CORRESPONDENCE bench study,1 a clinical trial of the use of patients have been associated with a hyp- sol with small particle aerosol generator and the vibrating mesh micropump technology oxia related to a decrease in functional re- vibrating mesh micropump aerosol technol- for ribavirin delivery. Although the use of sidual capacity, and they potentially increase ogies. Respir Care 2016;61(5):577-585. aerosolized ribavirin remains controversial, the chances of ventilator-associated pneu- the patient selection by Hartmann et al ap- monia. The crystallization found also raises Airway Pressure Release pears to be on target with the latest support- the question of loss of drug delivery and Ventilation May Result in Occult ing evidence of potential benefit. Therapies where else it may have been building up Atelectrauma in Severe ARDS offered within respiratory care are without that was not seen. Could it be in the endo- conclusive evidence. In these cases, it is a tracheal tube? Not turning the humidifier balance of risk versus benefit. For the most off likely reduced the crystallization after To the Editor: part, many of the therapies we provide are humidifier use to that seen before humidi- In a recent issue of RESPIRATORY CARE, low risk, and therefore the smallest of ben- fier use. Third, the team felt the need to Mireles-Cabodevila and Kacmarek1 did an eficial therapies are studied in clinical trials. double filter the expiratory limb with good excellent job reviewing the pros and cons of Continuous aerosol delivery in the me- intentions of protecting the integrity of the airway pressure release ventilation (APRV). chanically ventilated patient with traditional expiratory valve, transducers, and clinicians. The paper and the discussion session raised jet nebulizers or in the case of ribavirin de- However, the rationale of using 2 identical serious concerns about ARPV, particularly livery via the small-particle aerosol gener- high-efficiency particulate air (HEPA) fil- regarding imposed work of breathing and ator (SPAG) can place the patient at addi- ters designed to filter to the same micron excessive tidal volumes due to high peak tional risk of complications for a host of particle level does not appear to be logical. transpulmonary pressures. However, neither reasons surrounding the addition of flow This likely only adds resistance to expira- in the paper (except for a brief reference by within the ventilator circuit. Respiratory tory flow and potentially exposes the pa- Dr Mireles-Cabodevila) nor in the discus- therapists for years have compensated for tient to an additional risk. Changing a single sion was there an emphasis on what we this known complication by being physi- HEPA filter following the 2-h aerosol de- believe is the most dangerous risk of using cally present during intermittent aerosol drug livery would have alleviated concerns of an APRV as a rescue treatment for severe delivery to provide a near continuous clin- incompetent filter. Last, although this case ARDS: occult atelectrauma. ical assessment and if necessary interven- had an outcome we all would hope for, it is Although the use of APRV has almost tion. This continuous assessment is not pos- not clear whether this was the result of the disappeared at our institution (Cleveland sible when these treatments are prolonged aerosolized ribavirin therapy or a combina- Clinic), a few years ago there was a lot of past the traditional 10–15 min timeframe. tion of efforts. interest in the mode. As a result, we got Since the introduction of the vibrating We are a profession known to deliver involved in some research on the topic, pro- mesh micropump technology, the capability low-volume, high-risk therapies well. Yet, ducing a few papers and abstracts. What to monitor pulmonary mechanics and ven- we must consider the risk of every interven- always intrigued us was the frequent justi- tilator operation during aerosol delivery is tion and look beyond our 4 walls to deter- fication by supporters for using APRV that possible, because the vibrating mesh mi- mine whether there is a better method on “it works,” meaning that oxygenation im- cropump does not add additional flow to the the horizon. I applaud the Hartmann group’s proved in patients with severe ARDS, yet circuit. This improves the safety profile of use of the Journal and for assessing the risk patients with severe ARDS rarely die of hy- this delivery method. The vibrating mesh and benefits in this extreme situation. They poxemia (ϳ10%); they die of multi-organ micropump may provide a safer delivery appropriately interpreted our study and log- failure. Organ failure is linked to the release with similar results to that of SPAG, but we ically chose the more frequently used and of inflammatory mediators from the lung in cannot ignore the many unknowns of this perceived safer vibrating mesh micropump response to mechanical trauma. Hence, fo- drug therapy, nor should we let the use of over the less frequently utilized and poten- cusing on oxygenation as the main goal of this proposed delivery method blindly open tially higher risk SPAG device. Their criti- APRV at the expense of lung-protective ven- the door for widespread application without cal analysis of the literature allowed them tilation does not seem like the most rational further study. Let’s review a couple of un- to choose the appropriate patient, drug, and approach. knowns mentioned in the letter. First, the device, leading to a safe delivery of the po- Oxygenation problems are usually man- team switched ventilators according to a tentially beneficial aerosolized ribavirin and aged with PEEP. Patients with severe ARDS manufacturer recommendation. This intro- overall wonderful care. should have end-expiratory lung volume ma- duces a variation in practice, circuit change, nipulated by some form of optimal PEEP and exposes the patient with ARDS to a Brian K Walsh PhD RRT-NPS heuristic. However, the most vehement sup- disconnect from the ventilator circuit. Was RRT-ACCS RPFT AE-C FAARC porters for APRV recommend setting zero Boston Children’s Hospital, ϭ this really necessary, because the team was PEEP (ie, Plow 0cmH2O but setting Tlow properly filtering the expiratory gases? Sec- Boston, Massachusetts short enough to maintain adequate end-ex- ond, the team quickly identified crystalliza- piratory lung volume by means of auto- tion after the 22 mL (2-h drug delivery), PEEP).2 However, reliance on auto-PEEP DOI: 10.4187/respcare.05162 which lead them to change their practice instead of set PEEP may result in unknown and rinse after each therapy session. This and unstable lung volumes and hence un- again led to a disconnection in an immuno- REFERENCE stable mechanical support of gas exchange compromised patient with severe lung dis- 1. Walsh BK, Betit P, Fink JB, Pereira LM, (not to mention the uneven distribution of ease. Disconnects in mechanically ventilated Arnold J. Characterization of ribavirin aero- auto-PEEP in the lungs according to the dis-

1278 RESPIRATORY CARE • SEPTEMBER 2016 VOL 61 NO 9 CORRESPONDENCE tribution of unequal time constants of lung units, with the sickest units getting the least PEEP). Furthermore, changes in lung mechanics (eg, the need for suctioning, variable edema, and variable inspiratory effort) make auto- PEEP an unpredictable random variable in an individual patient. We have programmed a high-fidelity physical breathing simulator with actual values for resistance and com- pliance from patients with severe ARDS ventilated with APRV and demonstrated that ϭ with a ventilator connected to the simulator Fig. 1. Predicted auto-PEEP during airway pressure release ventilation with Phigh 30 ϭ and the actual ventilator settings used (ac- cm H2O and Plow 0cmH2O. Expiratory time constant (expressed in s) is the product cording to published APRV recommenda- of resistance ϫ compliance. tions), there was no auto-PEEP at all (un- published data). No PEEP in a patient with severe ARDS certainly suggests an in- APRV is more complex than it appears to concern is that mortality is substantially creased risk of ventilator-induced lung in- be. It requires a lot more knowledge and greater (67% vs 49%) in those with mod- jury due to atelectrauma. skill than may be apparent from descrip- erate to severe ARDS who develop acute Auto-PEEP can be estimated with math- tions in the literature. cor pulmonale,14 How this relates to a dis- ematical models.3,4 Patients with severe In contrast to expected auto-PEEP levels cussion of whether APRV is a prudent ap- ARDS have a wide range of respiratory sys- during APRV, “optimum” PEEP levels rec- proach to lung-protective ventilation in tem expiratory time constants (average 690 ommended by the ARDSNet higher-PEEP ARDS is seen by consideration of the so- Ϯ 280 ms).5,6 Assuming7 an expiratory time table8 for patients requiring F Ն0.60 bering evidence that elevated plateau pres- IO2 constant of 0.5 s, setting Tlow in the range range from 20 to 24 cm H2O. Thus, reliance sure 27–35 cm H2O) is linked to develop- of 0.2–0.8 s2 results in auto-PEEP ranging on auto-PEEP during APRV probably re- ing acute cor pulmonale and that an from2to20cmH2O with Phigh set at sults in suboptimal end-expiratory lung vol- additive effect of elevated plateau pressure 30 cm H2O (relative to atmospheric pres- ume at least some of the time. Add to this and cor pulmonale increases mortality ϭ 14,15 sure) and Plow 0 in a passive patient with the problem that some ventilators synch the risk. The deleterious effects of sus- ARDS (see Fig. 1). Lower Phigh values (ie, ending of Thigh with spontaneous expiration tained, high intrathoracic pressures produc- Ͻ 30 cm H2O) naturally result in lower during Phigh, and you get an unpredictable ing hemodynamic compromise probably 9,10 ranges of auto-PEEP. Auto-PEEP for this Tlow despite explicit settings. On the explain the unexpectedly higher mortality figure was estimated using the equation, au- other hand, triggering the transition from in those managed with high-frequency os- Ϫ ϭ Ϫ ϫ (Tlow/RC) 16,17 to-PEEP (Phigh Plow) e , Plow to Phigh using expiratory flow is help- cillatory ventilation. Others have also where RC is the expiratory time constant ful, but it does not avoid the interdepen- pointed out that the heterogeneous distribu- and e is the base of natural logarithms dence of auto-PEEP and tidal volume.4,11 tion of time constants in ARDS very likely (ϳ2.72). This equation describes a simple Here is the main point of our letter: The leads to heterogeneous distribution in re- exponential decay of pressure in response risk of atelectrauma (a distant and hidden gional lung volumes that not only may in- to a step change from Phigh to Plow (as- effect) is often discounted by clinicians in duce/exacerbate regional atelectrauma, as sumed to be zero in this case) over the pe- favor of the benefit of improved oxygen- we suggest, but also regional alveolar over- 14 riodofTlow. We have created a Microsoft ation (immediate and obvious effect). This distention. Excel-based simulator based on published is an example of the decision error known In summary, the use of ARPV (with ex- 3 ϭ equations that can be used to understand as present moment discounting (the ten- treme inverse I:E and Plow 0) as a rescue how ventilator settings during APRV affect dency for people to have a stronger prefer- strategy for ARDS results, at least theoret- lung volumes, flows, and pressures along ence for more immediate payoffs relative to ically, in (1) increased work of breathing, with estimated P and P for different later payoffs).12 Unfortunately, when out- (2) increased risk of volutrauma, (3) in- aCO2 aO2 values of passive lung mechanics (free to comes of ventilation with APRV are unsat- creased risk of atelectrauma, and (4) in- download at http://is.gd/G6hD0A), as isfactory, they are often rationalized on the creased risk of cor pulmonale, compared shown in Figure 2. It is very helpful in un- basis of poor prior probability of survival with other pressure control modes. To us, derstanding the interdependencies among due to ARDS (obvious) rather than the po- these theoretical risks outweigh the poten- P ,P ,T ,T , auto-PEEP, and tential for suboptimal ventilator manage- tial improvement in P . If there were no high low high low aO2 passive tidal volume (so-called “release ment (hidden). alternative, then the risks might be war- volume”). We would argue that the use of In addition, there has yet to be a thought- ranted. But as we have shown,4 you can such a simulator is the only practical way to ful discussion regarding the impact of obtain the same objectives as APRV (ie, the gain understanding of APRV, because APRV on right heart function. In ARDS, same values for mean airway pressure, equivalent experience with real patients acute cor pulmonale occurs in 22–25% of end-inspiratory lung volume, “release could take years and put a lot of people at patients and reaches 50% among those with volume,” and end-expiratory lung vol- risk. As we noted in a previous paper,4 severe presentations.13,14 Of even greater ume) using a known set PEEP value and

RESPIRATORY CARE • SEPTEMBER 2016 VOL 61 NO 9 1279 CORRESPONDENCE

tomatic selection of breathing pattern using adaptive support ventilation. Intensive Care Med 2008;34(1):75-81. 8. Chiumello D, Cressoni M, Carlesso E, Cas- pani ML, Marino A, Gallazzi E, et al. Bed- side selection of positive end-expiratory pressure in mild, moderate, and severe acute respiratory distress syndrome. Crit Care Med 2014;42(2):252-264. 9. Daoud E, Chatburn RL. Auto-PEEP during APRV varies with the ventilator model (ab- stract). Respir Care 2010;55(11):1516. 10. Haug K, Chatburn RL. Interactions among tidal volume, expiratory time, and total-PEEP in APRV. Respir Care 2014;59(10):OF12. 11. Chatburn RL, Babic S. Evaluation of Dra¨ger APRV with AutoRelease: a model study. Respir Care 2010;55(11):1593. 12. List of cognitive biases. http://en.wikipedia. org/wiki/List_of_cognitive_biases. Ac- cessed August 1, 2016. 13. Vieillard-Baron A, Schmitt JM, Augarde R, Fellahi JL, Prin S, Page B, et al. Acute cor pulmonale in acute respiratory distress Fig. 2. Microsoft Excel-based airway pressure release ventilation simulator. Courtesy Mandu syndrome submitted to protective ventila- Press. tion: incidence, clinical implications, and prognosis. Crit Care Med 2001;29(8):1551- 1555. 14. Boissier F, Katsahian S, Razazi K, Thille aTlow set long enough to avoid auto- Mr Chatburn has disclosed relationships with PEEP. This approach decouples the level IngMar Medical and DeVilbiss/Drive Medical. AW, Roche-Campo F, Leon R, et al. Prev- The other authors have disclosed no conflicts of alence and prognosis of cor pulmonale dur- of mechanical support from the level of interest. ing protective ventilation for acute respira- auto-PEEP, making clinical management tory distress syndrome. Intensive Care Med DOI: 10.4187/respcare.05099 easier and more predictable. 2013;39(10):1725-1733. Perhaps the last word should go to 15. Jardin F, Vieillard-Baron A. Is there a safe John Downs (the inventor of APRV) and REFERENCES plateau pressure in ARDS? The right heart colleagues who commented on the largest only knows. Intensive Care Med 2007; study of APRV to date.18 In a letter to the 1. Mireles-Cabodevila E, Kacmarek RM. 33(3):444-447. editor in the Journal of Trauma, they said Should airway pressure release ventilation 16. Malhotra A, Drazen JM. High-frequency be the primary mode in ARDS? Respir Care “Many clinicians use APRV as a rescue oscillatory ventilation on shaky ground. 2016;61(6):761-773. mode for the treatment of ARDS. No N Engl J Med 2013;368(9):863-865. 2. Habashi NM. Other approaches to open- 17. Repesse´ X, Charron C, Vieillard-Baron A. study supports…the use of APRV in that lung ventilation: airway pressure release Acute respiratory distress syndrome: the 19 way…” In their response to the letter, ventilation. Crit Care Med 2005;33(3 heart side of the moon. Curr Opin Crit Care the authors of the study stated that “We Suppl):S228-S240. 2016;22(1):38-44. do not believe that APRV should be used 3. Marini JJ, Crooke PS 3rd, Truwit JD. Deter- 18. Maxwell RA, Green JM, Waldrop J, Dart as a rescue mode either.” minants and limits of pressure-preset venti- BW, Smith PW, Brooks D, et al. A ran- lation: a mathematical model of pressure con- domized prospective trial of airway pres- Robert L Chatburn MHHS trol. J Appl Physiol 1989;67(3):1081-1092. sure release ventilation and low tidal vol- RRT-NPS FAARC 4. Sasidhar M, Chatburn RL. Tidal volume vari- ume ventilation in adult trauma patients with Respiratory Institute ability during airway pressure release venti- acute respiratory failure. J Trauma 2010; Cleveland Clinic and lation: case summary and theoretical analy- 69(3):501-511; discussion 511. Lerner College of Medicine of Case sis. Respir Care 2012;57(8):1325-1333. 19. Evans DC, Stawicki SP, Eiferman D, Reil- 5. Guttmann J, Eberhard L, Fabry B, Western Reserve University ley TE, Downs JB. Physiologically rele- Bertschmann W, Zeravik J, Adolph M, et Cleveland, Ohio vant application of airway pressure release al. Time constant/volume relationship of ventilation. J Trauma 2011;71(1):262-263; passive expiration in mechanically venti- Richard H Kallet MSc RRT FAARC author reply 263-264. lated ARDS patients. Eur Respir J 1995; Respiratory Care Services 8(1):114-120. San Francisco General Hospital 6. Kallet RH, Katz JA. Respiratory system We Agree!! San Francisco, California mechanics in acute respiratory distress syn- drome. Respir Care Clin N Am 2003;9(3): Madhu Sasidhar MD 297-319. In Reply Cleveland Clinic 7. Arnal JM, Wysocki M, Nafati C, Donati S, As we discussed in our article,1 the use Cleveland, Ohio Granier I, Corno G, Durand-Gasselin J. Au- of auto-PEEP to establish PEEP always

1280 RESPIRATORY CARE • SEPTEMBER 2016 VOL 61 NO 9