Integrated models combine clinical predictors together through analytically determined weights to yield a single patient score that can be mapped to an expected survival.
For example, Morita and colleagues developed the Palliative Prognostic Index (PPI) through creating a regression model that predicted patient survival from performance status and certain clinical signs and symptoms among a sample of 150 patients enrolled in palliative care programs. They then tested the approach on a second sample of 95 patients, finding that the PPI predicted 3-week survival with sensitivity of 83% and a specificity of 85% and 6-week survival with sensitivity of 79% and a specificity of 77%. The table below contains a description of the PPI scoring system and the following table , a summary of predictive relevance of PPI scores.
Several other groups have developed similar scoring systems that rely on integration of all or some of the previously described classes of prognostic indicators of patients with advanced cancer and under palliative care (Shimozuma 2000, Tamburini 1996, Maltoni 1999). Such scoring systems need to be sensitive to a variety of methodological concerns including whether or not they are applicable to patients different from the population in whom it was developed and validated (e.g., patients not yet enrolled in palliative care programs)
|