Obstructive Lung Disease and Dynamic Hyperinflation

Obstructive respiratory failure is a mismatch between the pathways available for expiration, the volume that must leave, and the time available before another breath. The resulting operating lung volume influences gas exchange, muscle effectiveness, ventilator interaction, and circulation. Understanding those links is more useful than treating a high airway pressure or carbon dioxide concentration as an isolated problem.

76.1 Similar Airflow Obstruction Can Arise from Different Structures

Asthma and chronic obstructive pulmonary disease, or COPD, can both produce severe expiratory obstruction, but the underlying mechanical changes need not be identical. Airway smooth-muscle contraction, inflammatory wall thickening, mucus, loss of parenchymal support, and destruction of gas-exchanging units contribute in different proportions. Acute infection or another insult can add several changes to a previously compensated system.

In asthma, narrowing may change rapidly and heterogeneously. Some pathways remain patent while others become severely narrowed or closed. In COPD, small-airway remodeling and emphysematous loss of recoil can coexist, so increasing airway caliber does not necessarily restore the previous pressure–flow relationship of the whole lung.

Hogg and colleagues examined surgically obtained tissue from 159 people spanning different COPD severity categories. Greater obstruction was associated with small-airway wall changes and inflammatory mucous material in the lumen. This human pathological study supports multiple structural contributors; it does not establish which one dominates an individual acute exacerbation or demonstrate a particular treatment's effect. 1

Loss of parallel pathways also matters. A lung with fewer functioning small airways is not equivalent to one uniformly narrowed tube. Remaining pathways can receive disproportionate tidal ventilation while poorly communicating regions retain gas or contribute little useful exchange.

The artificial airway adds another possible bottleneck in series. Tube narrowing, secretions, a kink, or an expiratory circuit problem can worsen emptying independently of the patient's bronchi. A sudden deterioration therefore requires reassessment of the whole pathway rather than attribution to a familiar diagnostic label.

These distinctions explain why wheeze, inspiratory resistance, and clinical history cannot specify the complete current mechanics. Very low flow can produce little audible wheeze despite serious obstruction. The physiological assessment asks which regions communicate, what limits their emptying, and how disease, equipment, and delivered ventilation interact.

76.2 Expiratory Flow Limitation Changes the Effect of Effort

Expiration requires a pressure gradient from alveoli toward the airway opening. With a fixed resistance, greater driving pressure increases flow. A deformable intrathoracic airway behaves differently when rising surrounding pressure narrows a critical segment: beyond a point, more expiratory effort may produce little additional flow at that lung volume.

During active expiration, pleural pressure raises alveolar pressure but also compresses susceptible airways. Pressure falls along the pathway toward the mouth. A segment can therefore encounter insufficient internal pressure to resist the surrounding pressure even though the alveolar-to-mouth gradient is large.

The equal-pressure-point explanation provides an intuitive picture of this dynamic compression. Wave-speed analysis additionally incorporates the airway's pressure–area behavior and gas velocity. Dawson and Elliott developed a theoretical account linking maximal flow to local airway properties, recoil, and gas density, while explicitly recognizing limits when friction and actual choke-point conditions are unknown. 2

In emphysema, reduced recoil lowers the pressure available to move gas and reduced tissue tethering permits compression. In other obstructed regions, luminal narrowing or closure may dominate. High resistance and flow limitation can coexist, but they are not interchangeable findings.

Increasing expiratory muscle activity cannot guarantee adequate emptying once flow is limited. It may raise intrathoracic pressure, consume energy, and alter cardiac loading without producing the desired extra exhaled volume. Conversely, persistent end-expiratory flow alone does not prove flow limitation: a resistive but nonlimited pathway can simply lack enough time.

This distinction also explains the conditional effects of pressure at the airway opening. Downstream pressure may have little effect on upstream flow over part of a flow-limited pathway's operating range. In another region it can immediately reduce the expiratory gradient. Neither COPD nor asthma guarantees one uniform response across the lung.

76.3 Rising Operating Volume Is Both an Adaptation and a Constraint

An obstructed patient can increase expiratory flow by operating at a higher lung volume, where recoil and airway traction are often greater. This helps explain how a new breathing equilibrium can be reached despite persistent obstruction. The adaptation carries costs because it leaves less room for the next inspiration and changes muscle geometry and vascular loading.

Static hyperinflation reflects an altered passive equilibrium, particularly when recoil is reduced. Dynamic hyperinflation is the additional elevation of end-expiratory volume caused by incomplete emptying under the current breathing conditions. Both can be present, and prolonging expiration cannot be assumed to reverse the static component or release every closed region.

At a stable elevated operating volume, each exhaled tidal volume can equal the preceding inspired volume. The system has stopped accumulating more gas, but it has not necessarily returned to its passive equilibrium. A reassuring tidal-volume balance therefore cannot exclude important hyperinflation.

Inspiratory capacity is the volume available between the current end-expiratory level and total lung capacity. If total capacity is sufficiently stable, a decline in inspiratory capacity can indicate a rising end-expiratory level. The inference also requires an adequate inspiratory maneuver; changing effort or inability to reach total capacity can distort it.

O'Donnell and colleagues examined operating-volume changes during exercise in 105 patients with COPD. Dynamic hyperinflation frequently curtailed tidal-volume expansion as demand increased, with substantial variation among patients. This is useful human evidence for the volume constraint, while exercise observations cannot be transferred directly into an ICU measurement protocol. 3

During critical illness, fever, metabolic demand, anxiety, gas-exchange inefficiency, and respiratory loading can increase the ventilation required. If breathing frequency rises while expiration is already constrained, operating volume can rise further. The patient then approaches an inspiratory ceiling despite increasing neural drive, making a small tidal volume compatible with intense effort.

76.4 Heterogeneous Emptying Makes Additional Time Unequally Valuable

A lung region's emptying depends on its resistance, compliance, airway patency, and surrounding pressures. In a simple passive linear compartment, the time constant is resistance multiplied by compliance. Real obstructed lungs contain a distribution of emptying behaviors, and those behaviors can change as volume falls.

For an illustrative passive interval, let each of two compartments initially contain 0.25 L above its own equilibrium. Assume fixed time constants of 0.5 seconds in the fast compartment and 2 seconds in the slow compartment, with both communicating to an unchanged downstream pressure.

Here t is elapsed expiratory time and τ the compartment's time constant, both in seconds; e is the base of natural logarithms. Volumes are in liters. After 1.5 seconds, the fast compartment retains 0.25 × e^(−3) ≈ 0.0124 L, while the slow compartment retains 0.25 × e^(−0.75) ≈ 0.1181 L. The combined remainder is approximately 0.1305 L.

After 3 seconds, the corresponding values are approximately 0.00062 and 0.0558 L, totaling 0.0564 L. The additional time mainly helps the slow compartment because the fast one has already nearly emptied.

These are constructed values for one interval, not a prediction of total steady-state trapped volume. Fixed resistance, compliance, equilibrium, and open communication were assumed. Dynamic airway compression or closure can invalidate the exponential description, and a closed compartment may not empty simply because more time is allowed.

The example nevertheless explains why neither one global time constant nor one expiratory-time ratio describes every region. It also explains diminishing return late in expiration: when residual flow is already very low, another second may release little volume. The relevant question is the measured volume and pressure response, rather than whether a familiar timing number has been reached.

76.5 Hyperinflation Increases Load While Reducing Effective Muscle Reserve

Intrinsic positive end-expiratory pressure, or intrinsic PEEP, can impose a threshold before inspiratory flow begins. In passive communicating regions, it is the end-expiratory alveolar pressure above applied PEEP. The muscles must lower alveolar pressure below pressure at the airway opening before gas reverses direction. This effort precedes the elastic and resistive work of the incoming tidal breath.

Suppose a constructed communicating region has end-expiratory alveolar pressure of 11 cm H₂O while airway-opening pressure is 4 cm H₂O. Their difference is 7 cm H₂O. An inspiratory effort initially reducing alveolar pressure by only 5 cm H₂O would not yet establish inward flow under those simplified conditions. Actual trigger requirements, regional communication, and respiratory muscle relaxation modify the event.

At the same time, a shortened, flattened diaphragm may generate less pressure for a given activation and confront a less favorable relationship between pressure and volume displacement. Increased drive can therefore coexist with limited useful tidal expansion. Chronic disease can produce muscular adaptations, so an unfavorable operating position should not automatically be equated with irreversible weakness or proven fatigue.

Smith and Marini studied ten mechanically ventilated patients with exacerbated chronic airflow obstruction under assisted and controlled conditions. Applied PEEP could narrow the end-expiratory alveolar-to-airway pressure difference and reduce inspiratory effort during assistance. This demonstrates a potential threshold-unloading mechanism, not that every obstructed patient benefits from added pressure. 4

Ineffective inspiratory attempts can occur when effort is insufficient to overcome the threshold before neural inspiration ends. The displayed ventilator rate can then underestimate the patient's actual effort frequency. A patient apparently breathing slowly on the monitor may be making frequent unsuccessful attempts.

The load–capacity balance consequently includes threshold pressure, resistance, elastic loading, respiratory frequency, muscle effectiveness, and systemic oxygen supply. Reducing visible respiratory rate without understanding these components does not establish adequate unloading. Chapter 27 develops the distinction between muscle force, work, pressure–time expenditure, weakness, and fatigue.

76.6 Pressure Measurements Reveal Only the Compartments That Communicate

During passive ventilation, a valid inspiratory hold estimates an equilibrated end-inspiratory pressure, whereas an expiratory hold estimates total PEEP in communicating regions. Their interpretation requires suitable timing, no important leak, and absence of substantial muscle activity. A displayed pause is not proof that those conditions were achieved.

Peak pressure during flow contains pressure losses across the tube, airways, and other resistive components. Raising inspiratory flow can increase that peak without an equivalent increase in end-inspiratory alveolar distension. Conversely, a high plateau suggests a different pressure burden, but still requires lung-volume and chest-wall context.

An expiratory hold does not necessarily reveal pressure behind closed airways. Leatherman and Ravenscraft described four ventilated patients with severe asthma in whom low measured auto-PEEP coexisted with findings suggesting substantial hyperinflation. Widespread closure was a proposed explanation for underestimation, not a directly measured pressure in every isolated region. 5

The opposite problem occurs when active expiration raises surrounding pressure. Lessard and colleagues studied assisted patients with expiratory muscle activity and showed that increased measured intrinsic PEEP could occur without an additional increase in end-expiratory lung volume. Pressure attributable to active muscles must therefore be distinguished from passive recoil. 6

Esophageal and gastric pressure measurements can help partition effort and identify expiratory recruitment, but their placement, calibration, and regional representativeness matter. An early inspiratory fall in esophageal pressure can include relaxation of expiratory muscles rather than exclusively new inspiratory pressure generation.

The clinician consequently needs several observations: delivered volume, flow throughout expiration, valid holds when feasible, effort, and hemodynamics. A low measured intrinsic PEEP cannot certify an uninflated lung, and a high value does not by itself quantify retained milliliters. Sudden changes also warrant examination of tube and circuit rather than calculation alone.

76.7 Delivered Timing Determines the Opportunity to Empty

Respiratory frequency, tidal volume, inspiratory flow, pauses, and patient-triggered breaths jointly determine expiratory opportunity. A low mandatory rate does not guarantee a long expiration if additional assisted breaths occur or machine inflation continues after neural inspiration has ended.

For a constructed constant-flow breath, assume tidal volume 0.50 L, inspiratory flow 0.50 L/s, no pause, and frequency 20 breaths/min. Delivery takes 1 second. The cycle lasts 60 ÷ 20 = 3 seconds, leaving 2 seconds for expiration. Doubling flow to 1 L/s shortens delivery to 0.50 seconds and leaves 2.50 seconds, provided every other assumption remains unchanged.

The resulting peak pressure may rise because resistive pressure loss increased. That displayed change can accompany improved emptying rather than greater retained volume. Tuxen and Lane demonstrated strong effects of ventilatory pattern on hyperinflation and circulation in nine sedated, paralyzed patients with severe obstruction; peak pressure did not reliably track retained lung volume. Their historical settings are physiological experiments, not present-day prescriptions. 7

Longer expiration also has limits. In twelve ventilated patients with severe asthma, Leatherman and colleagues found that reducing respiratory rate lowered plateau pressure, but further prolongation produced relatively modest additional effects when late flow was already low. The response depends on where in the emptying trajectory the extra time is added. 8

During pressure support, slow flow decay can delay cycling into expiration. Tassaux and colleagues studied ten intubated obstructive patients and found that earlier cycling reduced delayed inflation, intrinsic PEEP, and ineffective attempts under their tested conditions. The study established physiological responses rather than an outcome-validated universal cycling fraction. 9

Reducing delivered volume or frequency can decrease the amount requiring expiration but also decrease alveolar ventilation. Increasing assistance may relieve effort yet enlarge tidal breaths or prolong inflation. The appropriate balance must be assessed through actual delivery and response, not the nominal support level alone.

76.8 External PEEP Has a Conditional Relationship to Hyperinflation

External PEEP increases pressure at the airway opening. In an assisted patient it can reduce the pressure difference that must be overcome before triggering. Whether this occurs without additional hyperinflation depends on flow limitation, airway stability, regional heterogeneity, and the resulting breathing pattern.

In a simplified flow-limited pathway, downstream pressure can increase over a range without materially changing upstream expiratory flow. Beyond that range it can impede emptying and increase operating volume. Other regions may reach this transition at different pressures, and a nonlimited pathway can respond differently from the outset.

Ranieri and colleagues studied nine COPD patients during controlled ventilation and demonstrated a range of applied pressure that did not increase lung volume before higher pressures produced further inflation and adverse circulatory changes. The experimental relationship cannot be converted into one fixed fraction of measured intrinsic PEEP for every patient or ventilatory state. 10

Caramez and colleagues subsequently observed several responses in eight severely obstructed patients, including immediate overinflation, a threshold response, and paradoxical deflation under some conditions. This small physiological study shows why disease name and one baseline pressure cannot predict the response reliably. It does not establish routine benefit from escalating PEEP. 11

Subtraction can itself mislead. Suppose total PEEP remains 14 cm H₂O while external PEEP increases from 5 to 8. The calculated intrinsic component falls from 9 to 6 cm H₂O. That arithmetic alone does not demonstrate a fall in total pressure or lung volume, although threshold work might change in an appropriate assisted system.

A passive patient has no triggering effort to unload. The rationale and tolerance must then be judged through airway behavior, gas exchange, pressure, volume, and circulation. In either state, comparing valid plateau and total PEEP measurements at comparable tidal volumes is more informative than following the intrinsic subtraction alone, while effort and regional closure still limit interpretation.

76.9 Carbon Dioxide and Oxygen Reflect Distribution as Well as Ventilation

Minute ventilation is not synonymous with useful alveolar ventilation. Dead-space ventilation, heterogeneous perfusion, metabolic carbon dioxide production, and the distribution of tidal gas determine the ventilation required to control arterial carbon dioxide. Severe obstruction can increase that requirement while limiting the capacity to meet it.

At constant carbon dioxide production and under an interpretable steady-state approximation, carbon dioxide pressure varies inversely with effective alveolar ventilation. If illustrative ventilation falls from 4 to 3 L/min while initial arterial carbon dioxide is 45 mmHg, the proportional estimate becomes 45 × 4 ÷ 3 = 60 mmHg. This is a model comparison, not a precise prediction in a changing heterogeneous lung.

Pursuing the previous carbon dioxide through more frequent or larger breaths can increase hyperinflation, effort, and circulatory compromise. Conversely, tolerating a rise has consequences through pH, pulmonary vascular tone, cerebral circulation, and the patient's baseline condition. There is no universally safe carbon dioxide or pH boundary supplied by this tradeoff.

Oxygen-induced hypercapnia also has several contributors, rather than simply withdrawal of a single hypoxic drive. Robinson and colleagues measured ventilation and ventilation–perfusion distributions in 22 patients during COPD exacerbations. Responses to high inspired oxygen were heterogeneous; reduced ventilation characterized carbon dioxide retainers, while perfusion redistribution occurred in both response groups. Their observations discourage assigning every response to one mechanism. 12

Changes in hemoglobin's carbon dioxide carriage with oxygenation can contribute as well. The net arterial response depends on ventilation, distribution, carbon dioxide stores, and time. Adequate oxygenation remains necessary; the physiological lesson is to assess the gas-exchange response rather than assume that a higher oxygen concentration is always neutral.

Austin and colleagues' prehospital cluster-randomized trial found lower mortality with a titrated oxygen strategy than with the compared high-flow strategy in patients with presumed COPD exacerbations, including a confirmed-COPD subgroup. This supports that population-specific strategy comparison, not withholding oxygen during hypoxemia or transferring one protocol to every respiratory emergency. 13

76.10 Noninvasive Assistance Can Change the Load–Capacity Balance

Inspiratory assistance can supply part of the pressure needed to overcome resistance and inflate the respiratory system. Expiratory pressure may reduce threshold effort under suitable conditions. Together these effects can improve useful ventilation while allowing the patient's muscles to expend less pressure over time.

A mask preserves the natural airway but introduces leaks, interface pressure, dead space, and dependence on patient cooperation and airway protection. A device must distinguish patient effort from leak-related flow and terminate inflation appropriately. An apparent reduction in distress is meaningful only when accompanied by an acceptable overall trajectory.

Brochard and colleagues randomized 85 selected COPD-exacerbation patients recruited from a larger ICU population to noninvasive pressure support plus standard care or standard care alone. The noninvasive strategy reduced intubation and in-hospital mortality in that trial. Its selection and monitored setting are part of the evidence; the findings do not establish that every patient with obstruction can safely avoid an invasive airway. 14

Plant and colleagues studied 236 patients with COPD exacerbations and mild-to-moderate acidosis on general respiratory wards using a predefined support strategy. Fewer patients met criteria for treatment failure, and in-hospital mortality fell. These findings extended the evidence to that studied ward population. The intervention was an organized care strategy, not merely application of a pressure value. 15

The COPD evidence cannot be transferred automatically to acute asthma, severe airway-protection problems, or a different cause of respiratory failure. Anatomical obstruction, secretion handling, consciousness, gas exchange, cardiovascular state, and the ability to deliver effective assistance remain relevant to the method's feasibility.

Serial assessment matters because assistance can reduce effort, worsen delayed cycling, or fail to meet demand. More support may produce less useful unloading if it increases retained volume. A favorable mask display or oxygen saturation cannot substitute for observing effort, carbon dioxide/pH evolution, interaction, and circulation. This physiological interpretation supports timely reassessment without constructing a universal escalation algorithm.

76.11 Circulation and Transitions Can Expose Hidden Mechanical Burden

Hyperinflation can raise surrounding thoracic pressure and reduce the effective gradient for venous return. Increased lung volume can also raise right ventricular outflow load through pulmonary vascular effects. These processes can coexist, while their magnitude depends on lung–chest-wall mechanics, vascular volume, and ventricular reserve.

Measured venous pressure may rise as forward output falls. It then contains information about external pressure and congestion rather than proving useful preload. Large spontaneous inspiratory swings create another loading state, potentially increasing venous inflow and left ventricular transmural ejection load while imposing substantial respiratory work.

Transition to invasive ventilation changes effort, vascular tone, pressure transmission, and possibly medication exposure simultaneously. New hypotension should prompt assessment of hyperinflation alongside tube obstruction, pneumothorax, arrhythmia, and progression of disease. A response consistent with trapped volume does not exclude a second cause or justify one blanket disconnection maneuver.

Transition away from support can expose the opposite mismatch. The patient must again provide threshold, resistive, and elastic pressure and may shorten expiration through rapid shallow breathing. Failure cannot be assigned automatically to a weak diaphragm without examining how the load and operating volume changed.

Jubran and Tobin compared COPD patients who failed a spontaneous-breathing trial with those who succeeded. Failure was associated with worsening mechanics, increased intrinsic PEEP and pressure–time expenditure, and inefficient carbon dioxide clearance. Similar maximal inspiratory pressures before the trial did not predict identical load tolerance. This physiological comparison does not define a universal weaning target or prove that muscle fatigue caused every failure. 16

The important observation is the time course across a transition: what changed in delivered ventilation, effort, flow, pressure, gases, and circulation. Chapters 39–42 and 73 develop the cardiovascular mechanisms. Here they explain why a respiratory adjustment cannot be judged solely by oxygen saturation or a visually improved airway-pressure tracing.

76.12 Integrating an Obstructed Assisted Breath

Consider a constructed patient with exacerbated COPD receiving assisted ventilation. Recorded muscle signals show 28 inspiratory attempts/min, but only 20 generate machine-assisted breaths. Assuming no auto-triggering or double delivery, eight attempts are ineffective: 8 ÷ 28 × 100 ≈ 29% of recorded efforts. The denominator is all efforts, not delivered breaths, and this illustrative fraction is not a treatment threshold.

Expiration continues into the next inflation, tidal volumes vary, and some machine breaths extend beyond the end of neural inspiration. These findings suggest interacting emptying and synchrony problems. They do not establish whether narrowing, secretions, artificial-airway resistance, excess assistance, demand, or several factors are dominant.

Suppose peak pressure is high, but a technically valid passive assessment later shows a lower plateau and elevated total PEEP. The flow-related pressure component and retained end-expiratory pressure require separate explanations. If muscle activity persists during the maneuver, the assessment must instead retain that uncertainty rather than present the displayed value as passive mechanics.

Changes directed at the identified pathways may improve airway patency, shorten inappropriate inflation, reduce a triggering threshold, or reduce avoidable ventilatory demand. These actions address different parts of the sequence. Their success is assessed by useful tidal delivery, effort, emptying, gas exchange, and circulation rather than by one improved number.

After a change, delivered tidal volumes could become smaller while more neural efforts successfully trigger, or peak pressure could increase because inspiratory flow rose while expiration improved. Such responses are not contradictory. Conversely, better displayed carbon dioxide can accompany worsening inflation if greater ventilation was obtained at excessive mechanical cost.

The remaining questions are whether the patient is achieving adequate exchange with tolerable work, whether operating volume is stable at an acceptable physiological state, and whether cardiac flow and organ perfusion are preserved. Serial measurements must use comparable conditions or explicitly identify the condition that changed.

Obstructive respiratory failure becomes coherent when flow capacity, volume, time, effort, and circulation are followed together. The goal of physiological interpretation is to identify and reassess the limiting mechanism, not to force every patient into one pressure, timing, or gas target.

References

1. Hogg JC, Chu F, Utokaparch S, et al. The nature of small-airway obstruction in chronic obstructive pulmonary disease. N Engl J Med. 2004;350:2645–2653. Original human pathological study.

2. Dawson SV, Elliott EA. Wave-speed limitation on expiratory flow—a unifying concept. J Appl Physiol Respir Environ Exerc Physiol. 1977;43:498–515. Original theoretical study.

3. O'Donnell DE, Revill SM, Webb KA. Dynamic hyperinflation and exercise intolerance in chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2001;164:770–777. Original clinical physiology study.

4. Smith TC, Marini JJ. Impact of PEEP on lung mechanics and work of breathing in severe airflow obstruction. J Appl Physiol. 1988;65:1488–1499. Original clinical physiology study.

5. Leatherman JW, Ravenscraft SA. Low measured auto-positive end-expiratory pressure during mechanical ventilation of patients with severe asthma: hidden auto-positive end-expiratory pressure. Crit Care Med. 1996;24:541–546. Original clinical observations.

6. Lessard MR, Lofaso F, Brochard L. Expiratory muscle activity increases intrinsic positive end-expiratory pressure independently of dynamic hyperinflation in mechanically ventilated patients. Am J Respir Crit Care Med. 1995;151:562–569. Original clinical physiology study.

7. Tuxen DV, Lane S. The effects of ventilatory pattern on hyperinflation, airway pressures, and circulation in mechanical ventilation of patients with severe air-flow obstruction. Am Rev Respir Dis. 1987;136:872–879. Original clinical physiology study.

8. Leatherman JW, McArthur C, Shapiro RS. Effect of prolongation of expiratory time on dynamic hyperinflation in mechanically ventilated patients with severe asthma. Crit Care Med. 2004;32:1542–1545. Original clinical physiology study.

9. Tassaux D, Gainnier M, Battisti A, Jolliet P. Impact of expiratory trigger setting on delayed cycling and inspiratory muscle workload. Am J Respir Crit Care Med. 2005;172:1283–1289. Original clinical physiology study.

10. Ranieri VM, Giuliani R, Cinnella G, et al. Physiologic effects of positive end-expiratory pressure in patients with chronic obstructive pulmonary disease during acute ventilatory failure and controlled mechanical ventilation. Am Rev Respir Dis. 1993;147:5–13. Original clinical physiology study.

11. Caramez MP, Borges JB, Tucci MR, et al. Paradoxical responses to positive end-expiratory pressure in patients with airway obstruction during controlled ventilation. Crit Care Med. 2005;33:1519–1528. Original clinical physiology study.

12. Robinson TD, Freiberg DB, Regnis JA, Young IH. The role of hypoventilation and ventilation-perfusion redistribution in oxygen-induced hypercapnia during acute exacerbations of chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 2000;161:1524–1529. Original clinical physiology study.

13. Austin MA, Wills KE, Blizzard L, Walters EH, Wood-Baker R. Effect of high flow oxygen on mortality in chronic obstructive pulmonary disease patients in prehospital setting: randomised controlled trial. BMJ. 2010;341:c5462. Original cluster-randomized trial.

14. Brochard L, Mancebo J, Wysocki M, et al. Noninvasive ventilation for acute exacerbations of chronic obstructive pulmonary disease. N Engl J Med. 1995;333:817–822. Original randomized trial.

15. Plant PK, Owen JL, Elliott MW. Early use of non-invasive ventilation for acute exacerbations of chronic obstructive pulmonary disease on general respiratory wards: a multicentre randomised controlled trial. Lancet. 2000;355:1931–1935. Original randomized trial.

16. Jubran A, Tobin MJ. Pathophysiologic basis of acute respiratory distress in patients who fail a trial of weaning from mechanical ventilation. Am J Respir Crit Care Med. 1997;155:906–915. Original clinical physiology study.