Several studies have found a non-linear relationship between mental ill-health and BMI with higher rates in both the underweight and the obese. This study evaluated the shape of the relationship between BMI and distress, suicidal ideation and self-reported mental ill-health conditions in a large population sample.
Data were drawn from the South Australian Monitoring and Surveillance System (SAMSS) for the years 2002 to 2009 (n=46,704). SAMSS monitors population trends in state and national risk factors and chronic diseases. Samples are drawn from all households with a functioning number in the Australian White Pages. Computer assisted telephone interviews collected information on self-reported height and weight, demographic and health behaviours. Respondents completed the Kessler Distress and suicidal ideation scales and reported specific mental ill-health conditions. BMI was categorized into deciles to allow for assessment of the shape of any associations with other variables. Logistic regression was used to examine associations between each mental ill-health condition and BMI-decile controlling for age in the base model. This was followed by a full model that added SES and the health-adverse coping behaviours of smoking, alcohol and physical activity to test for changes from the base model.
Non-linear associations were observed between BMI-decile and mental ill-health but statistically significantly greater odds of mental ill-health were observed only in the obese and not in the underweight after controlling for age, health-adverse behaviours and socioeconomic status. The association between BMI and mental ill-health might best be described as 'threshold'. Elevated odds were apparent for middle-aged persons, whereas younger and older individuals had a significantly lower odds of having a mental ill-health condition.
In conclusion, this study has provided no support for the hypothesis of increased mental ill-health problems in the underweight but it has demonstrated the non-linear relationships between BMI and mental ill-health and between BMI and health-adverse behaviours. Non-linear relationships with BMI need to be recognized and addressed during analysis.
Read Full Text (PDF)
