This article summarises an analysis of how maternal BMI contributes to adverse pregnancy outcomes, the modifying effect of parity, and the difference between relative and absolute measures of risk.
The study used routinely collected data from the South Australian Perinatal Outcomes unit covering the years 2013 to 2017 inclusive. Records from 65,285 women were analysed to examine associations between BMI and a range of pregnancy outcomes. The analyses considered BMI both as standard BMI categories and as a continuous variable. The investigators reported both relative measures (adjusted relative risk, aRR) and absolute measures (adjusted risk difference, aRD; absolute risk estimates) and explicitly examined whether parity modified associations. The authors also discussed the potential influence of unmeasured confounding on observed associations.
When modelled as a continuous effect, a 5 kg/m2 increase in maternal BMI was associated with increased adjusted relative risks for several outcomes. For gestational diabetes, the aRR per 5 kg/m2 increase was reported between 1.3 and 1.4. For pre-eclampsia, the aRR per 5 kg/m2 increase differed by parity: 1.13 in nulliparous women and 1.43 in multiparous women. These relative measures show that increasing BMI is associated with higher proportional risk for these outcomes.
Although relative risks increased with higher BMI, the absolute risk differences reported were small for some outcomes. The adjusted risk difference for gestational diabetes associated with a 5 kg/m2 BMI increase was approximately 3%, irrespective of parity. For pre-eclampsia the aRD was reported as less than 0.1% for all women. The authors also provided absolute risk estimates: at a BMI of 20 the absolute risk of pre-eclampsia was 0.7% for nulliparous women and 0.2% for parous women, which the report states increased to 0.9% for nulliparous and 0.4% for parous women (values are those reported in the source). These absolute figures illustrate that even with measurable relative increases, the overall probability of some outcomes remains low.
The analysis found that parity substantially modifies the relationship between BMI and many pregnancy outcomes. The magnitude and sometimes the direction of risk differences differed between nulliparous and multiparous women. The authors illustrate this with an example: a woman with parity 1 and a BMI of 30 has about the same risk of pregnancy-induced hypertension, and a lower risk of pre-eclampsia, compared with a nulliparous woman with a BMI of 20. This emphasises that parity should be considered when communicating BMI-associated risks.
The authors note that while associations between higher BMI and many adverse outcomes were observed, a substantial portion of the apparent effects could be due to unmeasured confounding. The report cautions that routinely collected observational data are vulnerable to residual confounding from variables not captured or imperfectly measured in the dataset. The study does not claim causal proof and explicitly acknowledges this limitation.
The authors highlight two reporting issues that can mislead clinical interpretation:
Categorisation of BMI can mask risk gradients and may lead to underestimation of risk for some individuals and overestimation for others compared with continuous modelling.
Reporting only relative measures of effect (for example, aRR) without concurrent absolute risks can create an impression that BMI-associated risks are more dramatic than they are in absolute terms. The combination of relative and absolute measures provides a clearer clinical picture for counselling.
Given the small absolute increases for some outcomes despite higher relative risks, the authors argue for nuanced communication of risk to clinicians and pregnant women, incorporating both parity and absolute risk estimates.
In this large population dataset (65,285 women), increased maternal BMI was associated with higher relative risks for outcomes such as gestational diabetes and pre-eclampsia, but absolute risk increases were often small. Parity modifies many of these associations, and standard reporting practices—BMI categorisation and sole use of relative risk—can misrepresent individual risk. The authors conclude that unmeasured confounding cannot be ruled out, and they call for more nuanced presentation of BMI-related pregnancy risks to improve interpretation and counselling.
Keywords: body mass index; overweight and obesity; parity; pregnancy outcomes; risk.