Ağ Meta Analizinde Tutarlılık Varsayımını Etkileyen Faktörlerin Araştırılması: Bir Benzetim Çalışması

Abstract

Objective: This study aimed to examine the factors affecting consistency assumption with simulated data for network meta-analyses. Network pattern, number of studies per comparison, individual study sample sizes, and probabilities of the event were investigated that may affect the consistency assumption. Material and Methods: Data were produced with R 4.1.0 for the combination of different sample sizes (N: 100, 150, 200), the different success probability of three treatments (p1, p2, p3 changes in interval 0.01-0.90), and the number of study per comparison (M=5, M=10, M=20, M=30). Then the mean and standard error of ratios of odds ratios (ROR) were calculated and this process was repeated 1,000 times. Results: We found that the success probability of treatments and the number of studies in the network affected the inconsistency assumption more than the study sample size for N=100 and N=150. Also, the results indicated that in sample size 200, the study sample size affected the ROR values in addition to the other factors. Conclusion: Network meta-analysis is wide-spreading in recent years. Therefore, it is important to obtain reliable results with the providing assumptions. Especially, consistency assumption must be considered and the researchers may realize the affecting factors of the consistency assumption with this study.