Wens et al. (1) comment on the value of using multiple databases to search the literature for my report (2). However, they do not mention another technique that I have found to be valuable: checking the reference lists of articles found in database searches and other sources. This manual method overcomes reliance on electronic searches by key word. Relevant reports are often missed in electronic searches because they are focused on another end point (e.g., efficacy, adverse events, etc.), so the list of key words does not include terms for medication compliance/adherence or persistence. Careful reading of the article will lead to a relevant article containing pertinent data.
The results of the multimodal search noted by Wen et al. was not for adherence/compliance or persistence data but for another topic. They do not actually demonstrate flaws in the search that I used for my report on adherence with medications for diabetes. Had they performed the same search, they might have found fewer differences because my search required that articles include specific data on adherence/compliance rates or persistency duration. Of course, the time lag between performing a search and publication of the report is lengthy. I also found additional articles that were published during this period that have yet to be reported. The results of my previous literature searches on adherence/compliance rates have been consistent over time, with little difference in my overall finding that patients take approximately three-fourths of medication as prescribed (3), particularly when electronic dose monitoring is used as the method (4).