Estimating lung volume during high frequency oscillatory ventilation of neonates

A problem presented at the UK MMSG Nottingham 2000.

Presented by:
Mr Jag Ahluwalia (Rosie Maternity Hospital, Addenbrooke's NHS Trust, Cambridge)
Prof Oliver Jensen (School of Mathematical Sciences, University of Nottingham)
J Ahluwalia, OE Jensen, SL Waters, JAD Wattis

Problem Description

The primary goals of lung ventilation are to bring oxygen into the alveoli from where it can enter the blood and to eliminate carbon dioxide dissolved in blood. Since CO2 is much more soluble in blood than oxygen, O2 transport is diffusion-limited and a large alveolar surface area is the primary factor controlling its delivery. CO2 transport is flow limited, so that efficient gas transport and mixing in the lung primarily controls its elimination from the blood.

Artificial ventilators are often used to support premature infants suffering respiratory distress. With Conventional Mechanical Ventilation (CMV), which operates at normal tidal frequencies, a doctor can control six variables on a ventilator:

Increasing the difference PIPPEEP increases the tidal volume VT (the volume of air entering the lung during each breath). The efficiency of CO2 elimination is approximately proportional to the product VT f, although excessively rapid, large-amplitude breaths can be traumatic to airways. Raising PEEP keeps airways inflated at the end of each breath, preventing small airways and alveoli from collapsing, and ensuring good O2 exchange in alveoli. If PIP is too high there is a danger the lungs will over-distend, adversely affecting VT and CO2 clearance. The effectiveness of the ventilation can be measured by continuous monitoring of CO2 levels in the blood, hot-wire measurements of VT in the endotracheal tube and visual assessment of chest wall movements. These observations guide the doctor in setting appropriate values of PIP, PEEP, Ti, Te, Q and FiO2.

When CMV fails, High Frequency Oscillatory Ventilation (HFOV) may be used. Under HFOV, the ventilator variables that can be controlled are the mean airway pressure (MAP), the frequency of the oscillations f (typically around 10 Hz), the pressure amplitude of the pump oscillator (Pamp), and FiO2. Although Pamp is large, the resulting pressure fluctuations in the alveoli are small, and they fall in magnitude as f is increased in a manner controlled by the resistance and compliance of the lungs and of the ventilator itself. The tidal volume VT is also small, and it too falls as f is increased. MAP controls the mean lung distension: this should be large enough for small airways to be recruited, but not so large that lung compliance falls because airways become over-distended. O2 delivery is therefore controlled primarily by MAP and FiO2. CO2 elimination is controlled by the degree of mixing in the airways, and it correlates approximately with VT2 f. Continuous measurements of CO2 levels in the blood and VT are available to the doctor, as are occasional measurements of lung volume V by X-ray, but X-rays are inaccurate and potentially harmful. An alternative technique for estimating V is sought.

The study group is asked to consider whether V could be assessed directly from VT measurements, perhaps in combination with other techniques (such as measuring chest wall impedance from ECG, acoustic signals, or strain gauges placed on the chest wall).

Study Group Report

High frequency ventilation (HFOV) is often used to support premature neonates. During HFOV, occasional measurements of the lung volume V are available by X-ray, but X-rays are inaccurate and potentially harmful. This project aimed to determine whether V can instead be assessed directly from the data available from the ventilator. We also determined how the pressure fluctuations generated by the pump oscillator are attenuated through the system resulting in smaller pressure fluctuations at the alveoli.

In order to do this, we developed a simple lumped-parameter model of the system which captured the dominant characteristics of lung behaviour, in particular, resonance at certain frequencies. Moreover, while direct estimation of V proved difficult, we were able to determine lung compliance from pressure-flow curves. Ideally ventilation should occur in a regime where the compliance is greatest so that the least possible damage is done to the airways. The lung volume may then be inferred indirectly from the compliance.

The large number of parameters in the problem poses potential difficulties. The characteristics of the ventilator circuit should be quantifiable, and future work will identify the appropriate method for determining and describing the ventilator compliance: should this be in terms of either tubing compliance or air compressibility? Furthermore, the air leak from the end of the endotracheal tube, which is hard to quantify, may also be significant.

Download the full report