Inferior Vena Cava collapsibility detects fluid responsiveness among spontaneously breathing critically ill patients

Corl KA. Journal of Critical care 2017; 41: 130-137. DOI: 10.1016/j.jcrc.2017.05.008

Clinical Question

• In spontaneously breathing critically unwell patients undergoing resuscitation, does the collapsibility of the inferior vena cava (cIVC), when compared to administering a fluid bolus predict a >10% increase in cardiac output?

Background

• IV Fluid resuscitation remains one of the cornerstones of intensive care practice. However over-resuscitation has been associated with a worse outcome in critically unwell patients
• Conservative fluid strategies, where fluid is given only if the circulation requires it, have been designed around the concept of “fluid responsiveness”. This involves giving fluids only if measures suggest that the cardiac output will increase by a preset value following the fluid bolus
• Multiple methods exist to assess fluid responsiveness- static (e.g. CVP, PCWP) and dynamic (PPV, SVV, Aortic VTI, and IVC collapsibility)
• Assessing the collapsibility (or distensibility if the patient is mechanically ventilated) of the IVC during respiration using point of care ultrasound (POCUS) has been recommended by some clinicians as a means to assess fluid responsiveness, however this is controversial

Design

• Prospective observational study
• “Convenience” sample of spontaneously breathing patients with acute circulatory failure being admitted to the ICU
• Ultrasound were reviewed after they were performed by a reviewer blinded to the NICOM results
• intra- and inter-observer variability of IVC compared using ANOVA:
  • Intra-observer correlation coefficient 0.92 (95% CI: 0.89-0.95)
  • Inter-observer correlation coefficient 0.67 (95% CI: 0.56-0.76)
• ROC analysis used to compare baseline cIVC’s ability to predict fluid responsiveness
• Power calculation: A sample size of 124 patients needed to detect a difference between AUC for cIVC of at least 0.88 and an AUC of 0.70 (90% power, one sided alpha 0.05)

Setting

• 2 adult academic hospitals in the USA
• Medical ICU and emergency departments
• August 2014-July 2016

Population

• Inclusion:
  • Patients with acute circulatory failure, defined as:
  • 1- hypotension (a MAP <65mmHg or a systolic BP <90mmHg for >30 minutes)
  • 2- Decreased urine output (<0.5ml/kg/hr)
  • 3- Persistent tachycardia (HR >120 bpm for >30 minutes)
  • And/or- serum markers suggesting organ hypo perfusion (acidosis with a pH <7.3 or lactate >2meq/l)
• Exclusion:
  • Primary traumatic, cariogenic, obstructive or neurogenic shock
  • Age <18
  • Incarceration
  • Pregnancy
  • Hospitalisation for >36 hrs
  • Receiving non-invasive ventilation
  • Clinical impression of active pulmonary oedema
  • Clinical impression of further IV fluid administration posing clinical risk
• 287 screened, 154 consented, 124 included in final analysis
• When groups classified by fluid responsiveness, baseline characteristics mostly similar
• 11 COPD patients in “responder” group vs 1 in “non-responder group”
•  Average of 4L fluid given in each group before enrolment

Intervention

• IVC POCUS performed by one of 3 study physicians who had been trained in ultrasound
• Performed in subcostal long axis, 2D B-mode, at junction of RA and IVC, using low frequency phased array probe
• Measurements performed throughout native respiratory cycle
• Measurements taken 3cm caudal to IVC/RA junction. Maximum and minimum diameter identified through visual inspection
• 2 baseline ten-second measurements of IVC recorded 1 minute apart
• 3 minute passive leg raise performed; a third ten-second measurement of IVC recorded
• 500ml 0.9% saline bolus administered through participant’s largest gauge IV
• Final IVC measurement performed on completion of fluid bolus

Control

• In all patients Cheetah NICOM devices attached and calibrated for 3 mins at start of study
• Cardiac index recorded at one minute intervals throughout study
• Passive leg raise and fluid bolus performed as per IVC group

Management common to both groups

• If patients were on vasopressors, rate was held constant throughout study

Outcome

• 61 (49.2%) fluid responders
• Primary outcome: Discriminating ability of cIVC to detect fluid response: AUC 0.84 (95% CI: 0.76-0.91)
  • Optimal cut off for discrimination was cIVC of 24.6% (Sensitivity 88.5% w/ 95% CI 77.8-95.3; Specificity 81.0% with 95% CI 69.1-89.1)
  • +ve likelihood ratio of cIVC 25% 4.56 (95% CI 2.72-7.66)
  • -ve likelihood ratio of cIVC 25% 0.16 (95% CI 0.08- 0.31)
• Secondary outcomes:
  • Comparison of optimal cutoff of cIVC in this study to previous studies, using cIVC of 25% resulted in 16.1% misclassification rate, as opposed to 40% or 42% (34.7% misclassification rate, and 36.3% misclassification rate respectively)
  • Incorporating PLR with cIVC did not improve discriminatory characteristics

Authors’ Conclusions

• cIVC as measured by POCUS is able to detect fluid responsiveness and may be used to guide IV Fluid resuscitation among spontaneously breathing critically ill patients.

Strengths

• Real-world pragmatic design
• Standardised protocol for cIVC assessment
• Blinding of cIVC assessors
• Cardiac output monitor used to assess fluid responsiveness

Weaknesses

• Not randomised
• “Convenience” sampling resulted in only 125 patients being randomised over two years
• NICOM has been shown to be inaccurate in tracking CO in critically unwell and in tracking improvements with CO using PLR (https://academic.oup.com/bja/article/111/6/961/292109/Bioreactance-is-not-reliable-for-estimating)- this trial may well be comparing cIVC to an inaccurate monitor
• The flaws in NICOM could be why 11/12 of the COPD patients in the trial were “fluid responsive”- by relying on trans thoracic impedance large swings in intrathoracic pressure could skew the cardiac output results
• Inter-observer variability of cIVC demonstrated to be 0.67 between 3 trained clinicians; this seems rather low for an “accurate” test
• No long term, or patient focused, outcomes recorded- I would have liked to know if giving fluids was actually the right thing to do if they were fluid responsive
• The majority (50%) of fluid responders had septic shock, which is arguably not a hypovolaemic state
• The median amount of fluid given before enrolment was 4 litres over a median period of 16 hours- many clinicians are assessing fluid responsiveness much earlier than this, limiting the study’s external validity
• No mention of absolute IVC diameter- many clinicians use this as a guide to fluid responsiveness as well
• cIVC measurement in a single plane is not recommended by some experts- the IVC may be more complex than just a straightforward cylinder

The Bottom Line

• The lack of randomisation, lack of comparison to a true gold standard cardiac monitor, unavailability of patient-centred outcomes, and the use of a flawed method of measuring the IVC means that I cannot recommend cIVC in isolation as a measure of fluid responsiveness on the basis of this study
• I will cautiously and contentiously continue to use cIVC as part of my focused echo assessment, and will consider it in conjunction with other clinical parameters when deciding to give IV fluids

External Links

• [article] Corl KA et al. Inferior Vena Cava collapsibility detects fluid responsiveness among spontaneously breathing critically ill patients
• [Podcast] Justin Bowra- IVC Filling. The ultimate Myth
• [Blog] Marik PE. Vena Caval Ultrasound- Just don’t do it!
• [Blog] Rola P. IVC Assessment- let’s apply some common sense

Metadata

Summary author: Segun Olusanya
Summary date: 19th July 2017
Peer-review editor: Adrian Wong