FAQs About UpToDate Clinical Intelligence

UpToDate uses a systematic process to identify, evaluate, and synthesize new clinical evidence in the context of real clinical questions. More than 50 in-house physician deputy editors, trained in clinical epidemiology, help maintain the integrity and consistency of the editorial process that underpins UpToDate Clinical Intelligence.

UpToDate does more than retrieve or summarize available evidence. Its global network of over 7,600 clinical experts evaluates new findings in the context of established practice, reconciles uncertainty, and translates evidence into actionable recommendations. This governed expert judgment provides the trusted intelligence beneath both traditional decision support and AI-enabled experiences.

In the AI era, that principle remains essential: useful clinical intelligence must connect evidence with expert judgment, patient circumstances, and transparent reasoning rather than treat information retrieval alone as clinical decision-making.

UpToDate follows a hierarchy of evidence consistent with most evidence-based resources and we always use the highest quality clinical evidence available.

To create and update the clinical topics in UpToDate with new clinical evidence, our editorial staff performs comprehensive reviews of the medical literature and considers the quality of the study, the hierarchy of evidence, and its clinical relevance.

When current, high-quality systematic reviews are available, UpToDate topics and recommendations rely heavily on these reviews. When such reviews are unavailable, UpToDate summarizes the key studies bearing on the clinical issues at hand. Systematic reviews and the design of primary studies (e.g., randomized trials, observational studies) are often identified explicitly in the text, with the relevant data provided.

However, in cases where either the type of study or the data are not stated explicitly, users can click on the reference and bring up the Medline abstract to obtain this information.

Evidence is derived from a number of resources, including but not limited to:

UpToDate makes clinical recommendations that are always based on the best available evidence. Unsystematic clinical observation (i.e., clinical experience) is the weakest form of evidence, but it is evidence nonetheless. A critical component of our recommendations is that we are transparent about the quality of the evidence and the strength of each recommendation. In instances where there is poor, limited, or no data available, we still feel compelled to answer a clinical question based on clinical experience or pathophysiologic reasoning rather than avoid answering it altogether. Most clinical solutions do not have the depth of expertise from their contributors that UpToDate has, and therefore cannot adequately answer clinical questions that don’t have a strong evidence base (which accounts, unfortunately, for a substantial proportion of clinical questions).

Our world-renowned physician authors and editors use their expertise to understand the impact of new evidence on existing best practices. Their clinical knowledge enables UpToDate to apply new evidence to specific situations that arise during clinical practice. This does not mean that they express opinions that supersede the available evidence — their experience simply makes them better qualified to interpret new findings in the context of patient care.

In fact, a fundamental principle of evidence-based medicine — as described by Dr. Gordon Guyatt, who is credited with coining the term and is a thought leader in the field of EBM — is “Evidence alone is never sufficient to make a clinical decision. Decision makers must always trade the benefits and risks, inconvenience, and costs associated with alternative management strategies, and in doing so consider the patient’s values.”1 Expertise is thus required to move from evidence to recommendations.

Because recommendations for care need to account for all of the factors cited above, many evidence-based solutions avoid making specific recommendations for patient care. UpToDate has taken a different approach. It is the policy of UpToDate to make specific recommendations for patient care whenever possible.

Recommendations in UpToDate are based on a synthesis of evidence, including that obtained from clinical trials as well as clinical experience. And, whenever possible, the evidentiary basis for recommendations is stated explicitly. When there is no published systematic evidence available (e.g., prednisone dosing regimen in pulmonary sarcoidosis), recommendations are based on the unsystematic clinical observations of our experts and reviewers, and on pathophysiologic rationale. Our multiple layers of peer review prevent individual opinion from becoming a part of any recommendation in UpToDate.

UpToDate clinical intelligence is designed to keep trusted evidence and expert clinical judgment as the source of clinical authority. Rather than asking AI to independently interpret medical information and create clinical judgment, UpToDate leverages AI to help clinicians access, contextualize, synthesize, and communicate governed clinical intelligence.

This means the technology can evolve while the foundation remains grounded in expert-authored, peer-reviewed content, defined reasoning boundaries, transparent provenance, and human accountability. AI helps make trusted clinical intelligence more accessible without removing the clinical expertise behind it.

Since 2001, UpToDate has worked extensively with Dr. Guyatt to improve the clarity and transparency behind the clinical evidence we use and the strength of our recommendations. Dr. Guyatt is a founding member of the GRADE Working Group, which was created to develop “a common, sensible and transparent approach to grading quality of evidence and strength of recommendations.”² Many international leaders, including the World Health Organization, Cochrane and the British Medical Journal, collaborated in the development of this system and have adopted it.² The GRADE system and its modifications have become the most widely used clinical recommendation grading systems in the world.

In 2006, UpToDate began grading recommendations using the GRADE format. This is a continuing process, with thousands of graded recommendations in the program. Graded recommendations appear in the Summary and Recommendations sections of each topic.

UpToDate dedicates considerable resources to ensuring that our recommendations are clear, complete, and correct. In addition to our routine peer review process, our CoExecutive Editor and/or Deputy Editors review all assigned grades to confirm that they accurately reflect the quality of the evidence and the strength of the recommendation. In addition, Dr. Guyatt comes to the UpToDate office in Massachusetts regularly to continue to educate our physician editors about evidence-based medicine and grading It is important to understand the difference between grading evidence and grading recommendations. UpToDate does both, which makes it unique among point-of-care solutions, which typically grade only the quality of evidence or even just individual studies. However, there are occasions when strong recommendations may be made in the face of low-quality evidence, or weak recommendations in the face of high-quality evidence.

For example:

• Weak recommendations with high-quality evidence occur when individual patient values and preferences make a decision a close call. For example, we have high-quality evidence that lifelong anticoagulation after idiopathic DVT reduces the risk of recurrence, but many patients are unwilling to accept the ongoing burdens of warfarin therapy in return for such a benefit. In this case, UpToDate makes a weak recommendation for anticoagulation based on high-quality evidence.
• Strong recommendations based on low-quality evidence most often occur when there is high or moderate-quality evidence for either benefits or harms of a therapy but not for both. For example, we have moderate to high quality evidence that a number of different classes of antihypertensives are similarly effective, and low-quality evidence that ACE inhibitors are harmful in pregnancy. In such a situation, we would make a strong recommendation based on low-quality evidence not to treat hypertensive pregnant women with ACE inhibitors.

The GRADE system allows transparency in all of these situations.

UpToDate only grades recommendations about treatment or screening because we feel that no existing system for grading diagnostic recommendations works well enough to be useful to clinicians. Rather than grade diagnostic recommendations, UpToDate describes the evidence underlying such recommendations in the text of the topic. UpToDate is participating in efforts to create a clinically useful system for grading diagnostic recommendations.

 Validation should go beyond whether an AI-generated response sounds accurate. UpToDate evaluates whether the clinical substance of AI-enabled responses remains faithful to the underlying UpToDate source and whether claims or recommendations have been introduced, altered, or extended by the model.

UpToDate systematically measures performance across more than 25 medical specialties, including challenging clinical scenarios, and uses routine audits by human experts to identify weaknesses, monitor performance, and support continuous improvement. This helps ensure that clinical integrity remains a measurable requirement as AI capabilities evolve.

UpToDate is updated daily following a continual comprehensive review of the resources listed on the second page (peer-reviewed journals, clinical databases, etc.). Topics in UpToDate are revised whenever important new information is published, not according to any specific time schedule. Updates are integrated carefully, after extensive peer review, with specific statements as to how the new findings should be applied to clinical practice.

Updates that are deemed particularly important by our authors and editors (i.e., those that may change practice, drug alerts, etc.) are highlighted in our “What’s New” section and a subset that are considered practice-altering are featured in our “Practice Changing UpDates” section.

Trust in clinical AI requires more than citations. Clinicians should be able to understand the source of an answer, inspect the evidence and expert guidance supporting its clinical claims, and recognize where uncertainty remains.

UpToDate Expert AI provides line-by-line traceability between responses and underlying UpToDate content. This explicit provenance helps clinicians examine the basis of an answer rather than asking them to trust an AI-generated response at face value, supporting decisions they can understand, explain, and stand behind.