SEARCH

SEARCH BY CITATION

Keywords:

  • BMI changes;
  • draft board;
  • prospective study

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. References

Objective: To investigate whether intelligence and education are related to subsequent BMI changes and development and persistence of obesity in men from young adulthood through middle-age.

Research Methods and Procedures: Subjects were selected among men (median age, 19 years; examined between 1956 and 1977) appearing at Danish draft boards: a group with juvenile-onset obesity, including all men with a BMI of ≥31.0 kg/m2; and a nonobese group randomly selected as a 1% sample of the study population. The obese group and 50% of the nonobese group were invited to participate in follow-up studies between 1982 and 1984 and between 1992 and 1994. Among 907 men with juvenile-onset obesity and 883 nonobese men, age, examination region, intelligence test score, education, and BMI from baseline to first follow-up were analyzed by multiple linear and logistic regressions analyses.

Results: Education and intelligence, analyzed separately, were inversely related to BMI changes in both groups and to the development of obesity in the nonobese group. When adjusted for education, the association between intelligence score and BMI changes and development of obesity vanished, whereas the inverse relationship for education persisted only for BMI changes. Intelligence score was not associated with the persistence of obesity in the obese group, whereas inverse relationships were found for education.

Discussion: Intelligence test score was inversely related to risk of BMI changes and the risk of development of obesity, perhaps with education acting as a mediator or indicator of cognitive ability. Education, but not intelligence, was inversely associated with risk of remaining obese.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. References

Several studies have investigated the prospective associations between various social factors, such as school performance, educational attainment, occupation, and income, and BMI or weight changes. The results are inconsistent, with either inverse (1, 2, 3, 4, 5, 6) or no associations between these factors and BMI or weight changes (7, 8). Cross-sectional studies conducted on data from Danish draftees showed a clear inverse association between BMI above the median BMI and both intelligence test score and educational level. In particular, the obese subjects with BMI ≥31 kg/m2 had significantly lower intelligence test scores compared to the nonobese control group (9, 10, 11). The same tendencies have also been seen in a Chinese study in children, who were tested with Wechler's Intelligence Scale for Children (12). These cross-sectional observations raise the obvious question of whether cognitive ability and educational level are determinants or consequences of changes in body weight and development of obesity.

In the present study, we addressed the former question by investigating intelligence test score and educational level in young adult men in relation to their subsequent changes in weight and risk of development and persistence of obesity.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. References

Draft Board Examination and Study Population

At 18 years of age, all Danish men are required to register for eligibility in the military service; within the next few years after registration, they are required to appear before the local draft board for examination. The men undergo systematic health examinations, including measurement of weight in underwear, height without shoes, intelligence testing, and rating of educational level (11, 13). Reasons for delayed appearance are mainly ongoing education, disease, and traveling abroad. Reasons for exemption from service without appearing before the board are chronic diseases and disease sequelae, handicaps, and mental retardation that requires institutionalization; however, obesity is not a condition for exemption.

The background population for the actual study comprised 362, 200 men registered at the draft boards from 1943 through 1977 in the Copenhagen area (region 1) and from 1964 through 1977 in the remainder of Zealand (region 2). The data set in the present study was from the period of 1956 to 1977, the time during which intelligence testing was part of the examination. The study sample consisted of two groups: a group with juvenile-onset obesity and a nonobese control group, identified at draft board examination by manual screening of the files. The group with juvenile-onset obesity included 1564 men who were at least 35% overweight, according to a local national standard used for the original selection, which corresponded to a BMI of ≥31.00 kg/m2 (13). The nonobese control group was selected as a random 1% sample of the total study population in the same time period, and 50% of the controls were invited to the follow-up studies (n = 1191). Men with a BMI ≥31 kg/m2 (n = 21) and those without measures for height and weight (n = 171) were excluded from the control group.

For intelligence testing, the Børge Priens Prøve 1953 (BPP-53)1 test was applied. The BPP-53 test is a 45-minute test designed for group testing. It has remained unchanged since its incorporation into the draft board preinduction procedure in 1956. The test is still in use and is, therefore, not available to the public. The BPP-53 test comprises four subtests: 1) letter matrices (resembling “Ravens Progressive Matrices”); 2) verbal analogies (similar to the Miller's Analogies test); 3) number series, involving subjects giving the fifth number that follows a series of four; and 4) geometric figures, involving subjects decomposing a set of complex geometric shapes into smaller components. The test includes 78 items, with the sum of correct answers providing the test score. The test has been investigated extensively and has been proven to be a useful indicator of general intelligence and cognitive ability, with a high correlation to the Wechler Adult Intelligence Scale (WAIS) (14, 15, 16, 17). The BPP-53 test score was used for assessing cognitive performance skills and abilities in addition to the information about educational level.

Educational level at the time of examination was divided into nine categories, starting with primary school and ending with an academic university degree (11). For a sufficient number of subjects at each level, educational level was divided into the following four levels: 1) primary school (7 years) plus minor courses or elementary training; 2) primary school plus completed apprenticeship; 3) 9 or 10 years of school (ordinary level); 4) diploma, advanced, or academic level (≥12 years).

Follow-up Surveys

In collaboration with the Copenhagen City Heart study, two follow-up surveys were conducted, one between 1982 and 1984 and another between 1992 and 1994. Participants in the first follow-up survey included 907 (58%) of the 1564 individuals with juvenile-onset obesity at baseline and 883 (74%) of the 1191 nonobese controls. At the second follow-up, 747 individuals with juvenile-onset obesity (49% compared with baseline) and 730 nonobese controls (61% compared with baseline) participated. Some of the participants at the second follow-up did not participate in the first follow-up. The follow-up studies included information on age, height without shoes, and weight in underwear. Comparisons of the characteristics of the participants and nonparticipants at follow-up have been published previously (18).

Statistical Analyses

For descriptive statistics, median 5th and 95th percentiles were used. Data were analyzed in multiple linear and logistic regression models. All the analyses were run separately for the control and the obese group, and the results are presented in this report as regression coefficients (β) or odds ratios (OR) with 95% confidence intervals (CIs). The separation of the obese and the nonobese group in the analyses was derived from this study design (18, 19). Identifying the group of extremely overweight men gave new possibilities for studying the changes in BMI in the most extreme right of the distribution. Because of the sampling frame, the obese group represented a 200-fold higher share of the relevant part of the underlying population than the randomly selected nonobese control group. In the logistic regression analyses investigating development or persistence of obesity from baseline to first follow-up, the outcome binary variables were set at BMI ≥31 kg/m2 for the obese group and at BMI ≥30 kg/m2 for the control group. Subjects in the nonobese control group with baseline BMI between 30 and 31 kg/m2 (n = 2) were excluded from the logistic regression analysis. The intelligence test score was included both as a continuous and as a categorical variable divided into quintiles, based on the distribution of the control group. In the results, models including either education or intelligence test score and adjusted for baseline BMI, draft board region, and age at the actual study are referred to as basic models, whereas analyses including the same variables, but with both intelligence test score and educational level, are referred to as combined models. All the statistical analyses were performed in the SAS 8.0 software program (SAS Institute, Cary, NC).

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. References

Descriptive Analyses

Median age and BMI at baseline, first, and second follow-up for subjects included in the analyses are presented in Table 1. Because data from baseline were collected at draft board examinations during the time period between 1956 and 1977, the age range in the two follow-up studies, conducted during a few years, was much larger than at baseline. Within both groups, BMI increased significantly from baseline to both the first and second follow-ups. The median changes in BMI were greater in the nonobese group than in the obese group. There was considerable inter-individual variation in BMI change within each group; however, the variation was greater within the obese group than the nonobese group. Within the obese group, there were equal trends toward increases and decreases in BMI, whereas increases in BMI prevailed within the nonobese group (Tables 2 and 3). At baseline, the nonobese control group had a significantly higher median intelligence test score and educational level than the obese group.

Table 1.  Variables at baseline and first and second follow-up for the nonobese control and the obese groups with complete information
VariablesNonobese control group*Obese group
  • *

    Baseline BMI <31 kg/m2.

  • Baseline BMI ≥31 kg/m2.

  • For continuous variables, median and 5th and 95th percentiles are given.

  • §

    The intelligence test score range from 0 to 78 points.

Baseline  
 N883907
 Age, years19 (18 to 24)19 (18 to 24)
 BMI, kg/m221.3 (18.4 to 25.7)32.5 (31.1 to 37.9)
 Intelligence test score§41 (22 to 58)34 (16 to 55)
 Educational level 9 or more years of education (%)4828
First follow-up  
 N883907
 Follow-up time: baseline to first follow-up, years13 (5 to 23)11 (5 to 20)
 Age, years33 (25 to 42)31 (24 to 40)
 BMI at first follow-up, kg/m224.4 (20.0 to 30.4)33.6 (25.2 to 43.1)
Second follow-up  
 N730747
 Follow-up time: baseline to second follow-up, years24 (16 to 34)22 (16 to 30)
 Age, years45 (35 to 53)42 (35 to 51)
 BMI, kg/m225.5 (20.7 to 32.3)35.1 (27.6 to 45.9)
Table 2.  Intelligence and education of the randomly selected nonobese group in relation to median (5th to 95th percentiles) values for BMI at draft board examination and subsequent BMI changes until first follow-up, as well as estimates and OR (95% CI) for development of obesity
  Median BMI, kg/m2 (5th to 95th percentile)Linear regression BMI first follow-up [β (95% CI)]Logistic regression BMI ≥30 kg/m2 [OR (95% CI)]
VariablenAt draft boardΔBasic model*Combined modelBasic model*Combined model
  • *

    The basic regression model includes intelligence test score or educational level and baseline BMI, age at first follow-up, and draft board region.

  • The combined regression model includes baseline BMI, age at first follow-up, draft board region, intelligence test score, and educational level simultaneously.

  • ref, reference group.

Baseline BMI88321.3 (18.4 to 25.7)2.4 (−0.5 to 7.1)0.91 (0.84 to 0.97)0.94 (0.88 to 1.00)1.97 (1.69 to 2.33)2.03 (1.73 to 2.42)
 Intelligence test score (quintiles)       
  0 to 30 points18121.2 (17.8 to 26.2)3.3 (−0.3 to 8.1)0 (ref)0 (ref)1 (ref)1 (ref)
  31 to 38 points18821.5 (18.5 to 25.7)2.5 (−0.3 to 5.9)−0.89 (−1.34 to −0.45)−0.82 (−1.28 to −0.37)0.33 (0.11 to 0.88)0.29 (0.10 to 0.82)
  39 to 44 points18121.2 (18.3 to 26.5)2.2 (−0.6 to 7.0)−0.88 (−1.33 to −0.43)−0.58 (−1.08 to −0.09)0.51 (0.18 to 1.39)0.63 (0.19 to 1.95)
  45 to 51 points17921.6 (18.3 to 26.0)2.3 (−0.9 to 6.3)−1.01 (−1.46 to −0.56)−0.46 (−1.00 to 0.09)0.24 (0.08 to 0.70)0.36 (0.09 to 1.29)
  52 to 78 points15421.3 (18.4 to 26.0)1.7 (−0.6 to 6.5)−1.59 (−2.06 to −1.12)−0.70 (−1.33 to −0.08)0.16 (0.04 to 0.52)0.26 (0.04 to 1.70)
  p for trend   <0.00010.120.0010.11
 Educational level (baseline)       
  Primary school and few courses13521.4 (18.0 to 26.9)2.7 (−0.6 to 7.5)0 (ref)0 (ref)1 (ref)1 (ref)
  Primary school and apprenticeship32121.4 (18.5 to 25.8)2.8 (−0.3 to 7.5)−0.02 (−0.45 to 0.42)0.14 (−0.30 to 0.59)1.37 (0.55 to 3.77)1.85 (0.69 to 5.41)
  9 or 10 years (ordinary level)24721.3 (18.2 to 25.7)2.2 (−0.6 to 6.4)−0.69 (−1.14 to −0.23)−0.43 (−1.00 to 0.12)0.36 (0.10 to 1.22)0.62 (0.13 to 2.92)
  12 or more years of school18021.5 (18.1 to 26.1)1.6 (−1.0 to 5.2)−1.38 (−1.87 to −0.90)−1.09 (−1.72 to −0.46)0.39 (0.11 to 1.35)0.86 (0.13 to 5.13)
  p for trend   <0.00010.00020.010.72
Table 3.  Intelligence and education of the group of individuals with juvenile-onset obesity in relation to median (5th to 95th percentiles) values for BMI at draft board examination and subsequent BMI changes until first follow-up, as well as estimates and ORs (95% CI) for persistence of obesity
  Median BMI, kg/m2 (5th to 95th percentile)Linear regression BMI first follow-up [β (95% CI)]Logistic regression BMI ≥31 kg/m2 [OR (95% CI)]
VariableNAt draft boardΔBasic model*Combined modelBasic model*Combined model
  • *

    The basic regression model includes intelligence test score or educational level and baseline BMI, age at first follow-up, and draft board region.

  • The combined regression model includes baseline BMI, age at first follow-up, draft board region, intelligence test score, and educational level simultaneously.

  • Quintiles based on the distribution of the control group.

  • ref, reference group.

Baseline BMI90732.5 (31.1 to 37.9)0.5 (−7.0 to 9.2)0.71 (0.57 to 0.84)0.71 (0.58 to 0.85)1.20 (1.11 to 1.30)1.21 (1.11 to 1.32)
 Intelligence test score       
  0 to 30 points33132.7 (31.1 to 38.8)0.4 (−6.9 to 9.3)0 (ref)0 (ref)1 (ref)1 (ref)
  31 to 38 points22732.5 (31.1 to 36.8)1.0 (−7.3 to 9.4)0.08 (−0.76 to 0.92)0.15 (−0.71 to 1.01)1.05 (0.71 to 1.57)1.10 (0.73 to 1.67)
  39 to 44 points14932.7 (31.1 to 37.2)0.2 (−7.0 to 9.1)−0.44 (−1.40 to 0.52)−0.07 (−1.15 to 1.00)0.86 (0.55 to 1.34)1.05 (0.63 to 1.76)
  45 to 51 points10932.4 (31.1 to 35.8)0.9 (−6.5 to 8.8)−0.17 (−1.25 to 0.91)0.97 (−0.41 to 2.35)0.95 (0.58 to 1.57)1.67 (0.87 to 3.30)
  52 to 78 points9132.4 (31.1 to 38.0)−0.3 (−6.8 to 7.7)−1.25 (−2.41 to −0.09)0.72 (−0.90 to 2.34)0.61 (0.37 to 1.03)1.44 (0.68 to 3.10)
  p for trend   0.060.340.100.27
 Educational level (baseline)       
  Primary school and few courses31832.8 (31.1 to 39.7)0.9 (−8.4 to 9.2)0 (ref)0 (ref)1 (ref)1 (ref)
  Primary school and apprenticeship33432.4 (31.1 to 37.0)0.8 (−6.2 to 9.4)0.16 (−0.61 to 0.93)0.11 (−0.69 to 0.92)1.09 (0.76 to 1.58)1.05 (0.72 to 1.55)
  9 or 10 years (ordinary level)14932.7 (31.1 to 36.9)0.3 (−6.6 to 11.0)−0.15 (−1.11 to 0.82)−0.55 (−1.79 to 0.68)0.87 (0.56 to 1.37)0.69 (0.39 to 1.24)
  12 or more years of education10632.4 (31.1 to 35.8)−0.9 (−7.4 to 6.9)−2.16 (−3.26 to −1.06)−2.80 (−4.37 to −1.23)0.46 (0.28 to 0.74)0.33 (0.16 to 0.67)
  p for trend   0.0020.010.0040.01

As expected, the educational level and the intelligence test score were correlated (Spearman's ρ = 0.71 and 0.69 for the nonobese and juvenile-onset obesity groups, respectively). This did not cause co-linearity problems in the statistical modeling.

The Nonobese Control Group

BMI Changes

Baseline BMI was strongly and significantly associated with later increase in BMI (Table 2). A greater increase in BMI at first follow-up occurred in the lowest intelligence test-score quintiles and in the group of less educated individuals (Table 2). In the basic regression models, the highest intelligence test-score quintile had less increases in BMI for a given baseline BMI at first follow-up compared with the lowest test-score quintile (β = −1.59; 95% CI: −2.06 to −1.12), and there was a clear and highly significant inverse trend (Table 2). When the intelligence test score was used as a continuous variable, the regression coefficient for each five-point increase was −0.19 (95% CI: −0.25 to −0.13), corresponding to a score difference of 26 points for one BMI unit (p < 0.0001). Educational level was also inversely related to BMI changes (p < 0.0001). The highest-educated group had significantly less increase in BMI at first follow-up (β = −1.38; 95% CI: −1.87 to −0.90) compared with the least-educated group. In the combined models, the inverse trend between intelligence test score and BMI changes lost significance, with the test score both as a categorical variable and as a continuous variable (for each five-point increase: β = −0.06; 95% CI, −0.15 to 0.02), whereas the inverse relationship between educational level and BMI persisted.

The analyses of data from the second follow-up were almost the same for both the basic (p for inverse trend: <0.0001 for both intelligence and education) and the combined (p for trend = 0.22 and p = 0.001, respectively) models for intelligence test score and education.

Development of Obesity

The roles of intelligence test score and educational level as predictors of development of obesity (defined as BMI ≥30 kg/m2) were investigated in logistic regression analyses (Table 2). In the basic models, the OR of developing obesity at first follow-up was 0.16 (95% CI: 0.04 to 0.52) for the highest quintile compared with the lowest. The OR for the test score used as a continuous variable was 0.78 per five points (95% CI: 0.67 to 0.91; p = 0.001). Educational level also had a significant inverse effect on developing obesity. The results from the combined models were not so clear. The inverse relationship between educational level and the subjects’ odds of developing obesity at first follow-up lost significance. For the intelligence test score, an inverse trend was still present when included as a continuous variable (OR = 0.80; 95% CI: 0.65 to 0.99), but not if analyzed as quintiles (Table 2).

In the analyses of the second follow-up data, significant inverse relationships were found between both intelligence test score and educational level and obesity in the basic models (p for trend = 0.01 and p = 0.001, respectively). In the combined models, no trend between intelligence test score and the OR for becoming obese was seen (p for trend = 0.59), whereas it persisted for educational level (p for trend = 0.02).

The Juvenile-Onset Obesity Group

BMI Changes

There was no clear trend in the relationship between intelligence test score and the BMI changes at first follow-up, except for the group in the highest intelligence test-score quintile, in which a significant decline in the median BMI had taken place (Table 3). Within the four educational levels, a greater increase in BMI was seen in the least-educated groups, and, as for intelligence test score, there was a negative BMI change in the group with the highest educational level.

In the basic linear regression models for the first follow-up, the highest test-score quintile showed a tendency toward weight loss and was significantly different from the change in the lowest quintile, and there was a borderline inverse trend (Table 3). Using the test score as a continuous variable, the regression coefficient for each five-point increase was −0.13 (95% CI: −0.27 to 0.01; p = 0.06), corresponding to a score difference of 38 points for every 1 kg/m2 BMI. Educational level was significantly inversely related to BMI change at first follow-up. The highest-educated group showed a tendency toward weight loss, whereas the least-educated group showed a tendency toward weight gain (Table 3).

In the combined models, no relationship between intelligence test score and BMI change at first follow-up was seen, although the inverse trend for educational level and BMI changes persisted (Table 3).

At second follow-up, the associations with educational level, seen in the basic models at first follow-up, lost significance, and the borderline significant trend seen for intelligence test score disappeared as well. In the combined model, there was a significant inverse trend for educational level, whereas the relationship between intelligence test score and BMI changes seemed to change direction toward a positive association, which, however, was not significant (p for trend = 0.12).

Persistence of Obesity

The ORs for obesity persisting at first follow-up to at least the same degree as at the baseline examination (BMI ≥31 kg/m2) were lower for the three highest test-score quintiles than for the lowest quintile, but the differences were insignificant, and there was no significant trend. Using the test score as a continuous variable, the OR for each five-point increase was 0.95 (95% CI: 0.90 to 1.02; p = 0.14). On the other hand, the highest-educated group had less than half the odds of remaining obese compared with the lowest-educated group, and a significant trend was present (Table 3). In the combined models, the inverse relationship and significant trend between educational level and the persistence of obesity at first follow-up was still present, whereas a nonsignificant tendency toward higher ORs of maintaining obesity was found in the highest intelligence test-score quintiles, compared with the lowest quintile. At second follow-up, no relationship between intelligence test score and the risk of persistence of obesity, either in the basic or in the combined models, was found, whereas there was still an inverse trend between educational level and persistence of obesity.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. References

We investigated whether intelligence test score and educational level assessed in young adult men was associated with subsequent BMI changes and the risk of development and persistence of obesity. We found that both intelligence test score and educational level had an inverse effect on the subsequent BMI changes and risk of developing obesity. When investigated simultaneously, the effects of both the intelligence test score and education were reduced; however, the effects of intelligence test score were more reduced than those of education. Educational attainment showed a strong inverse relationship with the persistence of obesity in the juvenile-onset obesity group, whereas the intelligence test score had no effect.

The limitations of the study must be considered in the interpretation of the results. Possible effects of selective nonparticipation at follow-up should be considered. A higher rate of nonparticipation was seen among the men with juvenile-onset obesity, and in both groups, there were fewer nonparticipants among those with the lowest BMI, with the highest education, and with a higher intelligence test score (18). If weight gain and risk of development or persistence of obesity was greater among the nonparticipants, which seems likely, then the strength of the associations with educational level and intelligence test score were underestimated.

General BMI changes of a considerable degree were seen for both the juvenile-onset obesity and nonobese groups of men during the follow-up period, with a greater median increase in the nonobese group, but a greater variability in the obese group. Subjects who were obese at enrollment into the cohort allowed for observation of considerable weight loss, which was much less likely in the nonobese group, that is, as long as they remained healthy. On the contrary, the relatively low BMI of the nonobese group allowed for a considerable weight gain, which actually took place in some of them, particularly between baseline and the first follow-up. It seems plausible that the determinants of developing obesity may not be the same as the determinants of persistence of obesity. These reasons support the need for keeping the two groups separated during the analyses.

We found less consistent relationships when studying obesity as an outcome compared to studying general BMI change as an outcome. This may be due to the gradual development of obesity over time. Thus, only a few men (n = 42) in the nonobese group had reached the BMI threshold for obesity in the relatively short period preceding first follow-up, whereas at second follow-up, considerably more subjects had become obese (n = 92), improving the power of the analyses. On the other hand, with the longer follow-up, other influences may have weakened the associations.

The intelligence test score used in the present study has been evaluated against the WAIS, which is an individual intelligence test widely used and approved in psychological research (16). A high correlation was found, indicating that the group-based BPP-53 test measured the same general intelligence as the WAIS test. The BPP-53 has proven to be a suitable tool for epidemiological and demographic studies of intelligence (10, 11, 14, 15, 20). Although the test score correlates with educational attainment, it contributes important additional information about cognitive ability in another way than educational attainment. It is difficult to assess the degree to which good performance in the intelligence test is a consequence of previous schooling or whether the test measures an ability that determines the success in school performance, but there is evidence that a bi-directional interaction is operating (20, 21, 22). Danish adoption studies using the same type of data have found a stronger genetic effect on intelligence test score and stronger effects of familial environment on educational attainment (23). Overall, it seems justified to consider intelligence test score and educational attainment as separate entities despite the interaction and correlation between them.

This also seems to be valid in the study of the relationship between body size and intelligence test score and educational level. Thus, cross-sectional studies have shown equally strong and independent relationships for both variables (i.e., intelligence test score and education) and BMI or obesity, indicating that the two variables can account only partly for each other in their relationship with body size (10). Our findings of a somewhat stronger influence of educational attainment than of intelligence test score may suggest that educational attainment is acting as a mediating factor of the inverse effect of intelligence test score. Alternatively, the intelligence test-score performance in relation to body size may be a marker of educational attainment.

A few studies that have been published have not found these inverse associations between education and subsequent weight or BMI changes, although cross-sectional, inverse associations between educational level and BMI at baseline were found in these studies (7, 8). It has been proposed that this could be due to an effect of earlier weight gain on educational attainment (8). However, this is in contrast to our result. The same association found in our study has also been observed among children in whom school difficulties in the third grade were associated with overweight in young adults, irrespective of childhood BMI (5). On the other hand, obesity in childhood and adolescence has been found to influence later adult social class status (24, 25). Thus, there may well be bi-directional interactions between cognitive ability or education and body weight (26).

Several reasons for our findings may be considered. High ability in intelligence testing and educational attainment may be related to stronger expectations of a slim physical appearance and, therefore, a higher motivation for weight regulation or loss. Through their high cognitive skills, the well-educated might also have a better ability to receive and implement general-health guidelines into their everyday lifestyle, compared with the less-educated groups, who might not feel the same pressure to be slim (27). A recent British study suggested that part of the social gradient in obesity is due to a higher frequency of weight monitoring, a lower threshold for defining themselves as overweight, a higher deliberate effort at weight control utilized in more restricted dietary practices, and a higher level of physical activity in the higher social classes (28). Another, though speculative, possibility is that the biological pathways leading to weight gain and eventually obesity may also imply disturbances of the brain function at higher levels than the hypothalamic regulation of energy balance.

In conclusion, the intelligence test score in young men was inversely related to subsequent changes in BMI, perhaps with educational level acting as a mediator or indicator of the cognitive ability. Similar, but less consistent, relationships were found for the development of obesity. Persistence of obesity among men with juvenile-onset obesity was strongly inversely related to educational level, but not to the intelligence test score. These results encourage further studies of the mechanism of the interaction between cognitive psychosocial factors and obesity. Better understanding of this interaction may allow improved targeting and, thereby, more effective prevention and treatment of obesity.

Footnotes
  • 1

    Nonstandard abbreviations: BPP-53, Børge Priens Prøve 1953; WAIS, Wechler Adult Intelligence Scale; OR, odds ratio; CI, confidence interval.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Research Methods and Procedures
  5. Results
  6. Discussion
  7. References
  • 1
    Rissanen, A. M., Heliovaara, M., Knekt, P., Reunanen, A., Aromaa, A. (1991) Determinants of weight gain and overweight in adult Finns. Eur J Clin Nutr 45: 419430.
  • 2
    Martikainen, P. T., Marmot, M. G. (1999) Socioeconomic differences in weight gain and determinants and consequences of coronary risk factors. Am J Clin Nutr 69: 719726.
  • 3
    Kahn, H. S., Williamson, D. F. (1990) The contributions of income, education and changing marital status to weight change among US men. Int J Obes 14: 10571068.
  • 4
    Lahmann, P. H., Lissner, L., Gullberg, B., Berglund, G. (2000) Sociodemographic factors associated with long-term weight gain, current body fatness and central adiposity in Swedish women. Int J Obes Relat Metab Disord 24: 685694.
  • 5
    Lissau, I., Sørensen, TIA. (1993) School difficulties in childhood and risk of overweight and obesity in young adulthood: a ten year prospective population study. Int J Obes Relat Metab Disord 17: 169175.
  • 6
    Wagner, A., Simon, C., Ducimetiere, P., et al (2001) Leisure-time physical activity and regular walking or cycling to work are associated with adiposity and 5 y weight gain in middle-aged men: the PRIME Study. Int J Obes Relat Metab Disord 25: 940948.
  • 7
    French, S. A., Jeffery, R. W., Forster, J. L., McGovern, P. G., Kelder, S. H., Baxter, J. E. (1994) Predictors of weight change over two years among a population of working adults: the Healthy Worker Project. Int J Obes Relat Metab Disord 18: 145154.
  • 8
    van Lenthe, F. J., Droomers, M., Schrijvers, C. T. M., Mackenbach, J. P. (2000) Socio-demographic variables and 6 year change in body mass index: longitudinal results from the Globe study. Int J Obes Relat Metab Disord 24: 10771084.
  • 9
    Sørensen, T. I. A., Sonne-Holm, S., Christensen, U., Kreiner, S. (1982) Reduced intellectual performance in extreme overweight. Hum Biol 54: 765775.
  • 10
    Teasdale, T. W., Sørensen, T. I. A., Stunkard, A. J. (1992) Intelligence and educational level in relation to body mass index of adult males. Hum Biol 64: 99106.
  • 11
    Sørensen, T. I. A., Sonne-Holm, S. (1985) Intelligence test performance in obesity in relation to educational attainment and parental social class. J Biosoc Sci 17: 379387.
  • 12
    Li, X. (1995) A study of intelligence and personality in children with simple obesity. Int J Obes Relat Metab Disord 19: 355357.
  • 13
    Sonne-Holm, S., Sørensen, TIA. (1977) Post-war course of the prevalence of extreme overweight among Danish young men. J Chron Dis 30: 351358.
  • 14
    Teasdale, T. W., Owen, D. R., Sørensen, TIA. (1988) Regional differences in intelligence and educational level in Denmark. Br J Educ Psycol 58: 307314.
  • 15
    Teasdale, T. W., Owen, D. R. (2000) Forty-year secular trends in cognitive abilities. Intelligence 28: 115120.
  • 16
    Mortensen, E. L., Reinisch, J. M., Teasdale, T. W. (1990) Intelligence as measured by the WAIS and a military draft board test. Scand J Psychol 31: 315318.
  • 17
    Rash, G. (1960) Probabilistic Models for Some Intelligence and Attainment Tests Danish Institute for Educational Research, Copenhagen, Denmark Copenhagen, Denmark.
  • 18
    Sonne-Holm, S., Sørensen, T. I. A., Jensen, G., Schnohr, P. (1989) Influence of fatness, intelligence, education and sociodemographic factors on response rate in a health survey. J Epidemiol Community Health 43: 369374.
  • 19
    Sonne-Holm, S., Sørensen, T. I. A., Jensen, G., Schnohr, P. (1989) Independent effects of weight change and attained body weight on prevalence of arterial hypertension in obese and non-obese men. BMJ 299: 767770.
  • 20
    Teasdale, T. W. (1985) Familial Influences in Social Class, Educational Level and Intelligence. Ph.D. Dissertation Studiae Psychologicae et Paedagogicae 78 Liber Lund Sweden.
  • 21
    Teasdale, T. W., Owen, D. R. (1987) National secular trends in intelligence and education: a twenty-year cross-sectional study. Nature 325: 119121.
  • 22
    Neisser, U., Boodoo, G., Bouchard, T. J., Jr, et al (1996) Intelligence: knowns and unknowns. Am Psychol 51: 77101.
  • 23
    Teasdale, T. W., Owen, D. R. (1984) Heredity and familial environment in intelligence and educational level—a sibling study. Nature 309: 620622.
  • 24
    Gortmaker, S. L., Must, A., Perrin, J. M., Sobol, A. M., Dietz, W. H. (1993) Social and economic consequences of overweight in adolescence and young adulthood. N Engl J Med 329: 10081012.
  • 25
    Sonne-Holm, S., Sørensen, TIA. (1986) Prospective study of attainment of social class of severely obese subjects in relation to parental social class, intelligence, and education. Br Med J (Clin Res Ed) 292: 586589.
  • 26
    Stunkard, A. J., Sørensen, TIA. (1993) Obesity and socioeconomic status—a complex relation. N Engl J Med 329: 10361037.
  • 27
    Jeffery, R. W., French, S. A. (1997) Preventing weight gain in adults: design, methods and one year results from the Pound of Prevention study. Int J Obes Relat Metab Disord 21: 457464.
  • 28
    Wardle, J., Griffith, J. (2001) Socioeconomic status and weight control practices in British adults. J Epidemiol Community Health 55: 185190.