Pregnancy‐specific continuous reference intervals for haematology parameters from an Australian dataset: A step toward dynamic continuous reference intervals

Authors:
Publisher website
Google Scholar
Abstract
Background Pregnancy is a time of dynamic physiological changes occurring as a continuous spectrum. Smoothed centile curves describe the distribution of measurements as a function of age. There has been no application of centile charts in pregnancy for haematological parameters. Aims To derive gestational age‐specific centile curves for six haematological parameters and compare these with published reference intervals. Materials and Methods An LMS approach was used with haematology results from an obstetric hospital laboratory database. After application of exclusion criteria, smoothed centiles conditional on gestational age were obtained by a two‐step process: (i) finding the best model within four response distributions using Bayesian information criteria, and (ii) selecting the best model among the response distributions based on test‐dataset global deviance. Results In total, 11 255 deliveries were extracted from 10 813 patients. There was little difference between distributions, and Box‐Cox power exponential was selected overall. Red cell parameters showed similar trends: values fell until the second trimester and increased thereafter. Leukocyte and neutrophil counts rapidly increased and plateaued around 15 weeks. Platelets exhibited a gradual fall with advancing gestation. Conclusions This is the first study to use an LMS approach to model gestational age‐dependent variations in haematological parameters. Proposed haemoglobin reference intervals were lower than those published but reflect our patient population. Serial monitoring of antenatal patients, as is the standard of care, in conjunction with these centile charts, may highlight trends in red cell changes with advancing gestation, allowing early identification of adverse pregnancy outcomes and evolving anaemia.

This publication has 17 references indexed in Scilit: