key: cord-1037984-9nb8vja6
authors: Jones, J. G.
title: Oxygen saturation instability in suspected covid-19 patients; contrasting effects of reduced VA/Q and shunt.
date: 2020-12-19
journal: nan
DOI: 10.1101/2020.12.17.20248126
sha: db6a473bf6df09582a7e0195293928c7ad802557
doc_id: 1037984
cord_uid: 9nb8vja6

ABSTRACT. Patients in the UK at risk of Covid-19 pneumonia, but not needing immediate hospital attention, are to be given pulse oximeters to identify at home deterioration in oxygen saturation (SaO2 or SpO2). A recent finding in Covid-19 pneumonia is a dominant reduction in VA/Q. A mathematical model of gas exchange was used to examine the effect of reduction of VA/Q or increase in shunt on SaO2 stability inferred from the slope of the PIO2 vs SaO2 curve as it intersects the line representing ambient PIO2. Reduced VA/Q predicted SpO2 instability breathing air, e.g. a {+/-}1 kPa change in PIO2 gave an 8% change in SpO2 at a VA/Q of 0.4 but a 1.5% change in SpO2 with a 15% shunt. As a consistency check, two patients with pre-existing lung disease and 12 hour continuous SpO2 monitoring breathing air had gas exchange impairment analysed in terms of shunt and reduced VA/Q. The patient with 16% shunt and normal VA/Q had a stable but reduced SpO2 (circa 93%) throughout the 12 hr period. The patient with a VA/Q reduced to 0.48 had SpO2 ranging from 75-95% during the same period. SpO2 monitoring in suspected covid-19 patients should focus on SpO2 varying >5% in 30 minutes. Such instability in at risk patients is not diagnostic of Covid -19 pneumonia but this may be suspected from a dominant reduction in VA/Q if episodic hypoxaemia has progressed from a stable SpO2. Key words. Covid-19, Respiratory Measurement, Pneumonia, ARDS, VA/Q, Shunt, Oxygen Saturation.

Patients in the UK at risk of Covid-19 pneumonia, but not needing immediate hospital attention, are to be given pulse oximeters to identify at home deterioration in oxygen saturation (SaO2 or SpO2). A recent finding in Covid-19 pneumonia is a dominant reduction in VA/Q. A mathematical model of gas exchange was used to examine the effect of reduction of VA/Q or increase in shunt on SaO2 stability inferred from the slope of the PIO2 vs SaO2 curve as it intersects the line representing ambient PIO2. Reduced VA/Q predicted SpO2 instability breathing air, e.g. a ±1 kPa change in PIO2 gave an 8% change in SpO2 at a VA/Q of 0.4 but a 1.5% change in SpO2 with a 15% shunt. As a consistency check, two patients with pre-existing lung disease and 12 hour continuous SpO2 monitoring breathing air had gas exchange impairment analysed in terms of shunt and reduced VA/Q. The patient with 16% shunt and normal VA/Q had a stable but reduced SpO2 (circa 93%) throughout the 12 hr period. The patient with a VA/Q reduced to 0.48 had SpO2 ranging from 75-95% during the same period. SpO2 monitoring in suspected covid-19 patients should focus on SpO2 varying >5% in 30 minutes. Such instability in at risk patients is not diagnostic of Covid -19 pneumonia but this may be suspected from a dominant reduction in VA/Q if episodic hypoxaemia has progressed from a stable SpO2. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint NOTE: This preprint reports new research that has not been certified by peer review and should not be used to guide clinical practice.

200,000 patients in England at risk of complications with Covid-19, but not needing immediate hospitalisation, are to be given pulse oximeters to identify deterioration in oxygen saturation (SpO2) at home, [1] . This is to create "virtual Covid wards" of at risk patients who take SpO2 readings and relate these to their health teams. Mortality risk increases and admission to hospital is indicated if SpO2 falls to 94% or less, [1] .

A new finding may transform thinking about SpO2 monitoring in incipient Covid -19 Adult Respiratory Distress Syndrome (ARDS); a dissociation between severe hypoxemia and well-preserved alveolar gas volume in Covid-19 ARDS casing a reduced ventilation to perfused alveoli (VA/Q) with high compliance and radiological sparing. This is virtually never seen in most forms of ARDS where large shunt, extensive lung radio-opacities and low compliance dominate, [2] . A qualitative relationship has been reported between reduced VA/Q in patients with pre-existing lung disease and unstable SpO2 measured during 12 hour periods of continuous monitoring, [3, 4] .

A model of pulmonary gas exchange is used here to examine the effect of a reduction in VA/Q or increase in shunt on the position and slope of the PIO2 vs SpO2 curve as it intersects the line representing 21 kPa PIO2, i.e breathing air at sea level. From this will be inferred the stability of SaO2 with changes in distribution of ventilation or shunt. Newly available methods have enabled more precise discrimination of shunt and reduced VA/Q to examine the relationship between these entities and SpO2 stability from previous studies, [5, 6] . It is proposed that SpO2 instability is a sign of reduced VA/Q which may be a diagnostic feature of incipient Covid-19 ARDS.

VA/Q and shunt can be derived non-invasively using a computer algorithm to relate SpO2 to inspired oxygen pressure (PIO2) in a PIO2 vs SpO2 diagram, [4] [5] [6] [7] . An unstable SpO2 is is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint curve further to the right (green arrow and line) and as it crosses the 21kPa PIO2 line, representing air breathing, its gradient becomes increasingly steep. Small changes in VA/Q (or PIO2) cause large changes in SpO2. Increasing shunt bends the curve downwards, [4] [5] [6] so that large changes in PIO2 cause small changes in SpO2 emphasising the unreliability of the PaO2/FIO2 as an index of oxygen exchange, [8] .

An algorithm based on a three compartment lung model was recently developed to analyse the slope of the PIO2 vs SpO2 curve in pre term infants with broncho pulmonary dysplasia, [5] [6] [7] . This showed that slope was greatest in those with a homogeneous reduction of VA/Q to circa 0.4. Slope was considerably less with inhomogeneous pulmonary disease with or without increased shunt. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint

The effects of reducing VA/Q or increasing shunt on PIO2 vs SpO2 curves were derived using a mathematical model of pulmonary gas exchange described by Olszowska and Wagner, [9] . Datasets were generated using the equations implemented on a spreadsheet To provide a consistency check of this analysis two patients with pre-existing lung disease and 12 hour continuous SpO2 monitoring from a previous study [3] had their PIO2 vs SpO2 plots reanalysed with a newly developed algorithm based on a three compartment lung, [5, 6] .

The left panels in Fig is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint

and on the gradient of these curves as they intersect the 21kPa PIO2 line (right panels).

To illustrate the clinical effect of VA/Q vs Shunt on SpO2 stability are two patients with a modest impairment of gas exchange, one with increased shunt the other with decreased VA/Q re-analysed from Fig11 Ref 4. (Fig 3) . Left panels show oxygen dissociation (red) and normal lung PIO2 vs SpO2 curves (blue). Patients A and B had dark green PIO2 vs SpO2 curves derived by changing PIO2 stepwise and analysing the resulting PIO2 vs SpO2 data pairs with a new computer algorithm, [5, 6] . The curve plateau was displaced downwards in A by a moderate shunt (14%) but not shifted to the right. When breathing air the gradient of curve A as it intercepted the 21kPa PIO2 line was 0.5 and similar to the is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint normal lung. In B the whole curve was shifted to the right by the VA/Q reduced to 0.48. This patient had a shunt of 12% but the effect on SpO2 was outweighed by reduced VA/Q increasing curve gradient at the intercept to 2% SpO2 per kPa PIO2 which is similar to the result of the mathematical model in the right upper panel of is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint with an increased shunt showed a stable pattern with superimposed peaks varying 1-2% around the median, closely resembling the normal lung pattern [4] , but with a lower than normal median SpO2 circa 93-94%. In contrast Patient B with reduced VA/Q had broad peaks with SpO2 drifting downwards from >95 to 75% during the study. Large falls in SpO2 are consistent with small reductions in VA/Q below 0.5. Compared to the highly unstable SpO2 with a modest reductions of VA/Q to 0.5, it would be expected from Fig 2 that a patient with a 25% shunt breathing air would have a stable saturation down to 88%SpO2.

Instability of SpO2 breathing air with SpO2 varying >5% in 30 minutes is a sensitive marker of a reduction in VA/Q, e.g. to <0.5. The shape of the PIO2 vs SpO2 curve indicates more specifically the nature and magnitude of the underlying gas exchange abnormality; only a pencil and paper are required with PIO2 varied in three or more steps using a Ventimask, [10] In practice reduced VA/Q and and increased shunt often co-exist in different proportions giving complex effects on curve shape. A computer program plots the curve and generates a numerical estimate of VA/Q and Shunt from SpO2 vs PIO2 data pairs when PIO2 is changed stepwise, [5, 6] .

Other groups have reported on SpO2 monitoring of at risk Covid -19 patients in a domestic "virtual ward" setting. One chose a threshold for admission of 2% less than a 96% target, [11] . Another did not find such monitoring useful because of variability of implementation, [12] whereas a third reported on remote monitoring of 26 patients after discharge from hospital with Covid 19. In the latter group there were 5 "alerts" and 4 re-admissions with a median of 91% SpO2 and a lowest reading of 82%. None commented on unstable SpO2. is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprint

The copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint their health teams. An at risk Covid-19 patient with previously normal SpO2 but develop ing unstable SpO2 with episodes of profound hypoxaemia out of proportion to radiological change may well be compatible with the reduced VA/Q of Covid-19 pneumonia.

The author thanks Dr A Olszowka, Department of Physiology, University of Buffalo, New York, USA for providing his pulmonary gas exchange program and Mr Jon Brassey, Trip Database, for help in the preparation of this paper.

Covid-19: Patients to use pulse oximetry at home to spot deterioration

COVID-19 pneumonia: ARDS or not?

Causes of oxyhaemoglobin saturation instability in the post operative period

Discriminating between the effect of shunt and reduced VA/Q on arterial oxygen saturation is particularly useful in clinical practice

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

Influence of pulmonary factors on pulse oximeter saturation in preterm infants

Physiology and Predictors of Impaired Gas Exchange in Infants with Bronchopulmonary Dysplasia

Variation in the PaO 2 /F I O 2 ratio with F I O 2 : mathematical and experimental description, and clinical relevance

Numerical analysis of gas exchange

Profound hypoxemia in pulmonary patients in airline-equivalent hypoxia: roles of VA/Q and shunt

Validation of home oxygen saturations as a marker of clinical deterioration in patients with suspected COVID-19 doi

Remote home monitoring (virtual wards) during the COVID-19 pandemic: a living systematic review doi

Remote monitoring of oxygen saturation in individuals with COVID-19 pneumonia

It is made available under a perpetuity.is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprintThe copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint It is made available under a perpetuity.is the author/funder, who has granted medRxiv a license to display the preprint in (which was not certified by peer review) preprintThe copyright holder for this this version posted December 19, 2020. ; https://doi.org/10.1101/2020.12.17.20248126 doi: medRxiv preprint