In this study, we prospectively investigated the associations between the number of teeth and mortality in a region in Japan. We observed that people with < 20 natural teeth have a significant risk for all-cause mortality. To the best of our knowledge, this study is one of the largest studies covering a broad age range in the Japanese population, with one of the longest follow-up periods. This makes our study note-worthy.
The mechanism underlying the association between the presence of < 20 natural teeth and the risk for all-cause mortality is unclear; however, some hypotheses have been suggested. Events such as tooth loss, lead to hypoactivity of the masticatory system and ultimately, insufficient nutrient intake, which can have a negative influence on general health [13, 31]. Yoshihara et al. reported that participants with < 20 teeth had a significantly lower total protein, animal protein, sodium, vitamin D, vitamin B1, vitamin B6, niacin, and pantothenic acid intake than participants with ≥ 20 teeth . Furthermore, subjective mastication difficulties, particularly when eating hard food, begin to appear when there are < 20 teeth in the mouth [33, 34]. Tooth loss leads to reduced nutrient intake, which, in turn, may be associated with unhealthy conditions, such as being underweight or developing certain diseases [25, 32, 35, 36]. Ultimately, these factors may be linked to increased mortality.
Moreover, several studies have revealed the associations between poor masticatory function as a result of < 20 teeth and poor cognitive function , risk of incident falls , and requirement for nursing care , of which are associated with mortality. Furthermore, the association between tooth loss, poor masticatory function, and mortality has also been proven epidemiologically . Therefore, the association between < 20 teeth and risk of all-cause mortality is justified.
We would also like to discuss the underlying mechanisms that associate tooth loss with the risk of all-cause mortality, based on the causes for tooth loss. Although tooth loss is primarily caused by dental caries or periodontal disease, tooth loss due to periodontitis must affect mortality differently than that due to dental caries. Periodontal disease evokes a systemic inflammatory response and increases the risk for cardiovascular disease [10,11,12]. Further, chronic infection and inflammation associated with periodontal disease are also thought to affect the pathogenesis of several types of cancer . Moreover, the generation of carcinogens such as nitrosamines in periodontal disease is also suggested to increase the risk for cancer . In recent years, several studies have reported that bacterial colonization in periodontitis may be associated with not just oral cancer, but also gastrointestinal tract cancers, such as in the colon or pancreas [38, 39]. However, the underlying mechanisms behind these associations are yet to be scientifically proven , warranting further studies in this regard.
This study could not confirm significant associations between the number of teeth and cancer-related mortality, which is an extremely controversial topic. Goto et al. surveyed the associations between number of teeth and cancer-related mortality, such as lung cancer, upper gastrointestinal cancer, and orodigestive cancer; significant associations were confirmed only between number of teeth and lung cancer . Anzai et al. also reported similar results, confirming significant associations only between number of teeth and orodigestive cancer; the associations with other types of cancer (lung, stomach, pancreas, colon, and liver) were not significant . The mechanism underlying the association between number of teeth and cancer-related mortality is unclear. However, as previously discussed, chronic infection, inflammation, generation of carcinogens, and bacterial colonization accompanying periodontitis have been advocated as risk factors for carcinogenesis. Considering the mechanisms reported in the previous study [18, 21], the oral health status, including the periodontal and oral hygiene status, should play a more significant role in cancer-related mortality than the number of teeth. Future studies should attempt to assess not just the number of teeth, but also the oral health status. Furthermore, it may also be necessary to survey site-specific cancer-related mortality.
Interestingly, having < 20 teeth was more strongly associated with all-cause mortality among the non-diabetics and non-smokers. In general, DM has various complications such as renal disease and macroangiopathy, which are known to increase mortality risk. The higher prevalence of other mortality risk factors in the diabetic participants may explain why the predictive power of tooth loss for mortality was diminished in these participants. On the contrary, the predictive power of tooth loss for mortality may be enhanced in the non-diabetics as they have a lower prevalence of other mortality risk factors than the diabetic participants. Similarly, in the non-smoking participants, the predictive power of tooth loss for mortality was enhanced due to the lower prevalence of other mortality risk factors in the non-smokers than in the smokers. As such, there is a possibility that if an individual is relatively healthy, such as a non-smoker or a non-diabetic, the number of teeth in the individual may have a significant impact on his/her survival. However, a subgroup analysis interpretation should be performed with caution [41, 42]. Occasionally, a low statistical power may be problematic due to the reduced number of participants in the subgroup analysis. Furthermore, there was a possibility of cognitive bias in this study; i.e., the number of non-diabetics may have been under-reported. Moreover, there may be unknown confounding factors, such as economic status. Further studies are required to confirm the effects of number of teeth on relatively healthy people.
Nevertheless, the present study had several limitations. First, we surveyed the number of teeth using a self-reported questionnaire; we did not confirm the number of teeth by clinical examination. Furthermore, we did not attempt to validate the correlation between the self-reported number of teeth and that determined by clinical examination. However, several reports have revealed that the number of teeth determined by self-reports and that determined by clinical examinations showed strong correlations [43, 44]. Thus, our methodology for confirming the number of teeth may not have influenced our results to a large extent. However, since a discrepancy between the self-reported number of teeth and that determined by clinical examination may exist, a validation for the correlation between the two counts should have been performed.
Second, we did not obtain detailed information on eating ability. A Japanese prospective cohort study surveyed the associations between oral health and cancer-, cardiovascular disease- and respiratory disease-related mortality . That study showed significantly higher HRs for participants with ≤ 19 teeth and eating difficulty in comparison to those with ≥ 20 teeth. However, there were no significant HRs for participants with ≤ 19 teeth and those who could eat everything in comparison to those with ≥ 20 teeth. These results suggested that eating ability, rather than number of teeth, might affect respiratory disease- and cardiovascular disease-related mortality. These findings may elucidate our findings, which show a lack of significant association between < 20 teeth and cardiovascular disease-related mortality. Furthermore, another recent study suggested that the number of functional teeth was a stronger predictor for all-cause mortality than the total number of teeth among community-dwelling older adults . We should have surveyed not just the number of teeth, but also obtained details on eating ability. Furthermore, we did not investigate oral hygiene and periodontal status either. As previously discussed, oral health status has a greater potential to affect mortality than the number of teeth, especially cancer-related mortality. A comprehensive survey on the oral cavity may be required in the future.
Third, the interpretation of results in cases wherein p value was close to 0.05. In our main findings regarding the association between the number of teeth and all-cause mortality, the CI of the HR for all-cause mortality was revealed to be very close to the null value (1.007), and the p value also approached non-significance. Therefore, the results must be interpreted with caution. p values should be considered as continuous variables rather than dichotomous ones (limited to only “significant” and “not significant”) for the purpose of demonstrating how often these observations would occur by chance. If p values are truly marginal, we need to take precautions to not overlook any clinically significant findings.
Fourth, was the methodology for surveying the participants’ educational status. The participants were categorized into three groups (low, middle, and high educational status) based on their age during their final educational qualification. Some participants might have failed their senior high school entrance examination or repeated some of their education in senior high school; therefore, a possibility of discrepancy in the participants’ true educational status cannot be denied. However, almost all the junior high school students in Japan subsequently go to high school, and most high school students graduate from high school at 18 years . As such, the possibility of a discrepancy in the true educational status may not be problematic.
Fifth, the number of teeth in the participants may have changed over a period of time. There is a possibility that participants with < 20 teeth could have restored their teeth at any time during the follow-up period, changing their status from having < 20 teeth to ≥ 20 teeth through the use of prostheses, such as dental implants. There is also a possibility that participants with ≥ 20 teeth could have lost their teeth at any time during the follow-up period, changing their status from having ≥ 20 teeth to < 20 teeth. In this study, we only surveyed the number of teeth at baseline. Ideally, the change in the number of teeth should have been surveyed during the follow-up period as well.
Sixth, is selection bias. Our final participants may not be representative of the original target population, since 734 of 2942 participants were excluded due to incomplete data regarding their number of teeth. These excluded values may have affected the overall results.
Seventh, the participants’ smoking status was coded as binary variables; i.e., current smoker or non-smoker, which encompassed past smoking history. Several studies in the literature have indicated the importance of examining lifetime smoking history, rather than just examining the smoking history at a single point in time [45,46,47,48]. Smoking status should have been considered a continuous variable in the statistical analysis.