Objective To investigate if a statistical model of heart rate changes in preterm infants below 32 weeks gestation correlates with illness severity score and NICU LOS.
Design/Methods Infants <32 weeks gestation admitted to McMaster Children s Hospital were recruited. Routinely measured heart rate over the first 24 hours of stay was continuously recorded from the NICU s networked cotside monitors via the Infinity Gateway (Draeger Medical Canada Inc) software. The sampling interval was one minute. The heart rate was classified into four states: S1 80–100 bpm, S2 100–120 bpm, S3 120–150 bpm and S4>150 bpm. The data was processed using MATLAB 7.5 (Mathworks, Natick MA) software. The probability of these infants remaining in each state and changing from one state to another was calculated as a Markov chains probability model. SNAP-II, Perinatal Extension, Version II (SNAPPE-II) and Transport Risk Index of Physiologic Stability (TRIPS) scores were calculated.
Results 45 infants (21 females) with mean (SD) gestational age of 28.7 (2) weeks, mean (SD) birthweight of 1250 (346) g had a mean (SD) length of stay of 43 (45) days. None of the infants died. The strongest correlations were between length of stay and the change in states from S3–S4 (coefficient 0.52, p = 0.0003), and SNAPPE-II (coefficient 0.54, p = 0.0001). SNAP-II and TRIPS were not significantly correlated to LOS.
Conclusions Although the sample size is small, probability modeling using Markov chains could prove promising to correlate LOS with HR variations of preterm infants in the first day of life.