260

Assessment of Risk for Periodontal Disease. I. Risk Indicators for Attachment Loss
Sara G. Grossi, * Joseph J. Zambon, * Alex W. Ho, * Gary Koch,f Robert G. Eli E. Machtet, * Ola M. Norderyd, * and Robert J. Genco*

Dunford, *

Specific risk indicators associated with either susceptibility or resistance to severe forms of periodontal disease were evaluated in a cross-section of 1,426 subjects, 25 to 74 years of age, mostly metropolitan dwellers, residing in Erie County, New York, and surrounding areas. The study sample exhibited a wide range of periodontal disease experience defined by different levels of attachment loss. Therefore, it was possible to accurately assess associations between the extent of periodontal disease and patient characteristics including age, smoking, systemic diseases, exposure to occupational hazards, and subgingival microbial flora. Age was the factor most strongly associated with attachment loss, with odds ratios for subjects 35 to 44 years old ranging from 1.72 (95% CI: 1.18 to 2.49) to 9.01 (5.86 to 13.89) for subjects 65 to 74 years old. Diabetes mellitus was the only systemic disease positively associated with attachment loss with an odds ratio of 2.32 (95% CI: 1.17-4.60). Smoking had relative risks ranging from 2.05 (95% CI: 1.47-2.87) for light smokers increasing to 4.75 (95% CI: 3.28-6.91) for heavy smokers. The presence of two bacteria, Porphyromonasgingivalis and Bacteroides forsythus, in the subgingival flora represented risks of 1.59 (95% CI: 1.11-2.25) and 2.45 (95% CI: 1.87-3.24), respectively. Our results show that age, smoking, diabetes mellitus, and the presence of subgingival P. gingivalis and B. forsythus are risk indicators for attachment loss. These associations remain valid after controlling for gender, socioeconomic status, income, education, and oral hygiene status expressed in terms of supragingival plaque accumulation and subgingival calculus. Longitudinal, intervention, and etiology-focused studies will establish whether these indicators are true risk factors.

JPeriodontol 1994;65:260-267.

Key Words: Systemic diseases/complications; occupational diseases; risk factors; periodontal diseases/epidemiology; diabetes mellitus/complications; tobacco/adverse effects;

periodontal diseases/microbiology.

Risk assessment has become increasingly important in the prevention of chronic diseases and has recently being applied to oral diseases. This has mainly been due to the realization that severe forms of the two most prevalent dental diseases; i.e., caries and periodontal disease, are clustered in a minority of the population who are hypothesized as being at higher risk.1-3 This observation has triggered interest in identifying susceptible individuals as well as the factors that put them at higher risk to develop these diseases. This could be done by means of cross-sectional studies which allow the identification of risk indicators or mark*Periodontal Disease Research Center, Department of Oral Biology, School of Dental Medicine, State University of New York, Buffalo, NY. 'University of North Carolina, School of Public Health, Department of Biostatistics, Chapel Hill, NC.

the assessment of causation between attributes and disease outcome, identifying, therefore, risk factors. A logical extension of this definition of risk factor includes the ability to be modified or modulated resulting in a reduction in disease outcome. This rationale provides the basis for intervention strategies both as part of active treatments and health care planning. Consequently, a new paradigm is emerging to control dental disease which involves identification and targeting high risk individuals and intercepting the disease process. Therefore, a targeted prevention-based approach is proposed to replace more conventional treatment-based approaches.

by determining associations between attributes and disease with no inference on causality. A second experimental approach includes longitudinal studies which allow
ers

outcome

Volume 65 Number 3

GROSSI, ZAMBON, HO, KOCH, DUNFORD, MACHTEI, NORDERYD, GENCO

261

In concert with these goals, we conducted a cross-sectional study of a broadly defined population to investigate the role of systemic diseases, socio-economic factors, smoking, occupational hazards, and subgingival bacteria as risk indicators for attachment loss. The Erie County Study, an epidemiological survey designed to identify risk indicators and putative risk factors for periodontal disease, included a cross-sectional and a longitudinal component. The cross-sectional phase has been completed, providing the basis for this report. The longitudinal phase recently being initiated, will include two biennial examinations. All patient examinations and laboratory assessments were conducted entirely at the Periodontal Disease Research Center at Buffalo, NY.

MATERIALS AND METHODS

tablished the presence of teeth and no need for antibiotic prophylaxis for subacute bacterial endocarditis. The subjects were then asked to complete a questionnaire including the following sections; 1) demographic and socioeconomic (SES); 2) medical history; reporting symptoms and diagnosed systemic diseases; 3) dental habits and dental care utilization; and 4) history of cigarette smoking and exposure to occupational hazards. The demographic-medical questionnaire was administered to all subjects in the study. Marginally illiterate and individuals not fluent in English received the questionnaire in an interview format and/or in their native language. All subjects were examined at the Periodontal Disease Research Center. Periodontal examinations were carried out

study population consisted of a total of 1,426 subjects (741 females and 685 males) between the ages of 25 and 74. The individuals were all inhabitants of Erie County, New York and surrounding areas, essentially a metropolitan community. The study subjects were recruited using a combination of sampling methods including; 1) random selection from census tract data (365); 2) convenience, patients who presented for treatment at the State University of New York at Buffalo, School of Dental Medicine (325); and 3) respondents to advertisements in local newspapers (736). All subjects signed institutional informed consent forms. Approximately equal numbers of subjects were recruited in each of 5 age decades. Rather than population representation, the main objective of this cross-sectional sample was to display a broad variation in dental explanatory variables to assess the relationship between explanatory and outcome variables. Accordingly, the study sample included a wide range of periodontal disease severity defined by different levels of attachment loss. Subjects with no attachment loss were regarded as controls and those with attachment loss as cases, thereby allowing the estimation of odds ratios as a measurement of relative risk as in a case-control study. Initial screening to determine eligibility for the study es-

The

Subjects

votella intermedia.* The association between mean attachment loss and the various demographic, systemic, and microbial parameters was analyzed. For the purpose of analysis, the demographic and socioeconomic parameters were stratified in the following manner. Age was stratified into 5 decades (25 to 34, 35 to 44, 45 to 54, 55 to 64, and 65 to 74 years of age). The socioeconomic variables were stratified as gender-female or male, race-black, white or "other" which included Native-American, Asian, or Pacific Islander. Education and income were dichotomized as high school, completed high school education or lower or college, completed at least 1 year of college education; income, moderate, maximum annual income below US $30,000 or high, maximum annual income equal to or more than US $30,000. History of smoking was reported as the number of cigarettes smoked per day and the number of years smoked. Smoking was then quantified as a composite value of the number of packs of cigarettes smoked per day x number of years smoked; i.e., the number of packyears. The smokers exhibited a substantial range of packyears from 1 to 150. The group was then stratified in 5 ordered categories including nonsmoker, very light, or occasional smoker (>0 5.2 packyears); light (5.3 15.0 packyears); moderate (15.1 30.0 packyears); and heavy smokers (30.1 150.0 packyears). Occupational hazards (OH) assessed included chemicals, asbestos, radiation, and other. The assessment was done in a dichotomous manner; i.e., positive or negative self-reported history of exposure to the potential hazard.
-

by a team of trained and calibrated examiners and included assessment of supragingival plaque, gingival bleeding, subgingival calculus, probing depth (PD), and clinical attachment level (CAL) for all teeth present in the dentition as previously described.4 The mean CAL was computed for each subject, this value represented mean attachment loss (AL) and was used as the outcome variable. Two pooled subgingival plaque samples were taken using the paper point technique.5 One sample was taken from the mesio-buccal surface of 6 maxillary teeth (#3, 5, 7, 9, 12, 14) and one from the mesio-buccal surface of 6 mandibular teeth (#19, 21, 23, 25, 28, 30). The samples were assayed using immunofluorescence microscopy5-6 for the following "target" microorganisms: Actinobacillus actinomycetemcomitans, Bacteroides forsythus, Campylobacter rectus, Capnocytophaga species, Eubacterium saburreum, Fusobacterium nucleatum, Porphyromonas gingivalis, and Pre-

Data Analysis A preliminary step in the analysis involved stratifying AL into 5 ordered categories: healthy (0 to 1 mm AL); low (1.1 to 2.0 mm AL); moderate (2.1 to 3.0 mm AL); high (3.1 to 4.0 mm AL); and severe (4.1 to 8.0 mm AL). The range of AL observed in the sample was from 0.1 to 7.78 mm. Approximately 3% of the sample exhibited AL above 5.0 mm. Mean attachment loss, although not intended as

262

RISK ASSESSMENT FOR PERIODONTAL DISEASE. I. ATTACHMENT LOSS

J Periodontol March 1994

an indication of treatment needs, provides an estimate of the historical amount of periodontal destruction in a given patient.7'8 The variation in mean AL displayed by the sample enabled the case-control analysis as a gradient or continuum in severity of periodontal destruction. Contingency tables and Spearman rank correlations were used to examine the association of the ordered classification of attachment loss with age, systemic diseases, SES, smoking, occupational hazards, and subgingival microorganisms. Among the systemic diseases, only those with sufficient prevalence (at least 15 subjects with a positive history) to support approximately normal estimates for regression analysis were considered. An ordinal stepwise logistic regression was used adjusting for age because of its known strong association with attachment loss. The method was the proportional odds (cumulative logit) extension of logistic regression9 using SAS procedure LOGISTIC. For the ordinal stepwise logistic regression, those variables with a significance level of 0.10 or less were entered into the model in a stepwise approach. Estimates were obtained for the slope coefficients, odds ratios, and the 95% confidence intervals of the associations of severity of AL with systemic diseases, socioeconomic status, occupational hazards, and subgingival microorganisms. Finally, adjustments for smoking, plaque and calculus were made when appropriate for factors which could have been surrogates for these variables.

not add up to 100%. Our sample included 52% females and 48% males distributed proportionally among all 5 age categories. Caucasians were by far the most prevalent race comprising 90.3% of the total population, African-Americans included 6.2%, and all other races represented 2.5% of the entire population. Forty percent (40.8%) of the population had high school education or lower, and 58.4% had completed at least one year of college or higher. Thirty one percent (31.1%) had moderate income and 37.7% had high income. Thirty-nine percent (39.2%) of the subjects had never smoked, whereas the remaining 60.8% had engaged in different levels of cigarette smoking (Table 2). The proportion of non-smokers in each age category was relatively comparable ranging from 6.3% to 8.9%. The proportion of smokers, on the contrary, varied according to the age cat-

acteristics; i.e., education and income, do

The distribution of individuals according to attachment loss and age is shown in Table 3. Of the 79 "healthy" subjects, 68 (86%) were in age categories 1 and 2, while 331 (55%) of the 597 subjects with little AL were included in age categories 1 and 2, and 42 (34%) of the 123 subjects with severe AL were in age category 5. Therefore, with increasing age, the percent of subjects showing severe mean AL increased.

egories (Table 2).

RESULTS The demographic and socioeconomic characteristics of the study population are depicted in Table 1. Non-respondent cases are not reported, but nevertheless are accounted for by calculating individual percentages based on the total sample. Therefore, the percentages for some of the char-

Systemic Diseases The presence of systemic diseases was assessed including the following categories: 1) cardiovascular, 2) pulmonary, 3) allergy, 4) metabolic, 5) infectious, 6) neoplastic, and 7) sexually transmitted. The most prevalent self-reported diseases in our sample were allergies, asthma/hay fever, and hives; followed by high blood pressure, arthritis, and anemia (Fig. 1). There was a tendency for more systemic
County Study Population by Age

Table 1. Socio-Economic Status of the Erie

Decade

(Age

Category
Total Gender Female Male Race Black White Other Education High school

25-34

(%)

35-44

(%)

45-54

Years) 55-64 (%) (%)

275
147 128 20 248 3 146 125

in

65-74 263

(%)

Total

289
146 143 18 255 13

307

292

1,426

(50.5)* (49.5) (6.2) (88.2) (4.5) (30.8) (68.9) (36.3)5 (30.1)

160 147 15 283 8

91 214

(52.1) (47.9) (4.9) (92.2) (2.6) (29.6) (69.7) (30.9) (41.7)

155 137 25 256 9

(53.1) (46.9)

(53.5) (46.5)

133 130
11 245 3

(50.6) (49.4) (4.2) (93.2) (1.1) (58.9) (41.1) (42.6) (26.6)

741 685

(52.0)' (48.0) (6.2) (2.5)

(87.7) (3.1)
(34.6) (64.0) (20.2) (50.3)

(8.6)

(7.3) (90.2) (1.1)

(53.1) (45.5) (26.2) (38.5)

89

1,287 (90.3)
36
582 833
443

College

89 199

101 187
59 147

155 108 112 70

(40.8) (58.4) (31.1) (37.7)

Income

Moderate*

Highs

105 87

95 128

72 106

538

Maximum annual income below US$ 30,000. 5Not all subjects provided this data, therefore, percentages
group. "Maximum annual income
>

*Percentages are based on number of individuals within each age group. 'Percentages are based on a total 1,426 individuals.
are

based

on

number of individuals in each age

US$ 30,000.

Volume 65 Number 3
Table 2. Distribution of Cigarette Smokers

GROSSI, ZAMBON, HO, KOCH, DUNFORD, MACHTEI, NORDERYD, GENCO

by Age
in Years) 55-64 (%) (%) 82 (6.3) 101 (7.7) 21 (1.6) 31 (2.4) 30 (2.3) 47 (3.6) 50 (3.8) 33 (2.5) 66 (5.0) 56 (4.3) 266 (20.3) 251 (19.1) 45-54

263

Decade

(Age

Smoking Habit
None

25-34
117 70 49 28 8 272

Very light Light

Moderate

Heavy
Total

(%) (8.9) (5.3) (3.7) (2.1) (0.6) (20.7)

(%) 100 (7.6) 46 (3.5) 45 (3.4) 64 (4.9) 24 (1.8) 279 (21.3)

35-44

65-74 114 22 37 23 48 244

(%) (8.7) (1.7) (2.8) (1.8) (3.7) (18.6)

Total
514 190 208 198 202

(39.2) (14.5) (15.8) (15.1) (15.4)

1,312

Table 3. Distribution of

Age
Age,
Years

Subjects by Levels

of Attachment Loss and

Frequency
Allergy
Hives Asthma / Hay Fever High Blood Pressure Arthritis Anemia Cancer Gallbladder Disease Mononucleosis Kidney Disease Thyroid Disease Gout Venereal Disease

Attachment Loss

Healthy
40 28 5 5 1 79

Low
174 157 116 94

Moderate
60 77 112 113 112 474
mean

High
9 24 31 37 52 153

Severe
6 21 28 26 42 123

Total 289 307 292 275 263

25-34 35-44 45-54 55-64 65-74 Total

56 597

1,426
<

mm; low is defined as having overall mean probing attachment loss > 1 mm but < 2 mm; moderate is defined as having overall mean probing attachment loss > 2 mm but < 3 mm; high is defined as having overall mean probing attachment loss > 3 mm but < 4 mm; severe is defined as

Healthy

is defined

as

having

overall

probing

attachment loss

having overall

mean

probing

attachment loss

>

mm.

diseases in the individuals with more severe AL, Spearman correlation coefficient 0.10 (P <0.001). Twenty-five percent (25.3%) of the sample reported a history of allergies, 20.6% reported a history of hives, 18.4% asthma/hay fever and 16.8% reported hypertension. Diabetes was reported by 4.8% of our sample in nearly equal proportion by males (2.6%) and females (2.2%). Of the subjects with a history of diabetes, 27.0% reported therapy with insulin.
The relationship between mean packyears and the various levels of AL was examined using the Spearman rank correlation coefficient (Fig. 2). There were no periodontally healthy subjects who were, or who had been, heavy smokers. In addition, although for each level of AL different packyear levels were observed, there was a trend for increasing severity of AL with increasing packyears (Spearman 0.378, P <0.001). The healthy, low, moderate, high, and severe AL groups had 4.3, 7.2, 15.3, 24.9 and 28.3 mean packyears, respectively (Fig. 2). The correlation between age and packyears was also statistically significant (Spearman 0.135; P 0.001).
=

Hepatitis
Diabetes

Angina
Cataracts Heart Attack Malaria Heart Murmur Inflammatory Rheumatism Stroke Arteriosclerosis Liver Disease

Cigarette Smoking

Osteoporosis
Tuberculosis

Emphysema
Figure 1. Frequency distribution of systemic diseases. Values are reported as total number of positive cases for each of the individual diseases. Diseases shown below dotted line were not included in the logistic model.

Microbial Flora The percent of subjects positive for each of the "target" microorganisms in the different categories of AL is shown in Table 4. A patient was considered positive for a given species if they exhibited > 1% of the organism in any of the two subgingival plaque samples. The only two micro-

organisms showing increasing percentages of positive subjects from healthy to severe AL groups were B. forsythus and P. gingivalis. The differences in the proportions of positive subjects were statistically significant using a Chisquare test (P < 0.001). None of the other test organisms showed a significant difference in the proportion of positive individuals in the various categories of AL.

264
40

RISK ASSESSMENT FOR PERIODONTAL DISEASE. I. ATTACHMENT LOSS

J Periodontol March 1994

LU +1
CO
30

S>
Q.

20

CO

10

Liii
Healthy
Low

Table 6. Ordinal Stepwise Regression Analysis of Severity of Attachment Loss and Systemic Diseases, Adjusted for Age (Decade) (Systemic disease only model)

Variable

Allergy

Estimated Coefficient -0.55

-0.36 0.55 0.66 0.72 1.52 1.53 2.21

Standard Error 0.12 0.18 0.30 0.17 0.28 0.17 0.18 0.18

Estimated Odds Ratio 0.58 0.70 1.73 1.93 2.05 4.57 4.62 9.12

95% CI.

Anemia

Angina Age (35-44)

Diabetes

Moderate

High

Severe

Age (45-54) Age (55-64) Age (65-74)

(0.46- 0.73) (0.49- 0.99) (0.96- 3.12) (1.39- 2.70) (1.19- 3.56) (3.28- 6.38) (3.25- 6.57) (6.41-12.97)

Level of

Probing Attachment Loss

*The 95% confidence interval of the estimated odds ratio.

Figure 2. Cigarette smoking by level of probing attachment loss. Mean number of packyears SE associated with no attachment loss (healthy) and different categories of AL (low, moderate, high, and severe). There is a positive relation between the number of packyears smoked and increasing amount of AL.
The relationship between exposure to occupational hazards and levels of AL is indicated in Table 5. A total of 543 subjects (38.1%) of the entire study population reported that they were positive for exposure to at least one of the hazards that were investigated. The most common exposure was chemical, with 18.4% of the population positive, followed by asbestos exposure with 10.2% of positive subjects. The least number of positive subjects was seen in the healthy group, with various numbers of positive subjects represented in all four levels of AL. The mean AL was significantly greater for the individuals exposed to chemicals and asbestos compared to the non-exposed, using the SpearTable 4. Percent of Subjects Positive for by Levels of Attachment Loss

rank correlation test. The corresponding P values were 0.031 and 0.030 (data not shown).
man

Occupational Hazards

Ordinal Stepwise Logistic Regression The degree of association between mean AL and the

ex-

"Target" Microorganisms

Attachment Loss

Microorganism
A. actinomycetemcomitans B. forsythus C. rectus
E. saburreum F. nucleatum P. gingivalis P. intermedia

Healthy
3.9 18.2 2.6
15.6 5.2 7.8 5.2 36.4

Low

Moderate
4.3 45.6 4.8 16.3 8.4 12.0 19.9 42.3

High
4.0 56.8 4.8 17.5 9.5 19.1 26.2 44.4

Severe 4.7 63.2* 1.9 10.4 3.7 12.3 29.3* 34.0

2.3 26.2 3.5

15.4 10.1 10.1 8.2 34.9

investigated using the ordinal logistic regression model. For this analysis 14 systemic diseases whose observed prevalence was not sufficient to support approximately normal regression estimates were eliminated from the analysis. Age was controlled in the model since it is likely an important biologic factor and exposure to chemicals and asbestos were included in the model since preliminary association analysis revealed significant positive associations between these two factors and AL. The results of the ordinal logistic regression are shown in Tables 6 and 7. When only systemic diseases were correlated with AL, diabetes and angina appear as the only systemic diseases positively correlated with AL with an estimated odds ratio of 2.05 and 1.73 (95% CIs: [1.193.56] and [0.96-3.12], respectively). Allergy and anemia were negatively correlated with AL; i.e., if either of these two diseases were present, AL tended to be less severe.
Table 7. Ordinal Stepwise Regression Analysis of Severity of Attachment Loss and Potential Risk Indicators, Adjusted for Age (Decade) (model with SES, microflora, and occupational hazards)

planatory variables, age, smoking, systemic diseases, occupational hazards, and subgingival microorganisms was

Capnocytophaga spp.

Variable

*P

< 0.001.

Anemia Education

Capnocytophaga

spp.

Table S. Distribution of Subjects Occupational Hazards

by Levels of Attachment

Loss and

Male P. gingivalis

Allergy

Attachment Loss

Hazard Chemicals Asbestos Radiation Others Total

Healthy
12 4 2 4
22

Low 60 29 18 205

Moderate
41 36

High
38 24

Severe

Total

25
17

9
6

24 190

77

4 3 49

(%)* 262 (18.4) 146 (10.2) 80 (5.6) 55 (3.9) 543 (38.1)

Diabetes B. forsythus Moderate smoker

Age (35-44) Light smoker

Age (45-54) Age (55-64) Heavy smoker Age (65-74)

Estimated Standard Estimated Coefficient Error Odds Ratio 95% C.I.* 0.17 0.60 -0.51 (0.43- 0.84) 0.22 0.65 -0.44 (0.42- 0.99) 0.13 0.65 -0.42 (0.51- 0.85) 0.77 -0.27 0.14 (0.58- 1.00) 0.13 0.31 1.36 (1.06- 1.76) 0.18 0.46 1.59 (1.11- 2.25) 0.19 0.54 1.72 (1.18- 2.49) 0.17 0.72 2.05 (1.47- 2.87) 0.35 0.84 2.32 (1.17- 4.60) 0.14 0.90 2.45 (1.87- 3.24) 1.02 0.19 2.77 (1.91- 4.02) 1.10 0.21 3.01 (1.99- 4.53) 1.42 0.21 4.14 (2.74- 6.24) 1.56 0.19 4.75 (3.28- 6.91) 2.20 0.22 9.01 (5.86-13.89)

'Relative percentage of the total 1,426

subjects.

*The 95% confidence interval of the estimated odds ratio.

Volume 65 Number 3

GROSSI, ZAMBON, HO, KOCH, DUNFORD, MACHTEI, NORDERYD, GENCO

265

for the 65 to 74 year olds. The odds ratio for attachment loss increased with the amount of smoking. The very light or occasional smoker did not show increased risk compared to non-smokers. However, the light smoker; i.e., an individual who had smoked less than or equal to 15 packyears had an odds ratio of 2.05, (95% CI: 1.47-2.87), a level comparable to the diabetes risk. The heavy smoker; i.e., the individual who had smoked more than 30 packyears, had an odds ratio of 4.75 (95% CI: 3.28-6.91), more than double that of the light smoker. Of the 8 microorganisms examined, Capnocytophaga spp. was negatively associated with AL, odds ratio 0.60 (95% CI: 0.43-0.84), whereas P. gingivalis and B. forsythus were positively associated, with odds ratios of 1.59 (95% CI: 1.11-2.25) and 2.45 (95% CI: 1.87-3.24) respectively. DISCUSSION Our sample's reported prevalences of 16.8% for hypertension and 4.8% for diabetes are slightly lower than their respective nationwide prevalences.10'11 The 26.1% prevalence of smoking, on the other hand, is comparable to the 28.8% national U.S. prevalence for smokers and reflects the national trend in smoking cessation.12 Former smokers represented 32.3% of the total. Non-smokers represented 32.4% of the total and 9.2% of our sample did not answer questions on smoking. A history of smoking was highly correlated with severity of AL (Fig. 2) (Spearman correlation 0.378, P < 0.001) and, thus, our data support previous reports indicating a strong association between smoking and periodontal disease.3,13,14 We also find a dosedependent response; i.e., the more tobacco an individual uses and the longer it is used, the greater the severity of attachment loss. It is important to note that the strong association between smoking and more severe periodontal disease remains valid after controlling for age, sex, SES, education, income, and oral hygiene status, expressed in terms of supragingival plaque accumulation and calculus
scores.

subjects with a history of diabetes, the odds for more AL was twice that of subjects who did not have diabetes (odds ratio 2.05 [95% CI: 1.19-3.56]). When the systemic diseases were examined in conjunction with socioeconomic factors, occupational hazards, and microbial flora, allergy and anemia continued to be negatively correlated. However, angina failed to be an important variable in the model, while diabetes showed an even higher odds ratio, 2.32 (95% CI: 1.17-4.60). College level of education was inversely correlated with AL, and being male represented an odds ratio of 1.36 (95% CI: 1.06-1.76) for AL as opposed to being female. Asbestos exposure was not an important variable in the model. Age expressed in decades showed an odds ratio of 1.72 (95% CI: 1.18-2.49) for the 35 to 44 year olds increasing to 9.01 (95% CI: 5.86-13.89)
For
severe

The odds for

more severe

compared to non-smokers ranged from 2.05 (95% CI:

attachment loss in smokers 1.47-

Reduced oxygen levels are associated with a decrease in PMN mobility and an increase in the proportion of anaerobic bacteria in dental plaque.17 Tobacco contains cytotoxic substances, such as nicotine and cotinine (its major metabolite), which are detected in saliva and crevicular fluid,18 serum and urine,19 and root surfaces of periodontally involved teeth of smokers.20 The presence of cotinine on root surfaces may impair wound healing and alter the host response in periodontal disease. Fibroblast attachment to root surfaces in vitro is altered by the presence of nicotine. Smoking may also have a negative effect on the cellular turn over and repair of the periodontium. Asbestos exposure appeared highly associated with attachment loss in preliminary unadjusted regression analysis (odds ratio 9.2, data not shown). However, this association was explained to a great extent by smoking since when the model was corrected for different levels of smoking, asbestos exposure was no longer an important explanatory variable. Conversely, smoking did not explain the association between AL and the other variables which showed significance in the model; i.e., diabetes, P. gingivalis, and B. forsythus, since they retained their significance when the model was corrected for different levels of smoking. Diabetes mellitus was the only systemic disease studied which showed a positive association with attachment loss. This association was valid even after corrections were made for age, smoking, socio-economic factors, plaque, and calculus. Diabetic individuals were twice as likely to exhibit attachment loss than non-diabetics (odds ratio 2.32 [95% CI: 1.17-4.60]). A similar relative risk of 2.8 for Periodontitis was reported in Pirna Indians, a population exhibiting the highest reported prevalence of non-insulin dependent (type 2) diabetes mellitus.21 Diabetic status, age, and the presence of subgingival calculus were significantly associated with the prevalence and increased severity of peri-

2.5 and 3.3 for smokers versus non-smokers were reported by two independent case-control studies.14,15 Ismail et al.3 reported a significant association between smoking and poorer levels of periodontal status. No statistical differences were, however, reported between non-smokers and former smokers. The strong association between the amount of smoking and the severity of attachment loss could be explained by a number of biologic phenomena. Cigarette smoking, nicotine, and its byproducts have a vasoconstrictive effect, not only on peripheral circulation, but on coronary, placental, and gingival blood vessels as well.16 In addition, smoking may reduce the functional activity of leukocytes and macrophages in saliva and crevicular fluid, as well as decreasing Chemotaxis and phagocytosis of blood and tissue polymorphonuclear (PMN) leukocytes, thereby likely depressing phagocyte-mediated protective responses to periodontal pathogens.17 Tobacco smoking also reduces the shortterm oxidation-reduction potentials (Eh) in dental plaque.

2.87) for light smokers to 4.75 (95% CI: 3.28-6.91) for heavy smokers (Table 7). Odds for periodontal disease of

266

RISK ASSESSMENT FOR PERIODONTAL DISEASE. I. ATTACHMENT LOSS

J Periodontol March 1994

examined, subjects with diabetes had a higher of Periodontitis defined by either parameter.23 Other studies have also reported higher prevalence and severity of periodontal disease in individuals with diabetes.2427 However, no significant differences were observed in the number of sites with PD >3 mm and radiographic bone loss in a case-control study of individuals 19 to 25 years old with and without diabetes.28 This finding could be explained partly due to the young age of the study population. In the same study,28 diabetes severity, expressed as the presence of retinopathy and nephropathy, was associated with higher gingival inflammation, compared to uncomplicated diabetes. Allergies were negatively associated with attachment loss (relative risk 0.77 [95% CI: 0.58-1.00]). When allergies were present, attachment loss was less likely to be seen. La Vecchia et al.29 in a large case control study examining the relationship between systemic diseases and the risk for colorectal cancer, reported a significant risk for cholelithiasis (1.5 [1.1-2.1]) and diabetes (1.6 [1.1-2.3]) with protection derived from a history of allergies (0.6 [0.4-0.9]). On the contrary, a history of allergies was not protective but rather a risk factor (1.40 [1.10-1.77]) for lymphaticage groups

population.21-22 In another study on the population,23 two independent measures of periodontal destruction, attachment loss and radiographic bone loss, were utilized to determine periodontal status. For all
odontitis in this
same

Pirna

prevalence

tamines almost on a routine basis. In fact, our sample of allergic individuals reported taking antihistamines at three times higher levels than the non-allergic individuals. The chronic use of antihistamines could have a non-specific modulating effect on inflammation and host response to periodontal pathogens. The release of vasoactive amines, histamine and bradykinin as well as degranulation of mast cells could be chronically depressed by the constant use of antihistamines. This suppression could, in turn, ameliorate or reduce the inflammatory cascade in periodontal tissues, with less local tissue destruction or collagenolytic activity. These interpretations regarding the role of medications in allergic individuals should be regarded as a working hypotheses worthy of further consideration to explain the apparent protective effects of a history of allergies against periodontal disease. The second most frequent medication among the allergic group was systemic antibiotics (data not shown), which could suppress periodontal pathogens in the oral flora. Of the 8 "target" organisms examined, only P. gingivalis and B. forsythus were important explanatory variables in the logistic model. Their relative risks were 1.59 and 2.45 respectively. P. gingivalis was positively correlated

malignancies.30 The negative association between periodontal attachment loss and allergies has not been reported previously. One possible biological interpretation of this finding is that individuals suffering from allergies chronically use antihis-

hematopoietic (leukemia, lymphoma, myeloma)

Previous population-based studies have shown both a positive correlation between plaque and periodontal disease7 as well as no correlation.21 In the Pirna Indian study, when supragingival plaque was examined in the regression model together with other potential risk factors, the relative risk for plaque appeared lower compared to the other factors.21 The fact that the relative risks of the above mentioned risks indicators were not affected by the inclusion of these dental variables could be explained in part by these dental variables being intimately associated with, rather than explanatory for, periodontal disease. In a review assessing the role of calculus in periodontal disease, Mandel and Gaffar35 indicate that thorough scaling and root planing are required for successful maintenance of periodontal health which supports the view that calculus contributes to the chronicity and progression of the disease, rather than being responsible for its initiation. Calculus may also form more readily when

0.12-0.33]).

with AL in adults aged 65 and over (2.5 [95% CI: 2.033.02]).31 The presence of P. gingivalis and/or B. forsythus in subgingival plaque was significantly associated with disease severity in adults 55 and older.32 The validation by several independent studies31-32 of significant associations between these two microorganisms and periodontal disease in adults points to their usefulness as markers for diagnostic tests and as markers for intervention studies. Age was the single factor most associated with severity of attachment loss. The relative risks for the different age categories increased from 1.72 (95% CI: 1.18-2.49) at age 35 to 44 to a risk of 9.01 (95% CI: 5.86-13.89) at age 65 to 74. Numerous studies have reported higher prevalence and severity of periodontal disease with increasing age.7-33-34 More recently, a study of older adults aged 50 years and over8 identified age as a risk indicator for periodontal disease. In our study population when all other confounding factors were controlled in the logistic model, age emerged as the single variable most strongly associated with attachment loss. The mechanisms of how age is associated with increased severity of periodontal disease have not yet been explained. The combined odds for periodontal destruction for an individual exhibiting all the identified indicators, namely age older than 45, diabetes, moderate or heavy smoking, and presence of P. gingivalis or B. forsythus are 30 times higher than for an individual lacking such indicators. When local factors responsible for periodontal disease, namely supragingival plaque and subgingival calculus, were adjusted in the logistic model, the resulting odds ratios and confidence intervals for the risk indicators, age, diabetes and smoking remained virtually unaltered. The odds ratios were 9.12 (95% CI: 6.16-13.49), 2.01 (95% CI: 1.05-3.85), and 3.94 (95% CI: 2.77-5.60) respectively (data not shown). Subgingival calculus appeared significantly associated with attachment loss, with odds ratio of 7.10 (95% CI: 5.619.98). Supragingival plaque was inversely associated with severity of attachment loss (relative risk 0.20 [95% CI:

Volume 65 Number 3

GROSSI, ZAMBON, HO, KOCH, DUNFORD, MACHTEI, NORDERYD, GENCO

267

periodontal disease has occurred, and hence it may be a secondary phenomenon. This study finds that age, history of diabetes mellitus, smoking, and subgingival infection with P. gingivalis and B. forsythus are significant risk indicators for attachment loss in the adult population. The association between these indicators and attachment loss remained significant after controlling for confounding factors in the logistic model such as gender, race, education, income, socio-economic status, supragingival plaque, and subgingival calculus. These indicators are therefore worthy of further examination in longitudinal and intervention studies to determine if they
are true

risk factors.

Acknowledgments
The authors wish to thank Drs. Sebastiano Andreana, Lillian Carpio, Lars A. Christersson, Yoly M. Gonzalez, and Hans Preus who participated as additional examiners in this study and Dr. Michael Cummings for valuable discussions regarding self-reporting smoking. The expert technical assistance of Mrs. Frigg Levine in preparing the manuscript is appreciated. Supported in part by USPHS grants DE04898, DE07034, and DE00158.
REFERENCES 1. Beck JD, Lainson PA, Field HM, Hawkins BF. Risk factors for
various levels of periodontal disease and treatment needs of Iowa. Community Dent Oral Epidemiol 1984; 12:17-22. 2. Miller AJ, Brunelle JA, Carlos JP, Brown LJ, Le H. Oral health of United States adults. Bethesda, Maryland: National Institute of Dental Research, 1987; NIH publication no. 87-2868. 3. Ismail AI, Burl BA, Eklund SA. Epidemiologie patterns of smoking and periodontal disease in the United States. J Am Dent Assoc 1983; 106:617-623. 4. Machtei EE, Christersson LA, Grossi SG, Dunford R, Zambon JJ, Genco RJ. Clinical criteria for the definition of "established Periodontitis." JPeriodontol 1992; 63:207-215. 5. Zambon JJ, Reynolds HS, Chen P, Genco RJ. Rapid identification of periodontal pathogens in subgingival dental plaque: Comparison of indirect immunofluorescence microscopy with bacterial culture for detection of Bacteroidesgingivalis. J Petiodontol 1985; 59(Suppl.):3240. 6. Bonta CY, Zambon JJ, Genco RJ, Neiders ME. Rapid identification of periodontal pathogens in subgingival dental plaque: Comparison of indirect immunofluorescence microscopy with bacterial culture for detection of Actinobacillus actinomycetemcomitans. J Dent Res 1985; 64:793-798. 7. Le H, nerud , Boysen H, Smith M. The natural history of periodontal disease in man. The rate of periodontal destruction before 40 years of age. / Petiodontol 1978; 49:607-620. 8. Locker D, Leake JL. Risk indicators and risk markers for periodontal disease experience in older adults living independently in Ontario, Canada. J Dent Res 1993; 72:9-17. 9. Koch GG, Edwards S. Logistic Regression. In: Encyclopedia of Statistical Sciences, Vol 5. New York: John Wiley & Sons; 1985; 128133. 10. Roberts J, Rowland M. Hypertension in adults 25-74 years of age, United States, 1971-1975. Vital Health Stat 1981; 11:11-17. 11. Hadden WC, Harris MI. Prevalence of diagnosed diabetes, undiagnosed diabetes and impaired glucose tolerance in adults 20-74 years of age, United States, 1976-1980. Vital Health Stat 1987; 11:12-18.

1988; 164:258-260. 18. McGuire JR, McQuade MJ, Rossmann JA, et al. Cotinine in saliva and gingival crevicular fluid of smokers with periodontal disease. J Petiodontol 1989; 60:176-181. 19. Langone JJ, Gjika HB, Van Vunakis H. Nicotine and its metabolites. Radioimmunoassay for nicotine and cotinine. Biochem 1973; 12:50255030. 20. Cuff MJ, McQuade MJ, Scheidt MJ, Sutherland DE, Van Dyke TE. The presence of nicotine on root surfaces of periodontally diseased teeth in smokers. / Petiodontol 1989; 60:564-569. 21. Emrich LJ, Shlossman M, Genco RJ. Periodontal disease in noninsulin dependent diabetes mellitus. JPeriodontol 1991; 62:123-130. 22. Nelson RG, Shlossman M, Budding LM, et al. Periodontal disease and non-insulin-dependent diabetes mellitus in Pirna Indians. Diabetes Care 1990; 13:836-840. 23. Shlossman M, Knowler WC, Pettitt DJ, Genco RJ. Type 2 diabetes mellitus and periodontal disease. J Am Dent Assoc 1990; 121:532-536. 24. Bacic M, Plancak D, Granic M. CPITN assessment of periodontal disease in diabetic patients. J Petiodontol 1988; 59: 816-822. 25. Belting CM, Hiniker JJ, Dummett CO. Influence of diabetes mellitus on the severity of periodontal disease. J Petiodontol 1964; 35:476480. 26. Cianciola LI, Park BH, Bruck E, Mosovich L, Genco RJ. Prevalence of periodontal disease in insulin-dependent diabetes mellitus (juvenile diabetes). J Am Dent Assoc 1982; 104:653-660. 27. Glavind L, Lund B, Le H. The relationship between periodontal state and diabetes duration, insulin dosage and retinal changes. / Petiodontol 1968; 39:341-347. 28. Rylander H, Ramberg P, Blohme G, Lindhe J. Prevalence of periodontal disease in young diabetics. J Clin Petiodontol 1987; 14:38-43. 29. La Vecchia C, Davanzo B, Negri E, Franceschi S. History of selected diseases and the risk of colorectal cancer. EurJ Cancer 1991; 27:582586. 30. McWhorter WP. Allergy and risk of cancer. A prospective study using NHANES I follow up data. Cancer 1988; 62:451-455. 31. Beck JD, Koch GG, Zambon JJ, Genco RJ, Tudor GE. Evaluation of oral bacteria as risk indicators for Periodontitis in older adults. / Petiodontol 1992; 63:93-99. 32. Grossi SG, Dunford RG, Reynolds HS, Zambon JJ, Christersson LA. Clinical and microbiological parameters in the diagnosis of severe periodontal disease in older adults. J Dent Res 1990; 69(Spec. Is33. Abdellatif HM, Burt BA. An epidemiological investigation into the relative importance of age and oral hygiene status as determinants of Periodontitis. J Dent Res 1987; 66:13-18. 34. Beck JD, Koch GG, Rozier RG, Tudor GE. Prevalence and risk indicators for periodontal attachment loss in a population of older community-dwelling blacks and whites. J Petiodontol 1990; 61:521528. 35. Mandel ID, Gaffar A. Calculus revisited. A review. / Clin Petiodontol 1986; 13:249-257.
Send reprint requests to: Dr. Sara G. Grossi, Periodontal Disease Research Center, Department of Oral Biology, 3435 Main Street, Buffalo, NY 14214. Accepted for publication September 15, 1993.

12. Centers for Disease Control. Tobacco use by adults United States, 1987. Morbidity and Mortality Weekly Report 1989; 38:685-687. 13. Feldman RS, Alman JE, Chauncey HH. Periodontal disease indices and tobacco smoking in healthy aging men. Gerodontics 1987; 1:43-46. 14. Haber J, Kent RL. Cigarette smoking in a periodontal practice. J Petiodontol 1992; 63:100-106. 15. Bergstrm J. Cigarette smoking as risk factor in chronic periodontal disease. Community Dent Oral Epidemiol 1989; 17:245-247. 16. Baab DA, Oberg PA. The effect of cigarette smoking on gingival blood flow in humans. J Clin Petiodontol 1987; 14:418-424. 17. Palmer RM. Tobacco smoking and oral health: Review. Br Dent J
-

sue):361(Abstr. 2021).