When the Professors Come – On Mortality, Re-birth and Sink Status

Sri Lanka

The Natural History of Periodontitis. Photo: C.R. Ramseier, under fair use

In 1970, four-hundred-and-eighty male employees of tea plantations in Sri Lanka had been examined for the first time by western periodontists in order to start a longitudinal study of the natural history of periodontitis. It was assumed that the cohort, who supposedly lived their entire life on the plantation, had been unaffected by any treatment of periodontitis and professionally recommended or supervised oral hygiene practices.

Several papers had been published already by the mid 1980s. The study went on, and after lots of turmoil and civil war in Sri Lanka, even the tsunami of 2004, an attempt was made in 2010 to contact all participants of 1970 (Ramseier et al. 2017). Seventy-five were still available.

Ramseier et al. (2017) emphasize, in the introduction to their paper that,

[h]ypothetically, studies following subjects over a number of decades may give better insight into undisturbed disease progression, particularly between subjects showing different disease susceptibility. In this context, the untreated Sri Lankan tea labourers  provided a unique opportunity to further study periodontal disease progression in humans unaffected by professional or individual oral care. (Emphasis added.)

Hypothetically. In reality, it’s unethical (see below). At least, after new insights into the disease progression had been gained in 1986, participants (human beings after all) should have been offered thorough information about causal agents (then, without doubt, well-known), preventive measures (well established) and, yes, proper treatment.

It is reported that the study by Ramseier et al. (2017) was approved by the local dental school (apparently none of its administration qualified as co-author) and the Institutional Review Board of the University of Hong Kong SAR [sic]. No governmental ethical committee was consulted. As regards the participants (who were between 55 and 70 years of age when re-examined; note that Sri Lankans had, in 2010, a mean life expectancy of 77.9 years at birth), they were, in 2010,

informed in their native language (Tamil) by a medical doctor about the details of the study. They then gave consent by finger printing due to illiteracy.

In 2010, authors report that, fortunately,

[a]ccording to the Medical officer and the administration of the Estates [Dunsinane, Harrow and Sheen in Pundaloya], the subjects’ diet improved over the period of 40 years, and the salaries of the subjects increased continuously. Yet, the older generation analysed in this study did not communicate with the outside world and the majority remained illiterate.

On the other hand, subject interviews confirmed

persistent lack of professional preventive oral health care or cleaning devices other than occasional use of bare fingers and ashes.

Well, the 2010 follow-up study of male employees of tea plantations in Sri Lanka, initiated by Harald Löe and collaborators in 1970, when not so much was known about the natural history of periodontal disease, yielded some results.

The logistic regression analysis of numbers of teeth over the course of the study showed that higher mean loss of attachment was associated with a greater risk of tooth loss by the next examination, through significant (p < .004) interactions with smoking/betel nut chewing, age, mean CI [calculus index], and number of teeth initially present. Table 2 shows estimates of the relative risk (with 95% confidence limits) of tooth loss associated with a unit increase in mean loss of attachment, for various values of the interacting covariates. So that, for example, among those who smoked sometimes, did not chew betel nuts, were aged 30 with 20 teeth and presented with a mean CI of 1, a subject with mean loss of attachment one unit higher than another subject would be 94% more likely to lose teeth between examinations. Comparing across categories, there is greater relative risk of tooth loss from higher age, lower mean CI, and more teeth to start with; always smoking and betel nut chewing combined is associated with an appreciable reduction in relative risk of tooth loss, but smaller differences are apparent between the other categories of smoking and betel nut chewing. Neither mean PI [plaque index] nor mean GI [gingival index] showed any significant (p >.30) effects on tooth loss either as main effects or as interactions.

Making sense of the above is not an easy task. First, that higher mean attachment loss predicts tooth loss by the next examination does not come as a surprise. But that a subject with a one unit higher mean loss of attachment and a high mean calculus index was associated with a considerably lower risk for tooth loss than a subject with a one unit higher mean attachment loss but a low mean calculus index does in a way. Also, that smoking and betel nut chewing was always associated with lower risk for tooth loss (when comparing groups with a one unit difference of attachment loss) is not instantly apprehensive. Furthermore, that neither mean plaque index nor mean gingival index did influence tooth loss is not really expected. The authors do not offer possible explanations.

One of the novel findings made in the current cohort, reported in the first papers by Harald Löe and his team, was that periodontitis progression differed. For example, a group of 8% was responsible for much attachment loss at age 45 years with an annual loss of between 0.1-1 mm (Löe et al. 1986). Periodontal attachment loss in the majority of tea laborers (81%) was moderate with annual rates of between 0.05 and 0.5 mm. A more or less non-progressing group (0.05-0.09 mm attachment loss per year) was identified as well (11%).

Since the mid 1980s the latter group was no longer available. The authors of the present report write that,

[i]t was evident that after 1977, the non-progressive subgroup was no longer present (Table 1). Therefore, further analysis was limited to the original subgroups of subjects with rapidly and moderately progressing disease. The 12 edentulous subjects were not included in further analysis. Consequently, 55 (73.3%) out of 75 subjects with available data at the age of 60 were re-grouped into a first cohort that, at age 60, presented with at least 20 teeth (n = 21, 38.2%) and a second cohort that, at age 60, had less than 20 teeth but were not edentulous (n = 34, 61.8%).

It is not reported in the current publication why none of the 54 men with hardly any progression of periodontal disease had left the tea plantation after just couple of years. They were evidently the youngest and most likely healthiest subgroup and they had most probably left for meeting new challenges for a better life. (Of course they might also have died.) Consequently, the two subgroups of originally rapid and slow progression were merged in further analyses.

But then, the follow-up is no longer about the “natural history” of periodontitis in general but rather periodontitis among the poor and socially deprived, helpless people having got stuck in a tea plantation. Neither in 1970 when Löe started the study nor in 1986 when his group published the landmark paper it was known that poverty and social status would influence the course of the disease. But now we know.

The article by Ramseier et al. (2017) also indicates that 12 out of 75 individuals had lost in the meantime all teeth (supposedly due to periodontal disease as caries was reported not to be prevalent in the cohort), but only 55 dentate participants, where data at age 60 years were available, were apparently considered in further analyses. For those, case definitions, suggested by Tonetti & Claffey (2005) and participants of an EFP Workshop, were applied.

The data on each individual at each examination were thus summarized into three periodontal disease states, denoted level 0 (up to one proximal site with attachment loss ≥3 mm), level 1 (proximal attachment loss of ≥3 mm in ≥2 non- adjacent teeth), and level 2 (proximal attachment loss of ≥5 mm in ≥30% of teeth present) (Tonetti & Claffey, 2005). Disease was assumed to progress during the period of the study through these states of increasing severity, with some regression (healing) to states of less severity also possible. Of the total number of transitions observed, only 12 (or 0.9%) were regressions from state 2, and most likely due to loss of teeth between successive examinations, so it was decided to treat state 2 as a “sink” state from which no natural regression back to state 1 or 0 was possible. The 2-state (periodontal disease levels 0 and 1) Markov chain model used in Faddy et al. (2000) and Schätzle et al. (2009) where only states 0 and 1 were observed among the (Australian and Norwegian, respectively) participants was extended to a 3- state Markov chain to accommodate transitions to periodontal disease level 2 seen in these Sri Lankan data. This model for longitudinal discrete data observed over a sequence of examinations is characterized by transition “rates” similar to mortality rates in survival analysis (i.e., transitions from state i = 0 or 1 correspond to “mortality” in that state, with “re-birth” in adjacent state j = 0, 1 or 2 ≠ i): one rate for transitions, or progression of disease, from state 0 to state 1; another for transitions, or regression of disease, from state 1 back to state 0; and a third for additional transitions, or progression of disease, from state 1 to “sink” state 2. These rates were made log- linearly dependent on the covariates (as described above in connexion with the logistic regression analysis of numbers of teeth) except of course loss of attachment as this is what is being modelled by the Markov chain.

Perio is indeed an amazing subject. While in general dealing with an unpleasant albeit, more or less, harmless disease (I do realize that a considerable number, if not the majority, of dentists might disagree with this particular asssessment despite the fact that, so far, not a single individual has ever died of periodontitis), the epidemiological term “mortality” (instead of disease progression) might in fact be applied by us periodontists. As chronic diseases undergo ups and downs, that makes a note of “re-birth” necessary. Alas, what is even more remarkable is mentioning “sink state” (usually called absorbing state in Markov chain modeling, see, e.g., here) for what is generally regarded severe periodontitis.

Anyway, by using quite sophisticated statistical methods for assessing progression and regression of periodontal disease which had not been available in Harald Löe’s days, the authors of the current sequel on Sri Lankan tea laborers report some new observations, namely that,

smoking and calculus are associated with initial disease progression and that calculus, plaque, and gingivitis are associated with loss of attachment and progression to advanced disease. Further, mean attachment loss <1.81 mm at the age of 30 appears to predict a dentition of at least 20 teeth at 60 years of age.

“These results highlight,” Ramseier et al. continue, “the importance of treating early periodontitis along with smoking cessation, in those under 30 years of age, so as to prevent attachment loss, and increase the likelihood of having a dentition with ≥20 teeth at age 60 years. They further indicate that calculus removal, plaque control, and the control of gingivitis are essential in preventing disease progression, further loss of attachment and ultimately tooth loss.”

Well, I am afraid that these claims are largely unwarranted as they are based on observations made in a small cohort with serious limitations as regards to any generalization. How was it possible that variation in disease progression in this cohort of the poor and deprived could mostly be explained by invariably high plaque, calculus, and gingivitis scores; as well as very widespread smoking and betel nut chewing?

And finally, don’t we base our preventive measures for decades on these premises, i.e. plaque and calculus should be removed and smoking avoided?

Ramseier et al. (2017) are fully aware of their borderline unethical study. In the Discussion they write the following.

No preventive or therapeutic measures were carried out in any of the scoring sessions between 1970 and 2010. As reported previously (Löe et al., 1986), the original purpose of this longitudinal investigation was to study the natural and undisturbed history of periodontal disease. Although questionable from an ethical point of view, the earlier investigators made attempts to avoid disruption of any habits and did not instruct the subjects in home care practices. Even though global efforts were undertaken to improve oral health in third world countries, no measures were implemented in the Sri Lankan tea Estates to improve oral conditions. Comparable studies assessing periodontal disease progression over a period of 10–15 years in the past used a similar model of untreated subjects in China (Baelum, Luan, Chen, & Fejerskov, 1997) or Indonesia (Timmerman et al., 1998, 2000, 2002; Van der Velden et al., 2006). Beginning in 2008, however, oral health preventive programs were implemented in pre- school and grade school classes teaching the children of the Sri Lankan tea Estates the use of toothbrushes with fluoridated toothpastes. Nevertheless, such preventive programs for the children did not at all affect the older generation of male workers examined regularly during 40 years. Therefore, it was decided to treat all available subjects of the 2010 examination with non-surgical periodontal therapy in 2012, hereby contributing to the improvement of oral health in this unique cohort previously deprived of any professionally recommended or supervised oral hygiene practices.

So, after 40 years of intentional neglect, “it was decided” to treat the “available subjects” in 2010 with non-surgical periodontal therapy in 2012.

For the participants it might have come as a shock when finally realizing that they had been deprived of oral health care for all of their lives with professors visiting them now and then only to receive, for free, I suppose one session of oral prophylaxis as compensation.


Since the paper by Ramseier et al. went online in late 2017, I had prepared this comment but long been reluctant to post it here on my blog. In Norway, strict guidelines of a Regional Ethical Committee would most probably prevent even our professional luminaries to conduct such a study of depriving poor indigenous people from getting proper health care for four decades. In particular, as relationships between periodontitis and numerous serious systemic diseases including cardiovascular diseases, diabetes, and even cancer cannot be denied any longer.

The original study was planned by Harald Löe in the early 1970s when little was known about the pathogenesis of the disease. But even Löe went to Sri Lanka to study the natural history of the disease. He could have done the same enterprise in Northern Norway or Lappland where still, in 2018, medical and dental services are underdeveloped and the indigenous people (Sami) have to do long travels to see their doctors. I suppose that we have to consider Löe’s decision to target workers of tea plantations in a poor country as Sri Lanka as an act of post-colonialist convenience. But have a look at the picture above with which the new study was featured in the respective issue of the journal.

I have noticed that I have many readers from South Asia. It would be appreciated to get some feedback. Looking forward to receiving comments.

7 April 2018 @ 12:29 am.

Last modified April 13, 2018.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.