The pulse oximeter perfusion index as a predictor for high illness severity in neonates.

Abstract

Following the first development in its modern form in 1972 by Takuo Aoyagi [1], pulse oximetry has become invaluable for monitoring oxygenation and pulse rate [4]. A constant amount of light (DC) from the pulseoximeter is absorbed by skin, other tissues, and nonpulsatile blood, while a variable amount of light (AC) is absorbed by pulsating arterial inflow. The pulsatile signal indexed against the non-pulsatile signal and expressed as a percentage (AC/DC · 100) is commonly referred to as the perfusion index (PI). Skin colourimetry is related to illness severity in newborns [2] but its utility as a monitoring tool is partially limited by requiring direct caregiver assessment. As local skin vasoconstriction can be associated with skin colour changes, the usefulness of PI for assessing neonatal illness severity was similarly tested. A prospective study was carried out on 101 inborn or outborn Caucasian neonates (52 males, 49 females; gestational age 34.7±4.0 weeks, range 24.7–41.1 weeks; birth weight 2310±950 g, range 410–4170 g) during the first 24 h after admission. Illness severity was determined using the Score for Neonatal Acute Physiology (SNAP) [5] and the infants were categorised as either high or low severity of illness, defined by the presence of severe neonatal morbidity and/or a 24 h SNAP score >10 [2]. PI values were assessed using a Masimo SET Radical pulse-oximeter (Masimo Corp., Irvine, Calif., USA) with the sensor placed randomly on either foot. After the pulse wave was verified to be artifact-free, the PI values were manually captured by an operator, who was unaware of the infant illness severity group, at least every 0.3 min for a duration of 10.6±2.3 min (95% CI 9.9–11.4 min, range: 10–15 min). SpO2, pulse rate, body (skin and core) temperature, and blood pressure were also measured. Out of a total of 2,571 measurements, 2,546 (99.02%) artifact-free readings were obtained. The higher severity group showed a significantly higher frequency of severe neonatal morbidity (P=0.025), as determined by the presence of at least one of the following: sepsis or pneumonia; bronchopulmonary dysplasia; intraventricular haemorrhage grade 3 or more; periventricular leukomalacia grade 3 or more; retinopathy of prematurity grade 3 or more; and necrotising enterocolitis [2]. Predictive accuracy for identifying newborns with higher severity for different cut-off values of PI, SpO2, and pulse rate was calculated using receiver operating characteristic (ROC) curve [3]. Model calibration was evaluated using v to compare the expected values (according to the classification of infants into two severity categories) with the expected values (according to the SpO2, pulse rate and PI values). According to the predefined criteria, 43 neonates (42.6%) were allocated to the high severity group and 58 to the low severity group. Male to female ratio, gestational age, birth weight, body temperature, mean blood pressure, and use of peripheral vasoconstrictors and vasodilators were not significantly different between the two severity groups (P‡0.50). Mean PI values were 1.54±0.80 (range: 0.22–5.22). Significantly lower PI values (0.86±0.26 versus 2.02±0.70; P<0.0001), lower SpO2 (93.3±5.4% versus 95.1±3.9%, P<0.0001) and higher pulse rate Eur J Pediatr (2002) 161: 561–562 DOI 10.1007/s00431-002-1042-5