Received: 5 August 2016 | Accepted: 28 February 2017

DOI: 10.1111/cdoe.12299

ORIGINAL ARTICLE

The association of subjective orthodontic treatment need

with oral health-related quality of life

Lea Kragt1,2 | Vincent Jaddoe2,3,4 | Eppo Wolvius1,2 | Edwin Ongkosuwito1,2

1
Department of Oral & Maxillofacial
Surgery, Special Dental Care and Abstract
Orthodontics, Erasmus University Medical Objectives: The existing body of evidence reports an inconsistent association
Centre, Rotterdam, The Netherlands
2 between subjective and objective orthodontic treatment need. The concept of oral
Department of The Generation R Study
Group, Erasmus University Medical Centre, health-related quality of life (OHRQoL) might help to explain the differences in sub-
Rotterdam, The Netherlands
jective and objective orthodontic treatment need. Our aim was to investigate the
3
Department of Epidemiology, Erasmus
University Medical Centre, Rotterdam, association of subjective orthodontic treatment with OHRQoL in children.
The Netherlands Methods: This cross-sectional study was embedded in the Generation R Study, a
4
Department of Paediatrics, Erasmus
population-based prospective cohort study. OHRQoL and subjective orthodontic
University Medical Centre, Rotterdam,
The Netherlands treatment need were assessed by parental questionnaires. Questionnaire items were
individually compared among children with no, borderline and definite subjective
Correspondence
Lea Kragt, Department of Oral & orthodontic need. The association between subjective orthodontic treatment need
Maxillofacial Surgery, Special Dental Care
and OHRQoL was investigated in multivariate regression analysis with weighted
and Orthodontics, Erasmus University
Medical Centre, Rotterdam, The least squares. Differences by sex and levels of objective orthodontic treatment need
Netherlands.
were evaluated.
Email: l.kragt@erasmusmc.nl
Results: In total, 3774 children were included in the analysis. Children with borderline
Funding information subjective orthodontic treatment need and those with definite subjective orthodontic
Erasmus Medical Center, Rotterdam;
Erasmus University, Rotterdam; the treatment need had significantly poorer OHRQoL based on the fully adjusted model
Netherlands Organization for Health (adjusted regression coefficient (ab)= 0.49, 95% CI: 0.75, 0.30; (ab)= 1.58, 95%
Research and Development, Grant/Award
Number: VIDI 016.136.361; European CI: 1.81, 1.58, respectively). The association between subjective orthodontic treat-
Research Council, Grant/Award Number: ment need and OHRQoL was stronger in girls than in boys and stronger in children
ERC-2014-CoG-64916
with objective orthodontic treatment need than in those with none.
Conclusions: Oral health-related quality of life is poorer in children with subjective
orthodontic treatment need. This has not been investigated before in such a large-
population-based study and clearly offers an explanation for the lack of concurrence
between objective and subjective orthodontic treatment need.

KEYWORDS
dental health perceptions, oral health-related quality of life, orthodontics, public health

1 | INTRODUCTION The existing body of evidence shows a highly inconsistent associa-

tion between subjective and objective orthodontic treatment need.3
In 2013, a Dutch oral health report stated that 60% of young adults The concept of oral health-related quality of life (OHRQoL) was
have had orthodontic treatment.1 Reasons for providing orthodontic introduced in the orthodontic literature to help understand differences
treatment are based on prevention of oral diseases and improvement in subjective and objective orthodontic treatment need.4,5 Quality-of-
2
of aesthetics. The need for orthodontic treatment comes either life measures assess the impact of health on social, emotional and
subjectively from the patient or objectively from the care provider. functional aspects of life.6 OHRQoL measures the particular impact of

Community Dent Oral Epidemiol. 2017;17. wileyonlinelibrary.com/journal/cdoe 2017 John Wiley & Sons A/S. | 1
Published by John Wiley & Sons Ltd
2 | KRAGT ET AL.

oral conditions in terms of oral symptoms, functional limitations, emo- University Medical Centre Rotterdam (MEC-2012-165). Participating
tional and social well-being on daily life.7 Thus, OHRQoL measures parents have given written informed consent before the data
aim to capture subjective oral health in a more standardized way, so collection in children had started (n=7393). Information on childrens
8
that they can augment traditional measures of oral health. Naturally, OHRQoL was given by the parents of 3796 children (51.3%), of whom
various oral disorders influence OHRQoL. Whereas many studies have 3774 (51.0%) also provided subjective orthodontic treatment need.
focused on the association between objective orthodontic treatment Subjective orthodontic treatment need was assessed in parental
need and OHRQoL, the association between OHRQoL and subjective questionnaires with the question: Do you think your child needs
orthodontic treatment need has rarely been investigated. However, braces? The response to the question was given by the mothers on
this is of particular importance as treatment decisions are often for a 5-point Likert scale from strongly disagree to strongly agree. For
a big part influenced by what patients and their parents want. the analysis, subjective orthodontic treatment need was categorized
In the literature, objective orthodontic treatment need is into: No subjective orthodontic treatment need for children whose
assessed using clinical oral health features, such as with the dental mothers strongly or somewhat disagreed with the statement.
health component (DHC) of the index of orthodontic treatment need Borderline subjective orthodontic treatment need for children whose
(IOTN), or based on aesthetic impairments, such as with the IOTN mothers did not agree but also did not disagree with the statement
aesthetic component (AC). Studies on the association between and Definite subjective orthodontic treatment need for children
objective orthodontic treatment need and OHRQoL have shown whose mothers somewhat or strongly agreed with the statement.
weak and inconclusive associations between objective orthodontic Oral health-related quality of life was measured with the COHIP-
treatment need and OHRQoL.5,9-12 Subjective orthodontic treatment ortho.19 The COHIP-ortho is a questionnaire addressed to parents
need has been inconsistently assessed in a small number of existing measuring OHRQoL of the child with 11 questions, covering differ-
studies. Some studies used OHRQoL as a surrogate for subjective ent domains, including social-emotional well-being, functional well-
13,14
orthodontic treatment need. However, OHRQoL can be distin- being and school and peer interaction (Appendix Table S1).19 These
guished from subjective need, as OHRQoL is a dynamic concept that questions were answered on a 5-point Likert scale (never, almost
results from the interaction between health, social and contextual never, sometimes, fairly often, almost all the time). The responses
factors.7 Also, different studies have equated aesthetic impairment scored from 1 to 5 and were finally summed for each individual. The
13-15
and subjective orthodontic treatment need. However, there is total overall score of the COHIP-ortho ranges from 0 to 55, and
little evidence for this assumption and it might be wrong, because, higher scores correspond to higher OHRQoL. Missing values in the
for example, having a worse IOTN-AC score does not implicitly mean responses to the OHRQoL questionnaire (COHIP-ortho) were
having more perceived treatment need. In addition, dental attractive- replaced by the personal mean score of the remaining answers to
ness, which can be one of the reasons for subjective orthodontic the questions, as proposed by researchers who used the original ver-
treatment need, is not necessarily associated with OHRQoL.13,14,16,17 sion of the COHIP.20 If more than 30% of the answers were missing,
Although never evaluated, still the children with more aesthetic the participant was excluded from the analysis.
impairment might show a stronger association between subjective The association between childrens subjective orthodontic treat-
orthodontic treatment need and OHRQoL than children with less ment need and OHRQoL is most likely influenced by other factors,
aesthetic impairment. In summary, little is known about the associa- and so the following parental characteristics were considered as
tion between subjective orthodontic treatment need and OHRQoL. covariates: maternal educational level (low, high), household income
Accordingly, the aim of this study was to quantify the association (<2000, 2000-3200, >3200) and ethnicity (Dutch, other Western,
between subjective orthodontic treatment need, not assessed by an non-Western) as indicators for social economic status (SES), and the
objective index but a simple question, and OHRQoL. In particular, following childrens characteristics were considered as covariates:
we were interested in whether subjective orthodontic treatment sex, age and objective orthodontic treatment need. Objective
need in children is associated with poorer OHRQoL independent of orthodontic treatment need was assessed with the DHC and AC of
their objective orthodontic treatment need. The secondary aim of the IOTN. The IOTN was assessed from photographic and radio-
this study was to see whether the association between subjective graphic records of the children (median [90% range] age 9.78 [9.49-
orthodontic treatment need and OHRQoL varied by sex or different 10.45]). Assessment of the IOTN on a combination of photographic
degrees of objective orthodontic treatment need. and radiographic records has been validated previously.21 After
6 months, 10% of the photographs were reassessed to calculate the
intrarater reliability (linear weighted =0.84).
2 | METHODS Statistical analyses used Statistical Package for Social Sciences
(IBM SPSS statistics) version 21, SPSS Inc, Chicago, IL, USA. Charac-
This cross-sectional study was embedded in the Generation R Study, teristics of the participants were summarized and stratified by
a population-based prospective cohort study that previously has sex. Differences between males and females were investigated with
been described in detail.18 The study protocol and its conduct were chi-square tests and Mann-Whitney U tests. Mean scores for the
in accordance with the guidelines of the Declaration of Helsinki and individual items of the COHIP-ortho in the group of unsure and defi-
approved by the Medical Ethical Committee of the Erasmus MC, nite orthodontic treatment need were separately compared with the
KRAGT ET AL. | 3

mean scores for the individual questions of the no subjective

Adjusted only for age and additional for sex in the overall analysis. Adjusted for age, ethnicity, household income, maternal education and additional for sex in the overall analysis. cAdjusted for age, eth-
nicity, household income, maternal education, additional for sex in the overall analysis as well as the stratification on IOTN-DHC and IOTN-AC, and additional for IOTN-DHC and IOTN-AC in the overall
analysis as well as the stratification on sex. dP for trend for the fully adjusted model obtained by treating subjective orthodontic treatment need as continuous term. eObtained from interaction term
0.06)
0.17)
0.17)

1.45)
1.53)

1.38)
orthodontic treatment need group. To evaluate the differences in

AC>5 (n=1499)

0.55 ( 1.03,
0.64 ( 1.12,
0.64 ( 1.11,

1.88 ( 2.31,
1.93 ( 2.33,

1.81 ( 2.24,
the mean item scores between these groups, Cohens effect sizes
were calculated. Following Cohens suggestions, effect sizes of 0.2

<.001
were considered small, 0.5 were considered medium and 0.8 were

Ref
considered large.22 Differences between the groups were evaluated
with the Mann-Whitney U test (P<.05). Furthermore, weighted least
square (WLS) linear regression models were calculated with subjec-

0.21)
0.21)
0.19)

1.19)
1.21)

1.13)
tive orthodontic treatment need as the determinant and the sum-

AC5 (n=2275)

0.49 ( 0.77,
0.47 ( 0.74,
0.45 ( 0.72,

1.50 ( 1.80,
1.50 ( 1.79,

1.44 ( 1.74,
mary score for OHRQoL as the outcome. We used WLS regression

IOTN-AC
T A B L E 1 The association between subjective orthodontic treatment need and OHRQoL analysed overall, stratified by gender and both IOTN components
models, because of the heteroscedasticity in the OHRQoL data. In

<.001
.024
multivariate WLS regression analysis with (potential) confounders

Ref
(childs age and sex (crude model), childs ethnicity and other indica-
tors of socioeconomic status (Model 1) and finally orthodontic char-

entered into the crude model between subjective orthodontic treatment need and gender, resp. IOTN-DHC and IOTN-AC. ref, reference category.
acteristics (Model 2)) were added. The selection of covariates into

1.36)
1.47)

1.41)
0.30 ( 0.77, 0.17)
0.42 ( 0.87, 0.03)
0.42 ( 0.85, 0.02)
the model was based on the current orthodontic literature and sig-

DHC>3 (n=1645)

1.78 ( 2.21,
1.86 ( 2.25,

1.80 ( 2.19,
nificant associations between covariates with both subjective
orthodontic treatment need and OHRQoL. We also performed a test

<.001
for trend analysis by treating the categorized variable (subjective

Ref
orthodontic treatment need) as a continuous term. We tested for
differences in the association of subjective orthodontic treatment

0.30)

0.31)
1.17)

1.16)
need and OHRQoL between girls and boys and children with and

0.60 ( 0.89, 0.31)

0.58 ( 0.85, 0.31)
DHC3 (n=2129)
without objective orthodontic treatment need based on either the

0.57 ( 0.85,

1.45 ( 0.89,
1.46 ( 1.75,

1.45 ( 1.73,
IOTN-DHC or the IOTN-AC by including interactions terms in the
IOTN-DHC

model. For all variables, significant interactions were present

<.001
.039
Ref

(Table 1). Significant differences in the associations between the

strata were evaluated with a test for heterogeneity. For all analyses,
a P value <.05 was considered to be statistically significant.
0.07)
0.17)
0.15)

1.80)
1.86)

1.60)
Missing values for covariates were handled with multiple imputation
by using the Markov Chain Monte Carlo method. Objective orthodontic
Girls (n=1901)

0.42 ( 0.77,
0.51 ( 0.85,
0.49 ( 0.82,

2.13 ( 2.46,
2.18 ( 2.50,

1.93 ( 2.27,
treatment need had the largest amount of missing data (IOTN-AC

<.001
[22.9%], IOTN-DHC [20.3%], Table 2). We generated five independent
Ref

data sets with a fully conditional specified model, and we present the
pooled effect estimates (b (95% confidence intervals) [CI]). Rubins rules
were applied for pooling of the effect estimates.23 We generated five
b
0.27)
0.27)
0.23)

1.12)
1.12)

0.96)

independent data sets because the pooled effect estimates did not
change with more imputations and because based on Rubins rules, the
Boys (n=1873)

0.59 ( 0.90,
0.57 ( 0.86,
0.52 ( 0.82,

1.43 ( 1.74,
1.41 ( 1.71,

1.27 ( 1.58,

relative efficiency of five imputed data sets appeared sufficient, namely

Oral health-related quality-of-life score

<.001
.001

higher than 95.6% in case of 22.9% missing data.23 Imputations were

Sex

Ref

based on the associations between all variables used in this study, but
the main determinant (subjective orthodontic treatment need) and out-
come (OHRQoL) were not imputed.22 Finally, we also conducted a sen-
0.27)
0.30)
0.27)

1.53)

1.34)
1.76 ( 1.98, 1.55)

sitivity analysis in the original data set. The obtained effect estimates (b
[95%]) of the sensitivity analysis were comparable with the pooled
0.51 ( 0.74,
0.53 ( 0.75,
0.49 ( 0.71,

1.76 ( 1.99,

1.58 ( 1.81,

effect estimates on the relation between subjective and objective

(n=3774)

<.001

orthodontic treatment need (Appendix Table S2).

Total

Ref

To evaluate potential selection bias, children with missing data

Borderline need, b (95% CI)

Definite need, b (95% CI)

on OHRQoL and subjective orthodontic treatment need (n=3619)

were compared to those without missing data on OHRQoL and sub-
P for interactione

jective orthodontic treatment need (n=3774). Data on OHRQoL and

Adjustedb

Adjustedb

P for trendd
Adjustedc

Adjustedc

subjective orthodontic treatment need were more often missing in

Crudea

Crudea
No need

children from parents with lower socioeconomic status (for all

socioeconomic indicators P value <.001, Table 2).
a
4 | KRAGT ET AL.

T A B L E 2 Nonresponse analysis (n=7393) T A B L E 3 Characteristics of the study sample by sex (n=3774)

Included Excluded Boys Girls
Characteristics n=3774 n=3619 P value* Characteristics n=1873 n=1901 P value*

Sex Age in years

Boys (%)a 1873 (49.6) 1834 (50.7) Median (range) 9.8 (9.5-10.4) 9.9 (9.5-10.5) .643
a a
Girls (%) 1901 (50.4) 1785 (49.3) .374 Ethnicity (%)
Missing (%)b 0 (0.0) 0 (0.0) Dutch 1278 (68.2) 1295 (68.1)

Age in years Other Western 147 (7.8) 180 (9.4)

Median 9.78 (9.49-10.45) 9.86 (9.56-11.12) <.001 Non-Western 438 (23.4) 413 (21.7) .135
(range) a
Maternal education level
Missing (%)b 0 (0.0) 3252 (89.9)
Low 619 (33.0) 615 (32.3)
Ethnicity (%)
High 1143 (61.0) 1158 (60.9) .782
Dutch (%)a 2573 (68.6) 1619 (47.7)
Household incomea (%)
Other 327 (8.7) 247 (7.3)
<2000 321 (17.1) 298 (15.7)
Western (%)a
2000-3200 552 (29.5) 535 (28.1)
Non-Western (%)a 851 (22.7) 1529 (45.0) <.001
>3200 877 (46.8) 936 (49.2) .230
Missing (%)b 23 (0.6) 379 (10.5)
IOTN-DHCa (%)
Maternal education level
2 578 (30.6) 541 (28.5)
Low (%)a 1234 (34.9) 1191 (52.1)
3 364 (19.4) 392 (20.6)
High (%)a 2301 (65.1) 1093 (47.9) <.001
4 428 (22.9) 443 (23.3)
Missing (%)b 239 (6.3) 1490 (41.2)
5 128 (6.8) 135 (7.1) .449
Household income (%)
IOTN-aca (%)
<2000 (%)a 619 (17.6) 358 (31.8)
5 1045 (55.8) 1018 (53.5)
2000-3200 (%)a 1087 (30.9) 353 (31.3)
>5 394 (21.0) 451 (23.7) .049
>3200 (%)a 1813 (51.5) 415 (36.9) <.001
OHRQoL
Missing (%)b 255 (6.8) 2493 (68.9)
Median (range) 50.0 49.0 <.001
IOTN-DHC (%)
(43.00-53.00) (42.00-53.00)
2 (%)a 1119 (37.2) 632 (37.7)
Subjective treatment need (%)
3 (%)a 756 (25.1) 380 (22.7)
No 565 (30.2) 484 (25.5)
4 (%)a 871 (28.9) 496 (29.6)
Borderline 498 (26.6) 460 (24.2)
5 (%)a 263 (8.7) 168 (10.0) .185
Yes 810 (43.2) 957 (50.3) <.001
Missing (%) 765 (20.3) 1943 (53.7)
*Based on chi square test for categorical variables and t test or Mann-
IOTN-ac (%)
Whitney U test for continuous variables. aMay not add up to 3774, because
5 (%)a 2063 (70.9) 1144 (70.8) of missing values: maternal education: 6.3%; ethnicity: 0.6%; IOTN: 21.9%;
>5 (%)a 845 (29.1) 471 (29.2) .940 household income: 6.7%, IOTN-DHC: 20.2%; IOTN-AC: 22.9%.
Missing (%)b 866 (22.9) 2034 (56.2)
OHRQoL and 1049 (27.8%) did not perceive any subjective orthodontic treat-
Median (range) 50.0 (43.0-53.0) 49.0 (44.3-54.9) ment need. Boys had slightly higher OHRQoL and perceived less
Missing (%)b 0 (0.0) 3597 (99.4) .959 orthodontic treatment need than girls. These differences between
Subjective treatment need (%) boys and girls were significant (P<.001).
No (%)a 1049 (27.8) 1 (11.1) Table 4 shows the mean COHIP-ortho item scores of the chil-
Borderline (%) a
958 (25.4) 2 (22.2) dren with no perceived orthodontic treatment need, borderline per-
Yes (%)a 1767 (46.8) 6 (66.7) .429 ceived orthodontic treatment need and definite perceived
Missing (%)b 0 (0.0) 3610 orthodontic treatment need. Children with borderline perceived
orthodontic treatment need had lower scores than children with no
*Based on chi square test for categorical variables and t test or Mann-
Whitney U test for continuous variables. aPercentage of available data perceived need for the items about crooked teeth, discoloured
within the subgroup. bPercentage of missing data per subgroup. teeth and bleeding gums. Children with definite orthodontic treat-
ment need showed lower scores than children without perceived
3 | RESULTS orthodontic treatment need on all items except pain, bad breath
and attractiveness. Most of the effect sizes were small except for
In Table 3, the characteristics of the study sample are presented. In the item crooked teeth in the borderline perceived and definite
total, 3774 children were included in the final analysis, of whom orthodontic treatment need groups (d=.36, P.001; d=.98, P.001) as
1767 (46.8%) had definite subjective orthodontic treatment need, well as the item anxious in the definite perceived orthodontic
958 (25.4%) were unsure about their orthodontic treatment need treatment need group (d=.34, P.001).
KRAGT ET AL. | 5

T A B L E 4 COHIP-ortho scores by question for children with 95% CI: 0.85, 0.30) than in children with an IOTN-DHC >3 ([ab]
unsure or definite subjective orthodontic treatment need vs no = 0.42, 95% CI: 0.02, 0.85).
subjective orthodontic treatment need (n=3774)
COHIP-ortho mean scores per question (meanstandard
deviation)
4 | DISCUSSION
No Unsure Definite
subject subject Effect subject Effect
Questions need need sizea need sizea Our study findings suggest that subjective orthodontic treatment need is
Pain 4.8 (0.5) 4.8 (0.5) .02 4.8 (0.5) .02 associated with poor OHRQoL. We showed that more subjective
Crooked teeth 4.8 (0.6) 4.5 (0.8) .36** 3.8 (1.2) .98** orthodontic treatment need is associated with poorer OHRQoL in chil-
Discoloured teeth 4.7 (0.7) 4.6 (0.8) .13** 4.5 (0.9) .22** dren with and without objective orthodontic treatment need and that
Bad breath 4.5 (0.8) 4.5 (0.8) .05 4.8 (0.9) .07 this association is stronger in girls than in boys. Considering these
Bleeding gums 4.7 (0.6) 4.7 (0.6) .08* 4.6 (0.8) .22** marked associations, subjective orthodontic treatment need is not solely
Eating foods 4.9 (0.4) 4.9 (0.4) .00 4.8 (0.5) .15* related to objective orthodontic treatment need, but also related to
Anxious 4.8 (0.6) 4.7 (0.6) .05 4.5 (0.9) .34** OHRQoL. And thus, OHRQoL offers an explanation for the lack of con-
Speaking 5.0 (0.1) 5.0 (0.2) .00 5.0 (0.3) .14** currence between objective and subjective orthodontic treatment need.
Bullied 5.0 (0.3) 5.0 (0.2) .04 4.9 (0.4) .15** The main strength of the present study is the large and ethnically
Attractiveness 1.8 (1.2) 1.8 (1.2) .02 1.8 (1.1) .00 diverse study sample obtained from a population-based cohort study,
Pronunciation 5.0 (0.3) 5.0 (0.3) .04 4.9 (0.5) .16** which was designed to be representative for the general population in
a the Netherlands. However, the study findings should also be seen in the
Cohens effect size (d) for differences between either no subjective need
and borderline subjective need or no subjective need and definite subjec- light of several limitations. Nonresponse analysis showed a higher pro-
tive need. P values are based on Mann-Whitney U test for differences in portion of children without information on OHRQoL or subjective
mean scores *.05; **.001. orthodontic treatment need had parents of lower socioeconomic status.
This might have caused selection bias if the association between subjec-
In Table 1, the findings of the regression model for subjective tive orthodontic treatment need and OHRQoL would be different in
orthodontic treatment need and total COHIP scores are shown. In included and excluded participants. However, because we have no infor-
contrast to children without subjective orthodontic treatment need, mation on subjective orthodontic treatment need and OHRQoL in the
children with borderline orthodontic treatment need as well as chil- nonresponding subsample, this is difficult to ascertain. Another draw-
dren with definite subjective orthodontic treatment need had signifi- back of our study is that in this study, OHRQoL and subjective
cant lower total COHIP scores after adjustments for SES and orthodontic treatment need of the children was assessed by asking the
objective orthodontic treatment need (adjusted regression coefficient parents; thus, we assumed that parents are a valid proxy for childrens
(ab)= 0.49, 95% CI: 0.75, 0.30; (ab)= 1.58, 95% CI: 1.81, reports. This assumption was based on several studies that found par-
1.58, respectively). The trend estimates for the association between ents to be good proxies for childrens OHRQoL.24-26 Still, we cannot
subjective orthodontic treatment need and total COHIP scores were exclude an information bias including a social desirability bias. In addi-
significant (P<.001). In the group without subjective orthodontic treat- tion, we also had no information whether children already had started
ment need, girls had generally lower total COHIP scores than boys their orthodontic treatment or not which also might have contributed to
(Appendix Tables S4 and S5). In addition, the effect of definite subjec- an information bias in the main determinant. In the Netherlands, parents
tive orthodontic treatment need on OHRQoL was significantly and dentist start to concern with orthodontic treatment need around
stronger in girls than in boys ([ab]= 1.93, 95% CI: 2.27, 1.60 and the childrens age of nine, but it is rather uncommon that children start
[ab]= 1.27, 95% CI: 1.58, 0.96, respectively, P<.001). their orthodontic treatment so early. However, if they have started they
The associations between subjective orthodontic treatment need were definitely still in orthodontic treatment need, which we assessed
and OHRQoL stratified by objective orthodontic treatment need are and included in the analysis. Furthermore, a limitation of our study is that
also presented in Table 1. After stratification by objective orthodon- the IOTN was assessed from radiographic and photographic records due
tic treatment need based on the IOTN-AC, the association between to logistic reasons in such a large cohort study as the Generation R
subjective treatment need and total COHIP scores was stronger in study. This method is less valid than direct oral examination and might
children with an IOTN-AC >5 for the borderline and the defin- also have introduced some misclassification of participants orthodontic
ite subjective need group than in children with an IOTN-AC 5 treatment need. However, this method has been shown to be suffi-
(P value=.024). Similarly, after stratification by objective orthodontic ciently valid for research.21 Objective orthodontic treatment need was
treatment need based on the IOTN-DHC, the association between assessed with the IOTN. This measure was chosen because it was devel-
definite subjective treatment need and total COHIP scores was oped solely based on the opinion of orthodontists.27 In this way, the
stronger in children with an IOTN-DHC >3 than in children with an analysis would be adjusted only for professional-based objective
IOTN-DHC 3 (P=.039). In contrast, the association between border- orthodontic treatment need. The use of other orthodontic measures
line perceived subjective treatment need and total COHIP score was such as the Dental Aesthetic Index (DAI) might have been problematic,
significantly stronger in children with an IOTN-DHC 3 ([ab]= 0.57, because this index not only covers objective orthodontic treatment
6 | KRAGT ET AL.

need, but also social norms. Subjective orthodontic treatment need and follow-up, whether they were orthodontically treated or not.33 Thus,
OHRQoL are both influenced by social norms. Consequently, the use of although sex differences in oral health research are insufficiently inves-
the DAI to adjust the analysis might have resulted in a weaker associa- tigated yet, it is generally accepted that girls and boys differ in psycho-
tion between subjective orthodontic treatment need and OHRQoL. In logical variables as how they perceive themselves.34 Still, these
line with this, a recent meta-analysis showed that the association different studies suggest that the association between subjective
between objective orthodontic treatment need assessed with the DAI orthodontic treatment need with OHRQoL should be investigated at
and OHRQoL is highly heterogeneous, whereas this association assessed different ages and over time, also with regard to the differences
28
with the IOTN is not. Finally, like in every observational study, our between boys and girls, before valid conclusion can be drawn.
study findings might be affected by residual confounding, although we Our study is of clinical relevance in orthodontics, oral epidemiol-
were able to minimize confounding of the study findings by constructing ogy and community dentistry. The findings contribute to understand-
fully adjusted models including indicators for socioeconomic status and ing the importance of orthodontic treatment for the young
objective orthodontic treatment need. population in terms of quality of life. Our findings give an indication
In agreement with Kok et al.14, we think that aesthetics are limited for why boys might be less compliant with treatment, to be specific
in their ability to reflect subjective need for orthodontic care. For because they have generally higher OHRQoL. In this way, our find-
example, subjective orthodontic treatment need can arise when friends ings can support an effective communication between patient and
wear braces or when the opinion is influenced by the recommendation orthodontist. Furthermore, the regression analysis, in combination
of the dentist. Furthermore, children with a similar dental aesthetic with the item analysis, showed that subjective orthodontic treatment
impairment do not necessarily perceive the same subjective orthodon- need is associated with poorer OHRQoL, first independent of objec-
tic treatment need. Nevertheless, based on the stratification analysis, tive orthodontic treatment need, and second especially affecting
the association between subjective orthodontic treatment need and OHRQoL on the social-emotional and functional domain. Thus,
OHRQoL seemed indeed stronger in children with more dental aes- whereas the provision of orthodontic treatment nowadays is largely
thetic impairment. Next to this, our analysis showed that children with based on oral health factors such as susceptibility to caries or dental
an IOTN-DHC 3 do perceive more impact of borderline subjective trauma or functional problems such as temporomandibular disorders
orthodontic treatment need on OHRQoL than children with an IOTN- or difficulties with chewing, subjective problems such as avoiding to
DHC >3. Most likely, these are the children who are more aware of smile or to speak lie in the social-emotional domain and are those
their dentition and feel more impairment due to minor malocclusions. which were in the present study particularly associated with poorer
In this way, they could be a source for the divergent association OHRQoL and subjective orthodontic treatment need. Therefore, our
between subjective and objective orthodontic treatment need reported findings are also relevant for health education and policy decisions,
3
by other authors. Whether the perceived impairment due to minor especially in representing the patients perspective. Finally, this study
malocclusions is related to conditions like body dysmorphic disorder (as helps understanding the importance of OHRQoL as outcome mea-
suggested by several researchers) might be possible, but is probably sure in the orthodontic practice as well as Health Service Research.
not the case, given that it is a rare condition.12,29 In summary, we conclude that OHRQoL is poorer in children with
The sex differences shown in the present study reflect the dynamic, subjective orthodontic treatment need. This has not been investigated
context-specific character of OHRQoL. Based on the literature, we before in such a large-population-based study and clearly offers an
expected general poorer OHRQoL in girls.13,16 Surprisingly, the associa- explanation for the variability between objective and subjective
tion between definite subjective orthodontic treatment need and OHR- orthodontic treatment need. Further research should not only focus
QoL was stronger in girls, whereas the association between borderline on the association between subjective orthodontic treatment need
subjective orthodontic treatment need and OHRQoL was stronger in and OHRQoL in populations of different ages, but also investigate in
boys. In line with another study, this suggests that females might be more detail the role of personal and environmental factors other than
more conscious about their appearance, but boys might be more aware sex, such as socioeconomic status, on the association between OHR-
of their malocclusions.16,30 At the age of 9, girls might already be more QoL, subjective orthodontic treatment need and malocclusions.
aware of themselves and how they come across, compare themselves
more with their friends and feel more pressure to be like their peers,
ACKNOWLEDGEMENTS
for example by wanting braces. We saw in the sex-specific item analysis
that the effect of subjective treatment need on items about bullying We gratefully acknowledge the contribution of the participants, gen-
and pronunciation was stronger in girls than in boys (Appendix Tables eral practitioners, hospitals, midwives and pharmacies in Rotterdam,
S3 and S4), and those items belong to the peer interaction domain of the Netherlands. The Generation R Study was conducted by the Eras-
OHRQoL.31 Differences between girls and boys regarding the experi- mus Medical Center, Rotterdam, the Netherlands, in close collabora-
ence of oral health and its impacts on OHRQoL have been reported in tion with the School of Law and Faculty of Social Sciences of Erasmus
26
12-years-old children. Another study performed in adults showed University, Rotterdam; the Municipal Health Service, Rotterdam area;
that women perceive both the negative and positive impacts of oral the Rotterdam Homecare Foundation; and the Stichting Trombosedi-
health on OHRQoL more intensely,32 and a recently published study enst & Artsenlaboratorium Rijnmond, Rotterdam. The Erasmus Medi-
found that OHRQoL was worse in girls than in boys after a 3-years cal Center, Rotterdam; the Erasmus University, Rotterdam; and the
KRAGT ET AL. | 7

Netherlands Organization for Health Research and Development 18. Jaddoe VW, van Duijn CM, Franco OH, et al. The Generation R Study:
made the first phase of the Generation R Study financially possible. design and cohort update 2012. Eur J Epidemiol. 2012;27:739-756.
19. Kragt L, Tiemeier H, Wolvius EB, Ongkosuwito EM. Measuring oral
V.W.V.J. received an additional grant from the Netherlands Organiza-
health-related quality of life in orthodontic patients with a short ver-
tion for Health Research and Development (VIDI 016.136.361) and a sion of the Child Oral Health Impact Profile (COHIP). J Public Health
Consolidator Grant from the European Research Council (ERC-2014- Dent. 2015;76:105-112.
CoG-64916). The funders had no role in study design, data collection 20. Geels LMHJ, Prahl-Andersen B. Confirmatory factor analysis of the
dimensions of the child oral health impact profile (Dutch version).
and analysis, decision to publish or preparation of the manuscript. The
Eur J Oral Sci. 2008;116:148-152.
authors declare no potential conflict of interest with respect to the 21. Kragt L, Hermus AM, Wovius EB, Ongkosuwito EM. Three-dimensional
authorship and/or publication of this article. photographs for determining the Index of Orthodontic Treatment Need
in scientific studies. Am J OrthodDentofacial Orthop. 2016;150:64-70.
REFERENCES 22. Cohen J. Statistical Power Analysis for the Behavioral Sciences, 2nd
edn. New York: Academic Press; 2013.
1. Schuller AA, vanKempen IPF, Poorterman JHG, Verrips GHW. Kies 23. Rubin D. Multiple Imputation for Nonresponse in Surveys. United
Voor Tanden: Een Onderzoek Naar Mondgezondheid En Preventief States: John Wiley & Sons, Inc.; 1987.
Tandheelkundig Gedrag Van Jeugdigen. Leiden: Netherlands Organi- 24. Wilson-Genderson M, Broder HL, Phillips C. Concordance between
zation for Applied Scientific Research TNO; 2013. caregiver and child reports of childrens oral health-related quality of
2. Proffit WR, Fields HW, Ackerman JL, Bailey LT, Tulloch CJF. The life. Community Dent Oral Epidemiol. 2007;35:32-40.
orthodontic problem. In: Rudolf P, ed. Contemporary Orthodontics, 3rd 25. Jokovic A, Locker D, Stephens M, Kenny D, Tompson B, Guyatt G.
edn. London, United Kingdom: Elsevier Health Sciences; 2006:16-23. Measuring parental perceptions of child oral health-related quality of
3. Livas C, Delli K. Subjective and objective perception of orthodontic life. J Public Health Dent. 2003;63:67-72.
treatment need: a systematic review. Eur J Orthod. 2013;35:347-353. 26. Bos A, Hoogstraten J, Zentner A. Perceptions of Dutch orthodontic
4. Mandall NA, Wright J, Conboy FM, OBrien KD. The relationship patients and their parents on oral health-related quality of life. Angle
between normative orthodontic treatment need and measures of Orthod. 2010;80:367-372.
consumer perception. Community Dent Health. 2001;18:3-6. 27. Brook PH, Shaw WC. The development of an index of orthodontic
5. Zhang M, McGrath C, Hagg U. The impact of malocclusion and its treatment priority. Eur J Orthod. 1989;11:309-320.
treatment on quality of life: a literature review. Int J Paediatr Dent. 28. Kragt L, Dhamo B, Wolvius EB, Ongkosuwito EM. The impact of maloc-
2006;16:381-387. clusions on oral health-related quality of life in children-a systematic
6. Locker D, Allen F. What do measures of oral health-related quality review and meta-analysis. Clin Oral Investig. 2015;20:1881-1894.
of life measure? Community Dent Oral Epidemiol. 2007;35:401-411. 29. Anthony MT, Farella M. Body dysmorphic disorder and orthodon-
7. Sischo L, Broder HL. Oral health-related quality of life: what, why, ticsan overview for clinicians. Aust Orthod J. 2014;30:208-213.
how, and future implications. J Dent Res. 2011;90:1264-1270. 30. Peres KG, Barros AJD, Anselmi L, Peres MA, Barros FC. Does maloc-
8. McGrath C, Broder H, Wilson-Genderson M. Assessing the impact of clusion influence the adolescents satisfaction with appearance? A
oral health on the life quality of children: implications for research cross-sectional study nested in a Brazilian birth cohort. Community
and practice. Community Dent Oral Epidemiol. 2004;32:81-85. Dent Oral Epidemiol. 2008;36:137-143.
9. Liu Z, McGrath C, Hagg U. Associations between orthodontic treat- 31. Broder HL, McGrath C, Cisneros GJ. Questionnaire development:
ment need and oral health-related quality of life among young adults: face validity and item impact testing of the Child Oral Health Impact
does it depend on how you assess them? Community Dent Oral Epi- Profile. Community Dent Oral Epidemiol. 2007;35:8-19.
demiol. 2011;39:137-144. 32. Bedi R, McGrath CPJ. Gender variations in the social impact of oral
10. Liu Z, McGrath C, Hagg U. The impact of malocclusion/orthodontic health. J Ir Dent Assoc. 2000;46:87-91.
treatment need on the quality of life: a systematic review. Angle 33. Benson PE, Daas T, Johal A, et al. Relationships between dental
Orthod. 2009;79:585-591. appearance, self-esteem, socio-economic status, and oral health-
11. Dimberg L, Lennartsson B, Bondemark L, Arnrup K. Oral health- related quality of life in UK schoolchildren: a 3-year cohort study.
related quality-of-life among children in Swedish dental care: the Eur J Orthod. 2015;37:481-490.
impact from malocclusions or orthodontic treatment need. Acta 34. Kling KC, Hyde JS, Showers CJ, Buswell BN. Gender differences in
Odontol Scand. 2016;74(2):127-133. self-esteem: a meta-analysis. Psychol Bull. 1999;125:470.
12. Foster Page LA, Thomson WM, Ukra A, Baker SR. Clinical status in ado-
lescents: is its impact on oral health-related quality of life influenced by
psychological characteristics? Eur J Oral Sci. 2013;121:182-187.
13. Ghijselings I, Brosens V, Willems G, Fieuws S, Clijmans M, Lemiere J. SUPPORTING INFORMATION
Normative and self-perceived orthodontic treatment need in 11- to
16-year-old children. Eur J Orthod. 2014;36:179-185.
Additional Supporting Information may be found online in the sup-
14. Kok YV, Mageson P, Harradine NWT, Sprod AJ. Comparing a quality
of life measure and the Aesthetic Component of the Index of porting information tab for this article.
Orthodontic Treatment Need (IOTN) in assessing orthodontic treat-
ment need and concern. J Orthod. 2004;31:312-318.
15. Spalj S, Slaj M, Varga S, Strujic M, Slaj M. Perception of orthodontic
treatment need in children and adolescents. Eur J Orthod. How to cite this article: Kragt L, Jaddoe V, Wolvius E,
2010;32:387-394. Ongkosuwito E. The association of subjective orthodontic
16. Asma A, Alizae MM. Relationship of the dental aesthetic index to the treatment need with oral health-related quality of life.
oral health-related quality of life. Angle Orthod. 2016;86:337-342.
Community Dent Oral Epidemiol. 2017;00:18.
17. Calis EM, Geels LM, Prahl-Andersen B, Zentner A. Oral health-
related quality of life and dental esthetics in Amsterdam schoolchil- https://doi.org/10.1111/cdoe.12299
dren. J Dent Child (Chic). 2009;76:130-135.