How Common is Dry Mouth ? Systematic Rev iew and Meta-Regress ion Analysis of Prevalence Estimates

prevalence of xerostomia/hyposalivation in epidemiological studies. An electronic search was carried out up to February 2018 with no language restrictions. A total of 5760 titles were screened and just twenty-nine papers were included in review and the meta-analysis after a two independently reviewers applied the selection criteria. Data were extracted from PubMed and Web of Science databases. Eligibility criteria included original investigations from observational population-based studies that reported the prevalence of xerostomia or data that allowed the calculation of prevalence of xerostomia and/or hyposalivation. Studies conducted in samples with specific health conditions, literature reviews, case reports and anthropological studies, as conferences or comments were excluded. Sample size, geographic location of the study, study design, age of the studied population, diagnosis methods, and evaluation criteria used to determine xerostomia e/or hyposalivation were extracted for meta-analysis and meta-regression. Multivariate meta-regression analysis was performed to explore heterogeneity among studies. The overall estimated prevalence of dry mouth was 22.0% (95%CI 17.0-26.0%). Higher prevalence of xerostomia was observed in studies conducted only with elderly people. Despite diverse approaches to the condition’s measurement, just over one in four people suffer from xerostomia, with higher rates observed among older people. Moreover, the measurement methods used currently may overor underestimate xerostomia. These findings highlight the need for further work on existing and new clinical measure and will be useful to determine which one is more reliable in clinical and epidemiological perspectives. How Common is Dry Mouth? Systematic Rev iew and Meta-Regress ion Analysis of Prevalence Estimates


Introduction
Xerostomia and hyposalivation are two distinct and independent phenomena, which may manifest alone or in combination (1).Xerostomia is defined as the subjective sensation of dry mouth and it is diagnosed through selfreport (2).On the other hand, hyposalivation refers to an objectively measured low salivary flow (3).Either dry mouth situations may negatively affect oral health-related quality of life (4,5) and may cause oral health problems such as halitosis, impaired chewing and swallowing, and difficulties in prosthesis retention (6).There is evidence that lower salivary flow increases the risk of dental caries, due to an absence of the physical cleaning action and buffering capacity of saliva (7).This increasing risk is not exclusive of elderly and can occur in early age, specially in asthmatic patients under treatment.
Epidemiological studies of dry mouth situations from the last two decades have shown prevalence estimates ranging from 1% (8) to 62% (1).The high variability in estimates has been attributed to variations in measurement methods, populations investigated, sample representativeness, study design, and the age of individuals evaluated (9).Despite the research effort, much remains unanswered about the epidemiology of dry mouth.Even xerostomia that is measured through self-report (2,10), there are methodological differences, whereby some studies report only the presence or absence of dry mouth (7,11,12) and others investigate the frequency of such a sensation (5,13,14).New approaches use scales, such as the Xerostomia Inventory (XI) developed and tested by Thomson et al. (15), which includes a battery of seven xerostomia questions used by Locker (16) or a nine-item battery of questions on dry mouth-related symptoms and behaviors (17).
Hyposalivation is diagnosed through the assessment of salivary flow, which may be evaluated by collecting the fluid from individual salivary glands (or pairs of glands) and; also, total salivary flow may be evaluated by collecting whole saliva.The latter method is more common (9).However, there is a lack of consensus in the literature about which salivary flow rate indicates hyposalivation, ranging from less than 0.1mL/min unstimulated saliva (15) to 0.8 mL/ min stimulated saliva (18).
There is evidence of health conditions and risk behavior being determinants of xerostomia and hyposalivation.

Overall prevalence of dry mouth
Increasing age has been reported as a risk marker for xerostomia (11,19).Chronic diseases such as diabetes, autoimmune diseases, especially Sjögren's Syndrome (12), and polypharmacy (10,20,21) are recognized as major dry mouth associations.Not only polypharmacy, but specific medicines favor the occurrence of xerostomia and hyposalivation, especially inhaled antiastmatic drugs (22).The action of such medicines, even in early ages can significantly affect salivary flow.Considering the widespread occurrence of asthma since childhood (23) and the impact of its therapy in dry mouth, the effects of it therapy in a long-term are not clear and it must be considered in all ages as well.
A previous systematic review evaluating the prevalence of xerostomia in population samples was published in 2006 (24).However, that study included only estimates for xerostomia.Moreover, the study did not use a metaanalysis, so it was unable to obtain an overall prevalence estimate for xerostomia.Many epidemiological studies on dry mouth have been published since then.
This paper presents a systematic review of the literature on the prevalence of xerostomia and hyposalivation in order to obtain a global combined prevalence estimate for dry mouth, and to determine the factors behind the considerable variability in prevalence estimates.

Material and Methods
This systematic review was organized using the PRISMA statement and it was based on the following review question: "What is the estimated worldwide prevalence of xerostomia/hyposalivation?"

Eligibility Criteria
Original investigations from observational studies that reported the prevalence of xerostomia were included.Only population-based studies with representative samples, according to the Critical Appraisal Checklist for prevalence studies recommended by the Joanna Briggs Institute, were considered for this review.As a qualifying condition, all selected studies should have clearly reported the prevalence of xerostomia and/or hyposalivation or have included data allowing its calculation.
Studies conducted in samples with specific health conditions (asthma, cancer, depression, paralysis, syndromes -including Sjögren's Syndrome, and similar convenience samples) were excluded, as were literature reviews, case-control studies, retrospective studies, case reports, anthropological studies, in vitro and in situ studies, and comments or conference abstracts.Articles in other languages than English, Spanish, Portuguese, French or German were excluded.

Search Strategy
An electronic search was performed in the PubMed and Web of Science databases, with no initial date and language restrictions.Keywords included the following MeSH and free terms: (Xerostomia(Mesh)) OR (Xerostomia (all)) OR (Dry Mouth(all)) OR (Mouth Dryness(all)) OR (Hyposalivation(all)) AND Epidemiologic Studies(Mesh)) AND (Cross-Sectional Studies(Mesh)) AND (Longitudinal Studies(Mesh)) AND (Cohort Studies(Mesh)), which are presented in Table 1 in several combinations.
Reports were managed using the EndNote X7.4 software (Thomson Reuters, New York, NY, USA).Duplicate reports were excluded.Two reviewers (GOC and ERS) independently screened titles and abstracts, based on the aforementioned criteria.The screened lists were compared and differences were discussed and resolved by consensus.If there was no consensus, a third examiner was asked to decide on the inclusion or exclusion of the study.The same two reviewers also screened full text manuscripts.Reference lists from the eligible papers were reviewed according to the eligibility criteria.Gray Literature was not screened.

Data Extraction and Data Analysis
Information extracted from the studies included sample size, geographic location, study design, age of the studied population, diagnosis methods, and evaluation criteria used to define xerostomia and/or hyposalivation.Prevalence rates for xerostomia/hyposalivation were also collected (or calculated, if necessary).In case of missing data, the authors were contacted up by e-mail.When more than one method for measuring dry mouth was employed, prevalence rate of hyposalivation was included in the meta-analysis for dry mouth.Prevalence rates were categorized according to the age of the participants (adults or elders), when more than one available, the highest was used; however, when such information was not available, studies were grouped into mixed population (adults and elders).Considering that cohort studies could showed prevalences of xerostomia or hyposalivation in more than 1 time, it was established the use of the most recent values of prevalences.
The estimated global prevalence of dry mouth was calculated using fixed-and random-effect models.When heterogeneity was present (I2>50% or chi-square p value<0.05), the random-effect model was favored (25).The same criteria were adopted for individual meta-analysis of xerostomia and hyposalivation information of each study.Additionally, metaregression and subgroup analyses were conducted to investigate sources of between-study variability for each criterion.Characteristics were included in a multivariate meta-regression model.Variable selection was performed using the backward stepwise approach.Variables with p value<0.20 remained in the final model, but only those with p value<0.05 were considered significant in the final adjusted model.Explained heterogeneity was obtained from the adjusted R2 of the final model.Subgroup analysis was also conducted for each methodological variable included in the final meta-regression model.Sensitivity analyses were conducted to estimate the influence of each study on the pooled results.Funnel plot and the Egger test were used to test for any potential publication bias (17).All analyses were performed using the Stata 14.1 software (StataCorp, College Station, TX, USA).

Results
Electronic searches revealed 5760 studies.From those, 1346 were excluded for being duplicates.A total of 4414 articles were submitted to title and abstract screening and 114 of them remained for full-text evaluation, from which 85 were excluded after appraisal (Fig. 1, Table 2).A total of 29 articles met the inclusion criteria, among these articles, 26 reported the prevalence of xerostomia or data on it and 14 reported the prevalence of hyposalivation  or data on it.One study was included twice in the metaanalysis, because it presented separate data for different populations under study (14).The overall prevalence of xerostomia was estimated to be 23.0%(95%CI 18.0-28.0%),with high heterogeneity among studies (I2 99.8%; chi square p value<0.001; Fig. 2).The overall prevalence of hyposalivation was estimated to be 20.0%(95%CI 15.0 -25.0%) with high heterogeneity among studies (I2 99.4%; chi square p value<0.001; Fig. 3).The overall prevalence of dry mouth (xerostomia or hyposalivation) was estimated to be 22.0% (95%CI 17.0-26.0%),also with high heterogeneity among studies (I2 99.8%; chi square p-value<0.001;Fig. 4).

Overall prevalence of dry mouth
The final meta-regression analysis explained about 16% of the between-study variability.
Table 3 presents the main characteristics of the included studies.Some studies presented weaknesses under critical appraisal (Table 4), as follows: two studies (12,26) did not use an adequate sample frame for the target population, two studies (7,27) did not have an adequate sample size or were unclear, one study (3) did not describe their participants and the setting in detail, one study (28) did not conduct the data analysis with sufficient coverage of the identified sample did not avoid coverage bias in data analysis, one study (12) did not use an appropriate statistical analysis, and one study (7) did not have a sufficient participation rate.
Tables 5 and 6 shows the subgroup analysis according to the variables retained in the final adjusted meta-   Overall prevalence of dry mouth regression models.A higher prevalence estimate for dry mouth was noted in studies conducted with older people only and in studies conduceted in Americas (Table 5).The heterogeneity of the studies for each outcome was higher than 99.0%.The Egger test revealed the presence of publication bias (p-value=0.007),which was confirmed by the metafunnel analysis (Fig. 5).Sensitivity analysis demonstrated that the omission of any study would not significantly modify the estimated prevalence of dry mouth (Fig. 6).

Discussion
This meta-analysis of findings from epidemiological studies on dry mouth has found the overall estimated prevalence of dry-mouth from population-based studies tov be 22.0% (95%CI 17.0-26.0).Xerostomia and hyposalivation are two phenomena that may negatively affect the oral health of individuals and their quality of life (4,5,13).Dry mouth impairs oral function, chewing, and swallowing (29).Considering oral diseases, evidences from different study designs researches highlight the impact of salivary problems in caries experience (30,31).Moreover, a recent study showed that dry mouth could influence the occurrence of halitosis and consequently affect oral health-related quality of life (32).Besides oral manifestations, these problems may result in more general effects, including loss of appetite, malnutrition, impaired interpersonal communication and social interactions, and perhaps even depression, thereby negatively affecting the daily lives of sufferers (4,5).
When only xerostomia was considered the overall prevalence in the studies considered in this review ranged from 0.01% (8) to 1-Was the sample frame appropriate to address the target population?; 2-Were study participants sampled in an appropriate way?; 3-Was the sample size adequate?;4-Were the study subjects and the setting described in detail?; 5-Was the data analysis conducted with sufficient coverage of the identified sample?; 6-Were valid methods used for the identification of the condition?;7-Was the condition measured in a standard, reliable way for all participants?; 8-Was there appropriate statistical analysis?9-Was the response rate adequate, and if not, was the low response rate managed appropriately?45% (2).For hyposalivation, the prevalence rate ranged from 0.02% (3) to 40% (13).For both situations, ageing seems to be determinant of its occurrence, studies have shown what appears to be an increase in the prevalence of dry mouth with increasing age.Diverse factors have been investigated to clarify the potential association of age with such oral conditions.However, most studies were conducted in samples of older people and no population-based studies were conducted in children or adolescents sample.Moreover, a few studies conducted in both populations of young adults and older adults have found prevalence differences between them.There is evidence that the association with age is not just due to the aging process itself.Aging is associated with increases in comorbid chronic medical conditions, w h i c h c o n s e q u e n t l y increases the use of medications.Many of the drugs taken are associated with lower salivary flow (10,11,17,27,33).In this context, an interesting o b s e r v a t i o n w a s t h e association between the number of medications taken and the prevalence of xerostomia (34)(35)(36).The prevalence of xerostomia is usually higher in individuals who take more than one medication (4,37).Factors such as changes in saliva quality, underlying diseases, and medications should be  considered as the cause of higher subjective perception of dry mouth with aging (16).Not only polypharmacy but specific drugs present salivary flow and quality adverse  effects.Antiasthmatic drugs were suggested as the main mediator of high risk of dental caries in asthmatic children and adolescent due to its high impact on salivary conditions.Even few months using inhaled β2-agonist and corticosteroids could decrease significantly the salivary flow rate, increase dental plaque index, and decrease salivary pH (38)(39)(40), all of these consequences favor dental caries.Finally, the long-term effect of this group of medicines in salivary aspects is not known, and identify asthmatic population that used inhaled medicines in the past could be an alternative to understand life-course effects of medication on salivary flow rate.This study has several strengths that should be considered.The first one is the inclusion of populationbased studies only, excluding studies that investigated clinical or other biased samples, then the common occurrence of dry mouth in general population could be evaluated.Secondly, the analytical approach used, including the meta-analysis, allowed estimating the global prevalence of xerostomia.Such methodology has already been used to estimate the global prevalence of halitosis (32) and its use here by regarding dry mouth highlights the importance of the approach for obtaining global estimates of the impact of major oral conditions.The high heterogeneity found in all meta-regressions conducted evidence the lack of a standard criteria for population-based studies aiming to assess salivary conditions.However, we used meta-regression as a tool to explain heterogeneity in prevalence among studies.Although there was no statistical association, a considerable amount of heterogeneity were explained after considering age-groups and geographic location of the studies as potential sources of heterogeneity.Finally, the studies included showed high quality, since just six studies of all included do not fulfill all items as "yes" in the JBI critical appraisal checklist.
Besides the strengths stated, our findings should be interpreted with caution.Firstly, the Egger test and the funnel plot revealed publication bias, although the main databases for the outcome had been searched and an extensive search had been conducted this could suggest a lack of information of unpublished studies or grey literature.Even though, a great variety of studies were screened (4,414 articles).Moreover, the chosen databases included the main peer reviewed journals of the field and probably all high-quality population-based salivary studies were included in our research.Secondly, even adopting strategies to collect detailed information of all studies, as send e-mail to authors, eight potential articles were excluded due not provide prevalence data, and it is not clear their influence on the overall result.Moreover, language restriction was applied and some specific countries prevalences were not considered, but only three articles were excluded by such reason, being in Hebrew, Chinese and Hungarian.Finally, we cannot fully explain the heterogeneity found based on the included variables.Hence we encourage the development of further studies addressing other potential sources of heterogeneity in salivary research.

Overall prevalence of dry mouth
The studies included were conducted mostly in high-income countries, except for one study in Mexico.Considering that the occurrence of chronic diseases and aging are rather socially determined, and social inequality manifests differently and perhaps more acutely in poorer countries, it is possible that the overall dry mouth prevalence estimate may have been different had more data from those countries been available.It is also important to emphasize the need for more information from prospective cohort studies, including from younger populations (34) and consider the high presence of specific chronic diseases, such asthma that could influence xerostomia through medicines, in order to better understand the history of dry mouth and its effects (9).
Another important factor to consider is that the estimated prevalence rate highly depends on the method used to measure dry mouth.Xerostomia is not necessarily accompanied by lower salivary flow rate (7,10,13).In this context, Thomson et al. (15) reported that xerostomia and hyposalivation occurred together in only 6% of their overall sample, and that this was equivalent to only one in six of the 36% of individuals who had either condition.The measurement methods used may over-or underestimate xerostomia, a fact widely discussed by Thomson et al. (21).These findings emphasize the need for further work on existing and new clinical measures, including the most recent ones working with scales for epidemiological use to measure the prevalence of xerostomia.Alternatively, researchers may be required to reach a consensus on which of the many currently available measures should be used.
There is still much to find out about dry mouth and its associations (41).Despite diverse approaches to the condition's measurement, just over one in four people in adult age or older suffer from xerostomia, with higher rates observed among elderly.

Figure 1 .
Figure 1.Flowchart selection process for studies included in this systematic review.

Figure 2 .
Figure 2. The overall prevalence of xerostomia.

Figure 5 .
Figure 5. Funnel plot of the studies included in metaregression analysis.

Figure 6 .
Figure 6.Estimated prevalence of dry mouth and confidence interval with random effect after the omission of the study.

Table 2 .
Excluded studies and reasons for exclusion Gerdin et al. 2005Convenience/specific sample Ghezzi and Ship 2003 Non-representative sample Gilbert et al. 1993b Same sample of an included study Goaz et al. 1994 Comments/abstracts of conferences Gois et al. 2018 Convenience/specific sample Guignon and Novy 2015 Comments/abstracts of conferences Hahnel et al. 2014 Convenience/specific sample Hassel et al. 2010 Non-representative sample

Table 3 .
Main characteristics of included studies B. A. Agostini.*Valuesincluded in the meta-analysis of overall prevalence of dry-mouth (xerostomia or hyposalivation) of the studies with more than one measure of salivary conditions.
Table 6 presents overall prevalence estimates for xerostomia and hyposalivation separately.Considerating age group, the higher prevalence values were 27.2% for xerostomia in studies conducted with older people only; and 26.0% for hyposalivation in studies considering adults people only.

Table 4 .
Evaluation of included studies according to Joanna Brings Institute Critical Appraisal Checklist (5)mson et al. (15)found a prevalence of 20% of xerostomia in an older population and 10% in an adult population (4).Similar findings were observed by Benn et al.(5)in a nationally representative sample, in which the prevalence of xerostomia was 5% in the 18-24 age group and 26% in those aged 75 years or older, but there was not a consistent age gradient.These findings provide further evidence that xerostomia, even in divergent proportions, could not affect only older people, and that perhaps the lack of knowledge of its occurrence by young adults is because younger age groups have become aware of it only over the last decade or so.Hence, it is difficult to establish age-standardized population prevalences of dry mouth and what kind factors really modify salivary conditions in young population.

Table 5 .
Meta-regression of dry mouth and subgroup analysis according to methodological characteristics.
b : 15.81% a p value of the variable in the final meta-regression model.b including both variables in adjusted meta-regression.

Table 6 .
Overall prevalence of Xerostomia and Hyposalivation and subgroup analysis according to methodological characteristics a p-value of the variable in the final meta-regression model.b including both variables in adjusted meta-regression.