Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritis
Abstract Background The importance of adjusting for cross-study heterogeneity in control group response rates when conducting network meta-analyses (NMA) was demonstrated using a case study involving a comparison of biologics for the treatment of moderate-to-severe rheumatoid arthritis. Methods Baye...
| Main Authors: | , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2019-10-01
|
| Series: | BMC Medical Research Methodology |
| Subjects: | |
| Online Access: | http://link.springer.com/article/10.1186/s12874-019-0837-2 |
| _version_ | 1819072305680089088 |
|---|---|
| author | Chris Cameron Abhishek Varu Arthur Lau Mahdi Gharaibeh Marcelo Paulino Raina Rogoza |
| author_facet | Chris Cameron Abhishek Varu Arthur Lau Mahdi Gharaibeh Marcelo Paulino Raina Rogoza |
| author_sort | Chris Cameron |
| collection | DOAJ |
| description | Abstract Background The importance of adjusting for cross-study heterogeneity in control group response rates when conducting network meta-analyses (NMA) was demonstrated using a case study involving a comparison of biologics for the treatment of moderate-to-severe rheumatoid arthritis. Methods Bayesian NMAs were conducted for American College of Rheumatology (ACR) 50 treatment response based upon a set of randomized controlled trials (RCTs) identified by a recently completed systematic review of the literature. In addition to the performance of an unadjusted NMA, a model adjusting for cross-study heterogeneity of control group response rates using meta-regression was fit to the data. Model fit was evaluated, and findings from both analyses were compared with regard to clinical interpretations. Results ACR 50 response data from a total of 51 RCTs and 16,223 patients were analyzed. Inspection of cross-study variability in control group response rates identified considerable differences between studies. NMA incorporating adjustment for this variability was associated with an average change of 38.1% in the magnitude of the ORs between treatment comparisons, and over 64% of the odds ratio changed by 15% or more. Important changes in the clinical interpretations drawn from treatment comparisons were identified with this improved modeling approach. Conclusions In comparing biologics for moderate to severe rheumatoid arthritis, failure to adjust for cross-trial differences in the control arm response rates in NMA can lead to biased estimates of comparative efficacy between treatments. |
| first_indexed | 2024-12-21T17:35:37Z |
| format | Article |
| id | doaj.art-e8c5cb4e56674c089b5539f128ad1b83 |
| institution | Directory Open Access Journal |
| issn | 1471-2288 |
| language | English |
| last_indexed | 2024-12-21T17:35:37Z |
| publishDate | 2019-10-01 |
| publisher | BMC |
| record_format | Article |
| series | BMC Medical Research Methodology |
| spelling | doaj.art-e8c5cb4e56674c089b5539f128ad1b832022-12-21T18:55:47ZengBMCBMC Medical Research Methodology1471-22882019-10-0119111010.1186/s12874-019-0837-2Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritisChris Cameron0Abhishek Varu1Arthur Lau2Mahdi Gharaibeh3Marcelo Paulino4Raina Rogoza5Data Analytics & Evidence Synthesis, Cornerstone Research Group, Inc.Data Analytics & Evidence Synthesis, Cornerstone Research Group, Inc.Division of Rheumatology, McMaster UniversityAmgen Inc.Amgen Canada, Inc.Amgen Canada, Inc.Abstract Background The importance of adjusting for cross-study heterogeneity in control group response rates when conducting network meta-analyses (NMA) was demonstrated using a case study involving a comparison of biologics for the treatment of moderate-to-severe rheumatoid arthritis. Methods Bayesian NMAs were conducted for American College of Rheumatology (ACR) 50 treatment response based upon a set of randomized controlled trials (RCTs) identified by a recently completed systematic review of the literature. In addition to the performance of an unadjusted NMA, a model adjusting for cross-study heterogeneity of control group response rates using meta-regression was fit to the data. Model fit was evaluated, and findings from both analyses were compared with regard to clinical interpretations. Results ACR 50 response data from a total of 51 RCTs and 16,223 patients were analyzed. Inspection of cross-study variability in control group response rates identified considerable differences between studies. NMA incorporating adjustment for this variability was associated with an average change of 38.1% in the magnitude of the ORs between treatment comparisons, and over 64% of the odds ratio changed by 15% or more. Important changes in the clinical interpretations drawn from treatment comparisons were identified with this improved modeling approach. Conclusions In comparing biologics for moderate to severe rheumatoid arthritis, failure to adjust for cross-trial differences in the control arm response rates in NMA can lead to biased estimates of comparative efficacy between treatments.http://link.springer.com/article/10.1186/s12874-019-0837-2Systematic reviewNetwork meta-analysisRheumatoid arthritisBiologicsStatistical methods |
| spellingShingle | Chris Cameron Abhishek Varu Arthur Lau Mahdi Gharaibeh Marcelo Paulino Raina Rogoza Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritis BMC Medical Research Methodology Systematic review Network meta-analysis Rheumatoid arthritis Biologics Statistical methods |
| title | Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritis |
| title_full | Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritis |
| title_fullStr | Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritis |
| title_full_unstemmed | Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritis |
| title_short | Incorporating adjustments for variability in control group response rates in network meta-analysis: a case study of biologics for rheumatoid arthritis |
| title_sort | incorporating adjustments for variability in control group response rates in network meta analysis a case study of biologics for rheumatoid arthritis |
| topic | Systematic review Network meta-analysis Rheumatoid arthritis Biologics Statistical methods |
| url | http://link.springer.com/article/10.1186/s12874-019-0837-2 |
| work_keys_str_mv | AT chriscameron incorporatingadjustmentsforvariabilityincontrolgroupresponseratesinnetworkmetaanalysisacasestudyofbiologicsforrheumatoidarthritis AT abhishekvaru incorporatingadjustmentsforvariabilityincontrolgroupresponseratesinnetworkmetaanalysisacasestudyofbiologicsforrheumatoidarthritis AT arthurlau incorporatingadjustmentsforvariabilityincontrolgroupresponseratesinnetworkmetaanalysisacasestudyofbiologicsforrheumatoidarthritis AT mahdigharaibeh incorporatingadjustmentsforvariabilityincontrolgroupresponseratesinnetworkmetaanalysisacasestudyofbiologicsforrheumatoidarthritis AT marcelopaulino incorporatingadjustmentsforvariabilityincontrolgroupresponseratesinnetworkmetaanalysisacasestudyofbiologicsforrheumatoidarthritis AT rainarogoza incorporatingadjustmentsforvariabilityincontrolgroupresponseratesinnetworkmetaanalysisacasestudyofbiologicsforrheumatoidarthritis |