Barillot M J, Sarrut B, Doreau C G
Service d'Information Médico-Pharmaceutique, Pharmacie Centrale des Hôpitaux, Paris, France.
Ann Pharmacother. 1997 Jan;31(1):45-9. doi: 10.1177/106002809703100106.
To compare nine on-line bibliographic databases to obtain bibliographic references on specific drug interactions.
Seven bibliographic databases were selected for their ability to provide information concerning drug interactions: EMBASE, MEDLINE, TOXLINE, BIOSIS, Chemical Abstracts (CAS), PHARMLINE, and International Pharmaceutical Abstracts (IPA). Two French on-line bibliographic databases (i.e., PASCAL, BIBLIOGRAPHIF) were also tested to compare them with the other international databases. Twenty drug interactions were selected randomly using the journal Reactions Weekly 1993.
The total number of references, the number of potentially relevant references, the number of case report references, the number of unique references in the total number of references, and the number of unique references between potentially relevant references were analyzed by using the Friedman two-way ANOVA by ranks. For each database, relevance and relative recall were calculated.
For the total number of references, EMBASE was significantly more comprehensive then all other databases (p < 0.05). EMBASE had a significantly greater number of potentially relevant references than IPA, PHARMLINE, CAS, and BIBLIOGRAPHIF (p < 0.05). For the total number of case report references, only one significant difference, between EMBASE and BIBLIOGRAPHIF (p < 0.05), was observed. MEDLINE and TOXLINE had the lowest cost per potentially relevant reference.
To obtain bibliographic references on drug interactions, the first step should be to search MEDLINE or TOXLINE; the second step, for completeness, should be to search EMBASE.
比较九个在线书目数据库,以获取特定药物相互作用的书目参考文献。
选择七个书目数据库,因其能够提供有关药物相互作用的信息:EMBASE、MEDLINE、TOXLINE、BIOSIS、化学文摘(CAS)、PHARMLINE和国际药学文摘(IPA)。还测试了两个法语在线书目数据库(即PASCAL、BIBLIOGRAPHIF),以与其他国际数据库进行比较。使用《反应周刊》1993年随机选择了20种药物相互作用。
通过使用弗里德曼双向秩方差分析,分析参考文献总数、潜在相关参考文献数量、病例报告参考文献数量、参考文献总数中的唯一参考文献数量以及潜在相关参考文献之间的唯一参考文献数量。计算每个数据库的相关性和相对召回率。
对于参考文献总数,EMBASE比所有其他数据库都显著更全面(p < 0.05)。EMBASE的潜在相关参考文献数量比IPA、PHARMLINE、CAS和BIBLIOGRAPHIF显著更多(p < 0.05)。对于病例报告参考文献总数,仅观察到EMBASE和BIBLIOGRAPHIF之间的一个显著差异(p < 0.05)。MEDLINE和TOXLINE每潜在相关参考文献的成本最低。
为获取药物相互作用的书目参考文献,第一步应搜索MEDLINE或TOXLINE;第二步,为了完整性,应搜索EMBASE。