Background Obesity is a major public health concern because of the associated increased risk of type 2 diabetes, cardiovascular diseases, and various types of cancer. The causes of obesity are multifactorial, including modifiable risk behaviours such as physical inactivity and unhealthy diet. Such risk behaviours are known to cluster, but the effect of this clustering on body mass index (BMI) is not well understood. In a population sample, we aimed to establish the extent of risk behaviour clusters in association with BMI.

Methods We used data from the 1958 British birth cohort to assess four risk behaviours at age 42 years: physical inactivity (versus active), diet (low versus high fruit/vegetable consumption), smoking (current and former versus never smokers) and alcohol consumption (non/infrequent and binge versus moderate drinkers). The prevalence of the combinations of these behaviours were assessed, as well as the association with BMI at age 45 years (4455 men, 4544 women).

Results Among men at age 42 years, 7.54% were inactive, 5.57% were current smokers, 12.03% were former smokers, 2.92% were non/infrequent drinkers, and 8.37% were binge drinkers. Among women, the prevalence was 9.02%, 6.76%, 12.83%, 6.69%, and 2.95%, respectively. All risk factors, except diet, were associated with increased BMI, hence diet was excluded from our examination of risk behaviour clusters. Examining the combinations of inactivity, smoking and alcohol consumption, we found that 22.4% of men and 24.3% of women did not engage in any of the risk behaviours; 36.4% of men and 38.2% of women in one; and the remainder (41.2% and 37.5% respectively) in ≥2. The most common risk behaviour combinations were inactive and former smokers (5.3% among men, 5.2% among women), inactive and current smokers (4.3% and 5.4%), binge drinking and former smokers (6.3% and 2.4%), binge drinking and current smokers (5.3% and 2.8%), and inactive and non/infrequent drinkers (2.1% and 4.1%). These combinations were associated with increased mean BMI compared to that of individuals with no risk behaviours: difference ranging from -0.55 to 1.34 kg/m² (men) and from 0.12 to 2.59 kg/m² (women).

Conclusion Over a third of a general population sample had two or more risk behaviours for obesity in mid-adulthood. Whilst patterns of multiple risk factor clustering have been established for disorders such as cardiovascular diseases, identification of clustered behaviours for obesity is complex, partly because of U-shaped or opposing directions of associations for different risk behaviours.