Investigating and quantifying selection bias in research using placental pathology samples

Area of Science:

• Obstetrics and Gynecology
• Perinatal Medicine
• Pathology

Background:

• Placental pathology is crucial for understanding pregnancy outcomes.
• Provider-requested placentas are convenient for research but may introduce selection bias.
• Quantifying this bias is essential for accurate interpretation of findings.

Purpose of the Study:

• To compare placental pathology prevalence between provider-requested and non-requested samples.
• To quantify selection bias in research using requested placentas.
• To assess the impact of bias on associations between placental pathology and adverse pregnancy outcomes.

Main Methods:

• Prospective collection and examination of 575 placentas.
• Categorization of pathology into acute inflammation (AI), chronic inflammation (CI), fetal vascular malperfusion (FVM), and maternal vascular malperfusion (MVM).
• Comparison of pathology distribution and calculation of odds ratios (OR) and relative odds ratios (ROR) for preterm birth (PTB) and small for gestational age (SGA) infants.

Main Results:

• Maternal vascular malperfusion (MVM) prevalence was significantly higher in requested placentas (15% vs. 8%).
• MVM was associated with increased odds of PTB and SGA in both whole and requested cohorts.
• Relative odds ratios (ROR) indicated a 9% overestimation for PTB and 12% for SGA due to selection bias.

Conclusions:

• Maternal vascular malperfusion (MVM) is more prevalent in provider-requested placental samples.
• Selection bias in requested placentas can modestly overestimate associations with adverse pregnancy outcomes.
• Research findings remain largely consistent despite potential overestimation from convenience sampling.