Latest White Papers

Enhancing the Prediction Model

During the course of our use of the APACHE II system, including its use as a predictor of hospital outcome of individual patients, we came across several theoretical and practical problems, which our methodologies are designed to overcome.

Dynamic Pathophysiological Processes

We instituted daily APACHE II physiological assessments in order to reflect better the dynamic pathophysiological processes affecting ICU patients.

Reduction in Subjective Errors:

The APACHE II Risk of Death gives the probability of hospital death of an ICU patient and is derived from the APACHE II Score and a coefficient relating to the Specific Diagnostic Category or Major Organ System Failure which necessitated admission to the ICU. This choice is subjective and static. The correct choice of the Specific Diagnostic Category can be difficult especially among patients with multiple diagnoses and multiple organ system failures. Furthermore, during the course of a patient's stay within the ICU, other organ system failure may develop which may affect the final outcome. We have opted to use the Best Glasgow Coma Score over a 24 hour period as it is again a subjective assessment and an error in score can have a major effect on the APACHE II Score. Teasdale and Jannett, the originators of the Glasgow Coma Score have shown that the BEST score is the better predictor.

We have modified the APACHE II Score to take into account the effects of concomitant organ failures and their duration. Our proprietary algorithm calculates the Organ Failure Score from the APACHE II Score and a coefficient relating to the number and duration of acute organ system failures. The presence of organ failure is defined according to strict pathophysiological criteria.

OFS = APACHE II Score x (1 + Organ Failure Coefficient)

Dynamic Trend Analysis:

We use dynamic trend analysis of daily APACHE II Score corrected for the presence and duration of organ system failures to identify patients whose prognosis is hopeless. This reflects better the dynamic pathophysiological processes affecting ICU patients.

Outcome Prediction Model

Prognostic criteria (including the APACHE system) based on static analysis of group statistics do not help much in decisions to withhold or withdraw therapy from ICU patients too ill to benefit, since they do not provide adequate information on the features that distinguish non-survivors from survivors. For example, a patient with a certain disorder and APACHE II Score may be estimated to have a risk of dying of say 90% after 24 hours in the ICU. Would this risk of death be adequate information to support a decision to withdraw treatment? Can we blame the clinician for his difficulty in grasping probabilities and drawing conclusions from them? After all, the particular patient whom he decides not to treat may be one of the lucky 10% (or unlucky if treatment was withdrawn) who would have survived. Even more difficult is, say a patient with severe sepsis who had a risk of death of 60%. All clinicians would start treating such a patient: but up to what point? Can we know before death is imminent that a patient is no longer salvageable?

The solution lies in the development of dynamic models that would identify individual patients who would die during their course of illness. The prediction to make is death, not survival. This approach is important for strategic decision making. When a patient is admitted to the ICU the eventual outcome with the exception of low-risk patient being monitored is unknown. The aim of treatment is to achieve survival and this aim does not change as long as the outcome is unknown. However, when we know that a patient will die despite all treatment and withdraw treatment, the consequence of that decision is irreversible.

To develop a dynamic, individually predictive model, the key question to ask is p what are the features that distinguish the patients that died from those that survived? This question can be broken down to three subsidiary questions? What do most patients die from? Are there markers of this process (if a common one) that can be followed over time? Is there a reproducible pattern in the changes of these markers that distinguish survivors from non-survivors? ICU patients generally die from multiple organ system failures with death often the result of a breakdown in homeostasis. Since the function of major organ systems and homeostasis is reflected by changes in measurable physiological variables, selected variables can be used to follow the development and progression of major organ system failure.

As an illustration, let us imagine a ball in a bowl. When the ball is in the centre of the bowl, homeostasis and physiological status is normal. The size of the bowl represents the amount of physiological reserves. Pathological processes act to shift the ball away from the centre, whereas homeostasis mechanisms and effective treatment tend to shift the ball back to the centre. If the ball goes over the brim of the bowl, the threshold of physiological abnormality compatible with life is crossed and the organism dies. The APACHE II system fits this illustration well. The size of the bowl is represented by the reciprocal of the sum of age points and chronic health points. The acute physiological score tells us of the pathophysiological status of the patient – i.e. the position of the ball. However the original APACHE II score tells us only the position of the ball after 24 hours. It does not tell us about the changes in its position nor its rate of change on subsequent days. The rate of change is important, since with the position of the ball captured at a certain time point, it will allow us to predict whether the ball will go beyond the brim of the bowl. This emphasises the importance of dynamic analysis.

The Individual Outcome Prediction Model is switched off in our Clinical Audit Module. MASH has been researching this topic over the past 25 years or more. This topic has caused tremendous controversy amongst clinicians and wide debate. We can understand the moral and ethical issues surrounding research in this topic. It is akin to when the UK NICE (National institute for Clinical Excellence) decides to withhold expensive budget busting new high tech drugs from the general public as it is not proven to be cost effective.

Our aim is always to reduce mortality in the ICU.
 
canakkale canakkale canakkale truva search