American Journal of Respiratory and Critical Care Medicine

Critically ill patients requiring mechanical ventilation often develop intrinsic positive end-expiratory pressure (PEEPi). Methods for its detection include an expiratory flow waveform display (not always available), an esophageal pressure transducer (invasive), or a relaxed or paralyzed patient. We sought to determine the accuracy of clinical examination for detecting PEEPi. Examiners blinded to waveform analysis assessed patients for the presence of PEEPi by inspection/palpation and auscultation. If either inspection/palpation or auscultation demonstrated PEEPi, it was said to be present by clinical exam. Clinicians with various levels of experience (attending, resident, student) made 503 observations of 71 patients. Sensitivity (SENS), specificity (SPEC), positive predictive value (PPV), negative predictive value (NPV), and likelihood ratios were determined for inspection/palpation, auscultation, and clinical exam. PEEPi was present during 69.8% of observations. SENS, SPEC, and PPV of clinical exam were 0.72, 0.91, and 0.95 respectively for the examiners as a whole. Likelihood ratio for PEEPi detection by clinical exam was 8.35. Attending intensivists displayed SPEC and PPV of 1.0. NPV was only 0.58 (likelihood ratio 0.31). We conclude that the clinical exam is very good for detecting PEEPi at all experience levels; and further, that the clinical exam is only modestly useful for ruling out PEEPi, therefore, other tests should be used if PEEPi is not detected by clinical exam.

The adverse effects of intrinsic positive end-expiratory pressure (PEEPi) have been recognized since Pepe and Marini first reported its presence in 1982 (1). These effects include hemodynamic compromise (2, 3), misinterpretation of central venous and pulmonary artery catheter pressure measurements (1), erroneous calculations of static respiratory compliance (2, 4), and increases in work of breathing (5-8) which may interfere with attempts at liberation from mechanical ventilation (9, 10).

PEEPi can be measured in mechanically ventilated patients by several different means. End-expiratory airway occlusion (1, 11) is the simplest approach. However, this requires a cooperative patient who does not resist the occlusion maneuver. Typically, only heavily sedated or paralyzed patients can have PEEPi determined reliably with this approach. Simultaneous recording of flow and pressure at the airway opening is another approach. This method of measurement determines the increase from the unoccluded end-expiratory airway pressure required to initiate inspiratory flow (12). This method can, likewise, be used reliably only when the respiratory muscles are quiescent. In addition, it may be difficult to measure by visual analysis, usually requiring computer-assisted data analysis to ensure accuracy. Esophageal pressure manometry is a third way of measuring PEEPi, but it requires insertion of an esophageal balloon as well as equipment to transduce and display esophageal pressure.

Ventilator flow tracings are an accurate way of detecting the presence of PEEPi. Normally, expiratory airflow ceases before the subsequent inspiration. PEEPi is depicted by the presence of persistent end-expiratory airflow (2, 13). The flow-versus-time waveform allows a qualitative determination, but not a quantitative one.

Many ventilators do not have on-line waveform tracing capability. Therefore, a simple, noninvasive test for detecting the presence of PEEPi may be useful to clinicians caring for mechanically ventilated patients. Given the importance of recognizing PEEPi and the limitations of current methods for detecting it, being either invasive or requiring technology which is not available universally, we sought to evaluate the utility of bedside clinical examination.

All patients were mechanically ventilated with a Puritan Bennett 7200 ventilator (Carlsbad, CA) with waveform display or a Servo Siemens-Elema 900C ventilator (Sweden). Those ventilated with a Servo ventilator had waveforms displayed using a Bicore CP-100 pulmonary monitor (Irvine, CA). All waveform monitors were calibrated prior to use. This study was approved by the institutional review board at the University of Chicago. Demographic data (age, sex), respiratory rate, minute ventilation (V˙e), peak and plateau airway pressures, mode of ventilation, and primary cause of respiratory failure were recorded for all patients.

All patients were evaluated in the supine or semirecumbent position. Clinicians participating in the study were asked to examine the patient and to answer the question, “Does this patient have PEEPi?” (yes or no). Each examiner was asked to assess the patient by simultaneous inspection and palpation (inspection/palpation), followed by auscultation. For inspection/palpation, all examiners were instructed to look at and palpate the patient's chest during exhalation, looking for “continuous chest wall incursion (inward movement) which persisted up until the moment the next breath occurred.” For auscultation, all examiners were asked to auscultate the chest, listening for “the persistence of breath sounds during exhalation up until the moment the next breath occurred.” If an examiner answered “yes” to either inspection/palpation or auscultation, PEEPi was said to be present by combined clinical exam. If the answer was “no” to both inspection/ palpation and auscultation, it was said to be absent by clinical exam.

We studied examiners at different levels of training: attending critical care physicians, internal medicine residents, and fourth-year medical students. Each examiner evaluated a given patient no more than three times over the course of 4 d (no more than once per day). Clinicians were blinded to each others' assessments and to the ventilator waveform tracings as well as any arterial and venous pressure waveform tracings. A physician collecting data for the study (J.P.K.) analyzed simultaneous airway pressure-versus-time and flow-versus-time waveforms at the same time as the bedside examination. The reference standard for PEEPi was the presence of persistent expiratory airflow up to the next inspired breath as displayed on the waveform monitor. We attempted to quantify PEEPi, when present, using the expiratory port occlusion technique (1).

Data are expressed as mean ± SD. Interval data analysis was performed using unpaired, two-tailed t tests. Respiratory rate and V˙e were evaluated as a function of the primary cause of respiratory failure. These same two parameters were also evaluated as a function of mode of ventilation. Such comparisons were made with single-factor analysis of variance (ANOVA) with Newman-Keuls test for significance when appropriate. Primary cause of respiratory failure and PEEPi incidence, as well as mode of ventilation and PEEPi incidence were compared by χ2 analysis or Fisher exact test when appropriate. Bonferroni's correction for multiple comparisons was used. A p value of < 0.05 was considered significant. Sensitivity (SENS), specificity (SPEC), positive predictive value (PPV), and negative predictive value (NPV) were determined for each group of examiners, as well as all examiners together. Inspection/palpation, auscultation, and clinical exam were evaluated separately in each group of examiners. Likelihood ratios for the clinical exam as well as post-test probability of PEEPi in our patient population were determined for all groups of examiners.

A total of 503 observations were made on 71 different patients. The mean age was 61.1 ± 16.4 yr, with 36 male and 35 female patients. 238 observations were made by attending critical care physicians (7 examiners), 137 by residents (18 examiners), and 128 by students (15 examiners).

The primary causes of respiratory failure were distributed as follows: acute hypoxemic respiratory failure (AHRF)— 28%, ventilatory failure—23%, postoperative respiratory failure—15%, sepsis/shock—14%, airway protection/weakness— 20%. “Airway protection/weakness” was listed separately from other types of ventilatory failure because of the lower incidence (expected and observed) of PEEPi compared with other types of ventilatory failure such as obstructive lung disease.

Respiratory rate was 23.1 in those with PEEPi versus 18.7 in those without it (p = 0.0001). V˙e was 12.4 versus 9.9 L/min in those with and without it respectively (p < 0.0001). Average peak airway pressure was 38.2 versus 33.1 cm H2O in those with and without PEEPi respectively (p = 0.0003); the difference between peak and plateau airway pressures averaged 13.6 cm H2O in those with PEEPi versus 11.0 cm H2O in those without it (p = 0.003).

When grouped by cause of respiratory failure, respiratory rate was different when comparing the following groups: AHRF versus postoperative respiratory failure (24.3 versus 17.7; p = 0.0006), AHRF versus airway protection/weakness (24.3 versus 19.7; p = 0.019), sepsis/shock versus postoperative respiratory failure (24.5 versus 17.7; p = 0.0006), and sepsis/ shock versus airway protection/weakness (24.5 versus 19.7; p = 0.025). A similar comparison between cause of respiratory failure and V˙e (L/min) yielded the following differences: AHRF versus sepsis/shock (11.0 versus 13.6; p = 0.038), airway protection/weakness versus postoperative respiratory failure (10.4 versus 13.0; p = 0.035), and airway protection/ weakness versus sepsis/shock (10.4 versus 13.6; p = 0.007). The incidence of PEEPi varied greatly when patients were grouped by cause of respiratory failure. Only two such comparisons showed no difference in PEEPi incidence: AHRF versus postoperative respiratory failure (68.8% versus 56.8%; p = not significant [NS]) and ventilatory failure versus sepsis/ shock (83.7% versus 95.4%; p = NS). All other comparisons between cause of respiratory failure and PEEPi incidence were significant at p < 0.005. Table 1 summarizes the relationships between causes of respiratory failure and PEEPi incidence, respiratory rate, and V˙e.


No. of ObservationsPEEPi Incidence (%)e(L/min)Respiratory Rate
Ventilatory failure14783.712.121.2
Postoperative respiratory failure 3756.813.017.7
Sepsis/shock 6595.413.624.5
Airway protection/weakness11644.010.419.7

Respiratory rate was higher in patients ventilated in the pressure control mode compared with all other modes (p < 0.0001 when comparing pressure control to any other mode). A comparison between ventilator mode and V˙e (L/min) yielded the following differences: pressure control versus assist control (14.4 versus 11.1; p = 0.005) and pressure control versus continuous positive airway pressure (CPAP) (14.4 versus 10.7; p = 0.003). PEEPi was present in a greater number of patients ventilated with CPAP compared with all other modes (p = 0.02). The relationships between ventilation mode and PEEPi incidence, respiratory rate, and V˙e are listed in Table 2.


No. of ObservationsPEEPi Incidence (%)e(L/min)Respiratory Rate
Assist control19271.211.122.9
Intermittent mandatory ventilation21663.912.419.7
CPAP 6783.610.722.0
Pressure control 2871.414.534.1

PEEPi was present during 351 of 503 (69.8%) observations. When present on a ventilator mode where it could be measured (e.g., all modes except CPAP), it was quantifiable during only 86 of 283 (30.4%) observations; the majority of the time, it could not be quantified because of respiratory muscle effort by the patient during the end-expiratory occlusion maneuver. For all examiners together, those observations where PEEPi was quantified yielded 50 “yes” responses by clinical exam (39 for levels less than or equal to 5 cm H2O, 11 for levels greater than 5 cm H2O); there were 36 “no” responses by clinical exam (33 for levels less than or equal to 5 cm H2O, 3 for levels greater than 5 cm H2O; p = 0.14 by Fisher exact test for all examiners). When only attendings were evaluated, there were zero false-negative assessments for PEEPi levels greater than 5 cm H2O; 18 of 18 false-negative assessments came for levels less than or equal to 5 cm H2O. Conversely, 5 of 17 true-positive results were for levels greater than 5 cm H2O (p = 0.02 by Fisher exact test for attendings).

Observations for detection of PEEPi as performed by various examiners are summarized in Tables 3 and 4. Combined clinical exam showed a greater SENS and NPV than either inspection/palpation or auscultation alone for all groups of examiners. The likelihood ratio for clinical exam detecting PEEPi was 8.35 for all examiners, which translated to a 95.1% post-test probability of PEEPi (the pretest probability was 69.8% in our population). Among attendings, the likelihood ratio for a clinical exam detecting PEEPi approached infinity, yielding a 100% post-test probability (Table 4). Attendings fared better than residents and students in all categories, though SPEC and PPV were high at all levels of training. The NPV of the clinical exam was less reliable at 0.58. Accordingly, the likelihood ratio for the clinical exam not detecting PEEPi was 0.31, meaning that an answer of “no” by clinical exam brought the pretest probability of 69.8% down to a post-test probability of 41.9%.


Clinical exam


Clinical ExamPretest Probability (%)Test Result (Likelihood Ratio)Post-Test Probability (%)
Attending69.8“yes” (∞)100
“no” (0.25)   36.5
Resident69.8“yes” (3.72)   89.7
“no” (0.40)   47.8
Student69.8“yes” (5.13)   92.3
“no” (0.36)   45.4
Overall69.8“yes” (8.35)   95.1
“no” (0.31)   41.9

PEEPi is a common finding in critically ill, mechanically ventilated patients (2, 4, 14, 15). While anticipated in patients with ventilatory failure from obstructive lung disease, it may not be expected in other conditions such as AHRF, sepsis, or weakness, even though previous studies have shown it to be common in all mechanically ventilated patients (4, 16).

The presence of PEEPi in conditions other than obstructive lung disease may be due to higher respiratory rates leading to inadequate time for expiration (2, 4, 17). Indeed, we noted a higher mean respiratory rate as well as V˙e in those with PEEPi compared with those without it. Because there were no tidal volume differences (data not shown), the increased V˙e was due entirely to higher respiratory rates. Peak airway pressures as well as peak-plateau airway pressure gradients were higher in those with PEEPi. Such derangements in pulmonary mechanics are often signs of increased airway resistance—a factor which predisposes a patient to PEEPi. Other reasons mechanically ventilated patients may develop PEEPi include increased airway resistance due to lung injury (16) as well as extrinsic increases in resistance from the endotracheal tube (18), ventilator tubing, and attached devices (2). It is also possible that occult obstructive lung disease may have been present in some patients.

PEEPi incidence varied widely when grouped by cause of respiratory failure, being lowest in those in the airway protection/weakness and postoperative respiratory failure categories and highest in the ventilatory failure and sepsis/shock categories. The high incidence in the latter groups is most likely caused by higher respiratory rates and the presence of obstructive lung disease.

Patients ventilated in the CPAP mode had a higher incidence of PEEPi compared with other ventilator modes. The reasons for this observation are not clear. It may be that spontaneously breathing patients inspire with a lesser flow rate than is typically provided during intermittent mandatory ventilation or assist-control ventilation, leaving less time for expiration. Alternatively, because CPAP is most often used during attempts at liberation from mechanical ventilation, the lifting of sedation may predispose toward bronchospasm during such liberation attempts, though our data cannot confirm this speculation.

PEEPi was quantifiable in a minority of those in whom it was observed. We used 5 cm H2O as a clinically important “high” versus “low” cutoff, and sought to evaluate the utility of the clinical exam for detecting such levels. All examiners together showed a trend toward significance at detecting “high” compared with “low” levels of PEEPi (p = 0.14); attendings were able to detect levels greater than 5 cm H2O (p = 0.02), with no false-negative responses at high levels. While this suggests that clinical exam is useful for detecting higher levels of PEEPi, caution should be used in interpreting these results because they apply to a small subset of our observations.

Many ventilators do not have the capability to display flow waveforms. Because the majority of end-expiratory airway occlusion maneuvers in our study did not reliably distinguish the presence or absence of PEEPi, intensivists are often unable to detect its presence. We wondered if this simple, bedside exam available to all clinicians would detect reliably this common and relevant condition. The SPEC and PPV of clinical exam, inspection/palpation, and auscultation were each 1.0 for attending physicians. Though residents and students had less impressive results, SPEC and PPV were 0.82 or greater for all tests of PEEPi in all examiners. Clinical exam had a likelihood ratio of 8.35 for all examiners. Therefore, in our population, with a pretest probability of 69.8%, a positive clinical exam translated to a post-test probability of 95.1%, confirming this method to be very reliable for detecting PEEPi. Although residents and students did not perform as well as attendings, they were still able to detect PEEPi with reasonable likelihood ratios. This suggests that the skills required to detect PEEPi at the bedside of mechanically ventilated patients can be readily learned at all levels of training. The fact that 40 different individuals at various levels of training participated in the study further supports this notion. We conclude that clinical exam is a reliable tool for detecting PEEPi in mechanically ventilated patients.

Clinical exam was not nearly as reliable to rule out the presence of PEEPi. The likelihood ratio for all examiners was 0.31, decreasing our pretest probability of 69.8% to a post-test probability of only 41.9%. Clearly, the bedside clinical exam cannot stand alone as a test to rule out the presence of PEEPi, and other methods of detection, such as ventilator waveform analysis or end-expiratory airway occlusion, should be considered when clinical exam does not detect PEEPi.

We did not report the likelihood ratios for inspection/palpation or auscultation separately, because clinical exam results had better likelihood ratios than either inspection/palpation or auscultation for all examiners. This finding emphasizes the importance of a complete physical exam in mechanically ventilated patients. Few studies have evaluated physical exam or other noninvasive bedside tests for the management of patients either in the intensive care unit (19, 20) or elsewhere (21). Recently, others have expressed concern over the replacement of physical examination by high-technology diagnostic tests (22). Inspection/palpation—a portion of the physical exam at times disregarded—performed essentially as well as, or better than, auscultation in all cases.

One limitation of this study is our inability to quantify most PEEPi measurements; thus, we do not know what percentage of our patients had only small degrees of PEEPi. However, to the extent that any degree of PEEPi is associated with an increase in work of breathing (6), this detection may be useful to the clinician, particularly during attempts to discontinue mechanical ventilation (9). Another possible weakness is reliance on persistent expiratory airflow as a marker of PEEPi, which assumes the alveoli will be in communication with the flow transducer at the tip of the endotracheal tube. Leatherman and Ravenscraft reported patients with severe asthma exacerbations who showed only minimally elevated PEEPi levels as measured by end-expiratory airway occlusion (23). This was presumably due to complete airway closure preventing communication between alveoli and the proximal airway pressure transducer. Breen and coworkers have demonstrated similar findings in a dog model (24). To the extent that this problem is common, we might have expected to detect PEEPi in some patients by clinical exam who had none by waveform analysis. Yet the PPV was 1.0 (for attendings). Perhaps this is because in most patients with severe PEEPi, some obstructed airways communicate with the proximal airway, and so persistent expiratory flow demonstrating the qualitative presence of PEEPi is seen.

In conclusion, we have noted a high failure rate when using the end-expiratory occlusion method to detect PEEPi in nonparalyzed patients, and have demonstrated the utility of bedside physical examination for detecting the presence of PEEPi in a prospective, blinded manner. This assessment can be learned at all levels of training and should be part of the routine evaluation of mechanically ventilated patients. On the other hand, the clinical exam is not a reliable test to rule out the presence of PEEPi, and other methods should be used to detect it, when suspected.

1. Pepe P. E., Marini J. J.Occult positive end-expiratory pressure in mechanically ventilated patients with airflow obstruction: the auto-PEEP effect. Am. Rev. Respir. Dis.1261982166170
2. Rossi A., Gottfried S. B., Zocchi L., Higgs B. D., Lennox S., Calverley P. M., Begin P., Grassino A., Milic-Emili J.Measurement of static compliance of the total respiratory system in patients with acute respiratory failure during mechanical ventilation: the effects of “intrinsic” PEEP. Am. Rev. Respir. Dis.1311985672677
3. Brandolese R., Broseghini C., Polese G., Bernasconi M., Brandi G., Milic-Emili J., Rossi A.Effects of intrinsic PEEP on pulmonary gas exchange in mechanically-ventilated patients. Eur. Respir. J.61993358363
4. Bernasconi M., Ploysongsang Y., Gottfried S. B., Milic-Emili J., Rossi A.Respiratory compliance and resistance in mechanically ventilated patients with acute respiratory failure. Intensive Care Med.141988547553
5. Fleury B. D., Murciano C., Talamo M., Aubier R., Pariente R.Work of breathing in patients with chronic obstructive pulmonary disease in acute respiratory failure. Am. Rev. Respir. Dis.1311985822827
6. MacIntyre N. R., Cheng K. C. G., McConnell R.Applied PEEP during pressure support reduces the inspiratory threshold load of intrinsic PEEP. Chest1111997188193
7. Fessler H. E., Brower R. G., Permutt S.CPAP reduces inspiratory work more than dyspnea during hyperinflation with intrinsic PEEP. Chest1081995432440
8. Coussa M. L., Guerin C., Eissa N. T., Corbeil C., Chasse M., Braidy J., Matar N., Milic-Emili J.Partitioning of work of breathing in mechanically ventilated COPD patients. J. Appl. Physiol.75199317111719
9. Fiastro J. F., Habib M. P., Shon B. Y., Campbell S. C.Comparison of standard weaning parameters and the mechanical work of breathing in mechanically ventilated patients. Chest941988232238
10. Shikora S. A., Bistrian B. R., Borlase B. C., Blackburn G. L., Stone M. D., Benotti P. N.Work of breathing: reliable predictor of weaning and extubation. Crit. Care Med.181990157162
11. Zakynthinos S. G., Vassilakopoulos T., Zakynthinos E., Roussos C.Accurate measurement of intrinsic positive end-expiratory pressure: how to detect and correct for expiratory muscle activity. Eur. Respir. J.101997522529
12. Maltais F., Reissman H., Navalesi P., Hernandez P., Gursahaney A., Ranieri V. M., Sovilj M., Gottfried S. B.Comparison of static and dynamic measurements of intrinsic PEEP in mechanically ventilated patients. Am. J. Respir. Crit. Care Med.150199413181324
13. Rossi A., Brandolese R., Milic-Emili J., Gottfried S. B.The role of PEEP in patients with chronic obstructive pulmonary disease during assisted ventilation. Eur. Respir. J.31990818822
14. Patel H., Yang K. L.Variability of intrinsic positive end-expiratory pressure in patients receiving mechanical ventilation. Crit. Care Med.23199510741079
15. Rossi A., Polese G., Brandi G., Conti G.Intrinsic positive end-expiratory pressure (PEEPi). Intensive Care Med.211995522536
16. Wright P. E., Bernard G. R.The role of airflow resistance in patients with the adult respiratory distress syndrome. Am. Rev. Respir. Dis.139198911691174
17. Broseghini C., Brandolese R., Poggi R., Bernasconi M., Manzin E., Rossi A.Respiratory resistance and intrinsic positive end-expiratory pressure (PEEPi ) in patients with the adult respiratory distress syndrome. Eur. Respir. J.11988726731
18. Wright P. E., Marini J. J., Bernard G. R.In vitro versus in vivo comparison on endotracheal tube airflow resistance. Am. Rev. Respir. Dis.14019891016
19. Watters D. A., Wilson I. H., Sinclair J. R., Ngandu N.A clinical sickness score for the critically ill in central Africa. Intensive Care Med.151989467470
20. McIntyre K. M., Vita J. A., Lambrew C. T., Freeman J., Loscalzo J.A non-invasive method of predicting pulmonary-capillary wedge pressure. N. Engl. J. Med.327199217151720
21. Williams J. W., Simel D. L., Roberts L., Samsa G. P.Clinical evaluation for sinusitis: making the diagnosis by history and physical examination. Ann. Intern. Med.1171992705710
22. Fletcher R. H., Fletcher S. W.Has medicine outgrown physical diagnosis? Ann. Intern. Med.1171992786787
23. Leatherman J. W., Ravenscraft S. A.Low measured auto-positive end-expiratory pressure during mechanical ventilation of patients with severe asthma: hidden auto-positive end-expiratory pressure. Crit. Care Med.241996541546
24. Breen P. H., Becker L. J., Ruygrok P., Mayers I., Long G. R., Leff A., Wood L. D. H.Canine bronchoconstriction, gas trapping, and hypoxia with methacholine. J. Appl. Physiol.631987262269
Correspondence and requests for reprints should be addressed to Gregory A. Schmidt, M.D., Section of Pulmonary and Critical Care Medicine, MC 6026, University of Chicago, 5841 S. Maryland Ave., Chicago, IL 60637.


No related items
American Journal of Respiratory and Critical Care Medicine

Click to see any corrections or updates and to confirm this is the authentic version of record