Traditionally, any deviation from the pathway is deemed a "variance" from that pathway. Whether the variance was not checking vital signs X number of times this shift or an MI in a cardiac surgery patient, the variances were documented and analyzed.

We call this "drinking from a fire hose". Far too much data from which to extract meaningful information. This approach to variances terminated most variance systems, and many pathways programs.

Another approach to variances has been to only evaluate some set of outcomes, usually LOS or "costs". If the LOS for the patient population was less than before its pathway was implemented people celebrated. This approach greatly reduced the data collection effort, and eliminated any analysis of exactly what on the pathway caused the LOS reduction, or even if the pathway was the reason.

We take a broader view of variances. Task / intermediate outcomes data and final outcomes data are necessary for effective performance improvement. One set of data provides the context for the other and inform decision making.

We recognize two variances:

Compliance (Process) Variances:Are we doing what the pathway says? Is the patient progressing through the pathway as planned? The task and intermediate outcomes.

Effectiveness (Outcome) Variances: Was the care process effective, did it produce the desired outcomes? The final outcomes of care: clinical, health status, satisfaction, costs, LOS.

The task then becomes using process and statistical thinking tools to identify, understand, and correct - if necessary - the causal relationships between compliance and effectiveness variances.