Community Dent Oral Epidemiol 2015; 43; 116–127 Ó 2014 John Wiley & Sons A/S.

ohn Wiley & Sons A/S. Published by John Wiley & Sons Ltd
All rights reserved

Is frequency of tooth brushing a Heiko Zimmermann1, Nils Zimmer-

mann2, Daniel Hagenfeld2, Annette
Veile1, Ti-Sun Kim2 and Heiko Becher1

risk factor for periodontitis? A 1

Unit of Epidemiology and Biostatistics,
Institute of Public Health, University of
Heidelberg, Germany, 2Section of

systematic review and meta- Periodontology, Department of Operative

Dentistry, University Hospital Heidelberg
Germany

analysis
Zimmermann H, Zimmermann N, Hagenfeld D, Veile A, Kim T-S, Becher H. Is
frequency of tooth brushing a risk factor for periodontitis? A systematic review
and meta-analysis. Community Dent Oral Epidemiol 2015; 43: 116–127. © 2014
John Wiley & Sons A/S. Published by John Wiley & Sons Ltd

Abstract – Objectives: The epidemiology of periodontitis regarding oral-

hygiene practices particularly the frequency of tooth brushing has been the
subject of relatively few dedicated studies. This paper provides a systematic
review of available relevant epidemiological studies and a meta-analysis of the
effect of tooth brushing frequency on periodontitis. To review and to quantify
the risk for periodontitis associated with frequency of tooth brushing. Methods:
Systematic literature search was conducted in nine online resources (PUBMED,
ISI and 7 additional databases). Related and cross-referencing publications
were reviewed. Papers published until end of March 2013 reporting
associations between tooth brushing frequency and periodontitis were
considered. A meta-analysis was performed to quantify this association.
Results: Fourteen studies were identified. The test of heterogeneity for cross-
sectional studies was not significant (P = 0.31). A fixed-effects model yielded a
significant overall odds ratio estimate of 1.41 (95%CI: 1.25–1.58, P < 0.0001) for Key words: epidemiologic studies;
infrequent compared to frequent tooth brushing. For all fourteen studies, there epidemiology; meta-analysis; oral hygiene;
periodontitis; tooth brushing
was a slight indication for heterogeneity (I² = 48%, P = 0.02) and the
corresponding result with a random-effects model was 1.44 (95%CI: 1.21–1.71, Heiko Zimmermann, Unit of Epidemiology
and Biostatistics, Institute of Public Health,
P < 0.0001). Conclusions: There are relatively few studies evaluating the
University of Heidelberg, Germany
association between tooth brushing frequency and periodontitis. A clear effect Tel.: +49 6221 56 4647
was observed, indicating that infrequent tooth brushing was associated with Fax: +49 6221 56 5948
severe forms of periodontal disease. Further epidemiological studies are e-mail: h.zimmermann@uni-heidelberg.de
needed to precisely estimate the effect of key risk factors for periodontitis and Submitted 14 May 2014;
their interaction effects. accepted 9 August 2014

Periodontitis is a disease mostly plaque related
and caused by inflammation that affects the tissues
around the teeth of the oral cavity (1). The severity
and progression of periodontitis is related to risk
factors such as smoking (2), alcohol (3), diabetes
mellitus (4), as well as genetic predisposition (5, 6),
gender (7), age (8) and oral hygiene (9, 10). Oral
hygiene has recently shown to be associated with
diseases like chronic obstructive pulmonary
disease (11). In addition to the patient related risk
factors, the presence of dental caries (12, 13) and
insufficient dental restorations (12, 13) also seem to
have an influence on periodontitis. While studies
have shown that brushing, particularly with fluo-
ride toothpaste, reduces dental caries (14) the same
benefit of oral hygiene has not been clearly demon-
strated for periodontitis (10, 15). Attaining optimal
oral hygiene conventionally has been regarded as
strongly related to the prevention of gingivitis. A
lack of sufficient oral hygiene leads to an

116 doi: 10.1111/cdoe.12126


established gingival lesion (16). In a susceptible
host, this lesion can progress to bone loss and peri-
odontal disease (17).
Several studies have also indicated a relationship
between oral hygiene and severe chronic periodon-
titis (SCP) (18), while others have questioned this
association and suggested a genetic or other aetiol-
ogy (19). Many of these available studies were rela-
tively small, and the effect estimates therefore had
wide confidence intervals. Therefore, the aim of
this study is to review all existing epidemiological
studies that report on the association of the fre-
quency of tooth brushing (as a proxy for oral
hygiene) and periodontitis and to quantify through
meta-analysis the association between them.
Materials and methods
This systematic review and meta-analysis complies
with PRISMA (Preferred Reporting Items for Sys-
tematic Reviews and Meta-Analyses) reporting
guidelines (see Appendices S1, S2) (20).
Search strategy
A comprehensive literature search was performed
on different databases to identify English-language
articles evaluating the association between the fre-
quency of tooth brushing and periodontitis. Publi-
cations until the end of March 2013 were included.
The PubMed platform for accessing Medline
was searched systematically. The search strategy
relied on the medical subject headings (MeSH)
within Medline to find important and appreciable
data. ‘Dental Devices, Home Care’ and ‘Oral
Hygiene’ were used as synonyms of tooth brush-
ing. This standardized approach (MeSH) was
appropriate because of the variable definitions (21)
of periodontitis. ‘Clinical Attachment Loss’ and
‘Pocket-Probing Depth’ (not MeSH) also were used
to complete the list of search strings (22). The com-
plete search procedure was a combination of MeSH
terms and also free text words. The terms included
are listed in the two following paragraphs.
‘Periodontitis’ [Mesh] AND (‘Oral Hygiene’
[Mesh] OR ‘Toothbrushing’ [Mesh] OR ‘Dental
Devices, Home Care’ [Mesh]) AND (‘Epidemiolog-
ic Studies’ [Mesh] OR ‘Cohort Studies’ [Mesh] OR
‘Case–Control Studies’ [Mesh] OR ‘Cross-Sectional
Studies’ [Mesh] OR ‘Intervention Studies’ [Mesh])
AND (‘Alveolar Bone Loss’ [Mesh] OR ‘Tooth
Mobility’ [Mesh] OR ‘Periodontal Pocket’ [Mesh]
OR ‘Clinical Attachment Loss’ [txt words] OR
Review on frequency of tooth brushing and periodontitis
‘Pocket Probing Depth’ [txt words]). The search
was extended to eight other online libraries. The
complete search protocol for all databases is avail-
able from the first author (HZ) of this paper.
Study inclusion and exclusion criteria
Two authors (AV and HZ) independently searched
and screened all titles, and reviewed the abstracts to
assess eligibility. Studies were included in the
review if the publication reported an association
[e.g. in terms of odds ratio (OR)] of periodontitis or
alveolar bone loss with oral hygiene, resulting from
an original study using tooth brushing frequency as
an indicator for oral hygiene or if the study reported
data where an association could be derived (23, 24).
Reference lists of the included articles were checked
on related articles and cross-referenced publications.
All studies that only described conditions of peri-
odontitis and did not report ORs or prevalence ratios
(PR), or if these parameters could not be calculated
(25–27) were excluded.
Data extraction and information assessment
From selected papers, we extracted the relevant
study characteristics summarized in Tables 2 and
3. If necessary, effect measures and corresponding
confidence intervals (CIs) were recalculated from
the reported papers. As indicated in the results sec-
tion, some authors were contacted and asked to
provide additional analyses. The prevalence of
destructive periodontal disease given in the last
column of Table 3 was taken from the original
papers or, if not explicitly stated, was calculated
from the available data (24, 28–30).
Phenotyping
Different phenotype definitions were used in the
studies. The following were used, and in some stud-
ies, several were combined: community periodontal
index (CPI), clinical attachment loss (CAL), pocket-
probing depth (PPD) and alveolar bone loss (ABL).
CAL was assessed clinically, radiographically or both.
Assessment and categorization of tooth
brushing frequency
Tooth brushing frequency was commonly assessed by
direct questioning or by providing predefined answer
categories. These were different between studies, and
it was not possible to generate an identical dichotomi-
zation. A final classification was considered suffi-
ciently homogeneous across categories (Table 1). The
association between tooth brushing and periodontitis
was presented by the OR and a corresponding 95%
117
Zimmermann et al.

CI. If OR was available, but no CI was mentioned in
the publication, the approximate CI was calculated
from the corresponding p-value (31).
Meta-analysis methods
The basis for meta-analysis was the published, recal-
culated or provided ORs for infrequent tooth brush-
ing compared to frequent tooth brushing and the
corresponding 95% CI. Heterogeneity of study-spe-
cific effects was assessed using I² statistics and the
inverse variance method (32). Random-effects (RE)
and a fixed-effects model (FE) were calculated (33)
with Comprehensive Meta-Analysis Software (CMA).
Indication of publication bias was assessed based on
REVMANAGER (Version 5.2) with a funnel plot. Indica-
tion of heterogeneity was also tested with REVMANAG-
ER. A forest plot was provided to display the results
of the studies and of the meta-analysis (34).
remained relevant for thorough analysis. Three
studies of the 21 studies were excluded because of
the following reasons: one study reported on self-
rated oral health (35) in association with oral-
hygiene practices. One associated bad mouth
breath (36) with periodontitis, but that study was
not eligible for meta-analysis due to its inclusion
criteria. Another study (37) used oral hygiene as a
dichotomous variable determined by the dentist’s
judgment. Seven authors were asked to provide
supplementary data, full texts or information nec-
essary for analysis. From three authors, further
information could be obtained (see acknowledge-
ments’). In the other papers, no information on OR
for tooth brushing was given or could be calculated
from table data. Hence, we have excluded these
papers in the final meta-analysis (38–41). Fourteen
studies were included in the final review (Fig. 2).

Results
Literature search
The literature search yielded 2.063 articles (Fig. 1).
Of these, 197 duplicate articles were deleted and
1.866 titles and abstracts were reviewed. Among
that group, 42 met the inclusion criteria for full-text
review. Eight of the 42 articles were found through
cited references. Through this search, 21 studies
Characteristics of studies
The studies (Tables 2 and 3) were distributed all
over Europe, the Americas and Asia. There were
twelve cross-sectional studies (two nested into a
cohort) and two case–control studies (42, 43). The
sample sizes ranged from 94 (44) to 9203 (45).
Quality aspects of the studies
The Health Evidence Bulletin (HEB) WALES
checklist was used to address the critical apprai-

Table 1. Studies included in meta-analysis. Tooth brushing frequency and periodontitis: exposure categorization in the
original publications and in the meta-analysis
Study
Hansen et al., 1993 (49)
Gelskey et al., 1998 (43)
L

opez et al., 2001 (45)
Merchant et al., 2002 (29)
Teng et al., 2003 (42)
De Mac^edo et al., 2006 (23)
Wang et al., 2007 (30)
Julihn et al., 2008 (53)
Vysniauskaite and Vehkalahti,
2009 (44)
Herman et al., 2009 (28)
Han et al., 2009 (47)
Al Habashneh et al., 2010 (55)
Saxlin et al., 2011 (48)
Zini et al., 2011 (24)
OR, Odds Ratio.
a
Crude ORs.
Original/Final effect Final tooth brushing
Original tooth brushing categories measure categories
0–1/day versus ≥2/day OR ≤1/day versus ≥2/day
<1/day versus ≥1/day OR <1/day versus ≥1/day
<1/day versus 1/day versus >1/ OR ≤1/day versus >1/day
day
≤1/day versus 2/day versus >2/ OR ≤1/day versus ≥2/day
day
1/day or rarely versus 2/day OR ≤1/day versus ≥2/day
versus >2/day
≤1/day versus ≥2/day PR/OR ≤1/day versus ≥2/day
<2/day versus ≥2/day OR <2/day versus ≥2/day
<1/day versus ≥1/day OR <1/day versus ≥1/day
<1/day versus 1/day versus ≥2/ OR <1/day versus ≥1/day
day
≥1/day versus 0/day OR <1/day versus ≥1/day
<2/day versus ≥2/day ORa ≤1/day versus ≥2/day
No versus Yes ORa 0/day versus ≥1/day
<1/day versus 1/day versus ≥2/ ORa ≤1/day versus ≥2/day
day
<1/day versus 1/day versus 2/ OR ≤1/day versus >1/day
day

118
 Review on frequency of tooth brushing and periodontitis

Fig. 1. Flow diagram of the systematic

review. Search strategy, number of papers
at each level according to preferred
reporting items for systematic reviews
and meta-analyses (PRISMA).

Fig. 2. Forest plot and result of meta-analysis. Association between tooth brushing frequency and periodontitis – cross-
sectional studies only and all studies. *, **, ***, Calculated, combined OR. #, <1 = (never, seldom, sometimes, often but
less then daily) versus daily (≥1)

sal of observational studies (46). It comprises vari- each included study, an attempt was made to
ous aspects with several questions for cohort, answer the questions based on HEB Wales. These
case–control and cross-sectional study types. For were given in the Appendix S1. If a question

119
Zimmermann et al.

Table 2. Studies included in meta-analysis. Tooth brushing frequency and periodontitis: overview of study characteristics
Study
design, Study population/
Author(s), time of Sample size/Sex Age Ascertainment Confounder Control/
Country recruitment distribution range (restrictions) Adjustment
Hansen et al., Cross- 119/(52.9% female, 50 Dental Faculty members, Sex, previous periodontal
1993, Sectional, 47.1% male) university of Oslo treatment, OHI-S, tooth
Norway NA brushing frequency,
≤10 years at school
Gelskey et al., Case– 205 cases: moderate, 35 to Dentulous individuals, Age and sex
1998, Control advanced AP/205 87 dental teaching clinic,
Canada Study controls: healthy, Faculty of Dentistry,
1998 gingivitis or slight AP University of Manitoba,
(n.a.) Winnipeg
L

opez et al., Cross- 9203/(49.2% female, 12 to Pupils or students/98 Age, sex, smoking, dental
2001, Chile Sectional 50.8% male) 21 schools, Province of attendance, diabetic
1999 Santiago (regional) status, governmental
school support
Merchant Cross- 533/(71.5% female, 52 to Dentists and other health Age, smoking, diabetes,
et al., 2002, Sectional 28.5% male) 85 professionals enrolled in profession, history of
USA (nested ongoing cohort studies periodontal surgery,
into in the United States number of teeth present,
Cohort) (nationwide) case–control status (CHD)
1999
Teng et al., Case– 250 cases: chronic 33 to Hospital patients in Age, sex, occupation,
2003, Taiwan Control periodontitis/250 52 Changhua City education
Study controls: other oral (regional)
1998–1999 disease patients/
(46.0% female, 54.0%
male)
de Mac^edo Cross- 172/(62.2% female, 20 to Village inhabitants/ Age, sex, schooling,
et al., 2006, Sectional 37.8% male) 60 Matinha dos Pretos, monthly per capita family
Brazil 2000 State of Bahia, in the income, agglomeration,
North-East region of guidance with regard to
Brazil (regional) oral health, smoking/ex-
smoking, use of alcohol,
plaque index, missing
teeth, teeth present
Wang et al., Cross- 1590/(54.3% female, 26 to ~1/2 farmers, ~1/2 urban Age, sex, smoking,
2007, China Sectional 45.7% male) ~70a professionals/4 prophylaxis, bleeding,
2005 different regions and income and
(northwest, southwest, education levels
northeast and east) in
China (countrywide)
Julihn et al., Cross- 686/(52.0% female, 19 Individuals enrolled in Both parents’ educational
2008, Sectional 48.0% male) seven public dental level, occupational status
Sweden 2001 clinics in suburban
Stockholm (counties)
Vysniauskaite Cross- 94/(52.0% female, 60 to Dentate subjects, elderly Age, sex, level of
and Sectional 48.0% male) ~80a Lithuanians/Two public education, dental
Vehkalahti, 1999–2001 dental offices in cleanliness, number of
2009, Lithuania (regional one teeth remaining
Finland/ city)
Lithuania
Hermann Cross- 4153/(62.2% female, 18 to 304 survey locations from Age, geographic area, fixed
et al., 2009, Sectional 37.8% male) ~80a all Hungarian regions partial denture, smoking,
Hungary 2005–2006 (nationwide) dental floss use, toothpick
use, mouthwash use,
toothpaste, last dental
office visit, pulmonary
disease, hypertension,
education, dental office
attendanceb

120
 Review on frequency of tooth brushing and periodontitis

Table 2 Continued

Study
design, Study population/
Author(s), time of Sample size/Sex Age Ascertainment Confounder Control/
Country recruitment distribution range (restrictions) Adjustment
Han et al., Cross- 1328/(55.0% female, 15 to Randomly selected NA
2009, Korea Sectional 45.0% male) 84 subjects in two cities
(nested (Shiwha/Banwo), Korea
into
Cohort),
2005–2006
Al Habashneh Cross- 400 women 50 to Postmenopausal women/ NA
et al., 2010, Sectional 75 King Abdullah
Jordan 2008–2009 University Hospital,
Jordan University
Saxlin et al., Cross- 396/(64.4% female, 30 to Finland Health Survey NA
2010, Sectional 35.6% male) 59 2000 and 4 years
Finland 2000– Follow-Up Study
2001, 2004
–2005
Zini et al., Cross- 254/(50.0% female, 35 to Randomly selected Sex, age, education,
2011, Israel Sectional 50.0% male) 44 Jewish parents in the employment status, home
2008–2009 city Jerusalem (regional) density
NA, not available; AP, Adult periodontitis.
a
Upper limit not given.
b
Adjusted for a variety of variables: only generic terms are mentioned.

could be answered with a clear ‘yes’ (1), the study
has been given a point for this specific question.
For answers, which could not be answered clearly
(0) or explicitly answered with ‘No’ ( 1), we have
instead awarded 0 points (Table 4). The quality of
the studies was then categorized as follows:
high (14–16), mid-high (12 + 13), mid (10 + 11),
low-mid (8 + 9) and low (<8).
Critical appraisal on single studies
In all the five studies of Hermann (28), Zini (24),
Han (47), Saxlin (48) and Hansen (49), a dichoto-
mized CPI score was used to assess periodontal con-
ditions with cutpoint as low (<4) and high (=4) in
Hansen (49) and (<3) and (≥3) in the others. The
strength of the CPI lies in its high sensitivity if only
one tooth is tested positive for a given condition it
defines the whole sextant. Only the worst conditions
of each sextant is noted so that using the CPI leads
to an overestimation of disease prevalence. On the
other hand, a partial examination recording systems
might lead to some underestimation of disease. Nev-
ertheless, a recent review has shown a good corre-
spondence of partial examination and full mouth
scores (50). Another criticism has been its reliance on
just PD. The CPI was extensively tested (51, 52).
In this study from Vysniauskaite and Vehkalahti
(44), the prevalence was determined according to
mean PD measurements. These are clinically
important because they provide a useful overall
estimate of the depth of all periodontal pockets
and persons with generalized deep pockets get
well characterized. Localized forms of periodonti-
tis are difficult to recognize because of calculation
of the arithmetic mean. However, they give a
worse overall estimate of the amount of damage to
the periodontium than do CAL assessments, which
are on the other hand difficult to accurately mea-
sure.
The diagnosis of periodontitis in both studies
from L
opez (45) and Wang (30) was based on CAL,
which is primarily a disease sequelae and does not
reflect active disease.
A combination of PPD and CAL on six sites per
tooth was used in the two studies from Teng (42)
and de Mac^edo (23). These methods combine the
advantages and disadvantages of pure PD and
CAL recordings. Both studies from Merchant (29)
and Julihn (53) were based on radiographic assess-
ments, which also have previously been used to
measure severity and extent of periodontal disease.
The US study’s primary outcome was self-reported
periodontal disease among dental clinicians, which
was further evaluated with dental radiographs.
The study from Stockholm used bitewing radio-
graphs which are only suitable to report if a bone
loss of above 2 mm is present, because usually the
apical third of the tooth is not displayed. This

121
Zimmermann et al.

Table 3. Studies included in meta-analysis. Tooth brushing frequency and periodontitis: methods of dental examina-
tions and disease prevalence assessment
Author(s), Recording, clinical Destructive periodontal disease
year measurement(s) Instruments definition (DPDD) Prevalence
Hansen et al., Full mouth, (6 sextants) WHO probe Dichotomized CPI Score (low <4, NA
1993 CPI high 4) Score 4 represents ≥1
pockets >5.5 mm.
Gelskey et al., Radiographs (bitewings Viewbox, without Periodontitis if ≥1 teeth having NA
1998 and periapical magnification, 3 mm bone loss ≥4 mm from cemento
radiographs), PD for vernier caliper enamel junction and alveolar
confirmation crest, or ≥1 teeth having ≥7 mm
PD
L

opez et al., CAL at 6 sites on all Calibrated manual Periodontitis, if 1 tooth with CAL 4.5%
2001 teeth, except premolars periodontal probes: tip ≥3 mm all sites
and 3rd Molars diameter 0.5 mm, head
lamp
Merchant Self-reported Periodontal probe, Periodontitis, estimated (mean) 28.7%a
et al., 2002 periodontal disease, viewing box, radiographic bone loss
estimated radiographic magnifying loops
bone loss in mm at 2
sites
Teng et al., Full mouth, PD and Florida Probe Periodontitis, if two or more NA
2003 CAL at six sites per interproximal sites at different
tooth teeth with CAL ≥6 mm and ≥1
site with PD ≥5 mm
de Mac^edo Full mouth, PPD and Manual periodontal Periodontitis, if ≥4 teeth with ≥1 24.4%
et al., 2006 CAL at 6 sites per tooth probe, natural sites with PD ≥4 mm and CAL
light + flashlight ≥3 mm in the same places
Wang et al., Mean PD and mean Manual periodontal probe Periodontitis, mean value CAL 43%a
2007 CAL at 6 index teeth, 6 ≥3 mm (defined: moderate) each
sites per tooth subject
Julihn et al., Radiographic Magnifying lupe Incipient alveolar bone loss if 5.1%
2008 (convention and (conventional ≥2 mm
digital) measurement radiographs), manual
graded periodontal
probe
Vysniauskaite Half mouth, PD at 4 sites WHO probe with colour ≥1 pocket of ≥6 mm 70%
and per tooth band (3.5–5.5 mm), ball
Vehkalahti, of 0.5 mm diameter,
2009 operating light
Hermann Full mouth (6 sextants) WHO probe with colour Dichotomized CPI Score (low <3, 30.5%a
et al., 2009 CPI, 3rd Molars if they band (3.5–5.5 mm), ball high 3,4)
have the function of of 0.5 mm diameter,
2nd molars artificial spot light
Han et al., CPI at six sites at 10 WHO probe, Dental unit Dichotomized CPI Score (low <3, 28.5%a
2009 index teeth and light high 3,4)
Al Full mouth, PD at six Williams periodontal Periodontitis, if ≥2 interproximal 44.5%a
Habashneh, sites per tooth probe sites with ≥5 mm PD,
2010 radiographic bone loss, ≥6 mm
CAL
Saxlin et al., Full mouth, CPI, PD at WHO probe, headlamp Periodontitis, if ≥1 tooth with PD 58.5%a
2010 four sites per tooth, ≥4 mm
only deepest site
recorded
Zini et al., CPI at six sites at 12 WHO probe, in natural CPI - highest score recorded, 47.2%a
2011 index teeth light dichotomized CPI Score (low <3,
high 3,4)
CPI, Community Periodontal Index: generally defined as at least one pocket ≥ 3.5 mm; NA, Not Available; CAL,
Clinical Attachment Loss; PD, Probing Depth; PPD, Pocket-Probing Depth.
a
Not given in the original paper, recalculated (see text).

122
 Table 4. Final Quality Score. Single studies based on HEB WALES
Study Score Quality 2.1 2.2 2.3 2.4 3 4.1 4.2 5.1 5.2 5.3 5.4 5.5 6 7 8 9 10 11 12 13
c b
1 Hansen et al. 8 of 15 Low-Mid X – X X X 0 – 0 X 0 – – – X X X 0 0 1 0cd
2 Gelskey et al. 13 of Mid- X X X X X – X X – – X 0^ – X X X 0h X 1b X
16 High
3 L

opez et al. 12 of Mid- X – X X X X – X X 0b – – – X X X 0d X 1b X
15 High
4 Merchant et al. 11 of Mid X – X X X 0c – X X X – – – X X X 1c X 1b 0c
15
5 Teng et al. 15 of High X X X X X – X X – – X X – X X X X X 1b X
16
6 De Macedo et al. 10 of Mid X – X X X X – X X 0b – – – X X Xe 0cd 0f 1b 0cd
15
7 Wang et al. 13 of Mid- X – X X X X – X X 0b – – – X X X X X 1b X
15 High
8 Julihn et al. 10 of Mid X – X X X X – 0f X 0b – – – X X X 1d X 1b 0cd
15
9 Vysniauskaite and 8 of 15 Low-Mid X – X X X 1c – 0cf X 0b – – – X X X 1d 0d 1b 1cd
Vehkalahti
10 Herman et al. 13 of Mid- X – X X X X – X X 0b – – – X X X X X 1b X
15 High
11 Han et al. 12 of Mid- X – X X X X – 0f X 0b – – – X X X X X 1b X
15 High
12 Al Habashneh et al. 9 of 15 Low-Mid X – X X X 1c – X 1a 0b – – – X X X 0cd X 1b 0cd
13 Saxlin et al. 13 of Mid- X – X X X X – Xg X 0b – – – X X X X X 1b X
15 High
14 Zini et al. 12 of Mid- X – X X X X – 0f X 0b – – – X X X X X 1b X
15 High
– Not applicable.
a
Participation/response rate ≤50%.
b
No information given.
c
Limitations due to selection criteria, number of persons, selected study population, regions: sample might not be representative of it’s target population.
d
Differences of local (sex, persons age, socioeconomic profile), cultural (different ethnic types) and regional setting (restrictions to local settings), also in respect to studied
population.
e
Only minor information on statistical methods.
f
Confounding or bias or possible explanations largely considered.
g
Different focus than periodontitis, hence limited validity in this direction. However, generally important aspects considered.
h
Not clearly stated.
Review on frequency of tooth brushing and periodontitis

123
Zimmermann et al.
study (53) was included, although chronic peri-
odontitis is relatively rare in an age cohort of 19-
year-old individuals (54). In general, radiographs
have not been the predominant measurement in
epidemiological studies because of radiation expo-
sure, the cumbersome nature of radiography under
field conditions and magnification problems.
This study from Gelskey (43) used a combination
of PD and radiographic evaluation is used to con-
firm the diagnosis of persons with adult periodon-
titis allocated to their test group. Al Habashneh
(55) used PD, CAL and radiographic bone loss
jointly to define established periodontits.
Meta-analysis
The summary OR estimate for all cross-sectional
studies with the fixed-effects model is 1.41 (95%
CI = 1.25, 1.58). For all fourteen studies, the sum-
mary estimates for the fixed- and random-effects
models are 1.44 (95% CI = 1.29, 1.62) and 1.44 (95%
CI = 1.21, 1.71), respectively. The overall result of
the meta-analysis on the association of tooth brush-
ing frequency with periodontitis is shown in Fig. 2.
The cross-sectional studies showed homogeneous
results with ORs ranging from 1.01 to 3.96. The
tests for these studies showed no indication for het-
erogeneity (chi-squared = 13.90, P = 0.31,
I² = 14%). The test of heterogeneity with all study
types jointly showed a significant result (chi-
squared = 27.16, P = 0.02, I² = 48%).
Various combinations of selected studies (based
on the assessment of HEB Wales) were tested when
modelling the RE and FE to check whether they fit
with our final pooled estimate including all cross-
sectional studies jointly. Commonly testing the
studies that scored ‘Low-Mid’ and ‘Mid’ (Table 4)
resulted in an OR of 1.24 (95%CI: 0.93–1.66, chi-
squared = 6.74, I² = 26%; P = 0.24). Subgroup
analysis on studies that scored as ‘Mid-high’
resulted in OR of 1.41 and CI ranging from 1.20 to
1.65 (chi-squared = 8.93, P = 0.26, I² = 22%).
Hence, subgroup analysis showed similar results.
A funnel plot is given in Fig. 3. No publication
bias is apparent; however, due to small number of
studies it has limited power for detection.
Discussion
While most single studies showed an increased but
insignificant association of periodontitis with
infrequent tooth brushing, this meta-analysis pro-
vided a clear and highly significant effect
124
Fig. 3. Funnel plot of meta-analysis. Detection of possi-
ble publication bias.
(P < 0.0001) with an overall odds ratio of about 1.4.
For all cross-sectional studies combined, we
obtained a significant summary estimate for the
association of low-frequency tooth brushing with
periodontitis (OR = 1.41, 95% CI = 1.25–1.58) using
a fixed-effects model (s2 = 0.01; chi-
squared = 13.90, P = 0.31, I² = 14%). A random-
effects model with s2 = 0.05; chi-squared = 27.16,
P = 0.02, I² = 48% confirmed this with an OR of
1.44 and 95% CI 1.21–1.71.
There were no consistent clinical measurements
across studies. In our analysis, however, we tried
to clearly distinguish between studies evaluating
periodontitis and those evaluating calculus/gingi-
vitis only. The meta-analysis based on quality
ranking of HEB Wales for subgroups revealed sim-
ilar results. The prevalence of infrequent tooth
brushing varied largely between the studies from
about 9% to 57%. The risk attributable to infre-
quent tooth brushing therefore depends on the
considered population.
The studies included were quite heterogeneous
with respect to age groups considered, sample size
and regional setting. However, nearly all of these
studies have a sufficiently high quality assessed by
HEB Wales. The sample size varied by a factor of
100. The ages ranged from adolescents to seniors.
The most used statistical method was multivariate
logistic regression, although the variables included
for confounder adjustment were rather different
between studies (Table 2). It cannot be ruled out
that in some studies relevant confounders were not
assessed or not included in the final analysis
model. For example, high tooth brushing fre-
quency does not necessarily imply efficient tooth
brushing although a positive correlation is likely.
There are other parameters like dental visits, which

may also have a confounding effect (alone or in
combination with professional tooth cleaning,
including oral-hygiene instruction), when assess-
ing the effect of regular brushing on periodontitis.
These were only considered in few of the studies
included here. Any plaque indices or scores (56)
could be tools to quantify effectiveness of oral
hygiene. A relatively low plaque control record
(PCR) could be correlated to an effective oral
hygiene.
Except two case–control studies, all studies were
cross-sectional. This study type generally provides
limited evidence for assessing causal relations
between a factor and a disease in comparison to a
cohort study for example. Although a causal rela-
tion appears to be likely, the findings must be
interpreted with caution. Only four studies explic-
itly mentioned use of sample-size calculations (24,
28, 45, 53).
Every effort was made to include all relevant
studies. Hence, two authors (HZ and AV) per-
formed an independent search. In using MeSH
search, un-indexed studies could have been
missed, although the probability seems quite small.
Because findings are only based on observational
studies, the fact that studies are affected by system-
atic errors like recall or selection bias, cannot be
excluded.
The qualities of studies were assessed on the
basis of HEB Wales and accordingly, these were
classified differently. The assessment of the studies
depends on the subjective assessment of the
issues/evaluation of HEB Wales. As this evalua-
tion criterion is largely subjective, other evaluation
guidelines could also yield other results.
In studies that received a lower ranking score,
the chance is higher that possibly inaccuracies in
study designs, could weaken the real effect of tooth
brushing on periodontitis. As the studies were
viewed both from the epidemiological point of
view as well as from dental respects in relation to
periodontitis, it can also have happened, of course,
that studies with a different overall focus have
therefore been ranked less high.
A concern is the heterogeneity of the studies
regarding outcome assessment and clinical mea-
surements methods. At present in international
periodontitis literature, no agreement has yet been
found for a consistent periodontitis case definition
(57). Also in terms of severity of periodontal dis-
ease, very different criteria and metrics were used
(22, 57) in the studies. This is also related to the
index systems documented in the literature for
Review on frequency of tooth brushing and periodontitis
empirical detection of periodontal disease, such as
the community periodontal index of the World
Health Organization, with a focus on PD of 5 mm,
6 mm (58, 59) or the classification model of the
AAP for the measurement of potential attachment
loss in terms of their scope and significance (60).
These remain controversial. Combined evaluations
of PD and AL (i.e. the Centers for Disease Control
and Prevention/AAP index system), while pre-
ferred in the recent epidemiological research, were
not available across the studies included. Another
critical methodological aspect is that, in most of the
studies, only a partial mouth recording was per-
formed, leading to the possible under-reporting of
periodontal disease burden (61).
Conclusions
In summary, this meta-analysis has shown that
infrequent tooth brushing is statistically signifi-
cantly associated with periodontitis. However, this
association is relatively small and the risk attribut-
able to this factor is moderate. No large prospective
cohort study exists. Further prospective epidemio-
logical studies are needed to more precisely esti-
mate the effect.
Acknowledgements
We want to specially thank all authors that have sup-
ported us by contributing additional data or analyses of
their studies, namely M. Vehkalati (Finland), A. Mer-
chant (USA) and A. Zini (Israel). The study was sup-
ported by the Federal Ministry of Education and
Research of the Federal Republic of Germany (BMBF)
grant number 01ER1001B.
Conflict of interest
This work was supported by the German Ministry
of Education and Research (BMBF). The authors
declare no competing interests.
Authors’ contributions
HZ conducted the meta-analysis and drafted the
manuscript. TSK, DH, AV and NZ contributed to
the methodological part of the manuscript. HZ and
AV both did the literature search independently.
HB and TSK led the present project. HB partici-
125
Zimmermann et al.

pated substantially in writing the manuscript. All health status in older white men. J Dent Res
authors read and approved the final version. 1999;78:869–77.
16. Loe H, Theilade E, Jensen SB. Experimental gingivi-
tis in man. J Periodontol 1965;36:177–87.
17. Page RC, Kornman KS. The pathogenesis of human
periodontitis: an introduction. Periodontol 2000;
References 1997:9–11.
18. Hujoel PP, Cunha-Cruz J, Loesche WJ, Robertson PB.
1. Hayashi C, Gudino CV, Gibson FC 3rd, Genco CA.
Personal oral hygiene and chronic periodontitis: a
Review: pathogen-induced inflammation at sites dis-
systematic review. Periodontol 2000;2005:29–34.
tant from oral infection: bacterial persistence and
19. Baker PJ, Roopenian DC. Genetic susceptibility to
induction of cell-specific innate immune inflamma-
chronic periodontal disease. Microbes Infect
tory pathways. Mol Oral Microbiol 2010;25:305–16.
2002;4:1157–67.
2. Calsina G, Ram
on J-M, Echeverr
ıa J-J. Effects of
20. Liberati A, Altman DG, Tetzlaff J, Mulrow C, Gotz-
smoking on periodontal tissues. J Clin Periodontol
sche PC, Ioannidis JP et al. The PRISMA statement
2002;29:771–6.
for reporting systematic reviews and meta-analyses
3. Lages EJ, Costa FO, Lages EM, Cota LO, Cortelli SC,
of studies that evaluate healthcare interventions:
Nobre-Franco GC et al. Risk variables in the associa-
explanation and elaboration. BMJ 2009;339:b2700.
tion between frequency of alcohol consumption and
21. Leroy R, Eaton KA, Savage A. Methodological issues
periodontitis. J Clin Periodontol 2012;39:115–22.
in epidemiological studies of periodontitis–how can
4. Salvi GE, Carollo-Bittel B, Lang NP. Effects of diabe-
it be improved? BMC Oral Health 2010;10:8.
tes mellitus on periodontal and peri-implant condi-
22. Savage A, Eaton KA, Moles DR, Needleman I. A sys-
tions: update on associations and risks. J Clin
tematic review of definitions of periodontitis and
Periodontol 2008;35(8 Suppl):398–409.
methods that have been used to identify this disease.
5. Schaefer AS, Richter GM, Groessner-Schreiber B, No-
J Clin Periodontol 2009;36:458–67.
ack B, Nothnagel M, El Mokhtari NE et al. Identifica-
23. de Mac^edo TCN, da Conceicß~ ao N, Costa M, Gomes-
tion of a shared genetic susceptibility locus for
Filho IS, Vianna MIP, Santos CT. Factors related to
coronary heart disease and periodontitis. PLoS Genet
periodontal disease in a rural population. Braz Oral
2009;5:e1000378.
Res 2006;20:257–62.
6. Schaefer AS, Richter GM, Nothnagel M, Manke T,
24. Zini A, Sgan-Cohen HD, Marcenes W. Socio-eco-
Dommisch H, Jacobs G et al. A genome-wide associa-
nomic position, smoking, and plaque: a pathway to
tion study identifies GLT6D1 as a susceptibility locus
severe chronic periodontitis. J Clin Periodontol
for periodontitis. Hum Mol Genet 2010;19:553–62.
2011;38:229–35.
7. Timmerman MF, van der Weijden GA. Risk factors
25. Treasure E, Kelly M, Nuttall N, Nunn J, Bradnock G,
for periodontitis. Int J Dent Hyg 2006;4:2–7.
White D. Factors associated with oral health: a multi-
8. Eickholz P, Kaltschmitt J, Berbig J, Reitmeir P, Pretzl
variate analysis of results from the 1998 Adult Dental
B. Tooth loss after active periodontal therapy. 1:
Health survey. Br Dent J 2001;190:60–8.
patient-related factors for risk, prognosis, and quality
26. Kalsbeek H, Truin GJ, Poorterman JH, van Rossum
of outcome. J Clin Periodontol 2008;35:165–74.
GM, van Rijkom HM, Verrips GH. Trends in peri-
9. Amarasena N, Ekanayaka AN, Herath L, Miyazaki
odontal status and oral hygiene habits in Dutch
H. Tobacco use and oral hygiene as risk indicators
adults between 1983 and 1995. Community Dent
for periodontitis. Commun Dent Oral Epidemiol
Oral Epidemiol 2000;28:112–8.
2002;30:115–23.
27. Morris AJ, Steele J, White DA. The oral cleanliness
10. Bakdash B. Oral hygiene and compliance as risk fac-
and periodontal health of UK adults in 1998. Br Dent
tors in periodontitis. J Periodontol 1994;65(5 Sup-
J 2001;191:186–92.
pl):539–44.
28. Hermann P, Gera I, Borb
ely J, Fej
erdy P, Madl
ena M.
11. Liu Z, Zhang W, Zhang J, Zhou X, Zhang L, Song Y
Periodontal health of an adult population in Hun-
et al. Oral hygiene, periodontal health and chronic
gary: findings of a national survey. J Clin Periodon-
obstructive pulmonary disease exacerbations. J Clin
tol 2009;36:449–57.
Periodontol 2012;39:45–52.
29. Merchant A, Pitiphat W, Douglass CW, Crohin C,
12. Albandar JM, Buischi YA, Axelsson P. Caries lesions
Joshipura K. Oral hygiene practices and periodonti-
and dental restorations as predisposing factors in the
tis in health care professionals. J Periodontol
progression of periodontal diseases in adolescents. A
2002;73:531–5.
3-year longitudinal study. J Periodontol 1995;66:249–
30. Wang Q-T, Wu Z-F, Wu Y-F, Shu R, Pan Y-P, Xia
54.
J-L. Epidemiology and preventive direction of peri-
13. Broadbent JM, Williams KB, Thomson WM, Wil-
odontology in China. J Clin Periodontol 2007;34:
liams SM. Dental restorations: a risk factor for peri-
946–51.
odontal attachment loss? J Clin Periodontol
31. Altman DG, Bland JM. How to obtain the confidence
2006;33:803–10.
interval from a P value. BMJ 2011;343:d2090.
14. Marcus SE, Drury TF, Brown LJ, Zion GR. Tooth
32. Higgins JPT, Thompson SG. Quantifying heterogene-
retention and tooth loss in the permanent dentition
ity in a meta-analysis. Stat Med 2002;21:1539–58.
of adults: United States, 1988–1991. J Dent Res
33. Higgins JPT, Thompson SG, Deeks JJ, Altman DG.
1996;75:684–95.
Measuring inconsistency in meta-analyses. BMJ
15. Boehmer U, Kressin NR, Spiro A 3rd. Preventive
(Clinical research ed) 2003;327:557–60.
dental behaviors and their association with oral

126

34. Lewis S, Clarke M. Forest plots: trying to see the
wood and the trees. BMJ (Clinical research ed)
2001;322:1479–80.
35. Cascaes AM, Peres KG, Peres MA. Periodontal dis-
ease is associated with poor self-rated oral health
among Brazilian adults. J Clin Periodontol
2008;36:25–33.
36. Mumghamba E, Manji K, Michael J. Oral hygiene
practices, periodontal conditions, dentition status
and self-reported bad mouth breath among young
mothers Tanzania. Int J Dent Hyg 2006;4:166–73.
37. Abdellatif HM, Burt BA. An epidemiological investi-
gation into the relative importance of age and oral
hygiene status as determinants of periodontitis.
J Dent Res 1987;66:13–8.
38. Khader YS, Rice JC, Lefante JJ. Factors associated
with periodontal diseases in a dental teaching clinic
population in northern Jordan. J Periodontol
2003;74:1610–7.
39. Kushiyama M, Shimazaki Y, Murakami M, Yamash-
ita Y. Relationship between intake of green tea and
periodontal disease. J Periodontol 2009;80:372–7.
40. Yanagisawa T, Ueno M, Shinada K, Ohara S, Wright
FA, Kawaguchi Y. Relationship of smoking and
smoking cessation with oral health status in Japanese
men. J Periodontal Res 2010;45:277–83.
41. Ababneh KT, Abu Hwaij ZM, Khader YS. Prevalence
and risk indicators of gingivitis and periodontitis in
a multi-centre study in North Jordan: a cross sec-
tional study. BMC Oral Health 2012;12:1.
42. Teng H-C, Lee C-H, Hung H-C, Tsai DC-C, Chang
Y-Y, Yang Y-HC et al. Lifestyle and psychosocial fac-
tors associated with chronic periodontitis in Taiwan-
ese adults. J Periodontol 2003;74:1169–75.
43. Gelskey SC, Young TK, Singer DL. Factors associated
with adult periodontitis in a dental teaching clinic
population. Community Dent Oral Epidemiol
1998;26:226–32.
44. Vysniauskaite S, Vehkalahti M. Impacts of tooth-
brushing frequency on periodontal findings in a
group of elderly Lithuanians. Oral Health Prev Dent
2009;7:129–36.
45. L
opez DR, Fern
andez O, Jara G, Baelum V. Epidemi-
ology of clinical attachment loss in adolescents. J Pe-
riodontol 2001;72:1666–74.
46. Sanderson S, Tatt ID, Higgins JP. Tools for assess-
ing quality and susceptibility to bias in observa-
tional studies in epidemiology: a systematic review
and annotated bibliography. Int J Epidemiol 2007;
36:666–76.
47. Han DH, Lim SY, Sun BC, Janket SJ, Kim JB, Paik
DI et al. Mercury exposure and periodontitis
among a Korean population: the Shiwha-Banwol
environmental health study. J Periodontol 2009;80:
1928–36.
48. Saxlin T, Ylostalo P, Suominen-Taipale L, Mannisto
S, Knuuttila M. Association between periodontal
infection and obesity: results of the Health 2000 Sur-
vey. J Clin Periodontol 2011;38:236–42.
Review on frequency of tooth brushing and periodontitis
49. Hansen BF, Bjertness E, Gronnesby JK. A socio-eco-
logic model for periodontal diseases. J Clin Period-
ontol 1993;20:584–90.
50. Tran DT, Gay I, Du XL, Fu Y, Bebermeyer RD, Neu-
mann AS et al. Assessing periodontitis in popula-
tions: a systematic review of the validity of partial-
mouth examination protocols. J Clin Periodontol
2013;40:1064–71.
51. Barmes D. CPITN–a WHO initiative. Int Dent J
1994;44(5 Suppl 1):523–5.
52. Holmgren CJ. CPITN–interpretations and limita-
tions. Int Dent J 1994;44(5 Suppl 1):533–46.
53. Julihn A, Agholme MB, Modeer T. Risk factors and
risk indicators in relation to incipient alveolar bone
loss in Swedish 19-year-olds. Acta Odontol Scand
2008;66:139–47.
54. Heitz-Mayfield LJA, Sch€ atzle M, L€oe H, B€urgin W,

Anerud  A, Boysen H et al. Clinical course of chronic
periodontitis. II. Incidence, characteristics and time
of occurrence of the initial periodontal lesion. J Clin
Periodontol 2003;30:902–8.
55. Al Habashneh R, Alchalabi H, Khader YS, Hazza’a
AM, Odat Z, Johnson GK. Association between peri-
odontal disease and osteoporosis in postmenopausal
women in jordan. J Periodontol 2010;81:1613–21.
56. Burde S, Leary JF. Detection of individual human
chromosomes by chromosome in situ suppression
hybridization using PCR-amplified bacteriophage
library probes. Genet Anal Tech Appl 1992;9:64–7.
57. Page RC, Eke PI. Case definitions for use in popula-
tion-based surveillance of periodontitis. J Periodon-
tol 2007;78(7 Suppl):1387–99.
58. Ainamo J, Barmes D, Beagrie G, Cutress T, Martin
J, Sardo-Infirri J. Development of the World Health
Organization (WHO) community periodontal index
of treatment needs (CPITN). Int Dent J 1982;
32:281–91.
59. Page RC, Morrison EC. Summary of outcomes and
recommendations of the workshop on (CPITN). Int
Dent J 1994;44(5 Suppl 1):589–94.
60. Armitage GC. Diagnosis of periodontal diseases. J
Periodontol 2003;74:1237–47.
61. Kingman A, Susin C, Albandar JM. Effect of partial
recording protocols on severity estimates of peri-
odontal disease. J Clin Periodontol 2008;35:659–67.
Supporting Information
Additional Supporting Information may be found in the
online version of this article:
Appendix S1. HEB Wales for observational studies with
numbered questions. The adapted guideline of the HEB
Wales for observational studies that has been used in this
article
Appendix S2. PRISMA Checklist. The PRISMA (Pre-
ferred Reporting Items for Systematic Reviews and
Meta-Analyses) reporting guidelines.
127
This document is a scanned copy of a printed document. No warranty is given about the
accuracy of the copy. Users should refer to the original published version of the material.