|Home | About | Journals | Submit | Contact Us | Français|
This study searched two different online knowledge sources in order to evaluate their content coverage. UpToDate® was selected as the subscription resource and the National Guidelines Clearinghouse was selected as the publicly available resource. Using the Clinical Questions Collection, it was determined that UpToDate covered 10.4% more content although neither resources provides complete coverage. Current practice guidelines address at most two-thirds of clinical questions.
In 2001 Alper and his colleagues studied 2 experienced physician searchers and found that no single knowledge resource answered more than 70% of clinician questions.
This study will estimate how often answers may exist in current knowledge resources. We have two secondary aims. First, compare the difference in the content coverage of a subscription knowledge resource when compared to a publicly available knowledge resource. Second, determine how well any single resource can answer the questions.
Clinical questions were chosen at random from the NLM’s Clinical Questions Collection. The keywords from the questions were used to search UpToDate and NGC in a semi-automated manner. Results were stored in an Access™ database. Content found was recorded as a dichotomous variable, none found vs. content found.
UpToDate returned content in 75.9% of cases, whereas NGC returned content in 65.5% of cases. This 10.4% difference is significant at p<0.000. Combining the results, 89.9% of the content was covered. The only characteristic of the clinical question that predicted finding content was Disease Category. We observed that automated searching of the UpToDate was hindered by UpToDates’s use of a non-standard vocabulary and inability to accept two search terms simultaneously.
The difference in coverage may be explained by one resource having a controlled editorial process versus an open, undirected submission process. The 66% coverage by guidelines suggests an upper limit on the proportion of clinical questions that can be addressed with interventions such as CPOE and alerts. Difficulties in automated searching of resources with non-standard search methods may inhibit automated linkage to the EMR.
There may be a need to provide resources with a controlled editorial process, even if a subscription is required, to providers in medically underserved or other areas of need. Knowledge source producers need to be aware of content coverage lapses in certain disease categories and adjust their development efforts accordingly.