• Description of guideline
• How well is it followed?
• How was it developed?
• Do you agree with it?

Eddy suggests that guidelines are usually developed by "global subjective judgment", or based on "evidence" (without estimation of magnitude of outcomes), "outcomes" (without estimation of patient preferences) or "preferences."

He argues that a good method for developing a guideline should:

1. Be accurate
2. Be accountable
3. Be predictable: enable people to anticipate the consequences of the guideline (at individual and societal levels). This, he suggests, implies that guidelines must provide quantitative information about the magnitudes of risks and benefits.
4. Be defensible: Facilitate resolving conflicts among guidelines. This suggests the need for a metric in which to measure guidelines. How might we do this?
5. Be usable: Facilitate applying the guideline

He suggests that these requirements imply that an outcomes- or preferences-based approach to guidelines is necessary. We must objectively estimate the effects of an intervention on all relevant outcomes (possibly assessing patient preferences for the outcomes), and base our guideline on this assessment.

Eddy also says that guidelines should follow the same requirements as the methods for developing them. That is, a guideline should be accurate, accountable, predictable, etc. He even offers a nice outline of how a guideline should be written.

Do you agree with these recommendations? Have you seen guidelines developed this way? Why or why not?

A particular type of practice guideline is a prediction rule or scoring rule. In these guidelines, you make decisions based on a numerical index, often the sum of a set of observations or other judgments. We saw this early on with the CAGE rule for alcoholism -- add 1 point if you’ve thought about Cutting down, been Annoyed at people criticizing your drinking, felt Guilty about drinking, or had an "Eye-opener" in the morning. By setting a threshold on the CAGE score, it becomes a diagnostic test.

Ebell raises important issues in the use of decision rules. Was the rule designed with your population in mind? Are both the predictor variables and outcome variables precisely defined? (we know that physicians may have trouble consistently judging the predictor variables). Has the rule been appropriately validated?

Ebell lists many primary care decision rules. Are these familiar to the physicians or nurses in the class? Did any strike you as unusual, in appropriate, or less than useful?

Do you like this approach? Again, what is the role of the physician in using scoring rules? Where should clinical judgment come in?