The study, led by Liqiong Liu and Ling Chen of Southern Medical University’s Shenzhen Hospital together with colleagues across several Shenzhen institutions, recruited pregnant women during the antenatal period and tracked them through three time points. In total, 1,123 women completed the initial antenatal assessment, designated T0 in the study design, of whom 1,110 formed the valid baseline cohort. Of these, 984 returned for an assessment within 72 hours after delivery, a window the researchers labeled T1. Finally, 747 women completed the Birth Memories and Recall Questionnaire, known as BirthMARQ, at 6 to 12 weeks postpartum, the T2 endpoint where the memory outcomes were measured. This three-stage architecture is what gives the study its prospective power: because the key predictors were captured before or immediately around the birth, the researchers could examine how antecedent factors relate to later memory rather than relying on retrospective reports, which are notoriously vulnerable to distortion by current mood.

The BirthMARQ instrument is central to the study’s novelty. Rather than collapsing birth experience into a single score, it separates memory into six parallel domains, including emotional memory, sensory memory, recall, and coherence, among others. The research team treated each of these six domains as a separate outcome variable, running parallel statistical analyses for each one. This design choice matters because it allows the data to answer a question that global measures cannot: does a given prenatal factor, such as fear of childbirth or depressive symptoms, color the emotional intensity of a birth memory, the vividness of its sensory details, or the ease with which the memory comes to mind—or some combination of these that differs across dimensions?

To sift through a large pool of candidate predictors without overfitting their models, the researchers employed a modern variable-selection strategy. They compiled 36 candidate factors measured before the T2 memory assessment, spanning antenatal psychological measures, obstetric characteristics, delivery-related events, and early postpartum variables. For each of the six memory domains, they applied stacked least absolute shrinkage and selection operator regression—LASSO, in the standard shorthand—with participant-grouped 10-fold cross-validation. LASSO works by penalizing the inclusion of weak predictors, effectively shrinking irrelevant coefficients to zero and retaining only the variables that earn their place in the model. Grouping the cross-validation folds by participant guards against information leakage between correlated observations. After selection, the team fitted linear models in each of 20 multiply imputed datasets and pooled the results using Rubin’s rules, the standard technique for combining estimates across imputations while properly accounting for uncertainty introduced by missing data.

Because the study tested many predictors across six outcomes, the risk of false positives was substantial. The researchers addressed this with Benjamini-Hochberg false-discovery-rate adjustment, a statistical procedure that controls the expected proportion of erroneous discoveries among the declared significant findings. After this correction, a clear and coherent pattern emerged. Higher prenatal childbirth-fear scores were associated with higher emotional-memory scores, with an unstandardized coefficient of 0.180 and a 95 percent confidence interval running from 0.120 to 0.240. In other words, women who entered the delivery room more afraid of childbirth tended, weeks later, to report birth memories with a stronger emotional charge. Conversely, higher birth satisfaction was associated with lower emotional-memory scores, with a coefficient of −0.310 and a confidence interval from −0.398 to −0.221. The inverse direction suggests that satisfied mothers’ memories, while positive, may be less emotionally saturated—a finding that invites deeper investigation into how valence and intensity dissociate in autobiographical memory.

Complications affecting the infant left their mark on two separate dimensions. Women whose babies experienced complications showed higher sensory-memory scores, with a coefficient of 2.434 and a confidence interval from 1.003 to 3.866, and higher recall scores, with a coefficient of 1.275 and a confidence interval from 0.568 to 1.981. This dual signature fits with what memory science would predict: emotionally significant, threatening events tend to be encoded with heightened perceptual detail and retrieved more readily, a phenomenon related to the memory-amplifying effects of stress hormones on the amygdala and hippocampal systems. Prenatal depressive symptoms also emerged as a predictor, with higher antenatal depression scores associated with higher recall scores, at a coefficient of 0.098 and a confidence interval from 0.040 to 0.157. All of these associations survived the false-discovery-rate correction, lending them statistical robustness.

Not every dimension of birth memory proved tractable to prediction. No factor was selected for the coherence domain, meaning that the LASSO procedure found no prenatal, obstetric, or early postpartum variable among the 36 candidates that reliably forecast how coherent a woman’s birth narrative would be at 6 to 12 weeks. This null result is itself informative: it suggests that narrative coherence may depend on factors outside the measured set, on processes unfolding after the early postpartum window, or on idiosyncratic factors that resist population-level prediction. The authors also noted that concurrent birth-trauma ratings, collected at the same time as the memory questionnaire, were associated with several memory domains—but because trauma and memory were measured simultaneously, the temporal direction of those links could not be established, a limitation the prospective design was specifically built to avoid for the earlier predictors.

The study’s handling of attrition and missing data reflects careful methodological hygiene. Women who completed the final assessment and those who dropped out differed mainly in education and parity, while prenatal psychological measures showed little imbalance between the groups—a reassuring sign that differential attrition was unlikely to have systematically biased the psychological findings. Multiple imputation with 20 datasets allowed the analysis to retain information from participants with incomplete records rather than discarding them, and the pooling of post-LASSO estimates across imputations preserved honest standard errors throughout.

What do these findings mean in practice? The authors are careful to frame their conclusions as hypothesis-generating rather than practice-changing. The study, they write, adds prospective evidence from Chinese women that antecedent factors relate differently to distinct childbirth-memory dimensions, extending a literature that has focused predominantly on global birth appraisals and trauma-related outcomes. The domain-specific associations they identify—fear of childbirth with emotional memory, infant complications with sensory memory and recall, prenatal depression with recall—constitute concrete targets for longitudinal replication and for the evaluation of antenatal support programs and maternity care interventions. They explicitly caution that the findings do not establish intervention effects or justify immediate changes in clinical practice. Still, the implications are tantalizing: if prenatal fear reliably amplifies the emotional intensity of birth memories, then screening for fear of childbirth during pregnancy and offering targeted psychological support before delivery could, in principle, alter the memory trace women carry out of the delivery room.

The research also carries broader significance for the science of autobiographical memory. Childbirth offers a rare natural laboratory: an event that is universally significant, precisely datable, and rich in sensory and emotional content, occurring in a population that can be followed prospectively with relative ease. By demonstrating that the antecedents of memory are dimension-specific, the study challenges the assumption that a single appraisal of an event governs everything remembered about it. Fear, satisfaction, depression, and perinatal complications each sculpt a different facet of the recollection, much as separate chisels shape different contours of the same stone. As the authors and their collaborators within the INTERSECT Consortium continue this line of work, the six-dimensional map of childbirth memory they have begun to draw may prove valuable not only for improving maternity care but for understanding how the human mind transforms intense lived experience into lasting memory. The study was approved by the Medical Ethics Committee of Shenzhen Hospital of Southern Medical University, conducted under the Declaration of Helsinki, and funded by the Guangdong Basic and Applied Basic Research Foundation and several Shenzhen science programs, with the funders having no role in the design, analysis, or publication of the work.

Subject of Research: Prenatal and early postpartum predictors of multidimensional childbirth memory in Chinese women

Article Title: Prenatal and early postpartum factors associated with multidimensional childbirth memory among Chinese women: a prospective cohort study

Article References: Liu, L., Yang, Y., Huang, S., Qiu, Z., Li, K., Zhu, Y., Song, X., Deng, T., & Chen, L. (2026). Prenatal and early postpartum factors associated with multidimensional childbirth memory among Chinese women: a prospective cohort study. BMC Psychology. https://doi.org/10.1186/s40359-026-05743-4

Image Credits: AI Generated

DOI: 10.1186/s40359-026-05743-4

Keywords: childbirth memory, fear of childbirth, birth satisfaction, prenatal depression, infant complications, BirthMARQ, LASSO regression, prospective cohort study, postpartum, autobiographical memory, maternity care, Chinese women