Journal of Indian Association of Public Health Dentistry

REVIEW ARTICLE
Year
: 2021  |  Volume : 19  |  Issue : 4  |  Page : 240--254

Association between consumption of carbonated beverages and dental erosion – A systematic review


Abhijit Panda, Hemamalini Rath, Shilpa Mahapatra, Manikesh Mishra 
 Department of Public Health Dentistry, S.C.B Dental College and Hospital, Cuttack, Odisha, India

Correspondence Address:
Abhijit Panda
Department of Public Health Dentistry, S.C.B Dental College and Hospital, Mangalabag, Cuttack - 753 007, Odisha
India

Abstract

Diet and nutritional factors affects oral health in multiple ways. Carbonated beverages have an inherent acidity due to the presence of carbonic acid, phosphoric acid and citric acid that are added to stimulate taste. Dental Erosion is an irreversible loss of dental hard tissues by a chemical process without bacterial involvement. The objective of this systematic review and meta-analysis was to evaluate the association of carbonated beverages with dental erosion. MEDLINE, Science Direct and Google Scholar were searched for articles published between January 2000 to March 2020. Observational studies reporting odds ratios for comparing the risk of dental erosion in patients reporting a frequent consumption of carbonated drink versus those with infrequent consumption were included for review. Studies not reporting odds ratio or with insufficient information to calculate odds ratio were excluded from the study. Newcastle-Ottawa scale was used to assess the quality of the studies. Pooled odds ratios and 95% confidence interval were calculated using a random-effect, generic inverse variance method. A total of 20978 participants from 21 cross sectional studies and 1 cohort study were included in the analysis. The pooled crude odds ratio (19 studies) was found out to be 1.44 (95% CI, 1.23–1.68) and pooled adjusted odd ratio (12 studies) was found out to be 1.86(95% CI, 1.42–2.42). Several clinical and methodological variation among the studies contributed to statistical heterogeneity (I2 = 69%, P ≤ 0.01 for crude odds ratio and I2 = 88%, P ≤ 0.01 for adjusted odds ratio).This systematic review study demonstrates statistically significant associations between consumption of carbonated beverages and dental erosion. Because of the high clinical and methodological heterogeneity estimated pooled effect cannot be relied upon.



How to cite this article:
Panda A, Rath H, Mahapatra S, Mishra M. Association between consumption of carbonated beverages and dental erosion – A systematic review.J Indian Assoc Public Health Dent 2021;19:240-254


How to cite this URL:
Panda A, Rath H, Mahapatra S, Mishra M. Association between consumption of carbonated beverages and dental erosion – A systematic review. J Indian Assoc Public Health Dent [serial online] 2021 [cited 2022 Jan 26 ];19:240-254
Available from: https://www.jiaphd.org/text.asp?2021/19/4/240/332526


Full Text



 Introduction



Dental erosion (DE) is an irreversible loss of dental hard tissues by a chemical process without bacterial involvement.[1] An extensive literature search revealed a global prevalence of DE of permanent teeth in children and adolescents ranging from 20% to 41%.[2] Various risk factors have been identified for tooth erosion.[2],[3] Consumption of carbonated beverages(CB) increases the potential of developing dental erosion.[4],[5],[6] A systematic review done in 2015 to evaluate the influence of diet on dental erosion, showed that consumption of CB had an 69% increase in odds of developing dental erosion.[2] Numerous observational studies have been conducted after that to study the association of DE with CB with Odds Ratio(OR) varying between 0.94 to 3.56.[7],[8],[9],[10],[11] Given the lack of updated systematic review and meta-analysis on the effect of CB on DE and the wide variation in effect observed in recent observational studies, we aimed to systematically review the literature to evaluate the association between CB and DE.

 Search Strategy



Two investigators (AP, MM) independently searched three databases, MEDLINE, Science Direct, and Google Scholar for articles published between January 2000 and March 2020 [Annexure 1]. Studies identified through previous systematic review and cross references were also screened. The studies were screened first by title and abstract. After removing duplicates, full text of all potentially eligible studies was obtained and checked for eligibility by both the investigators. Reviewers resolved disagreement by discussion till they meet a consensus or if necessary with discussion with a third independent reviewer (HR).

Observational studies (cross-sectional or longitudinal) evaluating the association of DE in permanent dentition with consumption of CB reported as crudes or adjusted OR with a reference group comprising of participants who did not or infrequently consumed CB were included in the review. Studies published in English language were considered. Case control studies, randomised and non – randomised controlled trials and studies in which OR has not been reported or could not be extracted from reported data were excluded from the review.

 Data Extraction



A standardized data collection form was used to extract the following information: first author's name, year of publication, country name, age, study design, sampling method, sample size, sample size categories (>1000 and <1000), index used to assess DE, teeth assessed (index/all), tools for dietary record (food frequency questionnaire, brief dietary assessments, 24 h recall), mode of administration of the instrument (self-reported, interviewer administration), category of consumption of CB (weekly, daily, yes or no), calculated crude, and adjusted OR with upper and lower limit. The two investigators mentioned earlier independently performed this data extraction [Table 1].{Table 1}

 Quality Assessment



Assessment of the quality of studies was done using the Newcastle-Ottawa Scale (NOS) separately for cross sectional and cohort studies. Using the tool, each study was judged on eight items, categorized into three groups: the selection of the study groups; the comparability of the groups and the outcome.[12],[13] Studies were judged to be good if the total score was between 7 and 10, fair if total score was between 4 and 6, and poor when total score was between 0 and 4.

Statistical analysis

Statistical analysis was carried out by using R studio (meta and dmetar package). Point estimates and SEs of unadjusted and adjusted ORs were pooled separately using DerSimonian-Laird method and random effect model.[31] Statistical heterogeneity was assessed using the Cochran's Q test and I2 statistic, which quantifies the proportion of the total variation across studies that are due to heterogeneity rather than chance. A value of I2 of 0%–25% represents insignificant heterogeneity, 26%–50% low heterogeneity, 51%–75% moderate heterogeneity, and 75% high heterogeneity.[32]

Sensitivity analysis were performed for studies reporting both crude and adjusted OR and also after removing the outlier study. Sub group analysis and meta-regression were carried out for studies reporting adjusted OR to explain the heterogeneity on the basis of various categorical and continuous study level covariates.

The presence of publication bias was assessed by funnel plots of the logarithm of OR versus their SEs.[33]

 Results



Literature search yielded a total of 2337 articles. About 2220 articles were excluded based on title and abstract for not fulfilling inclusion criteria on the basis of the type of article, study design, population, or outcome of interest. Out of 117 articles, 37 duplicate articles were excluded. About 50 articles were excluded as they did not report the outcomes of interest. Thereafter, 30 studies were identified and included in data extraction and quality checking. Out of 30 eight studies were excluded due to low quality or data could not be extracted. A total number of 22 articles were included in final data analysis (21 cross sectional, 1 cohort) [Figure 1].{Figure 1}

 Assessment of Quality of the Study



Scores of the 21 cross-sectional studies ranged from 3 to 8 on the NOS for cross-sectional studies. About 28% of the studies scored 6 and 24% studies scored 7 on the scale. Score for the cohort study included in the current review was 7 on the NOS for cohort studies [Table 2] and [Table 3].{Table 2}{Table 3}

 Association of Carbonated Beverages and Dental Erosion Using Crude Odds Ratio



Considering crude OR (COR), 19 cross-sectional studies[7],[9],[10],[11],[14],[15],[16],[17],[18],[19],[20],[22],[23],[24],[25],[26],[27],[28],[30] with patients consuming CB were included in the data analysis for the risk of DE. The pooled OR of DE in individuals consuming CB was 1.44 (95% confidence interval [CI]: 1.23–1.68). The statistical heterogeneity was moderate with an I2 of 69%. [Figure 2] shows the forest plot of the included studies.{Figure 2}

 Association of Carbonated Beverages and Dental Erosion Using Adjusted Odds Ratio



Considering adjusted OR (AOR) 12 studies (one cohort studies[21] and 11 cross-sectional studies[7],[8],[11],[14],[15],[16],[17],[18],[23],[26],[29]) with patients consuming CB were included in the data analysis for the risk of DE. The pooled OR of DE in individuals consuming CB was 1.86 (95% CI, 1.42–2.42). The heterogeneity was high with an I2 of 88%. [Figure 3] demonstrates the forest plot of the included studies.{Figure 3}

 Sensitivity Analysis



Results of the sensitivity analysis have been summarised in [Table 4]. Pooled estimates for OR after sensitivity analysis showed significant association between DE and CB (z-value < 0.05). However, the estimates were different from the original analysis. Quantitative estimates could not be relied upon because of high heterogeneity.{Table 4}

 Subgroup Analysis



Subgroup analysis was performed based on various study level categorical covariate contributing to clinical and methodological heterogeneity using all the studies giving information about adjusted OR only [Table 5]. The test for subgroup differences indicated that there was no statistically significant subgroup effect for most of the subgroup analysis, suggesting that most of study level categorical covariates were not able to explain heterogeneity detected in the meta-analysis finding out association between CB and DE. However unequal distribution of the number of studies contributing data to the subgroups might have resulted in the analysis which was not be able to detect subgroup differences because of uneven covariate distribution.{Table 5}

 Meta-regression



Meta-regression was performed to find out the contribution of various study level continuous covariates (like sample size and year of publication) toward heterogeneity in the meta-analysis using adjusted OR. The β-coefficients of bivariate meta-regression was statistically significant only for sample size (β =0.0004, P = 0.012, R2 = 36.2).

 Evaluation of Publication Bias



The funnel plot [Figure 4] had included only 50% of studies deviating grossly from a symmetrical inverted funnel. Majority of the studies were lying on the right side of the funnel and six studies lying outside the funnel.{Figure 4}

 Discussion



Increased availability and reduced cost have resulted in drastic increases of CB consumption all over the world.[22] The carbonic acid formed by dissolution of carbon dioxide enhances the taste and counteracts the sweetness of the CB. Apart from this, other acids such as phosphoric acid and citric acid contribute to the low pH value even after the carbon dioxide has been blown off. Apart from the acidic pH, the erosive potential of a CB depends on many other factors such as the buffering capacity, the chelation property, the frequency, and duration of ingestion. The calcium, phosphate, fluoride content of the beverage, and pretreatment of the dental tissue also seem to be important factors influencing the association between CB and DE.[34],[35]

This current systematic review shows significant association between frequent consumption of CB and DE after pooling data from 21 cross-sectional and one cohort studies in both the meta-analysis models using COR and AOR. Pooled estimate for both COR was calculated to be 1.44 (1.23; 1.64) and confounder OR to be 1.85 (1.42; 2.42). The adjusted OR was found to be more than the COR. This variation in the amount of OR may be attributed to be attributed to the variation in the number of studies contributing data to both type of meta-analysis. Similar finding was also reported in the sensitivity analysis performed considering studies contributing data for both COR and AOR. Although significant association was also obtained in the sensitivity analysis even after excluding the single cohort study, the pooled OR and the values of heterogeneity were different. As the confounder adjusted estimate is of better interest for observational studies, further subgroup analysis was performed using studies showing AOR.

The heterogeneity reported in our meta-analysis was found to be high. There can be several causes contributing to such statistical heterogeneity such as clinical variability, methodological variability, and publication bias. In the present systematic review, the clinical heterogeneity was very high as there was variation in the age group of the selected participant, definition of exposure, definition of comparison group, as well as definition of the outcome DE. Although the age group of most of the study participants was between 11 and 30 years, in one study, there was no upper age limit. Diversity in the dietary assessment method and the tools utilized to evaluate consumption of CB can also be an important factor contributing to heterogeneity. Each instrument has a unique characteristic, advantages, and disadvantages. Four studies used FFQ,[7],[14],[18],[21] 17 studies used BDA,[8],[9],[10],[11],[15],[16],[17],[19],[20],[22],[23],[24],[25],[26],[27],[28],[29] one studies used 24 h recall.[21],[30] The FFQ is usually the preferred tool in such longitudinal assessment. Apart from this, it has the advantages of complete enumeration of consumed items. However, inherent limitations are lack of details with specificity and proneness to memory and recall bias. BDA and 24 h recall, in spite of having the advantage of less time consumption, provide very less information and might not be considered an appropriate tool for detecting minute within person variation of dietary intake.[36] Furthermore, methodological factors during the application of the instrument could also have influenced the assessment of diet such as the extension, the period, the quantity, of times required to record dietary habits, and even the days of the week that the measurement was performed. Most of the studies administered the questionnaire once were lacking in information regarding the exact type of carbonated drink as well as information about portion of consumption and hence were of limited scope. As the recall ability of individuals can vary from day to day, administration of the instrument more than once is advocated which was done in two studies reported in the review. The definition of exposure also varied to a great extent among the studies included in our review based upon the frequency of consumption of the CB.

There was also significant clinical heterogeneity in the measurement of outcome. Since erosion, attrition, and abrasion are difficult to distinguish in their initial stages, lesion can definitely be considered to be a result of an acid challenge confirmed by their localization and morphology.[37] The clinical diagnosis of DE in the studies was made on the basic of characterized deviation from original anatomical tooth morphology caused by acid-induced tissue loss. The earliest sign appears as a smooth silky shiny glazed surface located coronal to cementoenamel junction in smooth surface and nonfunctional occlusal and incisal surfaces area. In more advanced stage, flat regions whose width clearly exceeds its depth become evident, and ultimately, the original tooth morphology is altered to a great extent. The erosive tooth wear can be distinguished from attrition on the basic of their position corresponding to antagonist tooth and abrasion on the basis of their shape (not having v-shape) along with proper clinical history.[38] Several indices used in the studies with their dichotomization criteria of having a clinically evident erosion significantly contribute to heterogeneity. Most of these indices attempted to assess erosion clinically on the basis of two aspects, i.e., the morphological criteria[7],[8],[9],[10],[11],[14],[15],[16],[17],[18],[19],[20],[21],[22],[23],[25],[26],[27],[29],[30] and visual diagnosis of exposed dentine,[24],[28] as it might be difficult to clinically attribute the rounding and the cupping of the cups and groove on the incisal edge and occlusal surfaces erosive tooth wear.[37] Knowledge regarding how much erosive wear is physiological or can be accepted in relation to age.[39] The sensitivity of diagnosis of erosion of such indices becomes questionable. As far as the visual diagnosis of erosive tooth wear is concerned, it might be very difficult and accuracy might be poor.[37] One more methodological difference might be due to inclusion of number of teeth for clinical assessment. Few studies carried out a whole mouth examination but others included only index teeth.

There was a significant variability in the quality of the studies evaluated by NOS, though most of studies were of moderate to high quality.

In the meta-analysis with pooled adjusted OR, the variability might be because of number of variables included in the final model. As there is a significant difference in the number of confounders and adjustment methods among studies in the review, this might be one of the important contributors to statistical heterogeneity.

Subgroup analysis by tools used for measurement of exposure and study design returned with significant results which suggest that some of the heterogeneity can be explained by these two variables. Other study level covariates were unable to explain the statistical heterogeneity. The reason might be the presence of a very few number of studies in each category to illustrate a moderator effect, though there were adequate number of studies (i.e. 10), in the current meta-analysis for conducting subgroup analysis.[31] While interpreting subgroup analysis, it is important to consider number of studies and participant contributing to each subgroup. Although in three subgroup analysis, we acquired a result of low or less heterogeneity in a particular subgroup, presence of high clinical and methodological heterogeneity render the pooled result for the particular subgroup to be meaningless.

Meta-regression using sample size as a covariate also depicted a significant result explaining 36.2% of the overall heterogeneity.

There can be several causes explaining funnel plot asymmetry in the present review such as true heterogeneity (size of effect differs according to study size), reporting biases (publication bias, selective outcome reporting, selective analysis reporting), poor methodological quality leading to spuriously inflated effects in smaller studies (poor methodological design, inadequate analysis, fraud), artifactual (some circumstances sampling variation can lead to an association between the exposure effect and its standard error), chance (asymmetry may occur by chance).[40] Asymmetry should not be equated with publication bias particularly in context of observational studies.[33]

Similar finding was obtained in a previous meta-analysis with meta-regression to explain heterogeneity. The present review has an additional advantage of inclusion of additional 12 studies originating from different part of the world, which can enhance the generalizability. Clinical and methodological variation, publication bias, and statistical heterogeneity fail to produce a robust pooled effect size (OR) proving association between consumption of CB and DE. Subgroup and bivariate meta-regression analysis could not explain statistical heterogeneity because of small number of studies in each subgroup (not significant moderator effect). As most of the studies were cross sectional, it can indicate only association but not causation.

Limitations

Irrespective of getting a significant association between CB and DE, it is difficult to infer causal association, as most of the studies were cross sectional and greatly subjected to confounding. As both the variables are measured at one point of time, it is difficult to ascertain whether only CB consumption is contributing to individual's DE or other covariates could be responsible. This finding may help in facilitating hypothesis initiation to be further tested by various analytical studies. Although prospective cohort studies are better than cross-sectional studies to adjust for confounders, the current meta-analysis was able to include only one cohort study.

As erosion is multifactorial in nature, even after adjusting for covariates such as acidic diet, tooth brushing, frequent bouts of vomiting, frequency of lemon consumption etc. to calculate the AOR but still, there can be a possibility of existing residual confounding.

As all available databases were not searched and due to inclusion of articles published in English only, all published articles evaluating association of DE and CB might not have been included.

Implications for future research

Prospective studies can establish causality of association while eliminating the presence of potential confounders. In the synthesis of results of the current study, only one cohort study was available. In the future, more prospective studies should be conducted to establish the causal association between CB and DE.

 Conclusions



The current systematic review provides evidence that consumption of CB increases the odds of occurrence of DE in permanent dentition. Because of the high methodological heterogeneity pooled effect could not be estimated.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.

 Annexure



Annexure 1: Search methods for identification of studies

Electronic searches

The search strategy was based on the controlled vocabulary (MeSH terms) of the PubMed database along with free keywords; other electronic databases were also used. The detailed search strategy is presented below.

(i)-Search Strategy for Medline

Search (((((((((carbonated beverages) OR carbonated beverage) OR carbonated drink) OR carbonated drinks) AND soft drinks) OR soft drink) OR acidic drinks)) AND ((((((((((((epidemiological studies) OR cross sectional studies) OR cross-sectional study) OR prevalence studies) OR prevalence study) OR cohort study) OR cohort studies) OR incidence study) OR incidence studies) OR follow-up studies) OR observational studies)) AND ((((((((((Tooth wear) OR tooth erosion) OR tooth wears) OR dental wear) OR dental wears) OR tooth erosions) OR erosive tooth wear) OR dental erosion) OR tooth erosive lesions) OR dental erosions)

(ii)-Search strategy for science direct

In advance search

The first row, i.e., “find the article with these term” is filled with the word dental erosion OR erosion OR tooth wear

The last row, i.e., “title, abstract, author specific keyword” is filled with the word soft drink OR acidic drink OR carbonated beverages

(iii)-Search strategy for Google scholar

In advance search

First row, i.e., “with all of the words” is filled with dental erosion

Second row, i.e., “with the exact phrase” is filled with carbonated beverage

Third row, i.e., “with at least one of the words” is filled with soft drink acidic

The last row, i.e., “return article dates between” is filled from 2000

References

1World Health Organization. Risks to Oral Health and Intervention. Available from: https://www.who.int/oral_health/action/risks/en/. [Last accessed on 2008 Jul 03].
2Salas MM, Nascimento GG, Vargas-Ferreira F, Tarquinio SB, Huysmans MC, Demarco FF. Diet influenced tooth erosion prevalence in children and adolescents: Results of a meta-analysis and meta-regression. J Dent 2015;43:865-75.
3Habib M, Hottel TL, Hong L. Prevalence and risk factors of dental erosion in American children. J Clin Pediatr Dent 2013;38:143-8.
4Shenkin JD, Heller KE, Warren JJ, Marshall TA. Soft drink consumption and caries risk in children and adolescents. Gen Dent 2003;51:30-6.
5Imfeld T. Dental erosion. Definition, classification and links. Eur J Oral Sci 1996;104:151-5.
6Lussi A, Jäggi T, Schärer S. The influence of different factors on in vitro enamel erosion. Caries Res 1993;27:387-93.
7Martignon S, López-Macías AM, Bartlett D, Pitts N, Usuga-Vacca M, Gamboa LF, et al. The use of index teeth vs. full mouth in erosive tooth wear to assess risk factors in the diet: A cross-sectional epidemiological study. J Dent 2019;88:103164.
8Skalsky Jarkander M, Grindefjord M, Carlstedt K. Dental erosion, prevalence and risk factors among a group of adolescents in Stockholm County. Eur Arch Paediatr Dent 2018;19:23-31.
9Luciano LC, Ferreira MC, Paschoal MA. Prevalence and factors associated with dental erosion in individuals aged 12-30 years in a northeastern Brazilian city. Clin Cosmet Investig Dent 2017;9:85-91.
10Mafla AC, Cerón-Bastidas XA, Munoz-Ceballos ME, Vallejo-Bravo DC, Fajardo-Santacruz MC. Prevalence and extrinsic risk factors for dental erosion in adolescents. J Clin Pediatr Dent 2017;41:102-11.
11Maharani DA, Zhang S, Gao SS, Chu CH, Rahardjo A. Dental caries and the erosive tooth wear status of 12-year-old children in Jakarta, Indonesia. Int J Environ Res Public Health 2019;16:2994.
12Wells G, Shea B, O'Connell D, Peterson J, Welch V, Losos M, Tugwell P, et al. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies in meta-analyses. 2013. Available from: http://www.ohri.ca/programs/clinical_epidemiology/oxford.as. [Last accessed on 2020 Dec 12].
13Herzog R, Álvarez-Pasquin MJ, Díaz C, Del Barrio JL, Estrada JM, Gil Á. Are healthcare workers' intentions to vaccinate related to their knowledge, beliefs and attitudes? A systematic review. BMC Public Health 2013;13:154.
14González-Aragón Pineda ÁE, Borges-Yáñez SA, Lussi A, Irigoyen-Camacho ME, Angeles Medina F. Prevalence of erosive tooth wear and associated factors in a group of Mexican adolescents. J Am Dent Assoc 2016;147:92-7.
15Hamasha AA, Zawaideh FI, Al-Hadithy RT. Risk indicators associated with dental erosion among Jordanian school children aged 12-14 years of age. Int J Paediatr Dent 2014;24:56-68.
16Kumar S, Acharya S, Mishra P, Debnath N, Vasthare R. Prevalence and risk factors for dental erosion among 11- to 14-year-old school children in South India. J Oral Sci 2013;55:329-36.
17Chrysanthakopoulos NA. Prevalence of tooth erosion and associated factors in 13-16-year old adolescents in Greece. J Clin Exp Dent 2012;4:e160-6.
18Mulic A, Skudutyte-Rysstad R, Tveit AB, Skaare AB. Risk indicators for dental erosive wear among 18-yr-old subjects in Oslo, Norway. Eur J Oral Sci 2012;120:531-8.
19Vargas-Ferreira F, Praetzel JR, Ardenghi TM. Prevalence of tooth erosion and associated factors in 11-14-year-old Brazilian schoolchildren. J Public Health Dent 2011;71:6-12.
20Huew R, Waterhouse PJ, Moynihan PJ, Kometa S, Maguire A. Dental erosion and its association with diet in Libyan schoolchildren. Eur Arch Paediatr Dent 2011;12:234-40.
21Aidi HE, Bronkhorst EM, Huysmans MC, Truin GJ. Factors associated with the incidence of erosive wear in upper incisors and lower first molars: A multifactorial approach. J Dent 2011;39:558-63.
22Gurgel CV, Rios D, de Oliveira TM, Tessarolli V, Carvalho FP, Machado MA. Risk factors for dental erosion in a group of 12- and 16-year-old Brazilian schoolchildren. Int J Paediatr Dent 2011;21:50-7.
23Wang P, Lin HC, Chen JH, Liang HY. The prevalence of dental erosion and associated risk factors in 12-13-year-old school children in Southern China. BMC Public Health 2010;10:478.
24Bardolia P, Burnside G, Ashcroft A, Milosevic A, Goodfellow SA, Rolfe EA, et al. Prevalence and risk indicators of erosion in thirteen- to fourteen-year-olds on the Isle of Man. Caries Res 2010;44:165-8.
25Salas MM, Vargas-Ferreira F, Ardenghi TM, Peres KG, Huysmans MD, Demarco FF. Prevalence and associated factors of tooth erosion in 8 -12-year-old Brazilian Schoolchildren. J Clin Pediatr Dent 2017;41:343-50.
26Waterhouse PJ, Auad SM, Nunn JH, Steen IN, Moynihan PJ. Diet and dental erosion in young people in south-east Brazil. Int J Paediatr Dent 2008;18:353-60.
27Rafeek RN, Marchan S, Eder A, Smith WA. Tooth surface loss in adult subjects attending a university dental clinic in Trinidad. Int Dent J 2006;56:181-6.
28Milosevic A, Bardsley PF, Taylor S. Epidemiological studies of tooth wear and dental erosion in 14-year old children in North West England. Part 2: The association of diet and habits. Br Dent J 2004;197:479-83.
29Dugmore CR, Rock WP. A multifactorial analysis of factors associated with dental erosion. Br Dent J 2004;196:283-6.
30Arnadóttir IB, Saemundsson SR, Holbrook WP. Dental erosion in Icelandic teenagers in relation to dietary and lifestyle factors. Acta Odontol Scand 2003;61:25-8.
31Harrer M, Cuijpers P, Furukawa TA, Ebert D. Doing Meta-Analysis in R: A Hands-on Guide. 2019. Available from: https: //bookdown.org/MathiasHarrer/Doing_Meta_Analysis_in_R/. [Last acessed on 2020 Sep 17].
32Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ 2003;327:557-60.
33Dekkers OM, Vandenbroucke JP, Cevallos M, Renehan AG, Altman DG, Egger M. COSMOS-E: Guidance on conducting systematic reviews and meta-analyses of observational studies of etiology. PLoS Med 2019;16:e1002742.
34Hargitt B. Introduction. In: Steen DP, Ashurst PR, editors. Carbonated Soft Drinks Formulation and Manufacture. Australia: Blackwell Publishing Ltd.; 2006. p. 1-14.
35Prati C, Montebugnoli L, Suppa P, Valdrè G, Mongiorgi R. Permeability and morphology of dentin after erosion induced by acidic drinks. J Periodontol 2003;74:428-36.
36FAO. Dietary Assessment: A Resource Guide to Method Selection and Application in Low Resource Settings. Rome: FAO; 2018.
37Lussi A. Dental erosion clinical diagnosis and case history taking. Eur J Oral Sci 1996;104:191-8.
38Ganss C. Definition of erosion and links to tooth wear. In: Lussi A, editor. Dental Erosion From Diagnosis to Therapy, Monographs in Oral Science. Basel: Karger; 2006. p. 9-16.
39Bartlett D, Ganss C, Lussi A. Basic Erosive Wear Examination (BEWE): A new scoring system for scientific and clinical needs. Clin Oral Investig 2008;12 Suppl 1:S65-8.
40Sterne JA, Sutton AJ, Ioannidis JP, Terrin N, Jones DR, Lau J, et al. Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials. BMJ 2011;343:d4002.