Characterizing and Forecasting Individual Weight Changes in Term Neonates
Mélanie Wilbaux (1), Severin Kasser (2), Sven Wellmann (2), Olav Lapaire (3), Johannes N. Van Den Anker (1, 4), Marc Pfister (1, 5)
(1) Division of Pediatric Pharmacology and Pharmacometrics, University Children’s Hospital Basel (UKBB), Basel, Switzerland, (2) Division of Neonatology, University of Basel Children’s Hospital (UKBB), Basel, Switzerland, (3) Division of Obstetrics and Gynecology, University Hospital Basel, Basel, Switzerland, (4) Division of Pediatric Clinical Pharmacology, Children’s National Health System, Washington, DC, USA, (5) Quantitative Solutions LP, Menlo Park, CA, USA.
Objectives: As part of normal physiology newborns lose body fluid during the first days of life. The magnitude of initial fluid loss and consecutive weight gain vary strongly among neonates, and excessive weight loss in some newborns can result in long-term complications. Objectives of this work were to (i) develop a semi-mechanistic model that characterizes physiological weight changes during the first week of life; (ii) identify effects of maternal and neonatal factors, and (iii) provide an online tool to forecast individual weight changes.
Methods: Longitudinal weight data and individual characteristics from 1335 healthy term neonates exclusively breastfed were available up to 1 week of life. A semi-mechanistic turnover model was developed characterizing the weight change as a function of a changing net balance between time-dependent rates of weight gain (Kin) and first-order weight loss (Kout). Different time-dependent functions were tested such as linear, exponential or saturable functions. Population analysis was implemented using NONMEM 7.3. Model selection was based on statistical criteria, goodness-of-fit plots and simulation-based diagnostics. Clinically relevant covariates testing was performed utilizing a standard stepwise forward-backward covariate model building approach. Data from 300 additional term neonates were used for advanced evaluation of developed model.
Results: Kin was modeled as an exponential function of time. Kout was modeled with a saturable function to describe initial decrease due to fluid loss followed by an exponential time-dependent increase. Males had higher birth weights (WT0) than females. Gestational age had a positive effect on WT0 and Kin, whereas mother’s age had a positive effect on WT0 and a negative effect on Kin. Advanced evaluation demonstrated good predictive performance of the model (bias=0.01%, precision=0.52%). Furthermore, the model was able to accurately forecast individual weight changes up to 1 week with only 3 initial weight observations during the first 2 days (bias=-0.74%, precision=1.54%).
Conclusions: We developed the first model describing physiological weight changes in healthy term exclusively breastfed neonates during the first days of life. We provide a user-friendly online NeoWeight Prediction tool allowing caregivers to monitor individual weight changes, and further personalize and optimize care for neonates. This model will be expanded with data from preterm, formula-based fed and sick neonates.