The role of health-related behaviors in the socioeconomic disparities in oral health.

Abstract

This study aimed to examine the socioeconomic disparities in health-related behaviors and to assess if behaviors eliminate socioeconomic disparities in oral health in a nationally representative sample of adult Americans. Data are from the US Third National Health and Nutrition Examination Survey (1988-1994). Behaviors were indicated by smoking, dental visits, frequency of eating fresh fruits and vegetables and extent of calculus, used as a marker for oral hygiene. Oral health outcomes were gingival bleeding, loss of periodontal attachment, tooth loss and perceived oral health. Education and income indicated socioeconomic position. Sex, age, ethnicity, dental insurance and diabetes were adjusted for in the regression analysis. Regression analysis was used to assess socioeconomic disparities in behaviors. Regression models adjusting and not adjusting for behaviors were compared to assess the change in socioeconomic disparities in oral health. The results showed clear socioeconomic disparities in all behaviors. After adjusting for behaviors, the association between oral health and socioeconomic indicators attenuated but did not disappear. These findings imply that improvement in health-related behaviors may lessen, but not eliminate socioeconomic disparities in oral health, and suggest the presence of more complex determinants of these disparities which should be addressed by oral health preventive policies.

Full text

Page 1

The role of health-related behaviors in the socioeconomic disparities
in oral health
Wael Sabbah*, Georgios Tsakos, Aubrey Sheiham, Richard G. Watt
Epidemiology and Public Health, 1-19 Torrington Place, University College London, London WC1E 6BT, United Kingdom
a r t i c l e i n f o
Article history:
Available online 21 November 2008
Keywords:
Health behavior
Health status disparities
Oral health
Socioeconomic factors
USA
a b s t r a c t
This study aimed to examine the socioeconomic disparities in health-related behaviors and to assess if
behaviors eliminate socioeconomic disparities in oral health in a nationally representative sample of adult
Americans. Data are from the US Third National Health and Nutrition Examination Survey (1988–1994).
Behaviors were indicated by smoking, dental visits, frequency of eating fresh fruits and vegetables and
extent of calculus, used as a marker for oral hygiene. Oral health outcomes were gingival bleeding, loss of
periodontal attachment, tooth loss and perceived oral health. Education and income indicated socio-
economic position. Sex, age, ethnicity, dental insurance and diabetes were adjusted for in the regression
analysis. Regression analysis was used to assess socioeconomic disparities in behaviors. Regression
models adjusting and not adjusting for behaviors were compared to assess the change in socioeconomic
disparities in oral health. The results showed clear socioeconomic disparities in all behaviors. After
adjusting for behaviors, the association between oral health and socioeconomic indicators attenuated but
did not disappear. These findings imply that improvement in health-related behaviors may lessen, but not
eliminate socioeconomic disparities in oral health, and suggest the presence of more complex determi-
nants of these disparities which should be addressed by oral health preventive policies.
? 2008 Elsevier Ltd. All rights reserved.
Introduction
There is substantial evidence of socioeconomic disparities in
oral health (Locker, 2000; Watt & Sheiham,1999). Disparities in oral
health have been repeatedly demonstrated, using different markers
of oral health, different indicators of socioeconomic position, and in
different industrialized countries (Locker, 2000; Lopez, Fernandez,
& Baelum, 2006; Sabbah, Tsakos, Chandola, Sheiham, & Watt, 2007;
Sanders, Slade, Turrell, Spencer, & Marcenes, 2006a; Sanders,
Spencer, & Slade, 2006b; Thomson & Mackay, 2004; Watt &
Sheiham, 1999). There is also evidence of an inverse relationship
between unfavorable health-related behaviors and oral health
(Davis, 1980; Locker, 1989; Sanders et al., 2006b; Sheiham & Watt,
2000; Wamala, Merlo, & Bostrom, 2006). Health-related behavior is
defined as ‘‘.overt behavior patterns, actions and habits that relate
to health maintenance, to health restoration and to health
improvement’’ (Gochman, 1982). Two features of the unfavorable
health-related behaviors are that they tend to cluster together in
the same individuals. Second, they are more prevalent in those at
the lower, than those at the top of the social hierarchy (Davis,1980;
Jarvis & Wardle, 2006; Lantz et al., 2006; Locker, 1989).
The dual relationship of health-related behaviors with socio-
economic position on the one hand, and with oral health on the
other hand, implies that behaviors play an important role in the
socioeconomic disparities in oral health. Particularly as some oral
health enhancing behaviors, such as preventive dental visits, are
restricted by costs (Sanders et al., 2006b). It is plausible that posi-
tive changes in health-related behaviors and better access to
preventive dental services could eliminate or significantly decrease
socioeconomic disparities in oral health (Wamala et al., 2006).
Indeed this has been the dominant philosophy underpinning oral
health preventive programs for decades (Chen, 1995; Pine et al.,
2004). However, the significance and importance of the role played
by health-related behaviors in socioeconomic inequality in health
have been challenged in the medical literature (Jarvis & Wardle,
2006; Kivimaki et al., 2007; Lantz et al., 2006; McKinlay, 1993;
Syme, 1996) as well as in the dental literature (Sanders et al.,
2006b; Sheiham, 2000; Watt, 2007). A number of studies have
examined the role of behaviors in socioeconomic disparities in
general health (Jarvis & Wardle, 2006; Kivimaki et al., 2007; Lantz
et al., 2006; McKinlay, 1993; Sheiham, 2000; Syme, 1996) and
concluded that improvements in health enhancing behavior could
lessen inequality in health, but do not eliminate them. However,
there are very few studies which have examined the part played by
health-related behaviors in socioeconomic disparities in oral health
(Sanders et al., 2006b; Wamala et al., 2006).
* Corresponding author. Tel.: þ44 20 7679 5671; fax: þ44 20 7813 0280.
E-mail address: w.sabbah@ucl.ac.uk (W. Sabbah).
Contents lists available at ScienceDirect
Social Science & Medicine
journal homepage: www.elsevier.com/locate/socscimed
0277-9536/$ – see front matter ? 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.socscimed.2008.10.030
Social Science & Medicine 68 (2009) 298–303

Page 2

We set out to assess the effect of health-related behaviors on
socioeconomic inequality in oral health in the American pop-
ulation, using data from the Third National Health and Nutrition
Examination Survey (NHANES III) (US Department of Health and
Human Services, 1999). Our hypothesis is that poorer health-
related behaviors are more common among the poor and less
educated, and that behaviors account for the socioeconomic
differences in oralhealth status. The objectives of this studyare first
to assess whether there are income and education disparities in
health-related behaviors, and second, to assess whether certain
health-related behaviors account for socioeconomic disparities in
oral health in adult American population.
Materials and methods
Data source
For this study we used data from NHANES III, a cross-sectional
national survey which collected data on health, nutrition and
behaviors during the years 1988 through 1994 (US Department of
Health and Human Services, 1999). NHANES III used a stratified
multistage probability sampling design representative of the non-
institutionalized civilian American population. Data on the adult
population aged 17 years and older were used. The survey included
demographic and health-related behavioral data and a compre-
hensive dental examination. Detailed description of the survey and
the variables used in this study were published in a previous paper
(Sabbah et al., 2007).
Oral health outcomes
Periodontal status was assessed using the NIDR protocol (Miller,
Brunelle, Carlos, Brown, & Lo ¨e, 1987). Two variables, previously
used in NHANES studies (Sabbah et al., 2007; Slade & Beck, 1999)
were created to indicate the extent of (1) gingival bleeding and (2)
loss of periodontal attachment of 3 mm or more. They were
calculated as the percentage of tooth sites with the aforementioned
periodontal characteristic within the total number of examined
tooth sites. Another measureof oralhealth is the numberof missing
teeth. Missing tooth surfaces due to dental disease (dental caries or
periodontitis) was used to indicate tooth loss. In relation to
subjective oral health, NHANES III included a question on perceived
oral health. Participants were asked to rate their oral health as
excellent, very good, good, fair or poor. This variable was catego-
rized into two groups: excellent/very good/good versus fair/poor.
Behavioral variables
NHANES III did not include explicit variables on some important
oral health-related behaviors, such as tooth brushing and oral
hygiene practices. However, the database included data on impor-
tant health related behaviors linked to oral health such as smoking,
frequency of dental visits, frequency of eating fresh fruits and
vegetables. Smoking and dental visits are directly linked to oral
health (Ojima, Hanioka, Tanaka, & Aoyama, 2007; Sanders et al.,
2006b), while there is also evidence of an association between
consumption of fruits and vegetables and oral health (Burt et al.,
2006). Considering that health-related behaviors tend to cluster in
the same individuals (Ma, Betts, & Hampl, 2000; Petersen, Jiang,
Peng, Tai, & Bian, 2008), the consumption of fruits and vegetables
variable was included as a proxy behavior of diet. Smoking was
categorized into three groups, current smoker, non-smoker and
non-respondent. Frequency of visits to a dentist was categorized
into two groups (less than once a year versus once a year or more).
Questions on frequency of eating different fresh fruits and
vegetables were used to create an aggregated (summed) variable
indicating daily frequency of eating fresh fruits and vegetables.
Oral hygiene is associated with periodontal disease and tooth
loss (Treasure et al., 2001). On the other hand, dental plaque levels
are good indicators of effective oral hygiene (Maizels & Sheiham,
1987; Morris, Steele, & White, 2001). NHANES III did not have data
on dental plaque, but included the clinical assessment of calculus.
As calculus is calcified plaque, it was considered as a measure of
how long dental plaque had remained undisturbed byeffective oral
hygiene, and therefore was used as a surrogate measure of oral
cleaning behaviors (Maizels & Sheiham,1987). Although calculus is
not a direct observation of behavior, it is the best available proxy
indicator of oral hygiene practice. The presence of calculus has been
related to higher levels of periodontal disease in the individual
(Gilthorpe, Griffiths, Maddick, & Zamzuri, 2000; Timmerman & van
der Weijden, 2006). A variable indicating the extent of calculus
(ratio of all sites with calculus to all examined sites) was also
created.
Socioeconomic variables
Years of education and income, indicated by poverty-income
ratio, were used to indicate socioeconomic position. Years of
education was categorized into three groups: less than 12 years,12
years and more than 12 years. Poverty-income ratio is the ratio
between family income and poverty threshold (Federal poverty
level) and provides a better estimation of income that is compa-
rable throughout the six years of the survey (Sabbah et al., 2007).
Throughout the paper, the term income is used to indicate poverty-
income ratio. This variable was used as a continuous variable in the
regression models but was categorized into quartiles in the
descriptive analysis.
Covariates
Other variables included in the analysis were age in years, sex,
ethnicity (White, African, Mexican Americans and other ethnicities)
and availability of any dental insurance.
Data analysis
Data analysis was conducted using STATA survey commands.
Final sampling weights were used throughout the analysis (Slade &
Beck, 1999). The distributions of each of the health outcomes and
health-related behaviors were assessed within education groups
and income quartiles. To assess education and income disparities in
health-related behaviors, logistic regression models were con-
structed for smoking and frequency of dental visits, and linear
regression was used for frequency of eating fresh fruits and vege-
tables and for the extent of calculus. The first three models adjusted
for education, income, age, sex and ethnicity. The model for
calculus additionally adjusted for dental insurance, smoking and
dental visits, all considered important determinants of calculus
(Gilthorpe et al., 2000; Maizels & Sheiham, 1987).
To examine the effect of health-related behaviors on socioeco-
nomic disparities in oral health, regression models adjusting for
oral health-related behaviors and other covariates were compared
to those notadjusting for behaviors. This methodwouldaccount for
the direct and indirect effect of the explanatory variables (van Oort,
van Lenthe, & Mackenbach, 2005). Appropriate regression models
were conducted for each health outcome, namely linear regression
for the extent of gingival bleeding and extent of loss of periodontal
attachment, logistic regression for perceived oral health and
negative binomial regression for tooth loss. Three sets of regression
analysis were conducted for each health outcome, the first
adjusting for education, income, age, sex, ethnicity, and dental
W. Sabbah et al. / Social Science & Medicine 68 (2009) 298–303 299

Page 3

insurance,
frequency of dental visits, and frequency of eating fresh fruits and
vegetables. The third set additionally adjusted for the extent of
calculus. Since calculus is highly correlated with oral health
outcomes on the one hand, and with the other behaviors on the
other hand, it was added in the final model to allow for a better
estimation of the effect of other behaviors on disparities in oral
health.
thesecondadditionallyadjusting forsmoking,
Results
Overall, 12,051 individuals were included in the descriptive
analysis. Ethnic distribution was 80.1 percent White Americans
(95 percent confidence interval: 77.6, 82.4), 9.7 percent African
Americans (95 percent confidence interval: 8.4, 11.2), 4.2 percent
Mexican Americans (95 percent confidence interval: 3.3, 5.2) and 6
percent other ethnicities (95 percent confidence interval: 4.7, 7.7),
and 48.7 percent males (95 percent confidence interval: 47.7, 50.0).
The mean age was 39.8 years (95 percent confidence interval: 39.0,
40.7). Of the analyzed sample, 55.5 percent had dental insurance
(95 percent confidence interval: 52.2, 58.8). Additionally, 44.3
percent had less than 11 years of education (95 percent confidence
interval: 41.8, 47.0), 34.3 percent had 12 years of education (95
percent confidence interval: 32.6, 35.9) and 21.4 had more than 12
years of education (95 percent confidence interval: 19.6, 23.3). The
overall mean poverty-income ratio was 3.2 (95 percent confidence
interval: 3.0, 3.3). Poverty-income ratio was highest among people
in the top education group (mean: 4.0) (95 percent confidence
interval: 3.8, 4.1) and lowest among people in the bottom education
group (mean: 2.1) (95 percent confidence interval: 1.9, 2.2). Those
who reported having poorer perceived oral health, had higher
levels of gingival bleeding (mean: 12.8), loss of attachment (mean:
16.7) and tooth surface loss (mean: 28.3) compared to means of 7.4,
6.4, and 10.7 respectively in the group reporting good perceived
oral health.
In the whole analyzed sample, 29.2 percent reported being
a current smoker (95 percent confidence interval: 27.4, 30.0). The
percentage of persons reporting dental visits once a year or more
was 54.6 (95 percent confidence interval: 52.0, 57.2). The mean of
the daily frequency of consuming fresh fruits and vegetables was
3.3 (95 percent confidence interval: 3.3, 3.4). The mean of the
extent of calculus was 35.9 (95 percent confidence interval: 33.3,
38.4). The percentage of persons who reported poor or fair oral
health was 32.7 (95 percent confidence interval: 30.8, 34.7). The
means of the extent of gingival bleeding, loss of attachments and
loss of tooth surfaces were 9.4 (95 percent confidence interval: 8.3,
10.5), 9.9 (95 percent confidence interval: 9.2, 10.7) and 27.6 (95
percent confidence interval: 25.8, 29.5), respectively.
Table 1 demonstrates the distribution of oral health indicators
and health-related behaviors by income and education groups. The
fouroral health outcomes wereworse at lower than at higher levels
of education and income (Table 1). A very similar relationship was
found between the selected health-related behaviors by income
and education. That is, there were more unfavorable oral health-
related behaviors among poorer and less educated individuals,
while health enhancing behaviors were more common among the
more educated and more affluent (Table 1).
Table 2 shows the binary and adjusted associations of the four
health-related behaviors used in this study with education and
income. Higher levels of education and income were significantly
associated with higher levels of health-enhancing behaviors and
with lower levels of health-risk behaviors. Even after adjusting for
a number of covariates, education and income remained significant
determinants of behaviors.
The education and income disparities in all oral health indica-
tors persisted after adjusting for health-related behaviors (Table 3).
The extent of gingival bleeding for persons in the middle and
lowest education groups attenuated from 2.3 (95 percent confi-
dence interval: 1.4, 3.1) to 2.0 (95 percent confidence interval: 1.0,
3.0) and from 5.3 (95 percent confidence interval: 3.9, 6.6) to 4.9
(95 percent confidence interval: 3.5, 6.3), respectively after
adjusting for smoking, dental visits and frequency of eating fresh
fruits and vegetables. Additional adjustment for calculus resulted in
a greater attenuation in the extent of gingival bleeding to 1.3 (95
percent confidence interval: 0.4, 2.2) and 3.4 (95 percent confi-
dence interval: 2.0, 4.8) for the middle and lowest education
groups, respectively (Table 3). For each higher unit of income, the
extent of gingival bleeding was smaller by ?1.0 (95 percent confi-
dence interval: ?1.2, ?0.7), after adjusting for the four indicators
of behaviors, the extent of gingival bleeding at each higher unit
of income was smaller by ?0.8 (95 percent confidence interval:
?0.9, ?0.4).
The odds ratios for reporting poorer oral health for persons in
the middle and lowest education groups were 1.8 (95 percent
confidence interval: 1.5, 2.1) and 2.2 (95 percent confidence
interval: 1.8, 2.7), respectively. After adjusting for all indicators of
behaviorstheseoddsratiosattenuatedto1.4(95percentconfidence
interval: 1.2,1.6) and 1.6 (95 percent confidence interval: 1.3,1.9)for
the middle and lowest education groups, respectively (Table 3).
Similar trends were observed in all models for the other oral
health outcomes. The probabilities of oral diseases by income and
education attenuated after adjusting for health-related behaviors,
Table 1
Distribution of health outcomes and behaviors, by education groups and poverty-income ratio’s quartiles (N¼12,051).
EducationIncome indicated by poverty-income ratio
<12 years 12 years
>12 yearsLowest quartile Second lowest
quartile
Second highest
quartile
Highest
quartile
Oral health outcomes
Mean (95%CI) extent of gingival bleeding
Mean (95%CI) extent loss of periodontal
attachment ?3 mm
Mean (95%CI) number of missing
tooth surfaces
Percentage (95%CI) perceived oral
health (poor/fair)
13.4 (12.1, 14.7)
15.5 (14.3, 16.9)
9.5 (8.2, 10.8)
9.9 (8.9, 10.9)
6.7 (5.7, 7.7)
6.6 (5.8, 7.3)
13.5 (11.7, 15.3)
11.9 (10.2, 13.6)
11.9 (10.3, 13.5)
11.1 (9.8, 12.3)
9.2 (7.9, 10.4)
9.3 (8.4, 10.3)
6.7 (5.6, 7.7)
8.6 (7.7, 9.4)
25.4 (23.5, 27.2) 18.3 (17.2, 19.5)9.8 (8.9, 10.8)17.6 (15.4, 19.8) 20.2 (18.3, 22.1) 16.7 (15.1, 18.3)13.5 (12.1, 15.0)
50.0 (47.4, 52.2) 36.2 (33.2, 39.2) 21.8 (20.0, 23.6)50.5 (46.6, 54.4)44.4 (40.3, 48.6)32.4 (29.5, 35.5)23.0 (20.8, 25.4)
Health-related behaviors
Percentage (95%CI) dental visits once a
year or more
Percentage (95%CI) current smoker
Mean (95%CI) frequency of eating fresh
fruits and vegetables/day
Mean (95%CI) extent of calculus
34.2 (30.7, 38.0) 48.9 (45.1, 52.3)69.0 (66.1, 71.6) 26.3 (22.8, 30.1)36.1 (33.0, 39.3)54.2 (50.1, 58.3) 70.7 (67.5, 73.6)
37.6 (34.8, 40.5)
3.0 (2.9, 3.1)
34.7 (31.7, 37.9)
3.1 (3.0, 3.2)
20.8 (18.5, 23.3)
3.6 (3.5, 3.7)
38.2 (35.1, 41.4)
2.9 (2.8, 3.1)
36.2 (33.5, 39.0)
3.1 (3.0, 3.3)
28.7 (26.3, 31.3)
3.2 (3.1, 3.4)
24.0 (21.4, 26.7)
3.5 (3.3, 3.6)
49.0 (46.2, 51.9)37.3 (34.2, 40.5) 26.8 (23.8, 29.7) 47.8 (43.4, 52.3)42.9 (39.7, 46.1)35.9 (32.8, 38.9) 27.9 (24.4, 31.3)
W. Sabbah et al. / Social Science & Medicine 68 (2009) 298–303 300

Page 4

and calculus appeared to have a greater effect on the association
between oral health and socioeconomic indicators. However,
income and education remained significant determinants of all oral
health indicators, even after adjusting for all behaviors, including
calculus. The only exception was for the loss of periodontal
attachment, in the middle education groups (Table 3).
Discussion
In this study we have shown that in a nationally representative
sample of US adults poorer health-related behaviors were more
common among the less educated and the poorer, even after
adjusting for potential covariates. Income and education disparities
in all markers of oral health were attenuated after adjusting for
health-related behaviors, but did not disappear.
Socioeconomic disparities in health-related behaviors observed
in this study confirm findings from previous studies (Davis, 1980;
Jarvis & Wardle, 2006; Locker, 1989; Marmot, 1999). Similarly, the
finding that the effect of adjusting for health-related behaviors
altered, but did not markedly change socioeconomic disparities in
clinical and subjective oral health outcomes is consistent with
findings indicating that health-related behaviors lessen the
disparities in oral health but do not eliminate them (Sanders et al.,
2006b). However, this study has the advantage of using both clin-
ical and subjective indicators of oral health and more accurate
indicators of socioeconomic position than the aforementioned
study (Sanders et al., 2006b).
Others have argued that access to dental care explained most of
the socioeconomic disparities in oral health (Wamala et al., 2006).
This study refutes that viewpoint. When we used models adjusting
for frequency of dental visits along with other behavioral indicators
there were still significant socioeconomic disparities in oral health.
Frequency of dental visits is of particular importance, firstly
because it indicates a recommended health-related behavior, as
some visits are often for check-ups and can be considered
preventive in nature. Second, it is an indicator of utilization of
health services. Even regression models adjusting for two indica-
tors of utilization of health services, namely frequency of dental
visits and availability of dental insurance, still showed significant
socioeconomic disparities in oral health.
It could be argued that calculus is a confounding factor with oral
disease. Here it was used as a marker of oral hygiene behavior
(Maizels & Sheiham, 1987). Cleanliness of teeth, as measured by
plaque and calculus, plays an essential role in periodontal health
(Haffajee et al., 1991; Locker, 1989; Morris et al., 2001) and tooth
loss (Drake, Hunt, & Koch, 1995; Gilbert, Duncan, Crandall, Heft, &
Ringelberg, 1993; Treasure et al., 2001; Ylostalo, Sakki, Laitinen,
Jarvelin, & Knuuttila, 2004). Calculus is also associated with dental
plaque and oral hygiene related behaviors (Riley, Gilbert, & Heft,
2006; Timmerman & van der Weijden, 2006). However, the great
attenuation of socioeconomic disparities in oralhealth in this study,
after adjusting for calculus, should be interpreted with caution.
Nevertheless, despite the strong correlation between calculus and
socioeconomic indicators on the one hand, and oral health indi-
cators on the other, adjusting for calculus did not change the
direction or the significance of the association between socio-
economic and oral health indicators.
The persistence of socioeconomic disparities in all health-
related behaviors even after taking into account a number of
confounders, such as age and ethnicity, implies that the determi-
nants of these disparities are complex. They include factors such as
work related stress, job security, control at the work place and at
Table 2
Socioeconomic disparities in selected health-related behaviors.
Education (reference education >12 years)Higher income
12 years
<12 years
Smoking – odds ratio (95%CI)a
Dental visits – odds ratios (95%CI)b
Eating fresh fruits and vegetables – regression coefficient (95%CI)c
Extent of calculus – regression coefficient (95%CI)d
1.8***(1.4, 2.3)
0.5*** (0.4, 0.6)
?0.5*** (?0.6, ?0.4)
5.4***(3.5, 7.3)
2.2*** (1.8, 2.7)
0.3*** (0.3, 0.4)
?0.7*** (?0.8, ?0.5)
10.6*** (8.0, 13.4)
0.9** (0.8, 0.9)
1.4*** (1.3, 1.5)
0.1* (0.1, 0.1)
?1.2** (?2.0, ?0.5)
***p<0.001; **p<0.01; *p<0.05.
aOdds ratio for being a current smoker, model adjusted for education, income, age, sex and ethnicity.
bOdds ratio for dental visit once a year or more, model adjusted for education, income, age, sex and ethnicity.
cRegression coefficient for frequency of eating fresh fruits (continuous variable), model adjusted for education, income, age, sex and ethnicity at lower education level and
for a higher unit of poverty-income ratio.
dRegression coefficient for extent of calculus (continuous variable), model adjusted for education, income, age, sex, ethnicity, dental insurance, smoking and dental visits.
Table 3
Effects of adjustment for health-related behaviors on the socioeconomic disparities in oral health.
Education (reference education >12 years) Higher income
12 years
<12 years
Gingival bleeding – regression coefficient (95%CI)Model 1
Model 2
Model 3
Model 1
Model 2
Model 3
Model 1
Model 2
Model 3
Model 1
Model 2
Model 3
2.3*** (1.4, 3.1)
2.0*** (1.0, 3.0)
1.3** (0.4, 2.2)
3.0*** (1.9, 4.0)
1.7** (0.6, 2.8)
0.9NS(?0.2, 1.9)
1.8*** (1.5, 2.1)
1.5*** (1.2, 1.7)
1.4*** (1.2, 1.6)
2.0*** (1.7, 2.3)
1.8*** (1.5, 2.1)
1.7*** (1.4, 2.0)
5.3*** (3.9, 6.6)
4.9*** (3.5, 6.3)
3.4*** (2.0, 4.8)
8.2*** (6.8, 9.6)
6.5*** (5.1, 8.0)
4.8*** (3.3, 6.3)
2.2*** (1.8, 2.7)
1.8*** (1.5, 2.1)
1.6*** (1.3, 1.9)
2.3*** (1.8, 2.8)
1.9*** (1.6, 2.3)
1.7*** (1.4, 2.0)
?1.0*** (?1.2, ?0.7)
?0.8*** (?1.1, ?0.5)
?0.6*** (?0.9, ?0.4)
?0.8*** (?1.1, ?0.4)
?0.5** (?0.8, ?0.2)
?0.3* (?0.6, ?0.1)
0.8*** (0.7, 0.9)
0.9*** (0.8, 0.9)
0.9*** (0.8, 0.9)
0.9*** (0.8, 0.9)
0.9** (0.9, 0.9)
0.9** (0.9, 0.9)
Loss of periodontal attachment – regression coefficient (95%CI)
Perceived oral health (poor/fair) – odds ratios (95%CI)
Number of missing tooth surfaces – count rate ratios (95%CI)
Model 1: adjusting for education, income, age, sex, ethnicity, and dental insurance.
Model 2: adjusting for education, income, age, sex, ethnicity, dental insurance, smoking, frequency of visits to a dentist, and frequency of eating fresh fruits and vegetables.
Model 3: adjusting for education, income, age, sex, ethnicity, dental insurance, smoking, frequency of visits to a dentist, frequency of eating fresh fruits and vegetables and
extent of calculus.
***p<0.001, **p<0.01, *p<0.05.
W. Sabbah et al. / Social Science & Medicine 68 (2009) 298–303 301

Page 5

home, neighborhood characteristics, and ability to benefit from
recent knowledge of health enhancing factors (Abegg, Marcenes,
Croucher, & Sheiham, 1999; Brunner, 2002; Jarvis & Wardle, 2006;
Phelan & Link, 2005) and operate beyond the individual level. The
results also indicate the crucial importance of the education and
income.
A similar lack of a significant effect of health-related behaviors
on disparities in oral health (Sanders et al., 2006b) and general
health (Lantz et al., 2006) has been reported. The present analysis
reinforces these findings and also has the advantage of using
several clinical and subjective indicators of oral health in a repre-
sentative sample of the US population. The findings suggest that
changing individuals’ health-related behaviors is unlikely to
markedly affect socioeconomic disparities in oral health. Marmot
(2005) has argued that ‘‘if the major determinants of health are
social, so must be the remedies’’. If it is possible for some people to
achieve low levels of morbidity, then it is possible to achieve
similarly low morbidity in all groups. Hence, there is a need to
develop preventive policies from a more social and structural view
of the determinants of health (Wilkinson,1996). The findings of this
analysis suggest that reducing disparities in oral health necessitates
developing strategies which look beyond the proximal causes of
the disease and address the underlying determinants of oral health
and related behaviors (Watt, 2007).
Since it was based on a cross-sectional study, the results of the
present analysis cannot be used to draw conclusions about causal
relationships. The lack of data on specific oral health-related
behaviors such as tooth brushing and the use of calculus as
a surrogate marker of oral hygiene and related behaviors are other
limitations of this analysis. Additionally, the study did not examine
the effect of behaviors on disparities in dental caries, which should
be examined in future research. The lack of data on water fluori-
dation which could have influenced inequalities in tooth loss
examined in this study, is another limitation of the current analysis.
The lack of oral health specific behaviors, such as tooth-brushing, in
the NHANES necessitates the collection of these data in future
NHANES survey and calls for further research adjusting for these
behaviors.
The data used in this analysis was from a relatively old survey
(1988–1994). However, socioeconomic disparities still exist in oral
health in the US (Borrell & Crawford, 2008) despite possible
changes in health literacy. On the other hand, there is no evidence
on changes in the provision of dental services in US. This implies
that the findings of this study on the effect of behaviors on socio-
economic disparities in oral health in an old surveyare still relevant
today.
In conclusion, this study demonstrated the presence of socio-
economic disparities in health-related behaviors, similar to those
found in subjective and clinical oral health outcomes. Adjusting for
health-related behaviors attenuated but did not eliminate the
socioeconomic disparities in oral health. The findings indicate that
health-related behaviors explain part of the socioeconomic
disparities in oral health. The results of this study imply that the
determinants of oral health disparities are more complex than
explained by the proximal determinants such as health-related
behaviors. The findings suggest that oral health policies which aim
at changing behaviors are unlikely to completely eliminate dispar-
ities in oral health (Sanders et al., 2006b; Watt, 2007).
References
Abegg, C., Marcenes, W., Croucher, R., & Sheiham, A. (1999). The relationship
between tooth cleaning behaviour and flexibility of working time schedule.
Journal of Clinical Periodontology, 26(7), 448–452.
Borrell, L., & Crawford, N. (2008). Social disparities in periodontitis among United
States adults 1999–2004. Community Dentistry and Oral Epidemiology, 36(5),
383–391.
Brunner, E. (2002). Stress mechanisms in coronary heart disease. In S. Stansfeld, &
M. Marmot (Eds.), Stress and the heart: Psychosocial pathways to coronary heart
disease (pp. 181–199). London: BMJ Books.
Burt, B. A., Kolker, J. L., Sandretto, A. M., Yuan, Y., Sohn, W., & Ismail, A. I. (2006).
Dietary patterns related to caries in a low-income adult population. Caries
Research, 40(6), 473–480.
Chen, M. (1995). Oral health of disadvantaged populations. In L. Cohen, & H. Gift
(Eds.), Disease prevention and oral health promotion: Socio-dental science in
action (pp. 153–212). Copenhagen: Munksgaard.
Davis, P. (1980). Epidemiology, social structure and the public health tradition. The
social context of dentistry. London: Croom Helm Ltd. pp. 61–68.
Drake, C. W., Hunt, R. J., & Koch, G. G. (1995). Three-year tooth loss among black and
white older adults in North Carolina. Journal of Dental Research, 74(2), 675–680.
Gilbert, G., Duncan, R., Crandall, L., Heft, M., & Ringelberg, M. (1993). Attitudinal and
behavioral characteristics older floridians with tooth loss. Community Dentistry
and Oral Epidemiology, 21(6), 384–389.
Gilthorpe, M., Griffiths, G., Maddick, I., & Zamzuri, A. (2000). The application of
multilevel modelling to periodontal research data. Community Dental Health,
17(4), 227–235.
Gochman, D. S. (1982). Labels, systems, and motives: some perspectives on future
research. Health Education Quarterly, 9, 167–174.
Haffajee, A., Socransky, S., Lindhe, J., Kent, R., Okamoto, H., & Yoneyama, T. (1991).
Clinical risk indicators for periodontal attachment loss. Journal of Clinical Peri-
odontology, 18(2), 117–125.
Jarvis, M., & Wardle, J. (2006). Social patterning of individual health behaviours: the
case of cigarette smoking. In M. Marmot, & R. Wilkinson (Eds.), Social deter-
minants of health (pp. 224–337). Oxford: Oxford University Press.
Kivimaki, M., Lawlor, D., Davey Smith, G., Kouvonen, A., Virtanen, M., Elovainio, M.,
et al. (2007). Socioeconomic position, co-occurrence of behaviour risk factors,
and coronary heart disease: the Finnish Public sector study. American Journal of
Public Health, 97(5), 874–879.
Lantz, P., House, J., Lepkowski, J., Williams, D., Mero, R., & Chen, J. (2006). Socio-
economic factors, health behaviors, and mortality. Results from a nationally
representative prospective study of US adults. Journal of American Medical
Association, 279(21), 1703–1708.
Locker, D. (1989). Behavioural science and dentistry. London: Routledge.
Locker, D. (2000). Deprivation and oral health: a review. Community Dentistry and
Oral Epidemiology, 28(3), 161–169.
Lopez, R., Fernandez, O., & Baelum, V. (2006). Social gradients in periodontal disease
among adolescents. Community Dentistry and Oral Epidemiology, 34(3),184–196.
Ma, J., Betts, N. M., & Hampl, J. S. (2000). Clustering of lifestyle behaviours: the
relationship between cigarette smoking, alcohol consumption and dietary
intake. Am J Health Promot, 15, 107–117.
Maizels, A., & Sheiham, A. (1987). A new measure of teeth-cleaning efficiency and
periodontal disease. Journal of Clinical Periodontology, 14(2), 105–109.
Marmot, M. (1999). The social pattern of health and disease. In D. Blane, E. Brunner,
& R. Wilknson (Eds.), Health and social organization (pp. 42–67). London:
Routledge.
Marmot, M. (2005). Social determinants of health inequalities. Lancet, 365(9464),
1099–1104.
McKinlay, J. (1993). The promotion of health through planned sociopolitical change:
challenges for research and policy. Social Science & Medicine, 36(2), 109–117.
Miller, A., Brunelle, J., Carlos, J., Brown, L., & Lo ¨e, H. (1987). Oral health of the United
States adults. The national survey of oral health in US employed adults and seniors,
1985–86. Bethesda, MD: National Institute of Dental Research. NIH publication
number 87-2868.
Morris, A., Steele, J., & White, D. (2001). The oral cleanliness and periodontal health
of UK adults in 1998. A guide to the UK adult dental health survey 1998. London:
The British Dental Association. pp. 47–53.
Ojima, M., Hanioka, T., Tanaka, K., & Aoyama, H. (2007). Cigarette smoking and tooth
loss experience among young adults: a national record linkage study. BMC
Public Health, 7, 313. doi:10.1186/1471-2458-7-313, Available from. http://www.
biomedcentral.com/1471-2458/7/313.
van Oort, F., van Lenthe, F., & Mackenbach, J. (2005). Material, psychosocial, and
behavioural factors in the explanation of educational inequalities in mortality in
the Netherlands. Journal of Epidemiology and Community Health, 59(3), 214–220.
Petersen, P. E., Jiang, H., Peng, B., Tai, B. J., & Bian, Z. (2008). Oral and general health
behaviours among Chinese urban adolescents. Community Dentistry and Oral
Epidemiology, 36(11), 76–84.
Phelan, J. C., & Link, B. G. (2005). Controlling disease and creating disparities:
a fundamental cause perspective. [Special issue II]. Journals of Gerontology, 60B,
27–33.
Pine, C., Adair, P., Petersen, P., Douglass, C., Burnside, G., Nicoll, A., et al. (2004).
Developing explanatory models of health inequalities in childhood dental
caries. Community Dental Health, 21(Suppl. 1), 86–95.
Riley, J., Gilbert, G., & Heft, M. (2006). Dental attitudes: proximal basis for oral
health disparities in adults. Community Dentistry and Oral Epidemiology, 34(4),
289–298.
Sabbah, W., Tsakos, G., Chandola, T., Sheiham, A., & Watt, R. G. (2007). Social
gradients in oral and general health. Journal of Dental Research, 86(10), 992–996.
Sanders, A., Slade, G., Turrell, G., Spencer, A., & Marcenes, W. (2006a). The shape of
the socioeconomic-oral health gradient: implications for theoretical explana-
tions. Community Dentistry and Oral Epidemiology, 34(4), 310–319.
Sanders, A., Spencer, A., & Slade, G. (2006b). Evaluating the role of dental behaviour
in oral health inequalities. Community Dentistry and Oral Epidemiology, 34(1),
71–79.
W. Sabbah et al. / Social Science & Medicine 68 (2009) 298–303 302

Page 6

Sheiham, A. (2000). Improving oral health for all: focusing on determinants and
conditions. Health Education Journal, 59, 351–363.
Sheiham, A., & Watt, R. G. (2000). The common risk factor approach: a rational basis
for promoting oral health. Community Dentistry and Oral Epidemiology, 28(6),
399–406.
Slade, G., & Beck, J. (1999). Plausibility of periodontal disease estimates from
NHANES III. Journal of Public Health Dentistry, 59(2), 67–72.
Syme, L. (1996). To prevent disease: the need for a new approach. In D. Blane,
E. Brunner, & R. Wilkinson (Eds.), Health and social organisation: Towards
a health policy for the 21st century (pp. 21–31). London: Routledge.
Thomson, W., & Mackay, D. (2004). Child dental caries patterns described using
a combination of area-based and household-based socio-economic status
measures. Community Dental Health, 21(4), 285–290.
Timmerman, M., & van der Weijden, G. (2006). Risk factors for periodontitis.
International Journal of Dental Hygiene, 4(1), 2–7, Review.
Treasure, E., Kelly, M., Nuttall, N., Nunn, J., Bradnock, G., & White, D. (2001). Factors
associated with oral health: a multivariate analysis of results from the 1998
adult dental health survey. pp.19–27. A guide to the UK adult dental health survey
1998. London: The British Dental Association.
U.S. Department of Health and Human Services (DHHS). National Centre for Health
Statistics. (1999). Third National Health and Nutrition Examination Survey, 1988–
1994, NCHS CD-ROM Series 11, No 1 Version 1.22a. Public Use Data File Docu-
mentation Number 76200. Hyattsville, MD: Centers for Disease Control and
Prevention.
Wamala, S., Merlo, J., & Bostrom, G. (2006). Inequity in access to dental care services
explains current socioeconomic disparities in oral health: the Swedish national
surveys of public health 2004–2005. Journal of Epidemiology and Community
Health, 60(12), 1027–1033.
Watt, R. G. (2007). From victim blaming to upstream action: tackling the social
determinants of oral health inequalities. Community Dentistry and Oral Epide-
miology, 35(1), 1–11.
Watt, R. G., & Sheiham, A. (1999). Inequalities in oral health: a review of the
evidence and recommendations for action. British Dental Journal, 187(1), 6–12.
Wilkinson, R. (1996). Unhealthy societies: The afflictions of inequality. London/New
York: Routledge.
Ylostalo, P., Sakki, T., Laitinen, J., Jarvelin, M., & Knuuttila, M. (2004). The relation of
tobacco smoking to tooth loss among young adults. European Journal of Oral
Sciences, 112(2), 121–126.
W. Sabbah et al. / Social Science & Medicine 68 (2009) 298–303 303