Assessing the possible direct effect of birth weight on childhood blood pressure: a sensitivity analysis

Area of Science:

• Epidemiology
• Biostatistics
• Pediatric Health

Background:

• Estimating the direct effect of birth weight on blood pressure conventionally involves adjusting for current weight, a potential mediator.
• This adjustment can introduce collider-stratification bias, distorting the true relationship.
• Unmeasured common causes (U) can further confound this association.

Purpose of the Study:

• To evaluate the potential bias introduced by an unmeasured common cause (U) on the controlled direct effect of birth weight on blood pressure.
• To assess the impact of conditioning on current weight in the presence of unmeasured confounding.

Main Methods:

• Utilized data from a Swiss school-based study (n = 3,762; mean age = 12.3 years).
• Employed linear regression to analyze the association between birth weight and systolic blood pressure.
• Conducted sensitivity analyses to explore the influence of unmeasured common causes (U).

Main Results:

• A small negative association between birth weight and systolic blood pressure was initially observed (βbw = -0.3 mmHg/kg).
• This association strengthened upon adjustment for current weight (βbw|C = -1.5 mmHg/kg).
• Sensitivity analyses indicated that unmeasured confounding (U) could explain the conditional association only under specific conditions of strength and prevalence difference.

Conclusions:

• The observed negative association between birth weight and blood pressure, when adjusted for current weight, is not solely attributable to collider-stratification bias.
• Unmeasured factors may play a role, but require substantial association with blood pressure and differential prevalence across birth weight groups.
• Suggests a more complex interplay of factors influencing the birth weight-blood pressure relationship beyond simple mediation or collider bias.