Community Dent Oral Epidemiol 2017; 45; 74–83 Ó 2016 John Wiley & Sons A/S.

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

Lifecourse socioeconomic Dong-Hun Han1,2 and

Young-Ho Khang3,4
1
Department of Preventive and Social

position indicators and tooth loss Dentistry, Seoul National University School
of Dentistry, Seoul, South Korea, 2Dental
Research Institute, Seoul National
University, Seoul, South Korea, 3Department
in Korean adults of Health Policy and Management, Seoul
National University College of Medicine,
Seoul, South Korea, 4Institute of Health
Policy and Management, Seoul National
University Medical Research Center, Seoul,
Han D-H, Khang Y-H. Lifecourse socioeconomic position indicators and tooth South Korea
loss in Korean adults. Community Dent Oral Epidemiol 2017; 45: 74–83. © 2016
John Wiley & Sons A/S. Published by John Wiley & Sons Ltd.

Abstract – Aim: The aim of this study was to assess the association between
lifecourse socioeconomic position (SEP) indicators and tooth loss in a large
representative sample of the Korean adult population. Methods: Data from the
Fourth and Fifth Korea National Health and Nutritional Examination Survey
on 17,549 Korean adults aged 50 years or older were analyzed. The study
design was cross-sectional. Tooth loss was defined as either edentulism (0
teeth) or severe tooth loss (<20 teeth). Parental education, own education, and
income were used as indicators of early childhood SEP, adolescent SEP, and
adulthood SEP, respectively. The association between the lifecourse SEPs and
tooth loss was estimated with three different lifecourse models: the
independent impact model for three different lifecourse SEPs, the cumulative
impact model using the sum of the three SEP scores, and the social mobility
model using the pathway information of the three SEPs. The covariates were
demographic factors and health behaviors. The prevalence ratios (PRs), slope
index of inequality (SII), and relative index of inequality (RII) for tooth loss
were estimated with log-binomial regression analyses. Results: In the
independent impact model, significant associations of each SEP with
edentulism (SII=3.61%, 4.76%, 2.17% and RII=1.67, 1.52, 1.41 for parental
education, own education, and income, respectively) and severe tooth loss
(SII=8.03%, 17.73%, 12.73% and RII=1.26, 1.72, 1.42 for parental education, own
education, and income, respectively) were found after adjusting for all
covariates. For the cumulative impact model, stepwise patterns of PRs Key words: childhood; epidemiology; Korea;
according to the summed SEP scores were found. In addition, in the social socioeconomic factors; tooth loss
mobility model, the PR of tooth loss was greatest in the persistently low SEP Dong-Hun Han, Department of Preventive
group over the life course (PR=2.10 for edentulism and 2.33 for severe tooth and Social Dentistry, School of Dentistry,
Seoul National University, 101 Daehak-ro,
loss). The other mobile groups in lifecourse SEPs also presented higher PRs of
Jongno-gu, Seoul 110-749, South Korea
tooth loss than the persistently high SEP group. Conclusion: Parental education Tel.: +82-2-740-8780
was independently associated with edentulism in later life, and own education Fax: +82-2-765-1722
showed the strongest associations with tooth loss. The cumulative and social e-mail: dhhan73@snu.ac.kr
mobility models for the social impacts of life course on oral health should be Submitted 30 June 2015
considered in oral epidemiological studies. accepted 7 September 2016

Poor oral health is a major global health burden1–4,
and tooth loss is associated with future morbidity
and mortality5–7. Poor oral health in adulthood is
reportedly linked with socioeconomic position
(SEP) at different life stages8–13. Increasingly,
research on health in later life has employed life-
course approaches14,15. Several lifecourse models
have been suggested to examine the impacts of
early- and later-life conditions and their trajecto-
ries on health in later life16–18. The socioeconomic
condition at one time point over the life course
might have independent and long-lasting conse-
quences on later-life health (hereafter the ‘indepen-
dent impact model’). According to cumulative

74 doi: 10.1111/cdoe.12262

advantage and disadvantage perspectives19, social
inequalities in health start early in life and increase
with age as the initial advantages or disadvantages
accumulate across the life course (hereafter the ‘cu-
mulative impact model’). In addition, early-life
effects could be modified by later-life circum-
stances. Upwardly or downwardly, mobile indi-
viduals could present with different health
outcomes in later life (hereafter the ‘social mobility
model’). These lifecourse models have been used
in examining adulthood and elderly mortality and
mobidity16,20 but have not been commonly used in
examining oral health outcomes8,13,15,21–25.
Several lifecourse studies have examined the rela-
tionship between SEP and tooth loss but provided
mixed results. The Dunedin Multidisciplinary
Health and Development Study in New Zealand
found that a low SEP in early childhood and social
mobility was associated with tooth loss at 26 years
of age13. On the other hand, the Newcastle Thou-
sand Families Study in the United Kingdom
reported that parental social class as an indicator of
early social condition was not related with tooth
retention at 50 years of age21,22. One Finnish study8
suggested that edentulousness in adult life could be
more influenced by own education rather than par-
ental education. Prior investigations into the rela-
tionship between lifecourse SEP indicators and
tooth loss were largely carried out in the West. To
the best of our knowledge, no study has been car-
ried out that investigates the relationship between
lifecourse SEP and tooth loss among an Asian pop-
ulation. The association of early childhood SEP
indicators with tooth loss was examined in a prior
paper26; however, this study extended the research
questions to include the cumulative impact model
and social mobility model in the Republic of Korea
(hereafter ‘Korea’). The aim of this study was to
assess the association between different lifecourse
SEP indicators and tooth loss in a large representa-
tive cross-sectional dataset of the Korean popula-
tion: The Fourth and Fifth Korea National Health
and Nutritional Examination Survey (KNHANES
IV & V) between 2007 and 2012. We examined the
aforementioned lifecourse models for tooth loss: the
independent impact model, cumulative impact
model, and social mobility model.
Materials & methods
The conceptual framework of this study is pre-
sented in Figure 1. This study was a cross-sectional
Lifecourse socioeconomic position and tooth loss
Fig. 1. Conceptual framework for the association
between parental education and tooth.
approach to a lifecourse analysis. Parental SEP and
own SEP (own education and income) were main
interest of independent variables while tooth loss
was an outcome variable. Health behaviors were
considered as mediators in the relationship
between independent and outcome variables. Own
SEPs were also considered as mediators in the
association of parental SEP with tooth loss.
We explored any direct (independent) effects of
parental and own SEP indicators on tooth loss and
examined cumulative impact and social mobil-
ity models with different sets of lifecourse SEP
indicators.
The data consisted of a subset of data from the
Fourth and Fifth KNHANES. The KNHANES, con-
ducted periodically by the Korea Centers for Dis-
ease Control and Prevention (KCDC), provides
comprehensive data on sociodemographic factors,
pre-existing health problems, bio-clinical risk fac-
tors, health behaviors, and nutritional status each
year from about 200 primary sampling units in
Korea. The target population consists of nationally
representative noninstitutionalized civilians in
Korea. Each survey year includes a new sample of
about 10,000 participants aged 1 year and over.
The survey used stratified multistage probability
sampling units based on geographic area, gender,
and age. The final sample set for the KNHANES
included 21,400 households. KNHANES consists
of several component surveys: a health examina-
tion, health interview, and nutrition survey in the
Fourth and Fifth KNHANES27. The overall
response rates were from 65.8% to 82.8% during
KNHANES IV & V. Each participant provided
written informed consent. This survey was
approved by the KCDC Institutional Review
Board.
For this analysis, the data were combined from
separate annual surveys between 2007 and 2012
among adults aged 50 years or older. Because the
prevalence of edentulism was very low (0.02%)
among those who were younger than 50 years,
they were excluded from our analyses. The num-
ber of total participants aged 50 years or older in
KNHANES IV & V was 18,681 (7,991 men and
10,690 women). Of those, 17,549 individuals (7,459
75
Han & Khang
men and 10,090 women) were examined for the
number of teeth. Considering the missing values in
the health examination or questionnaires, multiple
imputation was used.
The KNHANES oral health examination was
performed by calibrated dentists with the partici-
pant seated on a dental chair with a light, mouth
mirror, and a World Health Organization (WHO)
periodontal probe. In addition, WHO criteria were
used when assessing oral health status28. Dentist
calibration training was conducted each year with
the results compared to a reference dentist. The
mean (range) kappa values for interexaminer relia-
bilities for the dental status measures by tooth were
from 0.711 to 0.95129–34. The number of teeth for
each participant was enumerated with the highest
number of teeth set to 28 excluding the wisdom
teeth. Edentulism was defined as no teeth in the
mouth. Because participants with at least 20 teeth
can be considered as having functional dentition,
severe tooth loss was defined as having less than
20 natural teeth excluding wisdom teeth.
Parental educational levels (father’s and
mother’s education) were considered as the child-
hood SEP measures20 and derived based on the fol-
lowing questions used at the interview: ‘What is
your father’s education level when you were
14 years old?’ and ‘What is your mother’s educa-
tion level when you were 14 years old?’ There
were eight possible responses: unknown or no
response; no education; Chinese classic (traditional
primary school for Chinese characters); primary
school; middle school; high school; college; gradu-
ate school or more. The education level of both par-
ents was ascertained, respectively. The education
level of the father was used to indicate parental
education considering the stronger association of
the education level of the father with tooth loss in
the elderly population in South Korea than the
education level of the mother26. The education
level of the mother was used if the father was not
present when the participant was 14 years old or
when data were missing on the status of the father.
When we only used the education level of the
father as the early childhood SEP, similar findings
were produced (data not shown here). Parental
education was then divided into four groups for
each age group, considering the rapid growth in
education levels in Korea35: I=no, II=Chinese clas-
sic, III=primary school (≤ 6 years), IV=more than
middle school (≥ 7 years) for 50- to 59-year-olds;
I=unknown, II=no, III=Chinese classic, IV=more
than primary school (≥ 6 years) for +60-year-olds.
76
An indicator for adolescent SEP was own edu-
cation. The own education level of participants
can be seen as the own SEP measure for early
exposure36. The own education level of partici-
pants was determined for four categories: Own
education I=primary school, II=middle school
(7-9 years), III=high school (10-12 years), IV=more
than college (≥ 13 years) for 50- to 59-year-olds;
I=no, II=primary school, III=middle school,
IV=more than high school (≥ 10 years) for +60-
year-olds. Because education levels among Kore-
ans have rapidly increased over the past dec-
ades35, the distribution of education level differed
greatly with generations. Therefore, we also cre-
ated four education levels for own education for
each age group.
Own income is a classic adulthood SEP mea-
sure37. In this study, monthly household income
was adjusted for the number of household mem-
bers and categorized into four groups: < 25%, 25–
49%, 50–75%, and > 75% of the total equalized
income (household income / [household size]0.5)
in each annual survey. Occupation was not used in
this study because many participants aged 50 and
older were not in the labor market.
Tooth loss can be seen as a terminal outcome of
multifactorial processes involving biological path-
ways as well as social and behavioral pathways
related with i) a SEP such as low education38, low
economic status39, and rural residence40; and ii)
health behaviors including cigarette smoking41,
toothbrushing habit, and infrequent dental
visits42. Demographic factors included age, gen-
der, survey year, and residence. The residence of
the participants was presented as urban versus
rural. Health behaviors were cigarette smoking,
frequency of tooth brushing, and recent dental
visits. Current smoking was defined as smokers
who smoked one or more packs per day. The fre-
quency of tooth brushing was divided into two
groups: less than two times a day and two or
more times a day brushing. Participants were
asked whether they had visited a dental clinic for
a regular check-up during the year prior to the
interview (response categories: yes or no). In
this study, health behaviors (current smoking,
toothbrushing, and regular dental visit) were
considered as mediators in the association
between SEP and tooth loss, while age, gender,
and place of residence were considered as
confounders.
All analyses were performed with SAS statistical
software (SAS Institute, Inc., Cary North Carolina)
 Lifecourse socioeconomic position and tooth loss

after taking into account the complex sample
designs including primary sampling units, stratifi-
cation, and sample weights from the KNHANES.
In Table 1, the characteristics of the study partici-
pants according to tooth loss associated preva-
lence. Relative index of inequality (RII) and slope
index of inequality (SII) for education presented in
Tables 2 and 3 were based on relative educational
position indicators ranging 0-1 for each age group.
This relative educational position indicator was
determined by calculating the relative position in
the cumulative population distribution of the cen-
tral subject in each group of age-specific educa-
tional hierarchy. For example, if 20% of the sample
fell into the highest education group in a particular
stratum, the group was assigned a relative rank of
0.10 (0.0 + 0.5 * 0.20). If the next highest educa-
tional group accounted for 30% of the sample, a
relative rank of 0.35 (0.20 + 0.5 * 0.30) was
assigned. The relative educational position variable
was then entered as an independent variable in our
regression analyses. Thus, the RII for education is
the relative risk of tooth loss at the lowest end
(age-specific relative educational rank = 1) of the
educational hierarchy as compared with the risk at
the top (age-specific relative educational rank = 0)
of the educational hierarchy, weighted by the pop-
ulation distribution of individuals across

Table 1. The numbers of total study participants by sociodemographic variables, health behaviors, and underlying
health problems and associated prevalences of edentulism and severe tooth loss: Findings from the Fourth and Fifth
Korea National Health and Nutrition Examination Survey
Edentulism, Severe tooth
Total N (%) N (%) Pa loss, N (%) Pa
Gender Male 7459 (%) 395 (4.3) 0.001 2505 (29.4) <0.001
Female 10090 633 (5.6) 3574 (33.0)
Parental education I (lowest) 3471 242 (5.7) <0.001 1240 (32.0) <0.001
II 4028 360 (8.2) 1852 (44.0)
III 4220 184 (3.6) 1220 (25.9)
IV (highest) 5559 194 (2.9) 1616 (26.0)
Own education I (lowest) 4133 401 (8.3) <0.001 1857 (41.2) <0.001
II 5899 378 (5.6) 2399 (37.1)
III 3623 100 (2.2) 811 (20.4)
IV (highest) 3623 101 (2.5) 861 (21.2)
Income I (lowest) 4262 317 (6.0) 0.002 1714 (35.8) <0.001
II 4314 255 (4.8) 1580 (33.2)
III 4283 230 (4.7) 1417 (29.6)
IV (highest) 4258 180 (3.9) 1182 (25.3)
Sum of SEP I (lowest) 1741 159 (7.0) <0.001 751 (38.5) <0.001
II 1920 184 (7.8) 875 (40.3)
III 2268 177 (6.9) 943 (37.9)
IV 2432 148 (5.3) 921 (34.9)
V 2375 102 (3.7) 766 (29.5)
VI 2178 71 (2.9) 647 (27.0)
VII 1744 47 (2.2) 417 (20.7)
VIII (highest) 2230 55 (2.1) 446 (17.3)
Path of SEP Low–low–low 3413 328 (7.8) <0.001 1586 (41.9) <0.001
Low–low–high 2104 185 (8.3) 901 (40.2)
Low–high–low 753 29 (3.1) 232 (27.2)
Low–high–high 1013 34 (3.3) 269 (25.3)
High–low–low 2358 137 (4.7) 930 (36.0)
High–low–high 1887 96 (4.7) 713 (34.5)
High–high–low 1940 60 (2.7) 484 (23.1)
High–high–high 3420 74 (1.7) 651 (16.3)
Place of residence Urban 11900 516 (3.9) <0.001 3500 (27.3) <0.001
Rural 5649 512 (8.0) 2579 (42.4)
Current smoking No 12576 657 (5.2) <0.001 4444 (35.3) <0.001
Yes 4695 322 (6.9) 1947 (41.5)
Toothbrushing < 2 times/day 3669 425 (10.5) <0.001 1767 (44.4) <0.001
≥ 2 times/day 13707 574 (3.5) 4225 (27.8)
Regular dental visit No 13546 923 (5.9) <0.001 5037 (33.7) <0.001
Yes 3830 76 (1.5) 954 (22.9)
a
P values are obtained from chi-square tests.

77
Han & Khang

Table 2. Association of individual socioeconomic position indicators with edentulism and severe tooth loss: slope index
of inequality (SII, %), relative index of inequalities (RIIs), and their 95% confidence intervals (CI)
Edentulism Severe tooth loss
SII (%, 95% CI) RII (95% CI) SII (%, 95% CI) RII (95% CI)
a
Model 1
Parental education 6.66 (0.89, 12.43) 2.06 (1.58, 2.69) 6.43 (4.27, 8.60) 1.31 (1.20, 1.43)
Own education 7.83 (1.18, 14.48) 2.01 (1.50, 2.71) 10.73 (8.63, 12.83) 1.66 (1.82, 1.52)
Income 3.78 ( 2.24, 9.80) 1.57 (1.22, 2.03) 9.35 (7.20, 11.51) 1.47 (1.35, 1.60)
Model 2b
Parental education 3.61 ( 3.30, 10.52) 1.67 (1.28, 2.17) 8.03 (4.81, 11.24) 1.26 (1.16, 1.38)
Own education 4.76 ( 2.87, 12.40) 1.52 (1.12, 2.05) 17.73 (14.16, 21.31) 1.72 (1.57, 1.89)
Income 2.17 ( 4.95, 9.29) 1.41 (1.08, 1.84) 12.73 (9.16, 16.30) 1.42 (1.31, 1.54)
a
Adjusted for survey year, age, gender, and place of residence.
b
Adjusted for survey year, age, gender, place of residence, smoking, toothbrushing habit, and regular dental visit.
Bold denotes statistical significance at p<0.05.

Table 3. Association of mutually adjusted socioeconomic position indicators with edentulism and severe tooth loss:
slope index of inequality (SII, %), relative index of inequalities (RIIs), and their 95% confidence intervals (CI)
Edentulism Severe tooth loss
SII (%, 95% CI) RII (95% CI) SII (%, 95% CI) RII (95% CI)
a
Model 1
Parental education 3.34 (-3.70, 10.38) 1.66 (1.25, 2.21) 2.64 ( 1.16, 6.44) 1.09 (1.00, 1.19)
Own education 5.13 (-2.76, 13.03) 1.51 (1.11, 2.07) 14.18 (10.14, 18.22) 1.52 (1.38, 1.68)
Income 1.78 (-5.14, 8.71) 1.36 (1.04, 1.77) 8.66 (4.84, 12.48) 1.34 (1.23, 1.46)
Model 2b
Parental education 1.77 (-5.95, 9.48) 1.48 (1.12, 1.96) 2.17 ( 1.70, 6.03) 1.08 (0.99, 1.18)
Own education 3.76 (-4.92, 12.44) 1.24 (0.90, 1.71) 14.44 (10.06, 18.82) 1.47 (1.34, 1.63)
Income 0.97 (-6.56, 8.49) 1.30 (0.99, 1.71) 8.22 (4.29, 12.15) 1.32 (1.21, 1.43)
a
Adjusted for survey year, age, gender, place of residence, parental education, own education, and income.
b
Adjusted for survey year, age, gender, place of residence, parental education, own education, income, smoking, tooth-
brushing habit, and regular dental visit.
Bold denotes statistical significance at p<0.05.

educational groups. The SII is the absolute version
of this summary measure. For income, similar rela-
tive income position indicator was used for RII and
SII. These inequality measures have been used
widely to summarize the effect of ordinal SEP indi-
cators (education and income) on health out-
come43. SII and RII were estimated by employing a
log-binomial regression using PROC GENMOD of
SAS. Missing data in confounding variables were
imputed with the Markov Chain Monte Carlo
method, which assumes that the variables with
missing data are multivariate normal and missing
at random.
In this study, a series of SIIs and RIIs for tooth
loss were estimated according to parental educa-
tion, own education, and income after adjusting for
covariates. Model 1 initially adjusted for survey
year, age, gender, and place of residence; next,
Model 2 additionally adjusted for health behaviors
such as smoking, tooth brushing habit, and recent
dental visits (Table 2). Then, we mutually adjusted
for all the SEP indicators to examine whether they
were independently associated with tooth loss in
the simultaneously adjusted models (Table 3).
These analyses (Tables 2 and 3) explored the inde-
pendent role of each SEP indicator in predicting
tooth loss (independent impact model), after taking
into account indirect effects via mediators. To
assess the cumulative nature of the SEP indicators
in predicting tooth loss, we summed up the scores
for each SEP variable coded as 1=the lowest SEP
and 4=the highest SEP. Log-binomial regression
analyses using PROC GENMOD in the SAS statisti-
cal software were conducted to estimate the preva-
lence ratio (PR) taking into account the complex
design (strata and sampling weights). We esti-
mated the PRs for the summed scores (range: 3-12)
for the SEP for the cumulative impact model
(Table 4). Finally, we examined the social mobility
model (Figure 2). Each SEP indicator was

78
 Lifecourse socioeconomic position and tooth loss

(sum = 11~12)
Table 4. Prevalence ratios (95% confidence intervals) of edentulism and severe tooth loss according to cumulative scores of lifecourse socioeconomic position (SEP) indica-
categorized into two groups (1 and 2=low, and 3
and 4=high). Eight possible trajectories were cre-
ated using parental education, own education, and
income and categorized into persistently high (ref-

1.00
1.00

1.00
1.00
VIII
erence group), upwardly mobile (low–high–high,
low–low–high), downwardly mobile (high–high–
low, high–low–low), fluctuated (high–low–high,
VII (sum = 10)

1.09 (0.69-1.73)
1.01 (0.63-1.63)

1.27 (1.05-1.55)
1.27 (1.04-1.54)
low–high–low), and persistently low group (low–
low–low). For each trajectory, the PR was esti-
mated.
1.33 (0.86-2.06)
1.18 (0.75-1.85)

1.72 (1.42-2.07)
1.68 (1.39-2.02)

Results
VI (sum = 9)

Table 1 shows graded patterns of the prevalence of

tooth loss according to own education and income.
Significant differences in edentulism and severe
tooth loss according to the covariates including
Adjusted for survey year, age, gender, place of residence, smoking, toothbrushing habit, and regular dental visit.

demographic factors and health behaviors were

1.35 (0.89-2.07)
1.12 (0.73-1.74)

1.73 (1.44-2.08)
1.65 (1.37-1.98)

found. We also found differences in the mean age

V (sum = 8)

according to the tooth loss status (61.9 versus

73.7 years in no edentulous group versus edentu-
lous group and 59.5 versus 69.1 years in no severe
tooth loss group versus severe tooth loss group; all
P < 0.05 by t-tests).
1.81 (1.18-2.78)
1.46 (0.94-2.27)

2.03 (1.69-2.44)
1.94 (1.61-2.34)

Table 2 presents the analysis results when an

IV (sum = 7)

individual SEP (parental education, own educa-

tion, and income) indicator was solely included in
the regression analysis. Parental educational, own
educational, and income inequalities in tooth loss
existed among Korean adults aged 50+ years old;
2.07 (1.38-3.12)
1.64 (1.07-2.50)

2.18 (1.81-2.63)
2.05 (1.70-2.49)

for example, the SIIs in the prevalence of eden-

III (sum = 6)

tulism based on parental education, own educa-

tion, and income were 3.61%, 4.76%, and 2.17%,
Adjusted for survey year, age, gender, and place of residence.

respectively. The SIIs in the prevalence of severe

tooth loss based on parental education, own educa-
tion, and income were 8.03%, 17.73%, and 12.73%
tors (parental education, own education, and income)

2.49 (1.63-3.79)
1.94 (1.26-3.01)

2.50 (2.07-3.03)
2.32 (1.91-2.83)

respectively, demonstrating greater prevalence in

II (sum = 5)

the lowest SEP group than the highest group; and

Bold denotes statistical significance at p<0.05.

the RIIs indicated that a move from the highest to

the lowest parental education, own education, and
income groups were associated with a 167%, 152%,
and 141% increases in the prevalence of eden-
tulism, respectively. The RIIs of parental educa-
2.63 (1.74-3.99)
1.95 (1.26-3.01)

2.89 (2.36-3.54)
2.65 (2.15-3.26)
I (sum = 3~4)

tion, own education, and income groups were 1.26,

1.72, and 1.42 in the prevalence of severe tooth loss,
respectively. The analysis results in Table 2 also
indicate that health behaviors modestly explained
Severe tooth loss

the relationships of individual SEP with tooth loss.

Appendix Table 1 shows the results on the PRs for
Edentulism

Model 2b

Model 2b
Model 1a

Model 1a

tooth loss when the original categories of individ-

ual SEP rather than the relative SEP variables (0-1)
were included in the analysis.
b
a

79
Han & Khang

a b

c d

Fig. 2. Prevalence ratios (PRs) for edentulism and severe tooth loss according to trajectories of three lifecourse socioeco-
nomic position (SEP) indicators (parental education, own education, and income). Reference was HHH. The vertical
lines represent the 95% confidence interval. Model 1 was adjusted for survey year, age, gender, and place of residence (a
and c. Model 2 adjusted for survey year, age, gender, place of residence, smoking, toothbrushing habit, and regular den-
tal visit (b and d).

Table 3 presents the associations of parental
education, own education, and income with tooth
loss in the mutually adjusted models. This analysis
explores any independent impact of each individ-
ual SEP indicator on tooth loss by simultaneously
adjusting for all three SEP indicators (independent
impact model). The SIIs of parental education, own
education, and income were 1.77%, 3.76%, and
0.97% for edentulism and 2.17%, 14.44%, and
8.22% for severe tooth loss in the fully adjusted
models, respectively. RIIs of parental education,
own education, and income were 1.48, 1.24, and
1.30 for edentulism and 1.08, 1.47, and 1.32 for sev-
ere tooth loss in the fully adjusted models, respec-
tively. Compared with the SIIs and RIIs in the
individually adjusted models for each SEP indica-
tor (Table 2), all corresponding SIIs and RIIs were
attenuated in the simultaneously adjusted models.
Especially, the magnitude of the attenuation was
noticeable in the SII and RII of severe tooth loss for
parental education. Similar to the findings in
Table 2, health behaviors partly explained the
association of lifecourse SEPs with tooth loss.
Appendix Table 2 presents the results of the simul-
taneously adjusted analysis on the PRs of tooth loss
by the original categories of lifecourse SEP indica-
tors. The PRs of severe tooth loss for parental
education were not statistically significant while all
the PRs for the other SEP indicators were statisti-
cally significant.
Table 4 presents the PRs of tooth loss according
to the cumulative scores of lifecourse SEP indica-
tors (cumulative impact model). The PRs of tooth
loss were higher in those with lower summed SEP
scores (PR of edentulism=1.95 and PR of severe
tooth loss=2.65 for the lowest level of the summed
SEP scores in the fully adjusted model). In these
cumulative models, health behaviors had a modest
role in explaining the associations of the summed
SEP scores with tooth loss while additional
explanatory roles of the health problems were not
evident.
The final model, the social mobility model, with
eight possible SEP trajectories are shown in Fig-
ure 2. The PRs of tooth loss for the trajectories of
the three SEPs were the highest in the persistently
low SEP group after controlling for all covariates
(2.10 for edentulism and 2.33 for severe tooth loss).
An interesting finding was that the associations of
edentulism with exposures to two or more consec-
utive low SEPs remained statistically significant in
the fully adjusted model. In addition, the associa-
tions between severe tooth loss and two or more
consecutive exposures to socioeconomic

80

disadvantage showed relatively high PRs. Incon-
secutive exposures to two low SEPs (fluctuating
exposures to low–high–low) showed relatively
smaller PRs of tooth loss than consecutive expo-
sures to two low SEPs (upward or downward
social mobility such as low–low–high or high–
low–low).
Discussion
This study is an extension of a prior paper26 exam-
ining the relationship between early-life indicators
of socioeconomic position (the education level of
the father and mother) with tooth loss. However,
in this paper, we created cumulative and pathway
indicators of lifecourse SEP and examined different
lifecourse models. We also used relative SEP indi-
cators to estimate the RII and produce the PR while
the prior paper used adjusted odds ratios26. The
number of study participants was larger in this
study because only elderly participants (aged 65+)
were included in the prior study26. Missing data in
confounding variables were also imputed in this
paper.
The findings of this lifecourse study showed that
low childhood, adolescent and adulthood socioeco-
nomic circumstances measured by parental educa-
tion, own education, and income had negative
influences on adulthood and elderly tooth loss.
Each individual lifecourse SEP indicator had an
independent association with edentulism and sev-
ere tooth loss when all three SEP indicators were
simultaneously adjusted for (independent impact
model). The cumulative impact model showed
graded patterns of PRs according to the summed
SEP scores. In addition, the risk of tooth loss was
greatest in the persistently low SEP group over the
life course in the social mobility model. Moreover,
exposures to two or more consecutive low SEPs
were detrimental to oral health in later life.
Prior studies have examined the associations
between childhood SEP indicators and oral health
outcomes8,13,21–26. However, only a few studies
have examined the impacts of parental and own
education levels on tooth loss8,26,39. A prior Korean
study on an elderly population from the
KNHANES showed that the education level of the
father rather than the education level of the mother
was independently associated with the edentate
status among elderly people after adjusting for
adulthood SEP indicators26. According to a Swed-
ish study, low educational attainment was
Lifecourse socioeconomic position and tooth loss
associated with a higher risk of tooth loss39. A pre-
vious Finnish study reported that parental educa-
tion and own education were associated with
edentulousness in adulthood8. This study pre-
sented a graded relationship between the summed
scores of cumulative SEP indicators and tooth loss.
A prior study showed a similar finding8.
In this study, the magnitude of the relationship
measured by the RIIs (Tables 2 and 3) and PRs
(Appendix Tables 1 and 2) between parental educa-
tion and edentulousness was as great as the magni-
tude of the relationships for own education and
income. However, the RIIs and PRs of severe tooth
loss (having less than 20 teeth) for parental educa-
tion were relatively smaller than those for own edu-
cation and income. In the simultaneously adjusted
models using the original SEP categories, the PRs of
severe tooth loss for parental education were not
statistically significant (Appendix Table 2). These
findings suggest that early-life socioeconomic influ-
ences on edentulism and severe tooth loss might be
different. One the other hand, a prior study showed
that edentulism in adult life was more influenced
by own education rather than parental education8.
This study showed that the risks of tooth loss
were relatively high among those exposed to two
or more consecutive low socioeconomic disadvan-
tages (low–low–high, high–low–low, or low–low–
low SEP exposure) during their life course while
two inconsecutive exposures (low–high–low expo-
sure) showed relatively low PRs. This pattern was
generally true for both edentulism and severe tooth
loss. This finding suggests that edentulism and
severe tooth loss require long-lasting exposures to
socioeconomic disadvantages during lifetime.
The findings of this study showed that health
behaviors modestly explained the relationships of
lifecourse SEP indicators and oral health outcomes.
This was true for the individually adjusted model
(Table 2) and the simultaneously adjusted model
(Table 3) and generally true for both outcome mea-
sures (edentulism and severe tooth loss). This find-
ing is expected because the health behaviors
considered in this study (cigarette smoking, tooth-
brushing habit, and regular dental visit) are impor-
tant predictors of oral health outcomes and have
been reported to be strongly associated with SEP
indicators44.
This study has strengths. First, we examined
three different lifecourse models (independent
impact model, cumulative impact model, and social
mobility model) with respect to oral health out-
comes. Only a few studies on oral health used
81
Han & Khang
multiple lifecourse models8,15,22,23,25. In addition, a
large nationally representative sample with fairly
good response rates was analyzed. The clinical
examination of missing teeth by dentists is another
strong point rather than self-reported data about
missing teeth. This study also has limitations. We
relied on recalled data on parental education for the
early childhood SEP. A prior study showed that the
use of historical records on early-life SEP rather
than retrospective recall data produced a greater
magnitude of association between early-life social
disadvantages with adulthood health outcomes45.
Thus, the impact of childhood SEP on tooth loss
might have been underestimated. In addition, early
childhood SEP indicators were limited to parental
education, and other indicators such as parental
occupation or living standards in early years were
not included in this study. Moreover, the cross-sec-
tional design prohibited us from inferring causal
relationships. Further well-designed prospective
investigations are required to determine the causal-
ity between early-life SEP and tooth loss to reduce
the above mentioned limitations.
In summary, the findings of this study showed
that socioeconomic disparities in tooth loss exist in
South Korea. Lifecourse SEP indicators measured
by parental education, own education, and income
were all independently associated with tooth loss
in a nationally representative sample. This study
also showed that tooth loss can be seen as a result
of cumulative exposures to social disadvantages
over the life course (cumulative impact model). In
addition, two or more consecutive exposures to
such disadvantages were important to create
greater likelihoods of having tooth loss (social
mobility model). This study provides further evi-
dence that lifecourse social influences are salient in
chronic diseases such as tooth loss. A recent study
showed that high-quality early childhood pro-
grams have substantial benefits in preventing dis-
ease and promoting adult health46. However, well-
designed, prospective interventions for oral health
are required to produce better evidence for the
causality between early-life environment and tooth
loss, to understand the mechanism of tooth loss
and to promote oral health in a population.
Acknowledgements
The authors wish to express their special appreciation to
all the examiners and the participants in KNHANES.
The authors declare no conflict of interest.
82
Conflict of interest and sources of
funding statement
None of the authors have a conflict of interest in
relation to this study.
References
1. Williams DM. Global oral health inequalities: The
research agenda. J Dent Res 2011;90:549–51.
2. Kassebaum NJ, Bernabe E, Dahiya M, Bhandari B,
Murray CJ, Marcenes W. Global Burden of Severe
Tooth Loss: A Systematic Review and Meta-analysis.
J Dent Res 2014;93:20S–8S.
3. Kassebaum NJ, Bernabe E, Dahiya M, Bhandari B,
Murray CJ, Marcenes W. Global burden of severe
periodontitis in 1990-2010: a systematic review and
meta-regression. J Dent Res 2014;93:1045–53.
4. Kassebaum NJ, Bernabe E, Dahiya M, Bhandari B,
Murray CJ, Marcenes W. Global Burden of Untreated
Caries: A Systematic Review and Metaregression. J
Dent Res 2015;94:650–8.
5. Schwahn C, Polzer I, Haring R, Dorr M, Wallaschof-
ski H, Kocher T, et al. Missing, unreplaced teeth and
risk of all-cause and cardiovascular mortality. Int J
Cardiol 2013;167:1430–7.
6. Tu YK, Galobardes B, Smith GD, McCarron P, Jef-
freys M, Gilthorpe MS. Associations between tooth
loss and mortality patterns in the Glasgow Alumni
Cohort. Heart 2007;93:1098–103.
7. Watt RG, Tsakos G, de Oliveira C, Hamer M. Tooth
loss and cardiovascular disease mortality risk–results
from the Scottish Health Survey. PLoS ONE 2012;7:
e30797.
8. Bernabe E, Suominen AL, Nordblad A, Vehkalahti
MM, Hausen H, Knuuttila M, et al. Education level
and oral health in Finnish adults: Evidence from dif-
ferent lifecourse models. J Clin Periodontol
2011;38:25–32.
9. Poulton R, Caspi A, Milne BJ, Thomson WM, Taylor
A, Sears MR, et al. Association between children’s
experience of socioeconomic disadvantage and adult
health: A life-course study. Lancet 2002;360:1640–5.
10. Sabbah W, Tsakos G, Chandola T, Sheiham A, Watt
RG. Social gradients in oral and general health. J
Dent Res 2007;86:992–6.
11. Sanders AE, Slade GD, Turrell G. John Spencer A,
Marcenes W. The shape of the socioeconomic-oral
health gradient: Implications for theoretical explana-
tions. Community Dent Oral Epidemiol 2006;34:310–9.
12. Sanders AE, Spencer AJ. Childhood circumstances,
psychosocial factors and the social impact of adult
oral health. Community Dent Oral Epidemiol
2005;33:370–7.
13. Thomson WM, Poulton R, Milne BJ, Caspi A,
Broughton JR, Ayers KM. Socioeconomic inequalities
in oral health in childhood and adulthood in a birth
cohort. Community Dent Oral Epidemiol
2004;32:345–53.
14. Kuh D, Shlomo YB. A life course approach to chronic
disease epidemiology. Oxford: Oxford University
Press, 2004; 4–14.

15. Schuch HS, Peres KG, Do LG, Peres MA. Can socioe-
conomic trajectories during the life influence peri-
odontal disease occurrence in adulthood?
Hypotheses from a life-course perspective. Med
Hypotheses 2015;84:596–600.
16. Hallqvist J, Lynch J, Bartley M, Lang T, Blane D. Can
we disentangle life course processes of accumula-
tion, critical period and social mobility? An analysis
of disadvantaged socio-economic positions and
myocardial infarction in the Stockholm Heart Epi-
demiology Program. Soc Sci Med 2004;58:1555–62.
17. Rosvall M, Chaix B, Lynch J, Lindstrom M, Merlo J.
Similar support for three different life course socioe-
conomic models on predicting premature cardiovas-
cular mortality and all-cause mortality. BMC Public
Health 2006;6:203.
18. Pudrovska T, Anikputa B. Early-life socioeconomic
status and mortality in later life: an integration of
four life-course mechanisms. J Gerontol B Psychol
Sci Soc Sci 2014;69:451–60.
19. Dannefer D. Cumulative advantage/disadvantage
and the life course: Cross-fertilizing age and social
science theory. J Gerontol B Psychol Sci Soc Sci
2003;58:S327–37.
20. Galobardes B, Lynch JW, Davey Smith G. Childhood
socioeconomic circumstances and cause-specific
mortality in adulthood: Systematic review and inter-
pretation. Epidemiol Rev 2004;26:7.
21. Pearce MS, Steele JG, Mason J, Walls AW, Parker L.
Do circumstances in early life contribute to tooth
retention in middle age? J Dent Res 2004;83:562–6.
22. Pearce MS, Thomson WM, Walls AW, Steele JG. Life-
course socio-economic mobility and oral health in
middle age. J Dent Res 2009;88:938–41.
23. Brennan DS, Spencer AJ. Income-based life-course
models of caries in 30-year-old Australian adults.
Community Dent Oral Epidemiol 2015;43:262–71.
24. Tsuboya T, Aida J, Kawachi I, Katase K, Osaka K.
Early life-course socioeconomic position, adult
work-related factors and oral health disparities:
cross-sectional analysis of the J-SHINE study. BMJ
Open 2014;4:e005701.
25. Listl S, Watt RG, Tsakos G. Early life conditions,
adverse life events, and chewing ability at middle
and later adulthood. Am J Public Health 2014;104:
e55–61.
26. Han DH, Khang YH, Choi HJ. Association of paren-
tal education with tooth loss among Korean Elders.
Community Dent Oral Epidemiol 2015;43:489–99.
27. Kweon S, Kim Y, Jang MJ, Kim Y, Kim K, Choi S,
et al. Data resource profile: The Korea National
Health And Nutrition Examination Survey
(KNHANES). Int J Epidemiol 2014;43:69–77.
28. Ainamo J, Barmes D, Beagrie G, Cutress T, Martin J,
Sardoinfirri J. Development of the World Health Orga-
nization (WHO) community periodontal index of
treatment needs (CPITN). Int Dent J 1982;32:281–91.
29. Kim JB. Standardization for oral health survey in
KNHANES(2007). Seoul: Korea Centers for Disease
Control & Prevention; 2008 Jan; p. 11
30. Park DY. Standardization for oral health survey in
KNHANES(2008). Seoul: Korea Centers for Disease
Control & Prevention, 2009 May; p. 57–65.
Lifecourse socioeconomic position and tooth loss
31. Park DY. Standardization for oral health survey in
KNHANES(2009). Seoul: Korea Centers for Disease
Control & Prevention; 2010 May. p. 53–6, 76-79.
32. Park DY.Standardization for oral health survey in
KNHANES(2010). Seoul: Korea Centers for Disease
Control & Prevention; 2011 May; p. 54–8, 77-80.
33. Park DY. Standardization for oral health survey in
KNHANES(2011). Osong: Korea Centers for Disease
Control & Prevention; 2012 Feb; p. 52–6, 75-79
34. Park DY. Standardization for oral health survey in
KNHANES(2012). Osong: Korea centers for disease
control & prevention; 2013 Feb; p. 49–53, 74.
35. Sorensen CW. Success and education in South Korea.
Comparative Edu Rev 1994;38:10–35.
36. Davey Smith G, Hart C, Hole D, MacKinnon P, Gillis
C, Watt G, et al. Education and occupational social
class: Which is the more important indicator of mor-
tality risk? J Epidemiol Community Health
1998;52:153–60.
37. Galobardes B, Shaw M, Lawlor DA, Lynch JW.
Davey Smith G. Indicators of socioeconomic position
(part 1). J Epidemiol Community Health 2006;60:7–
12.
38. Mendes DC, Poswar Fde O, de Oliveira MV, Haikal
DS, da Silveira MF, Martins AM, et al. Analysis of
socio-demographic and systemic health factors and
the normative conditions of oral health care in a pop-
ulation of the Brazilian elderly. Gerodontology
2012;29:e206–14.
39. Wennstrom A, Ahlqwist M, Stenman U, Bjorkelund
C, Hakeberg M. Trends in tooth loss in relation to
socio-economic status among Swedish women, aged
38 and 50 years: repeated cross-sectional surveys
1968-2004. BMC Oral Health 2013;13:63.
40. Vargas CM, Yellowitz JA, Hayes KL. Oral health sta-
tus of older rural adults in the United States. J Am
Dent Assoc 2003;134:479–86.
41. Arora M, Schwarz E, Sivaneswaran S, Banks E.
Cigarette smoking and tooth loss in a cohort of older
Australians: The 45 and up study. J Am Dent Assoc
2010;141:1242–9.
42. Dogan BG, Gokalp S. Tooth loss and edentulism in
the Turkish elderly. Arch Gerontol Geriatr 2012;54:
e162–6.
43. Harper S, Lynch J.Methods for Measuring Cancer
Disparities: Using Data Relevant to Healthy People
2010 Cancer-Related Objectives. NCI Cancer Surveil-
lance Monograph Series, Number 6. Bethesda, MD:
National Cancer Institute, 2005. 40–1 p. NIH Publica-
tion No. 05-5777.
44. Oh IH, Lee G, Oh CM, Choi KS, Choe BK, Choi JM,
et al. Association between the physical activity of
Korean adolescents and socioeconomic status. J Prev
Med Public Health 2009;42:305–14.
45. Kauhanen L, Lakka HM, Lynch JW, Kauhanen J.
Social disadvantages in childhood and risk of all-
cause death and cardiovascular disease in later life: a
comparison of historical and retrospective childhood
information. Int J Epidemiol 2006;35:962–8.
46. Campbell F, Conti G, Heckman JJ, Moon SH, Pinto
R, Pungello E, et al. Early childhood investments
substantially boost adult health. Science 2014;343:
1478–85.
83