Trans-oesophageal Echo

Journals Reviewed: A variety of Journals from 1999 and 2000
Abstracted by: Dr J van Schalkwyk; FCP (Critical Care) (Consultant, University of the Witwatersrand)

Summary of abstract

Here we look at several articles on TOE with relevance to anaesthesia and/or intensive care. (You may wish to briefly browse our editorial comment ).


1. TOE for Preload Assessment in the Critically Ill

The authors describe two groups of patients (twenty one patients post-cardiac surgery, and twenty ICU patients) who had both TOE and pulmonary artery catheters (PACs) in place. Baseline measurements were followed by fluid challenge (500 ml colloid over 15') and repeat measurements. The 'take home' message at the end of the abstract is worth quoting in full:

In a ventilated intensive care unit and cardiac surgical population, transesophageal echocardiography and pulmonary artery catheter are sensitive in detecting changes in preload after volume administration. Few patients demonstrate volume-recruitable increases in stroke volume when compared to cardiac surgical patients. It is not possible to establish an overall end diastolic threshold below which a large proportion of ventilated patients respond to volume administration".

On the surface, this study looks really interesting. We will first look at the authors objectives and conclusions, and then make a more careful examination of the body of the article, before deciding whether there 'take home message' should in fact be taken home!

A. Objectives
The objectives of the study are clearly stated:

  1. Determine the relationship, in an ICU population, between stroke volume (SV) and end-diastolic area (EDA) as assessed by TOE (using the transgastric mid-papillary short axis view);
  2. 'See whether an optimal EDA exists below which an increase in SV is observed after IV volume administration';
  3. 'Examine whether an ICU patient subgroup was different from postoperative cardiac surgical patients'.

B. The conclusions of the study (slightly re-arranged) are:

  1. 'There was a modest relationship between SV and EDA in all patients (r=0.60) ..';
  2. 'It was not possible to establish a threshold [EDA] below which a large proportion of patients respond to volume administration'.
  3. '.. few ICU patients responded to volume administration when compared with the post-operative cardiac surgical patients.. The ICU responders seemed to cluster at an EDA <= 12cm 2 ';
  4. PCWP increased significantly in responders (those who increased stroke volume (SV) by 20% or more) and non-responders alike;
  5. 'We found no correlation between EDA and PCWP' (pulmonary capillary wedge pressure);
  6. 'We found no correlation between SV and PCWP';
  7. 'This study provides evidence that the TEE gives information additive to the pulmonary artery catheter in the assessment of preload in an ICU population. Although there may not be a specific threshold EDA value that reliably predicts a response to fluid administration in all patients, the LV EDA may be useful in identifying some critically ill patients who could benefit from volume administration.'

C. A critique
In order to derive benefit (if any exists) from the following criticism, it's a good idea to have the article at hand, as we will not regurgitate the whole article verbatim. We have several problems with the study, apart from the usual limitations of ICU studies (smallish sample sizes, and the heterogeneous nature of ICU patients, both of which apply to this study). Our problems were:

  1. Objectives & conclusion :
    If we examine the objectives of the study, we have to ask ourselves several questions:
    • Do we expect a relationship between SV and EDA to exist? (Question 1)
    • Do we expect a 'magic number' for the EDA below which volume supplementation will work? (Question 2)
    • Do we expect the ICU subgroup to differ from the postoperative cardiac surgical group? (Question 3)

    The first question can be made more clear by paraphrasing it: "Do we expect a relationship between stroke volume and a proxy for preload measurement?". Stroke volume is determined by three things, (a) preload, (b) contractility and (c) afterload. So, on basic physiological principles, we can answer the authors first question with a "somewhat but not a lot" before we look at a single patient. We would indeed be surprised if there was any other result!

    Anyone who has been in ICU for some time should have acquired a healthy mistrust of magic numbers - we will therefore not be entirely flabbergasted if the authors don't come up with one (and we might be deeply suspicious if they do). Question number two answered, although perhaps not entirely satisfactorily.

    Let's turn to the cardiac group who were carefully selected with (i) uncomplicated surgery; (ii) good pre-operative ejection fractions, (iii) no recent myocardial infarction, and (iv) a 'stable condition' postoperatively. We compare these with a mix of ICU patients with sepsis (7), ischaemic heart disease (3), 'cardiac arrest' (2), pancreatitis (2) and a mixed bag of others with APACHE II scores from 10 to 33. We ask ourselves, "Do we expect the two groups to 'be different' ?". Clearly, the answer is again self evident. We would fall on our backs with surprise if the two groups didn't differ!

    The study is now starting to look a little less enticing. Even before we look at the results, we don't expect (if the authors adhere to their main objectives) that we will learn much that we couldn't predict (apart, perhaps, from an indication of what EDA values will be fluid responsive).

    Now let's re-examine the conclusions..

    1. The first conclusion, that SV relates modestly to EDA (r 2 of 0.36, indicating that EDA variation accounts for about one third of the total variation in SV) is as we expected;
    2. 'No magic EDA number' is far from surprising, especially given the heterogeneity of the case mix;
    3. The groups were different. (We explore this in more detail later).
    4. PCWP increased in most patients when fluid was given.
    5. The lack of correlation between EDA and PCWP has been shown in other studies - nothing remarkable here.
    6. The lack of correlation between SV and PCWP (in the face of a modest correlation between SV and EDP) therefore comes as no surprise;
    7. Conclusion B7 above is a bland pronouncement on preload, with minimal apparent scientific content.

    What then can we learn from this study? I believe that one can often learn more from a weak study than from a strong one, simply by analysing its faults in detail. Let's take an even closer look.

  2. The Methods
    The study has several good points in that ethical review appears appropriate, there were forty one cases (a fair number for an ICU-related study, although the reason why they stopped at 41 is not clear), and an attempt was made to acquire similar ICU cases (inotropic support, low-ish wedge, low urine output, adequate gas exchange, and thought to require fluid).

    There are however several questions we might ask:

    • What is "adequate gas exchange" and how was it determined?
    • How many potential candidates were excluded and why were they excluded (from both groups)?
    • Did anyone refuse consent?
    • How long were the patients in ICU before they were enrolled - was their enrolment a 'convenience' thing, or was there a trigger point at which they were enrolled? (What criteria did they have to satisfy and for how long, before enrolment?)
    • How much fluid did the patients receive before they entered the study?
    • How long did it take to repeat all the studies after the fluid was administered?
    • Let's see some basic demographics for the patients.
    • The ICU cases were broadly 'similar' but were they representative?

    It's disappointing to note that none of these questions appears to have been answered in the article! Of particular concern are the exclusions (if any) as a large number of exclusions could potentially contribute to a substantial selection bias. Even more important is to know 'where' the patients were when the assessment started ! A patient in ICU who is seriously ill is likely to have been subjected to a variety of interventions, each of which may alter their course and outcome. We get some indication of this from the startlingly diverse array of inotropes used. These patients all had pulmonary artery catheters inserted, so it is reasonable to assume the catheters were used for something before they were enrolled. This 'something' was surely goal-directed fluid administration by the physician looking after the patients! Or were the patients enrolled immediately after the PACs were inserted? We will probably never know!

    Because we have few demographics we have no idea of the patients' ages, apart from the bland statement that "there was no significant difference with respect to age, height and weight between the two groups". (We know that on average, women are smaller than men, so this statement, combined with the fact that there were eleven women in the ICU group, and none in the cardiac group should be a cause for concern, and not reassurance)! This brings us to another point - that all parameters are reported without reference to the sizes of the patients! We know that the conventional normalisation according to 'body surface area' (or weight 2/3 ) may well be incorrect, with Kleiber's work suggesting that scaling to weight 3/4 is more appropriate, but I am not aware of work saying it is right to completely disregard the size of the patient! All reported parameters that vary with body size (cardiac output, stroke volume, and EDA) must thus be regarded with deep suspicion! Their goal of a magic number for EDA may have been more closely realised had the EDA to some degree been adjusted for body size!

    In addition, no indication is given of the blood pressures of the patients. As Guyton has pointed out, we cannot become fixated with cardiac function -we need to look at how normal physiology works! The heart is a slave of the peripheral tissues which, if provided with a sufficient head of pressure, take what they need through the exquisitely complex process of vascular autoregulation. Although this process may be deranged in disease states, such derangement suggests that we should strive even harder to ensure that organs in the critically ill receive an adequate perfusion pressure, and not just ensure a high cardiac output!

    A brief note on the analysis. We note that 'linear regression was used to examine the relationship between SV and EDA '. We all know from basic physiology that the Starling curve relating stroke volume and ventricular end-diastolic volume is far from linear, especially as the curve flattens out at higher end-diastolic volumes. Perhaps I'm being silly, but is it appropriate to use linear regression for a non-linear curve?

  3. Results

    Only four of the twenty ICU patients responded to a fluid bolus! This is likely a confirmation of our initial suspicion that the PACs were placed some time before enrolment, and used by the attending physician to guide fluid therapy! It also (surely) precludes adequate analysis - but our authors lump the cases together with the cardiac surgical ones (despite testing for a 'difference' and indeed finding one)! We are tempted to say that you cannot have your cake and eat it.

    Let's look at our cases in a bit more detail. Unfortunately, much vital information is denied us - we've already alluded to lack of blood pressure information, but what about heart rate? We have been provided with stroke volume (SV) and cardiac output, so we can work out the heart rates for the various patients before and after fluid bolus administration. Can we derive any useful information from this calculation? With some cases, I think we can. For example, look at case six, on a noradrenaline + dopamine infusion, who starts off with a cardiac output of 5.00 l/min, and after 500 ml of fluid ends up with a cardiac output of 4.50 ! This doesn't make a lot of sense (one is tempted to conjure up the languishing spectre of a descending limb to the Starling curve), until we look at the heart rate. For some unknown reason, this decreased from 132/min to 116/min. Perhaps the patient's fever subsided? We will never know.

    What were the systemic vascular resistances of the patients? We know the cardiac and ICU groups were not comparable, and the responder group was so small as to make any analysis of significance of questionable value, but it would still give us a better "feel" for the cases if we knew the SVRs.

    Finally, we ask ourselves "If you have a patient with a cardiac output of 11.9 l/min (cases 1 and 10), a decent wedge (14 - 15), and an EDA of 19 - 20 cm 2 , what are you really hoping to achieve by whacking in 500 ml of colloid?" Has supra-normalisation suddenly come back into vogue, or are we just blindly chasing a protocol?

  4. Discussion
    Mention has already been made of the conclusions derived by the authors. What is interesting (although perhaps` not susceptible to statistical analysis, owing to the small numbers) is that the "responders" to fluid in the ICU group had EDAs of 9.8, 7.6, 10.4 and 11.7 cm 2 . The mean EDA in the remaining patients was 20.1 cm 2 .

    We are tempted to rewrite the 'take home message' of this study as follows:

    In a ventilated intensive care unit and cardiac surgical population, transesophageal echocardiography and pulmonary artery catheter are sensitive in detecting changes in preload after volume administration, but the clinical relevance of this observation is uncertain. In a selected and poorly characterised set of ICU patients few patients demonstrate volume-recruitable increases in stroke volume [when compared to cardiac surgical patients]. It is not possible to Owing to the small size, poor design and vigorous filling up of patients prior to enrolment we failed to establish an overall end diastolic threshold below which a large proportion of ventilated patients respond to volume administration although just eyeballing the data it would seem that round about 10 cm 2 is not a bad number, but we forgot to correct for patient size!

Article 1: The Use of Transesophageal Echocardiography for Preload Assessment in Critically Ill Patients
Anesth Analg 2000 90 351-5
Article type: Clinical Study
Authors: Tousignant CP, Walsh F & Mazer CD.


2. Mitral Regurgitation and General Anaesthesia

In a well designed study, the authors demonstrate quite convincingly that anaesthesia often has a substantial effect on the severity of mitral regurgitation. Every anaesthetist will know this already, but it's good to see quantitation, and be reminded that severe mitral regurgitation may be grossly mis-assessed in the anaesthetised patient!
 

Article 2: Effect of General Anesthesia on the Severity of Mitral Regurgitation by Transesophageal Echocardiography
Am J Cardiol 2000 85 199-203
Article type: Clinical Study
Authors: Grewal KS, et al.


3. Evaluation of Mitral Regurgitation

A beautiful article on systematic examination of the mitral valve using TEE. The authors demonstrate that meticulous, carefully planned mitral valve assessment is better than a (retrospective) "let's have a look" approach! Despite the small numbers, I believe the authors have made a significant contribution, that will permit improved surgical decision-making.
 

Article 3: Improved Evaluation of the Location and Mechanism of Mitral Valve Regurgitation with a Systematic Transesophageal Echocardiography Examination
Anesth Analg 1999 88 1205-12
Article type: Clinical study and tutorial
Authors: Lambert AS, et al.


4. TOE and Haemodynamics in children

This study superficially suggests that intra-abdominal pressures of up to 12 mmHg are safe in children, as there is only a transient and fairly insubstantial decrease in cardiac output in response to such pressures. Two comments are in order:

  1. The authors provide only mean ± SD data, not individual values. There is nothing to tell us (for example) whether in one of the children the cardiac index didn't plummet to an extremely low value, but that this is hidden in the average!
  2. This study should be contrasted with the next one, which we consider much more thought-provoking.

 

Article 4: Transoesophageal echocardiographic assessment of haemodynamic changes during laparoscopic herniorrhaphy in small children
BJA 2000 84(3) 330-4
Article type: Clinical study
Authors: Sakka SG, et al.


5. TOE, the elderly and laparoscopic cholecystectomy!

In laparoscopic cholecystectomy, middle hepatic venous blood flow plummets with pressures of 9-12 mmHg, from a baseline of 330 ± 64 ml/min, to 74 ± 25 ml/min! Such dramatic changes don't occur with open cholecystectomy. Recovery is slower in the elderly. Worrying stuff!
 

Article 5: Hepatic Blood Flow and Function in Elderly Patients Undergoing Laparoscopic Cholecystectomy
Anesth Analg 2000 90 1198-1202
Article type: Clinical Study
Authors: Sato K, Kawamura T & Wakusawa R.


Editorial thought-of-the-day!

Wouldn't it be nice if comprehensive patient data for each and every new clinical study were available on the Internet, perhaps in some stable repository? We could avoid the oh-so-frequent problem of inadequate information to really properly assess a study. Is any author brave enough?

By the way, there's a really remarkably good echo website at Echo In Context. Well worth a visit.

Ed      

Old scopes logo
Up