ECCO2R for NIV failure – ready for a proper look

ECCO2R NIVWe rightly like to avoid intubation in COPD exacerbations. Extracorporeal CO2 removal or ‘pulmonary dialysis’  has been around for a while and many feel it is beneficial in COPD (eg this case) but who is it best used for?

In an Italian matched cohort study Del Sorbo and colleagues looked at 200 patients with an exacerbation of COPD treated with NIV (up to 90yrs old!).




They focused specifically on the subset of 89 patients at risk of NIV failure, defined as:

  • 2 hours of continuous NIV
  • arterial pH less < 7.30
  • Paco2 greater than 20% of baseline
  • respiratory rate >30 or use of accessory muscles or paradoxical abdominal/chest movement.

However only 25 of the 89 ‘at risk’ patients intended for ECCO2R actually ended up with it, largely due to refusal of consent. In these 25 femoral vein (14F double lumen) pumped VV ECCO2R (?a converted haemofilter!) was popped in and heparinization commenced. (ECCO2R v ECMO  diagram for the uninitiated)

If after 2 hours they had a pH <7.25, rising oxygen requirement, were haemodynamically unstable or comatose they would be intubated.

The NIV-only control group was taken from 2 previous studies and propensity matched for ‘at risk’ criteria as well as other sensible parameters (APACHE, prior respiratory function, age etc). 21 matched patients were found.

 Primary outcome measure was intubation by 28 days.


3 of the ECCO2R ‘v’ 7 of the standard NIV group were intubated, ie RRR of over 70% and an ARR of 21%. Of course this was not statistically significant.

The ECCO2R group had better pH, CO2, oxygenation and respiratory rate.

But half the patients had adverse events. These were mainly circuit issues but haemorrhage was a problem in 4 of the 25.

There was a striking trend towards mortality 8% for ECCO2R,  ‘v’ 35% for NIV alone. Fewer ICU days too.

So, potentially useful in 1 in 8 COPD exacerbations requiring NIV, ECCO2R is not without risk but may avoid intubation or even be life saving. This now needs a closer look. Equipoise was clearly an issue in this study, and now surely shouldn’t be? Parallel effort to improve prediction of NIV failure is probably also warranted. 


ICU echo myths! Mayo et al both enthuse and warn.

Echo myths MayoMayo et al outline the misinformation regarding echocardiography use in the ICU. Some additional comments have been added!

1. Every echo exam should be comprehensive – False

Focused assessments are valuable. We are gathering evidence about powerful but easily achieved echo findings.

But stick to the protocol you’re competent with.

2. Echo is not a monitoring tool – False

Intermittent serial quantitative or qualitative measurements are often adequate for assessing trends and guiding therapy.

Serial assessment can be powerful, but this does come at the cost of being relatively labour intensive.

3. Intensivists do not need cardiologists – False

The echo skill sets of ICU doctors and cardiologists is diverging and an experienced echo-intensivist may be more familiar with assessing heart-lung interaction but the cardiologist will always be vital for many diagnoses and cardiac disease management plans.

Ability to acquire images does not imply ability to interpret images.

4. Echo is less reliable than other haemodynamic tools – FALSE

Every monitoring device has its limitations but there are occasions when echo  may be the more accurate (eg in the very unstable patient, active haemorrhage or anatomical abnormalities).

In addition no other modality gives rapid qualitative assessment of the heart.

5. In refractory or complicated shock, echo should be replaced by an alternative device – FALSE

Echo is an excellent diagnostic device in complex instability. Shock aetiology, fluid responsiveness, LV and RV systolic and diastolic function, preload/afterload, RV-ventilator interaction can all be addressed.

The roles of traditional monitoring and echo overlap. Often information really is power.

6. Training in transoesophageal echo is an optional part of advanced echo training – FALSE

This statement may depend on your definition of ‘advanced’ but some questions are very difficult to answer with TTE. Great vessels, some valves and intracardiac shunt (or even PAC placement) for example are better assessed with TOE.

As an advanced practitioner sometimes you’ll be frustrated you don’t have a TOE probe.

7. It’s easy to achieve competence in echo – FALSE

Adequate training is vital. In training you’ll learn as many limitations as applications. Some will be competent to accurately answer simple questions after 50 scans. There must be structured training and a means to maintain competency.

Echo images must be interpreted as meticulously as possible, potential confounders must be appreciated, and the findings put in clinical context.

8. There is a strong evidence base that supports echo – FALSE
9. Echo is academically supported worldwide – FALSE

Echo has been taken up worldwide due to its ‘intuitive obvious utility’ but an outcome benefit is yet to be demonstrated. There is a relatively small group of experienced enthusiasts who publish the bulk of the work in this area. The rise of echo despite this may also prove problematic when looking for equipoise in designing a study.

Do we even know the normal echo dataset for a ventilated patient or the expected effects of inotropes on echo values? We need to gather observational data to begin to generate hypotheses. Get studying!

10. There is no risk of medico-legal issues – FALSE

Train, stay competent and know your limits when interpreting findings.

Documentation about your focused assessment should emphasise that it was asking a specific question so that it is not later assumed that the exam was comprehensive. The patient has had a focused cardiac examination and not a formal echo.

 The multiple qualifications/accreditations are converging quickly and national standards are beginning to appear. It’s not acceptable to ‘dabble’ in echo!

E-CPR for all – just a dream, or the future?

CHEER ecmo tableA new meta-analysis of ECMO in cardiogenic shock and arrest showed an impressive 30-day survival rate of over 30% but at the cost of significant levels of complications (neurological/infective/renal).

A couple of months ago CHEER trial team published their results with a protocol that reads like science-fiction for many of us:

A senior team leader, 2 crit care doctors ready with femoral cannulae, another manning the IVC ultrasound, an ECMO nurse to run the mechanical CPR device and initiate the V-A ECMO circuit, and someone else to start the iced-saline.

Autopulse mechanical chest-compressor and cold infusion en-route to hospital. On arrival compressions are briefly paused while femoral cannulae go in (guidewire visualized in the IVC). Chest X-ray, heparinization then ECMO initiation. Coronary occlusion or pulmonary embolism is then addressed by PCI or thrombectomy. Cool for 24 hours and wean the ECMO using echo.

In-hospital as well as out of hospital arrests were included in this Melbourne single hospital prospective observational study. 24 of the intended 26 patients actually got ECMO. The arrests were largely VF and the aetiology was mainly ischemic but cardiomyopathies/channelopathies, PE and respiratory causes also featured. Fairly standard mix!

14 of the 26 got out of hospital. Unsurprisingly more in-hospital than out-of-hospital arrests survived (9/15 v 5/11). None survived asystole despite E-CPR. From arrest to starting ECMO  40 mins (median) passed in survivors and 78 mins in non-survivors (seems to take 15-30 mins from arrival of team to ECMO initiaition).

A highly-staffed process gives good ROSC and survival rates. Refine it by restricting to non-asystole and those who can get cannulated asap.

Given this protocol will never be achievable in smaller hospitals, is it more evidence for centralizing arrest management, or does it really emphasize that the priorities are:

  • shortening the time from arrest to arrest team
  • prompt definitive management thereafter.



Feed, or wait and eat, in pancreatitis

pancreas nutition pythonAnother addition to this year’s flurry of nutrition work.

Tradition has us bypassing the stomach and duodenum for nutrition in pancreatitis – intuitively sensible. To challenge this the Dutch PYTHON trial group has recently published its findings.

Multicentredly, they randomized 200-ish patients with acute ‘high-risk’ pancreatitis (defined as mGlasgow of >/= 3, CRP >150, APACHE >/=8) to start naso-jejunal feeding (25Kcal/Kg/dy if on ICU) within 24 hours or to wait up to 3 days and then encourage (the ‘on-demand’ group). Powered to detect 5% difference in major infection (pancreas or other) OR death within 6 months. The protocol was pretty well-adhered-to.

There was no difference in primary outcome (~30%), major infection rate (~25%), death rate (~10%), necrosis (~60%), mechanical ventilation (~12%). 5-10% got parenteral nutrition. 


  • Importantly only a fifth needed ICU care. So of those predicted to get severe pancreatitis presumably many didn’t.
  • Target calorie delivery was rarely achieved until day 3 or 4 in the early fed group and 6 or 7 in the other.
  • Interestingly, more than 2 thirds of of the ‘on-demand’ group didn’t need tube feeding.


So early NJ feeding doesn’t appear superior to delaying addressing nutrition until day 4 and starting oral intake.

How useful is this result? Accurately predicting severe pancreatitis is clearly problematic and for studies like this mean that the sample ends up being very heterogeneous. Regarding study design, mixing timing and route may have limited the applicability of these results, and many are inherently suspicious of composite outcome measures. 

We need a way to better predict severe pancreatitis. And then perhaps a 2×2 look at timing and route?

Shock taskforce – define, diagnose, monitor, treat it right.

My notes on this modern primer on shock by 11 well-known shock experts (shoxperts?). Interesting mix of evidence-base and expert opinion/enthusiasm. Definitely some echophiles amongst them.


Life-threatening, generalized, acute circulatory failure associated with inadequate oxygen utilization by cells (dysoxia leading to hyperlactataemia).


  • Insufficient circulating volume,
  • Pump failure (muscle problem or rhythm problem),
  • Obstruction (PE, pleural/pericardial effusion),
  • Distributive shock (vasodilatation – sepsis, anaphylaxis, spinal injury),

or a combination of the above.




Clinical, haemodynamic and biochemical.

3 Clinical ‘windows’
  1. Peripheral – cold, clammy, discoloured skin
  2. Renal – <0.5ml/Kg/hr
  3. Neurological – obtundation, delirium

Hence – Serial assessment HR, BP, temperature, skin features, urine output and mental state

Hypotension and shock

Hypotension (SBP<90, MAP<65, drop of 40)

  • NOT a prerequisite for diagnosing shock.


  • >1.5 mmol/l is associated with increased mortality, >2 usually. More predictive than almost any other marker (particularly inflammatory markers). But prognostic value does depend on the type of shock.
  • Elevation not simply due to abnormal oxidative phosphorylation.
  • Potentially protective and therefore a marker of severity rather than a ‘toxin’ needing to be removed.
  • Early reduction with therapy is a good sign.
  • Seems useful in guiding therapy, ie targeting lactate clearance (20%/2hr for first 8hr) – Jones and Jansen say so.


  • Low values (potentially inadequate O2 delivery) and high values (inadequate O2 use) are prognostically bad. But normal value can be misleading.
  • Elevated venoarterial CO2 difference ((ScvCO2-PaCO2) >6mmHg or 0.8KPa) may be a better marker of insufficient blood flow.


Serial lactate measurement is crucial.

If CVC in place then measure ScvO2 and venoarterial difference.



How/when to monitor cardiac function

Use cardiac monitors to 1) Identify shock 2) Select therapy 3) Evaluate response

Often a mixture of types of shock. Early echo quickly helps in decisions re fluid ‘v’ inotrope ‘v’ pressor ‘v’ definitive intervention.


Blood pressure target

Use an arterial line and CVC if not improving with initial treatment.

Individualize the target.

  • MAP >65 is default, but:
  • ‘higher’ if known hypertensive or if more seems to help.
  • ‘lower’ in uncontrolled bleeding

No new comment about targets in brain injury – ?stays at 90 mmHg.

Improving perfusion
  • Fluid early, as described below.
  • Inotropic agent (no specific recommendation) if BOTH inadequate cardiac output AND impaired cardiac function.
  • Do not use inotropes for impaired cardiac function with sufficient cardiac output.
  • Do not supranormalize oxygen delivery.
Evaluating response
  • Fluid responsiveness markers
  • End tidal CO2 changes (in ‘steady state’ ventilation)
  • Resolution of preload dependancy markers (where they were marked)
  • Arterial pressure, change in pulse pressure (not PPV), CVP, PAOP are all flawed.
  • Cellular dysoxia markers – theoretically nice but take longer to change.


Measure cardiac output if clinically not responding well to initial therapy.

Sequential assessment is key.



Preload and Fluid Responsiveness

Knowledge of preload is not knowledge of preload responsiveness.

Dynamic variables for fluid management are best:

  • Stroke volume variation
  • Pulse pressure variation
  • IVC variation
  • Passive leg raising

All have confounders (tidal volume, abdominal pressure, dysrhythmia, cardiac and pulmonaru state, open chests etc.) and SVV,

PPV and IVC variation are only valid in the absence of spontaneous breathing.


Echo may be best way to determine type of shock. If you don’t use it you may miss-manage 25% of septic patients.

Assess volume ideally using dynamic markers.

Static preload markers should not be used alone (CVP, PAOP, change in pulse pressure (not PPV)). However, immediate fluid is recommended if very low value of static markers.

Do not target LV filling pressure.

When deciding to give fluid use a measured fluid challenge and assess response, don’t just pile it in.

FLUID RESPONSIVENESS DOES NOT MEAN THEY NEED FLUID. Assess lung water where possible (TPTD or ultrasound).


Methods of monitoring


Echo can characterize the circulatory disturbance, aid therapy selection and assess response.

In addition to obvious structural abnormalities it is used to assess:

  • Direct stroke volume measurement using Doppler (velocity-time integral across an outflow tract)
  • Ejection fraction reflects the balance of contractility and afterload, so needs to be interpreted in the context of type of shock, and inotrope/pressor use.
  • Filling pressures can be estimated using pulsed Doppler and tissue Doppler. Combining these allows a ‘load-independant’ assessment (the E wave of LV inflow adjusted by the E’ wave of the MV annulus longitudinal movement). E/E’ >15 suggests high LV filling pressure, <8 suggests low. In practice it’s rarely outside this range.
  • RV size can be assessed with 2D echo, (ideally) using end-diastolic area values. RV:LV >0.6 suggests moderate RV dilatation, in turn suggesting acute cor pulmonale when accompanied byabnormal septal motion.
PA catheter
Provides a lot of numbers (RAP, PA pressure, PAOP, ScvO2, PvO2, O2 extraction etc.)
Accurate cardiac output monitor.
Use beyond specific indications is contentious (lack of benefit, possible harm, expense).
Transpulmonary thermo/lithium dilution
Less invasive. Also lots of numbers (global end-diastolic volume, extra-vascular lung water, cardiac function index).
Intermittent calibration with continuous waveform analysis in between.
Uncalibrated pulse contour analysis
Either via an arterial line or finger pleth. Complicated maths to estimate artery characteristics. Precision, accuracy and reliability is being questioned.
Oesophageal Doppler
Velocity time waveform of the descending aorta. Normogram for arterial cross-sectional area. Derived variables. Reliability issues largely relate to positioning.
Phase shift of high frequency current put across the thorax. Yet to prove itself.
Tissue microcirculation monitoring
Tissue CO2
Near-infrared spectroscopy occlusion test
Muscle O2 saturation changes with occlusion (DeOx and ReOx).
Microvideoscopic techniques, looking at capillary density, perfusion and flow heterogenity in eg tongue epithelium.
All currently research tools only.



Sequential echo is useful if not vital.

Do not use PAC routinely.

PAC or TPTD in severe shock.

Other uncalibrated and minimally invasive devices are yet to be validated.




Know your phases and master fluid management

ADQIThe ADQI renal authority turns its attention to fluid strategy in ADQI III’s BJA article.






Choose the right fluid balance target for the stage of critical illness:

  1. Rescue – first few minutes; generous fluid boluses to treat shock. Obviously, not all patients require this phase.
  2. Optimization – few hours; assessment of tissue perfusion, use of fluid responsiveness markers, working towards stability
  3. Stabilization – next few days; cover losses but otherwise minimize fluids and keep balance neutral-ish.
  4. De-escalation – mobilize fluid; allow or encourage it out; negative balance.

Different measurements and vital signs are valid (or used) at different stages. The utility of ScvO2, cardiac output monitoring, echo and fluid responsiveness markers for assessing volume status are largely confined to the ‘optimization’ period.

Lack of attention in the optimization and stabilization stage is common. Fluid in drugs and electrolyte replacement adds up and often passes relatively unchecked.

In addition a current article in Crit Care involved 492  patients in a prospective registry-collected observational study. Multivariate LR was used to assess whether fluid balance and hypotension (on RRT) was related to death or prolonged renal support. Positive fluid balance, or hypotension, in the first seven days of renal support was associated with greater mortality, but not long term renal support.

It’s difficult to draw any firm conclusions from this observational study. Although many variables were adjusted for, some can never be. Immortal time bias is a particular potential problem in this instance. However it as least adds more weight to ADQI’s assertions.

Immortal bias


More on ‘shaping’ fluid balance on the OXICM site.




Towards global consensus on end of life care principles in the ICU

End of life care 2The  WELPICUS study group have used an interesting approach (using the Likert scale to assess a number of statements) to try to gain international consensus (from more than 30 countries) on what matters in end of life care in the ICU.




Autonomy and decision making

There was considerable support for respecting advanced directives (AD), but not necessarily for doctors to encourage patients to write them. There is little current use AND low medical adherence. Patients with capacity may change their mind. Don’t break the law to follow an AD! So, know the law.

Decision making and capacity key elements reaffirmed.

  • Possession of necessary values,
  • ability to communicate and understand,
  • ability to reason.

Strong support for 3-tier model:

  1. Autonomous decision making where possible;
  2. Acknowledge previously communicated views when capacity is lost;
  3. Elsewhere, act in best interests.

Capacity is felt to be conditional – ICU collects complex cases and complex decisions for which a basic level capacity may be insufficient. (America/Europe historical disparity on this is lessening). Patients known preferences trump relatives preferences.

Gain consent for invasive procedures when possible.

Use shared decision making

Large international variation in involvement of surrogates in decisions about EOL. Europe very keen on shared decision making.

Life-sustaining treatments

Consensus gained on withdrawal of treatment when chances of surviving are very low or patient’s wishes against it are known. Includes CPR!

Not yet complete consensus on details of withdrawal/withhold/palliative issues due to legal issues in various countries. Again, know the law!

Decisions regarding end of life care should be made in by the multi-disciplinary ICU/home team AND in discussion with patient/family.

The intent must never be to shorten life, even if this is allowed by law.

Brain death

Near consensus on discontinuation of all therapy after BD. If the family does not accept brain death then most countries allow discontinuation anyway, but if their wishes are honoured then the patient is NOT required to remain on the ICU.

Palliative care

Physical, social, spiritual and emotional needs attended to – in order to facilitate as much autonomy as possible.

end of life consensus

Inadvisable actions

Non-beneficial treatments and investigations should be avoided. Difficulty with what ‘beneficial means’.


Document decisions and their rationale.


Everyone needs educating and teaching should be culture/law/religion specific.

Clinical ethics committees

These committees can be useful for advice where opinions differ.

Uncontested statements

Keep families included, informed and happy

Maximise life-saving but bear in mind long-term QOL


The global mind-set on end of life principles appears to be converging but some interesting differences remain.

Emphasis on shared-decision making is evident.


Personally, I find the comments about ‘very low chance of surviving’ unhelpful. On the whole we can make death on the ICU humane and largely dignified. But, in many cases we don’t have reliable markers of long-term outcome for individuals. We know that it almost always ends badly but to aim for the improbably tiny ‘bulls-eye’ of a good outcome you risk hitting the far larger surrounding zone of drawn-out death or unacceptable quality of life.

Consequently I often find, in discussions with patients and families regarding withdrawing, withholding or limiting treatment that the following elements need to be explained:

  • Death is most probable whatever we do.
  • There may be a tiny chance of survival, but…
  • what is much more likely, if we continue aggressive treatment, is that death will be protracted and potentially unpleasant or the quality of long-term health will be (unacceptably?) low.

Any advice on handling these discussions?

Be wiser, do less to be better.

Choose wiselyOver the last couple of years, 4 US critical care societies put together a literature-scouring taskforce as part of the Choosing Wisely campaign. After a lengthy process they came up with 5 cost-effective, tradition-questioning recommendations:






1. Do not order diagnostic tests at regular intervals (such as every day), but rather in response to specific clinical questions;

Avoid anaemia (and so transfusions). Avoid ultimately inconsequential incidental findings demand further work-up.


2. Do not transfuse red blood cells in hemodynamically stable, non-bleeding ICU patients with an Hb concentration greater than 7 g/dl;

A threshold of 7 mg/dl results in better survival, fewer complications, and reduced costs. And preserves a precious resource. But unelucidated subgroups may exist.


3. Do not use parenteral nutrition in adequately nourished critically ill patients within the first 7 days of an ICU stay;

If not malnourished prior to admission, early PN (even in those not tolerating EN) is possibly harmful and costs more.


4. Do not deeply sedate mechanically ventilated patients without a specific indication and without daily attempts to lighten sedation;

Protocol-based approaches can safely limit unnecessary deep sedation – eg lightest effective dose; analgesic before sedative; daily sedation holds.


5. Do not continue life support for patients at high risk for death or severely impaired functional recovery without offering patients and their families the alternative of care focused entirely on comfort

Routinely engage high-risk patients and their next of kin or advocate in discussions about limiting the level of aggression of treatment. Promotes patient’s/family’s values, improves the quality of death, and reduces family distress. Initiating palliative care pathways/team may be beneficial even where death is not necessarily expected.


Whether or not these would be your top five the ambition is laudable and these 5 recommendations, originally published in January, have generally withstood criticism over the last 9 months, and ‘savings’ are being made.


Older posts «

Follow on Feedly