123
Periodontal Disease in Non-InsulinDependent Diabetes Mellitus Lawrence J.
Emrich, *f Marc Shlossman, and Robert J. Gencox *
The relationship between diabetes mellitus and oral health status was determined in Pirna Indians from the Gila River Indian Community in Arizona. This tribe of native Americans has the world's highest reported incidence and prevalence of non-insulindependent (type 2) diabetes mellitus. The probing attachment level, alveolar bone loss, age, sex, Calculus Index, Plaque Index, Gingival Index, fluorosis, and DMFT as well as the diabetic status was assessed in 1,342 Pirna Indians who were at least partially dentate. The prevalence and severity of destructive periodontal disease was determined by measuring probing attachment loss and radiographically apparent interproximal crestal alveolar bone loss, two independent but correlated indicators of periodontal destruction. Only diabetic status, age, and the presence of subgingival calculus were significantly associated with both increased prevalence and greater severity of destructive periodontal disease in this population. Diabetic status was significantly and strongly related to both the prevalence and severity of disease after adjusting for the effects of demographic variables and several indices of oral health including the Plaque Index. Subjects with type 2 diabetes have an increased risk of destructive Periodontitis with an odds ratio of 2.81 (95% confidence interval 1.91 to 4.13) when attachment loss is used to measure the disease. The odds ratio for diabetic subjects was 3.43 (95% confidence interval 2.28 to 5.16) where bone loss was used to measure periodontal destruction. These findings demonstrate that diabetes increases the risk of developing destructive periodontal disease about threefold. Furthermore, diabetes increases the risk of developing periodontal disease in a manner which cannot be explained on the basis of age, sex, and hygiene or other dental measures. Periodontitis should be considered a potential complication of diabetes in evaluation of patients. / Periodontol 1991; 62:123-130.
Key Words: Indians, North American; diabetes, mellitus; oral health index; periodontal
diseases/epidemiology; periodontal diseases/etiology.
Diabetes is a chronic metabolic disorder that affects 6% of the American population. Much of the information currently available concerning the relationship between diabetes and oral conditions, and most notably periodontal disease, has been obtained from the study of convenience samples such as patients attending diabetic clinics.1-9 This method may favor selection of patients with symptoms and complications, and may under-represent those with asymptomatic carbohydrate intolerance, mild symptoms, or those who have not sought medical care. We sought to provide a description of the clinical characteristics of diabetes-related periodontal disease, especially in its earliest stages, by the systematic examination of these relationships in a *Roswell Park Cancer Institute, Buffalo, NY. fDr. Emrich died June 23, 1990. department of Oral Biology State University of New York at Buffalo, NY.
Buffalo,
the basis of their systemic oral disease status. The Pirna Indians of the Gila River Indian Community in Arizona have the highest recorded incidence and prevalence of type 2 diabetes in the world.10 Fifty percent of those above 35 years of age have diabetes.10-12 Diabetes mellitus in the Pirnas is exclusively type 2, or non-insulindependent diabetes mellitus, and exhibits progression and complications comparable to other populations with type 2 diabetes.13-14 This tribe has been under continuous epidemiologie investigation since 1965 by the National Institute of Diabetes, Digestive and Kidney Diseases (NIDDK) in
large population not selected on or
Phoenix, Arizona."
An epidemiologie study of the oral health of the Pirna Indians was added to the investigations in 1983. The objectives of this study were 1) to estimate the prevalence and severity of periodontal disease; 2) to study the relationship
J Periodontoi
124
PERIODONTAL DISEASE IN TYPE II DISEASE
between the prevalence and severity of periodontal disease and diabetes; and 3) to investigate other factors such as age, sex, and periodontal health indices (e.g., gingival index, plaque index, calculus index) with respect to their relationship with periodontal disease in the Pirna Indians. Shlossman et al.15 report the results of a study of 3,219 Pirnas who were examined during the first 5 years of this study. It was found that median attachment loss and alveolar bone loss is higher in diabetic subjects for all age groups and for both sexes. However, the potential mediating effects of other oral health measures such as the calculus, plaque, gingival condition, fluorosis, and decayed, missing, or filled teeth (DMFT) could not be assessed since the edentulous as well as dentate portion of the population was analyzed. The current paper presents results for a subset of 1,342 of these 3,219 subjects who were 15 years of age or older and who were at least partially dentate, so that relevant oral health measures could be measured to determine whether diabetes is associated with destructive periodontal disease, or whether the relationship can be explained by some or all of these other oral health factors. MATERIALS AND METHODS
Subjects The study population consists of 1,342 Pirna or TohonoO'odham (Papago) Indians, members of two closely related tribes from the Gila River Indian Community in Arizona. Approximately every 2 years, each member of the community who is > 5 years of age, regardless of health, is asked to participate in a standardized examination12 that includes a medical history, physical and laboratory examinations, glucose tolerance test, and dental examination. The informed consent of all subjects was obtained after the nature of the procedures and possible risks had been fully explained. Diabetes
Diagnosis
Diabetic
status
of the subject is determined by the results of a modified 2-hour oral glucose tolerance test performed according to the World Health Organization criteria.'6 After a fasting blood is drawn, the subject ingests 75g of carbohydrate.8 Two hours later blood is drawn for determination of the plasma glucose concentration. Subjects are classified as having normal glucose tolerance (NGT) if the 2-hour plasma glucose concentration is < 7.77 mmol/L (< 140 mg/dl); impaired glucose tolerance (IGT) if the 2-hour glucose is > to 7.77 mmol/L and less than 11.1 mmol/L (140 < 200 mg/dl); and diabetes if the 2-hour glucose is > to 11.1 mmol/L. A subject is considered to have diabetes if this criterion is met, or if a review of the medical record revealed documentation of a diagnosis of diabetes, as pre-
viously described.12 'Glucola,
Ames
Co., Elkhart, IN.
February
1991
Oral Examination DMFT were charted, and Dean's classification system for dental fluorosis (FI)'7 was scored by examining all the teeth in the patient's mouth. Information on third molars is not considered in any of the analyses. The periodontal status was assessed by clinical probing and by radiographie analysis. The clinical oral periodontal examination utilizes six "index teeth" (upper right first molar, upper left central incisor, upper left first premolar, lower left first molar, lower right central incisor, and lower right first premolar). '8·19 If missing, a substitute tooth immediately distal was used. If this too was missing, the tooth immediately mesial was substituted and this change noted. If an acceptable substitute was missing, no information was recorded for that tooth. The following parameters were recorded for three surfaces (buccal, mesial, and lingual) of each of the six index teeth: the Calculus Index (CI, presence or absence of subgingival and/or supragingival calculus), the Plaque Index (PI)20 and Gingival Index (GI).21 A standardized color-coded probe'1 was used for measuring the probing pocket depth (PPD) on six surfaces of the tooth (mesiobuccal, midbuccal, distobuccal, mesiolingual, midlingual, and distolingual) and probing attachment loss (PAL) from the cemento-enamel junction (CEJ) to the depth of the pocket or sulcus on four sites for each index tooth (midbuccal, midlingual, and the deepest mesial and distal
interproximal points). Interproximal alveolar bone loss,
as the percentage of bone loss from the CEJ to the tooth apex, was measured using a modified Schei ruler22-23 at the deepest point on the mesial or distal of every tooth on a panoramic radiograph.* Bone loss scores (BLS) were assigned as follows: 0 if there was no detectable loss of alveolar crest; 1 if detectable loss was less than 25% of the distance from the CEJ to the apex; 2 if the loss was 25% to 49%; 3 if the loss was 50% to 74%; and 4 if the loss was 75% or greater.
Criteria for Inclusion of Subjects This study deals with a subset of 1,342 subjects of the 3,219 subjects who participated in the oral health survey. Measurements for probing attachment loss, Calculus Index, Plaque Index, and Gingival Index on all 6 index teeth; and fluorosis index, DMFT, and bone loss score for the whole mouth, as well as information regarding age, sex, and glucose tolerance were required for inclusion in these analyses. Subjects were excluded if they were missing any of the six index teeth or their substitutes; had a history of rheumatic fever or other heart problems requiring prophylactic
antibiotics; were pregnant (since calculus obstructed the
available;
no radiographs were taken), CEJ; adequate radiographs were un-
or they were children with mixed dentition. The dental examinations and radiographie scoring were con-
•HuFriedy, PCP-10, Chicago, "Siemens, OP-5.
IL.
Volume 62 Number 2
EMRICH, SHLOSSMAN, GENCO
ducted by a single examiner the patient's diabetic status.
(MS)
without
knowledge
of
examined at a biennial examination were 14 to 21 day period for duplicate measurements (probing attachment loss and rescoring of radiographs). Intra-examiner reproducibility was assessed by the Kappa statistic.24 The Kappa statistic for probing attachment loss was 76%, and the Kappa statistic for the bone loss score was 88%. Since the population was examined over several years, the stability of measurements over time was estimated by evaluating the mean and standard error of the mean for attachment level measurements and bone loss scores. No major year-to-year differences in mean attachment level or bone loss score for the diagnostic subgroups or in the standard error of the mean were found. a
Statistical Methods Prevalence. The relationship between the prevalence of destructive periodontal disease and diabetic status and other potential prognostic factors was assessed using the linear
logistic model:25 P(Y=1) exp(B0 + B1X1 + + BkXk)/[l + exp(B()-rB1X1 + + BkXk)] where P(Y 1) denotes the probability that a subject has had destructive periodontal disease; X,, Xk are functions of the prognostic factors of interest; and B,„ B,, Bk are unknown regression coefficients. Since periodontal disease involves destruction of con...
...
=
nective tissue and bone, assessments of both were made to compare diabetic and non-diabetic subjects. PAL is a linear measurement of connective tissue loss made from a fixed point, and can be measured with considerable accuracy and reproducibility; the BLS is a measure of interproximal alveolar bony changes assessed semi-quantitatively on radiographs. Two definitions of destructive periodontal disease were employed for these analyses and for estimating the overall prevalence of destructive periodontal disease. Each subject (not tooth or surface) was classified according to the following definitions: (i) Y=l if PAL > 5 for any surface and Y 0 otherwise, or (ii) Y 1 if BLS > 2 for any tooth and Y 0 otherwise. These definitions of destructive periodontal disease do not necessarily estimate the absolute prevalence of Periodontitis. The PAL > 5 mm as a threshold value underestimates the prevalence of destructive periodontal disease as compared to the prevalence determined by bone loss as defined above. However, by systematically applying each definition to all three glucose tolerance groups (diabetic, IGT, and NGT), biases in comparing prevalence should be minimized, and comparisons between groups can be made. =
=
=
=
=
=
=
Thirty subjects
=
Regressor Variables
1 if the subject had impaired glucose tolerance (IGT) and X, 0 otherwise. 0 otherwise. 1 if the subject had diabetes and X, X2 Xj age (in years) at dental examination. 0 if the subject was male. 1 if the subject was female and X4 X4 X, percent of teeth with CI>2 (subgingival calculus) 1Ü0 (number of teeth with at least one CI>2)/6. X„ percent of teeth with PI>2 (visible plaque) 100 (number of teeth with at least one PI>2)/6. X7 percent of teeth with GI>2 (bleeding) 100 (number of teeth with at least one Gl>2)/6. 3 (severe fluorosis) Xx percent of teeth with Fl 100 (number of teeth with FI 3)/number of teeth scored. DMFT (number of decayed, missing, or filled teeth). X,,
X,
Reproducibility recalled within
Table 1: Definition of
125
= =
=
=
= =
=
=
=
=
=
=
=
=
Separate analyses were performed for each definition of periodontal disease. For each analysis, the regressor (X)
defined as in Table 1. By including the variables in the model, the principal tests of hypotheses concerning the relationship between diabetic status and periodontal disease were "adjusted" for the potentially confounding effects of these other factors. In addition, estimates of, and tests of hypotheses about, the relationship between periodontal disease and the factors listed in Table 1 also could be obtained directly from these analyses. We were able to carry out such analyses only on those subjects with the requisite teeth which were directly measured, realizing that by not including subjects who were edentulous, we were eliminating some of those who may have suffered from severe periodontal disease. The regression coefficients were estimated, as were the variances and covariances of the estimates, using standard maximum likelihood techniques, as implemented in the Statistical Analysis System (SAS) computer package. Statistical tests of hypotheses about the coefficients were performed using two-tailed tests.
variables
were
Severity. Techniques similar
to those described above for studies were prevalence employed to investigate the between the relationship severity of destructive periodontal disease and diabetic status and the other factors listed above, except that a linear model26 was employed. Two definitions of severity, one corresponding to each definition of destructive periodontal disease given above, were employed for these analyses: (i) Y square root of PAL maximum for the average subject square root of total (over teeth) maximum (over surfaces), PAL/6, and (ii) Y square root of average BLS for the subject square root of total BLS over all teeth/no. of teeth scored. The square root transformation was used to make the normal distribution assumptions for the linear model more tenable. Estimates of the regression coefficients (B,„ B,, Bk) and their estimated variances and covariances, were obtained using standard least squares/maximum likelihood techniques, as implemented in SAS.
the
=
=
=
=
J Periodontol 126
February
PERIODONTAL DISEASE IN TYPE II DISEASE
Table 2: Correlation Between Two Measures of Periodontal Destruction
of females seen in virtually all glucose tolerance and age groups. The prevalance of subjects with probing attachment loss > 5 mm is shown in Table 4. For each of the three glucose tolerance groups, there was an increasing prevalence of destructive disease seen with increasing age up to ages 45 to 54 (Table 4). There was a higher prevalence of periodontal disease in the diabetic subjects at all age groups as compared to the IGT and NGT subjects. The difference was greatest at the lower age groups, with diabetic subjects aged 15 to 24 suffering 4.8 times more periodontal disease than normal subjects, and diabetic subjects aged 25 to 34 had 2.3 times more periodontal disease than normal subjects. The IGT group was close to the normal in all age groups. The severity of destructive disease, as estimated by mean attachment levels, for all the subjects is depicted in Figure 1. Probing attachment loss is higher in the diabetic subjects as compared with normal subjects for all ages except for those 55 years or older. The differences were greater at younger age groups. The severity of periodontal disease as assessed by attachment loss in diabetic subjects 15 to 24 years of age was about twice that of the normal 15 to 24 year olds (Fig. 1). There was no difference from normal in attachment loss of those with impaired glucose tolerance. Similar results were noted when prevalence and severity were estimated by measuring radiographically apparent interproximal alveolar bone loss (Table 5 and Figure 2, respectively). The prevalence of radiographically detectable periodontal destruction increased with increasing age, and there was a markedly higher prevalence in the diabetic subjects in the 15 to 44 age groups as compared to the IGT and NGT subjects (Table 5). The severity of destructive disease as assessed by mean bone loss score for all subjects was higher in the diabetic subjects for all age groups (except those aged 55 or over) as compared to the IGT and NGT
BLS>2* No
Yes
No Yes
8Ü7 35
185 315
Totals 992 350
Totals
842
500
1,342
PAL>5
mmt
Agreement
=
Number of "No"
by both Number of "Yes" by both 807 + 315 .836 Total number of subjects 1,342 *Bone Loss Score > 2 refers to 25% or more interproximal alveolar bone loss on at least 1 tooth in the mouth. fProbing Attachment Loss > 5 mm refers to probing attachment loss of > 5 mm on at least 1 index tooth. _
RESULTS attachment level and alveolar bone loss measure destruction of different tissues in periodontal disease. The agreement between these two measures was 83.6% (Table 2). The "threshold levels" for each of these measures were arbitrarily selected and they differed in their ability to estimate the prevalence, with the bone loss on up to 28 teeth detecting more cases (500 out of 1,342 subjects) than the probing attachment level measured on 6 measured index teeth (350 out of 1,342 subjects). Even though as expected these two measures did not correlate exactly, they are useful in estimating the prevalence of periodontal disease in this dentate population and were both applied to all glucose tolerance categories.
Probing
Description of Population The distribution of the 1,342 subjects according to age, sex, and glucose tolerance is given in Table 3. The subjects in Table 3 are those among the total study group of 3,219 who fit the criteria for this study; i.e., index teeth must be present to complete all clinical measurements. From Table 3 it can be seen that 19% of these subjects had diabetes, 12% had impaired glucose tolerance, and 69% were normal with respect to glucose tolerance. The age distribution was markedly different for diabetic subjects as compared to normals. The diabetic subjects were found throughout the entire age range, whereas most of the normal glucose tolerant subjects were less than 35 years of age. There was a predominance Table 3: Distribution of the
1991
subjects (Fig. 2).
Table 6 shows the PI, GI, and CI scores, grouped by glucose tolerance and age for the 1,342 subjects. The mean percent of tooth surfaces with visible plaque (PI > 2) was very high, affecting half or more of the teeth in all diagnostic groups at all ages. Gingival bleeding (GI > 2) showed no age relationship and was very high, approaching 90% 1,342 Subjects by Glucose Tolerance, Age,
Age
at
Exam
15 to 24
25 to 34
35 to 44
_F__M_
_F_M_
_F_M_
and Sex
(In Years) 45
to
55 +
54
Diabetes
14
21
53
24
43
37
33
14
11
4
Totals 254
IGT
34
18
47
12
15
10
13
4
3
2
158
NGT
266
256
166
108
42
39
21
15
.3
14
930
_F_M
F__M
(19%)
(12%) (69% )
Totals
314
295
266
144
100
86
67
.33
17
20
1,342
Volume 62 Number 2
EMRICH, SHLOSSMAN, GENCO
Table 4: Prevalence of
Subjects
with
Probing Attachment
Loss
>
5
appropriate was detected when Chi-square goodness of fit tests were performed on the model. For both measures of periodontal destruction the factors which were highly significantly correlated to periodontal destruction were diabetes, age, and calculus (P < 0.0001 in each case). Destructive periodontal disease was weakly related to the DMFT when periodontal destruction was measured using the bone loss score (P 0.0387). As indicated in Tables 4 and 5 for PAL and BLS, respectively, destructive periodontal disease was significantly higher in subjects with diabetes than in either IGT or NGT subjects, while there was no significant difference between IGT or NGT subjects. Also, there was a significant increase in periodontal disease, as measured by either PAL or by BLS, with increasing age. Finally, again using either measure, periodontal disease increased with increasing levels of calculus. The risk indicators for Periodontitis that were highly statistically correlated (P < .0001) using both measures of periodontal tissue destruction are listed in Table 9. Diabe-
mm
Age Group (Years) Diabetes IGT NGT
%* % %
'"Subjects who have examination.
15 to 24 25.7 9.6 5.4
25 to 34 44.2 16.9 19.3
one or more
35 to 44 68.8 48.0 44.4
sites with PAL
>
45 to 54 85.1 64.7 75.0 5
mm at
55 + 86.7 80.0 76.5
the initial
=
for all groups, with no obvious differences related to glucose tolerance. Subgingival dental calculus increased through age 45, and then appeared to plateau. By the fifth decade, 60% to 70% of the teeth have subgingival calculus. From Table 6 it can be seen that diabetics appear to have higher plaque levels and at earlier ages more calculus; however, these differences did not reach statistical significance. The number of decayed, filled, or missing teeth were tabulated by age and glucose tolerance (Table 7). There was no apparent difference among the glucose tolerance groups with respect to DMFT, or decayed and filled (DF) teeth. However, the diabetic subjects had consistently more missing teeth than either of the other two groups. The mean percent of teeth with severe fluorosis, although relatively high, did not differ among the glucose tolerance groups
tes, age, and calculus were associated with an increased risk of Periodontitis. The diabetic subject had a risk of
Periodontitis 2.81 or 3.43 times higher than the non-diabetic using attachment loss or bone loss respectively as measures of destructive Periodontitis. Age was analyzed by decade and it
found that for each decade of life, the risk for developing Periodontitis was 3.10 and 4.28 using attachment loss and bone score respectively, as measures of Periodontitis. The relative risk associated with increasing amounts of calculus was assessed comparing the risk of calculus found on one versus three or six teeth. It was found that
(Table 7).
Statistical analysis using the linear logistic model to asthe relationship between the prevalence of destructive periodontal disease and diabetic status, and other potential contributing factors is presented in Table 8. Nothing insess
was
Attachment Loss and Glucose Tolerance 5
15 24 -
127
25 34
35 44
45 54
-
-
-
56+
Age
Figure 1: Mean probing attachment loss with respect to glucose tolerance and age for all 1,342 subjects. Diabetic subjects are those with a 2-hour post load glucose level greater than or equal to 200 mg/dl; IGT are those with 2-hour glucose level greater than or equal to 140 mg/dl and less than 200 mg/dl; and NGT refers to those with a 2-hour glucose level of less than 140 mg/dl. Error bars depict one standard error of the mean.
128
J Periodontol 1991
February
PERIODONTAL DISEASE IN TYPE II DISEASE
Table 5: Prevalence of
Subjects
with
a
Bone Loss Score
>
2
_Age Group (Years)_ 15 to 24
Diabetes IGT NGT
%*
'Subjects
who have
% %
28.6 17.3 7.7
25 to 34 67.5 44.1 34.7
one or more
teeth
35 to 44
45 to 54
55 +
88.8 60.0 67.9
95.7 88.2 86.1
93.3 100.0 100.0
exhibiting a radiographie
BLS
>
2, indicating greater than 25% loss of interproximal alveolar crestal height at the initial examination. The entire dentition was scored, except for third molars.
and dental health indices as they relate to the prevalence and severity of destructive periodontal disease, a subset of the 3,219 subjects who participated in the oral health survey were analyzed. The subset was selected on the basis of having six index teeth and all of the dependent and independent variables measured. This allows for measurement of variables which are tooth dependent, such as decayed and filled teeth, fluorosis, plaque, calculus, and gingival condition. The disadvantage of using this subset of 1,342 is that the subjects available for analysis represent that segment of the population who have not lost the teeth needed for measurement. The periodontal condition of the entire population is described elsewhere.15 The subset was generally comparable to the entire group; for example, the percentage of diabetic subjects in the subset was comparable (19% vs. 23%) to the overall sample. The validity of the findings concerning the relationship between the prevalence of destructive periodontal disease, age and glucose tolerance, as assessed in this subset of subjects with six index teeth, is supported by similar findings in the larger sample of 3,219 subjects.15 In addition to providing formal statistical tests of the relationship between the prevalence of destructive periodontal disease and diabetic status, age, sex, and the various oral health indices, the results of the linear logistic regression analysis can be used to estimate the risk of periodontal disease for this subset of subjects. Estimates of the magnitude of the effects of the variables, along with their statistical significance, can be accomplished by substituting status
;
the risk of Periodontitis increased markedly for both attachment loss and bone score as one to six additional teeth were found to harbor subgingival calculus. DISCUSSION The results of this study show that periodontal destruction, as measured by both the loss of probing attachment and radiographically apparent loss of alveolar bone, is more prevalent and of greater severity in subjects with diabetes than in those with normal glucose tolerance. The study was performed in the Gila River Indian Community, a wellstudied population of individuals who have the highest recorded prevalence of non-insulin-dependent diabetes in the world.11-12 Diabetes was found to be significantly related to both the prevalence and severity of periodontal disease in a manner which cannot be explained on the basis of age, sex, or several measures of oral health. In order to perform the statistical analysis of diabetic
Alveolar Bone Loss and Glucose Tolerance
1
H
i Diabetic IGT NGT
15-24
25-34
35-44
45-54
55+
Age
interproximal radiographie alveolar hone score with respect to glucose tolerance and age for all 1,342 subjects. Diabetic subjects are those with a 2-hour post load glucose level greater than or equal to 200 mg/dl; IGT are those with 2-hour glucose level greater than or equal to 140 mg/dl and less than 200 mg/ dl; and NGT refers to those with a 2-hour glucose level of less than 140 mg/dl. Error bars depict one standard error of the mean. Figure 2:
Mean
Volume 62 Number 2
EMRICH, SHLOSSMAN, GENCO
Table 6: Periodontal Indices. Visible
Plaque, Gingival Bleeding,
and Calculus
Age Group (Years) 15 to 24
25
to
34
35
45
to 44
to
55
54
-
Plaque Index* Diabetes IGT NGT Gingival Indext
Diabetes IGT NGT Calculus Indexi Diabetes IGT NGT
61.9 ±5.6 57.1 ±3.8 55.9±1.4
63.2 ±3.8 54.5 ±4.3 53.2±2.0
66.9±3.6 63.3 ±5.4 48.8 ±3.4
74.8 ±4.5 66.7 ±7.0 58.3±4.2
67.8 ± 9.5 40.0 ± 8.5 63.7± 8.2
94.8 ±2.3
97.6±0.7
94.4± 1.4
94.9±1.4
94.4+1.3
97.3 + 1.8
92.7 ±0.6
92.2± 1.0
89.3 ± 1.9
96.8± 1.4 91.2±3.5 92.6±2.2
95.6 ± 2.0 86.7± 6.2 93.1 ± 3.2
42.4±6.0 37.2±4.3 36.1 ± 1.3
57.3±3.9
62.5 ±3.8 58.7 + 6.7 59.3±3.7
69.9±4.6 64.7±7.9 67.6±5.6
67.8± 8.5 46.7± 12.2 75.5± 6.2
51.4 ±4.2 46.6±2.0
2 or 3) ± 1 SEM. "Data presented as mean % of teeth with visible plaque (PI 2 or 3), ± 1 SEM. 'Data presented as mean % of teeth with bleeding (GI *Data presented as mean % of teeth with subgingival calculus, ± 1 SEM. =
=
Table 7: Dental Indices.
Decayed, Missing,
or
Filled Teeth and Fluorosis
Age Group (Years) 15 to 24
25 to 34
35 to 44
45 to 54
55 +
7.5±0.7 7.0±0.5
10.7±0.5 9.8±0.6
10.0 + 0.6 9.8 ±1.9 9.4±0.6
10.3± 0.7 10.7± 1.4 9.7± 0.9
11.6+ 1.1 9.8± 1.3
7.1 ±0.5 7.9 ±0.9 7.5 ±0.5
5.9= 0.5 6.8± 0.9 6.3 ± 0.7
0.9 4.4 ± 1.2 5.2 ± 1.2
DMFT*
Diabetes IGT NGT
8.0±0.2
9,9+0.3
Diabetes IGT NGT
6.7±0.6 6.9 ±0.5 7.7 ±0.2
9.5 + 0.5
0.8 ±0.3 0.1 ±0.1
1.2 ±0.2 1.1 ±0.2 0.8±0.1
2.8 ±0.2
1.8±0.5 1.9±0.2
4.3 ± 0.4 3.8± 0.8 3.4± 0.5
6.3 ± 0.9 5.4± 1.8 4.7± 0.8
28.3±4.7 16.0 + 4.2 24.3 ±2.4
29.9±4.4 16.4 + 6.3 24.0 ±4.1
36.4± 5.9 44.1 ±10.3 28.5 ± 6.6
47.6± 9.7 12.1 + 12.1 43.1 ± 9.5
9.9± 1,3
1)1
Missing*
Diabetes IGT NGT
Fluorosist
Diabetes IGT NGT
8.7 ±0.6 9.0 ±0.2
0.3±0.0 15.8 + 5.6
12.9 ±4.0 16.8±1.5
'"Decayed (D), missing (M), or filled (F) teeth, data presented as 'Severe fluorosis, data presented as mean % of teeth ± 1 SEM.
mean ±
SEM.
Table 8: Estimated Regression Coefficients and P-Values* from the Linear Prevalence of Destructive Periodontal Disease
Probing Attachment Losst Variable
Intercept IGT Diabetes
Age
Sex Calculus
Plaque Gingival bleeding Fluorosis DMFT
SD
-7.4635 -0.0343 1.0335 0.1130 0.1523 0.0215 0.0036 0.0126 -0.0001
-0.0038
0.8498 0.2460 0.1965 0.0093 0.1701 0.0032 0.0032 0.0083 0.0022 0.0182
P-value
Periodontal Disease in Non-InsulinDependent Diabetes Mellitus Lawrence J.
Emrich, *f Marc Shlossman, and Robert J. Gencox *
The relationship between diabetes mellitus and oral health status was determined in Pirna Indians from the Gila River Indian Community in Arizona. This tribe of native Americans has the world's highest reported incidence and prevalence of non-insulindependent (type 2) diabetes mellitus. The probing attachment level, alveolar bone loss, age, sex, Calculus Index, Plaque Index, Gingival Index, fluorosis, and DMFT as well as the diabetic status was assessed in 1,342 Pirna Indians who were at least partially dentate. The prevalence and severity of destructive periodontal disease was determined by measuring probing attachment loss and radiographically apparent interproximal crestal alveolar bone loss, two independent but correlated indicators of periodontal destruction. Only diabetic status, age, and the presence of subgingival calculus were significantly associated with both increased prevalence and greater severity of destructive periodontal disease in this population. Diabetic status was significantly and strongly related to both the prevalence and severity of disease after adjusting for the effects of demographic variables and several indices of oral health including the Plaque Index. Subjects with type 2 diabetes have an increased risk of destructive Periodontitis with an odds ratio of 2.81 (95% confidence interval 1.91 to 4.13) when attachment loss is used to measure the disease. The odds ratio for diabetic subjects was 3.43 (95% confidence interval 2.28 to 5.16) where bone loss was used to measure periodontal destruction. These findings demonstrate that diabetes increases the risk of developing destructive periodontal disease about threefold. Furthermore, diabetes increases the risk of developing periodontal disease in a manner which cannot be explained on the basis of age, sex, and hygiene or other dental measures. Periodontitis should be considered a potential complication of diabetes in evaluation of patients. / Periodontol 1991; 62:123-130.
Key Words: Indians, North American; diabetes, mellitus; oral health index; periodontal
diseases/epidemiology; periodontal diseases/etiology.
Diabetes is a chronic metabolic disorder that affects 6% of the American population. Much of the information currently available concerning the relationship between diabetes and oral conditions, and most notably periodontal disease, has been obtained from the study of convenience samples such as patients attending diabetic clinics.1-9 This method may favor selection of patients with symptoms and complications, and may under-represent those with asymptomatic carbohydrate intolerance, mild symptoms, or those who have not sought medical care. We sought to provide a description of the clinical characteristics of diabetes-related periodontal disease, especially in its earliest stages, by the systematic examination of these relationships in a *Roswell Park Cancer Institute, Buffalo, NY. fDr. Emrich died June 23, 1990. department of Oral Biology State University of New York at Buffalo, NY.
Buffalo,
the basis of their systemic oral disease status. The Pirna Indians of the Gila River Indian Community in Arizona have the highest recorded incidence and prevalence of type 2 diabetes in the world.10 Fifty percent of those above 35 years of age have diabetes.10-12 Diabetes mellitus in the Pirnas is exclusively type 2, or non-insulindependent diabetes mellitus, and exhibits progression and complications comparable to other populations with type 2 diabetes.13-14 This tribe has been under continuous epidemiologie investigation since 1965 by the National Institute of Diabetes, Digestive and Kidney Diseases (NIDDK) in
large population not selected on or
Phoenix, Arizona."
An epidemiologie study of the oral health of the Pirna Indians was added to the investigations in 1983. The objectives of this study were 1) to estimate the prevalence and severity of periodontal disease; 2) to study the relationship
J Periodontoi
124
PERIODONTAL DISEASE IN TYPE II DISEASE
between the prevalence and severity of periodontal disease and diabetes; and 3) to investigate other factors such as age, sex, and periodontal health indices (e.g., gingival index, plaque index, calculus index) with respect to their relationship with periodontal disease in the Pirna Indians. Shlossman et al.15 report the results of a study of 3,219 Pirnas who were examined during the first 5 years of this study. It was found that median attachment loss and alveolar bone loss is higher in diabetic subjects for all age groups and for both sexes. However, the potential mediating effects of other oral health measures such as the calculus, plaque, gingival condition, fluorosis, and decayed, missing, or filled teeth (DMFT) could not be assessed since the edentulous as well as dentate portion of the population was analyzed. The current paper presents results for a subset of 1,342 of these 3,219 subjects who were 15 years of age or older and who were at least partially dentate, so that relevant oral health measures could be measured to determine whether diabetes is associated with destructive periodontal disease, or whether the relationship can be explained by some or all of these other oral health factors. MATERIALS AND METHODS
Subjects The study population consists of 1,342 Pirna or TohonoO'odham (Papago) Indians, members of two closely related tribes from the Gila River Indian Community in Arizona. Approximately every 2 years, each member of the community who is > 5 years of age, regardless of health, is asked to participate in a standardized examination12 that includes a medical history, physical and laboratory examinations, glucose tolerance test, and dental examination. The informed consent of all subjects was obtained after the nature of the procedures and possible risks had been fully explained. Diabetes
Diagnosis
Diabetic
status
of the subject is determined by the results of a modified 2-hour oral glucose tolerance test performed according to the World Health Organization criteria.'6 After a fasting blood is drawn, the subject ingests 75g of carbohydrate.8 Two hours later blood is drawn for determination of the plasma glucose concentration. Subjects are classified as having normal glucose tolerance (NGT) if the 2-hour plasma glucose concentration is < 7.77 mmol/L (< 140 mg/dl); impaired glucose tolerance (IGT) if the 2-hour glucose is > to 7.77 mmol/L and less than 11.1 mmol/L (140 < 200 mg/dl); and diabetes if the 2-hour glucose is > to 11.1 mmol/L. A subject is considered to have diabetes if this criterion is met, or if a review of the medical record revealed documentation of a diagnosis of diabetes, as pre-
viously described.12 'Glucola,
Ames
Co., Elkhart, IN.
February
1991
Oral Examination DMFT were charted, and Dean's classification system for dental fluorosis (FI)'7 was scored by examining all the teeth in the patient's mouth. Information on third molars is not considered in any of the analyses. The periodontal status was assessed by clinical probing and by radiographie analysis. The clinical oral periodontal examination utilizes six "index teeth" (upper right first molar, upper left central incisor, upper left first premolar, lower left first molar, lower right central incisor, and lower right first premolar). '8·19 If missing, a substitute tooth immediately distal was used. If this too was missing, the tooth immediately mesial was substituted and this change noted. If an acceptable substitute was missing, no information was recorded for that tooth. The following parameters were recorded for three surfaces (buccal, mesial, and lingual) of each of the six index teeth: the Calculus Index (CI, presence or absence of subgingival and/or supragingival calculus), the Plaque Index (PI)20 and Gingival Index (GI).21 A standardized color-coded probe'1 was used for measuring the probing pocket depth (PPD) on six surfaces of the tooth (mesiobuccal, midbuccal, distobuccal, mesiolingual, midlingual, and distolingual) and probing attachment loss (PAL) from the cemento-enamel junction (CEJ) to the depth of the pocket or sulcus on four sites for each index tooth (midbuccal, midlingual, and the deepest mesial and distal
interproximal points). Interproximal alveolar bone loss,
as the percentage of bone loss from the CEJ to the tooth apex, was measured using a modified Schei ruler22-23 at the deepest point on the mesial or distal of every tooth on a panoramic radiograph.* Bone loss scores (BLS) were assigned as follows: 0 if there was no detectable loss of alveolar crest; 1 if detectable loss was less than 25% of the distance from the CEJ to the apex; 2 if the loss was 25% to 49%; 3 if the loss was 50% to 74%; and 4 if the loss was 75% or greater.
Criteria for Inclusion of Subjects This study deals with a subset of 1,342 subjects of the 3,219 subjects who participated in the oral health survey. Measurements for probing attachment loss, Calculus Index, Plaque Index, and Gingival Index on all 6 index teeth; and fluorosis index, DMFT, and bone loss score for the whole mouth, as well as information regarding age, sex, and glucose tolerance were required for inclusion in these analyses. Subjects were excluded if they were missing any of the six index teeth or their substitutes; had a history of rheumatic fever or other heart problems requiring prophylactic
antibiotics; were pregnant (since calculus obstructed the
available;
no radiographs were taken), CEJ; adequate radiographs were un-
or they were children with mixed dentition. The dental examinations and radiographie scoring were con-
•HuFriedy, PCP-10, Chicago, "Siemens, OP-5.
IL.
Volume 62 Number 2
EMRICH, SHLOSSMAN, GENCO
ducted by a single examiner the patient's diabetic status.
(MS)
without
knowledge
of
examined at a biennial examination were 14 to 21 day period for duplicate measurements (probing attachment loss and rescoring of radiographs). Intra-examiner reproducibility was assessed by the Kappa statistic.24 The Kappa statistic for probing attachment loss was 76%, and the Kappa statistic for the bone loss score was 88%. Since the population was examined over several years, the stability of measurements over time was estimated by evaluating the mean and standard error of the mean for attachment level measurements and bone loss scores. No major year-to-year differences in mean attachment level or bone loss score for the diagnostic subgroups or in the standard error of the mean were found. a
Statistical Methods Prevalence. The relationship between the prevalence of destructive periodontal disease and diabetic status and other potential prognostic factors was assessed using the linear
logistic model:25 P(Y=1) exp(B0 + B1X1 + + BkXk)/[l + exp(B()-rB1X1 + + BkXk)] where P(Y 1) denotes the probability that a subject has had destructive periodontal disease; X,, Xk are functions of the prognostic factors of interest; and B,„ B,, Bk are unknown regression coefficients. Since periodontal disease involves destruction of con...
...
=
nective tissue and bone, assessments of both were made to compare diabetic and non-diabetic subjects. PAL is a linear measurement of connective tissue loss made from a fixed point, and can be measured with considerable accuracy and reproducibility; the BLS is a measure of interproximal alveolar bony changes assessed semi-quantitatively on radiographs. Two definitions of destructive periodontal disease were employed for these analyses and for estimating the overall prevalence of destructive periodontal disease. Each subject (not tooth or surface) was classified according to the following definitions: (i) Y=l if PAL > 5 for any surface and Y 0 otherwise, or (ii) Y 1 if BLS > 2 for any tooth and Y 0 otherwise. These definitions of destructive periodontal disease do not necessarily estimate the absolute prevalence of Periodontitis. The PAL > 5 mm as a threshold value underestimates the prevalence of destructive periodontal disease as compared to the prevalence determined by bone loss as defined above. However, by systematically applying each definition to all three glucose tolerance groups (diabetic, IGT, and NGT), biases in comparing prevalence should be minimized, and comparisons between groups can be made. =
=
=
=
=
=
=
Thirty subjects
=
Regressor Variables
1 if the subject had impaired glucose tolerance (IGT) and X, 0 otherwise. 0 otherwise. 1 if the subject had diabetes and X, X2 Xj age (in years) at dental examination. 0 if the subject was male. 1 if the subject was female and X4 X4 X, percent of teeth with CI>2 (subgingival calculus) 1Ü0 (number of teeth with at least one CI>2)/6. X„ percent of teeth with PI>2 (visible plaque) 100 (number of teeth with at least one PI>2)/6. X7 percent of teeth with GI>2 (bleeding) 100 (number of teeth with at least one Gl>2)/6. 3 (severe fluorosis) Xx percent of teeth with Fl 100 (number of teeth with FI 3)/number of teeth scored. DMFT (number of decayed, missing, or filled teeth). X,,
X,
Reproducibility recalled within
Table 1: Definition of
125
= =
=
=
= =
=
=
=
=
=
=
=
=
Separate analyses were performed for each definition of periodontal disease. For each analysis, the regressor (X)
defined as in Table 1. By including the variables in the model, the principal tests of hypotheses concerning the relationship between diabetic status and periodontal disease were "adjusted" for the potentially confounding effects of these other factors. In addition, estimates of, and tests of hypotheses about, the relationship between periodontal disease and the factors listed in Table 1 also could be obtained directly from these analyses. We were able to carry out such analyses only on those subjects with the requisite teeth which were directly measured, realizing that by not including subjects who were edentulous, we were eliminating some of those who may have suffered from severe periodontal disease. The regression coefficients were estimated, as were the variances and covariances of the estimates, using standard maximum likelihood techniques, as implemented in the Statistical Analysis System (SAS) computer package. Statistical tests of hypotheses about the coefficients were performed using two-tailed tests.
variables
were
Severity. Techniques similar
to those described above for studies were prevalence employed to investigate the between the relationship severity of destructive periodontal disease and diabetic status and the other factors listed above, except that a linear model26 was employed. Two definitions of severity, one corresponding to each definition of destructive periodontal disease given above, were employed for these analyses: (i) Y square root of PAL maximum for the average subject square root of total (over teeth) maximum (over surfaces), PAL/6, and (ii) Y square root of average BLS for the subject square root of total BLS over all teeth/no. of teeth scored. The square root transformation was used to make the normal distribution assumptions for the linear model more tenable. Estimates of the regression coefficients (B,„ B,, Bk) and their estimated variances and covariances, were obtained using standard least squares/maximum likelihood techniques, as implemented in SAS.
the
=
=
=
=
J Periodontol 126
February
PERIODONTAL DISEASE IN TYPE II DISEASE
Table 2: Correlation Between Two Measures of Periodontal Destruction
of females seen in virtually all glucose tolerance and age groups. The prevalance of subjects with probing attachment loss > 5 mm is shown in Table 4. For each of the three glucose tolerance groups, there was an increasing prevalence of destructive disease seen with increasing age up to ages 45 to 54 (Table 4). There was a higher prevalence of periodontal disease in the diabetic subjects at all age groups as compared to the IGT and NGT subjects. The difference was greatest at the lower age groups, with diabetic subjects aged 15 to 24 suffering 4.8 times more periodontal disease than normal subjects, and diabetic subjects aged 25 to 34 had 2.3 times more periodontal disease than normal subjects. The IGT group was close to the normal in all age groups. The severity of destructive disease, as estimated by mean attachment levels, for all the subjects is depicted in Figure 1. Probing attachment loss is higher in the diabetic subjects as compared with normal subjects for all ages except for those 55 years or older. The differences were greater at younger age groups. The severity of periodontal disease as assessed by attachment loss in diabetic subjects 15 to 24 years of age was about twice that of the normal 15 to 24 year olds (Fig. 1). There was no difference from normal in attachment loss of those with impaired glucose tolerance. Similar results were noted when prevalence and severity were estimated by measuring radiographically apparent interproximal alveolar bone loss (Table 5 and Figure 2, respectively). The prevalence of radiographically detectable periodontal destruction increased with increasing age, and there was a markedly higher prevalence in the diabetic subjects in the 15 to 44 age groups as compared to the IGT and NGT subjects (Table 5). The severity of destructive disease as assessed by mean bone loss score for all subjects was higher in the diabetic subjects for all age groups (except those aged 55 or over) as compared to the IGT and NGT
BLS>2* No
Yes
No Yes
8Ü7 35
185 315
Totals 992 350
Totals
842
500
1,342
PAL>5
mmt
Agreement
=
Number of "No"
by both Number of "Yes" by both 807 + 315 .836 Total number of subjects 1,342 *Bone Loss Score > 2 refers to 25% or more interproximal alveolar bone loss on at least 1 tooth in the mouth. fProbing Attachment Loss > 5 mm refers to probing attachment loss of > 5 mm on at least 1 index tooth. _
RESULTS attachment level and alveolar bone loss measure destruction of different tissues in periodontal disease. The agreement between these two measures was 83.6% (Table 2). The "threshold levels" for each of these measures were arbitrarily selected and they differed in their ability to estimate the prevalence, with the bone loss on up to 28 teeth detecting more cases (500 out of 1,342 subjects) than the probing attachment level measured on 6 measured index teeth (350 out of 1,342 subjects). Even though as expected these two measures did not correlate exactly, they are useful in estimating the prevalence of periodontal disease in this dentate population and were both applied to all glucose tolerance categories.
Probing
Description of Population The distribution of the 1,342 subjects according to age, sex, and glucose tolerance is given in Table 3. The subjects in Table 3 are those among the total study group of 3,219 who fit the criteria for this study; i.e., index teeth must be present to complete all clinical measurements. From Table 3 it can be seen that 19% of these subjects had diabetes, 12% had impaired glucose tolerance, and 69% were normal with respect to glucose tolerance. The age distribution was markedly different for diabetic subjects as compared to normals. The diabetic subjects were found throughout the entire age range, whereas most of the normal glucose tolerant subjects were less than 35 years of age. There was a predominance Table 3: Distribution of the
1991
subjects (Fig. 2).
Table 6 shows the PI, GI, and CI scores, grouped by glucose tolerance and age for the 1,342 subjects. The mean percent of tooth surfaces with visible plaque (PI > 2) was very high, affecting half or more of the teeth in all diagnostic groups at all ages. Gingival bleeding (GI > 2) showed no age relationship and was very high, approaching 90% 1,342 Subjects by Glucose Tolerance, Age,
Age
at
Exam
15 to 24
25 to 34
35 to 44
_F__M_
_F_M_
_F_M_
and Sex
(In Years) 45
to
55 +
54
Diabetes
14
21
53
24
43
37
33
14
11
4
Totals 254
IGT
34
18
47
12
15
10
13
4
3
2
158
NGT
266
256
166
108
42
39
21
15
.3
14
930
_F_M
F__M
(19%)
(12%) (69% )
Totals
314
295
266
144
100
86
67
.33
17
20
1,342
Volume 62 Number 2
EMRICH, SHLOSSMAN, GENCO
Table 4: Prevalence of
Subjects
with
Probing Attachment
Loss
>
5
appropriate was detected when Chi-square goodness of fit tests were performed on the model. For both measures of periodontal destruction the factors which were highly significantly correlated to periodontal destruction were diabetes, age, and calculus (P < 0.0001 in each case). Destructive periodontal disease was weakly related to the DMFT when periodontal destruction was measured using the bone loss score (P 0.0387). As indicated in Tables 4 and 5 for PAL and BLS, respectively, destructive periodontal disease was significantly higher in subjects with diabetes than in either IGT or NGT subjects, while there was no significant difference between IGT or NGT subjects. Also, there was a significant increase in periodontal disease, as measured by either PAL or by BLS, with increasing age. Finally, again using either measure, periodontal disease increased with increasing levels of calculus. The risk indicators for Periodontitis that were highly statistically correlated (P < .0001) using both measures of periodontal tissue destruction are listed in Table 9. Diabe-
mm
Age Group (Years) Diabetes IGT NGT
%* % %
'"Subjects who have examination.
15 to 24 25.7 9.6 5.4
25 to 34 44.2 16.9 19.3
one or more
35 to 44 68.8 48.0 44.4
sites with PAL
>
45 to 54 85.1 64.7 75.0 5
mm at
55 + 86.7 80.0 76.5
the initial
=
for all groups, with no obvious differences related to glucose tolerance. Subgingival dental calculus increased through age 45, and then appeared to plateau. By the fifth decade, 60% to 70% of the teeth have subgingival calculus. From Table 6 it can be seen that diabetics appear to have higher plaque levels and at earlier ages more calculus; however, these differences did not reach statistical significance. The number of decayed, filled, or missing teeth were tabulated by age and glucose tolerance (Table 7). There was no apparent difference among the glucose tolerance groups with respect to DMFT, or decayed and filled (DF) teeth. However, the diabetic subjects had consistently more missing teeth than either of the other two groups. The mean percent of teeth with severe fluorosis, although relatively high, did not differ among the glucose tolerance groups
tes, age, and calculus were associated with an increased risk of Periodontitis. The diabetic subject had a risk of
Periodontitis 2.81 or 3.43 times higher than the non-diabetic using attachment loss or bone loss respectively as measures of destructive Periodontitis. Age was analyzed by decade and it
found that for each decade of life, the risk for developing Periodontitis was 3.10 and 4.28 using attachment loss and bone score respectively, as measures of Periodontitis. The relative risk associated with increasing amounts of calculus was assessed comparing the risk of calculus found on one versus three or six teeth. It was found that
(Table 7).
Statistical analysis using the linear logistic model to asthe relationship between the prevalence of destructive periodontal disease and diabetic status, and other potential contributing factors is presented in Table 8. Nothing insess
was
Attachment Loss and Glucose Tolerance 5
15 24 -
127
25 34
35 44
45 54
-
-
-
56+
Age
Figure 1: Mean probing attachment loss with respect to glucose tolerance and age for all 1,342 subjects. Diabetic subjects are those with a 2-hour post load glucose level greater than or equal to 200 mg/dl; IGT are those with 2-hour glucose level greater than or equal to 140 mg/dl and less than 200 mg/dl; and NGT refers to those with a 2-hour glucose level of less than 140 mg/dl. Error bars depict one standard error of the mean.
128
J Periodontol 1991
February
PERIODONTAL DISEASE IN TYPE II DISEASE
Table 5: Prevalence of
Subjects
with
a
Bone Loss Score
>
2
_Age Group (Years)_ 15 to 24
Diabetes IGT NGT
%*
'Subjects
who have
% %
28.6 17.3 7.7
25 to 34 67.5 44.1 34.7
one or more
teeth
35 to 44
45 to 54
55 +
88.8 60.0 67.9
95.7 88.2 86.1
93.3 100.0 100.0
exhibiting a radiographie
BLS
>
2, indicating greater than 25% loss of interproximal alveolar crestal height at the initial examination. The entire dentition was scored, except for third molars.
and dental health indices as they relate to the prevalence and severity of destructive periodontal disease, a subset of the 3,219 subjects who participated in the oral health survey were analyzed. The subset was selected on the basis of having six index teeth and all of the dependent and independent variables measured. This allows for measurement of variables which are tooth dependent, such as decayed and filled teeth, fluorosis, plaque, calculus, and gingival condition. The disadvantage of using this subset of 1,342 is that the subjects available for analysis represent that segment of the population who have not lost the teeth needed for measurement. The periodontal condition of the entire population is described elsewhere.15 The subset was generally comparable to the entire group; for example, the percentage of diabetic subjects in the subset was comparable (19% vs. 23%) to the overall sample. The validity of the findings concerning the relationship between the prevalence of destructive periodontal disease, age and glucose tolerance, as assessed in this subset of subjects with six index teeth, is supported by similar findings in the larger sample of 3,219 subjects.15 In addition to providing formal statistical tests of the relationship between the prevalence of destructive periodontal disease and diabetic status, age, sex, and the various oral health indices, the results of the linear logistic regression analysis can be used to estimate the risk of periodontal disease for this subset of subjects. Estimates of the magnitude of the effects of the variables, along with their statistical significance, can be accomplished by substituting status
;
the risk of Periodontitis increased markedly for both attachment loss and bone score as one to six additional teeth were found to harbor subgingival calculus. DISCUSSION The results of this study show that periodontal destruction, as measured by both the loss of probing attachment and radiographically apparent loss of alveolar bone, is more prevalent and of greater severity in subjects with diabetes than in those with normal glucose tolerance. The study was performed in the Gila River Indian Community, a wellstudied population of individuals who have the highest recorded prevalence of non-insulin-dependent diabetes in the world.11-12 Diabetes was found to be significantly related to both the prevalence and severity of periodontal disease in a manner which cannot be explained on the basis of age, sex, or several measures of oral health. In order to perform the statistical analysis of diabetic
Alveolar Bone Loss and Glucose Tolerance
1
H
i Diabetic IGT NGT
15-24
25-34
35-44
45-54
55+
Age
interproximal radiographie alveolar hone score with respect to glucose tolerance and age for all 1,342 subjects. Diabetic subjects are those with a 2-hour post load glucose level greater than or equal to 200 mg/dl; IGT are those with 2-hour glucose level greater than or equal to 140 mg/dl and less than 200 mg/ dl; and NGT refers to those with a 2-hour glucose level of less than 140 mg/dl. Error bars depict one standard error of the mean. Figure 2:
Mean
Volume 62 Number 2
EMRICH, SHLOSSMAN, GENCO
Table 6: Periodontal Indices. Visible
Plaque, Gingival Bleeding,
and Calculus
Age Group (Years) 15 to 24
25
to
34
35
45
to 44
to
55
54
-
Plaque Index* Diabetes IGT NGT Gingival Indext
Diabetes IGT NGT Calculus Indexi Diabetes IGT NGT
61.9 ±5.6 57.1 ±3.8 55.9±1.4
63.2 ±3.8 54.5 ±4.3 53.2±2.0
66.9±3.6 63.3 ±5.4 48.8 ±3.4
74.8 ±4.5 66.7 ±7.0 58.3±4.2
67.8 ± 9.5 40.0 ± 8.5 63.7± 8.2
94.8 ±2.3
97.6±0.7
94.4± 1.4
94.9±1.4
94.4+1.3
97.3 + 1.8
92.7 ±0.6
92.2± 1.0
89.3 ± 1.9
96.8± 1.4 91.2±3.5 92.6±2.2
95.6 ± 2.0 86.7± 6.2 93.1 ± 3.2
42.4±6.0 37.2±4.3 36.1 ± 1.3
57.3±3.9
62.5 ±3.8 58.7 + 6.7 59.3±3.7
69.9±4.6 64.7±7.9 67.6±5.6
67.8± 8.5 46.7± 12.2 75.5± 6.2
51.4 ±4.2 46.6±2.0
2 or 3) ± 1 SEM. "Data presented as mean % of teeth with visible plaque (PI 2 or 3), ± 1 SEM. 'Data presented as mean % of teeth with bleeding (GI *Data presented as mean % of teeth with subgingival calculus, ± 1 SEM. =
=
Table 7: Dental Indices.
Decayed, Missing,
or
Filled Teeth and Fluorosis
Age Group (Years) 15 to 24
25 to 34
35 to 44
45 to 54
55 +
7.5±0.7 7.0±0.5
10.7±0.5 9.8±0.6
10.0 + 0.6 9.8 ±1.9 9.4±0.6
10.3± 0.7 10.7± 1.4 9.7± 0.9
11.6+ 1.1 9.8± 1.3
7.1 ±0.5 7.9 ±0.9 7.5 ±0.5
5.9= 0.5 6.8± 0.9 6.3 ± 0.7
0.9 4.4 ± 1.2 5.2 ± 1.2
DMFT*
Diabetes IGT NGT
8.0±0.2
9,9+0.3
Diabetes IGT NGT
6.7±0.6 6.9 ±0.5 7.7 ±0.2
9.5 + 0.5
0.8 ±0.3 0.1 ±0.1
1.2 ±0.2 1.1 ±0.2 0.8±0.1
2.8 ±0.2
1.8±0.5 1.9±0.2
4.3 ± 0.4 3.8± 0.8 3.4± 0.5
6.3 ± 0.9 5.4± 1.8 4.7± 0.8
28.3±4.7 16.0 + 4.2 24.3 ±2.4
29.9±4.4 16.4 + 6.3 24.0 ±4.1
36.4± 5.9 44.1 ±10.3 28.5 ± 6.6
47.6± 9.7 12.1 + 12.1 43.1 ± 9.5
9.9± 1,3
1)1
Missing*
Diabetes IGT NGT
Fluorosist
Diabetes IGT NGT
8.7 ±0.6 9.0 ±0.2
0.3±0.0 15.8 + 5.6
12.9 ±4.0 16.8±1.5
'"Decayed (D), missing (M), or filled (F) teeth, data presented as 'Severe fluorosis, data presented as mean % of teeth ± 1 SEM.
mean ±
SEM.
Table 8: Estimated Regression Coefficients and P-Values* from the Linear Prevalence of Destructive Periodontal Disease
Probing Attachment Losst Variable
Intercept IGT Diabetes
Age
Sex Calculus
Plaque Gingival bleeding Fluorosis DMFT
SD
-7.4635 -0.0343 1.0335 0.1130 0.1523 0.0215 0.0036 0.0126 -0.0001
-0.0038
0.8498 0.2460 0.1965 0.0093 0.1701 0.0032 0.0032 0.0083 0.0022 0.0182
P-value
