Clinical guidelines, or clinical practice guidelines, are "systematically developed statements to assist practitioner and patient decisions about appropriate health care for specific clinical circumstances" (1). These statements are evidence-based and attempt to support best practices for managing specific conditions (4). Recommendations often have a rating system based on the quality of their evidence. The highest quality evidence comes from randomized controlled trials, though all literature can be considered as well as consensus opinion (3). Guidelines differ from protocols (3). Guidelines are suggestions and recommendations designed to assist the provider. Protocols, however, are more rigid and comprehensive.
Elements of a guideline include (2, 3):
- Patient state/context - uses intrinsic variables to describe the patient.
- Eligibility criteria - decides which patients the guideline applies to.
- Goals - overall goals of the guideline.
- Classification schemes/Risk calculators - validated tools designed to calculate a patient's risk of an outcome given a certain clinical context, for example the Framingham Coronary Heart Disease Risk Score.
- Decisions & Actions - options offered to the provider, including recommendations on treatments, diagnostic tests, and alternatives.
- Evidence - primary literature supporting recommendations given by the guideline.
- Cost-benefit analysis
There are many different types of guidelines (5). They can provide recommendations on diagnosis and treatment. Treatments can involve behavioral modifications, medications, or surgical procedures. Ohno-Machado et. al. identified five broad categories of guidelines (5):
- Screening and prevention of disease
- Diagnosis and pre-diagnosis management
- Indications for surgery
- Appropriate use of tests
- Guidelines for care
GUIDELINE STANDARDIZATION & QUALITY
Many tools have been developed to promote standardization of guidelines, and to assess their quality (3). The AGREE instrument, for example, uses a four-point Likert scale to assess six quality domains: scope/purpose, stakeholder involvement, rigor of development, clarity/presentation, applicability, and editorial independence. Another instrument is the COGS (conference on guideline standardization) checklist, which similarly analyzes guidelines to assess their components.
The GLIF (guideline interchange format) specification was developed also with the goal of standardizing guidelines and easing their transition between institutions (5).
Characteristics of high quality guidelines include (3):
- Validity - do the recommendations lead to the intended outcome?
- Reliability - will different users interpret the recommendations similarly?
- Reproducability - would other guidelines offer similar recommendations given the same data?
- Implementability - can the recommendations be easily implemented in practice?
- Executability - do the recommendations communicate what to do?
- Decidability - do the recommendations state when to do it?
Computer-interpretable guidelines are the result of incorporating clinical guidelines into EHRs (2). This process is important to facilitate widespread and efficient use of guidelines. When building these tools, focus should be placed on flexibility so that the guideline can be as patient-specific as possible (1). This can be done by allowing entry into the guideline from a variety of different points based on varied eligibility criteria. Decision criteria should also be provided when considering switching between alternative plans (2). Regarding implementation, one strategy is to break the guideline down into its component parts, then implement these parts in various different clinical settings.
Attention should be paid to the formatting of the guidelines (5). Because guidelines involve the input and output of extensive patient health information, it's important that the data be easy to analyze and report on. This can be done through the use of tables, flowcharts, and if/then/else statements (5).
Medical logic modules (MLMs) are used to help implement computer interpretable guidelines (5). They typically consist of the data slot (maps terms between locations), logic slot (specifies what must be true for an action to occur), evoke slot (controls the logic slot), and the action slot (the recommended action). After creating multiple MLMs, they can be combined to represent a complex guideline. Arden syntax is typically used in the creation of MLMs (2).
One issue with guidelines is that many do not account for patient comorbidities that may affect the suggested pathway (1). For example, a guideline for chronic kidney disease may focus specifically on the renal system, but not take into account concomitant liver disease present in certain populations that could change the recommendations.
Another issue with clinical decision support in general which also affects guidelines is the curly braces problem (2). This is a problem in Arden syntax that arises when tools are attempted to be shared between institutions. Guidelines can suffer from similar issues when multiple institutions have specific needs.
(1) Greenes, Robert A. Clinical Decision Support: the Road Ahead. 1st ed., Academic Press, 2011.
(2) Sonnenberg, F., Hagerty, C. Computer-interpretable clinical practice guidelines. Where are we and where are we going? Yearb Med Inform 2006: 145-58.
(3) A. Latoszek-Berendsen, H. Tange, H. J. van den Herik, A. Hasman. From Clinical Practice Guidelines to Computer-interpretable Guidelines. Methods Inf Med 2010: 49(06).
(4) Qaseem A, Forland F, Macbeth F, Ollenschläger G, Phillips S, van der Wees P, et al. Guidelines International Network: Toward International Standards for Clinical Practice Guidelines. Ann Intern Med. ;156:525–531.
(5) Ohno-Machado L, Gennari JH, Murphy SN, et al. The GuideLine Interchange Format: A Model for Representing Guidelines. Journal of the American Medical Informatics Association : JAMIA. 1998;5(4):357-372.
Submitted by Christopher Tenore, MD