Table 3 Studies evaluating non-invasive and invasive LAP assessment in critical care populations

Invasive PAOP versus LVEDP studies

Lozman et al. [22]

Single centre, N = 5. Invasively ventilated post-operative cardiac surgical patients without ARDS

The relationship between PAOP and directly measured LAP was lost at PEEP levels above 15 cm H20

Not specified

Jardin et al. [23]

Single centre, N = 10. Invasively ventilated patients with ARDS. PAOP was measured at end expiration. LVEDP was measured with an LV catheter, defined as the pre-ejection diastolic plateau or onset of the ECG q wave

Below PEEPs of 10cmH20, PAOP correlated with invasively measured LVEDP. Correlation was diminished at PEEP values > 10cmH20 with PAOP being higher than LVEDP. Correlation values not provided

Not specified

Teboul et al. [24]

Single centre, N = 12. Patients with ARDS. Simultaneous measurement of PAOP and LVEDP at PEEP levels up to 20cmH20. PAOP, measured as the mean value at end expiration and averaged over 5 or more cycles. LVEDP measured at the ‘z’ point (i.e. at the end of the ‘a’ wave)

PAOP usually agreed with invasively measured post-A wave LVEDP by 1–2 mmHg. ‘Close correlation’ was seen between PAOP and LVEDP at PEEP levels up to 20cmH20

Authors suggested this observed correlation of PAOP and LVEDP is due to surrounding diseased lung preventing alveolar vessel compression

Contraindication to left heart catheterisation (aorto-femoral atherosclerosis, aortic stenosis, thrombocytopenia or coagulopathy)

Non-invasive Echo Doppler LAP versus PAOP

Brault et al. [29]

Prospective study across  two ICUs. N = 98. All mechanically ventilated. Pooled analysis of 3 prospective cohorts with simultaneously assessed LAP by echo and PAOP by PA catheter measured at end expiration and averaged over 5 cardiac cycles

The sensitivity and specificity of ASE/EACVI guidelines for predicting elevated PAOP ≥ 18 mmHg were both 74%. Agreement between echocardiography measured raised LAP and elevated PAOP (> 18 mmHg) was moderate (Cohen’s Kappa, 0.48; 95% CI, 0.39–0.70)

New simplified algorithm proposed: LVEF < 45% E/A cut off < 1.5 and LVEF > 45% lateral e’ cut off > 8 for predicting PAOP < 18 mmHg. Sensitivity and specificity of the proposed algorithm for predicting an elevated PAOP were 87% and 73%, respectively

Arrhythmia, severe mitral or aortic valvulopathy, merged Doppler mitral flow, or inadequate image quality for Doppler measurements

Vignon et al. [31]

Prospective, single-centre, two consecutive 3-year periods. N = 88 mechanically ventilated patients. Protocol A, n = 56 used to estimate Doppler parameters predicting PAOP ≤ 18 mmHg,

Protocol B, n = 32, derived Doppler values from protocol A were tested prospectively

In protocol B, mitral E/A ≤ 1.4, pulmonary vein S/D > 0.65 and systolic fraction > 44% best predicted an invasive PAOP ≤ 18 mmHg. Correlations between Doppler and PAOP values were consistently closer in the subset of patients with depressed LV systolic function

Lateral E/e′ ≤ 8.0 or E/Vp ≤ 1.7 predicted a PAOP ≤ 18 mmHg with a sensitivity of 83% and 80%, and a specificity of 88% and 100%, respectively. Areas under ROC curves of lateral E/e′ and E/Vp were similar (0.91 ± 0.07 vs 0.92 ± 0.07: p = 0.53)

Non-sinus rhythm, ‘relevant’ valvulopathy, AV conduction abnormality, TOE contraindication

Nagueh et al. [32]

Single-centre ICU. Complex design. N = 36 (20 mechanically ventilated) having adequate TTE Doppler tracings and PAC (initial study group). 32 patients were later enrolled (prospective study group, unspecified proportion mechanically ventilated)

Correlation of PAOP with E/A ratio (r = 0.75), IVRT (r =  − 0.55), DT (r =  − 0.5) and atrial filing fraction (r =  − 0.65). PA occlusion pressure equation derived incorporating E/A and IVRT and correlation assessed with invasive PAOP: r = 0.79 and r = 0.88 in initial and prospective groups, respectively

AF, inadequate Doppler recording, fusion of E/A

Mousavi et al. [33]

Retrospective, single centre. N = 40 patients with septic shock. TTE Doppler and PAC PAOP within 4 h. Methods for PAOP measurement not specified

Correlation between average E/e′ and PAOP (r = 0 .84, p < 0.05)

Not fulfilling criteria for septic shock

Dokainish et al. [34]

Prospective, single-centre ICU. N = 50, patients who had existing PAC. 21 invasively ventilated. Simultaneous measurements of echo Doppler and PAOP and BNP

Correlation between E/e′ and PAOP: r = 0.69 (p < 0.001)

E/e′ > 15 was the optimal cut off to predict PAOP > 15 mm Hg (sensitivity, 86%; specificity, 88%)

E/e′ was more accurate in those with cardiac disease

AF, paced rhythm, severe MR, MS, mitral prosthesis, severe MAC, acute MI, unstable angina, and CABG within 72 h

Combes et al. [35]

Prospective, single-centre ICU. N = 23, consecutive mechanically ventilated patients. TOE or TTE Doppler versus PAOP by PA catheter

PAOP and the lateral E/e′ correlation (r = 0.84) and medial E/e′ correlation (r = 0.76). The sensitivities and specificities of estimating PAOP > 15 mmHg were, respectively, 86% and 81% for lateral E/e′  > 7.5 and 76% and 80% for medial E/e′ > 9

Age < 18 years, non-sinus rhythm, mitral insufficiency greater than grade 2 and mitral stenosis, prosthetic mitral valve, tachycardia that prevented a distinct separation between the E and A waves

Bouhemad et al. [36]

Prospective, single-centre ICU. N = 60, admitted with septic shock and acute lung injury

Simultaneous comparison of echo Doppler with TOE and PAOP. All patients mechanically ventilated. PEEP was removed or reduced to 5cmH20 during study

Mean bias variation between invasive PAOP and PAOP measured with Doppler (using the equation 0.97x E/e′ + 4.34) was of 0.5 mmHg with a precision of 2.0 mmHg

ROC curves demonstrated that an E/e′ > 6 was an accurate predictor of a PAOP of ≥ 13 mmHg (AUC 0.98)

Changes in PAOP were significantly correlated to changes in E/e′ (Rho 0.84, p < 0.0001)

Unable to have TOE, lack of sinus rhythm, BBB, left ventricular systolic dysfunction, presence of a significant mitral pathology, CAD and segmental wall motion abnormality

Dabaghi et al. [37]

Prospective, single-centre ICU over 6-month period in consecutive patients requiring invasive haemodynamic monitoring and echocardiography. N = 49. PAOP performed at end expiration

Left ventricular filling pressure calculated non-invasively by: 46 − (0.22 − x IVRT) − (0.10 × AFF) − (0.03 × DT) − (2/[E/A]) + (0.05 × MAR)

Mean values 21 ± 8 vs 20 ± 8 mm Hg, for non-invasive and invasive, respectively. Correlation r = 0.88

Not in sinus rhythm, MS or prosthetic mitral valve. PEEP was < 10cmH20 in all patients