Guideline maintenance and revision. 50 years of the Jones criteria for diagnosis of rheumatic fever
R. N. Shiffman
Center for Medical Informatics, Yale School of Medicine, New Haven, Conn., USA.
OBJECTIVE: To understand better the factors that led to revisions of the
Jones criteria, a widely used diagnostic guideline for diagnosis of
rheumatic fever. DESIGN: The original publication of the Jones criteria and
the four revisions were examined to identify changes. A computer software
maintenance paradigm was applied, and modifications were categorized as
corrective (error correction), perfective (enhancements in response to user
needs), or adaptive (responses to new knowledge). RESULTS: Modifications of
the Jones criteria were primarily corrective and perfective. Disease
characteristics, originally characterized as major manifestations, were
subsequently categorized as minor manifestations and vice versa. Twenty
years after the initial publication, a requirement was added to enhance
specificity (evidence for antecedent streptococcal infection). Descriptions
of rheumatic manifestations became more detailed over time to eliminate
ambiguous definitions and provide information to help clinicians decide
about borderline cases. This emphasis on corrective and perfective
maintenance contrasts with an expectation that adaptive changes would
predominate, as with most knowledge-based systems. In fact, despite 50
years of technologic and methodologic advances in medicine, only
echocardiography and new antibody testing contributed new knowledge that
bears on the diagnosis of rheumatic fever. CONCLUSIONS: Corrective and
perfective maintenance can be avoided by making effective use of knowledge
that exists at the time a guideline is published. Despite the apparent
durability of the Jones criteria, carefully structured, evidence-based
guidelines should require less corrective and perfective maintenance.
Adaptive maintenance can be anticipated if the quality of evidence or the
level of consensus that supports each recommendation is explicitly
recorded.