CICM Online CCR Journal logo CICM logo

Full Text View

Original Article

Pressure support ventilation in intensive care patients receiving prolonged invasive ventilation

Wisam Al-Bassam, Tapan Parikh, Ary Serpa Neto, Yamamah Idrees, Mark A Kubicki, Carol L Hodgson, Ashwin Subramaniam, Mallikarjuna Ponnapa Reddy, Navya Gullapalli, Claire Michel, Madeline Coxwell Matthewman, Jack Naughton, Jason Pereira, Yahya Shehabi, Rinaldo Bellomo

Crit Care Resusc 2021; 23 (4): 394-402

  • Author Details
  • Competing Interests
    None declared
  • Abstract
    BACKGROUND: To our knowledge, the use and management of pressure support ventilation (PSV) in patients receiving prolonged (≥ 7 days) invasive mechanical ventilation has not previously been described.
    OBJECTIVE: To collect and analyse data on the use and management of PSV in critically ill patients receiving prolonged ventilation.
    DESIGN, SETTING AND PARTICIPANTS: We performed a multicentre retrospective observational study in Australia, with a focus on PSV in patients ventilated for ≥ 7 days.
    MAIN OUTCOME MEASURES: We obtained detailed data on ventilator management twice daily (8am and 8pm moments) for the first 7 days of ventilation.
    RESULTS: Among 143 consecutive patients, 90/142 (63.4%) had received PSV by Day 7, and PSV accounted for 40.5% (784/1935) of ventilation moments. The most common pressure support level was 10 cmH2O (352/780) observations [45.1%]) with little variation over time, and 37 of 114 patients (32.4%) had no change in pressure support. Mean tidal volume during PSV was 8.3 (7.0–9.5) mL/kg predicted bodyweight (PBW) compared with 7.5 (7.0–8.3) mL/kg PBW during mandatory ventilation (P < 0.001). For 74.6% (247/331) of moments, despite a tidal volume of more than 8 mL/kg PBW, the pressure support level was not changed. Among 122 patients exposed to PSV, 97 (79.5%) received likely over-assistance according to rapid shallow breathing index criteria. Of 784 PSV moments, 411 (52.4%) were also likely over-assisted according to rapid shallow breathing index criteria, and 269/346 (77.7%) having no subsequent adjustment of pressure support.
    CONCLUSIONS: In patients receiving prolonged ventilation, almost two-thirds received PSV, which accounted for 40.5% of mechanical ventilation time. Half of the PSV-treated patients were exposed to high tidal volume and two-thirds to likely over-assistance. These observations provide evidence that can be used to inform interventional studies of PSV management.
  • References
    1. Slutsky AS. Mechanical ventilation. American College of Chest Physicians’ Consensus Conference. Chest 1993; 104: 1833-59.
    2. Acute Respiratory Distress Syndrome Network; Brower RG, Matthay MA, Morris A, et al. Ventilation with lower tidal volumes as compared with traditional tidal volumes for acute lung injury and the acute respiratory distress syndrome. N Engl J Med 2000; 342: 1301-8.
    3. Amato MB, Meade MO, Slutsky AS, et al. Driving pressure and survival in the acute respiratory distress syndrome. N Engl J Med 2015; 372: 747-55.
    4. Sassoon CS, Zhu E, Caiozzo VJ. Assist-control mechanical ventilation attenuates ventilator-induced diaphragmatic dysfunction. Am J Respir Crit Care Med 2004; 170: 626-32.
    5. Levine S, Nguyen T, Taylor N, et al. Rapid disuse atrophy of diaphragm fibers in mechanically ventilated humans. N Engl J Med 2008; 358: 1327-35.
    6. Brochard L. Less sedation in intensive care: the pendulum swings back. Lancet 2010; 375: 436-8.
    7. Boles JM, Bion J, Connors A, et al. Weaning from mechanical ventilation. Eur Respir J 2007; 29: 1033-56.
    8. MacIntyre NR. Respiratory function during pressure support ventilation. Chest 1986; 89: 677-83.
    9. Brochard L, Rauss A, Benito S, et al. Comparison of three methods of gradual withdrawal from ventilatory support during weaning from mechanical ventilation. Am J Respir Crit Care Med 1994; 150: 896-903.
    10. Esteban A, Frutos F, Tobin MJ, et al. A comparison of four methods of weaning patients from mechanical ventilation. Spanish Lung Failure Collaborative Group. N Engl J Med 1995; 332: 345-50.
    11. Pinto Da Costa N, Di Marco F, Lyazidi A, et al. Effect of pressure support on end-expiratory lung volume and lung diffusion for carbon monoxide. Crit Care Med 2011; 39: 2283-9.
    12. Thille AW, Cabello B, Galia F, et al. Reduction of patient-ventilator asynchrony by reducing tidal volume during pressure-support ventilation. Intensive Care Med 2008; 34: 1477-86.
    13. Brochard L, Harf A, Lorino H, Lemaire F. Inspiratory pressure support prevents diaphragmatic fatigue during weaning from mechanical ventilation. Am Rev Respir Dis 1989; 139: 513-21.
    14. Pletsch-Assuncao R, Caleffi Pereira M, Ferreira JG, et al. Accuracy of invasive and noninvasive parameters for diagnosing ventilatory overassistance during pressure support ventilation. Crit Care Med 2018; 46: 411-7.
    15. Rose L, Presneill JJ, Johnston L, et al. Ventilation and weaning practices in Australia and New Zealand. Anaesth Intensive Care 2009; 37: 99-107.
    16. Jabaley CS, Groff RF, Sharifpour M, et al. Modes of mechanical ventilation vary between hospitals and intensive care units within a university healthcare system: a retrospective observational study. BMC Res Notes 2018; 11: 425.
    17. Esteban A, Anzueto A, Alia I, et al. How is mechanical ventilation employed in the intensive care unit? An international utilization review. Am J Respir Crit Care Med 2000; 161: 1450-8.
    18. Simonis F, Binnekade J, Barber A, et al. Effect of a low vs intermediate Tidal Volume Strategy on Ventilator-Free Days in Intensive Care unit Patients Without ARDS. A Randomised Clinical Trial. JAMA 2018; 320 (18): 1872-1880.
    19. Clark FJ, von Euler C. On the regulation of depth and rate of breathing. J Physiol 1972; 222: 267-95.
    20. Jubran A, Van de Graaff WB, Tobin MJ. Variability of patient-ventilator interaction with pressure support ventilation in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med 1995; 152: 129-36.
    21. Leung P, Jubran A, Tobin MJ. Comparison of assisted ventilator modes on triggering, patient effort, and dyspnea. Am J Respir Crit Care Med 1997; 155: 1940-8.
    22. Petrof BJ, Hussain SN. Ventilator-induced diaphragmatic dysfunction: what have we learned? Curr Opin Crit Care 2016; 22: 67-72.
    23. Al-Bassam W, Dade F, Bailey M, et al. “Likely overassistance” during invasive pressure support ventilation in patients in the intensive care unit: a multicentre prospective observational study. Crit Care Resusc 2019; 21: 18-24.
Invasive mechanical ventilation can be delivered via a mandatory or spontaneous ventilation mode. 1  Choice of mode depends on patient factors, familiarity and clinician preference. Choice may also vary according to country and centre because there is no high quality evidence to guide practice. 1  During mandatory ventilation, a tidal volume (Vt) of 6–8 mL/kg of predicted bodyweight (PBW), a plateau pressure of less than 30 cmH20, and a driving pressure of less than 15 cmH2O are recommended, especially in patients with acute respiratory distress syndrome. 2, 3  However, for pressure support ventilation (PSV), there is no strong evidence to guide the level of pressure support (PS) and target Vt.
The proposed advantages of PSV over mandatory ventilation include its potential to improve respiratory muscle strength, 4, 5  reduce sedation requirements, 3, 6 and assist clinicians in determining the readiness of patients to be liberated from the ventilator. 7  However, in contrast to mandatory ventilation, there is no evidence on the use and management of PSV in patients receiving prolonged ventilation. There is also little information to guide clinicians in terms of optimal PSV practice. 8, 9, 10 Some authors suggest targeting a low Vt, 11, 12 but others recommend assessment of accessory muscle activity to determine the adequacy of a given PS level. 13 In patients receiving prolonged ventilation, there are no data on key aspects of management. These include the timing of PSV initiation; the duration of PSV use; the level of PS applied; the delivered Vt size and respiratory rate; the changes in respiratory variables after transition from mandatory ventilation to PSV; and the percentage of time spent on PSV. Yet this information is essential for designing interventional studies that aim to improve the management of PSV.
Accordingly, we conducted a retrospective observational multicentre study of patients receiving invasive mechanical ventilation for more than 1 week. We aimed to test three hypotheses regarding patients receiving prolonged ventilation: the primary hypothesis that more than a third of total ventilation time would be spent on PSV; and the secondary hypotheses that Vt would be significantly greater during PSV than during mandatory ventilation, and that likely over-assistance would be common.


Study design

This was a multicentre retrospective observational study of invasive ventilation practice in five mixed medical–surgical adult intensive care units (ICUs) in Victoria, Australia, with a focus on PSV. Three ICUs were in tertiary hospitals and two were in regional hospitals. The study was approved by the ethics committees of all the participating institutions with a waiver of informed consent (ethics approval number, LNR/17/Austin/265).


We included ≥ 20 consecutive adult patients (aged ≥ 18 years) admitted to each participating ICU who received invasive ventilation from March 2017 to August 2019. The main inclusion criterion was prolonged ventilation, defined as receiving ventilation for 7 or more days. We excluded patients receiving extracorporeal membrane oxygenation, patients receiving palliative care, and heart and lung transplant patients. We only considered first (index) ICU admission data.

Data collection

We collected data for 7 consecutive days, starting on the first day of ventilation, and at two “moments” per day (8am and 8pm). Thus, all patients had 14 measurements of ventilatory variables, sedative doses (midazolam, morphine, fentanyl and propofol), and arterial blood gas levels. In addition, we collected baseline data and admission diagnoses. We collected all data from electronic health records.


For every patient, the number of hours on PSV, on synchronised intermittent mandatory ventilation (SIMV), on T-piece ventilation, on other modes of ventilation (eg, assist control ventilation), or without invasive ventilation (for those re-intubated during the 7-day study period) were calculated, with the maximum being 168 hours (7 days). In addition, we calculated the percentage of time spent on PSV, defined as the number of hours on PSV divided by 168.
We calculated PBW in kg using a standard formula: 50 + 0.91 × (height [cm] – 152.4) for males; and 45.5 + 0.91 × (height [cm] – 152.4) for females. We calculated the rapid shallow breathing index (RSBI) in breaths/min/L as: respiratory rate (breaths/min) ÷ tidal volume (L). We calculated the cumulative doses of sedatives as the sum of all doses available in the 14 moments of measurement per patient. Moments without the use of the sedative of interest were coded as 0. We calculated the mean daily doses of sedatives, and the mean values for arterial blood gas levels, as means of all measurements available for these variables. Finally, we recorded the lowest and highest values for all measurements.
To understand the prevalence of over-assistance during PSV, we used two definitions of “likely over-assistance” as previously reported: respiratory rate ≤ 17 breaths/min and/or RSBI ≤ 37 breaths/min/L. 12, 14 In addition, we calculated the variation in PS levels during the period of observation as the coefficient of variation per patient, defined as the standard deviation of the PS level divided by its mean (presented as percentage).


The primary outcome of the study was the percentage of invasive ventilation time spent on PSV compared with other modes in the first 7 days of ventilation. Secondary outcomes were Vt size delivered during mechanical ventilation (before and during PSV compared with mandatory ventilation) and the prevalence of likely over-assistance during PSV. In addition, we report data on ICU length of stay, hospital length of stay, ICU mortality and hospital mortality.

Statistical analyses

We report continuous variables as median and interquartile range (IQR; quartile 25% to quartile 75%) and categorical variables as number and percentage. Each measurement was classified according to mode of ventilation at the time it was taken (PSV, SIMV or other). We compared the cumulative doses of sedatives, the mean, lowest and highest values for ventilatory variables, and arterial blood gas levels, according to use of PSV at time of measurement (PSV versus no PSV) using median differences and 95% confidence intervals. We calculated median differences using a mixed-effect quantile model considering a τ = 0.50 and an asymmetric Laplace distribution. We obtained P values after 1000 bootstrap samples. Due to within-patient data clustering, we included patients as a random effect. We compared continuous variables between three or more groups using the Kruskal–Wallis test.
We further divided patients according to first ventilation mode, and according to the most used mode (the most frequently reported mode in the 14 assessments available). To expand the findings, we assessed the changes in ventilatory variables due to transition to PSV. We compared moments classified as immediately before transition (moment exactly before the first transition to PSV), immediately after transition (moment exactly after the first transition to PSV), and 24 hours after transition (moment 24 hours after the first transition to PSV and considered only when PSV was still the mode being used).
We present ventilatory variables in cumulative distribution plots considering an empirical cumulative distribution function. This type of plot represents the cumulative distribution of the variable of interest according to all measurements assessed. We performed all analyses using R version 4.0.2 (R Core Team).

Availability of data and material

The datasets used and analysed during the current study are available from the corresponding author on reasonable request.



We studied 143 consecutive patients. Of them, 94 (65.7%) were men, the median age was 62.0 years, and the median APACHE III score was 68.0 (IQR, 39.5–89.5). Most were medical patients admitted due to a respiratory condition (Table 1). ICU and hospital mortality rates were 27.9% and 44.3%, respectively.

Mechanical ventilation practice

The ventilation mode was available for 1935 available moments (96.7%), with 784 (40.5%) classified as PSV, 884 (45.7%) as SIMV, and 267 (13.8%) as other modes of ventilation. At the start of study period, 16 patients (11.3%) were already on PSV, 98 (69.5%) were already on SIMV, and 27 (19.1%) were already on other modes of ventilation. General clinical characteristics and clinical outcomes were similar between the groups. However, the APACHE III scores were lower in patients who were classified as being on other modes of ventilation, and surgical patients were less frequently ventilated with SIMV (Online Appendix, eTable 1). We found a similar pattern according to the most used ventilation mode (Online Appendix, eTable 2).
Over the first 7 days, there was a progressive increase in the percentage of time spent on PSV. Overall, 39.8% of total ventilation moments were on PSV (median [IQR] duration of PSV, 61.0 [26.5–97.5] hours) and 47% were on SIMV (median [IQR] duration of SIMV 82.0 [23.5–119.0] hours) (Figure 1; Online Appendix, eTable 3). By Day 7, 63.4% of study patients were on PSV, 25.3% were on SIMV, and 9.1% were on other modes of ventilation.

Pressure support ventilation practice

The most common PS level was 10 cmH2O (352/780) [45.1%] observations), followed by 12 cmH2O (103/708) [13.2%] observations) and 5 cmH2O (95/780 [12.2%] observations). There was little variation in PS level over time (mean coefficient of variation per patient, 10.7 ± 24.4%), with 37 of 114 patients (32.4%) having no change in PS level over the entire observation period (Figure 2; Online Appendix, eFigure 1 and eFigure 2).

Ventilatory variables during pressure support ventilation

The median (IQR) Vt during PSV was 8.3 (7.0–9.5) mL/kg PBW, compared with 7.5 (7.0–8.3) mL/kg PBW with SIMV, and 7.7 (7.2–9.1) mL/kg PBW with other modes of ventilation (P < 0.001) (Table 2). Mean respiratory rate, minute ventilation and RSBI were higher during PSV (Table 2). A comparison of moments with PSV versus moments with no PSV is shown online (Online Appendix, eTable 3 and eFigure 3). In total, 40% (315/784) of the measurements during PSV had a Vt ≥ 8 mL/kg PBW, compared with 25% (293/1151) during moments with no PSV (Figure 3). When Vt was > 8 mL/kg PBW, clinicians did not change the PS level in 75% of moments (247/331, they increased the PS level in 10% of moments (32/331), and they decreased the PS level in 16% of moments (52/331) (Online Appendix, eFigure 4).
During PSV, 53% (417/784) and 42% (326/784) of the measurements had an RSBI of ≤ 37 breaths/min/L and a respiratory rate of ≤ 17 breaths/min, respectively (Figure 3) implying likely over-assistance. Overall, in 78% of RSBI moments (269/346), clinicians did not change the PS level, in 9% (30/346) they increased it and in 14% (47/346) they decreased it (Online Appendix, eFigure 4). Comparisons of measurements in PSV, SIMV and other modes of ventilation are shown online (Online Appendix, eFigure 5). While an increase in PS level led to a decrease in Vt size (median difference, –0.13 [95% CI, –0.25 to –0.01]; P = 0.025), we found no relationship between RSBI or respiratory rate and PSV levels (Online Appendix, eFigure 6). However, the relationships between the changes in PS level and the Vt and RR measurements were highly variable (Online Appendix, eFigure 7).

Transition to PSV

After transition to PSV, there were increases in minute ventilation and absolute Vt (Online Appendix, eTable 4). After 24 hours of PSV, absolute Vt and Vt corrected for PBW were similar to values before transition. However, there were increases in respiratory rate, minute ventilation, RSBI and pH (Figure 4; Online Appendix, eTable 5 and eFigure 8) even though sedative doses did not change after transition. After transition to PSV, there was a significant increase in the number of patients receiving a Vt > 8 mL/kg PBW and having a low respiratory rate (Online Appendix, eFigure 9). However, after 24 hours, the Vt and respiratory rate patterns were similar to those in the moments immediately before transition.

Likely over-assistance

Among 122 patients exposed to PSV, the vast majority had at least one moment of likely over-assistance according to respiratory rate (94 patients [77.0%]) and RSBI (97 patients [79.5%]). Of 784 PSV moments, 326 (41.6%) were in the likely over-assistance range according to the respiratory rate and 411 (52.4%) according to the RSBI (Figure 3).

Sedation and arterial blood gases

The cumulative and mean daily doses of fentanyl and propofol were lower during moments on PSV than during moments not on PSV (Online Appendix, eTable 6 and eTable 7). While Fio2 and Paco2 levels were lower during moments on PSV, pH and HCO3 levels were higher. Oxygenation parameters (Spo2 levels and Pao2/Fio2 ratios) were similar.


Key findings

We assessed the current practice of PSV in a cohort of patients in Australian ICUs who received prolonged invasive mechanical ventilation (more than 1 week). We found that clinicians applied PSV during 40.5% of the total ventilation moments, and that more than 60% of patients were exposed to PSV during the study period. The initiation of PSV resulted in a higher mean Vt, with 40% of the measurements under PSV showing a Vt of ≥ 8 mL/kg PBW, significantly more than with SIMV. In addition, the most common PS level by far was 10 cmH2O, with little variation during the study period. Finally, likely over-assistance was evident in half of PSV moments, and PS level adjustment in response to this was uncommon.

Relationship to previous studies

To our knowledge, this is the first study of PSV practice in patients receiving prolonged invasive ventilation. However, more than 10 years ago, a single 1-day point prevalence study was performed in Australia and New Zealand and showed that PSV was used in 41% of patients. 15  This study did not report data on Vt or over-assistance during PSV. In contrast, a multicentre observational study that was conducted in 12 ICUs in the United States and published in 2018 showed that PSV was used in only 10% of patients, but it also provided no additional data on PSV use. 16  A multinational 1-day point prevalence study conducted in nine countries, which included 1638 patients from 412 ICUs across North America, South America and Europe, showed that PSV was used for maintenance ventilation in only 6% of patients.
If the above findings are representative, then the practice and choice of invasive mechanical ventilation mode is markedly different in Australia compared with the United States and perhaps elsewhere. It is unclear why overseas ICUs appear to avoid PSV, while Australian ICUs appear to embrace it. 17  However, the recent PReVENT study, which allowed the use of PSV, found that PSV was applied to 20% of the patients in the first 3 days, implying growing use. 18  On the other hand, in these patients, it was impossible to achieve the target low Vt — a possible reason for the avoidance of this mode of ventilation in some, and perhaps many, ICUs.
Although the clinical impact of higher Vt during PSV on clinical outcomes has not been assessed, a large Vt size during PSV has been shown to decrease the neural inspiratory time and inspiratory efforts. 19  Moreover, a high PS level can: lead to overdistension, prolonged inspiratory time and increased patient–ventilator dys-synchrony; 12, 20, 21 stretch the diaphragm; and worsen ventilator-induced diaphragm dysfunction. 22  Our study showed that in 17% of the measurements during PSV, Vt was ³ 10 mL/kg of PBW. Despite this, the PS level of 10 cmH2O appeared stereotypical, and was rarely adjusted, similar to what was seen in a previous study of patients ventilated for a shorter period. 23
In a recent study of over-assistance during PSV — which was defined as the occurrence of a work of breathing level of < 0.3 J/L, or an incidence of ineffective inspiratory efforts of > 10% 14  — it was found that a respiratory rate of ≤ 17 breaths/min, or a RSBI of ≤ 37 breaths/min/L, accurately predicted over-assistance. The occurrence of such likely over-assistance was evaluated in a multicentre Australian study of patients receiving short-term PSV, where it occurred in 40–53% of the measurements. 23  In our study, we found a similar rate of likely over-assistance.

Implications of study findings

Our findings imply that PSV is common in patients receiving prolonged invasive ventilation in Australian ICUs, where it accounts for close to half of overall ventilation time in the first 7 days of ventilation. Moreover, our data imply that the transition to PSV is associated with more patients experiencing higher Vt, and that PSV appears to be prescribed in a stereotypical way, with little adjustment. Finally, they imply that likely over-assistance occurs frequently during PSV.

Study strengths and limitations

To our knowledge, this is the first multicentre study describing the practice details of PSV applied to patients receiving prolonged ventilation in an ICU setting. It involved three metropolitan ICUs and two regional ICUs with wide varieties of diagnoses, and it likely represented the ventilation practice of more than 100 ICU specialists, fellows and senior residents with more than 2000 observations.
Our study has some limitations. It was performed in the state of Victoria and may not represent national practice. However, the use of ventilation modes was similar to the ventilation modes in a previous study conducted in Australia and New Zealand. Thus, our findings are likely to reflect Australian and New Zealand use of PSV in patients receiving prolonged ventilation.
In addition, we used a retrospective study design. However, we obtained documented ventilation data for 7 consecutive days, from different hospitals and from a wide variety of patient populations at different time intervals. The definitions of over-assistance were extrapolated from a previous study, but we found that Vt size in mL/kg of PBW was high, which is also consistent with likely over-assistance. Moreover, we used the term “likely” because we did not use oesophageal manometry to measure the work of breathing.


In patients receiving prolonged invasive ventilation in Australia, PSV appeared to account for almost half of mechanical ventilation time during the first 7 days. Moreover, it appears that about half of the patients who received PSV were exposed to large Vt and likely over-assistance. In moments of large Vt or likely over-assistance, there was little response by clinicians, with few changes in PS levels. These observations imply that PSV management could be improved. They also provide the epidemiological background needed to plan interventional studies aimed at optimising the quality of PSV.