Prognostication and intensive care unit outcome: the evolving role of scoring systems

Prognostic scoring systems remain important in clinical practice. They enable us to characterize our patient populations with robust measures for predicted mortality. This allows us to audit our own experience in the context of institutional quality control measures and facilitates, albeit imperfectly, comparisons across units and patient populations. Practically, they provide an objective means to characterize case-mix and this helps to quantify resource needs when negotiating with hospital administrators for funding. Prognostic scores also help to stratify patient populations for research purposes. To be used accurately and effectively, one must have a good understanding of the limitations that are intrinsic to these prognostic systems. It is important to understand the details of their derivation and validation. The population of patients that is used to develop the models may not be relevant to your patient population. The model may have been derived several years before and may no longer reflect current practice patterns and treatment. These models may become obsolete over time. As with all scoring systems, there are potential problems with misclassification and more serious, systematic error, in data collection. One needs to rigorously adhere to guidelines about how these data are to be collected and processed; the persons who collect the data require regular updates and ongoing training. In their current form, the systems should not be used to prognosticate in individual patients, nor should they be used to define medical futility. The prognostic models should be viewed as being in evolution. Many patient and ICU characteristics that seem to have an important impact on mortality have yet to be incorporated into any of the current models. As an example, these may include the genetic characteristics of the patients and the ICU's organizational structure and process of care [51, 52]. Because the organ dysfunction measures are able to be obtained daily they give a much more complete understanding of the patient's entire ICU course as opposed to the initial 24-hour period. Daily scores also help to capture the intensity of resource use and may help us gain a better understanding of what is truly ICU-acquired organ dysfunction. These measures may also be used for research to better characterize the natural history and course of a certain disease group or population. Also, they may be used in innovative ways to predict ICU mortality and post-ICU long-term morbidity. These current and developing applications will help us to further understand the link between ICU severity of illness and long-term morbidity as we move beyond survival as the sole measure of ICU outcome.