A search of only four key databases would identify most randomized controlled trials of acupuncture: A meta-epidemiological study

Little research has been conducted to assess which specific databases should be searched when performing a systematic review (SR) on acupuncture. The current study aimed to identify key databases and the optimal database combination to retrieve randomized controlled trials (RCTs) on acupuncture for inclusion in SRs. A systematic search for SRs in the field of acupuncture was conducted in order to identify target databases and RCTs were extracted from the SRs that had searched all target databases. The proportions of SRs that had achieved 100%, 95%, or 90% recall of RCTs and the total recall of RCTs in various combinations of target databases were calculated. Sensitivity analysis was performed on those SRs that included 10 or more RCTs. CNKI, WanFang, VIP, PubMed, CENTRAL and Embase were regarded as target databases. A total of 4349 acupuncture RCTs were extracted from 286 SRs. Searching all six target databases retrieved 99.3% of all RCTs while 99.1% were recalled by searching the combination of CNKI, WanFang, PubMed and CENTRAL. There were no significant differences on total recall of RCTs (p = 0.549) or in the proportion of SRs with 90% recall of RCTs (97.2% vs. 97.6%; p = 0.794) between searching the above four and the full six target databases. Most results were similar in the sensitivity analysis. The combined retrieval power of CNKI, WanFang, PubMed and CENTRAL was considered an efficient choice to retrieve acupuncture RCTs included in SRs.