Birth timing and delivery-mode differences in next-day neonatal weight loss: a retrospective nursing-record study

Background Routine next-calendar-day weighing does not ensure equal postnatal observation intervals. We examined whether delivery-mode differences in recorded neonatal weight loss changed after accounting for birth-clock time and record source. Methods This retrospective study used 4801 nursing-record rows from a Beijing maternity hospital. Investigators confirmed routine next-calendar-day morning weighing, rather than individual 24-hour measurements; individual weighing timestamps were unavailable. Percentage weight loss was calculated from birthweight and next-day weight after investigator-confirmed corrections. Median regressions used the same complete-record sample. A non-cyclic birth-clock spline with source adjustment was the workflow-aligned reference; cyclic and linear alternatives, missingness bounds and paired birth-date cluster resampling assessed robustness. Results Numeric weight pairs were available for 4382 records (91.27%); median loss was 2.71% (interquartile range 1.47–4.14). Morning births were more frequent after cesarean than vaginal delivery. In 4376 complete records, the cesarean–vaginal conditional median difference was 0.70 percentage points (95% confidence interval 0.55–0.84) after sex and birthweight adjustment, 0.67 (0.53–0.81) with source adjustment alone, and 0.20 (0.06–0.33) with spline and source adjustment. Cyclic and linear alternatives yielded differences of 0.19 and 0.38 points. In 3608 known-date records, the base-minus-spline/source coefficient difference was 0.510 points (95% paired date-cluster bootstrap interval 0.399–0.641; 1000 replications). Results were similar after reducing duplicate candidates, while weight-pair availability differed by delivery mode and transfer/discharge notation. Conclusions Accounting for birth-clock time reduced delivery-mode differences in next-day recorded weight loss, although residual estimates depended on model form. Routine calendar-day observations should not be treated as measurements at a uniform postnatal age. These findings support timing-aware interpretation of nursing records, not causal attribution or a 24-hour clinical reference standard.