Guideline Development Policies
American College of Gastroenterology
Practice Parameters Committee
ACG Guideline Creation Process
Below is a brief guide to the process of creating an ACG guideline which was approved by the ACG Board of Trustees in January 2010.
The steps to guideline creation are as follows:
- Authors will have a conference call with subgroup of the ACG Practice Parameters Committee. During this call, the group will isolate the central issues in the diagnosis and management of the relevant disease state; for example, “what is the role of biological agents in the maintenance of remission of Crohn’s disease?” These pragmatic issues will serve as the headings of the sections of the document. Generally, they will number between 8-12 items, collapsing issues into larger headings as needed.
- For each of the crucial clinical questions, the authors will identify recent, relevant systematic reviews of the evidence.
- In cases where a recent systematic review is available with which to assess the evidence, the authors will cite and use this review as the evidence base for the question. When a recent systematic review is not available, the authors will perform an explicit review of the literature, stating search terms, databases accessed, period searched, “hits” obtained, relevant papers extracted, and describe the process by which the content was filtered for relevant articles (for example, “we considered only randomized controlled trials and case-control studies”). Because there is substantial heterogeneity in the literature with respect to available data for clinically relevant questions, for some questions, it is expected that cohort studies and even case series may form the foundation of the data available. For other questions, the authors may be able to limit their data to randomized controlled trials and meta-analyses. Whatever the state of the evidence, the authors will describe how they found it, and that they did it. In some circumstances, the authors may wish to perform formal systematic review of the literature, with two abstractors independently extracting the evidence. While this process is time-consuming and costly, and not feasible for every clinical question posed, it may be desirable for certain highly contentious, rapidly evolving, or extremely important issues.
- After assimilating and reviewing the data, the authors will generate a number of recommendations regarding the diagnosis and management of subjects with the condition. Generally, it is expected that between 1-4 recommendations will be generated from each document section, although more or fewer may be appropriate in unusual situations.
- Each recommendation will be generated by a process known as the “nominal group technique.” In this process, the authorship group first discusses the goal of the recommendation. Then each member of the group writes one or more statements that they feel best expresses the goal of the recommendation. These statements are disseminated, without attribution of author, among the authors, who then rank the statements, first, second, third, and so on. The statement with the lowest point total is deemed to best express the consensus of the group, and is endorsed.
- Concurrent with the generation of the recommendation statements, the statements are also graded for both strength of evidence, and strength of recommendation. These ratings are related, but distinct. For instance, a recommendation that has a great deal of high quality evidence substantiating a very small benefit at some cost or risk to the patient may generate only a weak recommendation. Alternatively, a recommendation based on only weak evidence, but with little risk or cost may still generate a strong recommendation in its favor.
- The authors will categorize recommendations using the GRADE system rankings. This system grades evidence into the following categories:
- High – further research is very unlikely to change our confidence in the estimate of effect;
- moderate – further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate;
- low – further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate; and,
- very low – any estimate of effect is very uncertain.
- The recommendations are then categorized into two areas: strong or weak (it is permissible to use “conditional” or “discretionary” in place of the term “weak”).
- The guidelines, in addition to grading strength of evidence and strength of recommendation, will allow for dissent from the majority opinion by one or more authors. This will simply be recorded by 0 dissent, 1 dissent, etc.
- Flow diagrams demonstrating recommended diagnostic and/or therapeutic algorithms are encouraged. The guidelines should explicitly answer, “What is the best management of these patients?”
- TIMELIME: The first iteration of the guidelines document will be expected to be returned 6 months from the commissioning of the guideline. In an effort to make our new guidelines as “fresh” as possible when published, we have created a special guideline review process, involving members of the Board of Trustees, Practice Parameters Committee and the American Journal of Gastroenterology. It is our goal to review the guideline, allow you to revise the guideline, and re-review the guideline within 6 months of first submission. Therefore the entire process should take 1 year from commission to finished, accepted guideline.