• The experience of pregnancy is associated with physiological and psychological change that is shown to promote sedentary behaviour and/or lower levels of physical activity (PA).
• PA in pregnancy has historically been deemed a risk factor for miscarriage, though this may result from reverse causality. Those feeling well enough to stay active early in pregnancy might already be in the early stages of miscarriage and thus experience fewer symptoms like nausea and fatigue. The perception of Pa as a risk may have led to the promotion of sedentary behaviour and reduced PA during pregnancy by clinicians and medical professionals in fear of causing harm.
• The majority of epidemiological analyses to-date, however, have reported that higher PA and reduced sedentary time are associated with improved health outcomes for pregnant people and their offspring. Furthermore, guidelines in several countries encourage safe levels of activity and exercise during pregnancy.
• Since PA in pregnancy may have benefits for the mother and/or offspring, it is important to better understand its effects in pregnancy on a range of pregnancy and perinatal outcomes.
• When conducting conventional multivariable regression analyses, sources of bias induced by confounding factors, including confounding by undiagnosed existing disease (reverse causation), present challenges when inferring causality.
• Mendelian randomisation (MR) exploits the random distribution of genetic variants from parents to offspring, independent of the influence from other traits, to reduce susceptibility to these confounding factors. However, MR may be biased by weak instruments or horizontal pleiotropy.
• Negative control designs, which use paternal exposures as the negative control, can reveal bias in associations of maternal exposures with adverse pregnancy and perinatal outcomes. The paternal exposure is unlikely to affect these outcomes but may be associated with unmeasured confounders in a similar way to the maternal exposure.
• Employing a triangulation approach using multivariable regression, Mendelian randomisation, and a paternal negative control design, this investigation seeks to explore the causal effects of PA in pregnancy on pregnancy and perinatal outcomes.