Quality Goals vs. Thresholds
What Should Labs Tell Customers?
Director of Quality and Compliance, ARUP Laboratories, Salt Lake City, Utah
A client questioned our STAT turn-around time goal of 95%. They would like us to set our goal at 100%. We understand that 95% is generally an accepted benchmark. What is your advice?
The client is right—a labs’ goal should always be perfection. What you have been advertising is your STAT turnaround time threshold. Advertising your threshold for quality monitoring, instead of your performance goal is misleading to customers.
What is the difference between a goal and a threshold? The Joint Commission introduced the term threshold into its 10-step quality process in 1987. The Commission’s “Primer on Indicator Development and Application” in 1990 included an in depth discussion of thresholds in the chapter on Indicator Data Evaluation. Here, threshold was defined as: (1) the point or level at which a stimulus is strong enough to signal the need for organizational response to the data, and (2) the beginning of the process of determining why the threshold has been reached.
Take for example, a sentinel event indicator, which always has a threshold of zero. The classic sentinel event indicator for laboratories is a blood transfusion-related death. Each sentinel event triggers the need for prompt and thorough investigation.
Setting rate-based or non-zero thresholds is more complex. Approaches include expert consensus, definition of individual organization objectives, and use of a derived range around a statistical mean. Expert consensus relies on valid and available data and may come from multiple sources. For laboratories, perhaps the most comprehensive benchmarking data has been compiled and published through the programs of the College of American Pathologists. Of course, each setting has its unique characteristics such as patient mix and operations; therefore, benchmarks must be studied to ensure comparisons are valid. Organizational experts may convene to determine a reasonable threshold that is continuously refined as data is collected and analyzed and the process performance improves.
Rate-based thresholds may indeed be the goal when set as the internal objective. This must be approached with caution as the threshold must signal a rational response to the data even while in pursuit of the goal. Ongoing accumulation of data and calculation of the statistical mean and a range around that mean will provide a central data baseline as a starting point for continued monitoring and analysis of performance.
Data quality is paramount. Ideally, the data that informs whether a threshold is being exceeded should include only the uncomplicated process. This is particularly important when it is made available to customers. In the question about STAT turnaround times, if the collected data also reflects instances with submission problems (e.g., mislabels, inadequate specimens) or difficulties with the analytical or the reporting process, the additional time to resolve these issues will ultimately skew the compiled information. This could lead to a tainted reaction and incorrect action.
Thresholds are valuable, and as described by Gary Clark, have been “a realistic management tool for dealing with natural variation of indicator data and for establishing a rational basis for triggering investigation” (1). Data trending and analysis of patterns must also be included in the monitoring process. These can reveal issues of concern or need for intervention even if a threshold has not been exceeded.
Readers may find a recent review of quality monitoring interesting (2). The most important message a laboratory can extend to its workforce is, “Do the right thing, in the right way, every time.”
- Clark G. Systematic Quality Management. Chicago: ASCP Press, 1995.
- Shangian S, Snyder SR. Laboratory Medicine Quality Indicators. Am J Clin Path. 2009;131:418–431.