Ann Thorac Surg 1996;62:1358-1359
© 1996 The Society of Thoracic Surgeons
Invited Commentary
Invited Commentary
Richard P. Anderson, MD
Section of Cardiothoracic Surgery, The Virginia Mason Clinic, 1100 Ninth Ave, PO Box 900, Seattle, WA 98111-0900
See also page 1351.
Medicine is being industrialized by managed care. This paradigm shift has been accompanied by the introduction of industrial quality improvement into the hospital and clinic. Whether termed continuous quality improvement or total quality management, the goal is to realize optimum output with minimum variability and conservation of resources. When medical practitioners first encounter quality improvement the reaction is likely to be "We're treating patients, not making widgets!" However, anyone will soon forget about widgets who has ever seen a hospital staff reenergized, chronic problems solved through unprecedented interdepartmental collaboration, and patients delighted because their expectations are exceeded.
One of the fundamental techniques of quality improvement is measure/adjust/remeasure. Surgeons, and particularly thoracic surgeons, have a long and distinguished record in measuring and analyzing the outcomes of their work. As a group, I believe we are more knowledgeable about statistics than most clinicians, partly because we are so often called upon to challenge statistical misapplications by government and the insurance industry. Perhaps this is why thoracic surgeons are often in the forefront of quality management initiatives and tend to form easy working relationships with hospital epidemiologists and statisticians. Measure/adjust/remeasure is not new to us.
In the foregoing article Dr Shahian and his colleagues give an excellent introduction to an ingenious industrial technique for process measurement termed statistical process control. They demonstrate its application in their clinical practice to a relatively large number of patients over a relatively long period of time. Statistical process control is based on the observation that the outcomes of standardized processes, including certain surgical processes, tend to vary randomly over time. What a disappointment for the beancounters who practice scorecard rate comparisons-"Your mortality rate is 50% higher this year than last year; what's wrong?" Random occurrences over time tend to follow certain recognized patterns of distribution and by using these patterns the laws of statistical probability permit estimation of whether such occurrences have or have not occurred by chance. A chance variation in a complication-"We had a bad quarter for perioperative myocardial infarctions"-calls for quite a different response than one not likely to have occurred by chance-"We had better modify our myocardial protection techniques."
The Comment merits a very careful reading because Dr Shahian and his colleagues have clearly spelled out the limitations of statistical process control. Several of these bear repeating. Statistical process control does not study patients, it studies processes. It may not be very useful in comparisons between institutions, although the direction of trends may be meaningful. It will not be useful for all classes of patients, particularly the infrequent high-risk patients. Risk stratification promotes the necessary homogeneity of study groups but may create inadequate sample sizes. Time periods must be selected with subgroup sizes in mind, and at least 10 time periods are necessary for statistical validity. This is a longer time horizon than desirable, and initially corrective actions will be triggered by identification of "extrinsic" sources of variability. With accumulating experience "intrinsic" sources of variability, identified by the rules given, can become targets for correction. This is a neat system-measure/adjust/remeasure. Can it have value in your practice and mine? There is only one way to know. Check it out.
Related Article
-
Applications of Statistical Quality Control to Cardiac Surgery
- David M. Shahian, Warren A. Williamson, Lars G. Svensson, Joseph D. Restuccia, and Richard S. D'Agostino
Ann. Thorac. Surg. 1996 62: 1351-1358.
[Abstract]
[Full Text]
