Nosocomial infections in surgical patients: developing valid measures of intrinsic patient risk.

For surgeons or hospitals to compare their rates of wound infection meaningfully, the analysis must first control for the mix of intrinsic infection risk of their patients. Research over the past century has led to the development of several intrinsic risk indexes that can be used to stratify the wound infection rates so that valid comparisons can be made within risk strata. For an intrinsic risk index to be useful for comparing rates, it must control for all of the important intrinsic risk constructs; merely being statistically associated with infection rates does not ensure that a risk index will be useful. Understanding how a risk index can be both parsimonious and comprehensive requires consideration of the competing principles of multicollinearity and orthogonality. Various techniques of multivariate analysis are used to develop multivariate risk indexes, but the success of the process depends on having all of the important orthogonal risk constructs represented in the pool of predictor variables available for the analysis, either directly by variables in the pool or by demonstrated multicollinearity. Despite recent advances in risk measurement, many important questions remain.