1 School of Public Health, Faculty of Medicine, Université de Montréal, Montréal, QC, Canada; Research Center of Centre Hospitalier Universitaire Sainte-Justine, Montréal, QC, Canada; Research Centre of Centre Hospitalier Universitaire de Montréal, Montréal, QC, Canada.
2 School of Public Health, Faculty of Medicine, Université de Montréal, Montréal, QC, Canada; Research Centre of Centre Hospitalier Universitaire de Montréal, Montréal, QC, Canada.
3 Ingram School of Nursing, Faculty of Medicine, McGill University, Montréal, QC, Canada.
4 Research Center of Centre Hospitalier Universitaire Sainte-Justine, Montréal, QC, Canada; Department of Family Medicine, Faculty of Medicine, McGill University, Montréal, QC, Canada.
5 School of Kinesiology and Physical Activity Sciences, Faculty of Medicine, Université de Montréal, Montréal, QC, Canada; Research Center of Centre Hospitalier Universitaire Sainte-Justine, Montréal, QC, Cana

Background: Uncertainty remains regarding the causal effect of physical activity and sedentary behaviours on the development of type 2 diabetes in children. We aimed to estimate average treatment effects of physical activity and sedentary behaviours on risk of type 2 diabetes in individuals who are at risk during childhood and adolescence.

Methods: We used data from the Quebec Adipose and Lifestyle Investigation in Youth (QUALITY) cohort of children of western European descent (white non-Hispanic race or ethnicity) with a parental history of obesity (defined as a BMI of 30 kg/m² or more, or a waist circumference of more than 102 cm in men and 88 cm in women) evaluated at the ages of 8-10 years (baseline), 10-12 years (first follow-up cycle), and 15-17 years (second follow-up cycle) in Québec, Canada. We measured moderate-to-vigorous physical activity (MVPA) and sedentary time by accelerometry, and leisure screen time by questionnaire at each cycle. Outcomes included fasting and 2 h post-load glycaemia and validated indices of insulin sensitivity and insulin secretion. We estimated average treatment effects of MVPA, sedentary time, and screen time on markers of type 2 diabetes using longitudinal marginal structural models with time-varying exposures, outcomes, and confounders from the ages of 8-10 to 15-17 years and inverse probability of treatment and censoring weighting. We considered both the current and cumulative effects of exposures on outcomes.

Findings: 630 children were evaluated at baseline (age 8-10 years) between July, 2005, and December, 2008, 564 were evaluated at the first follow-up (age 10-12 years) between July, 2007, and March, 2011, and 377 were evaluated at the second follow-up (age 15-17 years) between September, 2012, and May, 2016. Based on cumulative exposure results, estimated average treatment effects for MVPA were 5·6% (95% CI 2·8 to 8·5) on insulin sensitivity and -3·8% (-7·1 to -0·5) on second-phase insulin secretion per 10 min daily increment from the ages of 8-10 years to age 15-17 years. Average treatment effects for sedentary time and reported screen time resulted in reduced insulin sensitivity (-8·2% [-12·3 to -3·9] and -6·4% [-10·1 to -2·5], respectively), increased second-phase insulin secretion (5·9% [1·9 to 10·1] and 7·0% [-0·1 to 14·7], respectively), and higher fasting glycaemia (0·03 mmol/L [0·003 to 0·05] and 0·02 mmol/L [0·01 to 0·03], respectively) per additional daily hour from the ages of 8-10 years to 15-17 years.

Interpretation: Using modern causal inference approaches strengthened the evidence of MVPA and sedentary behaviours as key drivers of development of type 2 diabetes in at-risk children and adolescents, and should be considered as key targets for prevention.

Funding: Canadian Institutes of Health Research, Heart and Stroke Foundation of Canada, and Fonds de Recherche du Québec-Santé.

Translation: For the French translation of the abstract see Supplementary Materials section.