Lung function and associations with multiple dimensions of dental health: a prospective observational cross-sectional study

Background Epidemiological data suggest an association between respiratory diseases and periodontal health. However, the link between the overall dental status and single lung function measures, within a practical clinical context, is not well studied. Methods Following a prospective cross-sectional design, consecutive adult patients were evaluated. Next to spirometry, anthropometric data, profession, smoking status, symptoms, self-rated exercise performance, comorbidities, allergies and medication were determined. Assessment of dental status comprised carious lesions, dental fillings, missing teeth, dentures, insufficient fillings/dentures, implants, oral mucosa diseases, calculus, decayed-missed-filling-teeth (DMF-T)-index, periodontal screening-index, and orthopantomograms. Results Among 587 adult patients considered, 206 were included (119 female; median age 42.0 years; 56 % smoking history). Most patients had dental fillings (86.9 %), fix/mobile dentures (66.5 %), missing teeth (56.8 %) and calculus (84.0 %), the overall DMF-T being 15 (9; 21). Periodontitis was present in 53.9 %, an abnormal orthopanthomogram in 47.9 % of subjects. Regarding spirometric indices expressed as % predicted, dentures, missing teeth, oral mucosal diseases and a DMF-T > 15 (median) were associated with lower maximal expiratory flows at 25 % of vital capacity (MEF25) (p < 0.05 each). In adjusted logistic regression analyses, only dentures were associated with low MEF25 % predicted and with the ratio of forced expiratory volume in 1 s to forced vital capacity (FEV1/FVC; p < 0.05 each). However, periodontitis and DMF-T were linked to age (p < 0.001) and packyears (p < 0.05) only. Conclusion Within a real-life clinical setting, only the presence of dentures showed weak associations with lung function, suggesting small airways dysfunction and obstruction. Most of the associations were explained by smoking habits and age.

evoked by inflammatory oral processes, might affect the lower respiratory system [5], in analogy to cardiovascular disorders [9].
These considerations render a link between oral health and especially COPD plausible. Systemic inflammation and recurrent exacerbations of chronic inflammation, often due to bacterial infections, are of major importance for the progression of COPD. Conversely, the presence of COPD and its associated frequent infections could impair oral health [8], but such interactions could be substantially confounded by socioeconomic status, lifestyle, education, dental care, medication and smoking habits.
As yet, most data refer to the association between COPD and periodontal disease [10][11][12][13][14][15][16][17]. These studies have provided conflicting results and used different techniques, outcome measures or populations for the assessment of periodontitis. If there should be a relevant link between dental health and lung function, it would be reasonable to expect that oral bacteria and systemic inflammation lead to impaired respiratory function prior to the manifestation of COPD. Additionally it could be hypothesized that lung function is associated with other dimensions of dental health, e.g. caries and endodontic lesions [8].
Many of the investigations were retrospective [2-5, 7, 14, 18, 19] or differed with regard to study design and populations, methods and/or outcome measures used for dental assessment. Recently it has been reported that periodontal health (dental plaque, gingival bleeding and PD) was worse in smokers versus non-smokers, but in a multivariate model no major differences between smokers with versus without COPD occurred [18]. Similarly, in an earlier analysis of the NHANES-I population, no association was noted between any respiratory disease and the periodontal index [7]. In the larger NHANES III population comprising 7625 subjects, also no significant association between COPD and periodontitis has been found for former and non-smokers, except for current smokers [19].
Based on this it was concluded that the risk to develop periodontitis is predominately explained by smoking. In contrast, clinical attachment loss (CAL), PD, and number of missing teeth were found to be associated with reduced lung volumes, airflow limitation and hyperinflation but not with diffusing capacity [10]. Moreover, in very severe COPD, chronic marginal periodontitis was more frequent than in other severe respiratory diseases [14]. In line with this, CAL, PD and an oral hygiene test were higher in COPD patients as compared to a control population, after adjustment for covariates [11]. In a Chinese population plaque index was the most important periodontal factor related to COPD [12]. As a result, the most recent meta-analysis stated a significant association between periodontitis and COPD but also a substantial heterogeneity between studies [20]. Moreover, the funnel plot was markedly asymmetrical, which is considered to be indicative of a publication bias.
As a further relevant issue, the majority of studies focused on the presence of COPD as defined by a fixed FEV 1 /FVC <70 % [11,12,19,[21][22][23] or did not report lung function in detail [7,14,24]. In the present study, we covered all common spirometric lung function measures and did not restrict the analysis to the mere presence of COPD. We also explored the associations with dental health by using spirometric measures in absolute or % predicted values, applying established and, if possible, the novel GLI reference values. In particular we included MEF 50 and MEF 25 which are traditionally considered as being related to small airway dysfunction and indicative for the onset of obstructive lung disease. This was complemented by a comprehensive assessment of dental health as far as maximally feasible under the conditions of routine clinical practice.
We investigated the possible link between lung function and several dimensions of dental health beyond the periodontal status in a prospective "real-life" setting. For this, spirometry as well as a comprehensive, standardized dental examination by established techniques in dental practice were performed. We also included the decayedmissed-filled-teeth (DMFT)-index as an established score of dental health [25] which never has been explored regarding respiratory diseases in adults.

Study Protocol
Following the study protocol, consecutive, potentially eligible subjects consulting a general dental practice in Mühldorf, Bavaria, Germany, from July 2011 to January 2013 were asked for study participation. Inclusion criteria were age ≥18 years and written informed consent. Subjects with known chronic infections (e.g. HIV or tuberculosis) or malignant diseases were excluded. We recorded anthropometric data and assessed medical history by routine dental exploration and a standardized questionnaire. Patients were asked for pulmonary symptoms (cough, sputum production, wheezing), comorbidities (pulmonary, cardiovascular, other), medication, the presence of type I allergies, smoking status (current, previous, packyears), the profession including possible workplace exposures, self-estimated exercise capacity, dyspnoea level, and previous periodontal treatment.
Spirometry (Masterscope/Labmanager; Carefusion, Höchberg, Germany) was performed according to the guidelines of the American Thoracic Society (ATS) [26] and evaluated using established [27] and, when possible, the novel global lungs initiative (GLI) reference values [28]. Forced expiratory volume in 1 s (FEV 1 ), forced vital capacity (FVC), peak expiratory flow (PEF) and maximal expiratory flow rates at 50 and 25 % (MEF 50 and MEF 25 ) of vital capacity were recorded. Clinically relevant bronchial obstruction was assumed at a ratio of FEV 1 / FVC <70 % or, alternatively, <lower limit of normal (LLN) according to GLI. The spirometry was done by one single trained dental practice technician.

Dental health assessment
The comprehensive dental examination was also done by the same experienced dentist who performed the spirometry and who assessed the following items with regard to the presence and the number of respective teeth: carious lesions, dental fillings, missing teeth; dentures (fix, mobile, implants), insufficient fillings/dentures; presence of oral mucosal diseases (fistula, aphthous ulceration, gingivitis) and calculus. Finally, the DMFT-index was determined (with the highest value of 28, as third molars are not considered) [25].
Periodontal status was evaluated by the periodontal screening-index (PSI) using an established WHO-probe (DB767R, PCP-11.5C, Aesculap, B. Braun, Melsungen, Germany) with a marker of length and a hemisphere at its end. The probe was introduced cautiously in four positions (mesiobuccal, distobuccal distopalatinal, mesiopalatinal) of each single tooth and probing depth (PD), bleeding and roughness of surface evaluated. Codes were chosen as follows [29]: (0) black marker (PD 3.5 mm) completely apparent during probing, no calculus and/or insufficient edge of restauration occur, gums are healthy, no bleeding after probing; (1) black marker completely apparent during probing, no calculus and/or insufficient edge of restauration occur, minor bleeding after probing; (2) black marker completely apparent during probing, manifest calculus and/or insufficient edge of restauration occur; (3) black marker only partially apparent during probing, any other previously mentioned criteria may exist; (4) black marker disappearing completely during probing (PD > 5.5 mm).
For documentation the maxilla and mandible was divided in sextants: maxillary (S1-S3, right-posterior, front, left-posterior) and mandibular (S4-S6, left-posterior, front, right-posterior), and the highest value was registered for each sextant. Periodontal status was categorized as follows: 0 in all sextants: no periodontitis; 1 or 2: no significant periodontitis but gingivitis; 3: moderate periodontitis; 4: severe periodontitis. The presence of periodontitis was defined according the following criteria: PSI ≥ 3 or previous treatment of periodontitis, or bone loss ≥1.
X-rays (orthopanthomogram), if available or required for medical reasons (e.g. dental surgery, search for apical lesions/dental focal points), were evaluated with regard to apical lesions, exposed furcation, displaced teeth, and horizontal loss of alveolar bone. The horizontal loss of alveolar bone was categorized as follows: grade 1: bone loss <1/3 of alveolar bone; grade 2: bone loss >1/3 but <2/3, or grade 3: bone loss >2/3 of alveolar bone. Following the study protocol, approved by the ethics committee of the Ludwig-Maximilians-University, Munich, Germany (study number 066/11), no X-rays were taken only for the study. Only if there was a medical reason for the comprehensive dental examination, new X-rays were taken.

Statistical analyses
Data were continuously documented and finally analyzed with SPSS software (IBM, Version 19.0, Chicago, IL, USA). Due to skewed distributions (Shapiro-Wilk Test) continuous variables were summarized as median values (quartiles), while using the non-parametric Mann-Whitney U test to assess intergroup differences. In case of categorical variables the Fishers exact test was applied (two-sided). Bivariate correlation analyses (Pearson or Spearman) were employed to assess associations between spirometric measures and patients' characteristics. For the identification of independent associations with dental health, a logistic (dichotomous variables) or multivariate linear regression analysis was performed, whereby the single dimensions of dental health were used as dependent variables. Smoking, parameters being significant regarding the respective dimension of dental health, and single lung function indices (one by one) were introduced as explanatory variables. When using % predicted values of lung function we did not additionally introduce age, gender and height, as these were already taken into account by the calculation of % predicted values. P values <0.05 were considered statistically significant.

Study population
Among the 578 adult patients asked for study participation, 206 were included; 372 patients refused participation, delivered insufficient spirometry, or were excluded according to the study protocol. Anthropometric data, comorbidities, profession, job-related exposures, smoking history, respiratory symptoms, and exercise capability are shown in Table 1. Regarding medication, most often prescribed were thyroxine/iodine (n = 19; 9.2 %), angiotensin-converting enzyme inhibitors and β-blockers (each n = 7; 3.4 %).
There were inverse correlations between the number of packyears and FEV 1

Association between periodontitis and spirometry
Subjects with periodontitis (n = 111) had higher age, BMI, weight and more packyears (Table 4), but less type-I allergies. Moreover, the absolute values of FEV 1 , FVC, FEV 1 /FVC and MEF 25 were significantly lower (Table 5), while this was not the case for % predicted values.
In multivariate logistic regression analyses taking the presence of periodontitis as dependent variable together with the absolute value of FEV 1 , FVC, FEV 1 /FVC or PEF (one by one), only age (p < 0.001) and packyears (p < 0.05) were statistically significant. When introducing the absolute value of MEF 50 /MEF 25 , only age remained significant (p < 0.001). In multivariate analyses including % predicted values, only packyears (p ≤ 0.01 each) were significantly associated with periodontitis.
In adjusted bivariate logistic regression analyses dentures were significantly associated with MEF 25 % predicted (Table 6), and marginally failed statistical significance when also including age (p = 0.067). No statistical associations with lung function indices (absolute or % predicted) were found in adjusted multivariate analyses including dental fillings, missing teeth, oral mucosal disease, calculus, carious lesions, insufficient fillings/dentures and implants.
In a bivariate logistic regression analysis, taking DMF-T index (either < versus ≥ 15) as dependent variable together with the absolute of FEV 1 , FVC, FEV 1 /FVC or PEF, again only age (p < 0.001) and packyears (p < 0.05) were significant. When introducing absolute values of MEF 50 /MEF 25 , only age remained significant (p < 0.001). In multivariate analyses including % predicted values, packyears were still significant (p ≤ 0.01 each). In a multivariate regression analysis, taking DMF-T as dependent variable, similar results were obtained.
In an adjusted bivariate logistic regression analysis, taking the presence of any radiologic finding as dependent variable, significant association were only detected for age (p < 0.001 each) when including absolute values, or packyears (p < 0.05 each) when including % predicted values.

Association between abnormal dental health and the presence of COPD
Patients with versus without FEV 1 /FVC <70 % or <LLN did not differ with regard to periodontitis (Table 4), the DMF-T index or the other single dimensions of dental health. However, these patients more often were male, had more packyears and current smoking history (p < 0.05 each). In adjusted multivariate analyses, FEV 1 / FVC < 70 % was significantly associated with the presence of dentures (Table 8).

Table 7 Spirometric data in relation to the DMF-T index
Comparing subjects with DMF-T index either < versus ≥ 15 (median) Italic characters are used for statistically significant comparisons (p < 0.05) Table 1 Parameter The major results have been presented at the annual meeting of the European Respiratory Society 2014 [30].

Discussion
The present study is the first prospective cross-sectional study evaluating the associations between spirometric measures and multiple dimensions of dental health in a random population of adult patients from a general dentist's practice. Using % predicted values, the associations of the different indices of dental health with lung function appeared to be rather weak. While patients with dentures, missing teeth, oral mucosal diseases and a higher DMF-T-index presented with lower MEF 25 , in adjusted multivariate analyses, only the presence of dentures was significantly associated with MEF 25 % predicted and FEV 1 /FVC < 70 %. Particularly periodontitis and DMF-T were mainly linked to age and smoking.
As expected we found strong correlations between age and dental health, particularly regarding periodontitis and the composite DMF-T index. There was also a consistent association with smoking habits, in line with the literature [18,19]. In contrast, no significant associations between periodontitis and spirometric measures were present after adjustment for covariates. Recently, Holfreter and colleagues [10] found that measures of periodontitis were significantly linked to nearly all dynamic and static lung function indices. However, these were retrospective analyses, and dental health and lung function were assessed at different time points. The associations were clearly weaker when smoking was included as covariate. Moreover, the adjustment for body height reduced the correlations of periodontitis with lung function considerably. Therefore authors hypothesized that smaller persons might have a higher susceptibility to develop severe periodontitis. However, their findings could also be interpreted in the way that the use of % predicted values of lung function, correcting for age, gender and height for each individual index, probably would have resulted in less strong associations.
In the present analysis we also explored the associations between other dimensions of dental health beyond periodontitis. Most spirometric indices expressed in absolute values were significantly different with regard to specific characteristics of dental health (carious lesions, dentures, missing teeth, oral mucosa disease, calculus). Using % predicted values, dentures, missing teeth, oral mucosa diseases and a DMF-T-index > median were associated with lower MEF 25 only. In further adjusted multivariate analyses, accounting for the difference between groups, only dentures were significantly linked to MEF 25 % predicted. Considering a potential causal relationship between reduced lung function and impaired dental health, it seems pathophysiologically plausible that the association is particularly reflected in the small airways where the initial changes in obstructive respiratory diseases occur [31]. Interestingly, Bergström and colleagues [18] recently found a weak but significant correlation between the occurrence of emphysema in CT scan or reduced carbon monoxide diffusing capacity and pocket depth and/or teeth loss. They hypothesized that this points towards an association between dental health and lung pathology not reflected in spirometry. Unfortunately they did not assess MEF 50 or MEF 25 , which are capable of detecting minor spirometric airway abnormalities.
The association between dentures and lung function has been described in the latest literature by Linuma and co-workers [6]. Some authors described associations between missing or remaining teeth and lung function [6,10] or the presence of COPD [12,18,22,23]; these associations were not significant in our analysis after adjustment for covariates. From a dentist's perspective, both findings are not unexpected as dentures or missing teeth often reflect poor overall dental health status. As known, carious lesions acquired in early childhood or dentistry phobia often leads to early tooth loss and dentures; this phenomena is particularly observed in socially disadvantaged families [32][33][34]. This also fits to studies which found a link between oral hygiene status and COPD [11,24,35]. Surprisingly, the introduction of the composite DMF-T index did not enhance the associations in the present analysis, although patients with a higher DMF-T-index showed worse MEF 25 % predicted values. The associations probably have been weakened by the "F" (=fillings) subscore, which reflects a restored condition and not so much an inflammatory focus potentially linked to lung function impairment. In line with this, patients with dental fillings did not have worse lung function.
Oral mucosal disease was also associated with MEF 25 % predicted, but this association again was not significant after adjustment. This was most likely to the fact that the overall number of patients having oral mucosal diseases was rather low. In a larger population, gingival index was associated with airway obstruction in former smokers [21]. Moreover, a significant inverse correlation between gingival index and FEV 1 % predicted was recently reported in patients with COPD after adjustment for covariates [11]. Only few authors have considered radiologic abnormalities assessed by orthopantomogramms with regard to lung function; these studies addressed alveolar bone loss with regard to the presence of COPD. In the investigation by Wang and co-workers [22], the higher alveolar bone loss in COPD marginally failed statistical significance in univariate analysis, whereas in another study [12] it was very well associated with COPD. Similarly, Leukfeld and colleagues [14] observed that a mean marginal bone level ≥4 mm was associated with COPD independent of other risk factors for periodontitis. We found radiologic abnormalities in 47.9 % of subjects but no significant associations with lung function. Similar results were obtained when including patients with alveolar bone loss of ≥1 only (41.3 %). Possibly, radiologic abnormalities only occur in very advanced respiratory disease [14], but not in a general population as studied by us.

Limitations
An obvious limitation is the relatively low number of study participants. Many patients were not interested in lung function testing, medical research, or could not perform spirometry adequately. However, we consider it likely that this happened randomly and that we still investigated a representative sample. Since the study was performed in a dentist's practise, we had to dispense from postbronchodilator measurements. Thus, the presence of COPD via FEV 1 / FVC < 70 % might have been overestimated. MEF 50 and MEF 25 could not be determined in all patients, however a full determination would presumably enhance the significance of the associations. As the study is observational, it does not allow inferring causal relationships. The determination of serum inflammatory markers or oral bacterial colonization would have been desirable regarding potential interactions between lung and dental health, although in a recent analysis fibrinogen and high-sensitive C-reactive were not suggestive of links between systemic inflammation and lung function [10]. We characterised patients comprehensively with regard to dental health, but the protocol had to remain feasible and we might have missed some factors. Actually, tooth brushing times and method, experience of dental floss use, dental visits and oral health knowledge have been demonstrated to be significantly related to the risk of COPD [14,22]. We could not adjust the analysis for these factors, similarly as, e.g., Holfreter and collegues in their impressive analysis [10]. Whether the inclusion of the profession (academic versus non-academic) or school education [10] is sufficient to account for these factors, is unclear.

Conclusion
In conclusion, this prospective cross-sectional study assessed the associations between dental health and spirometric measures in a general real-life population, using a broad panel of dental health measures. The major result was that dentures, missing teeth, oral mucosa diseases and a higher DMFT were weakly linked to impaired age-adjusted indices of small airways dysfunction in terms of MEF 25 % predicted, while dentures were the only dimension remaining significant in multivariate analyses. Our data suggest that at least phenomenologically, the link between lung function and dental health is weak, compared to the needs for relevant predictions in individuals. Further prospective studies could benefit from the information on specific assessments of dental health provided by our study. These studies might also address respiratory disorders beyond the diagnosis of COPD, particularly with regard to small airway dysfunction.

Authors' contributions
The contributions of the authors are as follows. CH collected the data, participated in the design of the study, participated in the statistical evaluation and helped in drafting and writing the manuscript. RAJ participated in the statistical evaluation of the data, their interpretation and in writing the manuscript. SB carried out and designed the study, performed part of the data evaluation and drafted the manuscript. All authors had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the analysis. All authors read and approved the final manuscript.