How to structure clinical practice guidelines for continuous quality improvement?

The purpose of this study was to evaluate the relevance of available practice guidelines to clinical quality improvement programs. A sample of 19 guidelines was evaluated in four prominent primary care areas. Two research assistants independently coded the clinical conditions and recommended/not recommended procedures abstracted from the guidelines (Cohen's kappa .67 and .50, respectively). An average of 35.1 (+/- 25.8) medical conditions and 48.4 (+/- 41.5) clinical procedures were defined by the guidelines. Most conditions were defined by using ICD-9-CM, age/sex group, or therapy, but 29% of definitions included symptoms which are not coded routinely. CPT codes alone were unable to identify most procedures. AHCPR guidelines mentioned significantly more procedures (p < .001) and fewer symptoms (p < .001) per clinical condition than other guidelines. The difficulty of finding codes for conditions and procedures, the high rate of non-codable items, and the lack of recommended measures limit the applicability of published clinical practice guidelines to continuous quality improvement programs.