New Prediction Tool Flags Postpartum Fatigue With Surprising Accuracy

For millions of new mothers, the exhaustion that follows childbirth is dismissed as an unavoidable rite of passage, a tiredness that will simply fade with time. But a new study from China suggests that postpartum fatigue is far more than ordinary sleep deprivation, and that clinicians may soon be able to predict which women are most at risk using just four simple variables. The research, published in the journal Nursing Open, presents a statistically validated nomogram, a graphical calculation tool that converts a patient’s individual characteristics into an estimated probability of fatigue, designed to be used during the routine six-week postpartum check-up.

The scale of the problem is striking. In the study, nearly 69 percent of the 744 women assessed met the criteria for postpartum fatigue, a condition defined as overwhelming exhaustion that impairs physical and mental functioning and is not relieved by rest. That figure aligns with earlier estimates suggesting that more than half of new mothers experience significant fatigue after delivery. Unlike typical tiredness, postpartum fatigue has been linked to poorer maternal recovery, reduced breastfeeding success, diminished competence in infant care, weakened mother-baby bonding and disrupted family adjustment. Despite this burden, the condition remains chronically under-recognized, because both clinicians and mothers often mistake it for a normal feature of the postpartum period.

The research team, led by investigators affiliated with a tertiary hospital in Guangdong, recruited 780 women attending their routine six-week postpartum visit between January and May 2025. Participants were aged 20 or older, recovering from a singleton live birth, and assessed between 42 and 7 days either side of the six-week mark. Women with major postpartum complications such as eclampsia or severe haemorrhage requiring intensive care, significant mother-infant separation, or pre-existing severe medical or psychiatric illness were excluded. After removing 36 questionnaires with implausibly short completion times or patterned responses, 744 valid responses remained, a response rate of 95.4 percent. The sample was then randomly split, with 522 women assigned to a derivation set for building the model and 222 to an internal validation set for testing it.

Fatigue was measured using the 14-item Postpartum Fatigue Scale, which covers physical fatigue, mental fatigue and the consequences of fatigue, each rated on a five-point scale. A standardized score above 40 defined fatigue. The scale showed excellent internal consistency in this sample, with a Cronbach’s alpha of 0.95. Depressive symptoms were assessed with the 10-item Edinburgh Postnatal Depression Scale, and perceived social support with the 12-item Perceived Social Support Scale, both validated in Chinese populations. Data on delivery mode, labour duration and intrapartum blood loss came from medical records, while sociodemographic and infant-care information came from a structured questionnaire administered electronically, with mandatory fields and duplicate-submission blocking to protect data integrity.

The modelling strategy reflected current best practice in clinical prediction research. Rather than relying on stepwise selection, which risks overfitting, the team applied least absolute shrinkage and selection operator regression, known as LASSO, with ten-fold cross-validation to screen 25 candidate factors down to the most informative subset. Multicollinearity was checked using variance inflation factors, and continuous variables were entered in their original form rather than arbitrarily categorized. The final multivariable logistic regression model retained four independent predictors: postpartum depression score, maternal sleep quality, breastfeeding ability and intrapartum blood loss. Perceived social support, although significantly associated with fatigue in univariate analysis, dropped out of the final model, suggesting its apparent effect is largely explained by its overlap with mood and sleep variables.

The statistical weights behind the nomogram tell a coherent biological story. Each one-point increase on the depression scale raised the odds of fatigue by roughly 30 percent, with an odds ratio of 1.297. Each step down in sleep quality carried an odds ratio of 1.534, while each decline in breastfeeding proficiency carried an odds ratio of 1.899, the strongest single predictor in the model. Intrapartum blood loss contributed an odds ratio of 1.419 per additional 100 millilitres. The full model takes the form of a logistic equation with a constant of minus 3.757, from which an individual probability of fatigue can be computed and read off the nomogram.

Performance metrics were encouraging. In the derivation set, the model achieved an area under the receiver operating characteristic curve of 0.821, and in the internal validation set it held at 0.801, values generally considered indicative of good discrimination. Using an optimal probability cut-off of 0.682, derived from the maximum Youden index, sensitivity was 75.2 percent and specificity 76.8 percent in the derivation set, declining only modestly to 74.4 percent and 73.5 percent in validation. Positive predictive values exceeded 86 percent in both sets, although negative predictive values hovered near 57 percent, meaning the tool is better at confirming risk than at ruling it out. Calibration was strong, with Hosmer-Lemeshow tests showing no evidence of poor fit in either set, and Brier scores of 0.163 and 0.155. Decision curve analysis demonstrated favourable net benefit across a wide range of threshold probabilities, indicating the tool could be clinically useful without generating excessive false alarms.

Each retained predictor points toward a concrete intervention pathway. The link between depression and fatigue is likely bidirectional: abrupt postpartum hormonal shifts can destabilize neuroendocrine regulation, while depressive symptoms erode the psychological reward of infant care and disrupt circadian rhythms through neurotransmitter disturbances involving serotonin and norepinephrine, producing early-morning awakening that no amount of nighttime help can fix. Sleep quality, meanwhile, is battered by the fragmented nights of newborn care, which reduce slow-wave and REM sleep and impair physical recovery, cognition and hormonal regulation. Breastfeeding difficulties such as poor latch, nipple pain and perceived insufficient milk supply add physical strain and emotional burdens of anxiety and self-doubt. Blood loss acts through a purely physiological channel, since postpartum anaemia reduces oxygen delivery to tissues and produces lethargy, dizziness and weakness, making accurate blood-loss assessment and timely iron supplementation a plausible preventive strategy.

The authors are careful to spell out the limitations. The cross-sectional design means the associations cannot be interpreted causally, and because predictors and outcome were measured simultaneously through self-report questionnaires, common-method bias may have inflated some associations. Convenience sampling at a single hospital, and the absence of external validation, mean the nomogram’s performance in other populations and health systems remains unproven. The team calls for longitudinal, multicentre studies with larger and more diverse samples before the tool can be considered generalizable.

Even with those caveats, the study represents a meaningful step toward making postpartum fatigue visible in routine care. The four inputs it requires, a depression screen, a sleep-quality rating, a breastfeeding assessment and a figure for intrapartum blood loss, are all readily available at a standard postpartum visit, which means the nomogram could be deployed without new tests or expensive equipment. If external validation confirms its accuracy, the tool could help obstetric clinicians move from reactive reassurance to targeted prevention, identifying high-risk women weeks before exhaustion hardens into a chronic problem, and directing them toward lactation support, sleep-sharing arrangements, anaemia treatment or mental health care. For a condition that affects roughly two in three new mothers and has long been written off as normal, that shift could reshape the standard of postpartum care.

Subject of Research: Development of a clinical nomogram for identifying postpartum fatigue in postpartum women

Article Title: Development and Internal Validation of a Nomogram for Identifying Postpartum Fatigue in Postpartum Women: A Cross‐Sectional Study

Article References: Guo, Z., Xu, M., & Cheng, S. (2026). Development and Internal Validation of a Nomogram for Identifying Postpartum Fatigue in Postpartum Women: A Cross‐Sectional Study. Nursing Open, 13(10), Article e70907. https://doi.org/10.1002/nop2.70907

Image Credits: AI Generated

DOI: 10.1002/nop2.70907

Keywords: postpartum fatigue, nomogram, postpartum depression, sleep quality, breastfeeding, intrapartum blood loss, logistic regression, LASSO, prediction model, maternal health, cross-sectional study, nursing