to the editor: As Dr. Ebell points out, it is well documented in the literature that questions arise from patient encounters and that most of these questions are not researched and are left unanswered. However, what he neglects to address is that when the questions are pursued, the literature provided answers to more than 80 percent of these questions.^1,2

Dr. Ebell comments that PubMed should not be the first place to search for answers because “there is so much information it is like trying to drink from a firehose.” This phrase, coined by Clifford Stoll in an interview about his 1995 book Silicon Snake Oil,³ was made in reference to searching the World Wide Web—not for searching a well-constructed, refined database such as PubMed. Indisputably, PubMed is a large database; it contains more than 18 million citations and indexes about 5,200 journals. It may also be one of the most underappreciated and underused resources freely available to physicians and researchers.

Although time will always be a limiting factor or barrier for physicians in their pursuit of answers to clinical questions, a lack of searching skills should not stop them. As expert searchers, medical librarians are in a unique position to teach and demonstrate how to effectively navigate PubMed and other databases. Learning how to fine-tune literature searches takes practice and patience. Proper use of MeSH, Clinical Queries, My NCBI, and other tools can improve searching speed, success, and satisfaction.

In addition to using PubMed and Google to keep abreast of new studies and current trends, a subscription to a news reader is invaluable. A reader is a free online service that can be tailored to find blogs, RSS (Really Simple Syndication) feed subscriptions, journal tables of contents, clinical trials, health news, or virtually any topic of interest. These feeds can then be read from a single Web site. Medical librarians can offer advice and assistance in using the readers, managing subscriptions, and honing topics of interest.

IN REPLY: I share Ms. O'Brien's appreciation for the depth and breadth of the PubMed database. However, it is important to note that a study found that it took physicians an average of 24 minutes to answer a question that remained unanswered at the end of a day of practice using PubMed.¹ As I noted in my article, utilizing the Clinical Queries and Related Articles features on PubMed is much more useful for the busy physician. But, it should not be the first place we look for answers. Sources of pre-appraised, pre-filtered evidence are simply much more efficient, especially when they are regularly updated and incorporate calculators and interactive decision support tools.

I disagree that a steady diet of preconfigured PubMed searches for areas of interest, RSS feeds, and e-mailed tables of contents is a good medical information strategy for most practicing physicians. Those sources can tell us what is new, but they cannot tell us what is most important or useful. Because most physicians read no more than 30 to 60 minutes per week, it is critical that they focus on information that is relevant (reports patient-oriented outcomes) and valid (free of intentional and unintentional biases) in a way that takes as little time as possible.^2,3 Although medical librarians can be important partners for researchers and for physicians when a search is particularly difficult or problematic, it is unrealistic to think that they have enough time to help physicians answer all or even many of their questions. The typical physician has about 10 questions per day, of which about one half go unanswered;^4–6 that is more than 2 million unanswered questions per week by family physicians alone. Obviously, there is much work to be done by physicians, medical librarians, and informaticists to help us do a better job of answering more of these questions with the best available evidence.