Mechanical ventilation: past lessons and the near future

The ability to compensate for life-threatening failure of respiratory function is perhaps the signature technology of intensive care medicine. Unchanging needs for providing effective life-support with minimized risk and optimized comfort have been, are now, and will be the principal objectives of providing mechanical ventilation. Important lessons acquired over nearly half-a-century of ICU care have brought us closer to meeting them, as technological advances in instrumentation now effectively put this hard-won knowledge into action. Rising demand in the face of economic constraints is likely to drive future innovations focused on reducing the need for user input, automating multi-element protocols, and carefully monitoring the patient for progress and complications.


Introduction
Mechanical ventilation is instrumental in the rescue and maintenance of the patient with failing cardiorespiratory function. With passing time, the goals of ventilatory support have been refi ned to include not only eff ective life-support, but also minimized iatrogenesis and improved coordination between patient needs or demand and machine-delivered breathing cycles. Th e capacity of mechanical ventilators to ventilate and oxygenate eff ectively has steadily improved, while the caregiver has become aware of its potential to cause infection, hemodynamic consequences, and ventilator-induced lung injury. Once an inherently uncomfortable process that invariably required deep sedation and even paralysis to maintain, modern machines provide diverse options to reduce breathing work load, improve comfort, and enhance coordination. In this discussion I recount the important lessons we have learned during the positive pressure ventilation era, describe current developments, and suggest remaining problems and innovative approaches that point toward future progress.

Mechanical ventilation: a brief look back
Although primitive forms of mechanical ventilation were suggested or implemented in the fi rst half of the 20th century, ventilation with positive pressure emerged as an everyday technology only with the birth of the modern ICU in the early 1960s [1]. About that time, ventilatory equipment transitioned from negativepressure tanks that surrounded the patient to the familiar positive-pressure machines attached only through the airway and facilitate patient access. At fi rst, the ventilator or respirator was envisioned essentially as a push-pull bellows pump with which to move conditioned gas into and out of the lungs. In the fi rst decades of the 1900s, newly developed electric motor-driven pistons allowed enclosures for the patient's thorax and abdomen but prevented caregiver access without interrupting ventilatory support. Drinker-Shaw and Emerson machines were introduced into medical practice in relatively small numbers around 1930, and these came to be known as iron lungs [2]. By the early 1950s, relatively advanced tank-style ventilators were employed success fully during the polio epidemic; however, these negative pressure devices were cumbersome, worked best when the patient was suffi ciently conscious to prevent upper airway closure, and could not hope to support a patient with full-blown oxygenation failure. Spurred by this experience and by two war-time confl icts that occurred in rather quick succession, the value of deploying improved life-support technology became evident for both civilian as well as military applications. Th e roots of positive end-expiratory pressure (PEEP) and noninvasive ventilation also can be traced back to these early years [3].
Th e 1960s were a pivotal decade in the development of positive pressure ventilation, infl uenced by advances in physiology and surgery and the need to address the problems of postoperative atelectasis and the traumatic lung injuries of battlefi eld confl ict. Pressure cycled devices that delivered intermittent positive pressure were utilized on the general wards with the intent of helping a variety of patients breathe more deeply, aiding cough ing effi ciency, forestalling basilar collapse and im prov ing deposition of therapeutic aerosols. Simul taneously, machines that allowed the infl ation and defl ation phases to be unlinked (separately regulated) and that were expressly designed for sustained life-support of the critically ill were introduced into the newly formed ICUs [4]. Among the more purpose-refi ned of these early ICU machines was the Puritan-Bennett® MA-1, introduced in 1967. Th ese powerful units, less bulky and more purposedesigned than some contemporary anesthesia-based alter natives, were innovative and durable. But by today's standards they were infl exible, off ered only time-cycled, fl ow-regulated breathing, and provided simply a calibrated exhalation bellows for tidal volume deter mination and a needle gauge for airway pressure monitoring. Durable circuits were re-usable, airway suctioning was performed only during ventilator disconnections, fl ow was not displayed, and key ventilation alarms were attached externally.
Looking back, it is interesting to note that these MA-1 machines off ered scheduled sighs to be added when delivering breaths of lower amplitude [5]. Primed by the need to prevent atelectasis in healthy lungs during surgery, large tidal volumes of 10 to 20 ml/kg were the entrenched prescription at that time and normal blood gases were targeted, even in patients with catastrophic respiratory failure [6]. Th e design engineers were also clearly attempting to mimic natural breathing in their off ering of sinusoidal and square wave inspiratory fl ow patterns -those that are associated with the spontaneous selections made by the normal patient during unassisted breathing and by the patient with serious airfl ow obstruction. Expiratory retard could be applied in the latter cohort in the attempt to avert tidal expiratory airway collapse and to mimic pursed-lip breathing.
Th e clinician could manipulate only one variable at a time, so that a change of the imposed breathing pattern required sequential rather than simultaneous adjustment of frequency, fl ow rate, and tidal volume. Pressureassisted modes of ventilation suitable for the severely ill were not available. In those early days, PEEP -if used at all -was added externally, using valves with high resistance rather than integrated within the ventilator circuit [7]. Th e most popular mode of ventilation was assistcontrol with square wave fl ow, essentially because it was the only form of triggered assistance available for the adult with critical illness.
In the late 1960s, the syndrome of adult respiratory distress (ARDS) and its treatment by PEEP were described [8,9]. Pediatricians had primed adult intensivists by their experience with surfactant defi ciency-caused acute lung injury in neonates, but their well-developed and justifi ed concern for the problems of barotrauma and the use of pressure-based modes of ventilation in this population did not translate into adult caregiving until much later. Intubation for the prolonged periods needed to support respiratory failure using tubes sealed to the airway with high pressure gave rise to serious and often permanent laryngeal and tracheal injuries. Attempts to treat the lung gently during ARDS by undertaking extracorporeal gas exchange proved ill-fated, as the patients rescued with extracorporeal membrane oxygenation were very severely aff ected and late in their disease course. Materials and techniques of the time infl icted unacceptable injury [10].
Treatment of ARDS was one central driver of new approaches to respiratory failure, but clearly not the only one. How to provide partial support, recondition the respiratory muscles, and gauge readiness of the patient to assume the entire ventilatory workload were also preoccupy ing concerns of the day [11,12]. As adult clinicians gained more experience in managing such challenging problems, the need to address them effi ciently drove the incorporation of better monitoring as well as the radically new modes of assistance such as (synchronised) intermittent mandatory ventilation and PEEP without assisted breathing [13,14]. Over a relatively brief period of time, microprocessor controls and electronic waveform displays of pressure and fl ow became embedded into the machines them selves, enabling discoveries related to work of breathing, synchrony, and the eff ects of adjustments in frequency, PEEP, peak fl ow, and triggering paradigm on eff ort and dynamic hyperinfl ation [15,16].
Th e importance of improved monitoring and mode fl exibility became evident throughout the 1970s and 1980s, as laboratory and clinical investigations revealed the full potential for the ventilator to cause both obvious and hidden forms of lethal injury [17,18]. Awareness of the key roles of maximum transalveolar pressure and high tidal volume led to the approach of accepting higher partial pressure of carbon dioxide (permissive hypercapnia) as a necessary consequence of using smaller and safer tidal volumes to support, fi rst, intubated asthmatics [19] and later those with ARDS [20]. High-frequency jet ventilation and high-frequency oscillation were developed and tested as strategies for limiting the lung-damaging potential of maximum tidal pressure while recruiting the unstable lung units of infants with infant respiratory distress syn drome. Although jet ventilators were available early on, adult use of high-frequency oscillation awaited the development of capable machines in the late 1990s [21]. Inhalation of vasodilatory gas mixtures (nitric oxide) that promoted gas exchange through patent lung units fi rst gained popularity in the 1990s [22].
Pressure-regulated modes of ventilation (pressure support, pressure control, and their modern variants) were developed to address with relative safety the varying fl ow demands of the patient with cardiopulmonary disease. Th e ability to respond to the patient's changing fl ow demands, as well as the need to cycle in timely fashion into the exhalation phase, was introduced to machinery developed in the mid-1970s in the form of pressure support (pressure support ventilation) [21]. At fi rst, time-cycled pressure control (pressure control ventilation) was often implemented as inverse-ratio venti lation in the treatment of ARDS [23], an approach that has since faded from favor. In contrast, pressure support, assist-control, and synchronized inter mittent mandatory ventilation with either fl ow-con trolled or pressure-controlled breaths have become en trenched as the fl exible standard modes of ventilation for more than 30 years.
Observational studies and clinical trials testing the worth of traditional and innovative approaches to lung protection and gas exchange effi ciency characterized scientifi c eff orts in mechanical ventilation through the 1990s and into the fi rst decade of the 21st century [24,25]. Current-generation technology has responded admirably to emerging knowledge concerning iatrogenic upper airway damage, lung parenchymal injury, and the consequences of dys-synchrony [26]. Present-day approachesfor example, proportional assist ventilation and neurally adjusted ventilatory assist -are immeasurably more eff ective than before, but still need to eliminate imperfect integration with the patient's neural demands and underlying physiologic needs. Safety and coordination remain the frontiers for scientifi c investigation and technological development in this fi eld.

The invasive interface
Among the fi rst harsh lessons of invasive ventilation was that the protracted presence of an endotracheal tube not only increased the resistance through the upper airway, but also provided a pathway for infection and often damaged irreversibly the delicate tissues of the larynx and trachea. Even today, the problem of airway debris is diffi cult to contend with, as the biofi lm that lines the unperfused endotracheal tube combined with interruption of the mucociliary escalator and a disrupted coughing mechanism predisposes to retention of contaminated airway secretions [27]. Accumulation of airway debris causes increased work of breathing, impaires gas exchange, and pre dis poses to bronchopulmonary infec tions. Better materials, lower cuff pressures, and improved nursing practices have addressed some of these problems, but clearly not all of them. In-hospital use of noninvasive ventilation was born from the need to address such issues, and with continually improving interfaces now allows for intubation avoidance, improved sleep quality, and safer treatment of patients with diverse cardio pulmonary problems of moderate severity [28].

Patient-ventilator interactions
Also learned relatively early in the experience with positive-pressure ventilation was the fact that controlling fl ow rather than pressure could result in high eff ort and could predispose to breath timing dys-synchrony [29]. Furthermore, insistence on targeting near-normal pH and partial pressure of carbon dioxide in patients with airfl ow ob struc tion often produces dynamic hyper infl a tion and auto-PEEP [15]. Th is pervasive gas-trapping pheno menon, which is non homogeneously distributed, impairs breath triggering, increases work of breathing, and may impede venous return. In patients with expira tory fl ow limitation, counter balancing auto-PEEP with added PEEP can improve the sensitivity of breath trigger ing, improve the homogeneity of ventilation, and reduce dyspnea without further lung distention, hemodynamic compromise, or disadvantage to the muscles of the res pira tory system [30,31].

Ventilator-induced lung injury
High airway pressures and tidal volumes have been shown to damage both healthy and diseased lungs of laboratory animals since the 1970s. Investigations into the causative relationships among mechanical forces, machine settings and cofactors continues to the present day. It is generally understood, however, that the repeti tive application of transalveolar pressures and tidal swings of pressure (driving pressure) that substantially exceed those normally encountered during normal tidal breath ing will give rise to hemorrhagic edema and infl ammation that mimic ARDS [17]. Sustained re-open ing of collap sible lung units that are points of stress focusing is, in general, desirable. But debate continues as to the feasibility and relative importance of fully recruiting all collapsed units, as the latter requires that alveolar pressures do not fall below a high threshold that initiates closure of refractory-dependent units [32]. Because re cruit ing unstable alveoli can dramatically reduce the incidence of ventilator-induced lung injury, a persuasive rationale exists for recruiting maneuvers, prone position ing, and the early use of high-level PEEP -the latter obligating use of relatively small driving pressures and accepting resultant hypercapnia when necessary.
We have learned only slowly to account for the important infl uence of the chest wall on measured values of pressure at the airway opening. Th e lung may thus be exposed to lower or higher transalveolar pressures than suggested by the unmodifi ed plateau pressure or PEEP. Even when considering alveoli in diff erent sectors, stresses and strains upon tissues almost undoubtedly vary greatly, in part because of variations in the environment surrounding those lung regions.

Complexity and clinical trials
Few rules governing mechanical ventilation apply across all phases and severities of acute illness; choices must be conditioned by stage and by patient response. Many of the tested questions have sought 'yes or no, toggle switch' answers ( Figure 1). Yet even those interventions that seem amenable to such dichotomous testing are nuanced by considerations of their dose, duration, timing of use, and patient selection. Complexity of co-morbidities, timing stages, and co-interventions requires the clinician to weigh and integrate all important factors before making a decision, and then to employ short-loop feed back with frequent mid-course corrections [33] (Figure 2). Knowing these key principles of eff ective clinical practice, it is wise to remember that few clinical trials have been undertaken with detailed or proven knowledge of the underlying mechanism driving the outcome variable or have accounted for the complexity and timing of pathophysiology and management. As a simple example, none of the multicenter cooperative trials of mechanical ventilation yet conducted has assured passivity of the study cohort, despite the implications of muscular eff ort for the transalveolar pressures that lie at the root of ventilator-induced lung injury.
Without question, we have learned substantially from the conduct of clinical trials. But, as with physiologic principles gathered from laboratory models of disease, results from population-based clinical trials are only a starting point to guide thinking in many matters related to mechanical ventilation of the critically ill. In many instances, I believe we are well served by unproven experience-based rules (Table 1). Examples abound; high levels of PEEP are relatively helpful in the early stage of ARDS management when the lung is relatively wet and recruitable and benefi t outweighs hazard [34,35]. During this initial period of support, recruiting maneuvers (in themselves may be only transiently eff ective) are often required to set optimal PEEP, which is best selected using defl ation characteristics and functional gas exchange responses. Later in the patient's course (or when the lung is poorly recruitable for other reasons), PEEP simply adds to the peak and average airway pressures, both accentuating stresses and strains asso ciated with a given tidal volume and creating deadspace. Advisability of prone positioning may also be time and severity dependent. Meta-analysis of clinical trials data indicates that prone positioning seems to reduce mortality only in those patients who are both severely aff ected and in the early stage of illness [36]. Persuasive evidence suggests that we are learning similar lessons regarding timing and empiricism when using glucocorticoids [37,38] and recruiting maneuvers [39] in the management of such patients.

Conditional benefi ts of spontaneous eff orts
Another important lesson learned is that there is a need to strike a balance between the benefi ts of spontaneous breathing and the dangers of oversedation and neuromuscular paralysis. Ventilator-induced diaphragmatic dys function should clearly be of concern when fully controlled ventilation is imposed for extended periods [40,41]. Furthermore, unlabored spontaneous patterns of breathing (not accompanied by dyspnea or expiratory muscular eff ort), appear to be more mechanically efficient than are those administered to a passive patient [42,43]. Yet taking control of ventilation during the earliest phase of life-threatening sepsis and ARDS may enable reductions in potentially damaging mechanical forces arising from high cardiac output and minute ventilation [44,45]. Brief use of paralytics during the most vulnerable early period of illness is not necessarily associated with delayed neuromuscular recovery or ventilator-associated diaphragmatic dysfunction. Th at  being said, it is now strongly suspected that sustained suppression of awareness by large uninterrupted doses of sedatives without periodically returning the patient to consciousness extends the likelihood of prolonged mecha nical ventilation, delirium, inability to wean, and consequent adverse clinical outcomes [46].

Unproven rules of ventilator management
Self-evident rules regarding mechanical ventilation have emerged from decades of our collective experience at the bedside. But as yet these rules remain unproven by rigorous clinical trials -and some may never be proven. Ventilatory manage ment of the acute phase of ARDS provides several good examples of our unproven folk wisdom. A major step forward in the prevention of lung damage was to relate tidal volume to predicted (lean) as opposed to measured body weight [47]. Using predicted weight helps scale tidal volume to the underlying anatomical dimen sion of the lung. Yet the simple rule of 6 ml/kg predicted body weight that has gained traction for protocols used in daily practice is not suffi cient for all situations relating to body stature and ventilation demand. Th e guideline may need to be adjusted upward when patients are small and ventilation demands are high ( Figure 3). On the fl ip side, 6 ml/kg is not always a safe device. Because tidal volume enters only the aerated compartment, it may (depending on compartmental capacity) generate an inadvisably high specifi c tidal volume and consequently excessive trans alveolar pressures and strain during passive infl ation. Any inspira tory muscle activity adds further to actual mechanical stress on delicate tissue.
We should also modify therapy according to the patient's physiological need. For example, employing a guideline-approved small tidal volume without reducing a high minute ventilation demand may incur dyspnea as well as inappropriate high breathing frequencies. Whenever possible, we should attempt to reduce the ventilation intensity as well as the patient's demand for support. Reducing agitation, pain, body temperature, and meta bolic acidosis are often addressable. Sedation may also be required to tolerate permissive hypercapnia. Refocusing on the pressure diff erence across the lung is important, as the peak and driving transpulmonary (transalveolar) pressures are those that count with respect to the causation of iatrogenic lung damage [48]. In theory, know ing the func tional residual capacity and the trans alveolar (as opposed to plateau) static pressure would be necessary to interpret the safety of our tidal volume selection.
Th oughtful clinicians seek ways other than modifying the tidal volume and PEEP to ventilate protectively. From the viewpoint of clinical trial evidence, most methods remain unproven. One aspect of management that may have received insuffi cient attention in ARDS management is the need to reduce the eff ects of high fl ow on tidal shearing forces. Because the baby lung has a reduced number of open airways, fl ows that would be tolerable in a larger, high-capacity, fully open lung can theoretically result in un acceptable rates of tissue opening. For example, venti la tion of 10 l/minute equates to 40 l/ minute and an inspiratory average fl ow velocity of at least double that value in the typical patient whose actual functional residual capacity is reduced to one-quarter of normal. Whereas the open conducting channels may not be directly injured, units that open quickly during infl ation may be more vulnerable to epithelial shearing. Moreover, the popularity of pressure control ventilation promotes very high peak in spi ratory fl ows that occur just at the time during which unstable units have yet to be opened. Some experimental evidence in small and large animals strongly implicates high peak fl ow and delivery profi les as key to generating or avoiding ventilatorinduced lung injury [49][50][51].
Although stretching, shearing, and small airway trauma have been demonstrated to occur when transpulmonary pressures are excessive, tissue tension cannot be directly measured. Unfortunately our reliance on airway pressures alone (PEEP and plateau pressures) -which merge infor mation from all aircontaining sectors, are distorted by chest wall stiff ness, and are infl uenced by the presence or absence of spontaneous breathing eff orts -glosses over such realities (Figure 4). Experienced clinicians are aware that airway pressures alone may be misleading when the chest wall is stiff ened by obesity, surgery, trauma, or disease as well as when the patient makes forceful inspiratory and expira tory eff orts. Even measuring trans pulmonary pressure with the aid of an esophageal balloon catheter may not be enough [52][53][54]. A challenging aspect of managing the stresses and strains developed within a mechanically heterogeneous lung is the amplifi cation (or stress focusing) that occurs at the interfaces between fully open and closed lung units [55]. Marini Critical Care 2013, 17(Suppl 1):S1 http://ccforum.com/content/17/S1/S1

Well-intentioned but dysfunctional practices
It is humbling to consider that practices which have gone many years unquestioned might contribute to the genera tion or extension of the primary disease we are trying to resolve. Acute illness progresses through phases. In general, we have not taken into account that the under lying pathophysiology varies with disease stage, and that such physiological diff erences should factor strongly into our management. Here is one possible example: in the early stage of pneumonia treatment, the intubated patient is typically hydrated, given antibiotics, and repositioned frequently to avoid decubitus ulceration of the skin and to improve comfort.
We often encourage such patients to breathe spontaneously, with each forceful call for and assisted breath resulting in the delivery of relatively high transpulmonary pressure and tidal volume. PEEP is not considered helpful in lobar disease unless maintenance of adequate oxygenation requires it. With the patient's ability to expel secretions impaired by intubation, we suction the airway frequently and promote coughing in the process. Yet we may need to rethink our approach in this earliest stage of pneumonia [56]. Th in proteinaceous and mediator-laden fl uids with great potential for spreading through the airway network characterize this earliest phase [57,58]. During these early post-intubation hours these mobile biofl uids can extend injury, cause metastatic lobar infection [59], and even propagate a process that culminates in diff use lung injury that we label primary ARDS ( Figure 5). Before propagation happens and focal lung disease becomes generalized, implementing moderate PEEP to peripheralize liquid, small tidal volumes, inhibited coughing, enforced quiet breathing, and dependent positioning of the aff ected side may be the most rational strategy to contain the pneumonia to its region of origin [55] ( Figure 6). Later, the well-intentioned suctioning, movement, and lower PEEP with higher tidal volume are perfectly rational in helping to expel the thickened and less dangerous biofl uids known as sputum. One must emphasize that this 'propagation prevention with injury avoidance' hypothesis is mechanisti cally plausible but unproven. A clinical trial to determine its validity would be informative.

Emerging technologies
Important challenges remain in current practice ( Table 2). Although we have learned important lessons much too slowly regarding the dangers of protracted endotracheal intubation, ventilator-induced lung injury, sedation issues, and breathing dys-synchrony, our cumulative experience has given rise to advances with potential for better care of the ventilated patient. Greatly improved noninvasive ventilation may obviate the need for more invasive approaches in many patients. For the foreseeable future, however, intubation will continue to be required to protect the airway, to extract retained secretions, to allow deep sedation, and to control ventilation for purposes of manipulating the airway or performing cardiothoracic surgery. Minute ventilation infl uences on breathing frequency for two patients of diff ering body size (50 kg and 85 kg). Using the 6 ml/kg predicted body weight guideline, a small patient would be obligated to breathe at an unacceptably high frequency as minute ventilation increases (15 and 20 l/minute). Observing the commonly used upper limit of 30 breaths/minute, tidal volumes far in excess of the 6 ml/kg criterion would be required to ventilate the smaller patient in this higher range (500 and 670 ml, as opposed to 300 ml). br, breaths; pbw, predicted body weight. Secretion retention will therefore probably remain a vexing source of complications so long as invasive intubation is required. Th e unperfused biofi lm that lines the tube is inaccessible to host defenses, providing a safe haven for large infective inoculums to form and later seed the lung. Nonetheless, approaches that minimize or remove the infective endotracheal biofi lm, visualize the proximal airways, reduce secretion impaction, and assist with sputum elimination by attention to inspiratory fl ow modifi cation, percussive vibration of the air column, and mechanically aided coughing promise to minimize secretion-related complications [60][61][62][63][64].
Genuine progress has also been made in the attempt to link appropriate patient demands for ventilatory assistance with synchronous triggering and power. Initial bene fi ts from pressure support and pressure control have paved the way for recently released innovations such as proportional assist ventilation and neurally adjusted venti latory assist [65,66]. With better monitoring of mechanics and gas exchange, automated goal-directed algorithms integrated into the machine circuitry may enable automated upregulation and downregulation of power assistance, fraction of inspired oxygen, and PEEP, according to demands and patient response. Th ese algorithms have only recently gained traction in the clinical setting but clearly are steps in the right direction.
Concerns regarding ventilator-induced lung injury continue, of course, but deployment of the laboratoryproven and venerated esophageal balloon monitoring of pleural pressure may now enable routine determination of transpulmonary pressure -a value that comes a step closer to the calculation of eff ective stress upon the lung itself during spontaneous breathing and that helps select the PEEP that must be applied to keep it positive so as to avoid collapse [54] (Figure 7). Direct measurement of functional residual capacity allows estimation of the size of the baby lung, which does not always coincide with estimates based on trans pul mo nary pressure [67].
Regarding the force amplifi cation at points of stress focusing, there is still a considerable gap that needs closure. Here too, however, tools needed for regional and dynamic monitoring of the heterogeneous lung are becom ing available in the form of bedside regional imaging methodologies such as electrical impedance tomography and ultrasonic probing of the diseased lung [68,69]. Th ese methods currently off er impressive qualitative insights, even if they lack quantitative precision at this time.
Reducing the need to ventilate and to generate high pressures for ventilation, lung recruitment, and oxygenation with the patient remaining fully conscious and with spontaneously breathing has been a clear but elusive goal that is now much closer to widespread implementation. Prudently administered pharmaceuticals and judicious use of renal replacement therapies applied in a timely    [70,71]. Such methodologies were urgently and successfully applied in the treatment of severely aff ected patients with H1N1 lung injury [72].

A few predictions
As we progress through this early part of the 21st century, emerging economic realities will help drive our approach to bedside care (Table 3). We will probably have fewer personnel deployed per patient for both observation and intervention. Caregivers will be aided by electronic information handling, but it is unclear at this time how well prepared the individual caregiver will be to think analytically when managing the required information stream and knowledge base. Hospital administrations are likely to demand faster hospital throughput while emphasizing the priorities of safety, timely intervention, and avoidance of complications. Aggressive attempts will be made to protocolize many aspects of care. Such needs may spawn a variety of future innovations in mechanical ventilation (Table 4). Smarter machines will reduce the need for user input and monitoring. Flexible equip ment will be needed to address patients of all sizes and conditions and to apply multi-element protocols automatically while carefully monitoring the patient for unanticipated deviations and complications. To make such automation safely possible, advanced ventilators will not only monitor pressures and fl ows, but also exhaled gas analysis and inputs from the hemodynamic side. I anticipate that machines of the future will be goaldirected and self-adapting, fully capable of integrating mechanics, gas exchange, and cardiovascular information to achieve the clinical targets. Remote reporting and machine adjust ment are a clear and natural evolution. Past lessons and future needs will shift the ventilatory paradigm (Table 5).

Conclusion
Unchanging needs for providing eff ective life-support with minimized risk and optimized comfort have been, are now, and will remain the principal objectives of Regional pressure recorded within the esophagus (P es ) and along the sagittal and coronal planes that intersect it may be representative of pressures relevant to the stress-focused and relatively unstable units at the aerated and airless interface. VILI, ventilator-induced lung injury.   Marini Critical Care 2013, 17(Suppl 1):S1 http://ccforum.com/content/17/S1/S1 mechanical ventilation. Important lessons acquired during almost half a century of ICU care have brought us closer to meeting these elusive goals. Perhaps the overarching theme of our education, however, is that a solid understanding of organ system physiology is the fundamental and irreplaceable tool for guiding our progress.