Predicting healthcare outcomes in prematurely born infants using cluster analysis.

Victoria MacBean, Alan Lunt, Simon B. Drysdale, Muska N. Yarzi, Gerrard F. Rafferty, Anne Greenough

    Research output: Contribution to journalArticlepeer-review

    10 Citations (Scopus)
    174 Downloads (Pure)

    Abstract

    Aims Prematurely born infants are at high risk of respiratory morbidity following neonatal unit discharge, though prediction of outcomes is challenging. We have tested the hypothesis that cluster analysis would identify discrete groups of prematurely born infants with differing respiratory outcomes during infancy. Methods A total of 168 infants (median (IQR) gestational age 33 (31‐34) weeks) were recruited in the neonatal period from consecutive births in a tertiary neonatal unit. The baseline characteristics of the infants were used to classify them into hierarchical agglomerative clusters. Rates of viral lower respiratory tract infections (LRTIs) were recorded for 151 infants in the first year after birth. Results Infants could be classified according to birth weight and duration of neonatal invasive mechanical ventilation (MV) into three clusters. Cluster one (MV ≤5 days) had few LRTIs. Clusters two and three (both MV ≥6 days, but BW ≥or <882 g respectively), had significantly higher LRTI rates. Cluster two had a higher proportion of infants experiencing respiratory syncytial virus LRTIs (P = 0.01) and cluster three a higher proportion of rhinovirus LRTIs (P < 0.001) Conclusions Readily available clinical data allowed classification of prematurely born infants into one of three distinct groups with differing subsequent respiratory morbidity in infancy.
    Original languageEnglish
    Pages (from-to)1067-1072
    Number of pages6
    JournalPediatric Pulmonology
    Volume53
    Issue number8
    Early online date23 May 2018
    DOIs
    Publication statusE-pub ahead of print - 23 May 2018

    Fingerprint

    Dive into the research topics of 'Predicting healthcare outcomes in prematurely born infants using cluster analysis.'. Together they form a unique fingerprint.

    Cite this