My main concern when using TCI is the variation in anaesthetic demand between individual patients.

Once this is settled, TCI offers advantages of quick and easy adjustments of anaesthetic depth.

Each patient is unique, and a such, I really have dificulty accepting TCI. I find continuous infusions giuded by patient responses the way to go. TCI is fancy and all, but for all practical purposes, gives a false impression of control.

Once you have settled the patient´s need, TCI still offers quicker control of adjusting anesthetic depth.

I think PKPD visualization tools are the way to go.

Obviously, biological variation places a limit on any model one uses.

That's why we need anesthesiologists.

But if you paralyze the patient and use IV agents, you should always use an EEGe derived parameter to monitor depth because of model misspecifiation.

I also believe that feedback from the patient is the way to go. We need tools telling us the patient's status, like bispectral index or any other deep anaesthesia monitor, specially when no nitrous oxide or volatile agents are used.

TCI works fine in the vast majority of patients, but only one recalling the surgical procedure is enough to save the day.

..what we need is a way to monitor the level pf analgesia. So far all suggested tools (EEG power spectra, EEG high frequency waves, AEP etc ) have failed or are not specific. One item is in the pipeline: sensory evoked potentials. but that will last till we see this device in the clinical environmnet. At least prelimenary data demonstratethat with this tool the individual level of analgesia (not anesthesia-because it also iniolves sedation) seem possible.

I have an issue with the semantics here. We give anesthesia to obliviate pain, so what do we really mean if we say "the level of analgesia"?

I agree with all contributors that population models will always be hampered by the fact that you do not know in advance whether your individual patient will behave as an average patient or rather will behave like an outlier. However, remember that it is a biological fact that chances are high of having a patient that behaves like someone close to the middle of the gaussian distribution rather than an outlier. (by definition: outliers are rare :-) )

TCI has major limitations due to their population derived background. BUT at least they are applied through IV pumps. This means that the maximum infusion speed of the pumps can be standardized between patients. It is well known that differences in speed of injection are a cause of variability in plasma and effect-site concentration. In my opinion the standardised infusion speed of the bolus dose in each TCI pump results in THE major advantage of TCI versus manually controlled bolus en CI techniques. Even with the model misspecifications included, using TCI allows a better control of the timecourse of drug titration towards a predefined effect, whereas a bolus + continuous infusion technique may end up with larger variability in plasmaconcentrations and effects using the same dose in the same patient but at different speed of injection.

The problem of model estimation error becomes less relevant if we start considering TCI as a concept of improved reproducibility in drug titration rather than an attempt to predict individual responses perfectly. Here in Groningen, we are currently focussing our investigator initiated studies towards anesthetic interactions and have found amazing and promising results on predictability of BOTH hypnotic and analgesic endpoints during anesthesia.

(REF: Best Abstarct awards competition: Euroanesthesia, Barcelona 2013 Hannivoort et al...Drug interaction models are better predictors of tolerance/response to noxious stimuli compared to individually measured parameters)

Offcourse, confirmation from other researchers will be mandatory (and are on the way as we speak). The unique thing in these studies is that we do not focus on the misspecifications of each individual TCI model, but rather use the models to control the time course of drug administration better. I am convinced that combined use of TCI with multiple drugs will help to target equipotent drug combinations more accurately than ever before.

@ Jan Hendrickx: How are you doing? Still very active in science I notice! About level of analgesia: I only understand "dose or concentration of analgetics". That is something we can control. As such: the correct question remains: Which individual dose/concentration of opioids do we need to evoke tolerance to noxious stimulation. We can only answer such questions by studying population averages first, I guess? What we really hope to reach is the ability to predict in advance whether a balanced combination of hypnotics or analgesics will/or will not evoke tolerance to a certain noxious stimulus, aren't we? But that is exactly the catch 29! As both changing hypnotic and analgesic drugs will have an effect on the total probability that a patient will respond to a noxious stimulus. Both drug types will always be linked in clinical practice. As such, I am convinced that this topic can only be studied in well constructed interaction studies.

To spice things up... I did not yet mention this major second source of unpredictability in responsiveness ... how noxious is each noxious event during surgery? How can we compare that?

I guess we are only at the beginning of an era where we start gathering data on these topics. It is my conviction that we have to find a usefull measure of potency of noxious events in their ability to evoke some sort of arousal during steady state anesthesia? Coincidently, we just have a new paper in press that makes such an attempt. ;-) Watch "Anesthesiology" for: Heyse et al... Response Surface model Approach for continuous measures of hypnotic and analgesic effect during sevoflurane-remifentanil anesthesia. Quantifying the Pharmacodynamic shift evoked by stimulation. To be continued...

Hugo,

I am convinced that "The problem of model estimation error becomes less relevant if we start considering TCI as a concept of improved reproducibility in drug titration rather than an attempt to predict individual responses perfectly. Here in Groningen, we are currently focussing our investigator initiated studies towards anesthetic interactions and have found amazing and promising results on predictability of BOTH hypnotic and analgesic endpoints during anesthesia. " Congratulations on Hannivoort. Maybe a combination of your model with an EEG derived parameter will work better still. Nick Franks mentioned a measure of cortical effect (EEG) and subcortical effect (MAP, HR - mediated by medulla oblongata) may do the trick. Still, I think concepts like the SMartPilot and Navigator are the way to go - that's why I include it in our upcoming meeting www.navat.org

hello again, Dr. Vereeke - and thanks again for your splendid Copenhagen presentation in 2011! Interesting to hear about the importance of bolus infusion speed. Can you refer to studies of this phenomenon?