Comparative performance of six bedside dynamic tests for predicting fluid responsiveness in mechanically ventilated adults with shock: a prospective single-center diagnostic accuracy study

Introduction

Fluid administration is central to the resuscitation of circulatory shock, but not every hypotensive patient will benefit from additional fluid. In intensive care unit (ICU) cohorts, approximately half of hemodynamically unstable patients increase stroke volume or cardiac output after a fluid challenge, and fluid accumulation itself has been associated with worse outcomes (1-5).

Bedside hemodynamic assessment has therefore shifted from static filling pressures to dynamic tests that ask a more clinically relevant question: will flow increase if preload is augmented? The answer depends on physiology. Dynamic indices do not perform equally across patient populations, and their usefulness is shaped by heart rhythm, ventilation mode, tidal volume, spontaneous breathing activity, right ventricular function, intra-abdominal pressure, and the ability to monitor cardiac output in real time (4,6).

Several bedside tools are used in contemporary practice, including passive leg raising (PLR), end-expiratory occlusion test (EEOT), stroke volume variation (SVV), pulse pressure variation (PPV), inferior vena cava distensibility index (IVC-DI), and respiratory variation in left ventricular outflow tract velocity-time integral (respiratory ΔVTI) (7-11). Because pooled estimates from heterogeneous studies do not fully answer the bedside question of which test is most informative when several options are available in the same patient, within-protocol evaluation remains clinically useful.

We therefore prospectively evaluated and descriptively compared the diagnostic performance profiles of six bedside dynamic tests in mechanically ventilated adults with shock, using the change in cardiac index after a standardized 500-mL saline bolus as the protocol reference standard. We hypothesized that PLR would provide the most consistent overall diagnostic profile under strict physiological applicability conditions. We present this article in accordance with the STARD reporting checklist (available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-2026-0015/rc).

Methods

Study design and setting

This prospective single-center diagnostic accuracy study was performed in the Critical Care Medicine Department, Faculty of Medicine, Cairo University, Cairo, Egypt, between March 2020 and March 2021. Data collection was planned before performance of the index tests and the reference standard.

The target condition was fluid responsiveness, defined as a clinically relevant increase in cardiac index after a standardized fluid challenge. The intended clinical role of the six index tests was bedside prediction of fluid responsiveness before administering additional fluid.

Patient selection

Adults aged 18 years or older were eligible when they had shock with persistent hypotension after initial resuscitation, mean arterial pressure <60 mmHg, serum lactate ≥2 mmol/L, and required volume-controlled invasive mechanical ventilation with tidal volume ≥8 mL/kg. Potentially eligible participants were identified among ICU admissions with shock on the basis of clinical diagnosis, persistent hypotension, lactate elevation, and ventilatory status. Eligible patients were enrolled as a consecutive series when all protocol criteria were met and surrogate consent was available.

Patients were excluded if they had arrhythmia, spontaneous breathing activity, inability to obtain any study measurement, significant valvular disease, right-sided heart failure, abdominal compartment syndrome, pregnancy, acute respiratory distress syndrome requiring low-tidal-volume ventilation, or a clinical need for immediate fluid administration that would not allow completion of pre-fluid index testing. No positive end-expiratory pressure (PEEP)-specific exclusion threshold was prespecified; limitations related to unavailable PEEP-range data are addressed in the Discussion.

Hemodynamic assessment and dynamic tests

All patients underwent clinical assessment, electrocardiography, laboratory testing, arterial blood gas analysis, lactate measurement, and bladder-based assessment of intra-abdominal pressure in the supine position (12). Hemodynamic monitoring was established with PiCCO after standard transpulmonary thermodilution calibration (13).

During the same pre-fluid assessment window, six index tests were recorded before the reference fluid challenge. PLR was performed by moving the patient from a semirecumbent position to one with the trunk horizontal and the legs elevated to 45 degrees for approximately 1 minute. EEOT consisted of a 15-second end-expiratory hold. In the prespecified protocol, both PLR and EEOT were considered positive when they increased PiCCO-derived cardiac index by >15%.

SVV and PPV were obtained directly from PiCCO arterial waveform analysis using prespecified protocol thresholds >12% for each. IVC-DI was measured in M-mode from the subcostal long-axis view approximately 2 cm from the right atrium, with a prespecified positive threshold >18%. Respiratory ΔVTI was measured by pulsed-wave Doppler from the apical 5-chamber view, with a prespecified positive threshold >12%. These cutoffs were selected from prior validation literature and the prespecified study protocol. The testing sequence and reference classification are summarized in Figure 1.

Figure 1 STARD-style study flow and protocol for comparative assessment of six bedside dynamic tests and the reference fluid challenge. The study database included 40 eligible, consented, enrolled mechanically ventilated adults with shock. All enrolled patients completed the six index tests and the reference fluid challenge. The number of screened but non-enrolled patients was not retained in the study screening log. This figure is original and was created from the study workflow. EEOT, end-expiratory occlusion test; IVC-DI, inferior vena cava distensibility index; MAP, mean arterial pressure; PiCCO, pulse contour cardiac output monitoring; PLR, passive leg raising; PPV, pulse pressure variation; STARD, Standards for Reporting Diagnostic Accuracy Studies; SVV, stroke volume variation; ΔVTI, respiratory variation in left ventricular outflow tract velocity-time integral.

Performers of index tests were aware of the clinical eligibility information required for bedside measurement but were unaware of the post-fluid reference-standard classification because the reference fluid challenge had not yet been performed. The reference-standard classification was based solely on the predefined PiCCO-derived cardiac-index response and was not adjudicated using index-test results.

Reference standard

After all six index tests were completed, each patient received 500 mL isotonic saline over 20 minutes. The reference standard for fluid responsiveness within this protocol was a prespecified ≥15% increase in PiCCO-derived cardiac index after fluid administration. This reference standard was selected because it directly measured the protocol-defined hemodynamic response to a fluid challenge; no error-free biologic gold standard for fluid responsiveness exists. Stroke-volume-based reclassification was not part of the prespecified analysis, and this limitation is explicitly acknowledged below.

Handling of indeterminate and missing data

All enrolled patients completed the six index tests and the reference fluid challenge. No indeterminate index-test or reference-standard results occurred, and no missing index-test or reference-standard data were present in the diagnostic-accuracy study dataset.

Statistical analysis

Statistical analysis was performed with IBM SPSS Statistics Version 24.0 (IBM Corp., Armonk, NY, USA). Continuous variables are presented as mean ± standard deviation (SD), and categorical variables are presented as number (percentage). For the prespecified protocol thresholds, diagnostic performance was summarized as true positive, false positive, false negative, and true negative counts, together with sensitivity, specificity, and accuracy. Accuracy was defined as (true positives + true negatives)/total patients. Ninety-five percent confidence intervals (CIs) for sensitivity, specificity, and accuracy were estimated using the Wilson method. The target sample size was 40 patients as defined in the original single-center feasibility protocol; no formal sample-size calculation was performed for pairwise comparison of the six correlated index tests.

Exploratory receiver operating characteristic analysis was used to estimate study-derived thresholds and the corresponding areas under the curve (AUCs) with 95% CIs. Because these thresholds were generated from and evaluated within the same dataset, they were considered hypothesis-generating rather than definitive. Pairwise DeLong comparisons of AUCs and McNemar comparisons of paired binary classifications were not included in the original analysis plan. Given the small cohort and the 15 possible pairwise comparisons among six correlated tests, such post hoc analyses would be underpowered and vulnerable to multiplicity-related misinterpretation. Therefore, the results are presented as descriptive within-cohort performance profiles rather than statistically ranked superiority claims. An exploratory Pearson correlation matrix for the continuous index-test values and the post-fluid cardiac-index response is provided as Supplementary Table S1. No prespecified subgroup, variability, inter-test agreement, or combined-test diagnostic model was planned.

Ethical consideration

The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Research Ethics Committee of the Faculty of Medicine, Cairo University (Kasr Al Ainy), Cairo, Egypt (approval No. md-137-2020; approval/response date 6 December 2020). Written informed consent was obtained from all participants and/or their legally authorized representatives prior to enrollment.

Results

The study database included 40 eligible, consented, consecutively enrolled mechanically ventilated adults with shock. All 40 patients completed the six pre-fluid index tests and underwent the reference 500-mL fluid challenge. No enrolled patient was excluded after enrollment, no index-test or reference-standard result was missing or indeterminate, and no adverse event related to the index tests or reference fluid challenge was reported in the study database.

Mean age was 59.47±12.2 years (range, 25–85 years). Twenty-one patients (52.5%) were women and 19 (47.5%) were men. Septic shock was the most common presentation (21/40, 52.5%), followed by mixed septic-cardiogenic shock (16/40, 40.0%) and cardiogenic shock (3/40, 7.5%) (Table 1).

By the reference fluid challenge, 25 of 40 patients (62.5%) were fluid responders and 15 of 40 (37.5%) were nonresponders. Because the target condition was a physiological response to fluid challenge rather than a disease entity, formal disease-severity categories and alternative diagnoses for nonresponders were not applicable; clinical shock phenotypes for the cohort are reported in Table 1.

Diagnostic performance at the protocol-defined thresholds is summarized in Table 2. PLR and IVC-DI had the highest overall accuracy (75.0% each), but their operating characteristics differed. PLR correctly classified 30 of 40 patients (true positive, 20; true negative, 10) and favored sensitivity over specificity (80.0% vs. 66.7%). IVC-DI also classified 30 of 40 patients correctly (true positive, 18; true negative, 12) and showed a more specific profile (sensitivity, 72.0%; specificity, 80.0%). Classification counts for all tests are displayed in Figure 2, and the corresponding sensitivity, specificity, and accuracy are shown in Figure 3.

Figure 2 Classification outcomes for each dynamic test at the prespecified protocol thresholds. The stacked bars show true positives, false positives, false negatives, and true negatives for PLR, EEOT, SVV, PPV, IVC-DI, and respiratory ΔVTI, using the standardized post-fluid cardiac-index response as the reference standard. EEOT, end-expiratory occlusion test; FN, false negative; FP, false positive; IVC-DI, inferior vena cava distensibility index; PLR, passive leg raising; PPV, pulse pressure variation; SVV, stroke volume variation; TN, true negative; TP, true positive; ΔVTI, respiratory variation in left ventricular outflow tract velocity-time integral.

Figure 3 Protocol-threshold sensitivity, specificity, and accuracy for the six dynamic tests. Values are percentages and correspond to the diagnostic-classification data reported in Table 2. EEOT, end-expiratory occlusion test; IVC-DI, inferior vena cava distensibility index; PLR, passive leg raising; PPV, pulse pressure variation; SVV, stroke volume variation; ΔVTI, respiratory variation in left ventricular outflow tract velocity-time integral.

EEOT correctly classified 29 of 40 patients (accuracy, 72.5%) and, like IVC-DI, favored specificity over sensitivity. SVV matched PLR sensitivity (80.0%) but had lower specificity (53.3%), producing more false-positive classifications. PPV and respiratory ΔVTI were less discriminative overall, with accuracies of 65.0% and 57.5%, respectively.

Exploratory receiver operating characteristic analysis using study-derived cutoffs yielded the highest AUC for PLR (0.815; 95% CI, 0.676–0.954), followed by SVV (0.775; 95% CI, 0.628–0.921) and PPV (0.743; 95% CI, 0.582–0.903). The ROC curves are shown in Figure 4. Derived in-sample cutoffs were 14.0% for PLR, 14.5% for EEOT, 11.0% for SVV and PPV, 16.0% for IVC-DI, and 18.0% for respiratory ΔVTI (Table 3). Sensitivity and specificity at derived cutoffs are not emphasized because these thresholds were generated and evaluated in the same small dataset.

Figure 4 Exploratory receiver operating characteristic curves for the six dynamic tests. Areas under the curve and study-derived cutoffs are reported in Table 3. Because the cutoffs were generated and evaluated in the same 40-patient dataset, these ROC analyses should be interpreted as exploratory. EEOT, end-expiratory occlusion test; IVC-DI, inferior vena cava distensibility index; PLR, passive leg raising; PPV, pulse pressure variation; ROC, receiver operating characteristic; SVV, stroke volume variation; ΔVTI, respiratory variation in left ventricular outflow tract velocity-time integral.

Discussion

In this single-center cohort of sedated, mechanically ventilated adults with shock, PLR provided the most consistent overall bedside estimate of fluid responsiveness when six dynamic tests were applied under the same controlled conditions. IVC-DI and SVV were useful complementary tests, but their strengths differed: IVC-DI offered greater specificity, whereas SVV and PLR were more sensitive. EEOT, PPV, and especially respiratory ΔVTI were more variable in this dataset.

The finding that PLR performed best is physiologically plausible and consistent with prior literature. Because PLR functions as a reversible autotransfusion, it directly challenges preload reserve without giving fluid and remains one of the most reproducible dynamic maneuvers across studies (5,7,14). The AUC for PLR in this cohort (0.815) was lower than some published estimates. This may reflect the small sample size, mixed shock phenotypes, use of a cardiac-index rather than stroke-volume reference definition, and the strict protocol threshold used for test positivity. These factors can narrow the range of observed responses and reduce apparent discrimination.

IVC-DI and SVV likely benefited from the deliberate physiologic selection of our cohort. Both tests perform best in patients with controlled mechanical ventilation, sinus rhythm, and absent spontaneous effort, which were imposed by the protocol (5,9,10,15,16). In practice, however, their applicability is narrower than that of PLR. SVV in particular should not be used in isolation to justify additional fluid because lower specificity means that false-positive classifications remain clinically relevant.

EEOT showed moderate protocol-threshold accuracy but a lower exploratory AUC. This should not be interpreted as a contradiction. Protocol-threshold accuracy describes performance at one prespecified cutoff, whereas the AUC reflects discrimination across the available range of values. In a small dataset, a test may perform reasonably at a clinically chosen cutoff yet have limited global discrimination, especially when values cluster around the threshold. For this reason, we removed emphasis on sensitivity and specificity at study-derived cutoffs and interpret the exploratory ROC findings cautiously. Prior systematic and echocardiographic evidence supports the use of EEOT to predict fluid responsiveness in appropriately selected mechanically ventilated patients (17,18).

Respiratory ΔVTI was the weakest test in this study. The most plausible explanation is methodological rather than purely physiologic. Echocardiographic dynamic measurements are attractive because they are noninvasive, but their reliability depends heavily on image quality, operator expertise, and the ability to detect small beat-to-beat changes in flow. In ventilated ICU patients, narrow Doppler envelopes or poor acoustic windows can easily translate into misclassification when diagnostic thresholds are small (11,19,20).

The bedside implication is not that one test should replace all others. The diversity of test profiles can be clinically useful. A more sensitive maneuver such as PLR may be preferred when the cost of missing fluid responsiveness is high and the patient is likely to tolerate additional fluid. A more specific profile, such as that observed with IVC-DI in this dataset, may be attractive when fluid overload would be particularly harmful. In practice, the decision to administer fluid should integrate dynamic-test results with tissue perfusion, fluid tolerance, right ventricular function, ventilatory conditions, and overall resuscitation goals. A patient may be fluid responsive and still not be a good candidate for more fluid.

Limitations

This study has several limitations. It was a small single-center investigation and should be interpreted as hypothesis-generating. The cohort was intentionally narrow: all patients were fully sedated, receiving controlled ventilation with tidal volume ≥8 mL/kg, in sinus rhythm, with normal intra-abdominal pressure and an intact thorax. The results therefore cannot be generalized to the broader ICU population, especially patients with spontaneous breathing, arrhythmia, low-tidal-volume ventilation, acute respiratory distress syndrome, elevated intra-abdominal pressure, or PEEP settings outside the range encountered in this cohort.

Shock phenotypes were mixed, but the study was not powered for subgroup analyses, and only three patients had isolated cardiogenic shock. The number of screened but non-enrolled patients was not retained in the study screening log and could not be reliably reconstructed retrospectively, which limits assessment of pre-enrollment selection flow. Several clinically important baseline and paired hemodynamic variables, including SOFA score, prior fluid balance, vasopressor and inotropic dose, PEEP range, central venous pressure, stroke volume, systemic vascular resistance, and left ventricular ejection fraction, were not captured in the study analysis file. This limits clinical characterization and assessment of generalizability.

The reference standard was based on PiCCO-derived cardiac index rather than stroke volume. Although cardiac index is clinically relevant for global flow, many prior validation studies used stroke-volume response. Stroke-volume-based reclassification was not part of the prespecified analysis. Echocardiographic measurements were operator dependent, and interobserver reproducibility was not assessed. The study was not designed or powered for formal pairwise superiority testing between the six correlated index tests using DeLong or McNemar methods, and such analyses were not prespecified. Thus, comparisons among tests should be interpreted as descriptive performance profiles rather than definitive statistical rankings. Finally, study-derived ROC cutoffs were generated and evaluated in the same dataset and should not be adopted without external validation.

Conclusions

In mechanically ventilated adults with shock studied under strict applicability conditions, PLR provided the most consistent overall bedside estimate of fluid responsiveness. IVC-DI and SVV were useful complementary tools with different sensitivity-specificity profiles, whereas EEOT, PPV, and respiratory ΔVTI showed more variable performance in this cohort. Exploratory correlations supported partial overlap among several tests, especially SVV and PPV, but did not justify ranking one strategy as statistically superior. Larger multicenter diagnostic accuracy studies using harmonized cardiac-index and stroke-volume definitions, formal paired comparisons, inter-test agreement analyses, and external validation of thresholds are needed before comparative rankings can be generalized.

Acknowledgments

We thank the nursing and ICU teams at Cairo University (Kasr Al Ainy) for their support in protocol implementation and bedside monitoring.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-2026-0015/rc

Data Sharing Statement: Available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-2026-0015/dss

Peer Review File: Available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-2026-0015/prf

Funding: None.

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://jeccm.amegroups.com/article/view/10.21037/jeccm-2026-0015/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Research Ethics Committee of the Faculty of Medicine, Cairo University (Kasr Al Ainy), Cairo, Egypt (approval No. md-137-2020; approval/response date 6 December 2020). Written informed consent was obtained from all participants and/or their legally authorized representatives prior to enrollment.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Vincent JL, De Backer D. Circulatory shock. N Engl J Med 2013;369:1726-34. [Crossref] [PubMed]
2. Evans L, Rhodes A, Alhazzani W, et al. Surviving sepsis campaign: international guidelines for management of sepsis and septic shock 2021. Intensive Care Med 2021;47:1181-247. [Crossref] [PubMed]
3. Boyd JH, Forbes J, Nakada TA, et al. Fluid resuscitation in septic shock: a positive fluid balance and elevated central venous pressure are associated with increased mortality. Crit Care Med 2011;39:259-65. [Crossref] [PubMed]
4. Marik PE, Monnet X, Teboul JL. Hemodynamic parameters to guide fluid therapy. Ann Intensive Care 2011;1:1. [Crossref] [PubMed]
5. Monnet X, Marik PE, Teboul JL. Prediction of fluid responsiveness: an update. Ann Intensive Care 2016;6:111. [Crossref] [PubMed]
6. Cecconi M, De Backer D, Antonelli M, et al. Consensus on circulatory shock and hemodynamic monitoring. Task force of the European Society of Intensive Care Medicine. Intensive Care Med 2014;40:1795-815.
7. Monnet X, Teboul JL. Passive leg raising: five rules, not a drop of fluid! Crit Care 2015;19:18. [Crossref] [PubMed]
8. Monnet X, Osman D, Ridel C, et al. Predicting volume responsiveness by using the end-expiratory occlusion in mechanically ventilated intensive care unit patients. Crit Care Med 2009;37:951-6. [Crossref] [PubMed]
9. Marik PE, Cavallazzi R, Vasu T, et al. Dynamic changes in arterial waveform derived variables and fluid responsiveness in mechanically ventilated patients: a systematic review of the literature. Crit Care Med 2009;37:2642-7. [Crossref] [PubMed]
10. Barbier C, Loubières Y, Schmit C, et al. Respiratory changes in inferior vena cava diameter are helpful in predicting fluid responsiveness in ventilated septic patients. Intensive Care Med 2004;30:1740-6. [Crossref] [PubMed]
11. Feissel M, Michard F, Mangin I, et al. Respiratory changes in aortic blood velocity as an indicator of fluid responsiveness in ventilated patients with septic shock. Chest 2001;119:867-73. [Crossref] [PubMed]
12. Malbrain ML. Different techniques to measure intra-abdominal pressure (IAP): time for a critical re-appraisal. Intensive Care Med 2004;30:357-71. [Crossref] [PubMed]
13. Sakka SG, Reuter DA, Perel A. The transpulmonary thermodilution technique. J Clin Monit Comput 2012;26:347-53. [Crossref] [PubMed]
14. Cavallaro F, Sandroni C, Marano C, et al. Diagnostic accuracy of passive leg raising for prediction of fluid responsiveness in adults: systematic review and meta-analysis of clinical studies. Intensive Care Med 2010;36:1475-83. [Crossref] [PubMed]
15. Yang X, Du B. Does pulse pressure variation predict fluid responsiveness in critically ill patients? A systematic review and meta-analysis. Crit Care 2014;18:650.
16. Long E, Oakley E, Duke T, et al. Does Respiratory Variation in Inferior Vena Cava Diameter Predict Fluid Responsiveness: A Systematic Review and Meta-Analysis. Shock 2017;47:550-9. [Crossref] [PubMed]
17. Si X, Song X, Lin Q, et al. Does End-Expiratory Occlusion Test Predict Fluid Responsiveness in Mechanically Ventilated Patients? A Systematic Review and Meta-Analysis. Shock 2020;54:751-60.
18. Georges D, de Courson H, Lanchon R, et al. End-expiratory occlusion maneuver to predict fluid responsiveness in the intensive care unit: an echocardiographic study. Crit Care 2018;22:32. [Crossref] [PubMed]
19. Jozwiak M, Mercado P, Teboul JL, et al. What is the lowest change in cardiac output that transthoracic echocardiography can detect? Crit Care 2019;23:116. [Crossref] [PubMed]
20. Wang J, Zhou D, Gao Y, et al. Effect of VTILVOT variation rate on the assessment of fluid responsiveness in septic shock patients. Medicine (Baltimore) 2020;99:e22702. [Crossref] [PubMed]