| ObjectiveBiomedical research literature databases are one of the important sources for searching evidence of clinical practice. Clinical end - user searching for related articles has risen dramatically for recent years, spurred by the development of databases on CD and on web. The most frequendy used databases searching for Chinese medical literatures are Chinese Biomedical Literature Analysis and Retrieval System ( CBM ) , China National Knowledge Infrastructure ( CNKI) and Chongqing VIP information (VIP). The clinical end-uses find that the imprecise search skills always bring difficulties to seek answers to clinical questions. This will lead to both missing sound studies in searching (low sensitivity) and retrieving many citations of studies that are not sound (low specificity and low precision ). This problem is especially prominent when searching for evidence-based literatures ( EBM). In order to searching EBM literatures effectively, PubMed has established a set of built - in methodological search filters, Clinical Queries , which can improve the detection of high quality studies on four clinical categories: therapy, diagnosis, etiology, and prognosis. This search filter has been widely utilized by clinicians for its easy use and accurate searching results. However, there are few reports on the searching methodology of clinical archives in Chinese. This may account for the low quality of Chinese journal articles in their research design and paper writing, and of abstract and indexing of Chinese literature databases. Our objectives are firsdy design a set of clinical literatures search strategies on etiology, diagnosis, therapy and prognosis for CBM and for Chinese online databases. Secondly design a set of search strategies for articles met key methodologic criterion on etiology, diagnosis, therapy and prognosis forthe two kinds of databases. Thirdly study on methodology of information retrieval. Material and methodThe manual review of the literature is served as the " gold standard" against which database search strategies (the diagnostic tests). We selected original articles about disease cause, diagnosis, treatment and prognosis from 31 journals. The articles were downloaded from CBM. We selected potentially useful words through a word frequency analysis and determined the frequency of all the words in the title, abstract, and subject indexes. All the selected journals functioned as a closed database. The sensitivity, specificity, precision of all the high frequency words were calculated and the high frequency words of large sensitivity x precision were viewed as final searching words. All the combined searching strategies resulted which consisted of all searching words and title field and abstract field! Meanwhile, the sensitivity, specificity, precision, and NNR(number needed to read) were calculated. Among the strategies, those comprised of all searching words would be used in Chinese disc database, those of title and abstract words in Chinese Web databases. The best strategies were the one of high sensitivity and the one of high specificity in each group.Result2570 articles were selected. Of those met key methodological criterion, 10 were on etiology, 10 on diagnosis, 45 on treatment, 7 on prognosis. In terms of subject, 203 were on etiology, 287 on diagnosis, 577 on therapeutics, and 81 on prognosis. The sensitivity of the final strategies emphasizing sensitivity was 0.96 -1.00, specificity 0.49 - 0.90, NNR 2.7-11.7. The specificity of the final strategies emphasizing specificity was 0.93 - 0.99, precision 0.24 - 0.71, sensitivity 0.40 - 0.95, NNR 1.4-4.1.ConclusionThe method was optimal to the design of Chinese search filter. The result of our study was reliable compared with those related reports.
|
PDF Full Text Request