To the Editor:

Thank you for including UpToDate in Dr. Brian Alper’s article, “Practical Evidence-Based Internet Resources.” We would like to address a few points about UpToDate’s editorial processes.

The article states that UpToDate does not include the Cochrane Library; however, the data cited in UpToDate is derived from a number of resources, including The Cochrane Database, Clinical Evidence, consensus guidelines and over 270 peer-reviewed journals. The article also states that UpToDate does not describe any systematic process for selecting the research included in our reviews, but we do have a systematic method for identifying and analyzing relevant literature.

Each topic has an author who is an expert in the area and at least two other physician reviewers. This group works together to screen and select studies for presentation based upon the following hierarchy of evidence:

• Systematic reviews of randomized trials,

• Single randomized trial,

• Systematic review of observational studies,

• Single observational study,

• Unsystematic clinical observations.

The recommendations are carefully linked to the available evidence, and when there is insufficient evidence from controlled clinical trials and other sources of data, UpToDate explicitly states that the recommendations are the authors’.

All of the topic reviews in UpToDate are revised whenever important new information is published. Updates are carefully integrated into UpToDate, with specific statements as to how the new findings should be applied clinically. Each topic review has a date indicating the most recent author review, and approximately 30 percent are updated during each four-month cycle. The new material in each topic review is underlined for easy identification, and updated and new topic reviews can be recognized by choosing the “Show Updated Topics” option in the table of contents. In addition, UpToDate is currently working with Gordon Guyatt at McMaster University to standardize our presentation of evidence.

Author’s response:

A systematic process is necessary to identify the best available evidence. Selecting evidence based on the systematic evaluation of study methodology reduces bias that occurs when expert authors select articles based on individual choices. One approach is systematic searching using explicit search protocols. Clinical Evidence and PDxMD (for some content) use this approach. DynaMed, InfoRetriever and UpToDate do not use this approach for most content because of the increased effort required.

Another approach is systematic literature surveillance using explicit protocols for article selection. With this approach, the authors and reviewers may modify the content, but article selection is protocol-driven. DynaMed and InfoRetriever use this approach, which allows coverage of a broader range of content yet systematically selects for the best available evidence. This is different from a process that identifies articles but allows authors to selectively choose what and how to cite, negating the systematic approach.

For example, a CDC position-paper states that antibiotics produce only a small benefit in acute sinusitis, based on five placebo-controlled randomized trials and two meta-analyses. It also states that broad-spectrum antibiotics are no more effective than narrow-spectrum antibiotics, based on three meta-analyses.^1,2 However, the UpToDate review of acute sinusitis states there is solid evidence supporting antibiotics to reduce illness duration based on a single citation, which is neither a scientific study nor systematic review.^3,4 It recommends broad-spectrum antibiotics, citing disagreement with the “consensus guidelines” without disclosing their underlying evidence.

Hopefully UpToDate’s work with Gordon Guyatt will lead to standardization in the way evidence is selected and not just the way it is presented.