Table 2 General characteristics of included studies

Corl³⁸

2012

USA

Single centre, academic ED

Diagnostic

To determine the accuracy of the caval index to detect fluid responsiveness

4

• High risk of bias

• Convenience sample, unclear description of population

• Gold standard (TEB) questionable

• Patients were excluded from analysis due to incomplete data

Jung⁴²

2012

Korea

Single centre, academic ED

Diagnostic

To determine the validity of corrected flow to predict fluid responsiveness

4

• High risk of bias

• Not presented as a diagnostic test study

• Unclear patient selection method

• No blinding to reference test measurements.

• No pre-specified threshold

Feissel⁴⁰

2013

France

Single centre, ED

Diagnostic

To determine whether plethysmographic variability index can predict

fluid responsiveness

3

• High risk of bias

• Not presented as a diagnostic test study

• Unclear patient selection method.

• Inappropriate exclusions

• Patients were excluded from analysis due to incomplete data

• No pre-specified threshold

Coen³⁷

2014

Italy

Single centre, major metropolitan ED

Treatment

To investigate the reliability of caval index and lung ultrasound to guide fluid infusion

4

• High risk of bias

• No control group

• No sample size calculation

de Valk³⁹

2014

Netherlands

Single centre, academic ED

Diagnostic

To investigate the reliability of caval index to predict fluid responsiveness

4

• High risk of bias

• Convenience sample

• No pre-specified threshold.

• Gold standard test questionable

• No blinding to index test measurements

• Patients were excluded from analysis due to incomplete data

• No pre-specified threshold

Duus⁴¹

2015

USA

Single centre, adult ED

Diagnostic

To determine the reproducibility of PLR and fluid bolus monitored by bioreactance in predicting fluid responsiveness

3

• High risk of bias

• Observational

• Convenience sample.

• Consecutive measures may influence each other

• No blinding to index test measurements

Hou³⁶

2016

USA

Multi centre

Treatment

(RCT)

To evaluate the impact of a fluid responsiveness protocol in decreasing organ failure

4

• Lack of blinding

• Relatively few patients per center

• Only collected 10% of planned sample. Very likely Type II error

• Downgraded due to imprecision

Kuan³⁵

2016

Singapore

Single centre, academic ED

Treatment

(RCT)

To Evaluate a non-invasive heamodynamic algorithm compared to standard care

2

• Lack of blinding

• Underpowered: high difference in primary outcome – ARR 25%. RRR approx. 40%