ABO Blood Group Distribution and Infertility Risk in Southern China: A Retrospective Analysis of the Ganzhou Region

Blood type O was overrepresented in infertile couples (39.5% vs 34.2%), but effect sizes were weak and key findings didn't survive correction

Journal: Journal of Assisted Reproduction and Genetics | Published: 2025-12-19 | Type: Retrospective cohort | PMID: 41417294 Authors: Liu Min, Zhou Jiedong, Hu Shian, Ouyang Yong, Xiong Yuting, Zhao Ying (First Affiliated Hospital of Gannan Medical University, Ganzhou, China) Funding/COI: Funding not listed. Authors declare no competing interests.

Summary

A single-center retrospective study of 696 infertile couples in southern China found blood type O overrepresented among both infertile men and women compared to a 14.41-million-person regional reference population. O/O couples appeared more than expected (16.81% observed vs. 11.70% expected). A proposed sperm morphology gradient—where type O men had the worst morphology and AB men the best—did not survive correction for multiple comparisons. The authors' own disclosure buries the lede: the headline association evaporates under Bonferroni adjustment.

Claims

• Blood type O was present in 39.51% of infertile men vs. 34.21% in the reference population (P = 0.008 uncorrected; adjusted P = 0.064, not significant after Bonferroni correction)
• Blood type O was present in 41.67% of infertile women vs. 34.19% in the reference population (P < 0.001)
• AB was underrepresented in both infertile men (6.18% vs. 8.91%) and women (6.47% vs. 8.91%)
• O/O couple pairings: observed 16.81%, expected 11.70% under random pairing (O/E = 1.44, P < 0.001)
• Sperm morphology gradient across blood types: O < A < B < AB (P = 0.02 uncorrected)—explicitly noted by the authors as not surviving multiple comparison correction
• Effect size across all associations: Cramér's V = 0.134 (weak by convention)

Study Quality

This is a single-center retrospective study with 696 infertile couples drawn from a tertiary reproductive medicine center—an already-selected, help-seeking population. Comparisons were made against a massive 14.41-million-person reference dataset, but that dataset was not matched for age, BMI, socioeconomic status, environmental exposure, or any other variable that might confound fertility. The sample is large enough to generate statistically significant p-values even for trivial differences, which is exactly what appears to have happened: Cramér's V of 0.134 means ABO blood type explains roughly 1.8% of the variance in group membership. The authors' disclosure that the male distribution difference loses significance after Bonferroni correction is buried mid-results and contradicts the abstract's framing.

The sperm morphology finding is explicitly labeled exploratory in the discussion, with the authors acknowledging it didn't survive multiple comparison adjustment. That's scientifically honest—but it also means the paper's most mechanistically interesting finding is not a finding.

Red Flags

• Primary association (blood type O overrepresentation in infertile men) is not significant after multiple comparison correction (adjusted P = 0.064), yet the abstract implies statistical significance throughout
• No adjustment for confounders in the comparison against the reference population—age, BMI, region-specific diet, occupational exposures, etc. are uncontrolled
• Reference population (14.41 million) and study population (696 patients at one clinic) are not comparable cohorts; selection bias is structural, not incidental
• Sperm morphology gradient did not survive correction for multiple comparisons—author-acknowledged, but still highlighted in abstract and conclusion
• Proposed biological mechanisms (anti-A/anti-B antibodies attacking sperm; glycosyltransferase-mediated glycan variation) are speculative, with the authors confirming "no direct mechanistic evidence" exists
• Small subgroup counts for "protective" combinations (A/B, B/AB) limit statistical power for those findings
• No funding source disclosed; impossible to assess independent replication incentive

Strengths

• Authors are transparent about the Bonferroni correction failure and weak effect sizes—the paper reads more honestly than its abstract suggests
• Reference population (n ≈ 14.41 million, drawn from a 2022 multi-city Chinese dataset) is the most rigorous comparator available for this region
• Use of O/E ratio and Cramér's V provides effect size context beyond p-values alone
• Eight-year data window (2016–2024) with consistent inclusion/exclusion criteria
• Excludes azoospermia and chromosomal abnormalities, reducing etiological noise in semen analysis comparisons

Verdict

This paper reports weak associations with small effect sizes, at least one primary finding that doesn't survive standard statistical correction, and a mechanistic hypothesis with no supporting evidence. The authors are admirably candid about these limitations in the discussion—but the abstract overstates the case. Worth noting as a hypothesis-generating regional dataset, not as evidence that blood type meaningfully predicts infertility risk. The call for multicenter, prospective, confounder-adjusted studies is correct, and until those exist, this paper contributes a data point, not a conclusion.