Systematic review of longitudinal studies on the association between cluster of health-related behaviors and tooth loss among adults

Objectives To systematically review longitudinal studies on the association between cluster of/multiple health-related behaviors and tooth loss among adults. Materials and methods Inclusion criteria were prospective and retrospective longitudinal studies; adults; multiple or cluster of behaviors; tooth loss, one or more tooth lost and complete tooth loss. Exclusion criteria were intervention studies; cross-sectional studies; case-control studies; children under 18years-old; single behavior. Two reviewers searched three databases up to April 2023. Open Grey and Google Scholar were searched for grey literature. Results Twelve longitudinal studies were included in this review. Nine studies had good quality, two had poor quality, and one had fair quality according to New-Castle-Ottawa Scale. According to ROBINS-E tool, nine studies were judged as moderate risk of bias while two studies were at low risk of bias and one study had serious risk of bias. One study assessed cluster of behavior, while others examined a number of separate health-related behaviors in relation to tooth loss. Meta-analysis was not feasible because of the high heterogeneity in exposure, measure of outcomes, covariates, sample size, and follow-up time. The research found an association between tooth loss and oral hygiene practices (two studies), dental attendance (four studies), smoking (six studies), and alcohol consumption (three studies). Conclusion This review provides evidence of a longitudinal association between cluster of/multiple health related-behaviors and tooth loss.


Introduction
One of the important indicators of poor oral health is the number of teeth present in the oral cavity.Tooth loss can have a significant impact on quality of life, can lead to inadequate intake of essential nutrients, and discourage people from engaging in social events [1].Accounting for the known behavioral factors that affect tooth loss and the common risk factors for oral health and general health have been the main focus of preventive oral health research [2].There are multiple factors that contribute to tooth loss including oral conditions such as periodontal diseases and dental caries [3,4], and general conditions [5].These factors can be increased by engaging in risk behaviors.Health-risk behaviors are defined as any harmful act that can increase the possibilities of diseases or delay healing [6].Several studies have found a correlation between tooth loss and oral health behaviors such as toothbrushing, and frequency of dental visits [7].Others have suggested that tooth loss has a strong link with other behavioral factors such as smoking [8,9] and consumption of alcoholic drinks [10,11].However, other behavioral factors could have an indirect and less obvious association with tooth loss.Recent research findings reported an association between better oral health and physical activity, and argued that being active promotes a positive inflammatory response and potentially protects against tooth loss [12].Other findings suggested a link between diet and oral health and reported that consumption of fruits and vegetables may halt periodontitis progression and may prevent tooth loss [13].
Risk behaviors frequently occur in groups and clusters rather than in isolation [14,15].Engaging in multiple risk behaviors is more common among people with poorer oral health [16,17].Exploring the effect of multiple health-related behaviors rather than single behavior provides a more comprehensive view of the factors that contribute to tooth loss.This is particularly important due to the complexity of the determinants of tooth loss which could include different factors such as diet, lifestyle, and overall health.For instance, researchers have found that people who smoke and have poor oral hygiene are at higher risk of tooth loss [18,19].Developing an understanding of these combined risks and how health behaviors co-occur (different behaviors occurring among the same population) or cluster (different behaviors cluster together among certain population) is important to gain knowledge about the contributing effect of different health-related behaviors on tooth loss.Previous systematic reviews have assessed the relationship between oral health and individual behaviors such as smoking [18,20], beverage consumption [21,22], diet [13,23], and physical activity [24].However, previous reviews mainly focused on one behavior at a time.Little is known about the effect of multiple behaviors on tooth loss among the same population.To our knowledge, there is no systematic review that assessed the longitudinal association between multiple health-related behaviors and tooth loss in the same population.Thus, the aim of this paper is to systematically review existing longitudinal studies on the association between cluster of/multiple health-related behaviors and tooth loss among adults.

Materials and methods
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA).The study protocol was listed in the International Prospective Register of Systematic Reviews (PROSPERO) (Registration number CRD42022367174).

Criteria for study consideration
Eligibility was determined based on PRISMA 2020 guidelines, which include PECO, as 'P, ' population 'adults' (aged 18 and over at baseline), 'E, ' exposure 'cluster of/multiple health-related behaviors, ' 'C, ' comparison group 'not exposed to multiple behaviors' , and 'O, ' outcome 'tooth loss' .

Inclusion criteria: •
Prospective and retrospective longitudinal studies.

•
Participants aged 18 years-old and more at baseline.

•
Exposure included cluster or multiple health-related behaviors.

•
The outcome is tooth loss, one or more tooth lost, and complete tooth loss (clinically examined or self-reported).

Exclusion criteria:
• Randomized and non-randomized controlled trails.

Study selection and data extraction
The literature review was conducted by two independent reviewers through three databases (MEDLINE via PubMed, EMBASE via Ovid and LILACS via BIREME) using Medical Subject Headings (MeSH) terms and text words around the main topics: the outcome (tooth loss) and exposure (cluster of health-related behaviors).The search in each data source was until April 2023.The three databases (MEDLINE, EMBASE, and LILACS) were deemed to be sufficient to identify relevant literature and consistent with the common recommendation for systematic reviews for searching biomedical literature.The articles were filtered for relevancy by their titles and by their abstracts before they were finally included by reading the full article.All relevant papers were referenced using Endnote X9.

Selection process
Two reviewers independently evaluated eligibility in a standardized, blinded manner.Following the PRISMA flow diagram, a flowchart was made that shows the number of studies at each stage of the evaluation and the reasons for exclusion after determining their eligibility.Reviewers' disagreements were settled through discussion and by a decision of a third reviewer to reach an agreement.

Data collection process
Two reviewers extracted data on the study design, authors, publication year, country, participants' characteristics (sample size, age, follow-up duration), exposures including cluster or multiple health-related behaviors (oral hygiene practices, dental service utilization, smoking status, and alcohol intake), outcomes (complete tooth loss and/or partial tooth loss), covariates including socioeconomic condition (education), and social networking (marital status and living alone), results, and conclusions were all collected from the included publications.

Study risk of bias assessment
Two independent reviewers evaluated the included studies' risk of bias using the New-castle-Ottawa Scale (NOS) for longitudinal research [25].This tool uses three groups divided into eight criteria to evaluate each study.First group is the selection of the study sample (representativeness of exposed, selection of non-exposed, ascertainment of exposure, and determining if the outcome was not present at the start).The second group is the comparability (controlling for confounders and additional factors).The last group is relevant to the outcome of interest (assessment of outcomes, follow-up length, adequacy of follow-up).There are three quality levels: good, fair, and poor.For each of the article, the assessment groups are assigned between zero to nine stars, with a higher number indicating higher-quality research.A good quality score is three to four stars for the selection, one to two stars for the comparability, and two to three stars for the result group.Two stars in the selection group, one or two stars in the comparability group, and two or three stars in the outcome group, are considered to be of fair quality.A study is considered to be of poor quality if it receives a score of zero or one in any of the categories of selection, comparability, outcome, or exposure.
To ensure adequate assessment of risk of bias, we also used Risk of Bias in Nonrandomized Studies of Exposures (ROBINS-E) tool [26].Each study was assessed on seven items of the ROBINS-E: (1) confounding bias, (2) selection bias, (3) exposure bias, (4) departures from intended exposures bias, (5) missing data bias, (6) measurement of outcomes bias, (7) reported result bias.The ROBINS-E questions were answered with options 'Yes, ' 'Probably yes, ' 'Probably no, ' or 'No.' before finally judging the risk of bias at study-level and at item-level as 'low, ' 'moderate, ' 'serious, ' or 'critical' .

Effect measures
Odds ratio, relative risk, rate ratio, and regression coefficient were used to represent the effect measures for the outcome (tooth loss).

Synthesis method
Data were gathered from the different studies on exposure (cluster or multiple health-related behaviors), outcome (tooth loss), covariates (socioeconomic condition and social networking), and the effect measure.We also gathered information on follow-up time, sample size, and demographic variables.The review adopted a qualitative synthesis, and individual studies were presented in tables that display the characteristics and the results of the selected papers.It was not possible to do a meta-analysis of the included studies due to their high degree of heterogeneity, notably differences in exposure (health behaviors), measure of outcome (total tooth loss, partial tooth loss), covariate, sample sizes, and follow-up times.

Report bias of assessment
risk of bias was assessed according to the New-Castle-Ottawa Scale (NOS) for longitudinal research [25] which looked for the outcome and the adequacy of follow-up time and ROBINS-E tool [26] which assessed different sources of bias, such as confounders, exposures, and selection measurements bias.The bias of unpublished research was assessed by searching for grey literature for unpublished results.

Certainty assessment
Confidence intervals was reported for all the studies.

Study selection
The study selection summary is shown in Figure 1.A total of 721 references were identified from three databases (MEDLINE via PubMed, EMBASE via Ovid and LILACS via BIREME).After duplicates were removed, 669 references were included for the title and abstract screening, and 547 articles were found to be irrelevant.Only 12 of the remaining 122 studies met the inclusion criteria after the full reports were evaluated.The number of studies at each stage of the review and the exclusion criteria are depicted in a flowchart (Figure 1).At the final stage, some related papers were excluded because some of them used a cross-sectional analysis of a longitudinal data, and some used a general assessment of oral health rather than a specific assessment of tooth loss.Table 1 displays the included studies' methodological evaluations based on (NOS) criteria.Table 2 shows the individual studies risk of bias using ROBINS-E tool.

Study criteria
Table 3 shows the characteristics of the included studies.The studies included adults aged 40 and above except for two studies which included participants who were in their 20s at the baseline [28,32].Most studies had a follow-up period of 10 years and more except for five which had 4, 5, 6, 7, and 8 years of follow-up [30,32,35,36,38].Tooth loss was clinically examined in 7 studies [27,28,[31][32][33][34][35], and self-reported in 5 studies [29,30,[36][37][38].Tooth loss was defined as the number of missing teeth in 5 studies [27,28,31,33,38], ratio of teeth lost at the follow-up time to teeth present at base line in 1 study [35], complete tooth loss in 1 study [30], missing one or more teeth in 1 study [32], loss of 3 teeth or more during the follow-up period in 1 study [34], lost any teeth in 1 study [29], and lost many or all teeth in 2 studies [36,37].The major health-related behaviors included in the studies were smoking, alcohol consumption, oral health habits (toothbrushing, flossing and professional oral hygiene prophylaxis) and dental visits.

Risk of bias in the included studies
According to the Newcastle-Ottawa Scale (NOS), nine of the included papers were considered good quality (scored 9,7,6) [27,29,31,[33][34][35][36][37][38], two poor quality (scored 6 with low score in the outcome section) [30,32], and one fair quality (scored 7 with 2 stars in the selection group) [28].The differences in the score were because of differences in the selection domain which requires 3 or 4 stars to be considered good quality while the poor score indicates having 0 or 1 star in the outcome domain.

Cluster of behaviors
One study used cluster analysis of oral hygiene behaviors to determine whether certain preventive practice combinations had a stronger association with tooth retention than others [28].The study divided patients into 5 groups '(1) people who brush at least once per day; (2) people who brush at least once per day and get annual professional oral hygiene prophylaxis but do not floss; (3) people who flossed and brush their teeth at least once a day but did not get any regular prophylaxis; (4) individuals who engaged in each of the three recommended practices; and (5) those who did none of them.' There was an almost 49%, 63%, 56%, and 67% reduction in the risk of tooth loss among men who brushed alone, who brushed and received prophylaxis (professional oral hygiene) regularly, who brushed and used floss only, and who brushed, flossed and received regular dental prophylaxis compared to non-regular participant, respectively.In addition to oral hygiene behaviors, the researcher also examined other behavior such as smoking.It was found that persistent smokers (using a pack or more daily) had 1.9 times more likely to lose teeth compared to non-smokers.

Multiple behaviors
Five other papers included oral hygiene practices along with other behaviors such as smoking [31,34,35] and dental visits [27,31,32,34,35].In a Japanese study, it was found that frequency of toothbrushing had no association with tooth loss.However, there was a significant association between tooth loss and no regular dental visits when adjusted for periodontal treatment [31].The study also assessed smoking and found a positive association between being a current smoker and the number of missing teeth.Another study conducted among the same population examined oral hygiene behaviors and dental visits and found no significant associations with the frequency of brushing or dental visits in relation to the number of missing teeth over time [27].A different study conducted among Swedish population included toothbrushing frequency and smoking but did not report their association with tooth loss [34].However, the study included other behaviors such as dental visits and the use of interdental devices, but the respective results were not reported in the multivariate model, and they were only reported in the crude model with no association.A Brazilian study examined risk factors for tooth loss and found an association with dental visits, but no association with dental floss in relation to tooth loss [32].However, these associations were only reported in the crude model.A different study conducted among Brazilian population explored different predictors for tooth loss and found no associations between tooth loss and multiple behaviors including (smoking, access to health services, and toothbrushing frequency) [35].
Two Swedish studies assessed predictors of tooth loss and reported that tooth loss was positively associated with refraining from dental visits [29,37].The studies also discussed smoking and found that being a current smoker was significantly associated with tooth loss.
Two studies reported alcohol consumption along with other behaviors including dental visits [30] and smoking [30,33].The first study used data from USA Health Retirement Study examined factors associated with becoming edentulous and reported that being an alcohol drinker has a protective effect on tooth loss compared to being a non-drinker (OR: 0.75) [30].The same study found that compared to regular dental attenders, non-attendants had nearly three times 2021 [36] tooth loss (self-reported as missing many teeth and edentate).

Education
Education, used as an indicator of early life course social and recoded into (0) higher education, (1) lower education, and other category (5).
Frequency of dental attendance recorded as (0) Attendance at least once a year, (1) less than once a year.
Use of fluoridated toothpaste recorded into (0) at least twice daily and (1) daily or less than daily.toothbrushing recorded into (0) ≤ once daily, and (1) t least twice daily.
Use of alcohol recorded into (1) several times a week, and (0) more seldom.
in norwegian cohort: Smoking: (OR = 0.5: 95% ci: 0.4-0.6).Alcohol: (OR = 1.3: 95% ci: the risk of becoming edentulous [30].Furthermore, the study also examined smoking and found that current smokers were more likely to become edentulous than individuals who never smoked.The second study conducted among two different USA populations showed that alcohol consumption was associated with tooth loss as a risk factor when the populations combined [33].However, when comparing the two cohorts separately, the direction of the relation was different showing that alcohol consumption was a risk factor for tooth loss in one population whilst in the other it had a protective effect.The same study also found an association between smoking and tooth loss among both populations. Two studies discussed multiple behaviors as mediators of the association between social factors and tooth loss [36,38].The first study demonstrated that multiple behaviors (including smoking, oral hygiene practices, alcohol drinking, and dental visits) were associated with tooth loss and education across the cohorts [36].The results showed education gradient among Swedish and Norwegian cohort participants who reported tooth loss, with the prevalence of tooth loss higher among those with lower education levels.Adjusting for behaviors attenuated the education gradients in tooth loss especially among Norwegian cohort participants.The second study conducted among Chinese older adults showed that social isolation was significantly associated with fewer remaining teeth, and only tooth brushing frequency was significantly associated with the number of remaining teeth [38].After the adjustment for mediators including all health behaviors (smoking, alcohol intake, tooth brushing frequency) the association did not change and remained significant.

Results of syntheses
In general, the review's findings showed that tooth loss is influenced by multiple behaviors.According to (NOS), nine of the included studies were found to be of good quality [27,29,31,[33][34][35][36][37][38], two had poor quality [30,32], and one had fair quality [28].According to ROBINS-E risk of bias tool, nine studies were judged as at moderate risk of bias [27,[29][30][31][32][35][36][37][38], one study at serious risk of bias [28], and two studies as at low risk of bias [33,34].Meta-analysis was not possible because of the high heterogeneity of the included studies regarding differences in exposure (health behaviors), measure of outcome (total tooth loss, partial tooth loss), covariates (socioeconomic factors and social networking), sample sizes, and follow-up times.Only one study used cluster analysis to explore the effect of oral hygiene practices on tooth loss [28], and two studies explored the mediating role of multiple behaviors [36,38].

Discussion
This systematic review found twelve longitudinal studies from Japan, China, Sweden, Norway, Brazil, and the United States which examined the relationship between cluster of/multiple health-related behaviors and tooth loss.Overall, the research showed evidence of a longitudinal link between cluster of/ multiple health-related behaviors and tooth loss.
To our knowledge, no other systematic review assessed the longitudinal relationship between cluster or multiple health-related behaviors and tooth loss.However, there are multiple systematic reviews that examined the association between individual behavior and tooth loss [13,18,[20][21][22][23]39,40].It is worth noting that there was unbalance between identified studies that assessed multiple behaviors and studies that used cluster of behaviors.Only one study used cluster analysis to explore the association between oral health behaviors and tooth retention [28], and the rest of the included studies (eleven studies) explored the effect of multiple behavioral factors on tooth loss in the same population.Hence, most of the discussion is relevant to multiple behaviors in the same population.
In this review we found one study suggesting that men who used multiple hygiene practices including brushing, flossing, and regular professional hygiene prophylaxis had better chances of retaining more teeth [28] while four studies found no association between toothbrushing and tooth loss [27,31,35,36].Another study reported that more frequent tooth brushing was related to higher chances of having more remaining teeth [38].There are no systematic reviews linking oral hygiene practices and tooth loss.However different reviews have reported that good oral hygiene is related to lower chances of poor oral health outcomes such as dental caries and periodontal diseases [39,40].One systematic review of observational studies showed that fair and poor oral hygiene practices increase the risk of periodontal diseases [39].While a Cochrane review stated that the use of interdental devices in addition to toothbrushing may reduce gingival diseases more effectively than toothbrushing alone [40].
Previous systematic reviews and meta-analysis of cross-sectional and longitudinal studies found smoking to be a predictor of tooth loss and suggested that smoking cessation may lower the risk of tooth loss [18,20].This is similar to our findings that showed a link between smoking and tooth loss in six studies [28][29][30][31]36,37].It was argued that smoking has an impact on the number of teeth through periodontal diseases [41].While tooth decay is often the main cause of tooth loss, alveolar bone resorption is also an important cause.Hence, nicotine use or cigarette smoking can accelerate alveolar bone loss [42] and smoking cessation may help maintain periodontal and alveolar bone status [43].Others argued that lower socioeconomic factors impact both smoking and tooth loss, for example, those with lower education levels are more likely to smoke compared to people with higher education levels [44].This implies that tooth loss among smokers might be a result of their socioeconomic conditions.
In this review, we did not find any study that examined other behaviors and different types of beverages in relation to tooth loss.Only alcohol intake was a risk factor for tooth loss in different studies.However, a recent review of longitudinal and cross-sectional studies investigated the relationship between oral health and beverage consumption and found that tooth loss was cross-sectionally associated with sugary beverages, and coffee consumption [22].The review also found that alcohol intake can be a protective factor against tooth loss among certain populations (Denmark and Japanese population) [22], a finding consistent with this review.Alcohol intake was found to be a protective factor against tooth loss in two studies among USA population [30,33].It was argued that consumption of alcohol instead of other beverages, such as sugar-sweetened beverages, could partially explain the observed protective relationship [30].Sugary drinks can be a risk factor for dental caries and if untreated may lead to tooth loss.Another reason could be the type of alcoholic drinks consumed.Previous evidence reported a relationship between different types of alcoholic drinks and social conditions.Wine drinking was linked to higher socioeconomic status [45] and healthier eating habits [46].As a result, the current findings might be because of higher socioeconomic status and a healthier lifestyle rather than being the result of alcohol itself [47].Contradictory, the same systematic review found that alcohol intake can also be a risk factor for tooth loss in other populations (USA, Sweden, Japanese population) [22].Our review found one paper suggesting that alcohol intake was a risk factor for tooth loss among Swedish and Norwegian populations [36].As tooth loss is mainly the result of periodontal diseases and tooth decay, it was suggested that drinking can harm the soft tissues by slowing down the salivary flow and accumulating dental plaque which increases the risk of tooth decay and periodontal problems [48].The flow of saliva aids in the neutralization of acids created by plaque, which helps prevent tooth decay.Lack of saliva enables acids to build up, resulting in gum disease, tooth decay, and periodontal disease [11].
While previous studies linked tooth loss with other behaviors such as diet and physical activity, we did not find any longitudinal study that examined multiple behaviors that included diet and physical activity.However, multiple systematic reviews examined the relationship between fruit and vegetable consumption, as a single behavior, with different oral health outcomes [13,23].Moreover, a recent systematic review reported an association between physical activity, better oral health (periodontal disease), and oral health behaviors [24].
As behavioral factors are associated with both socioeconomic status and tooth loss, they can be potential mediators of the relationship between oral diseases and social status.The review found two studies that assessed the mediating role of multiple behaviors (including smoking, oral hygiene practices, alcohol drinking, and dental visits) between social factors and tooth loss [36,38].The studies explained that behavioral factors could explain in some way the social differences in tooth loss.Previous evidence suggested that behavioral and psychological factors could be potential mediators between social factors and dental diseases and could explain the social inequalities in oral health [49][50][51].
The major impact of tooth loss on oral and general health is a public health concern, especially among older adults.Tooth loss can have a negative impact on the quality of life of patients, change masticatory function, impair nutritional intake, and affect aesthetics [52].Finding different mechanisms for tooth loss impact on older adults can guide our prevention towards more practical strategies and behavioral changes to reduce the negative influences of tooth loss [2].
This systematic review has some limitations that should be addressed.First, it was not possible to perform a meta-analysis because of variances in behavioral characteristics, the measure of outcome, sample size, and follow-up duration that prevented the pooling of the results.Second, the discussion of this review was limited due to the lack of studies that examined the relationship between clustering of health-related behavior and tooth loss.Third, the limitations of the included studies such as the limited generalizability of the sample, the variations of the follow-up period, and the different methods of collection of data, have impacted the quality of the evidence produced by the review.
The review has some implications.The findings highlight the importance of exploring the aggregate effect of multiple behaviors on tooth loss to help maintain teeth at an older age.Given the fact that multiple behaviors affect other health outcomes, it is necessary to address and target multiple behaviors, particularly those that are not directly linked to tooth loss and may also have other health benefits.Future research on cluster of health-related behavior and tooth loss should investigate explanatory pathways to demonstrate how health-behaviors affect oral health over time and assess the impact of cluster of health behaviors on tooth loss.

Conclusion
The majority of previous research on tooth loss have examined the effect of multiple health related behavior and not cluster of behaviors.The overall findings of the review imply that multiple health-related behaviors were longitudinally associated with tooth loss in the same population.The included studies varied according to exposures, sample sizes, and follow-up times.Further research is required to explore the different factors that contribute to the association between health-related behaviors and tooth loss.

Figure 1 .
Figure 1.Flowchart of the selection of studies for the review.

Table 1 .
Methodological assessment of included studies using the newcastle-Ottawa scales (nOS) with converting scales.

Table 2 .
ROBinS-E risk of bias assessment.

Table 3 .
characteristics of the longitudinal studies on the association of cluster health-related behaviors and tooth loss in adults.

Table 4 .
Association between cluster of health-related behavior and tooth loss.