Table 5 Risk of bias for included cross-sectional studies*

Izuora, [46]

2019,

U.S.A.

• Consecutive sample of patients admitted to urban university hospital.

• No sample size calculation or response rate included.

Predictor: examiners calibrated for number and health of the teeth, no kappa scores reported.

Outcome:

history of stroke or heart disease defined as atherosclerotic heart disease or congestive heart failure. Responses verified through review of medical records.

Included: demographics, BMI, smoking, previous hospital admissions, DM history, duration of DM, history of CVD, prescription medication, frequency of tooth brushing, flossing, and using mouthwash.

Not included:

diet, dental care utilization, alcohol use.

p ≤ 0.05

OR (95%CI) Multivariable analysis using linear and logistic regression.

• Consecutive sample from urban university hospital.

• Small sample size.

• High prevalence of DM (34%) in study participants higher than in U.S. population overall.

Itakura, [47] 2018,

Japan

• Convenience sample from 2 urban nursing homes.

• No sample size calculation or response rate included.

Predictor: Eichner index classification of bite stability. Examiner training and calibration not included.

Outcome:

Japan Diabetes Society guidelines for diagnosis of DM.

Included: demographics, BMI, serum albumin, blood pressure, chest X-ray, electrocardiogram, dementia, activities of daily living, grip strength, repetitive swallowing, comorbidities.

Not included:

daily oral hygiene, diet, dental care utilization, duration of DM, smoking, alcohol use, prescription use for hypertension, DM, and dyslipidemia.

p < 0.05

OR (95%CI)

Multivariable logistic regression analysis.

• Convenience sample from 2 nursing homes may have included healthier participants able to consent and undergo testing.

• Small sample size.

• 9% of participants excluded due to missing data.

• Japanese study participants may be less diverse than U.S. population.

Shin, [48]

2017,

Korea

• National probability sample of non-institutionalized adults.

• Large sample size.

Predictor: examiners trained to count teeth, excluded 3rd molars, kappa scores not included.

Outcome: International Diabetes Federation Task Force on Epidemiology and Prevention guidelines for metabolic syndrome diagnosis.

Included:

demographics, SES, general health status, oral health status, daily toothbrushing frequency, periodontitis, prescription use for hypertension, DM, and dyslipidemia, smoking, alcohol use, physical activity.

Not included:

dental care utilization, diet, duration of DM, comorbidities.

p < 0.05

OR (95%CI)

Multivariable logistic regression analysis.

• National probability sample of noninstitutionalized adults.

• 49% of participants excluded due to missing data.

• Korean population may be less diverse than U.S. population.

Song, [49]

2017,

Korea

• National probability sample of non-institutionalized adults.

• Large sample size.

Predictor: examiners trained to count teeth, excluded 3rd molars, kappa scores not included.

Outcome:

DM diagnosis if fasting blood sugar level was > 126 mg/dL or the participant was currently using antidiabetic medications.

ETDRS severity scale for diabetic retinopathy diagnosis.

Included: demographics, BMI, waist circumference, blood pressure, smoking, alcohol use, exercise, HbA1c, duration of DM, prescription use of DM medication, frequency of brushing, flossing, mouthwash, interdental brush, electric brush.

Not included: periodontal disease, diet, dental care utilization.

Two-sided p < 0.05

OR (95%CI)

Multivariable logistic regression analysis.

• National probability sample of noninstitutionalized adults.

• 20% of participants excluded due to missing data.

• Korean population may be less diverse than U.S. population.

Jung,[50]

2015,

Korea

• District population-based sample of adults aged ≥ 50 years.

• Nested cross-sectional study within cohort study.

Predictor: periodontal measures reported for pocket depth, gingival recession, bleeding on probing, clinical attachment loss, number of teeth. Trained and calibrated examiners with kappa scores included for periodontal measures.

Outcome:

2010 American Diabetes Association guidelines for diabetes diagnosis.

Included: demographics,

blood lipids, blood glucose, HbA1c, BMI, blood pressure, periodontitis, medical history, lifestyle, smoking, year of survey; prescription use for hypertension, DM, and dyslipidemia.

Not included:

daily oral hygiene, diet, dental care utilization, alcohol use, duration of DM, comorbidities.

p < 0.05

OR (95%CI) Multinomial logistic regression analysis.

• District population-based sample of adults aged ≥ 50 years.

• 27% response rate.

• 1.5% of participants excluded due to missing data.

• Korean population may be less diverse than U.S. population.

Vedin,[51]

2015,

Global

• Nested cross-sectional study of baseline data from global clinical trial of participants aged ≥ 60 years with stable CHD.

• Registered with ClinicalTrials.gov.

Predictor:

number of teeth by self-report, included 3rd molars.

Outcome:

DM which requires pharmacotherapy.

Included: demographics, SES, smoking, alcohol use, periodontal disease by self-report of gum bleeding after brushing, physical activity, stress, waist circumference, blood pressure, lipid panel, blood glucose, hs-CRP, WBC, eGFR.

Not included:

daily oral hygiene, diet, prescription use of DM medication, dental care utilization, duration of DM.

p = 0.05

OR (95%CI)

Multivariable logistic regression.

• Large global sample of participants aged ≥ 60 years, who have stable CHD, from both developed and developing countries.

Furukawa,[52]

2007,

Japan

• Convenience sample from 2 urban outpatient diabetes clinics.

• No description of inclusion/exclusion criteria.

• No sample size calculation or response rate included.

Predictor:

DMFT excluding 3rd molars, pocket depth, no information provided for examiner training or calibration.

Outcome:

serum lipid panel for atherogenic factors.

Included: demographics, BMI, blood pressure, smoking, oral hygiene score, periodontal pocket depth, HbA1c, duration of DM, creatinine, urinalysis.

Not Included:

daily oral hygiene, diet, dental care utilization, alcohol use, prescription use of DM medication, comorbidities.

Two-sided p-values

Spearman correlation coefficients.

Did not report multivariable regression analysis.

• Convenience sample.

• Small sample size.

• Japanese study participants may be less diverse than U.S. population.

Huang,[53] 2013,

U.S.A.

• Nationally representative sample of participants aged ≥ 65 years.

• Large sample size.

Predictor:

number of teeth by self-report, included 3rd molars.

Outcome:

health-related quality of life by Healthy Days Core Module.

Included: demographics, smoking, dental care utilization, DM by self-report, healthcare utilization.

Not included:

daily oral hygiene, alcohol use, diet, comorbidities, duration of DM, prescription use of DM medication, periodontal disease.

p < 0.001

OR (95%CI)

Multivariable linear and logistic regression.

• Large nationally representative sample of U.S. population aged ≥ 65 years.