Non-evidence-based preconception information on Xiaohongshu and delayed assisted reproductive care attendance: an observational study with propensity score matching

<title>Abstract</title> <p> <bold>Background</bold> Infertility is recognised as a major global reproductive public health concern, and timely access to standardised assisted reproductive technology (ART) care is a core prerequisite for improving pregnancy outcomes. Social media has become integrated into the health information‑seeking pathways of people of reproductive age. As a leading domestic image‑and‑text social platform in China, Xiaohongshu serves as a core information source for individuals planning pregnancy. Nevertheless, non‑evidence‑based preconception content is widely disseminated on the platform, and the quantitative association between such content and delayed ART attendance among people with infertility remains unclear. <bold>Methods</bold> This cross‑sectional observational study was based on publicly available text content from the Xiaohongshu platform. A total of 470 valid samples were retrieved and enrolled. Two trained professionals conducted independent double‑blind coding. Six covariates were selected under the theoretical framework of the Health Belief Model (HBM). 1:1 propensity score matching (PSM) with a caliper width of 0.02 was performed to balance confounding between groups. McNemar’s test and conditional binary logistic regression were applied to examine the association between non‑evidence‑based information exposure and delayed care attendance. <bold>Results</bold> A total of 130 matched pairs were successfully generated via PSM. After matching, the standardised mean difference (SMD) for all covariates was below 20%, indicating satisfactory baseline balance between groups. The rate of delayed care attendance in the non‑evidence‑based exposure group was 30.77%, which was significantly higher than the 7.69% recorded in the evidence‑based control group (McNemar’s χ² = 22.29, <italic>P &lt; 0.001</italic> ). Non‑evidence‑based information exposure was significantly and positively associated with delayed care attendance (OR = 5.33, 95% CI: 2.53–11.23, <italic>P &lt; 0.001</italic> ). The median DISCERN score in the non‑evidence‑based group was 9, significantly lower than the median score of 15 in the control group (Mann–Whitney U = 6269, <italic>P</italic> &lt; 0.001). A moderate positive correlation was observed between information quality and delayed care attendance (r = 0.47, <italic>P</italic> &lt; 0.001). Stratified subgroup analyses by age and infertility duration, together with sensitivity analyses using multiple matching parameters, consistently confirmed the robustness of the primary findings. <bold>Conclusions</bold> Among active users planning pregnancy on Xiaohongshu, exposure to non‑evidence‑based preconception information has a statistically significant positive association with self‑reported delayed attendance for assisted reproductive care. This study provides quantitative empirical evidence for social media reproductive health governance and clinical reproductive health literacy interventions. </p>