The association of intelligence with morbidity, mortality and disability pension : epidemiological studies in a cohort of Swedish men

Abstract

Intelligence measured early in life is associated with various health outcomes later in life, but the mechanisms underlying the associations are not fully known. The overarching aim of the thesis was to increase the understanding of such associations, and in particular to investigate some potential mechanisms that might explain them.

The Swedish conscription cohort 1969-70 was used in all four studies. This cohort of 49 321 men, born 1949-51, was linked to several nation registers with information spanning from the men‘s childhood up to age 59. The exposure variable, intelligence, was measured at conscription when the men were 18-20 years old. Studies I-III were based on the full cohort. Study IV was based on a subsample comprising the 2156 men who were diagnosed with coronary heart disease (CHD) after 1990. The associations between intelligence and the primary outcomes were analysed in survival analysis, using Cox proportional hazards model, or by logistic regression. Potential explanatory factors (confounders and mediators) were retracted from conscription, national populations and housing censuses, the LOUISE/LISA registers and other national registers.

In general, lower intelligence in early adulthood was associated with poorer outcomes. This includes most indicators of health status in early adulthood (study III, in cross-sectional analyses), suicide and suicide attempt up to age 57 (study I), disability pension granted between age 40 and 59 (study II), case-fatality in first CHD between age 40 and 58 (study IV) and mortality up to age 59 among the men who survived a first CHD (study (IV). In studies I and II, the associations were graded across the intelligence span. However, no association between intelligence and recurrence in CHD was found (study IV), and some diagnoses in early adulthood were not, or positively, associated with intelligence (study III). Psychiatric diagnoses showed the strongest association with intelligence compared to all somatic diagnoses and health indicators. Characteristics and health indicators measured in early adulthood, including personality aspects, an indicator of social problems, substance use and BMI, attenuated the associations in all analyses where such data were introduced. In addition, socioeconomic and social factors in adulthood contributed in attenuating the associations. However, childhood socioeconomic factors had no or minimal impact on the associations.

The results add to, and extend, previous findings of associations between intelligence early in life and health outcomes across the life span. The associations between intelligence and disability pension in late middle age and several diagnoses and health indicators in youth had not been investigated previously. Of the covariates studied, individual characteristics and health behaviours in youth, and socioeconomic circumstances in adulthood, were the most important contributors to the associations in this male cohort. Although causality cannot be determined, the findings are compatible with the notion of an accumulation of risks associated with intelligence across the life course. The findings also suggest that intelligence might not be as important when a chronic disease such as CHD is already established.