ImportanceEicosanoids have a pathophysiological role in atopic dermatitis (AD), but it is unknown whether this is affected by prenatal ω-3 long-chain polyunsaturated fatty acid (n-3 LCPUFA; ie, fish oil) supplementation and genetic variations in the cyclooxygenase-1 (COX1) pathway.ObjectiveTo explore the association of n-3 LCPUFA supplementation during pregnancy with risk of childhood AD overall and by maternal COX1 genotype.Design, Setting, and ParticipantsThis prespecified secondary analysis of a randomized clinical trial included mother-child pairs from the Danish Copenhagen Prospective Studies on Asthma in Childhood 2010 birth cohort, with prospective follow-up until children were aged 10 years. In the trial, maternal and child COX1 genotypes were determined, and urinary eicosanoids were quantified when the child was 1 year of age. The present study was conducted from January 2019 to December 2021, and data were analyzed from January to September 2023.InterventionA total of 736 pregnant women at 24 weeks’ gestation were randomized 1:1 to 2.4 g of n-3 LCPUFA (fish oil) or placebo (olive oil) per day until 1 week post partum.Main Outcomes and MeasuresRisk of childhood AD until age 10 years overall and by maternal COX1 genotype.ResultsAt age 10 years, 635 children (91%; 363 [57%] female) completed the clinical follow-up, and these mother-child pairs were included in this study; 321 (51%) were in the intervention group and 314 (49%) in the control group. Pregnancy n-3 LCPUFA supplementation was associated with lower urinary thromboxane A2 metabolites at age 1 year (β, –0.46; 95% CI, –0.80 to –0.13; P = .006), which was also associated with COX1 rs1330344 genotype (β per C allele, 0.47; 95% CI, 0.20-0.73; P = .001). Although neither n-3 LCPUFA supplementation (hazard ratio [HR], 1.00; 95% CI, 0.76-1.33; P = .97) nor maternal COX1 genotype (HR, 0.94; 95% CI, 0.74-1.19; P = .60) was associated with risk of childhood AD until age 10 years, there was evidence of an interaction between these variables (P < .001 for interaction). Among mothers with the TT genotype, risk of AD was reduced in the n-3 LCPUFA group compared with the placebo group (390 mother-child pairs [61%]; HR, 0.70; 95% CI, 0.50-0.98; P = .04); there was no association for mothers with the CT genotype (209 [33%]; HR, 1.29; 95% CI, 0.79-2.10; P = .31), and risk was increased among offspring of mothers with the CC genotype (37 [6%]; HR, 5.77; 95% CI, 1.63-20.47; P = .007). There was a significant interaction between n-3 LCPUFA supplementation and child COX1 genotype and development of AD (P = .002 for interaction).Conclusions and RelevanceIn this secondary analysis of a randomized clinical trial, the association of prenatal n-3 LCPUFA supplementation with risk of childhood AD varied by maternal COX1 genotype. The findings could be used to inform a personalized prevention strategy of providing supplementation only to pregnant individuals with the TT genotype.Trial RegistrationClinicalTrials.gov: NCT00798226