Contribution of Midparental BMI and Other Determinants of Obesity in Adult Offspring




The aim of this study was to evaluate midparental BMI among intergenerational factors associated with obesity in adult offspring. The data was from an unusual two-generational observational design of 1,477 married couples from Renfrew and Paisley in Scotland who were aged 45–64 years when screened in 1972–1976, and 1,040 sons and 1,298 daughters aged 30–59 years when screened in 1996. BMI was categorized as normal (<25 kg/m2), overweight (25–29.9 kg/m2), and obese (≥30 kg/m2) in offspring and parents. Midparental BMI was defined as the mean of the mother's and father's BMI. Low physical activity, nonsmoking status, higher cholesterol level, and manual social class were all associated with increased BMI in offspring. The effect of reported dietary intake was less clear. Offspring of obese parents (defined by midparental BMI) were over four times more likely to be obese than offspring of normal weight parents. Midparental BMI had a strong effect on offspring BMI, independent of social class, smoking habit, physical activity, and reported dietary intake. Adding midparental BMI to the regression model more than doubled the explained variation of offspring BMI from 7.7 to 17%. Every 1 kg/m2 increment in midparental BMI was associated with a BMI greater by 0.51 kg/m2 in offspring. We conclude that midparental BMI is a useful simple tool to predict offspring BMI. Whether it represents genetic or environmental family effects, it is easily ascertained by the individual and could be used in health promotion and clinical settings to target individuals who are at increased risk of becoming obese.


The prevalence of obesity is increasing in many populations, as average nutritional intakes of energy exceed average energy expenditure. The reason why some people become obese, while others do not, despite similar excesses of energy intake over expenditure, is partly attributed to increased genetic susceptibility (1). Although more than 200 genetic factors have been postulated as having a role in the pathophysiology of obesity, the precise role of genetic factors, and their interaction with behavior and environmental factors, is unclear. Single gene defects are extremely rare and polygenic influences account for ∼20% of variance in BMI (2).

In this study, we analyzed some environmental, behavioral, and familial factors associated with obesity in adult offspring in an unusually informative two-generation study.

Methods and Procedures

In 1972–1976, residents of Renfrew and Paisley (7,049 men and 8,353 women), comprising 79% of the general population aged 45–64 years and including 4,064 married couples, completed a questionnaire and attended a cardiorespiratory examination (3). In 1996, the second component of the study took place, screening the adult offspring of the married couples. A total of 3,202 offspring from 1,767 families lived locally and formed the eligible population for this study. Details of the study have been described previously (4). A total of 1,040 sons and 1,298 daughters aged 30–59 years from 1,477 families participated in the 1996 study (response rate of 73% for individuals 84% for families).

In the parental study, height without shoes was measured to the nearest centimeter and weight was measured to the nearest kilogram. In the offspring study, height was measured without shoes using a Holtain stadiometer. A single measurement was made after participants inhaled and stretched to reach their maximum height. Measurements were recorded to the nearest millimeter. Weight was measured to the nearest 0.1 kg using Seca digital scales (Hamburg, Germany) in stockinged feet and wearing indoor clothes. BMI was calculated as the weight divided by the square of height and classified into three categories, according to World Health Organization criteria: <25 kg/m2, normal weight and underweight; 25–29.9 kg/m2, overweight; and ≥30 kg/m2, obese. BMI categories were calculated for sons, daughters, mothers, and fathers. Also the midparental BMI was calculated as the mean of each offspring's mother's and father's BMI. Plasma cholesterol was measured in a nonfasting venous blood sample (5).

The offspring completed a questionnaire which included questions on smoking habit, occupation, physical activity, and diet. Three categories of smoking habit were defined: never smoker, current smoker, and former smoker. Social class was coded from the occupation, using the Registrar General's classification of occupation (6). This classifies occupations into six categories denoted by Roman numerals with class III further subdivided into nonmanual and manual. Social class I represents professional and similar occupations, social class II managerial and technical occupations, social class III nonmanual skilled nonmanual occupations, social class III manual skilled manual occupations, social class IV partly skilled occupations, and social class V unskilled occupations. Women's own occupation was used unless they were housewives/homemakers and did not give a previous occupation, in which case their husband's or father's occupation was used. Nonmanual social class was defined as social classes I, II, or III nonmanual, and manual social class was defined as social classes III manual, IV, or V Participants were asked to report their daily activity as one of four categories: very physically active, fairly physically active, not very physically active, or not at all physically active. They also reported how often per week they were physically active in their nonworking time, based on the then current UK guidelines (more than three times, two to three times, once, less than once, and never). These categories were combined. Participants who were not very physically active or not at all physically active during their usual daily activities and who were physically active outside work less than once a week or never, were classified as having no exercise. Nutrient intake was estimated using an established food frequency questionnaire (7,8) which comprised questions on the weekly frequency of consumption of commonly eaten foods (9,10). Nutrient intakes were calculated by a computer program, which multiplied the food frequency by standard portion size, and by nutrient values from UK food composition tables.

Statistical procedures were performed using STATA release 9 (Stata-Corp, College Station, TX), adjusting for clustering of offspring within families. Means and percentages were standardized by 5-year age group using the distribution of sons and daughters combined. Tests for trends across offspring or midparental BMI categories were calculated using regression analysis (continuous variables) or logistic regression analysis (discrete variables), with offspring or midparental BMI and age as continuous variables. Regression analysis of the correlates of offspring BMI used only participants with complete data on all the variables used and included interaction terms between each risk factor and sex to test for any differences in effects between the sexes.

The sample size was 2,338. BMI was missing for 17 daughters, giving 2,321 participants. In addition, one son and two daughters had fathers with missing BMI, two sons and one daughter had mothers with missing BMI, leading to two sons and three daughters with missing midparental BMI. After excluding offspring with missing data used in the regression analyses, there were 2,162 participants for those analyses.

Ethical approval for the study was obtained from the Local Research Ethics Committee of relevant Health Boards.


The prevalences of overweight and obesity were respectively 44.2 and 17.9% in sons, and 31.6 and 18.1% in daughters (Table 1). There was an increasing trend in age with BMI category in sons but not in daughters. Obese sons and daughters were more likely to be in manual social classes. Lighter offspring were more likely to be current smokers and less likely to be former smokers. There were strong relationships with physical activity in usual daily activities, with the heaviest offspring being more likely to do no exercise. Physical activity outside work did not show clear relationships, but the combination of the two measures showed that normal weight offspring were least likely to report physical inactivity, followed by overweight offspring, with obese offspring being most likely to report physical inactivity. There were no clear relationships between reported total daily energy intake, fat intake, and carbohydrate intake, although there was a suggestion of an inverse relationship between carbohydrate intake and BMI. In both sons and daughters, there was a positive relationship between reported protein intake and BMI. Plasma cholesterol level was positively related to BMI.

Table 1.  Age adjusted characteristics of sons and daughters by BMI category
inline image

The prevalence of obesity was 8.6% in sons of parents with normal weight (based on midparental BMI), 20.3% in sons of overweight parents, and 43.8% in sons of parents who were obese (Table 2). The corresponding figures for daughters were 9.2, 20.5, and 42%, respectively. Thus, sons of obese parents were five times more likely to be obese than sons of normal weight parents and daughters of obese parents were over four times more likely to be obese than daughters of normal weight parents. Only 18.4% of obese sons had normal weight parents, while 60.5% had overweight parents and 21.1% had obese parents. For all sons, irrespective of personal BMI, 38.2% had normal weight parents, 53.2% had overweight parents, and 8.6% had obese parents. Similar proportions were seen for daughters; therefore, obese offspring had half the prevalence of normal weight parents and double the prevalence of obese parents. There were very few normal weight sons and daughters who had obese parents (3.3 and 4.5%). Similar patterns were observed between sons and daughters and their fathers and mothers.

Table 2.  Distribution of offspring BMI category by parental BMI categories
inline image

Midparental BMI category was the same as mother's BMI category for 70.2% of mothers, higher in 14.6% and lower in 15.2%. Equivalent percentages for fathers were 66.9, 14, and 19.1%. Alternatively, analyzing with parents in three categories according to the number of obese parents (0, 1, or 2), 14.1, 25.6, and 63.2% of offspring were obese.

Regression analyses of correlates of offspring BMI in 2,162 sons and daughters with complete data are shown in Table 3. Model 1 showed that manual social class and current smoking were significantly associated with BMI. The contribution of current smoking was negative, whereas that of manual social class was positive. Adding the combined exercise variable, and reported fat, protein, and carbohydrate intakes to the model (model 2) showed that fat did not contribute to the model, protein contributed positively, and carbohydrate contributed inversely. Doing no exercise was associated with increased BMI. These additional variables on diet and exercise increased the explained variation in BMI from 3.6% in model 1 to 7.7%. In model 3, midparental BMI was positively associated with offspring BMI and increased the explained variation to 17%. Every 1 kg/m2 increment in midparental BMI was associated with a BMI greater by 0.51 kg/m2 in offspring. There was a significant interaction between sex and the exercise variable in model 3. Regression analyses performed for sons and daughters separately showed the coefficient of the exercise variable to be small and nonsignificant for sons (0.48 (95% confidence interval −0.09 to 1.04), P = 0.10), and larger and statistically significant for daughters (1.38 (95% confidence interval 0.77–2.0), P < 0.0001).

Table 3.  Regression analysis of correlates of BMI (N = 2,162)
inline image

Fewer correlates of BMI in sons and daughters were associated with midparental BMI than with their own BMI (data not shown). Sons' age, proportion in manual social classes, proportion of participants who were former smokers, and protein intake were positively associated with midparental BMI. In daughters, the proportion in manual social classes and protein intake were positively associated with midparental BMI.

Correlates of contrasting parental BMI in obese offspring

Obese offspring of obese parents reported higher mean intakes of total protein than obese offspring of normal-weight parents (95.3 vs. 87.4 g/day, P = 0.026) and included more current smokers (28.9% vs. 11.7%, P = 0.005) (Table 4). Obese offspring of normal weight parents reported less exercise outside work (70.5% vs. 50.4%, P = 0.007) and less exercise with the two measures combined (46.4% vs. 28.6%, P = 0.026) than obese offspring of obese parents.

Table 4.  Age adjusted characteristics of offspring by own and midparental BMI category
inline image

Correlates of obesity in offspring with obese parents

Obese offspring of obese parents had higher cholesterol levels (5.48 vs. 4.77 mmol/l, P < 0.0001) and reported less physical activity in usual daily activities (47.7% vs. 28.0%, P = 0.034) than offspring of obese parents with normal BMI.

Correlates of obesity in offspring with normal-weight parents

Obese offspring of normal weight parents reported fewer current smokers (11.7% vs. 30.6%, P < 0.0001), more former smokers (34.8% vs. 20%, P = 0.009), and less physical activity by all three measures (P = 0.001, P = 0.001 and P < 0.0001, respectively) than offspring with normal BMI of parents with normal BMI. Although there were no differences in reported dietary intake, obese offspring had significantly higher cholesterol levels than normal weight offspring (5.54 vs. 5.13 mmol/l, P = 0.0002).


This study included a limited number of measures among which we found that manual social class, smoking habit, physical activity, cholesterol level, and protein intake were associated with obesity. Cross-sectional analyses in the Scottish Health Survey have revealed similar influences on BMI and waist from age and smoking status (11). This study had an unusual two-generation design, which allowed familial influences to be separated from secular time trends in BMI. Midparental BMI was important with sons of obese parents being five times more likely to be obese than sons of normal weight parents, and daughters of obese parents being four times more likely to be obese. The addition of midparental BMI to the regression model more than doubled the explained variation of BMI to 17%.

Obese offspring of obese parents had the highest reported energy intake of all offspring/parental groups, although this was only statistically significantly higher than one group (normal weight offspring of overweight parents, P = 0.03). Offspring of normal weight but with obese parents reported more physical activity and had lower cholesterol than obese offspring of obese parents. Similarly, although most offspring of normal weight parents were also of normal weight, the minority who were obese reported less physical activity, less current and more former smoking, and had higher cholesterol. These findings are consistent with established explanations of obesity, but appear insufficient to explain the high prevalence of obesity in some families fully, suggesting that additional explanations, such as genetic factors may also have a role to play.

Benefits of the study design: The design of this study, with offspring aged 30–59 years, maximized the potential for predispositions to normal weight and obesity in adulthood to be expressed. A weakness of many family studies is that offspring are too young for predispositions to be fully expressed (12,13). Measurements of weight and height were taken in clinical settings, and not self-reported.

Limitations: Behaviors were all reported, and likely to be biased, particularly by the systematic underreporting of food intakes by people who are overweight and obese (14). Obese people may underreport fatty foods and foods rich in carbohydrates rather than their total dietary intake (15). Samaras et al. reported up to 65% of obese women underreporting their energy intake (16). For underreporters, the reported diet was lower in fat, similar in carbohydrates, and higher in protein, as a percentage of energy intake (16). Lara et al. found that over half of a group of women said they would intentionally mis-report food intake (17). An additional limitation is that in general, and also specifically with our instrument, food frequency questionnaires do not measure portion size, but assume an average-sized portion for everyone. This means that the nutrient intakes calculated will underestimate the intakes of participants who have larger portion sizes. These are more likely to be in overweight or obese individuals. Food frequency questionnaires also measure diet at screening, at one point in time, not diet earlier, including childhood and adolescence when many of the dietary habits were established. Despite this, food frequency questionnaires may be the most practicable method for studying the eating patterns of populations (18).

Midparental BMI benefited from being measured accurately compared to the self-reported diet and exercise variables. This could contribute to midparental BMI being a successful measure. Midparental BMI can mask parents having different BMI categories, for example one parent of normal weight and one obese could give the midparental BMI category of overweight, but ∼70% were in the correct categories. However, it is a good measure encompassing both parents, which is easily ascertained by the individual and could be used in health promotion and clinical settings to target individuals who are at increased risk of becoming obese. More regular monitoring of height and weight in the primary care setting would be of value in predicting future obesity of individuals and of benefit to their offspring. We also observed strong increases in offspring obesity when looking at whether neither, one, or both parents were obese. Recent discovery of the association of a common variant in the FTO gene with BMI and obesity could similarly identify individuals susceptible to weight gain (19), but would be more complex than simply asking about parental BMI.

The importance of midparental BMI to offspring BMI could be due to genetic or environmental factors, or both. Eating habits, learnt in the family, involve such variables as suitable food types, portion size, snacking, and eating outside the home (14). The Victorian Family Heart Study suggested that familial covariation in height was mostly due to genetic factors, but correlations between weight and BMI in first degree relatives varied, suggesting a role for shared environmental factors such as diet and physical activity (12). Physical activity, with regard to both walking and use of public transport, rather than using the private car, and to recreational exercise, is also likely to be influenced by the family (20). Physical inactivity, such as television watching, is also likely to be influenced by family norms with more television watching also contributing to increased energy intake through snacking (20,21,22).

The observed aggregation of normal weight within some families, and obesity within others, could be explained by shared environments and behaviors, shared genetic susceptibility, or a combination of these factors. This finding may be important for intervention and treatment programs, where identification of individuals at risk of obesity and the identification of those likely to be resistant to dietary intervention and hence, requiring perhaps, specialist advice is important. Midparental BMI is a powerful, largely unexplained determinant of BMI in offspring and further investigation of families, including genetic factors is warranted.


Victor Hawthorne initiated the original study. The offspring study was supported by grants from the Wellcome Trust and the National Health Service Research and Development Programme. C.L.H. was supported by VHS Health Scotland.


The authors declared no conflict of interest.