frequency of vomiting practices.

Conversely, dietary habits differed significantly between groups. Adolescents who presented severe bulimic symptomatology were identified with higher daily consumption of acidic foods and drinks and higher intakes of citric fruits were the factor mainly associated with higher prevalence of ETW. In addition, it is interesting to notice that the teeth sites most affected by erosive lesions were labial/buccal surfaces of upper anterior teeth and the lingual or palatal surfaces were the least affected. The location of erosive lesions varies according to the source of acid. Among patients who vomit regularly, the distribution of erosion is typically on the palatal surfaces of the upper anterior teeth [10, 17], whereas extrinsic sources of acid, such as acidic diet, typically affect labial surfaces of teeth, specially the incisors, as it has been observed in the present study.

The importance of evaluating dietary habits in patients with BN was reported in 1998, when it was observed that patients with BN were more likely to consume potentially erosive food, such as herbal tea, soft drinks and apple vinegar than healthy controls [18]. The recent findings have detected higher prevalence and severe lesions of ETW in patients with EDs, who eat more sweetened and acidic foods [10, 19].

A clinical study conducted to evaluate the association between self-induced vomiting and ETW found higher prevalence and more severe lesions among individuals with low consumption of acidic beverages than in those with high consumption [20]. However, no statistical analysis was conducted to control for the confounding variables (intrinsic and extrinsic acids) and it is known that both can contribute to the occurrence of ETW [10].

To control confounding variables, the present study conducted an adjusted Poisson regression analysis, which revealed that higher consumption of an acidic food was the major factor related to ETW. In the adjusted model it was observed that the variable ‘ketchup’ lost significance, which occurred due to the collinearity of this item with the consumption of diet soda.

The present sample was grouped according to the severity of the bulimic symptomatology. Interestingly, a research conducted in 2014 also evaluated dietary habits and oral health among patients with EDs, but the sample was grouped based on the ETW severity [19]. This reverse methodology detected significantly more patients with mild ETW lesions consuming substantial amounts of water before self-inducing vomit and more persons with severe lesions routinely consuming sweetened food or carbonated beverages [19].

The present findings brought up a question: why did adolescents with severe bulimic symptoms present a higher consumption of acidic foods? Frequent consumption of carbonated drinks has been identified in patients with EDs to help decreasing the reflex stimulus for hunger, by increasing dilation of the stomach [21,22,23]. Furthermore, it has been reported that consumption of vinegar and lemon juice contributes to eliminate the gustatory mechanism phase of hunger regulation [24]. A study identified that individuals with a lifetime diagnosis of BN consumed substantial amounts of diet soda as compared to controls. It was argued that diet soda was consumed to satiate the appetitive drive without adding calories, combining high weight concerns and high appetitive drive among patients with BN [25].

Another point that must be mentioned is related to the consumption of foods that trigger reflux. In the literature, this dietary connection has not been raised in the field of ED, but the question is worth discussing: do these BN risk groups "learn" to detect which foods contribute to reflux and consume them more frequently to facilitate the induction of vomiting? According to the American Society for Gastrointestinal Endoscopy, some foods are known to trigger symptoms of gastroesophageal reflux, such as citric fruits and juices, tomatoes and processed tomato-based products and even carbonated beverages [26]. Those drinks cause gaseous distension of the stomach and increase the pressure on the lower oesophageal sphincter, causing acid reflux [26, 27].

The present study had several strengths. First, risk behaviour (symptoms) was evaluated instead of researching diagnosed patients. Understanding the habits of risk groups is important to plan strategies and prevent potential harm to the individuals [28, 29]. Validated measures of bulimic symptomatology were used and adjusted analyses were conducted to control for confounding variables.

Some limitations also merit discussion. First, self-induced vomiting practices and its frequency were based on self-reports according to the last three months and it might be associated with uncertainties [29]. Thus, the absence of significant associations related to vomiting practices among the participants might have occurred because self-report was flawed and not precise enough for accurate stratification of purging frequency. In addition, gastroesophageal reflux, which is an important confounding factor for ETW, was not evaluated and the frequency of dietary habits was stipulated according to the food consumption during a week, instead of daily/non-daily frequency of consumption. We also acknowledge that other potentially erosive foods and beverages might have not been included in the present data collection. Finally, the present study was conducted among a non-specific population; due to the small number of outcomes identified no extrapolation of the present findings is possible.

The primary factor leading to ETW among patients with BN is frequent self-induced vomiting, but the present findings emphasise the need for also considering the compounding effect of extrinsic sources of acids. The frequency and pattern of consumption of acidic diets have been pointed out as the major aetiological factors in the prevalence and incidence of ETW [30,31,32]. Even though, there is scientific evidence supporting that some dietary components, such as acid snacks/sweets, carbonated beverages and natural fruit juices increase the prevalence of ETW, whereas milk presents a protective effect [33].

In the present sample, higher consumption of citric fruits was the main factor associated with higher prevalence of ETW. However, despite knowing that acidic foods, such as citric fruits seem to have an association with ETW, it must be clear that the consumption of fruits should not be discouraged, when it is part of a balanced diet. The World Health Organisation recommends a consumption of at least 400 g of fruits and vegetables per day, in order to avoid the onset of chronic conditions [34]. Therefore, monitoring dietary habits is clinically valuable and strategies must be properly addressed to prevent ETW (see Box 1).

It is important to know when, how, how often and how much a beverage or foodstuff is ingested. Therefore, an accurate clinical examination and anamnesis must be conducted by the clinician, to identify possible erosion-related risk factors and plan strategies to control them [35]. The patient should be instructed to record a food diary for a distinct period (e.g., 4 days), including specific eating and drinking habits. Health care providers should check whether the food diary suggests that extrinsic erosive sources are an important causal factor, then dietary counselling should be provided [36] (see Box 1).

It is known that behaviour changes are not easily achieved. However, offering alternatives and targeted behavioural interventions may increase treatment success [30]. Monitoring dietary habits might be clinically valuable, especially considering persons who self-induce vomit, once the diet can exacerbate the medical consequences of the disorder.

What is already known on this subject?

There is consensus that the primary factor leading to erosive tooth wear among patients with bulimia nervosa is frequent self-induced vomiting. However, extensive studies have typically been conducted with patients already diagnosed with bulimia nervosa, thereby limiting knowledge related to high-risk groups.

What this study adds?

A higher consumption of citric fruits was the main factor associated to higher prevalence of erosive tooth wear among the sample. It was the food choices and not purging practices that differed among adolescents with varying severity of bulimic symptomology. These findings emphasise the need for also considering the compounding effect of extrinsic sources of acids.