Hujoel PP, Drangsholt M, Spiekerman C, DeRouen TA. Periodontal Disease and Coronary Heart Disease Risk. JAMA. 2000;284(11):1406-1410. doi:10.1001/jama.284.11.1406
Author Affiliation: Department of Dental Public Health Sciences, School of Dentistry (Drs Hujoel, Drangsholt, Spiekerman, and DeRouen), Department of Epidemiology, School of Public Health (Dr Hujoel), Department of Oral Medicine, School of Dentistry (Dr Drangsholt), Department of Biostatistics, School of Public Health (Dr DeRouen), and Department of Medicine, Cardiovascular Health Research Unit (Dr Spiekerman), University of Washington, Seattle.
Context Research has suggested a relationship between periodontal disease and
coronary heart disease (CHD), but data on the association between these 2
common conditions are inconclusive due to the possibility of confounding.
Objective To evaluate the risk of CHD in persons with periodontitis, gingivitis,
or no periodontal disease.
Design Prospective cohort study.
Setting The First National Health and Nutrition Examination Survey Epidemiologic
Follow-up Study, conducted in 1982-1984, 1986, 1987, and 1992.
Participants A total of 8032 dentate adults aged 25 to 74 years with no reported
history of cardiovascular disease, including 1859 individuals with periodontitis,
2421 with gingivitis, and 3752 with healthy periodontal tissues.
Main Outcome Measure First occurrence of death from CHD or hospitalization due to CHD, or
revascularization procedures, obtained from death certificates and medical
records, by baseline periodontal status.
Results During follow-up, 1265 individuals had at least 1 CHD event, including
CHD fatality (n = 468) or at least 1 hospitalization with a diagnosis of CHD
(n = 1022), including coronary revascularization procedures (n = 155). After
adjustment for known cardiovascular risk factors, gingivitis was not associated
with CHD (hazard ratio, 1.05; 95% confidence interval, 0.88-1.26), while periodontitis
was associated with a nonsignificant increased risk for CHD event (hazard
ratio, 1.14; 95% confidence interval, 0.96-1.36).
Conclusion This study did not find convincing evidence of a causal association
between periodontal disease and CHD risk.
Several infectious diseases have been implicated as possibly causing
myocardial infarction (MI).1 In related research,
periodontal disease has also been related to coronary heart disease (CHD).
Several observational studies have indicated that periodontitis, a chronic
inflammatory periodontal disease that results in the breakdown of bone that
surrounds teeth, may be associated with an increased risk for MI.2- 4 At least one cohort
study indicated that gingivitis, an inflammatory periodontal disease without
the breakdown of supporting bone, also increased the risk for fatal MI.5 Both the chronic low-level bacteremia that occurs
with brushing or chewing and the elevation of inflammatory mediators in response
to the bacterial biofilm growing on teeth6- 11
have been suggested as possible causal pathways for the increased risk of
Since periodontal disease and heart disease are common, quantifying
their association is of significant public health importance. The interpretation
of the reported associations is difficult.12
On the one hand, the associations could be interpreted as causal, which could
imply, as has been suggested, that reducing periodontal disease with interventions
may have the additional benefit of reducing the risk of cardiovascular disease
(CVD).13 On the other hand, these data could
be interpreted as being artifacts, that is, the result of biases caused by
confounding.14 Since periodontitis and myocardial
disease share common risk factors, such as increasing age, smoking, stress,
socioeconomic status, and body fat content, the potential for confounding
is substantial. One recent meta-analysis suggested that incomplete adjustment
for socioeconomic status may be responsible for the observed weak associations.15
The primary goal of this study was to evaluate 3 periodontal conditions
(periodontitis, gingivitis, and periodontal health [no gingivitis or periodontitis])
at baseline and the incidence of the first subsequent CHD event observed within
the First National Health and Nutrition Examination Survey (NHANES I) Epidemiologic
The design and sampling of NHANES I (1971-1975) and the epidemiologic
follow-up study as it relates to dental and cardiovascular studies have been
reported.2 Briefly, a US population-based probability
sample of civilian noninstitutionalized individuals was obtained. Low-income
groups, women of childbearing age, and elderly persons were oversampled. The
NHANES I Epidemiologic Follow-up Study is a prospective study of the NHANES
I participants who were aged 25 to 74 years at baseline: 8032 dentate individuals
who had both a medical and dental examination, reported no history of CVD,
and had 4 completed longitudinal follow-ups: 1982 to 1984, 1986 (only those
individuals who were 55-74 years at baseline), 1987, and 1992.
The baseline information included a medical examination, a standardized
medical history and dental examination, laboratory tests, and a single 24-hour
dietary recall. Demographic variables in the risk models of each examined
person included age at baseline, sex, race (white, African American, and other),
education, poverty index (defined in Table
1), and marital state (ever married vs never married). Cardiovascular
risk factors evaluated at the baseline clinical examination included systolic
and diastolic blood pressure, serum cholesterol level, history of diabetes
mellitus, physical activity (individuals were defined as physically active
if they reported being either very active in their usual day, aside from recreation,
or if they reported much exercise for recreation), height and weight, alcohol
consumption (glasses per day), smoking duration (years), the average number
of cigarettes smoked per day during the smoking years, and a history of a
nervous breakdown. For smoking, information regarding the duration of smoking
and the average number of cigarettes smoked prior to the baseline examination
(1971-1975) was derived from 16 questions asked during the interview in 1982
to 1984. Validation studies have indicated that surrogate response and self-response
on cigarette smoking obtained approximately 10 years after the baseline interview
(1971-1975) were not remarkably different from the follow-up interview (1982-1984).16,17
A CHD event was defined as 1 of the following outcomes: (1) death with
an underlying cause of death coded 410-414 using the International
Classification of Diseases, Ninth Revision (ICD-9); (2) a hospital
stay with a discharge diagnosis code 410-414 using the ICD-9CM; or (3) either of the following coronary revascularization
procedures: 36.10-36.19 (coronary revascularization) or 36.00-36.09 (removal
of coronary obstruction). The first occurrence of any of these 3 events (fatalities,
hospitalization because of CHD, or hospitalization because of revascularization)
was used as the defining event.
Three mutually exclusive periodontal classifications were defined based
on the Russell Periodontal Index18: periodontitis
(n = 1859), gingivitis (n = 2421), and periodontal health (Periodontal Index
<.05; n = 3752). Individuals with periodontitis had a periodontal pocket
with attachment loss (ie, not merely a deepened gingival crevice due to swelling
in the free gingiva). Individuals with gingivitis had an overt area of inflammation,
which may have completely circumscribed the tooth and which may have been
associated with pseudopockets. The following signs of overt inflammation were
separately assessed during the dental examination: bleeding gums, diffuse
marginal inflammation, and swollen red papillae. Individuals with either periodontitis
or gingivitis were subdivided into groups with and without any of these 3
overt signs of clinical inflammation. Based on the Russell Periodontal Index,
4 levels of periodontal disease severity were defined, and dose-response relationships
Cox proportional hazard models were fitted to assess whether individuals
with periodontitis or gingivitis at baseline were at higher risk for a CHD
event than individuals with no signs of periodontal disease at baseline. Time-on-study
was used as the time scale for all time-to-event analyses. Potentially confounding
variables were included in the model using a forward elimination process.19 With this approach, potentially confounding variables
representing competing hypotheses were added to the model. Since the use of
ratios such as body mass index (calculated as weight in kilograms divided
by the square of height in meters) and pack-years can induce spurious correlations,20 factors and interaction terms rather than ratios
were modeled. Three different approaches for taking into account the sampling
design were evaluated: a model-based analysis assuming the sample was a simple
random sample,21 a design-based analysis taking
into account the stratification and clustering but ignoring the sampling weights,22 and a design-based analysis incorporating the clustering,
stratification, and the sampling weights.23
Which analysis is appropriate is a subtle question22
that depends on a trade-off between efficiency and bias. The primary results
reported in the tables take into account the sampling design but not the sampling
weights. This approach was selected because the sampling weights are primarily
determined by design variables, such as age, race, poverty census enumeration
district, and family income.21 Since these
design variables were partially captured by the socioeconomic variables included
in the statistical models, not using the sampling weights in the analyses
provides a good compromise between bias and efficiency.22
Since the associations between periodontitis and CHD were small and sensitive
to the analytic approach selected, all 3 approaches were presented for the
key results so that the robustness of the conclusions could be evaluated.
Analyses that adjusted for the sampling design and/or weights were performed
using SUDAAN software.24,25 Individuals
with evidence of prior CVD (a report of a prior MI, stroke, heart failure,
or use of medication for a weak heart) were excluded from the primary analyses.
Baseline differences were assessed using analysis of variance models for continuous
variables and logistic regression models for binary variables. Post hoc power
estimates were computed based on the normal distributions of the regression
coefficient estimates as 1 − β = Φ (1.96 + |log(hazard ratio
Periodontal status was significantly associated with demographic factors,
lifestyle characteristics, and medical conditions (Table 1). When compared with individuals with a healthy periodontium,
individuals with periodontitis and gingivitis were significantly more likely
to be male, less educated, African American, and poorer (P<.005). Individuals with periodontitis were also older and significantly
different from individuals with a healthy periodontium (P<.005) with respect to most cardiovascular risk factors. These
individuals also were more likely to have diabetes mellitus, be overweight,
consume more alcohol and cigarettes, have higher systolic and diastolic blood
pressure, have higher serum cholesterol levels, and have had a nervous breakdown
(P<.005 for all comparisons). Individuals with
gingivitis were similar to individuals with a healthy periodontium with respect
to many cardiovascular risk factors with the exception of body mass index,
weight, alcohol consumption, and pack-years of smoking (P<.005 for all comparisons).
During the follow-up, 1265 individuals had at least 1 CHD event: CHD
fatality (n = 468), hospitalization for CHD (n = 1022), or hospitalization
for coronary revascularization procedure (n = 155). Unadjusted for any potentially
confounding variables and excluding individuals with evidence of prior CVD,
individuals with periodontitis had an HR of 2.66 (95% confidence interval
[CI], 2.34-3.03) for a CHD event when compared with those individuals with
a healthy periodontium (Table 2).
After adjustment for confounders and sampling design, the HR of CHD for individuals
with periodontitis decreased to 1.14 (95% CI, 0.96-1.36). Adjustment for the
sampling weights increased the HR by 6% to a value of 1.21 (95% CI, 0.98-1.50).
Various dose-response relationships and subgroup analyses were explored
(Table 3). When the analyses were
restricted to individuals with periodontitis, no dose-response relationships
were found between the number of teeth at baseline and CHD risk. Among individuals
with periodontitis, the presence of either swollen papillae, diffuse marginal
inflammation, or bleeding gums did not elevate the risk for CHD (HR, 0.98;
95% CI, 0.74-1.29). No obvious dose-response relationships were detected between
the severity of the periodontitis and CHD risk (Table 3). Individuals within the upper level of periodontal disease
severity were not at elevated risk for CHD (HR, 1.28; 95% CI, 0.87-1.88).
When the analyses were limited to nonsmoking individuals at baseline,
the HR was attenuated (HR, 1.06; 95% CI, 0.84-1.34). When the analyses were
limited to individuals younger than 50 years, the HR associated with periodontitis
was 1.36 (95% CI, 0.97-1.92). When the analyses were stratified by 10-year
age groups, no simple age-related patterns became apparent. Stratifying by
measures of socioeconomic class did not lead to any substantial changes in
the HR estimates. Individuals within the lowest quartile of the poverty index
had an HR of 1.09 (95% CI, 0.80-1.48) associated with periodontitis, while
individuals within the upper quartile of the poverty index had an HR of 1.12
(95% CI, 0.82-1.53) associated with periodontitis.
When only fatalities were evaluated, periodontitis was associated with
an HR for cardiovascular mortality of 1.20 (95% CI, 0.90-1.61) (Table 2). For cardiovascular mortality, the sampling weights increased
the HR by 21.7% to 1.46 (95% CI, 0.94-2.27).
Unadjusted for confounding variables, gingivitis was associated with
a 20% increased risk for CHD events when compared with a healthy periodontium
(HR, 1.20; 95% CI, 1.05-1.39) (Table 2).
After adjustment for potentially confounding variables and the sampling design,
gingivitis was not associated with an increased risk of CHD (HR, 1.05; 95%
CI, 0.88-1.26). Inclusion of the sampling weights led to a slightly protective
association (HR, 0.88; 95% CI, 0.69-1.11). The presence of overt signs of
clinical inflammation in combination with gingivitis did not increase CHD
risk (HR, 0.99; 95% CI, 0.75-1.31) (Table
3). There was no association between gingivitis and fatal CHD disease
(HR, 1.17; 95% CI, 0.84-1.61) (Table 2).
Separate analyses were performed to evaluate whether signs of periodontal
inflammation, regardless of the diagnosis of gingivitis or periodontitis,
were associated with CHD risk. After adjustment for confounding variables
and sampling design, there was no association between CHD and the presence
of red swollen papillae (HR, 1.06; 95% CI, 0.87-1.28) and between CHD and
the presence of diffuse marginal inflammation (HR, 1.05; 95% CI, 0.88-1.25).
Bleeding gums were associated with an nonsignificantly increased CHD risk
(HR, 1.24; 95% CI, 0.77-2.00).
Post hoc power calculations indicated that the statistical model used
to relate CHD events (fatal MI, nonfatal MI, and revascularization procedures)
had greater than 90% power to detect a 30% increased CHD risk associated with
periodontitis or gingivitis (type I error rate, 5%).
The results of this study do not provide convincing evidence that periodontitis
and gingivitis are associated with CHD. Gingivitis was not associated with
CHD. Periodontitis was associated with a nonsignificant increased risk for
CHD. Markers of periodontal inflammation associated with either periodontitis
or gingivitis, such as swollen red papillae, bleeding gums, or diffuse marginal
inflammation, were not associated with an increased risk for CHD. No obvious
dose-response relationships were present between the severity of periodontitis
and CHD risk. The findings of this study do not provide the kind of consistent
evidence needed to support the hypothesis of a causal relationship between
periodontal disease and CHD.
Gingivitis was not associated with CHD. Subgroup analyses by gingivitis
severity further corroborated this conclusion with more severe gingivitis
not being associated with any increased risk for CHD. This finding is important
since an estimated 50% of the US adult population has gingivitis.26 Even a small elevated risk for CHD associated with
gingivitis would result in a substantial attributable proportion. This finding
is in contrast to the findings based on the Nutrition Canada Survey in which
gingivitis was associated with an increased fatal MI risk.5
Slightly positive associations between periodontitis and CHD events
were identified. The relevance, if any, of these associations is uncertain.
No clear statistical significance or dose-response trends were observed. Our
study is currently one of the largest investigations on the association between
periodontal disease and CHD. With 1265 events, long follow-up, extensive information
on potentially confounding variables, and sufficient statistical power (>90%)
to detect small associations (HR, 1.3), this study had a good chance to reliably
identify small associations. Not observing clearly significant associations
despite the extensive data manipulations in the search of significance suggests
that the true association between periodontitis and CHD is either absent or
so small that even larger studies are required to identify them.
Various subgroup analyses and dose-response relationships between periodontitis
and CHD were explored. Coronary heart disease risk did not increase substantially
with the severity of periodontitis. No relationship was identified between
the number of teeth in patients with periodontitis and CHD risk. There was
no association between markers of clinical inflammation and CHD, which is
inconsistent with the current hypothesized causal role of periodontal inflammation6 in CHD.
When sampling weights were included in the analyses, periodontitis tended
to increase CHD risk. Conversely, inclusion of sampling weights led to small
protective effects associated with gingivitis. These small risk increases
and decreases, depending on the particular analytic approach used, further
strengthened the impression of no association.
A review of recent studies does not provide convincing support in favor
of a causal association.15 Two large cohort
studies in socially homogeneous populations with similar health awareness
and rigorous control for confounders reported no associations between periodontitis
and CHD.27,28 Less well-controlled
studies, studies in socially heterogeneous populations, such as this one,
without control for health awareness, or studies with few events reported
In particular, insufficient control for smoking history, socioeconomic status,
and health awareness may have induced small positive associations.
While this study did provide convincing evidence regarding the absence
of a moderate-to-large association between periodontitis and CHD, a small
causal association could not be ruled out. Several biological mechanisms through
which periodontal disease may cause CHD have been proposed as follows: the
invasion into endothelial coronary cells by oral microorganisms,32,33
the harmful cardiovascular effects of inflammatory response to periodontitis,6 or contributions of oral microorganisms to acute thromboembolic
events during bacteremia.34 If these biological
mechanisms are responsible for a slight risk increase, better controlled and
still larger studies will be required to identify them. Such efforts may be
important because of the high prevalence of periodontitis.