7
Francesco Mojoli Giorgio A. Iotti Ilaria Curro ` Marco Pozzi Gabriele Via Aaron Venti Antonio Braschi An optimized set-up for helmet noninvasive ventilation improves pressure support delivery and patient–ventilator interaction Received: 20 April 2012 Accepted: 8 August 2012 Ó Copyright jointly held by Springer and ESICM 2012 F. Mojoli ( ) ) Á G. Via Á A. Braschi Dipartimento di Emergenza Urgenza, S.C. di Anestesia e Rianimazione I, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy e-mail: [email protected] F. Mojoli Á I. Curro ` Á M. Pozzi Á A. Venti Á A. Braschi Dipartimento di Scienze Clinico-chirurgiche, Diagnostiche e Pediatriche, Sezione di Anestesia Rianimazione e Terapia Antalgica, Universita ` degli Studi di Pavia, Pavia, Italy G. A. Iotti Dipartimento di Emergenza Urgenza, S.C. di Anestesia e Rianimazione II, Fondazione IRCCS Policlinico San Matteo, Pavia, Italy Abstract Objective: To test the effects on mechanical performance of helmet noninvasive ventilation (NIV) of an optimized set-up concerning the ventilator settings, the ventilator circuit and the helmet itself. Subjects and methods: In a bench study, helmet NIV was applied to a physical model. Pressurization and depressurization rates and minute ventilation (MV) were measured under 24 conditions including pressure support of 10 or 20 cmH 2 O, positive end expiratory pressure (PEEP) of 5 or 10 cmH 2 O, ventilatorcircuit with ‘‘high’’, ‘‘inter- mediate’’ or ‘‘low’’ resistance, and cushion deflated or inflated. In a clinical study pressurization and depressurization rates, MV and patient–ventilator interactions were compared in six patients with acute respiratory failure during conven- tional versus an ‘‘optimized’’ set-up (PEEP increased to 10 cmH 2 O, low resistance circuit and cushion infla- ted). Results: In the bench study, all adjustments simultaneously applied (increased PEEP, inflated cushion and low resistance circuit) increased pressurization rate (46.7 ± 2.8 vs. 28.3 ± 0.6 %, p \ 0.05), depressur- ization rate (82.9 ± 1.9 vs. 59.8 ± 1.1 %, p B 0.05) and patient MV (8.5 ± 3.2 vs. 7.4 ± 2.8 l/min, p \ 0.05), and decreased leaks (17.4 ± 6.0 vs. 33.6 ± 6.0 %, p \ 0.05) compared to the basal set- up. In the clinical study, the optimized set-up increased pressurization rate (51.0 ± 3.5 vs. 30.8 ± 6.9 %, p \ 0.002), depressurization rate (48.2 ± 3.3 vs. 34.2 ± 4.6 %, p \ 0.0001) and total MV (27.7 ± 7.0 vs. 24.6 ± 6.9 l/min, p \ 0.02), and decreased ineffective efforts (3.5 ± 5.4 vs. 20.3 ± 12.4 %, p \ 0.0001) and inspiratory delay (243 ± 109 vs. 461 ± 181 ms, p \ 0.005). Conclusions: An opti- mized set-up for helmet NIV that limits device compliance and ventila- tor circuit resistance as much as possible is highly effective in improving pressure support delivery and patient–ventilator interaction. Keywords Helmet noninvasive ventilation Á Mechanical ventilation Á Patient–ventilator interaction Introduction The helmet is an interface originally conceived for applying continuous positive airway pressure [1, 2]. Although more than 10 years have passed since the helmet was also proposed for delivering noninvasive ventilation (NIV) [36], this latter application is still discussed and not widely accepted [7, 8]. The very high tolerability of the helmet should suggest that it is the best NIV interface when prolonged and continuous assistance is needed in patients with acute respiratory failure [36, 9]. However, bench tests and studies in healthy Intensive Care Med DOI 10.1007/s00134-012-2686-x ORIGINAL

An optimized set-up for helmet noninvasive ventilation improves pressure support delivery and patient–ventilator interaction

Embed Size (px)

Citation preview

Francesco MojoliGiorgio A. IottiIlaria CurroMarco PozziGabriele ViaAaron VentiAntonio Braschi

An optimized set-up for helmet noninvasiveventilation improves pressure support deliveryand patient–ventilator interaction

Received: 20 April 2012Accepted: 8 August 2012

! Copyright jointly held by Springer andESICM 2012

F. Mojoli ()) ! G. Via ! A. BraschiDipartimento di Emergenza Urgenza,S.C. di Anestesia e Rianimazione I,Fondazione IRCCS Policlinico San Matteo,Pavia, Italye-mail: [email protected]

F. Mojoli ! I. Curro ! M. Pozzi ! A. Venti !A. BraschiDipartimento di Scienze Clinico-chirurgiche,Diagnostiche e Pediatriche,Sezione di Anestesia Rianimazionee Terapia Antalgica, Universita degliStudi di Pavia, Pavia, Italy

G. A. IottiDipartimento di Emergenza Urgenza,S.C. di Anestesia e Rianimazione II,Fondazione IRCCS Policlinico San Matteo,Pavia, Italy

Abstract Objective: To test theeffects on mechanical performance ofhelmet noninvasive ventilation (NIV)of an optimized set-up concerning theventilator settings, the ventilator circuitand the helmet itself. Subjects andmethods: In a bench study, helmetNIV was applied to a physical model.Pressurization and depressurizationrates and minute ventilation (MV)were measured under 24 conditionsincluding pressure support of 10 or20 cmH2O, positive end expiratorypressure (PEEP) of 5 or 10 cmH2O,ventilator circuit with ‘‘high’’, ‘‘inter-mediate’’ or ‘‘low’’ resistance, andcushion deflated or inflated. In aclinical study pressurization anddepressurization rates, MV andpatient–ventilator interactions werecompared in six patients with acuterespiratory failure during conven-tional versus an ‘‘optimized’’ set-up(PEEP increased to 10 cmH2O, lowresistance circuit and cushion infla-ted). Results: In the bench study, alladjustments simultaneously applied(increased PEEP, inflated cushion andlow resistance circuit) increasedpressurization rate (46.7 ± 2.8 vs.

28.3 ± 0.6 %, p \ 0.05), depressur-ization rate (82.9 ± 1.9 vs.59.8 ± 1.1 %, p B 0.05) and patientMV (8.5 ± 3.2 vs. 7.4 ± 2.8 l/min,p \ 0.05), and decreased leaks(17.4 ± 6.0 vs. 33.6 ± 6.0 %,p \ 0.05) compared to the basal set-up. In the clinical study, the optimizedset-up increased pressurization rate(51.0 ± 3.5 vs. 30.8 ± 6.9 %,p \ 0.002), depressurization rate(48.2 ± 3.3 vs. 34.2 ± 4.6 %,p \ 0.0001) and total MV(27.7 ± 7.0 vs. 24.6 ± 6.9 l/min,p \ 0.02), and decreased ineffectiveefforts (3.5 ± 5.4 vs. 20.3 ± 12.4 %,p \ 0.0001) and inspiratory delay(243 ± 109 vs. 461 ± 181 ms,p \ 0.005). Conclusions: An opti-mized set-up for helmet NIV thatlimits device compliance and ventila-tor circuit resistance as much aspossible is highly effective inimproving pressure support deliveryand patient–ventilator interaction.

Keywords Helmet noninvasiveventilation ! Mechanical ventilation !Patient–ventilator interaction

Introduction

The helmet is an interface originally conceived forapplying continuous positive airway pressure [1, 2].Although more than 10 years have passed since thehelmet was also proposed for delivering noninvasive

ventilation (NIV) [3–6], this latter application is stilldiscussed and not widely accepted [7, 8]. The very hightolerability of the helmet should suggest that it is the bestNIV interface when prolonged and continuous assistanceis needed in patients with acute respiratory failure [3–6,9]. However, bench tests and studies in healthy

Intensive Care MedDOI 10.1007/s00134-012-2686-x ORIGINAL

volunteers and patients have clearly shown thatmechanically helmets are less effective than face masksin delivering NIV [4, 7, 10, 11]. The main reason for thisis the high compliance of the helmet, that behaves like adamper interposed between the patient and the ventilator,thus weakening and delaying the inspiratory pressuresupport.

To increase the effectiveness of helmet NIV, theselection of higher PEEP and pressure support than nor-mally applied during mask NIV has been suggested,together with the fastest pressurization rate provided bythe ventilator [12]. However, slow inspiratory pressuri-zation is not the only problem of helmet NIV, andpossibly not the most important one. After ventilatorcycling-off to exhalation, pressure inside the helmetexhibits a typical slow decrease towards the PEEP level[12, 13], thus braking the patient’s expiratory flow. Fur-thermore, the helmet may promote dyssynchrony bydampening signals used by the triggering system of theventilator [7, 11–13].

Therefore, besides the choice of ‘‘pushing more’’ withthe ventilator, we considered it worthwhile to investigatehow to make the helmet a more responsive interface inorder to optimize not only inspiration but also exhalation,and to improve patient–ventilator synchronization. Wetherefore performed a series of bench tests to re-evaluatethe effects of increased PEEP and pressure support, and toevaluate the effects of two more adjustments aiming toimprove mechanical performance of helmets: inflation ofan internal cushion and major reduction of ventilatorcircuit resistance. Adjustments showing favourableeffects in the in vitro experience were combined into anoptimized set-up for helmet NIV that was introduced intoour clinical practice and evaluated in patients with acuterespiratory failure.

Methods

Bench study

Experimental set-up

We applied a NIV helmet (CaStar R, Starmed, Mirandola,Italy) to a physical model simulating a passive patient(Fig. 1). The model was made up of an expanded poly-styrene head connected to a plate simulating theshoulders. The head was provided with a proximal airwaythat was connected below the plate to a passivemechanical lung simulator composed of a main airway, aY-piece and two elastic balloons (flow resistance8 cmH2O/l/s, aggregate compliance 70 ml/cmH2O). Thehelmet, provided with an inflatable neck cushion, wasplaced around the head and secured to the plate by straps.When the helmet was pressurized and ventilated, its softcollar adhered to the neck and plate, thus providing a

reasonably sealed connection with some leaks allowed, ashappens in the clinical setting.

The simulator was ventilated with an Engstrom Car-estation ventilator (GE Healthcare, Madison, WI)connected with two ports to the helmet. We used threedifferent ventilator circuits, with high, intermediate andlow resistance. The high resistance circuit (flow resistanceof each limb 4.2 cmH2O at 1 l/s) was a standard circuitfor invasive ventilation (22-mm inner diameter and160-cm length; Mallinckrodt Dar, Mirandola, Italy)connected to the helmet ports by interposing two anti-microbial filters (Mallinckrodt Dar). The same circuit,directly connected to the helmet without filters, was usedas the intermediate resistance circuit (2.4 cmH2O at 1 l/s).The low resistance circuit (1.6 cmH2O at 1 l/s) was madewith shorter tubes (22-mm inner diameter and 65-cmlength; Teleflex Medical, High Wycombe, UK) directlyconnected to the helmet.

Experimental protocol

Pressure-controlled ventilation was applied to the physi-cal model. Respiratory rate was 20 bpm, the inspiratory to

Fig. 1 Experimental set-up of the bench study. Physical modelmade up of an expanded polystyrene head supported by a platesimulating the shoulders. a helmet cushion, b noise filter, c venti-lator inspiratory port (pressure measurement point), d ventilatorexpiratory port (flow volume measurement point), e helmet inletport (pressure measurement point), f carina of the main airway ofthe simulator (flow volume measurement point)

expiratory time ratio was 1:2 and the rise time was 0 ms(maximum pressurization rate) throughout the study. Wetested the effect of two levels of PEEP (5 or 10 cmH2O),two levels of pressure support over PEEP (10 or20 cmH2O), two states of the internal cushion (deflated orinflated at a pressure of approximately 120–150 cmH2O),three different ventilator circuits (with high, intermediateor low resistance) giving overall 24 different conditionstested.

Measurements

The Engstrom ventilator is provided with standardinternal sensors for pressure and flow located close tothe inspiratory and expiratory ports, as well as with anexternal flow sensor and an auxiliary pressure sensor formeasurements proximal to the patient (Fig. 1). Thestandard internal sensors were used to measure helmetventilation; total helmet minute ventilation (MV) wasmeasured at the inspiratory port of the ventilator, leakswere calculated by comparing the MV at the two portsof the ventilator and expressed as percentage of totalhelmet MV. The external flow sensor was connected tothe main airway of the simulator to measure patient MV.The auxiliary pressure sensor was connected to anauxiliary port to measure pressure inside the helmetwhich was analysed for pressurization and depressur-ization speed. Pressurization rate was calculated byintegration of pressure over time during the first 500 msfollowing the inspiratory valve opening (red area inFig. 2) and expressed as percentage of its ideal value(equal to the set inspiratory pressure times 0.5 s; i.e. thered rectangle in Fig. 2, on the left). Depressurization ratewas calculated similarly during the first 500 ms fol-lowing the expiratory valve opening (green area inFig. 2) and expressed as percentage of the ideal value(equal to the actual end-inspiratory pressure times 0.5 s;i.e. the green rectangle in Fig. 2, on the right). Dedicatedsoftware and a personal computer were used for con-tinuous recording of real-time signals. All measurementswere done during the last 60 s of a 5-min continuousrecording.

Statistics

To evaluate the effect of the different adjustments, the 24conditions were divided into groups according to PEEP (5vs. 10 cmH2O), pressure support (10 vs. 20 cmH2O),cushion (deflated vs. inflated), type of ventilator circuit(high vs. medium vs. low resistance). Groups werecompared with repeated measures ANOVA, the pairedt test and the Mann-Whitney test for normally and non-normally distributed data, respectively.

Clinical study

Experimental protocol

Following the favourable results of the bench study, wedecided to introduce an optimized set-up for helmet NIVin our ICU (Rianimazione 1, Fondazione IRCCS S.Matteo, Pavia, Italy). For safety reasons, the first patientssubmitted to the new protocol were extensively studied, inparticular by comparing in each of them the effect ofswitching from the previous, conventional set-up to thenew, optimized one. Basal FiO2, PEEP and pressuresupport were as prescribed by the on-duty physician. Theconventional set-up consisted of the highest pressuriza-tion rate, a standard ventilator circuit provided with filtersat the helmet ports, and deflated internal cushion. Theoptimized set-up differed from the conventional one forthe following adjustements: a low resistance circuit (asdefined in the bench study), inflation of the internalcushion, increase of PEEP up to 10 cmH2O if lower inconventional set-up, or otherwise unchanged; FiO2,pressure support and trigger sensitivity were unchanged.The study was approved by the ethics committee of ourinstitution.

Measurements

Patients were ventilated with an Engstrom Carestationventilator. The measurement set-up was the same as in the

Fig. 2 Pressurization and depressurization rates inside the helmet.Pressurization rate was calculated by integration of pressure overtime during the first 500 ms following the inspiratory valve opening(red area) and expressed as percentage of the ideal value (redrectangular area). Similarly, depressurization rate was calculatedduring the first 500 ms following the expiratory valve opening(green area) and expressed as percentage of the ideal value (greenrectangular area) (Phelmet pressure inside the helmet)

bench test, except for patient MV, that was not measured.Each condition was applied for at least 20 min, duringwhich data were continuously recorded. The results wereobtained as average values for the last 10 min of eachrecording. As a surrogate for oesophageal pressure, weanalysed flow and pressure tracings to detect autotrig-gered breaths and unassisted efforts and to calculateinspiratory delays [14, 15].

Statistics

To compare optimized helmet NIV set-up and the con-ventional one, we used paired t test and the Mann-Whitney test for normally and non-normally distributeddata, respectively.

Results

Bench test

Each of the adjustments intended to speed up theresponsiveness of the helmet (increased PEEP, inflatedcushion and decreased ventilator circuit resistance)resulted in moderate but significant increases in bothpressurization and depressurization rates (Table 1).Each of these adjustments also resulted in moderate butsignificant increases in patient MV on the lungsimulator.

Compared to the basal set-up, when all the adjust-ments were applied simultaneously, we observed majorincreases in pressurization rate (46.7 ± 2.8 vs. 28.3 ±0.6 %, p \ 0.05), depressurization rate (82.9 ± 1.9 vs.59.8 ± 1.1 %, p B 0.05) and patient MV (8.5 ± 3.2 vs.7.4 ± 2.8 l/min, p \ 0.05), as well as a major decrease inleaks (17.4 ± 6.0 vs. 33.6 ± 6.0 %, p \ 0.05). Figure 3shows favourable cumulative effects on instantaneouspressure inside the helmet associated with progressiveintroduction of the different adjustments tested: inflationof the cushion, then increase in PEEP, and finally venti-lator circuit switched to the lowest resistance.

Leaks were significantly reduced by higher pressuresinside the helmet; this effect was observed both withhigher pressure support and higher PEEP (Table 1).Inflation of the internal cushion resulted in a significantdecrease in leaks only when PEEP was low (24.9 ± 8.3vs. 35.2 ± 13.0 % for inflated vs. deflated cushion,p \ 0.01). The increase of pressure support from 10 to20 cmH2O over PEEP was associated with an increase inpeak pressure inside the helmet (26 ± 3 vs. 18 ± 2cmH2O, p \ 0.001), total helmet MV and actual patientMV, with slight worsening of both pressurization anddepressurization rates (Table 1).

Clinical study

We enrolled six consecutive patients (four men, age71 ± 7 years) who underwent helmet NIV for acuterespiratory failure. Five patients had chronic obstructivepulmonary disease. The causes of acute decompensationwere tracheopulmonary infection in three, gastroenteritisin one and major surgery in one. One patient wasrecovering from myocardial infarction. In four patientshelmet NIV was the sole form of respiratory support, andin two it was used for weaning from mechanical venti-lation after extubation. PEEP and pressure support were,respectively, 6.5 ± 1.2 cmH2O (range 5–8 cmH2O) and18.2 ± 2.6 cmH2O (range 13–20 cmH2O) in the con-ventional set-up, and 10.0 ± 0.0 cmH2O and 18.2 ±2.6 cmH2O in the optimized set-up.

Compared to the conventional one, the optimized set-up was associated with major improvements in pressuri-zation rate, depressurization rate and patient–ventilatorinteraction (Table 2; Fig. 4). Inspiratory delay wasreduced by nearly 50 % and unassisted efforts greatlydecreased, becoming very rare. On average, leaks (whichwere already low with the conventional set-up) wereunchanged. A single patient showed moderate leaks(13 %), that decreased to 3 % with the optimized set-up.

Discussion

Our results demonstrate that the mechanical performanceof the helmet interface for NIV can be greatly improvedby specific adjustments of the ventilator settings, theventilator circuit and the helmet itself. The bench studyconfirmed the known favourable effect of increasingPEEP [12] and showed for the first time the advantagesassociated with inflation of the helmet internal cushionand with major reduction in ventilator circuit resistance.Each of these adjustments had an independent favourableeffect on the speed of inspiratory pressure rise and expi-ratory pressure drop within the helmet, thus increasing theventilatory action on the respiratory system obtained by agiven pressure support level. The clinical study confirmedthat a combination of these adjustments greatly improvedthe speed of pressure rise and drop within the helmet, andinterestingly showed a major improvement in patient–ventilator interaction, with easier triggering and a majorreduction in the frequency of unassisted efforts.

When used as interface for NIV, the head helmet istolerated very well [3–6, 9], but its mechanical effec-tiveness is limited [4, 7, 10–13]. Intriguingly, the reasonsfor the good and the bad are the same: the soft collar andthe high internal volume. Also prolonged applications ofthe helmet are generally well tolerated because the softcollar provides a good seal while remaining atraumatic:no skin lesions occur and there are no draughts that irritate

the eyes of the patient. Furthermore, inside the helmet thepatient can freely drink, cough, speak, look around and evenwear his/her glasses. On the other hand, the soft collar and

the high volume of compressible gas make the helmet muchmore compliant mechanically than standard face masks.During helmet NIV, pressure is transmitted between themechanical ventilator and patient by a long time constantsystem that delays and decreases pressure support [10],brakes patient’s exhalation thus favouring hyperinflationand intrinsic PEEP, and finally interferes with the ability ofthe ventilator to detect spontaneous activity by the patient ina timely manner [11–13]. These mechanical limitations areadditional to the fact that the high internal volume favourssignificant CO2 rebreathing [4, 11] when the amount offresh gas flowing through the helmet is not well matchedwith the patient’s CO2 production [16–18].

Whether the effectiveness of the helmet can beimproved without changing the favourable characteristicsof its construction is a matter of debate. A simple option is

Table 1 Bench study: mechanical performance of helmet NIV in relation to different adjustments of ventilator, respiratory circuit and helmet

Performanceparameter

Pressure support (cmH2O) PEEP (cmH2O) Cushion inflation Circuit resistance

10 (n = 12) 20 (n = 12) 5 (n = 12) 10 (n = 12) Deflated(n = 12)

Inflated(n = 12)

High(n = 8)

Intermediate(n = 8)

Low(n = 8)

Pressurizationrate (%)

40.4 ± 7.8 37.0 ± 5.0# 33.8 ± 4.1 43.7 ± 4.6* 36.5 ± 7.2 40.9 ± 5.4* 35.5 ± 5.5 39.1 ± 7.3# 41.5 ± 6.5*"

Depressurizationrate (%)

72.4 ± 7.4 68.5 ± 6.9* 64.5 ± 3.4 76.4 ± 4.8* 69.1 ± 7.4 71.8 ± 7.2# 67.6 ± 7.0 69.6 ± 6.8# 74.0 ± 7.4*"

Patient MV(l/min)

6.0 ± 0.3 10.2 ± 0.5* 8.0 ± 2.2 8.3 ± 2.3* 8.1 ± 2.3 8.2 ± 2.3* 7.7 ± 2.1 8.3 ± 2.3* 8.4 ± 2.4*§

Total helmetMV (l/min)

14.9 ± 2.4 23.6 ± 1.5* 20.8 ± 4.4 17.8 ± 5.0* 20.0 ± 4.9 18.6 ± 4.9# 18.5 ± 4.9 19.2 ± 5.1# 20.2 ± 5.1*"

Leaks (%) 28.3 ± 12.8 16.5 ± 6.9* 30.1 ± 11.7 14.8 ± 4.8* 23.8 ± 15.1 21.0 ± 7.4 21.3 ± 9.8 22.0 ± 12.7 24.0 ± 13.8

* p \ 0.001, # p \ 0.01 in comparison with pressure support 10 cmH2O, or PEEP 5 cmH2O, or cushion deflated, or high resistance circuit; § p \ 0.001," p \ 0.01 in comparison with intermediate resistance circuit.

Fig. 3 Airway pressure inside the helmet (Phelmet) duringrespiratory cycles with pressure support of 10 and 20 cmH2O:cumulative effects of three different adjustments tested at thebench. Black line conventional set-up; green dotted line cushioninflated; blue dashed line cushion inflated ? increased PEEP; red

line cushion inflated ? increased PEEP ? low resistance circuit;grey dashed line ideal pressure trace. Pressurization rates (P %) anddepressurization rates (deP %) expressed as percentages of theideal value under the four different conditions are also shown

Table 2 Clinical study: pressure support delivery and patient–ventilator interaction with the conventional and optimized helmetNIV set-up

Conventionalset-up

Optimizedset-up

p

Pressurization rate (%) 30.8 ± 6.9 51.0 ± 3.5 \0.002Depressurization rate (%) 34.2 ± 4.6 48.2 ± 3.3 \0.0001Total helmet MV (l/min) 24.6 ± 6.9 27.7 ± 7.0 \0.02Leaks (%) 5.4 ± 4.1 2.6 ± 1.3 nsInspiratory delay (ms) 461 ± 181 243 ± 109 \0.005Unassisted efforts (%) 20.3 ± 12.4 3.5 ± 5.4 \0.0001Autotriggering (%) 4.1 ± 5.4 2.2 ± 2.7 ns

to push more with the mechanical ventilator, in this waycounteracting, although not treating, the dampening effectof the helmet. Vargas et al. [12] significantly improved theunloading of respiratory muscles during helmet NIV byincreasing both PEEP and pressure support by 50 % andby selecting the fastest pressurization rate available on theventilator. This ‘‘specific setting’’ made the degree ofventilatory assistance provided by the helmet similar tothat obtained with a face mask, but provided small bene-ficial effects on pressurization time and patient–ventilatorinteraction. Moreover, benefits obtained by increasingpressure support have some limits. Our bench studyshowed that an increase in pressure support is associatedwith a decrease in pressurization rate and with inspiratorypressures lower than expected (Table 1; Fig. 3), thusconfirming previous observations that the higher the setinspiratory pressure, the greater the gap between actualand ideal pressurization inside the helmet [11]. Moreover,the option of pushing with higher pressure can be appliedonly in patients with a low respiratory demand, as seen inthe study by Vargas et al. [12], in which baseline pressuresupport was quite low (10 cmH2O).

As an alternative for patients with a high demand, wetested a strategy aimed at improving the response time of thehelmet and limiting its dampening effect. With higherPEEP, inspiratory pressurization finds the helmet collaralready tensed and stiffened. Once inflated and well tensed,the cushion decreases the volume of compressible gas insidethe helmet. Both higher PEEP and an inflated cushionincrease the adhesion of the collar to the skin surface,eventually limiting leaks. Using a very short ventilator

circuit without filters reduces the overall time constant of thesystem as a result of a major decrease in resistance to flow.When tested separately in the bench study, each of theseadjustments was associated with slight but significantimprovements in both pressurization and depressurizationrates in the helmet. The combination of these effects wasassociated with a clear additive effect resulting in majorimprovement during both inspiration and exhalation(Fig. 3). The clinical application of our optimized set-up forhelmet NIV confirmed the in vitro results: rates of pres-surization and depressurization approached 50 % of idealvalues in the first 500 ms, with a relative improvement ofabout 50 %, thus approaching the mechanical effectivenessof a well-managed mask NIV [10, 13, 19]. These changeswere associated with major improvements in patient–ven-tilator interaction that can be explained first by bettertransmission to the ventilator of pressure and flow changesgenerated by the patient’s inspiratory muscles, and secondby faster expiratory depressurization of the helmet, allowingfaster expiratory flow thus limiting dynamic hyperinflationand the patient’s triggering workload [19, 20].

Of note, when a nonoptimal ventilator circuit is used,the airway pressure curve on the ventilator monitor can becompletely misleading: a square pressure wave, appar-ently ideal, can be associated with very slow pressuretransmission when resistance to flow between ventilatorand helmet is not negligible (Fig. 4). With a low resistanceventilator circuit, airway pressures measured inside theventilator reflect much better the course inside the helmet.

Leaks were moderate in our bench test, despite the factthat the model was suboptimal in mimicking a patient’sskin surface. Interestingly, both higher PEEP and pressuresupport led to decreases in leaks, probably because of abetter fit of the soft collar. Our clinical study confirmedthat, once the helmet was well pressurized, leakage wasminimal. This behaviour of helmets is completely dif-ferent from that of face masks, whose leakage typicallyincreases with pressure [21]. Our results confirm previousobservations that with helmets it is possible to selecthigher PEEP values than with masks, without major leaksor significant patient discomfort [3]. Because higherPEEP values also improve pressure support delivery, aPEEP of C8 cmH2O is not only possible but even usefuland advisable. The favourable effects of cushion inflationunderline the importance of limiting the helmet internalvolume as much as compatible with patient comfort; thisalso means that large helmets typically used for contin-uous positive airway pressure must definitely be avoidedfor NIV. In our optimized set-up, we removed filters,whose insertion in the circuit was originally suggestedto decrease noise perceived by subjects inside the hel-met [22]. A further reduction of circuit resistance byshortening both limbs provided additional improvementin helmet mechanics. Therefore, our advice is to use veryshort (even if sometimes troublesome) ventilator circuitsand to manage the increase in noise necessarily associated

Fig. 4 Airway pressure as displayed by the ventilator (Pvent,dashed line) and as simultaneously measured inside the helmet(Phelmet, continuous line) with the conventional set-up (A) andoptimized set-up (B) in a patient with chronic obstructivepulmonary disease and acute respiratory failure. With the conven-tional set-up (A), pressurization and depressurization inside thehelmet are very slow and patient–ventilator interaction is poor(arrows indicate unassisted inspiratory efforts); the slow waves ofthe Phelmet trace are poorly reflected by the square waves of thePvent trace. With the optimized set-up (B), pressurization, depres-surization and patient–ventilator interaction are improved, whilethe Pvent trace reflects the actual Phelmet trace much better

with the removal of filters with ear-plugs [22] and/orcircuits with a smooth internal surface.

This study had some limitations. We analysed a smallnumber of subjects, and almost all were patients withobstructive disease. We cannot assume the same results in adifferent population. Moreover, we did not assess patientcomfort. Finally, oesophageal pressure was not measured,and therefore we did not directly prove that faster andbetter-synchronized pressure support translated into greaterunloading of the patient’s inspiratory muscles. For instance,

in the presence of significant hyperinflation part of thepressure support delivered by the ventilator could substan-tially be wasted. On the other hand, our optimized set-upallowed easier triggering and this was consistent withdecreased or at least unchanged intrinsic PEEP.

In conclusion, an optimized set-up for helmet NIVaimed at limiting the compliance of the device and theresistance of the ventilator circuit as much as possible washighly effective in improving both pressure supportdelivery and patient–ventilator interaction.

References

1. Villa F, Cereda M, Colombo E, Berra L,Nacoti M, Foti G, Pesenti A (1999)Evaluation of four non invasive CPAPsystems. Intensive Care Med S 66:A246

2. Tonnelier JM, Prat G, Nowak E,Goetghebeur D, Renault A, Boles JM,L’her E (2003) Non-invasivecontinuous positive airway pressureventilation using a new helmetinterface: a case–control prospectivepilot study. Intensive Care Med29:2077–2080

3. Antonelli M, Conti G, Pelosi P,Gregoretti C, Pennisi MA, Costa R,Severgnini P, Chiaranda M, Proietti R(2002) New treatment of acutehypoxemic respiratory failure:noninvasive pressure support ventilationdelivered by helmet. A pilot controlledtrial. Crit Care Med 30:602–608

4. Antonelli M, Pennisi MA, Pelosi P,Gregoretti C, Squadrone V, Rocco M,Cecchini L, Chiumello D, Severgnini P,Proietti R, Navalesi P, Conti G (2004)Noninvasive positive pressure ventilationusing a helmet in patients with acuteexacerbation of chronic obstructivepulmonary disease. A feasibility study.Anesthesiology 100:16–24

5. Rocco M, Dell’Utri D, Morelli A,Spadetta G, Conti G, Antonelli M,Pietropaoli P (2004) Noninvasiveventilation by helmet or face mask inimmunocompromised patients. A case–control study. Chest 126:1508–1515

6. Conti G, Cavaliere F, Costa R, Craba A,Catarci S, Festa V, Proietti R, AntonelliM (2007) Noninvasive positive-pressure ventilation with differentinterfaces in patients with respiratoryfailure after abdominal surgery: amatched-control study. Respir Care52:1463–1471

7. Navalesi P, Costa R, Ceriana P, CarlucciA, Prinianakis G, Antonelli M, Conti G,Nava S (2007) Non-invasive ventilationin chronic obstructive pulmonary diseasepatients: helmet versus facial mask.Intensive Care Med 33:74–81

8. Crimi C, Noto A, Princi P, Esquinas A,Nava S (2010) A European survey ofnoninvasive ventilation practices. EurRespir J 36:362–369

9. Principi T, Pantanetti S, Catani F, EliseiD, Gabbanelli V, Pelaia P, Leoni P(2004) Noninvasive continuous positiveairway pressure delivered by helmet inhematological malignancy patients withhypoxemic acute respiratory failure.Intensive Care Med 30:147–150

10. Chiumello D, Pelosi P, Carlesso E,Severgnini P, Aspesi M, Gamberoni C,Antonelli M, Conti G, Chiaranda M,Gattinoni L (2003) Noninvasivepositive pressure ventilation deliveredby helmet vs. standard face mask.Intensive Care Med 29:1671–1679

11. Racca F, Appendini L, Gregoretti C,Stra E, Patessio A, Donner CF, RanieriVM (2005) Effectiveness of mask andhelmet interfaces to deliver noninvasiveventilation in a human model ofresistive breathing. J Appl Physiol99:1262–1271

12. Vargas F, Thille A, Lyazidi A, CampoFR, Brochard L (2009) Helmet withspecific settings versus facemask fornoninvasive ventilation. Crit Care Med37:1921–1928

13. Moerer O, Fischer S, Hartelt M, KuvakiB, Quintel M, Neumann P (2006)Influence of two different interfaces fornoninvasive ventilation compared toinvasive ventilation on the mechanicalproperties and performance of arespiratory system. A lung model study.Chest 129:1424–1431

14. Georgopoulos D, Prinianakis G, KondiliE (2006) Bedside waveformsinterpretation as a tool to identifypatient–ventilator asynchronies.Intensive Care Med 32:34–47

15. Mojoli F, Venti A, Pozzi M, Via G,Braschi A (2009) Patient-ventilatorinteraction during pressure supportventilation: how to monitor and toimprove it. Minerva Anestesiol75:533–536

16. Patroniti N, Foti G, Manfio A, CoppoA, Bellani G, Pesenti A (2003) Headhelmet versus face mask for non-invasive continuous positive airwaypressure: a physiological study.Intensive Care Med 29:1680–1687

17. Taccone P, Hess D, Caironi P, BigatelloLM (2004) Continuous positive airwaypressure delivered with a ‘‘helmet’’:effects on carbon dioxide rebreathing.Crit Care Med 32:2090–2096

18. Mojoli F, Iotti GA, Gerletti M, LucariniC, Braschi A (2008) Carbon dioxiderebreathing during non-invasiveventilation delivered by helmet: a benchstudy. Intensive Care Med34:1454–1460

19. Vignaux L, Tassaux D, Jolliet P (2007)Performance of noninvasive ventilationmodes on ICU ventilators duringpressure support: a bench model study.Intensive Care Med 33:1444–1451

20. Fabry B, Guttmann J, Eberhard L,Bauer T, Haberthiur C, Wolff G (1995)An analysis of desynchronizationbetween the spontaneously breathingpatient and ventilator during inspiratorypressure support. Chest 107:1387–1394

21. Schettino GP, Tucci MR, Sousa R,Barbas CS, Amato MB, Carvalho CR(2001) Mask mechanics and leakdynamics during noninvasive pressuresupport ventilation: a bench study.Intensive Care Med 27:1887–1891

22. Cavaliere F, Conti G, Costa R, ProiettiR, Sciuto A, Masieri S (2004) Noiseexposure during noninvasive ventilationwith a helmet, a nasal mask and facialmask. Intensive Care Med30:1755–1760