Introduction
Acute respiratory failure (ARF) may be divided into two broad types: hypoxemic and ventilatory. Reference Koutsoukou and Roussos1 Hypoxemic ARF entails impaired diffusion of alveolar oxygen into pulmonary capillary blood and/or mismatch between alveolar ventilation and pulmonary capillary blood flow (V/Q mismatch). Ventilatory failure entails alveolar ventilation, which is insufficient to excrete metabolic carbon dioxide (CO2) and results in its accumulation. Often, the syndrome is a combination of both mechanisms.
Patients with ARF may require mechanical ventilatory support. The mainstays of such support are enhanced alveolar ventilation and oxygen-enriched positive pressure air. These are coupled with various mechanical adjustments, such as Positive End-Expiratory Pressure (PEEP), Reference Briel, Meade and Mercat2 variations of Inspiratory Time (Ti), and Inspiratory to Expiratory ratios (I/E Ratio). Reference Müller-Redetzky, Felten and Hellwig3
The basic assumption in nations with well-developed health care systems is that patients with ARF are treated by experts in an appropriate environment with state-of-the-art mechanical ventilators, and that enough medical oxygen is always available. However, in medically underserved nations and in very extensive disasters (eg, epidemics, earthquakes, wars, or other humanitarian catastrophes), the number of potentially salvageable patients requiring mechanical ventilatory support may outstrip the number of available ventilators Reference Dickson, Hotchkin and Lamm4 and the quantity of available medical oxygen.
The causes of respiratory failure in disasters are multiple, and they too may be divided into hypoxemic, hypercapnic, and mixed. Examples include viral pneumonitis in influenza, Reference Daugherty, Branson and Rubinson5 chemical pneumonitis with chemical agent exposure, Reference Abou-Donia, Siracuse, Gupta and Sobel Sokol6 near drowning with tsunami wave, and post-operative ventilation.
The issue of the supply of medical oxygen has not been discussed much in the literature. Oxygen is usually generated in commercial plants, though it may be generated locally. In a major disaster, this supply chain cannot be guaranteed. Oxygen is thus a critical resource in disasters, and its unlimited availability cannot be taken for granted. Reference Blakeman and Branson7 Blakeman notes that there are essentially no national disaster reserves of oxygen in the US, and the system relies on private vendors. These may be unable to supply all the demand, or may have transportation issues, resulting in a total or partial disruption in oxygen supply. It will be the incident manager’s task to ration this critical, yet suddenly scarce, commodity. Surprisingly, a paper describing a European consensus process for pandemic preparedness does not even mention oxygen supply issues. Reference Belfroid, Timen, van Steenbergen, Huis and Hulscher8
Given an oxygen shortage, the instruction may go out to minimize use of oxygen. Methods to decrease oxygen utilization may include: decreasing flow to oxygen masks; decreasing fraction of inspired oxygen (FiO2) to patients receiving continuous positive airway pressure (CPAP) and non-invasive ventilation; avoiding oxygen-hungry continuous-flow CPAP systems; discontinuing oxygen as soon as clinically acceptable; and decreasing FiO2 to mechanically ventilated patients to the minimum compatible with acceptable oxygen saturation, or even to 0.21.
The concept of using the minimum amount of oxygen in ventilated patients in order to avoid oxygen toxicity is an old one. Reference Hafner, Beloncle, Koch, Radermacher and Asfar9 Nevertheless, there is no consensus requiring that FiO2 not be decreased to less than 0.4. In recent reviews, this is not even mentioned among evidence-based recommendations. Reference Blakeman and Branson7,Reference Fan, Brodie and Slutsky10 Even in a study looking specifically at a disaster situation and testing ventilators aimed at such situations, oxygen was used liberally. Reference Dickson, Hotchkin and Lamm4 Ventilating with FiO2 = 0.21 is not an option even discussed in textbooks or most studies. Reference Fan, Brodie and Slutsky10
In ventilated patients, FiO2 is usually increased if partial pressure of oxygen (PaO2) drops and does not respond to alveolar recruitment maneuvers such as increased PEEP, increasing inspiratory time up to and including inverting I/E ratio, muscle paralysis, prone position, switching to Pressure Controlled Ventilation (PCV), and Airway Pressure Release Ventilation (APRV). Reference Siegel and Hyzy11
However, significant variability exists in practice, and often mechanically ventilated patients are ventilated at FiO2 of at least 0.4. A study of the effect of a restrictive FiO2 regimen (median FiO2 = 0.36) Reference Girardis, Busani and Damiani12 showed lower mortality and complications in an intensive care unit (ICU). A clinical study of very low FiO2, and even room air, in ICU patients showed no negative (and no positive) effects of a restrictive FiO2 regimen that also included FiO2 = 0.21 compared to usual care. Reference Mackle, Bellomo and Bailey13 Finally, a meta-analysis concluded that high-quality evidence shows that liberal oxygen therapy increases mortality without improving other patient-important outcomes. Reference Chu, Kim and Young14
The objective of the present study was to show in an Acute Respiratory Distress Syndrome (ARDS) model in pigs, that mechanical ventilation with FiO2 = 0.21 is feasible. Pigs are similar to humans in terms of anatomy, genetics, and physiology. As in humans, the porcine lung has extensive inter- and intra-lobular connective tissue, which connects the major vessels and the bronchi to the pleural surface. Hence, the swine lung is considered an excellent model and has served to study lung development, reperfusion injury, and hyperoxia-induced acute lung injury, as well as other diseases. Reference Katalan, Falach and Rosner15
Methods
Experimental Definition of ARDS
Acute Respiratory Distress Syndrome was defined when two conditions were met: peripheral capillary oxygen saturation (SpO2) <90% at FiO2 = 0.4, PEEP = 5cmH2O and PaO2/FiO2 ratio <300.
Animals
Experiments were carried out in accordance with Israeli law and approved by the Institutional Animal Care and Use Committee (IACUCs) at the Israel Institute for Biological Research (Ness Ziona, Israel). Treatment of animals was in accordance with regulations outlined in the US Department of Agriculture (USDA; Washington, DC USA) Animal Welfare Act and the conditions specified in the National Institute of Health (NIH; Bethesda, Maryland USA) Guide for Care and Use of Laboratory Animals (2015).
Female piglets (Topigs 20, 14-24kg, aged 10-15 weeks, n = 17) were obtained from an approved commercial source (van Beek; Netherlands), fed standard pig diet, and housed in a purpose-built animal holding facility for four to eight days prior to the beginning of the experiment. Animals were allowed access to water ad libitum and to four percent body weight food per day. Fifteen hours before the experimental procedure, food was withdrawn, while water remained freely available.
Drugs and Chemicals
Drugs used were: Ketamine (Vetoquinol Lure; France), Midazolam (Rafa Laboratories; Israel), Xylazine (Eurovet Animal Health; Netherlands), epinephrine and Standard Solution [0.9% NaCl + 5% glucose] (Teva; Israel), as well as oleic acid (OA) and ethanol (Sigma-Aldrich, Chem-impex, purity 99.5%).
Surgical Preparations
Pigs were anesthetized with intramuscular ketamine and xylazine (20mg/kg and 2mg/kg, respectively). Anesthesia was maintained with inhaled isoflurane (2%-4% in 100% oxygen) during insertion of arterial and venous catheters. After vascular catheterization, anesthesia was switched to a continuous infusion of Ketamine (3mg/kg/hr) and Midazolam (0.1mg/kg/hr) using a syringe pump (GH Plus, Alaris; Franklin Lakes, New Jersey USA). Five ml/kg/20 minutes infusion 0.9%NaCl/5%Dextrose (“Standard”) solution was followed by continuous infusion at 3ml/kg/hr. Endotracheal intubation was performed with a cuffed endotracheal tube (5.5-6.0mm). Tube positioning was verified by capnography and auscultation. A single lumen catheter (16GX20cm, Biometrix; Jerusalem, Israel) was percutaneously introduced into the femoral artery and a triple lumen catheter (8.5FrX20cm, Biometrix) into the femoral vein.
The Improved Technique of OA ARDS Induction (Figure 1)
Repeated intravenous (IV) boluses of OA/saline emulsion is a well-established ARDS model. Reference Grotjohan, Van der Heijde, Jansen, Wagenvoort and Versprille16 However, according to the literature and to preliminary experiments, it results in wide variability of individual response. Following a pilot study on four pigs, it has been established that administering a fresh solution of 50% OA in ethanol using syringe-pump into the distal lumen of the central-vein line catheter, simultaneously with standard solution through the proximal lumen of the catheter, is more repeatable and convenient than giving recurrent doses of stirred OA/saline emulsion. This technique allowed a rate-controlled administration adjusted to the individual physiological response of each animal.
Experimental Paradigm.
A fresh solution of OA in 96% ethanol (1:1) was prepared on the day of the experiment and was administered using syringe pump at 0.17mg/kg/hr. Initial dose was 0.075ml/kg of pure OA. If the ARDS criteria were not met, additional doses of 0.0125ml/kg OA were infused at five-minute intervals, with reassessment of ARDS status. The OA infusion often caused rapid blood pressure (BP) drops; whenever systolic blood pressure (SBP) decreased below 90mmHg, OA administration was stopped for a few minutes, and if SBP dropped below 80mmHg, 5ml/kg fluid bolus was given. If SBP fell below 60mmHg, 0.025mg IV epinephrine boluses were administered. Forty-five minutes after the last administration of OA, to allow development of ARDS, PaO2/FiO2 ratio was calculated again before reducing the FiO2 to 0.21 and the beginning of the experiment.
Experimental Period
After ARDS induction, FiO2 was reduced from 0.4 to 0.21. When SpO2 decreased to <80%, PEEP (initially 5cmH2O) was increased by 3cmH2O every three minutes, up to 20cmH2O. If SpO2 remained <80%, Ti was increased by 0.2 seconds every three minutes (up to 1.4 seconds). If these maneuvers failed to result in SpO2 > 80 (only in three of 17 pigs), FiO2 was increased by 0.05 increments every three minutes. Hemodynamic variables were monitored (Carescape B650, GE Healthcare; Rehovot, Israel), collected onto a laptop computer at 10-second intervals, via S/5 collection software (GE Healthcare). Ventilator data were recorded manually every 10 minutes from the machine screen. Arterial blood gas (ABG) values and serum lactate were measured using an i-STAT Blood Gas Analyzer (Abbott Laboratories; Chicago, Illinois USA). Data were collected for one hour after room air ventilation was started.
Physiologic Monitoring
Mechanical ventilation was provided with a Hamilton C1 ventilator (Hamilton Medical; Bonaduz, Switzerland). Initial ventilator settings were Synchronized Intermittent Mandatory Ventilation (SIMV), 20 breaths/minutes, FiO2 = 0.4, mandatory tidal volume 10ml/kg, Ti = 1.0 second, PEEP = 5cmH2O.
Threshold SpO2
The lower limit of SpO2 was arbitrarily set at 80%, in order to simulate rather severe disease, which would challenge caregivers in a disaster or low-resource situation and require a resource-allocation decision. The model was thus calibrated to “start low” and show that PEEP alone or slight elevation of FiO2 will maintain life. There is physiological evidence that saturation in the 80% range does not adversely affect brain metabolism, Reference Ainslie, Shaw and Smith17 and clinicians encounter these levels often.
Statistical Analysis
Data are presented in Figure 2 and Figure 3 as means and 95% confidence intervals (CI). For each pig, the momentary measurements of the hemodynamic variables were aggregated into 15-minute groups. Such a one-quarter of an hour group, t, contained all the measurements taken in the time range [t–7.5 min, t+7.5 min). For each group t, a temporal average value was calculated. Then, for each one-quarter of an hour t, the mean and standard deviation (SD) of the temporal means over the pigs were calculated. The statistical calculations were performed with Microsoft Office Professional Plus 2013 Excel (Microsoft Corp.; Redmond, Washington USA) using the functions: AVERAGEIF and STDEV.
Cardiorespiratory Data of 14 Pigs [mean (95% CI)] During and After Oleic Acid-Induced Acute Respiratory Distress Syndrome.
Oxygen Saturation and Positive End Expiratory Pressure of 14 Pigs [mean (95% CI)] During and After Oleic Acid-Induced Acute Respiratory Distress Syndrome.
Ethical Approval and Consent to Participate
Animal Study—Experiments were carried out in accordance with Israeli law and approved by the IACUC at the Israel Institute for Biological Research.
Availability of Data and Materials—The data generated or analyzed during this study are included in this published article (and its supplementary information file; available online only).
Results
Induction of ARDS
The experimental paradigm is depicted in Figure 1. Mean PaO2/FiO2 ratio after OA infusion was 169 (SD = 42). Forty-five minutes later, the mean PaO2/FiO2 ratio was 163 (SD = 34). Thirty minutes after FiO2 was reduced to 0.21, mean PaO2/FiO2 ratio was 237 (SD = 41), and 30 minutes later 248 (SD = 59; Table S1 [available online only]). Peak airway pressure (Ppeak) was 17.7 (SD = 3.4) cmH2O immediately after OA infusion and increased to 30 (SD = 6) cmH2O. Static lung compliance (Cstat) was reduced by approximately one-half, from 17.2 (SD = 3.8) to 9.4 (SD = 2.3) ml/cmH2O (Table 1).
Ventilation Parameters of 14 Pigs
Note: The parameters exhibited as Mean (SD).
Abbreviations: Cstat, Static Lung Compliance; OA, oleic acid; Pmean, Mean Airway Pressure; Ppeak, Peak Airway Pressure.
Cardiorespiratory Responses
Intermittent, rapid BP drops were noted during OA infusion (Figure S1; available online only), then BP stabilized. Figure 2 shows the average heart rate (HR), mean arterial pressure, and end tidal CO2 (EtCO2) of 14 pigs that maintained SpO2 > 80% at FiO2 = 0.21 (additional data are provided in the supporting information; available online only).
Arterial Blood Gas (ABG) Analysis
No acidosis or hypercarbia were detected during the entire process, with arterial pH starting at 7.48 (SD = 0.03) and being 7.42 (SD = 0.05) at the end of the experiment (Table 2). As expected, the PaO2 decreased after the FiO2 decrease. Lactic acid increased during induction of ARDS from 0.7 (SD = 0.3) mmol/L to 4.7 (SD = 2.5) after OA, then remained unchanged at 4.9 (SD = 3.0).
Arterial Blood Gas (ABG) Analysis (N = 14)
Note: The parameters exhibited as Mean (SD).
Abbreviations: OA, oleic acid; PaCO2, partial pressure of carbon dioxide; PaO2, partial pressure of oxygen.
SpO2 versus Interventions
The SpO2 decreased during ARDS induction period, then it further dropped sharply with FiO2 reduction to 0.21 (Figure 3). Increasing PEEP resulted in increased SpO2 back to >80%. In some pigs, when the SpO2 increased, even a reduction in PEEP was possible. Only in three of 17 pigs, PEEP elevations and Ti prolongation were not sufficient and a minimal elevation of FiO2 was required (Figure S19; available online only).
Discussion
The present study aimed to employ a validated animal ARDS model to determine the feasibility of using room air to mechanically ventilate subjects with moderately severe ARDS. The conceptual framework of the study was the need to conserve scarce medical oxygen in situations such as major disasters, prolonged humanitarian crises, and in regions suffering from chronic medical oxygen shortages in normal times, as well as bolster the emerging understanding of the need to minimize FiO2.
Medical oxygen scarcity is a reality. Mowafi found that in 2015, most Syrian hospitals in the civil war zone did not have reliable access to medical oxygen. Reference Mowafi, Hariri and Alnahhas18 Contini reported that 30% of rural Afghan hospitals do not have oxygen supplies on an assured basis. Reference Contini, Taqdeer and Cherian19 Tran showed that less than one-half of the hospitals in Haiti had a reliable oxygen supply in 2015, Reference Tran, Saint-Fort and Jose20 as did only one-third of hospitals in Gambia in 2011, Reference Iddriss, Shivute and Bickler21 and 17% of sub-district hospitals in Bangladesh in 2017 Reference Loveday, Sachdev and Cherian22 or Cameroon. Reference Kouo-Ngamby, Dissak-Delon, Feldhaus, Juillard, Stevens and Ekeke-Monono23 The experience of the first author in this study (PH) while operating an ICU in Turkey during the 1999 earthquake there indicated that critical oxygen shortages occurred frequently and were often unpredictable. Reference Halpern, Rosen and Carasso24
The issue of “permissive hypoxemia” is still under study, with variable results. Reference Abdelsalman and Cheifetz25 The present study was, however, explicitly undertaken to address a scenario whereby multiple patients require respiratory support, mechanical or manual ventilators are available, but oxygen is not or is in limited supply. The alternatives are, to put it bluntly, either to allow many or most of these patients to die, or to provide them with a viable chance of survival, if research shows this option to be feasible at all. The authors thus chose a rather arbitrary lowermost value of SpO2 of 80% as a pragmatic, though physiologically acceptable, threshold for the study.
The two most important initial interventions when mechanically ventilated patients become hypoxemic are increasing FiO2 and increasing PEEP. In a study by Lawless Reference Lawless, Tobias and Mayorga26 in an OA-induced ARDS model in pigs exposed to a simulated altitude of 2440 meters, PEEP was increased gradually from 5cmH2O to 12.5cmH2O, or FiO2 was increased from 0.21 to 1.0. Fifty percent of FiO2 interventions did not reach target PaO2 while all PEEP animals did. The authors concluded that increases in PEEP are more reliable than increases in FiO2 for correcting altitude-induced hypoxia in this model. Admittedly, the response to PEEP is variable and difficult to predict. Among other variables, it depends on the recruitable lung fraction. Reference Gattinoni, Caironi and Cressoni27 A systematic review Reference Goligher, Hodgson and Adhikari28 showed that both PEEP and lung recruitment maneuvers were beneficial in ARDS.
Most studies on ARDS, as well as reviews and textbooks, do not mention the possibility, and certainly not the need, to decrease FiO2 to very low levels even when possible. In spite of the many studies cited above indicating that lowering FiO2 in critically ill patients may be beneficial, it seems that it is usually assumed that oxygen is non-toxic at FiO2<0.4 and there is therefore no incentive, and certainly no imperative, to decrease FiO2 to below 0.4, even when clinically feasible. Helmerhorst, et al showed that physicians in ICUs believe they should use the lowest possible FiO2, yet analysis of actual data indicated that the average FiO2 administered in >100,000 instances was 0.4-0.5, higher than the studied physicians indicated as optimal. Reference Helmerhorst, Schultz and Van der Voort29
In disasters and in underserved regions, some casualties may perhaps not need oxygen supplementation at all, or only very little, provided that mechanical ventilatory maneuvers are employed. If it were to be shown that at least some patients may not require oxygen enrichment or only minimal amounts, precious oxygen might be saved. Additionally, lives may be saved by not triaging potentially salvageable casualties as “expectant” (ie, not expected to survive) because of the perceived futility of instituting mechanical ventilation in the face of limited oxygen supplies.
The actual use of FiO2 = 0.21 in ventilated patients has rarely been tested, and no studies have been found of a validated laboratory ARDS model looking at this issue. The purpose of the present study was therefore to test the assumption that in a moderately-severe ARDS model in pigs, mechanical ventilation with room air is feasible in most instances, or that only slight increases in FiO2 may be sufficient when mechanical ventilatory maneuvers, such as adjusting PEEP and Ti, are not sufficient.
In the present study, PaO2/FiO2 ratio improved during the experimental procedure. This is either attributable to the increasing PEEP or to the well-known effect of FiO2 on the P/F ratio, whereby decreasing FiO2 to below 0.4 results in an increase in the ratio. Reference Aboab, Louis, Jonson and Brochard30 Also, other indicators of the severity of ARDS remained constant, such as decreased lung compliance.
In a small subgroup of animals, increasing PEEP was insufficient to increase SpO2 to the predetermined level, but a slight increase of FiO2 was enough. This group exemplifies the need for dynamic clinical decision making with such patients, but does not invalidate the basic message of the study.
Limitations
From practical reasons, the study period was relatively short and did not address possible further deterioration of ARDS for extended period of time. Also, other lung volume recruiting maneuvers were not applied, which might potentially further facilitate oxygen sparing. Further studies need to look at these issues, as well as use other models of respiratory failure (eg, infectious or primary ventilatory failure).
Conclusions
The present study indicates that in a relevant and well-established animal model of hypoxemic respiratory failure, mechanical ventilation with room air or with small amounts of added oxygen was feasible. Admittedly in humans, some patients with respiratory failure go on to develop worsening ARDS due to the original insult, Ventilator Associated Pneumonia (VAP), or lung atelectasis. This may necessitate additional oxygen supplementation, even after an initial period of satisfactory oxygenation on FiO2 = 0.21. Regardless, the data shown here indicate that automatic use of high FiO2 and the high oxygen utilization it entails may not be obligatory. Increased PEEP may be used in lieu of increased FiO2, as well as other interventions that had not been explored in the present study, such as high I:E ratios, ventilatory modes such as PCV, APRV, the prone position, and other lung volume recruiting maneuvers. The present study thus addresses the practical basis for the ethical and practical paradigm of mechanical ventilation in disasters and underserved areas, which assumes that oxygen is mandatory in respiratory failure and is a rate-limiting factor in care capacity allocation. Further studies are needed before any paradigm changes are considered.
Acknowledgments
The authors would like to thank Dr. Ziv Klausner for the statistical consultation.
Authors’ Contribution (using the CRediT Taxonomy)
PH: Conceptualization, Methodology, Validation, Investigation, Writing - Original Draft, Writing - Review & Editing, Visualization. MG: Conceptualization, Methodology, Validation, Formal Analysis, Investigation, Data Curation, Visualization. KB: Conceptualization, Methodology, Investigation. AR: Validation, Resources. AW: Methodology, Validation, Formal Analysis. GY: Methodology, Validation, Resources, Data Curation. SK: Conceptualization, Methodology, Validation, Investigation, Writing - Original Draft, Writing - Review & Editing, Visualization, Supervision, Project Administration. All authors read/approved final manuscript.
Conflicts of interest
none
Supplementary Material
To view supplementary material for this article, please visit https://doi.org/10.1017/S1049023X20001016
