PT - JOURNAL ARTICLE AU - G David Batty AU - Ian J Deary AU - Sally Macintyre TI - Childhood IQ in relation to risk factors for premature mortality in middle-aged persons: the Aberdeen Children of the 1950s study AID - 10.1136/jech.2006.048215 DP - 2007 Mar 01 TA - Journal of Epidemiology and Community Health PG - 241--247 VI - 61 IP - 3 4099 - http://jech.bmj.com/content/61/3/241.short 4100 - http://jech.bmj.com/content/61/3/241.full SO - J Epidemiol Community Health2007 Mar 01; 61 AB - Objective: A series of studies have shown an association between high childhood IQ scores and reduced rates of total mortality in adulthood. Several mechanisms have been advanced to explain these relationships, including mediation via established risk factors. This study examines the association between childhood IQ and a range of established physiological and behavioural risk factors for premature mortality in adulthood. Design, setting and participants: In 1962, 12 150 children took part in a school-based survey when their IQ scores were extracted from educational records. When re-surveyed forty years later (n = 7183; 63.7% response), they self-reported information on risk factors for premature mortality (smoking, heavy alcohol consumption, obesity, height, hypertension and diabetes). Main results: In sex-adjusted analyses based on an analytical sample of 5340 (2687 women), higher childhood IQ scores were associated with a decreased prevalence of ever having smoked regularly in adulthood (ORper SD increase in IQ (95% CI): 0.77 (0.73 to 0.81)), heavy alcohol consumption (0.89 (0.84 to 0.94)), obesity (0.78 (0.72 to 0.83)) and overweight (0.86 (0.81 to 0.91)). Higher IQ scores were similarly related to a reduced prevalence of short stature and higher rates of smoking cessation in smokers; effects that were stronger in women (p value for interaction: ⩽0.04). Adjusting for indicators of early and, particularly, later-life socioeconomic circumstances led to heavy attenuation of these gradients with statistical significance at conventional levels lost in most analyses. Conclusions: The IQ–risk factor gradients reported may offer some insights into the apparent link between high pre-adult IQ and reduced mortality rates.