|
Diagnostic accuracy of passive leg raising for prediction of fluid responsiveness in adults: systematic review and meta-analysis of clinical studies |
Cavallaro F, Sandroni C, Marano C, La Torre G, Mannocci A, De Waure C, Bello G, Maviglia R, Antonelli M |
|
|
CRD summary This review concluded that changes in cardiac output induced by passive leg raising could reliably predict fluid responsiveness regardless of ventilation mode, underlying cardiac rhythm, and technique of measurement. Weaknesses in the review process and unclear data quality mean that this conclusion should be interpreted cautiously. Authors' objectives To assess the ability to predict fluid responsiveness from changes in cardiac output and changes in arterial pulse pressure, induced by passive leg raising (PLR). Searching MEDLINE, EMBASE, and the Cochrane Database of Systematic Reviews were searched; search terms were reported, but the dates were not. The bibliographies of included studies were screened for additional articles, and the MEDLINE alert system was used to identify studies published during the review process. No language restrictions were applied, but only studies published or accepted for publication as full-text articles were included. Study selection Studies assessing the ability of haemodynamic response to PLR (the index test), in adults, to predict an increase in cardiac output after a subsequent fluid infusion, were eligible for inclusion. Included studies had to report one of the following: the sensitivity and specificity of the index test in identifying patients who subsequently responded to fluids (responders); the difference in the mean value of the index test between responders and non-responders; the correlation between haemodynamic response to PLR and the increase in cardiac output after fluid infusion.
The authors did not state how many reviewers assessed studies for inclusion.
All the included studies were conducted in intensive care units, in patients who were in shock (various causes). The threshold increase in cardiac output used to define a responder ranged from 12% to 15%. Most studies executed PLR from a semi-recumbent position, and the time between PLR and index test varied from one to five minutes. Assessment of study quality Methodological quality was assessed using the 14-item Quality Assessment of Diagnostic Accuracy Studies (QUADAS) tool, and an overall quality score, with a maximum of 14, was calculated.
The authors did not state how many reviewers assessed the quality. Data extraction Two authors independently extracted the following data, where reported: the number and percentage of responders; the mean and standard deviation of the index test measure in responders and non-responders; the correlation coefficient (Spearman or Pearson) between the index test value and cardiac output after fluids; and the sensitivity, specificity, best diagnostic threshold and area under the receiver operating characteristic curve (AUC).
Study authors were contacted for additional data, where needed. Methods of synthesis Separate analyses were conducted for PLR-induced changes in cardiac output and PLR-induced changes in pulse pressure. A summary receiver operating characteristic (SROC) curve was drawn using the Moses and Littenberg model. The sensitivity, specificity, diagnostic odds ratio and AUC values from included studies were pooled using random-effects models.
Pooled estimates were calculated for the difference in the mean value of the index test between responders and non-responders, and the correlation between haemodynamic response to PLR and the increase in cardiac output after fluid load.
All values were reported as point estimates with 95% confidence intervals. The unit of analysis was a fluid bolus, rather than the patient number.
Between-study heterogeneity was assessed using Cochran Q and I2 tests, and Spearman correlation between sensitivity and specificity was used to assess the threshold effect. Subgroup analyses were used to compare the index test performance in: patients with sinus rhythm and those with arrhythmias; patients adapted to ventilator and those with inspiratory efforts; and studies where PLR was performed from a semi-recumbent position and those where it was performed from a supine position. Results of the review Nine studies (353 patients and 366 fluid boli) were included in the review. The sample size ranged from 15 to 89 participants. The median QUADAS score was 13 (range 12 to 14).
Changes in cardiac output: In eight studies, with 351 boli, the pooled estimate for sensitivity was 89.4% (95% CI 84.1 to 93.4) and for specificity was 91.4% (95% CI 85.9 to 95.2). The pooled diagnostic odds ratio was 89.0 (95% CI 40.2 to 197.3) and the AUC was 0.95 (95% CI 0.92 to 0.97). Heterogeneity was not significant and no significant differences were found between any of the subgroups.
Changes in pulse pressure: In three studies, with 139 boli, the pooled estimate for sensitivity was 59.5% (95% CI 47.4 to 70.7) and for specificity was 86.2% (95% CI 75.3 to 93.5). The pooled diagnostic odds ratio was 10.8 (95% CI 4.4 to 26.1) and the AUC was 0.76 (95% CI 0.67 to 0.86). Heterogeneity was not significant.
The pooled correlation coefficients and pooled mean difference (index test) between responders and non-responders were reported. Authors' conclusions PLR-induced changes in cardiac output reliably predicted the fluid responsiveness regardless of ventilation mode, underlying cardiac rhythm, and technique of measurement. PLR-induced changes in pulse pressure could be a viable alternative, but had a lower predictive ability. CRD commentary The review provided a clear objective and defined inclusion criteria. A range of sources were searched for relevant studies and no language restrictions were applied, but the restriction to published studies means that publication bias cannot be ruled out. The data were extracted by two reviewers to minimise the potential for error and bias, but it was unclear whether similar precautions were applied to study selection and quality assessment. The results of the quality assessment were reported as a summary score only, which is not recommended by QUADAS guidelines and the reliability of the included study results cannot be fully assessed. The methods used to pool studies were broadly appropriate, but other SROC models (hierarchical or bivariate) are now generally recommended.
The authors' conclusions reflected the data presented, but were based on a relatively few, small studies. Weaknesses in the review process and unclear data quality mean that these conclusions should be interpreted cautiously. Implications of the review for practice and research Practice: The authors stated that PLR-induced changes in cardiac output could be recommended for the routine assessment of fluid responsiveness in most patients in intensive care units.
Research: The authors did not provide recommendations for future research. Bibliographic details Cavallaro F, Sandroni C, Marano C, La Torre G, Mannocci A, De Waure C, Bello G, Maviglia R, Antonelli M. Diagnostic accuracy of passive leg raising for prediction of fluid responsiveness in adults: systematic review and meta-analysis of clinical studies. Intensive Care Medicine 2010; 36(9): 1475-1483 Indexing Status Subject indexing assigned by NLM MeSH Adult; Aged; Aged, 80 and over; Animals; Blood Pressure; Central Venous Pressure; Confidence Intervals; Critical Illness /therapy; Fluid Therapy /methods; Hemodynamics; Humans; Leg /blood supply; Middle Aged; Odds Ratio; Respiration, Artificial /methods; Supine Position; Treatment Outcome AccessionNumber 12010006381 Date bibliographic record published 17/11/2010 Date abstract record published 11/05/2011 Record Status This is a critical abstract of a systematic review that meets the criteria for inclusion on DARE. Each critical abstract contains a brief summary of the review methods, results and conclusions followed by a detailed critical assessment on the reliability of the review and the conclusions drawn. |
|
|
|