Abstract

To investigate the impact of omitting Chinese-language databases from systematic literature reviews (SLRs) on the results of subsequent meta-analyses. This research will inform the feasibility and effect on quality of including Chinese-language databases in global reviews. A published SLR and meta-analysis, on the treatment of negative symptoms in patients with schizophrenia using repetitive transcranial magnetic stimulation (rTMS) (Wang J et al. Shanghai Arch Psychiatry. 2017;29(2):61-76) was identified. Wang et al. searched both English (MEDLINE, Embase, Cochrane Library and Web of Science) and Chinese (Chinese National Knowledge Infrastructure, Wanfang Data, VIP, SinoMed and Airiti Library) language databases and included similar proportions of English (13/29) and Chinese (16/29) studies. Using the same approach as Wang et al., the meta-analysis was repeated with Chinese studies omitted. As in the original study, overall there was a significant difference between rTMS plus antipsychotic drugs compared to control (standardized mean difference [SMD] -0.4039, CI -0.6543– -0.1535; p<0.0001). When the Chinese and English studies were analysed separately, the overall effect was significant in the Chinese studies (SMD -0.5772, CI -0.8505– -0.3040; p<0.0001) but not significant in the English studies (SMD -0.1501, CI -0.6253–0.3251; p=0.5359). There was no clear relationship between study language and baseline age, disease course or Positive and Negative Syndrome Scale score, suggesting it was unlikely that the differing results were due to varying baseline characteristics between Chinese and English studies. Simulations were also performed to assess whether these results could be seen by chance and this was found to be unlikely. The inclusion of Chinese language studies in SLRs and meta-analyses may substantially influence the results and subsequent decisions based on these data. Future SLRs from a global perspective may therefore consider the inclusion of Chinese-language databases in their search strategies to ensure maximal relevant data are captured.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call