Now Reading
A 20-year cohort study of health gain from orthodontic treatment: Psychological outcome

A 20-year cohort study of health gain from orthodontic treatment: Psychological outcome

American Journal of Orthodontics and Dentofacial Orthopedics, 2007-08-01, Volume 132, Issue 2, Pages 146-157, Copyright © 2007 American Association of Orthodontists

Introduction: Despite the widespread expectation that orthodontic treatment improves psychological well-being and self-esteem, there is little objective evidence to support this. The aim of this study was to compare the dental and psychosocial status of people who received, or did not receive, orthodontic treatment as teenagers. Methods: A prospective longitudinal cohort design was adopted. A multidisciplinary research team evaluated 1018 participants, aged 11 to 12 years, in 1981. Extensive assessments of dental health and psychosocial well-being were conducted; facial and dental photographs and plaster casts of dentition were obtained and rated for attractiveness and pretreatment need. No recommendations about orthodontic treatment were made, and an observational approach was adopted. At the third follow-up, 337 subjects (30-31 years old) were reexamined in 2001. One-way ANOVA was used to explore differences between the 4 groups (need/no need; treatment/no treatment). Results: The percentage changes in index of complexity, outcome and need scores for the 4 groups were need/no treatment (12.7%), no need/no treatment (−17.1%), need/treatment (31%), and no need/treatment (−11.4%). Participants with a prior need for orthodontic treatment as children who obtained treatment had better tooth alignment and satisfaction. However, when self-esteem at baseline was controlled for, orthodontic treatment had little positive impact on psychological health and quality of life in adulthood. Conclusions: Lack of orthodontic treatment when there was need did not lead to psychological difficulties in later life.

In 1962, a report from the World Health Organisation (WHO) International Expert Committee on Dental Health stated that an anomaly should be regarded as requiring treatment if the disfigurement or functional defect is, or is likely to be, an obstacle to the patient’s physical or emotional well-being. If this is true for malocclusion, there would be clear justification for orthodontic treatment and an expectation of measurable health gain. However, a review conducted in the late 1970s indicated that, despite the high demand for orthodontic treatment, the literature contained little evidence of benefit for oral health or psychological well-being. This was the stimulus for a research protocol that subsequently gained competitive peer review funding.

The multi-disciplinary research program began in 1981 as a prospective longitudinal cohort study to investigate oral health, the social and psychological effects of malocclusion, and the effectiveness of orthodontic treatment (separate reports in this series will consider periodontal health, caries, and temperomandibular dysfunction). Extensive baseline data were recorded for 1018 children in 1981 when they were 11 to 12 years old. Reports from the earlier stages of this study explored the relationship between attractiveness, dental status, and psychosocial variables. These provided some empirical support for the relationship between attractiveness and teacher expectations, attractiveness and self-esteem, and social class and the uptake of orthodontic treatment, but the initial results provided little support for the major hypothesis that children with visible malocclusion were likely to be socially and psychologically disadvantaged. Several points of caution were made with regard to this conclusion, one of which related to measures that might not have been best suited to show a psychological cost of malocclusion.

The operational definition of psychological well-being heavily depends on the measures available to the health psychologist, and, over the past 20 years, the number of validated measures of psychological health has increased substantially. In 1981, self-esteem was considered to have a major causal influence on psychological well-being in children and was one of a few validated health measures available at that time. Self-esteem can be both cause and effect, and the importance of the person in the causal relationship between attractiveness and self-esteem was addressed. Because the effects of low self-esteem have been assumed to be damaging, there has been a particular interest in identifying its causes, and the hypothesis that orthodontic treatment will lead to higher self-esteem (with a prior need) or lack of treatment can lead to low self-esteem in adulthood appears well grounded.

Material and methods

In 1981, access was granted to 23 of 29 South Glamorgan Education Authority schools in South Wales, United Kingdom. In the preliminary screening, 4810 children were identified, consent to participate was denied for 390 children (8.1%), and 651 children (13.5%) were absent from school at the screening. Nonwhite children (163; 3.4%) and children already wearing orthodontic appliances (186; 3.9%) were excluded, leaving 3420 potential subjects for the study.

The final allocation of children to the study was determined by disproportionate stratified sampling to ensure that occlusal conditions of low prevalence but high orthodontic interest would be well represented in the cohort ( Table I ). This method of sampling produced a total of 663 children with specific occlusal arrangements, approximately one fifth of the number screened; in addition, one tenth of the children without these traits were included (355), giving a final total of 1018 children. No recommendations about orthodontic treatment were made by the investigating team. Instead, an observational approach was adopted, with no attempt to influence the children’s attitudes to orthodontic treatment.

Table I
Proportions retained from 1981 to 2001 by sex, original social class, and screening category
1981 2001
n (%) n (%)
Male 508 (50) 146 (43)
Female 508 (50) 191 (57)
Family social class
Manual 445 (50) 166 (49)
Nonmanual 391 (44) 158 (47)
Unemployed/unknown 51 (6) 13 (4)
Screening category
Control 352 (35) 142 (40)
Deepbite 65 (6) 22 (7)
Prominent incisors 79 (8) 27 (8)
Partial anterior crossbite 162 (16) 55 (17)
Total anterior crossbite 22 (2) 0 (0)
General anterior spacing 37 (4) 10 (3)
Midline space 13 (1) 6 (2)
Missing incisor 68 (7) 14 (4)
Exposed upper gingivae 38 (4) 9 (3)
Severe anterior crowding 174 (17) 51 (15)

The primary assessments at age 11 to 12 years, completed in 1981, provide extensive baseline measures for each child before (for some children) orthodontic treatment.

Oral health examinations were conducted for the entire study group in a mobile dental clinic in 1981 and at the Cardiff Dental School in 2001. The psychological and social components of the study were developed to investigate the relationships among occlusal status, attractiveness, psychological well-being, social status and achievement, and quality of life. Questionnaires were distributed to the participants at each stage of the study, and to their parents and teachers in 1981. Each participant was interviewed individually. Further details of the initial sampling and methodology are described elsewhere.

From 1999 to 2001, 733 participants for whom contact details were available were targeted for the 20-year follow-up. They included people residing in the United Kingdom, Ireland, and several other countries. Ethical approval was granted, and all participants returning in 2000 and 2001 completed consent forms.


The index of complexity, outcome and need (ICON) was used because it has been shown to be a reliable and valid international occlusal index ( Appendix .). Published in 2000, it represents the consensus opinion of 98 orthodontists in 9 countries obtained from 1997 to 1999. Participants whose total ICON scores were 44 and above were classified as having a prior need for orthodontic treatment in 1981, and those with scores less than 44 as having no (or low) need for orthodontic treatment in 1981.

In 1981 and 2001, much data in addition to oral health were collected from subjects in the sample. The psychological component was executed by experienced interviewers under the supervision of a chartered health psychologist and included assessment of (1) physical appearance, including self- and objectively judged attractiveness and dental status; (2) psychological health, including emotional health, social anxiety, self-esteem, quality of life, stress, and life events; (3) health status and health-related quality of life, including beliefs about health and illness; and (4) individual and demographic characteristics.

The psychological measures were completed by the participants before the clinical examinations. A booklet was prepared and sent to participants, who were asked to complete the questionnaires in the week preceding their clinical appointment at the Cardiff Dental School. Photographs and dental casts were obtained during the clinical assessment.

A list of the data obtained that are relevant to this report is given in Tables III-V . An extended account of the measures used, the data obtained, and the statistical analysis are presented elsewhere.

Table III
Orthodontic, psychological, and demographic variables for subjects with and without orthodontic treatment need (1981) who either received or did not receive orthodontic care (means and standard deviations)
No orthodontic treatment (n = 181) Received orthodontic treatment (n = 150)
Need in 1981 (ICON = 44+) n = 124 Mean (SD) No need in 1981 (ICON = <44) n = 57 Mean (SD) Need in 1981 (ICON = 44+) n = 138 Mean (SD) No need in 1981 (ICON = <44) n = 12 Mean (SD)
Psychosocial variables Group 1 Group 2 Group 3 Group 4 F (2316) P
Physical appearance and dental status
ICON total score in 1981 64.82 (13.30) 33.60 (7.74) 75.34 (15.17) 32.83 (8.18) 196.74 <.001
ICON total score in 2001 56.61 (19.84) 39.33 (15.13) 51.97 (14.71) 36.58 (14.88) 20.33 <.001
Satisfaction with teeth (0-100) 49.03 (22.31) 58.33 (18.29) 55.04 (20.89) 63.00 (20.62) 4.66 .010
Satisfaction with teeth line up and fit 2.56 (0.67) 3.04 (0.63) 2.85 (0.69) 3.00 (0.60) 11.19 <.001
Peer comparison of teeth lineup and fit 2.90 (0.74) 3.32 (0.60) 3.09 (0.66) 3.33 (0.78) 7.46 .001
Perception of general oral health 3.15 (0.88) 3.11 (0.94) 3.30 (0.83) 3.33 (0.78) 1.49 NS
ICON esthetic component in 2001 5.27 (2.15) 3.35 (1.77) 4.69 (1.75) 3.33 (1.23) 19.51 <.001
Judged attractiveness of teeth 32.32 (9.87) 39.97 (8.06) 34.74 (8.65) 43.97 (7.09) 13.74 <.001
Self-rating of teeth (0-100) 43.44 (17.87) 49.60 (17.07) 48.96 (18.71) 63.33 (19.56) 3.79 .024
Self-rating of others’ view of teeth 42.23 (18.58) 50.32 (16.65) 49.10 (19.00) 61.67 (22.48) 4.40 .013
Dissatisfaction with appearance of teeth 2.50 (0.64) 2.82 (0.66) 2.72 (0.60) 2.92 (0.67) 6.50 .002
Peer comparison: appearance of teeth 2.93 (0.83) 3.23 (0.68) 3.04 (0.63) 3.17 (0.72) 3.39 .035
Self-rating of attractiveness (0-100) 55.13 (17.53) 50.36 (16.80) 53.16 (16.27) 52.00 (21.38) 1.58 NS
Self-rating: attractiveness to others 52.33 (17.08) 49.29 (16.97) 51.70 (16.17) 49.50 (19.29) 0.66 NS
Satisfaction with facial appearance 62.81 (17.73) 61.89 (15.68) 62.51 (16.95) 67.17 (13.29) 0.06 NS
Satisfaction: overall appearance 61.81 (19.70) 59.21 (17.65) 60.78 (16.71) 61.58 (21.68) 0.41 NS
Judged attractiveness: face (0-100) 36.36 (9.37) 37.40 (9.13) 36.84 (8.69) 36.00 (7.82) 0.27 NS
Judged attractiveness: smiling face 35.05 (7.72) 37.21 (7.39) 35.82 (8.48) 37.25 (5.11) 1.40 NS
Psychological health
General psychological health: GHQ-12 2.01 (2.83) 1.51 (2.38) 1.70 (2.29) 1.67 (2.02) 0.89 NS
RSE 2001 (reversed score 10-40) 31.40 (4.83) 31.63 (4.84) 32.99 (4.25) 32.25 (5.41) 4.33 .014 §
Depression: CES-D scale (0-60) 11.30 (10.00) 10.75 (9.35) 9.36 (7.80) 10.33 (10.40) 1.58 NS
Perceived stress: PSS-14 (0-56) 22.79 (7.77) 22.39 (7.26) 21.71 (6.34) 21.33 (7.29) 0.77 NS
Satisfaction with life scale (5-35) 22.85 (7.55) 24.30 (6.89) 25.07 (6.12) 26.33 (6.51) 3.47 .032 §
Health status and health-related QOL
Overall quality of life facet (1-5) 4.05 (0.76) 4.05 (0.69) 4.25 (0.65) 4.25 (0.62) 3.06 .048 §
Physical domain (4-20) 16.32 (2.48) 16.47 (2.24) 17.09 (1.80) 16.71 (1.40) 4.49 .012 §
Psychological domain (4-20) 14.44 (2.58) 14.75 (2.53) 15.30 (1.96) 14.83 (2.13) 4.61 .011 §
Environment domain (4-20) 14.45 (1.98) 15.09 (1.97) 15.16 (1.92) 14.71 (2.07) 4.85 .008
General health facet (1-5) 3.70 (0.97) 3.79 (0.84) 3.91 (0.91) 3.67 (0.89) 1.73 NS
Social relationships domain (4-20) 14.79 (3.34) 15.36 (3.37) 15.69 (3.11) 14.22 (3.18) 2.44 NS
Iowa-Netherlands comparison orientation (INCOM) (11-55) 34.20 (7.76) 34.02 (7.39) 32.59 (7.64) 31.17 (7.04) 1.64 NS
Social interaction anxiety (0-80) 23.65 (14.54) 22.09 (12.13) 21.16 (12.57) 23.50 (13.51) 1.16 NS
Social phobia: SPS (0-80) 13.46 (12.65) 13.84 (13.03) 11.40 (10.93) 11.67 (8.44) 1.31 NS
Self efficacy: GSES (10-40) 30.35 (4.91) 30.54 (4.17) 30.78 (4.31) 30.92 (4.32) 0.29 NS
Life events weighted score: LEI 259.4 (173.4) 230.3 (131.8) 260.7 (144.8) 205.2 (105.3) 0.88 NS
Value attached to health (4-28) 20.06 (4.48) 20.19 (4.92) 19.93 (4.23) 21.25 (3.70) 0.07 NS
Belief in dental health (4-28) 18.95 (3.79) 18.82 (3.36) 18.53 (4.02) 19.17 (3.29) 0.42 NS
SF-36 v2 health survey
General health perception (0-100) 72.78 (18.03) 74.59 (16.34) 78.28 (16.13) 80.58 (10.64) 3.51 .031 §
Reported health transition (1-5) 3.28 (0.79) 3.12 (0.71) 3.18 (0.59) 3.00 (0.43) 1.19 NS
Physical functioning (0-100) 92.12 (15.74) 91.31 (16.57) 93.45 (14.01) 92.50(13.23) 0.48 NS
Role physical–limitations (0-100) 90.02 (18.76) 91.89 (18.86) 92.51 (15.02) 86.46 (25.39) 0.69 NS
Bodily pain (0-100) 79.48 (22.92) 79.96 (23.01) 84.99 (17.76) 82.33 (24.96) 2.56 NS
Vitality (0-100) 60.50 (18.49) 57.86 (18.72) 61.58 (15.53) 57.29 (20.09) 0.93 NS
Social function (0-100) 82.46 (21.44) 85.09 (22.59) 88.42 (18.61) 87.50 (23.23) 2.76 NS
Role emotional–limitations (0-100) 88.58 (18.63) 89.77 (14.94) 91.67 (14.49) 95.83 (9.73) 1.18 NS
Mental health (0-100) 73.42 (16.45) 73.59 (15.72) 75.71 (14.76) 76.25 (17.34) 0.79 NS
Social class
NS-SEC 2001 (1-8) 3.75 (2.34) 4.07 (2.51) 3.58 (2.28) 3.53 (2.55) 0.86 NS
NS , not significant; QOL , quality of life; NS-SEC , national statistics socioeconomic classification.

All groups: 1-2, 1-3, 2-3;

groups 1-2, 1-3;

groups 1-2;

§ groups 1-3.

Table IV
Significant effects of prior need and orthodontic treatment on outcome variables, controlling for self-esteem in 2001
No orthodontic treatment n = 181 Treated n = 138
Need in 1981 (ICON = 44+) n = 124 E mean (SD) No need in 1981 (ICON ≤44) n = 57 E mean (SD) Need in 1981 (ICON = 44+) n = 138 E mean (SD)
Psychosocial variables Group 1 Group 2 Group 3 F (2315) P
Physical appearance
Dental status
ICON total score in 2001 56.58 (1.54) 39.32 (2.25) 52.01 (1.46) 20.27 <.001
Satisfaction with teeth (0-100) 49.82 (1.85) 58.87 (2.71) 54.11 (1.76) 4.00 <.01
Satisfaction with way teeth line up and fit 2.58 (0.06) 3.05 (0.09) 2.83 (0.06) 10.47 <.001
Peer comparison of teeth lineup and fit 2.92 (0.06) 3.33 (0.09) 3.08 (0.06) 7.17 <.01
ICON aesthetic component in 2001 5.28 (0.17) 3.36 (0.25) 4.67 (0.16) 19.53 <.001
Self-rating of others’ view of teeth 43.83 (1.63) 50.72 (2.39) 48.40 (1.55) 3.51 <.05
Judged attractiveness of teeth 32.42 (0.82) 40.02 (1.21) 34.63 (0.77) 13.59 <.001
Dissatisfaction with appearance of teeth 2.51 (0.06) 2.83 (0.08) 2.70 (0.05) 5.87 <.01
Peer comparison: appearance of teeth 2.95 (0.06) 3.24 (0.09) 3.01 (0.06) 3.44 <.05
E mean , Estimated mean.

Table V
Standardized regression coefficients for simultaneous regression of self-esteem at outcome upon psychosocial and dental variables
Predictive variable Beta t Significance
Quality of life: psychological domain 0.245 4.19 < .001
Self-efficacy 0.243 6.17 < .001
Life satisfaction 0.212 4.72 < .001
Depression −0.151 2.96 .01
Social interaction anxiety −0.118 2.85 .01
Quality of life: social relationships domain −0.099 2.24 .02
SF-36v2: emotional role limitations 0.089 2.26 .02
Self-rating of attractiveness 0.087 2.27 .02
ICON total score in 1981 0.076 2.27 .02
SF-36v2: general health perception 0.068 1.89 .05
Overall predictive power of model = 65%

Subjects who had orthodontic treatment received it at specialist orthodontic offices or at Cardiff Dental School, and details of treatment were obtained from the provider. The sample naturally divided into those who had received and those who had not received orthodontic treatment by 2001, providing 4 groups of participants: (1) treatment needed in 1981 but not received, (2) treatment not needed in 1981 and not received, (3) treatment needed in 1981 and received, and (4) treatment not needed in 1981 but received.

Statistical analyses

One-way ANOVAs were conducted on groups 1 through 3 across the psychological outcome measures obtained in 2001 (group 4 was excluded because of insufficient data). The significance level was set at <.05, since the variables were considered to be independent measures of psychological outcome. Significance levels less than .05 were considered not significant. Least significant difference post-hoc comparisons were used to identify pairs of means that differed significantly.

Parametric statistics were used because no variable exhibited significant skewness (with the exception of 1 measure, SF-36v2 health survey). Although the data on some scales assessing psychological pathology and health status (SF-36v2) departed from statistical normality, this was in all cases the result of the restricted range of the data, as might be expected in an essentially healthy population of 31-year-old participants. Transforming these data, particularly when there is a powerful floor effect, is unlikely to bring the distribution into normality. “Distribution-free” statistics offer no advantage in relation to this problem, and, because parametric statistics are more powerful and are generally robust to departures from normality, the use of parametric statistics was thought to be justified.


Three hundred thirty-seven (46% of the target) returned to participate in 2000 and 2001; 284 (62%) of these last attended in 1989, and a further 53 last attended in 1984. Thus, of the original 1981 sample of 1018, 33% returned after 20 years ( Table I ).

Complete psychosocial data were collected from 332 participants. One participant declined to take part in the psychological component of the study, and 4 participants who failed to return their completed data were classified as missing after 3 follow-up reminders. The ages of these participants ranged from 29.67 to 32.42 years, with a mean age of 31.25 years (SD, 0.62). There were 144 men (mean age, 31.23 years; SD, 0.67) and 188 women (mean age, 31.26 years; SD, 0.58).

Of the 332 participants, 181 had not received orthodontic treatment by 2001, and 150 had ( Table III ). There were no differences in social class between the groups.

There were statistically significant differences between ICON scores for the 3 groups in 1981 and 2001 ( P <.001; Table III ). (As noted above, group 4 was excluded from significance testing because of its small size.) For those with a prior need for orthodontic treatment who received it, the mean long-term improvement was 23.37 ICON points ( Table III ). For those with a prior need that was unmet, the occlusion also improved but only by 8.21 points. However, low initial scores tended to worsen irrespective of whether orthodontic treatment had been received (deteriorations of 3.75 points in those without prior need who received treatment and 5.73 points in those without prior need who did not). The percentage change in ICON scores for the 4 groups were need/no treatment (12.7%), no need/no treatment (−17.1%), need/treatment (31%), and no need/treatment (−11.4%).

The ICON score was significantly reduced by an average of 21 points in the group that received orthodontic treatment compared with an average reduction of 4 points in the group that did not (regardless of prior need). The data are best visualized in the scatter plot ( Fig ). The regression lines are shown for treated and untreated groups. The regression analysis confirms that orthodontic treatment achieved a long-term reduction in malocclusion for higher initial scores but not for low scores. The difference at 2001 is equivalent to a gain of 3 points on the ICON esthetic component 10-point scale.

Scatterplot and regression lines for ICON scores for treated and untreated groups.
Scatterplot and regression lines for ICON scores for treated and untreated groups.

Among those with prior need for orthodontic treatment, those who received it were significantly more satisfied with their teeth than those who did not ( P <.01, Table III ). There were also higher ratings in the perceived relative appearance of the participants’ teeth ( Table III ), although those who had no prior need and no history of orthodontic treatment scored the highest. Those with unmet needs scored the lowest ( P = .001).

Table II
Orthodontic, psychological, and demographic variables used
No of items Scale
Physical appearance and dental status/ICON 5 7-120 (≤30 acceptable; ≥43 need treatment)
Self-rating scale of attractiveness
  • 100-mm visual analogue scale: 0, very unattractive, to 100, very attractive

  • Self-attractiveness

  • Attractiveness to others

  • Attractiveness of their teeth

  • Attractiveness of their teeth to others

Satisfaction with appearance 100-mm visual analogue scale: 0, very unattractive, to 100, very attractive
Judged assessment of facial and dental attractiveness 100-mm visual analogue scale: 0, very unattractive, to 100, very attractive
Awareness and satisfaction with dental status 100-mm visual analogue scale: 0, very unattractive, to 100, very attractive
Perception of general oral health 5-point scale
Esthetic appearance of teeth 5-point scale
Psychological health
General health questionnaire (GHQ-12) 12 0-12, higher scores reflect psychological distress; >3, merits psychological intervention
Rosenberg self-esteem scale (RSE) 10 10-40, higher scores indicate greater self-esteem
Centre for epidemiological studies depression scale (CES-D) 20 0-60, >16 considered to indicate depressive disorder
Perceived stress scale (PSS) 14 0-56, higher scores indicate greater perceived stress
Health status
World Health Organisation WHOQOL-BREF quality of life scale 26 1-15, higher scores indicate higher quality of life
Iowa-Netherlands comparison orientation measure (INCOM) 11 11-55, low scores indicate low social comparison tendency
Social interaction anxiety scale (SIAS) 20 0-80
Social phobia scale 20 0-80, higher scores indicate greater anxiety about being observed
Generalized self-efficacy scale (GSES) 10 10-40, higher scores indicate greater generalized sense of self-efficacy
Life events inventory (LEI) 55 0-2879, higher scores indicate greater stress
Health status and health related quality of life (health value scale) (HVS) 4 4-28, higher scores indicate higher value attached to health
Dental health beliefs (dental health beliefs model) (HBM) 4 4-28, higher scores indicate stronger belief in preventative dental health behavior
SF36v2 health survey (standard version 2.0) 36 0-100, higher scores indicate better health status
Individual and demographic differences
Demographic information Age, sex, marital status, number and ages of children, residential status, educational level, academic qualifications, professional training, occupation, dental attendance, orthodontic treatment received, smoking habits, and current state of health
Classifications of occupations
  • 5 classes, 6 categories

  • I professional

  • II intermediate occupations

  • IIIn skilled nonmanual

  • IIIm skilled manual

  • IV partly skilled occupations

  • V unskilled populations

  • VI unemployed

Socioeconomic classification Codings based on person’s occupation—classification into 9 major groups based on skill

There were no significant differences in self-perceived attractiveness beyond the dentition and no differences between the groups in their perception of general oral health.

The ICON esthetic score and the score for attractiveness of teeth were highest for the group without prior need and no history of orthodontic treatment. These scores were lowest for the group with unmet orthodontic needs. However, there were no significant differences in the objective ratings of the smiling face photographs.

For most of these measures, no statistically significant differences emerged between the subgroups of interest ( Table III ), including general psychological health, depression, and stress, or in social interaction phobia, social phobia, and self-efficacy. However, differences did emerge for self-esteem, the satisfaction with life scale, and some elements of the WHO quality of life scale.

Thus, the group with a prior need for orthodontic treatment that was met had a higher level of self-esteem in 2001 ( P = .014) and a higher score for satisfaction with life ( P = .032) than the group with an unmet need. Participants with a prior need who had received treatment also reported the highest overall quality of life (WHOQOL-BREF) and high quality of life in the domains of physical, psychological, and environment; they also reported better general health (SF-36).

An analysis of covariance (ANCOVA) was performed to assess the effect of prior need and orthodontic treatment (3 levels) on self-esteem in 2001 with self-esteem in 1981 covaried (rescaled 1981 Piers Harris score). There was a significant main effect of self-esteem in 1981 ( F [1250] = 19.71, P <.001) but no longer a significant effect of orthodontic treatment (by prior need) on self-esteem in 2001 ( F [2250] = 1.91, not significant).

The association between orthodontic treatment and self-esteem in adulthood was therefore accounted for by self-esteem at baseline. Participants with a prior orthodontic need who were subsequently treated had higher self-esteem before treatment, and the observed increase in self-esteem in this group was paralleled by increases in the other 2 groups (difference in self-esteem 1981-2001 by prior need/treatment: F [2251] = 0.86, not significant).

Because self-esteem at 11 years of age was found to have an influence on health outcome 20 years later, it was hypothesized that current self-esteem could be exerting an influence on present psychological health. ANCOVAs with self-esteem in 2001 as the covariate were performed to reexamine the significant findings related to psychological outcome.

Self-esteem in 2001 had a significant effect on all psychological variables relating to attractiveness, psychological health, and health status ( P <.01). When self-esteem was covaried, there were no longer any significant main effects of treatment; orthodontic treatment did not have a significant effect on any psychological variables when self esteem at the 2001 follow-up was controlled for.

When the effects of prior need and orthodontic treatment were reanalyzed, self-esteem in 2001 had a significant effect on the psychological health and health status variables ( Table III ): life satisfaction, quality of life (WHOQOL-BREF), and SF-36v2 general health perception (all at P <.001), and the effects of treatment by prior need were no longer significant when self-esteem in 2001 was covaried. However, the significant effects of treatment by pretreatment need remained for the variables related to physical appearance; dental status, and attractiveness. In all cases, those with prior need who received treatment had more positive scores than those with prior need who did not receive treatment.

Table IV shows the estimated means and standard errors for variables when the effects of treatment by prior need remained significant after controlling for current level of self-esteem; the means are estimated after evaluation of the covariate.

An exploratory stepwise multiple regression analysis was performed on all 2001 outcome study quantitative psychosocial measures, the measures of dental status, and the ICON scores in 1981 and 2001 to identify which variables, if any, made significant predictive contributions to self-esteem.

The analysis produced a 10-step solution that predicted 63% of the variance in self-esteem in 2001 (adjusted R 2 , 0.632; SE estimate, 2.77). ANOVA indicated a linear relationship between the variables ( F [10,307] = 55.54; P <.001).

The 10 variables identified as making significant contributions to the prediction were then entered as independent variables into a simultaneous multiple regression analysis. There were significant linear relationships between the variables ( F [10,314] =59.75; P <.001), and the model significantly predicted 65% of the variance in self-esteem in 2001 (adjusted R 2 , 0.645; SE, 2.77). All 10 variables made significant contributions to the predictive power of the model. The beta coefficients and the t values for these variables are shown in Table V . Good psychological quality of life, general satisfaction with life, and strong self-efficacy beliefs were predictive of higher self-esteem, as were lower levels of depression and less anxiety in social interactions. Beta values in Table V for social relationships, perceptions of general health, emotional health, and attractiveness, with severity of malocclusion in 1981, indicated that the contributions of these variables to predicting self-esteem in 2001, although significant, were negligible.

The skewness of some variables entered into the regression analysis indicated departures from normality. The data from these variables were transformed, and the regression analysis was rerun with the transformed data; there was no change in the statistical conclusions (adjusted R 2 , 0.653; SE estimate, 2.74), and the same 5 variables made the greatest contributions to the prediction of self-esteem.


Our results indicate that persons with a prior need for orthodontic treatment that was met have a more favorable perception of, and higher satisfaction with, their own teeth than those with an unmet need. However, although initial analysis indicated that these were accompanied by higher levels of self-esteem and reported quality of life, these gains were lost when account was taken of pretreatment self-esteem in 1981. Thus, the results on 31-year-olds confirm those when the cohort was in adolescence: visible malocclusion has no discernable negative effect on social and psychological well-being, and orthodontic treatment has no discernable positive effect.

Other attempts to construct comparisons between groups of subjects who either obtained or failed to obtain orthodontic correction of malocclusion reached equivocal conclusions. Two randomized trials evaluated short-term psychological outcomes. In an American setting, Albino et al found that dental-specific evaluations appear to be influenced by treatment, whereas more general psychosocial responses are not. In their randomized trial of early treatment of Class II malocclusion, subjects in general did not have below-average self-concepts, and those who received treatment did not experience improvement in social competency or self-esteem. In contrast, children who received early treatment in a recent randomized trial in the United Kingdom of Class II Division 1 malocclusion had increased self-esteem. In a 15-year follow-up in 30-year-old Danes, some subjects with untreated severe malocclusion experienced continuing feelings of dissatisfaction with their teeth and memories of teasing that persisted into adulthood. However in their progression through life with respect to personal accomplishment and achievement, and more general psychological well-being, they did not appear disadvantaged.

The finding that dental status did not make a greater contribution to the prediction of self-esteem led us to explore the contribution of other variables to self-esteem in adulthood. In this 2001 study, 65% of the variance in self-esteem was predicted by psychological variables related to perception of quality of life, life satisfaction, affective state (depression and social anxiety), emotional health, and perception of attractiveness. Interestingly, poor quality of life with respect to social relationships was not predictive of low self-esteem, whereas higher levels of depression and social anxiety, as might be expected, did predict low self-esteem.

The fact that severity of malocclusion assessed in childhood made a weak contribution to the prediction of self-esteem 20 years later is noteworthy, although prior need and treatment received were not considered in this analysis. However, the contribution of dental status at age 11 to self-esteem at age 30, although statistically significant, appears to be of minor importance in a model that includes other psychosocial variables.

Our findings demonstrate that analyzing the long-term effects of orthodontic treatment on psychological health without considering intervening factors might lead to invalid conclusions regarding the efficacy of treatment. The finding that participants who received orthodontic treatment had significantly higher scores on some psychological health variables than those who did not receive treatment appeared to be an artefact. When prior need for treatment 20 years earlier was taken into account, there was little objective evidence to support the belief that orthodontic treatment improves psychological well-being in adulthood. In addition, lack of orthodontic treatment when there was a prior need did not lead to psychological difficulties in later life in the form of nonpsychotic mental disorders, depressive disorders, social anxiety, or social phobia.

The observed effect of prior need and orthodontic treatment on self-esteem at outcome was accounted for by self-esteem at baseline. This finding emphasizes the need for longitudinal analysis of baseline and outcome data when evaluating the effect of treatment on psychological variables. In our sample, participants’ self-esteem increased over 20 years but not as a result of receiving orthodontic treatment and regardless of whether there was a prior need for treatment at baseline. Orthodontic treatment alone did not have a significant effect on any psychosocial variables when self-esteem at outcome in 2001 was controlled for.

The principal uncertainty surrounding our study was anticipating the dropout rate. From the outset, this was an observational study, and the subjects, recruited as schoolchildren, were offered neither treatment nor other benefit by the researchers. Thus, successful follow-up depended entirely on goodwill and the effectiveness and persistence of our recall efforts. In practical terms, retention became a greater challenge after 1989 when our subjects no longer attended school.

Some indication of the success of sequential recall efforts is shown by the proportion of subjects lost to follow-up over each interval. Between 1981 and 1984, 22% of the subjects failed to return, equal to 7.3% per year. Between 1984 and 1989, the rate of loss was 8.4% per year. From 1989 to 2001, the rate of loss fell to 2.2% per year, in spite of the longest gap.

Inevitably, however, the total level of dropouts (about two-thirds of the initial 1018) raises possibilities for bias. Although the sample remaining in the study at the 20-year follow-up was similar for the main characteristics recorded for the entire sample at the outset ( Table I ), the generalizability of the findings to the entire sample cannot be guaranteed.

Another issue concerns factors that determined division into the 2 groups with severe malocclusion—those who received treatment and those who did not. As the analyses indicate, those who received treatment did not appear to differ in major respects in the 1981 recorded characteristics from those who did not. Family social class in 1981 was not associated with the uptake of orthodontic treatment, although social class in 2001 was. However, a low valuation of dental alignment or unwillingness to comply with treatment must undoubtedly have influenced entry to treatment and might have diminished any potential for psychological benefit to be realized. Interestingly, subjects with a high objective need for treatment that was unmet had higher levels of decayed, missing, or filled teeth at age 11, suggesting some inherent difference in dental health behavior or attitude in the untreated group.

You're Reading a Preview

Become a DentistryKey membership for Full access and enjoy Unlimited articles

Become membership

If you are a member. Log in here