Transition dynamics in experimental gingivitis in humans

B. Danielsen', R Manji^ N. Nagelkerke^ O. Fejerskov^ and V. Baelum^

Transition dynamics in experinnental gingivitis in humans

'Department of Oral Anatomy, Dental Pathology and Operative Dentistry, Royal Dental College, Aarhus, Denmark. 'Kenya Medical Research Institute, Medical Research Centre, Nairobi, Kenya and ^Department of Periodontology and Oral Gerodontology, Royal Dental College, Aarhus, Denmark

Danielsen B. Manji F, Nagelkerlce N, Fejerskov O, Baelum V: Transition dynamics in experimental gingivitis in humans. J Periodont Res 1989; 24: 254-260. This study was conducted in order to investigate the dynamics of the gingival itiflatntnatory status during periods of plaque accumulation and thorough oral hygiene. After a period of prophylaxis, 42 volunteers were asked to abstain from all oral hygiene measures for 2 weeks, whereafter oral hygiene was reinstituted. In the absence of oral hygiene, plaque was found at virtually all sites after 7 days and the number of sites with gingivitis increased simultaneously. A reduction in gingival inflammation occurred subsequent to plaque removal. During phases of both plaque accumulation and thorough oral hygiene, sites were fcund to convert from non-inflamed to inflamed status concurrently, as in the reverse direction. No association between plaque and gingivitis was revealed. The gingival status of a single site was a poor predictor of its status on the subsequent occasion. The proportion of inflamed sites converting to non-inflamed status was greater than the proportion of non-inflamed sites converting to inflamed status at any time. The estimated incidence rate remained fairly constant during both the plaque accumulation phase and the oral hygiene phase, whereas the estimated recovery rate was considerable lower during the plaque accumulation phase compared to oral hygiene phase. The clinical appearance is the outcome of the dynamics between these rates. The steadystate prevalences derived from the estimated "incidence" and "recovery" rates were quite similar to the actual fmdings after 14 d of plaque accumulation and the subsequent ID d of thorough oral hygiene.

Introduction The observation that withdrawal of oral hygiene leads to an increase in the occurrence of gingival inflammation led to the concept that dental plaque causes gingivitis (1). Although several "experimental gingivitis" studies have drawn broadly similar conclusions (2-5), very little attention has been paid to the dynamics of the response of the gingival tissues to plaque accumulation. It is often assumed that the only response of the tissues to the presence of plaque is a transition, with time, from healthy to inflamed status. However, some studies indicate that the state of the gingival tissues fluctuates between inflamed and non-inflamed status (6, 7) even in the presence of an apparently constant plaque load (8). Although this might reflect transient changes in the quantity or quality of plaque at a given site in time, such findings might also suggest that the response of the gingival tissue to plaque may not necessarily be a constant state of inflammation.

Accepted for publication March 28, 1989

Following this concept, it is to be expected that what we clinically define as a "healthy" or "diseased" tissue is a reflection of the relative balance of forces, at any given time, between those factors which militate towards disease and those which contribute towards health. Where plaque is permitted to accumulate, therefore, we would expect that the subsequent development of clinically visible inflammation takes place under conditions in which the relative rate of conversion to and from "disease" status has been altered. The purpose of the present study was to investigate the transition dynamics involved in an experimental gingivitis study in young adults. Material and Methods Forty-five dental students at the Royal Dental College, Arhus, volunteered to participate in the experiment. The volunteers were allowed to enter the study provided that they were healthy (no present or past history of serious illnesses), exhibited no

Dynamics of experimental gingivitis Table 1. Summary of the design of the experiment showing the number of individuals and sex distribution at each examination 3 0 7 10 14 17 21 24 Day: Ora I hygiene -• .- No oral hygiene -* *- Oral hygiene Plaque t » t t Gingiva t t I i i t 16 16 Men 16 16 16 9 14 10 Women 26 26 26 26 26 10 25 14 42 42 42 42 42 19 39 24 Total

periodontal pockets > 3 mm, and had not taken antibiotics during the previous 10 weeks. All volunteers were informed as to the nature of the experiment according to the Helsinki Declaration II. Prior to the start of the experiment, all volunteers underwent thorough prophylaxis, which was repeated on three occasions over a period of 3 wk. On each occasion plaque was disclosed, oral hygiene instructions provided, and the teeth were thoroughly pohshed. Only those subjects who demonstrated the capacity to maintain good oral hygiene between each visit and who, on examination, were found to have a Plaque Index value (9) less than 0.20 were permitted to enter the experiment. Consequently, 3 subjects were excluded. The mean age of the remaining 42 subjects was 22.4 (s.d. 3.0) years, and there were 26 women and 16 men. The design of the experiment is outlined in Table 1. On d 0, each subject's gingival status was recorded and each was provided with a thorough cleaning to ensure complete plaque removal from all sites. For the next 14 d, the subjects were required to refrain from all oral hygiene measures. The chewing of gum was prohibited for the first 3 d. Plaque status was then measured on d 3, 7, 10 and 14, while gingival status was measured on d 7 and 14. On the 14th d, after recording of plaque and gingival status, each subject was provided with thorough prophylaxis (plaque removal), and oral hygiene measures were re-instituted. Thereafter, gingival status was recorded on d 17, 21 and 24. Until d 14 of the experiment, all 42 subjects participated. However, the full complement of individuals was not available thereafter (Table 1). Recordings for plaque and gingivitis were made Table 2. Mean Plaque Index and mean Gingival Index Mean Plaque Index Mean Gingival Index (S.D.) (S.O) DayO 0.11 (0.09) Day 3 1.17 (0.40) Day 7 1.70(0.37) 0.41 (0.37) Day 10 1.86(0.18) Day 14 1.82(0.29) 0.51 (0.39) Day 17 0.26 (0.21) Day 21 0.21 (0.27) Day 24 0.14(0.14)

255

from five sites on the buccal aspect of all permanent teeth (excluding 3rd molars): at each papilla (mesially and distally), the most apical point, and two points mid-way between these. Plaque was recorded using the Plaque Index: the presence of plaque was, however, detected using a spoon-excavator scraping from the gingival margin towards the center of the tooth surface. Gingival status was recorded according to the Gingiva! Index (10) using a constant force probe (F = 65 g, with rounded tip of 0.65 mm diameter). All examinations were performed by BD, and on each occasion fresh data forms were used to ensure that knowledge of previous measurements would not influence subsequent scoring. Seven individuals (with a total of 970 sites) were re-examined to assess intra-examiner consistency. The observed and expected proportional agreement for Plaque Index scores were, respectively, 0.880 and 0.528 (Kappa = 0.75), and for the Gingival Index 0.839 and 0.535 (Kappa = 0.65). The asymmetric uncertainty coefficient (11) calculated for the presence or absence of gingivitis at each examination was 0.46. statistical analysis

Plaque scores 1 and 3 were relatively rarely assigned, as was gingival score 2, while gingival score 3 was absent on all occasions (Fig. 1 a & b). In the following analysis, we have therefore considered only the presence or absence of plaque and gingivitis as the operational values of these variables, except in the calculations of the mean Plaque Index and Gingival Index scores. The asymmetric uncertainty coefficient has been calculated as a measure of the extent to which "uncertainty" in the dependent variable is reduced by knowledge of the independent variable, and has a possible range of 0 (completely uncertain) to 1 (complete predictabihty). In the following analysis we have applied the model proposed by Bekessy et al. (1976) for analysis of transition dynamics of a stochastic process in which changes occur from negative to positive states and vice versa over time. In order to estimate the transition rates from non-inflamed to inflamed status (incidence rate = h) and from inflamed to Table 3. Correlation coefficient between the individuals Plaque Index (PlI) and their Gingival Index (GI) at the same or a later occasion GI Day 7 GI Day 14 0.312 0.026 Pll Day 3 0.026 -0.186 PlI Day 7 0.068 Pll Day 10 0.142 Pll Day 14 p>0.05

256

Danielsen et al.

non-inflamed status (recovery rate=r), it is necessary to assume that stability of the underlying process has been attained (i.e. there are constant transition rates in either direction), that the probability of making one transition in one time unit is small and that the probability of making two or more transitions in one time unit is neghgible. This assumption was very nearly confirmed on two occasions in the experimental period, namely between d 7 and d 14, and between d 21 and d 24. We have Table 4a-f 4a. Gingival score on day 7 according to score on day 0. Gingivitis day 7

Frequency Row pet Col pet

DayO + Column Total

Row Total

4357 81.1 95.1 226 56.5 4.9

1013 18.9 85.3 174 43.5 14.7

5370 93.1

4583 79.4

1187 20.6

5770 100.0

400 6.9

expressed these rates in units of d"'. In other words, each statistic expresses the estimated proportion of sites converting (negative to positive h, or, positive to negative r) each day during each of the periods. The statistics li and r as previous described (12): h =

1 1 - (alpha + beta) 1 1 - (alpha+beta)

alpha t (alpha + beta) beta t (alpha+beta)

4 d. Gingival score on day 21 according to score on day 17. Gingivitis day 21

Frequency Row pet Col pet

Gingivitis Day 17

+ Column Total

Row Total

1721 93.9 84.0 328 66.5 16.0

111 6.1 40.2 165 33.5 59.8

1832 78.8

2049 88.1

276 11.9

2325 100.0

493 21.2

Uncertainty coefricient=0.02

Uncertainty coefncient = 0.1

4b. Gingival (icore on day 14 accordiing to seore on day 7.

4 e. Gingival score on day 24 according to score on day 21,

Frequency Row pet Col pet

Gingivitis Day 7 -t- .. . Column Total

Gingivitis day 14

Row Total

-

+

3800 82.9 91.6 350 29.5 8.4

782 17.1 48.3 838 70.5 51.7

4582 79.4

4150 71.9

1620 28.1

5770 100.0

1188 20.6

Gingivitis day 24

Frequency Row pet Col pet

Row Total

Gingivitis Day 21 +

2121 94.9 76.1 666 72,4 23,9

114 5.1 31.0 254 27,6

Column Total

2787 88.3

368 11.7

2235 70.8 920 29.2

m.a 3155 100.0

Uncertainty coefficient=0.I8

Uncertainty coefneient=0.i3

4c. Gingival score on day 17 according to score on day 14.

4 f. Gingiva! score on day 24 according to score on day 14.

Gingivitis day 17

Frequency Row pet Col pet

Row Total 124 11.0 20.9 468 31.7 79.1

1127 43.3

Gingivitis Day 14

1003 89.0 49.8 1010 68.3 50.2

Column Total

2013 77.3

592 22.7

2605 100.0

Uncertainty coefficient=0.06

1478 56.7

Frequency Row pet Col pet

Gingivitis day 24

Row Total 73 4.4 19.5 301 18.5 80.5

1667 50.6

Gingivitis Day 14

1594 95.6 54.6 1326 81.5 45.4

Column Total

2920 88.6

374 11.4

3294 100.0

Uncertainty coeflieient=0.07

1627 49.4

Dynamics of experimental gingivitis where: alpha is the ntimber of sites converting from noninfiamed to inflamed status (positive conversions) between two examinations expressed as a proportion of the sites "at risk" to converting in that direction; and: beta is the number of negative conversions expressed as a proportion of the sites that could convert in that direction; and (is the time (in days) between the two examinations. Assuming that the rates of conversion are constant, then, irrespective of the initial prevalence of infiammation, the eventual prevalence of inflammation is given by h/(h + r). We call this prevalence the "steady-state prevalence" which can be estimated from the observed transition rates long before the steady state has actually been attained. Results

The mean Plaque Index and mean Gingival Index at the different examinations are shown in Table 2. After 3 d of plaque accumulation the mean Plaque Index value was 1.17, rising to 1.70 at d 7, whereafter only minor changes occurred. The mean, Gingival Index score rose from 0.11 to 0,41 during the 1st wk and to 0.51 after the 2nd wk. Three days after the re-institution of oral hygiene the mean Gingival Index was 0.26 and after 10 d it was 0.14. Correlation coefficient for the association between the individual's Plaque Index on one occasion and the Gingival Index on the same or later occasions are shown in Table 3. No association between these parameters was apparent. Likewise no statistically significant association between plaque and gingival inflammation was observed for any

i1I 7

to t4 Day of Experiment •

site when analysis was performed taking each site as the unit of analysis. The gingival status of a site at any one time in the experiment was a poor predictor of its status on the subsequent occasion, reflected in the values of the uncertainty coefficients (which ranged between 0.02 and 0.18) (Table 4a-0. Of those sites which were recorded as sound on d 0, 81.1% remained sound on d 7, during which period plaque accumulation was marked (Table 2); 18.9% of previously sound sites (d 0) had converted to inflamed status after 7 d; of those sites recorded as inflamed on d 0, 56.5% were found to be non-inflamed by d 7; of those sites which were found to be inflamed on d 7, 14.7% had previously been recorded as inflamed on d 0 (Table 4a). Of those sites recorded as sound on d 7, 17.1% had converted to inflamed status by d 14 (and thus 82.9% remained healthy), whereas of those recorded as inflamed on d 7, 29.5% were recorded as sound by d 14. Of those sites found to be inflamed on d 14, half (51.7%) had previously been recorded as inflamed on d 7 (Table 4b). Table 4c compares the gingival status of each site on d 17 (i.e., after prophylaxis and reinstitution of oral hygiene) with the status on the last day of plaque accumulation (d 14) (just before prophylaxis and oral hygiene). Of those sites which were sound before prophylaxis, 11.0% became inflamed subsequently. Of those sites which were inflamed on d 14, 68.3% became healthy. Of those sites which were recorded as inflamed on d 17, however, 79.1 % were inflamed prior to prophylaxis and oral hygiene. Of those sites which were recorded as sound on d 14 (the last day of plaque accumulation, before prophylaxis and oral hygiene were re-instituted), 4.4% were found to be inflamed on the last day of

Frequency of Gingival Index Scx)fes

Frequency of Plaque Index Scores

1

257

17

Score 0

21

24

7

10 14 Day of Experiment

17

21

24

ED Score 1

^ig- I. The frequency distribution of Plaque Index scores (a) and Gingival Index scores (b) for all sites recorded during the -xperiment.

258

Danielsen et al.

the experiment (d 24) (Table 40. Of those sites which were inflamed on d 14, 81.5% were recorded as being sound by d 24. However, 80.5% of those sites which were recorded as inflamed on the last day of the experiment had been inflamed on d 14. Fig. 2a-f shows the cumulative frequency distributions of the alpha and beta values for each individual. Between each of the examinations, a greater proportion of individuals had higher beta than alpha scores - i.e. the proportion of inflamed sites converting to non-inflamed status was greater than the proportion of sites converting to infiamed status. The difference between the distribution of alpha and beta scores was marked between d 0 and d 7, but was considerably reduced between d 7 and d 14. Thereafter, once oral hygiene measures had

been reinstituted, there was a tendency for alpha to decline. Beta values increased between d 14 and d 17 (relative to d 7 to d 14) and remained considerably higher than alpha for all individuals. Based on the data for sites converting to or from inflamed status between d 7 and 14 (mid- and endpoints of the period of plaque accumulation), we estimated (over all sites) the incidence rate to be 0.0328 (s.e. + 0.0013) d ' whereas the recovery rate was 0.0566 (s.e.-1-0,0032) d"'. In other words, we estimate that approximately 3.3% of non-inflamed sites per d were converting to inflamed status, whereas 5.7% of inflamed sites per d were converting to non-inflamed status. At the end of the experiment, the estimates of the incidence and recovery rates between d 21 and d 24 were, respectively, 0.0327 (s.e.-I-0.0033) d ' and 0.4644 (s.e.-l0,0213) d', i.e., the estimate of the incidence rate was approximately the same as that obtained between d 7 and d 14, while approximately 46.4% of inflamed sites per d were converting to non-infiamed status. The cumulative frequency distribution of the estimate of the steady-state prevalence for each subject between d 7 and 14, and between d 21 and 24, is shown in Fig. 3. For the period between d 7 and d 14 (the last period of plaque accumulation) over 50% of the individuals had an estimated steadystate prevalence exceeding 23% and 17% of the individuals were estimated to have 100% of the sites affected. For the period between d 21 and d 24 (the last period with supervised oral hygiene) less than 25% of the individuals had an estimated steady-state prevalence exceeding 23% and no one was found to have more than 55% of sites inflamed in the steady-state condition. Discussion

Fig. 2. The cumulative frequency distribution of alpha and beta values of each individual based on conversions between d 0 and d 7 (a), between d 7 and d 14 (b), between d 14 and d 17 (e). between d 17 and d 21 (d). between d 21 and d 24 (e) and between d 17 and d 24(0.

In principle, the findings of the present study are in accordance with those of previous experimental gingivitis studies (1-5). Hence, the number of sites with gingival inflammation increases when plaque is allowed to accumulate, and decreases once plaque is removed. It seems of interest, however, that while previous experimental gingivitis studies have not attempted to statistically verify the extent of association between plaque and gingivitis, we found no statistically significant evidence of such associations in the present study. Our study has shown that, at any given time, even in the presence of plaque accumulations, sites which were previously inflamed apparently convert to non-inflamed status (recovery) concurrently as sites previously non-inflamed convert to inflamed status (incidence). That is to say, the dynamics of gingivitis in the presence of plaque includes tran-

Dynamies of experimental gingivitis

259

tZ],

90

ao

\ (

50

3 J

*C

S

V

\

30

50

\

20 10

20

30

40

60

60

h / h ' r

70

K^

' 0

X

30

40

SO

60

70

60

* OF SURFACES PER INDIVIDUAL

90 100

Q

10

20

30

40

60

60

70

80

90 100

» OF SURFACES PES INDIVIDUAL

Fig. 3. The cumulative frequency distribution of the proportion of inflamed sites per individual in steady-state estimated from the incidence and recovery rates during d 7 & 14 and d 21 & 24 (a), at d 7 & 14 (b) and at d 21 & 24 (c).

sitions in both directions. Similarly, subsequent to oral hygiene there continue to be transitions in both directions. It is the difference between the transition rates (from non-inflamed to inflamed, and from inflamed to non-inflamed) which determines the outcome - i.e., whether the "prevalence" of gingivitis in a given individual is high or low, inereases or decreases. Our fmdings indicate, moreover, that when plaque is permitted to accumulate over time the incidence rate does no apparently change very much. What changes markedly, however, is the recovery rate. These findings suggest that the effect of plaque is to suppress the capacity of the tissue to recover from inflammation. It is this suppressed recovery which results in the chnical findings of an apparent development of gingivitis. The fact that there is plaque accumulation does not mean that gingival inflammation will necessarily be observed - or that once gingival inflammation has occurred it will remain until plaque is removed. Under conditions of both plaque accumulation and of good oral hygiene, the status of many sites may, in fact, remain unchanged. The study demonstrates that the status of a given site at any one point in time provides no information about its likely condition on a subsequent day irrespective of whether there has been plaque accumulation or good ora! hygiene prevailing in the intervening period. It might be argued that the findings here may be attributed to examiner variation in diagnosis. The size of the asymmetric uncertainty coefflcient obtained from the intra-examiner test data suggest that this is highly unlikely. Furthermore, the finding that the empirical observations were entirely consistent with the findings of previous experimen-

tal gingivitis studies (1-5) indicates to us that the observations cannot be accounted for by variations only in application of the diagnostic criteria. In order to estimate the incidence and the recovery rates, we have assumed that constant conversion rates have been achieved. Had we prolonged the period of plaque accumulation, however, it is likely that the eventual distribution of sites with gingival inflammation would have had a shape not dissimilar to that estimated in the steady-state prevalence (Fig. 3 a-c). Comparing the observed distribution of gingivitis on either d 7 or on d 14 with the steady-state prevalence, it appeared that this assumption of constancy had very nearly been fulfilled. The same was the case for the period d 21 to d 24. It should be noted, however, that the estimates of the incidence and recovery rates during the last part of the experiment were based on data obtained from only 23 of the 42 subjects. It is quite possible that there may have been some consequent selection bias in the data and that different conclusions might have been drawn had the data been available from the full complement of individuals. The model for the analysis of data may, however, be fruitfully applied in future experimental gingivitis studies. The present way of analyzing experimental gingivitis data does indicate that in young healthy Danes who are otherwise exhibiting good oral hygiene the induced, initial stages of gingival inflammation may only be transient. This may reflect the particular "steady-state" which an individual has attained and, very importantly, the findings may explain why inflammation is occasionally observed at various sites in individuals who are practising good oral hygiene (13).

260

Danielsen et al.

The approach taken in this paper will be applied on experimental data obtained from studies where various factors thought to affect gingivitis have been studied, in order to see whether the concept of transition dynamics can improve our understanding of the nature of the processes involved.

Acknowledgments Our thanks to the participants, the clinical staff and the Director, Kenya Medical Research Institute. This study was supported by grants from the Danish Dental Association Research Fund (FUT).

References t. Loe H, Theilade E, Borglum Jensen S. Experimental gingivitis in man. J Periodontol 1965; 36: 177. 2. Theilade E, Wright WH. Borglum Jensen S, Loe H. Experimental gingivitis in man. 11 A longitudinal clinical and bacteriological investigation. J Periodont Res 1966; 1; 1. 3. Weiderman Von W, Lahrsow J, Naujoks R, Uber den einfluss der parodontalen resistenz auf die experimentelle gingivitis, Dtsch Zahnarztl Z 1979; 34: 6, 4. Holm-Pedersen P, Agerbsek N, Theilde E. Experimental gingivitis in young and elderly individuals. J Clin Periodontol \915,1: 14. 5. Winkel EG, Abbas F, Van der Velden U, Vroom TM, Scholte G, Hart AAM. Experimental gingivitis in relation

to age in individuals not susceptible to periodontal destruction, J Clin Periodontol 1987; 14: 499, 6. Larato DC, Stahl SS, Brown Jr R, Witkin GR. The effect of a prescribed method of toothbrushing on the fluctuation of marginal gingivitis, J Periodontol 1969; 40: 142. 7. Suomi JD, Smith LW, McCiendon BJ. Marginal gingivitis during a sixteen-week period. J Periodontol 1971; 42: 268, 8. Hoover DR, Lefkowitz W. Fluctuation in marginal gingivitis, J Periodontol 1965; 36: 310. 9. Silness J, Loe H. Periodontal disease in pregnancy, II Correlation between oral hygiene and periodontal condition. Acta Odontol Scand 1964; 22: 121. 10. Loe H, Silness J. Periodontal disease in pregnancy, I Prevalence and severity. Acta Odontol Scand 1963; 21: 533. 11. Nie NH, HuU CH, Jenkins JG, Steinbrenner K, Bent DH. Statistical package for the social sciences, 2nd edn. New York: McGraw-Hill, 1975, pp 226-227. 12. Bekessy A, Mulineau L, Storey L. Estimation of incidence and recovery rates of plasmodium falciparum parasitaemia from longitudinal data. Bull WHO 1976; 54: 685, 13. Lindhe J, Socransky S, Wennstrom J. Design of clinical trials of traditional therapies of periodontitis. J Clin Periodontol 1986; 13: 488. Address: Bo Danielsen Department of Oral Anatomy Dental Pathology and Operative Dentistry Royal Dental College Vennelyst Boulevard DK 8000. Aarhus C Denmark