Unstable SpO2 in preterm infants: The key role of reduced ventilation to perfusion ratio.

Introduction: Instability of peripheral oxyhemoglobin saturation (SpO2) in preterm infants is correlated with late disability and is poorly understood. We hypothesised that a reduced ventilation to perfusion ratio (VA/Q) is the key predisposing factor for SpO2 instability. Methods: We first used a mathematical model to compare the effects of reduced VA/Q or shunt on SaO2 stability (SaO2 and SpO2 are used for model and clinical studies respectively). Stability was inferred from the slope of the SaO2 vs. inspired oxygen pressure (P IO2) curve as it intersects the 21 kPa P IO2 line (breathing air). Then, in a tertiary neonatal intensive care unit, paired hourly readings of SpO2 and P IO2 were recorded over a 24 h period in week old extremely preterm infants. We noted SpO2 variability and used an algorithm to derive VA/Q and shunt from the paired SpO2 and P IO2 measurements. Results: Our model predicted that when VA/Q < 0.4, a 1% change in P IO2 results in >8% fluctuation in SaO2 at 21 kPa P IO2. In contrast, when a 20% intrapulmonary shunt was included in the model, a 1% change in P IO2 results in <1% fluctuation in the SaO2. Moreover, further reducing the VA/Q from 0.4 to 0.3 at 21 kPa P IO2 resulted in a 24% fall in SaO2. All 31 preterm infants [mean gestation (±standard deviation) 26.2 (±1) week] had VA/Q < 0.74 (normal >0.85) but only two infants had increased shunt at 1.1 (±0.5) weeks' postnatal age. Median (IQR) SpO2 fluctuation was 8 (7)%. The greatest SpO2 fluctuations were seen in infants with VA/Q < 0.52 (n = 10): SpO2 fluctuations ranged from 11%-17% at a constant P IO2 when VA/Q < 0.52. Two infants had reduced VA/Q and increased shunt (21% and 27%) which resolved into low VA/Q after 3-6 h. Discussion: Routine monitoring of P IO2 and SpO2 can be used to derive a hitherto elusive measure of VA/Q. Predisposition to SpO2 instability results from reduced VA/Q rather than increased intrapulmonary shunt in preterm infants with cardiorespiratory disease. SpO2 instability can be prevented by a small increase in P IO2.

Evaluation of a computer program for non-invasive determination of pulmonary shunt and ventilation-perfusion mismatch.

We describe a three-compartment model (shunt and two perfused compartments) to analyse the relationship between inspired oxygen (FIO2) and arterial oxygen saturation (SaO2) in terms of pulmonary shunt and ventilation-perfusion ratio (VA/Q). The program was tested using 24 exact datasets, each with six pairs of FIO2 and SaO2 data points with known VA/Q and shunt, generated by a complex calculator of gas exchange. Additional datasets were created by adding noise and rounding the exact sets, and by reducing the number of data points per dataset. The importance of the oxyhaemoglobin dissociation curve and the arterio-venous difference in oxygen content (avDO2) were also tested. Analysis using the three compartment model was more accurate than the two compartment model and less affected by data degradation. The absolute error in shunt estimation was never more than 2.2 % for the exact and rounded datasets, but the error in VA/Q estimation was -29 to 19 % of the true value (10th-90th centiles). The characteristics of the well-ventilated compartment were not determined accurately. At extremes of cardiac output, an assumed value of avDO2 resulted in significant errors. It is probably advantageous to correct for foetal haemoglobin in neonatal datasets. Analysis of FIO2 versus SaO2 datasets using a three compartment model provides accurate estimates of shunt and VA/Q when arterio-venous difference in oxygen content is known. The estimates may have value as objective measures of gas exchange, and as a visual guide for oxygen therapy.

Feasibility and safety of delivering xenon to patients undergoing coronary artery bypass graft surgery while on cardiopulmonary bypass: phase I study.

BACKGROUND: Postoperative neurocognitive deficit is prevalent after cardiac surgery. Xenon may prevent or ameliorate acute neuronal injury, but it also may aggravate injury during cardiac surgery by increasing bubble embolism. Before embarking on a randomized clinical trial to test the safety and efficacy of xenon for postoperative neurocognitive deficit, we undertook a phase I study to investigate the safety of administering xenon to patients undergoing coronary artery bypass grafting while on cardiopulmonary bypass and to assess the practicability of our xenon delivery system. METHODS: Sixteen patients scheduled for coronary artery bypass grafting surgery with hypothermic cardiopulmonary bypass gave their informed consent to participate in an open-label dose-escalation study (0, 20, 35, 50% xenon in oxygen and air). Xenon was delivered throughout surgery using both a standard anesthetic breathing circuit and the oxygenator. Gaseous and blood xenon partial pressures were measured five times before, during, and after cardiopulmonary bypass. Middle cerebral artery Doppler was used to assess embolic load, and major organ system function was assessed before and after surgery. RESULTS: Middle cerebral artery Doppler showed no evidence of increased emboli with xenon. Patients receiving xenon had no major organ dysfunction: Troponin I and S100beta levels tended to be lower in patients receiving xenon. Up to 25 l xenon was used per patient. Xenon partial pressure in the blood tracked the delivered concentration throughout. CONCLUSIONS: Xenon was safely and efficiently delivered to coronary artery bypass grafting patients while on cardiopulmonary bypass. Prevention of nervous system injury by xenon should be tested in a large placebo-controlled, randomized clinical trial.

Predicting the behaviour of G-RAM networks.

A Generalising Random Access Memory (G-RAM) neuron is distinguished from conventional neuron models by the fact that its tolerance to departures in similarity from its training pattern is variable. Details of this are given in this paper as it affects the behaviour a class of digital probabilistic neural networks which have been achieving attention in the neural networks literature for some years now. Such systems are also called n-tuple systems, weightless systems or p-RAM systems. After reviewing the literature on such networks, a novel simple combinatoric analysis of the most likely behaviour of recursive GRAM networks is described. The best network performance, measured by a key parameter called 'radius of retrievability' (first defined by Wong and Sherrington [J. Phys. A 22 (1989) 2233] as the error in the input that still allows evolution of the dynamic network to the correct attractor state), is obtained with a training set composed of random data patterns. Increasing the size of the training set reduces this radius of retrievability in a predictable manner. Changing the nature of the training set to non-random patterns also reduces the radius of retrievability to an extent that we show can be estimated from a measure of the diversity of the elements of the training set (we refer to this as the 'mean intra-set Hamming distance of the training set'). As mentioned earlier the feature of G-RAMs (indicated by the G) is that there exists a generalization parameter which determines how far a neuron input vector can stray from a training input for the neuron to respond in the trained way. It is shown that when this generalization parameter is reduced, then the radius of retrievability is also reduced but it is then stable in the face of an increase in size, or change in nature, of the training set. This is a novel prediction of the behaviour of systems and of the robustness of such behaviour in the face of varying the size and correlation properties of the training set.