On 2015 Jul 17, Andrea Messori commented:

Cumulative pairwise meta-analysis to evaluate the risk of infection in patients with rheumatoid arthritis treated with biological agents as compared with DMARDs: the difference is not significant using the random-effect frequentist model.

By Andrea Messori

HTA Unit, Regional Health Service, 50100 Firenze (Italy)

In the article by Singh and coworkers [1], biological agents were compared with traditional disease-modifying antirheumatic drugs (DMARDs). This comparison showed an increase in serious infections for patients treated with biological agents. To reach this conclusion, a frequentist cumulative meta-analysis (shown in Figure 4 of ref.1; 59 trials) was carried out in which the summary odds-ratio (calculated according to the Mantel-Haenszel fixed-effect [FE] model) was estimated to be 1.27 (95% confidence interval [CI]: 1.05 to 1.52).

Since Singh et al did not test one of the most commonly used meta-analytic models (i.e. the random-effect [RE] model according to Der Simonian and Laird[2]), we have re-analysed the data shown in Figure 4 of Singh’s article by running the above mentioned RE model as implemented in the Open Meta-Analyst software (version 4.16.12, Tufts University, url http://tuftscaes.org/open_meta/).

Our results are presented in the figure available at http://www.osservatorioinnovazione.net/papers/singhs-reanalysis.jpg

The odds-ratio that we obtained (1.25; 95%CI: 0.92 to 1.71; p=0.16; Figure) differs from the values reported by Singh et al. [1] mainly because “our” odds-ratio remained far from the threshold of statistical significance, whereas all the odds-ratios reported by Singh et al. satisfied this criterion.

References

[1] Singh JA, Cameron C, Noorbaloochi S, Cullis T, Tucker M, Christensen R, Ghogomu ET, Coyle D, Clifford T, Tugwell P, Wells GA. Risk of serious infection in biological treatment of patients with rheumatoid arthritis: a systematic review and meta-analysis. Lancet. 2015 May 11. pii: S0140-6736(14)61704-9. doi:10.1016/S0140-6736(14)61704-9. [Epub ahead of print]

[2] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–87.

On 2015 Jul 17, Andrea Messori commented:

Cumulative pairwise meta-analysis to evaluate the risk of infection in patients with rheumatoid arthritis treated with biological agents as compared with DMARDs: the difference is not significant using the random-effect frequentist model.

By Andrea Messori

HTA Unit, Regional Health Service, 50100 Firenze (Italy)

In the article by Singh and coworkers [1], biological agents were compared with traditional disease-modifying antirheumatic drugs (DMARDs). This comparison showed an increase in serious infections for patients treated with biological agents. To reach this conclusion, a frequentist cumulative meta-analysis (shown in Figure 4 of ref.1; 59 trials) was carried out in which the summary odds-ratio (calculated according to the Mantel-Haenszel fixed-effect [FE] model) was estimated to be 1.27 (95% confidence interval [CI]: 1.05 to 1.52).

Since Singh et al did not test one of the most commonly used meta-analytic models (i.e. the random-effect [RE] model according to Der Simonian and Laird[2]), we have re-analysed the data shown in Figure 4 of Singh’s article by running the above mentioned RE model as implemented in the Open Meta-Analyst software (version 4.16.12, Tufts University, url http://tuftscaes.org/open_meta/).

Our results are presented in the figure available at http://www.osservatorioinnovazione.net/papers/singhs-reanalysis.jpg

The odds-ratio that we obtained (1.25; 95%CI: 0.92 to 1.71; p=0.16; Figure) differs from the values reported by Singh et al. [1] mainly because “our” odds-ratio remained far from the threshold of statistical significance, whereas all the odds-ratios reported by Singh et al. satisfied this criterion.

References

[1] Singh JA, Cameron C, Noorbaloochi S, Cullis T, Tucker M, Christensen R, Ghogomu ET, Coyle D, Clifford T, Tugwell P, Wells GA. Risk of serious infection in biological treatment of patients with rheumatoid arthritis: a systematic review and meta-analysis. Lancet. 2015 May 11. pii: S0140-6736(14)61704-9. doi:10.1016/S0140-6736(14)61704-9. [Epub ahead of print]

[2] DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177–87.