Jump to comment:

• Author Response to Letter to the Editor
  Allison Low, Devan Kansagara
  Published on: 7 March 2019
• Re: Comparative Effectiveness and harms of intravitreal antivascular endothelial growth factor agents for three retinal conditions: a systematic review and meta-analysis
  Francisco-Javier Carrera-Hueso, Pedro Vazquez-Ferreiro and Jaime Poquet-Jornet
  Published on: 27 January 2019

• Published on: 7 March 2019

  Author Response to Letter to the Editor

  • Allison Low, Research Assistant Portland Evidence-based Synthesis Program, VA Portland Health Care System, Portland, Oregon, USA
  • Other Contributors:
    • Devan Kansagara, Director of the Portland Evidence-based Synthesis Program, Staff Physician

  To the Editor,

  We appreciate Francisco-Javier Carrera-Hueso, Pedro Vazquez-Ferreiro, and Jaime Poquet-Jornet's careful reading of this paper. This commissioned review had a necessarily broad scope in order to summarize benefits and harms across three available therapies for the most common clinical indications. We agree there was quite a bit of information to present and that doing so in a succinct format is a challenge. However, we disagree with their contention that we did not follow current methodologic systematic review standards. We did indeed follow PRISMA reporting guidelines, as described in the Methods.

  Regarding Table 1, the studies included in the summary table are the same as those described within the text and meta-analyses; we apologize for any confusion. In terms of format for the listing of studies in the meta-analyses, since studies are known primarily by their acronym, we used them in the figures whenever possible. The trials without specific names or acronyms were listed according to author and year.

  Biswas 2011 only reported the percentage of patients gaining ≥15 letters at the 18-month endpoint, not at 12 months, so the study could not be included in the 12-month analysis for this outcome. The study did report mean change in BCVA at both endpoints, so it is included in both the 12 month and 18-24 month analyses in Figure 2. In terms of analyzing cost-effectiveness, only two trials meeting inclusion criteria (CATT and DRCR) di...

  Show More

  To the Editor,

  We appreciate Francisco-Javier Carrera-Hueso, Pedro Vazquez-Ferreiro, and Jaime Poquet-Jornet's careful reading of this paper. This commissioned review had a necessarily broad scope in order to summarize benefits and harms across three available therapies for the most common clinical indications. We agree there was quite a bit of information to present and that doing so in a succinct format is a challenge. However, we disagree with their contention that we did not follow current methodologic systematic review standards. We did indeed follow PRISMA reporting guidelines, as described in the Methods.

  Regarding Table 1, the studies included in the summary table are the same as those described within the text and meta-analyses; we apologize for any confusion. In terms of format for the listing of studies in the meta-analyses, since studies are known primarily by their acronym, we used them in the figures whenever possible. The trials without specific names or acronyms were listed according to author and year.

  Biswas 2011 only reported the percentage of patients gaining ≥15 letters at the 18-month endpoint, not at 12 months, so the study could not be included in the 12-month analysis for this outcome. The study did report mean change in BCVA at both endpoints, so it is included in both the 12 month and 18-24 month analyses in Figure 2. In terms of analyzing cost-effectiveness, only two trials meeting inclusion criteria (CATT and DRCR) directly compared cost outcomes between any of the 3 agents; both were included in our review.

  Regarding the differing figures for mean difference calculated by the letter authors, we state in the Data Synthesis and Analysis portion of the text that “we used the mean difference (MD) between arms reported in the study whenever available; otherwise, mean between-group differences were calculated based on reported data.” For example, Kodjikian 2013 reported a mean difference of 2.36 (95% CI, -0.72 to 5.44) analysis for the GEFAL study, so that data was used in our analyses.

  The use of quantitative estimation of publication bias is controversial and not without its own drawbacks (1). Publication bias is a type of reporting bias in which positive (often smaller) studies are preferentially published over negative studies. We examined the possibility of including unpublished research by searching, as mentioned in the manuscript, trial registries, and regulatory agency sites to identify all relevant registered and in-progress trials. We also requested unpublished data from drug manufacturers, but did not identify any additional trials.

  We describe qualitatively the results contributed by each study according to its risk of bias. We found no indication of consistent differences in results according to ROB. Statistical measures also suggested minimal heterogeneity in the meta-analytic estimates presented.

  We thank the authors of the letter for their interest in our article, and appreciate the opportunity to respond to their concerns.

  Sincerely,

  Allison Low
  Devan Kansagara

  References:
  1. Guyatt GH, Oxman AD, Montori V, et al. GRADE guidelines: 5. Rating the quality of evidence—publication bias. J Clin Epidemiol. 2011;64(12):1277-1282.

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top
• Published on: 27 January 2019

  Re: Comparative Effectiveness and harms of intravitreal antivascular endothelial growth factor agents for three retinal conditions: a systematic review and meta-analysis

  • Francisco-Javier Carrera-Hueso, pharmacist Hospital Dr Moliner
  • Other Contributors:
    • Pedro Vazquez-Ferreiro, ophthalmologist
    • Jaime Poquet-Jornet, pharmacist

  To the Editor,
  Intravitreal antivascular endothelial growth factor (VEGF) agents undeniably have many clinical applications and we read with great interest the recent meta-analysis published in your journal by Low et al1 comparing the effectiveness and harms of these agents in three retinal disorders.
  We would first like to thank the authors for their exhaustive review and synthesis of the evidence in this area. The conclusions they reached served to confirm what many of us had already suspected.2 Nevertheless, the article features some important methodological flaws and inadequate reporting of data that we would like to highlight to ensure that readers are in a position to interpret the findings of the meta-analysis correctly.
  In relation to reporting issues, we were surprised to see that Table 1, which is quite creative and unique in terms of systematic review tables, does not include a list of the studies analyzed for each section. The authors, for example, state that they included two clinical trials comparing aflibercept and ranibizumab, but they do not specify which ones. This detracts from the transparency of the study and makes it difficult to review the findings. We also noticed a lack of uniformity within the figures, as some of the studies are listed by author name and others by author name and year of publication. In addition, Figure 3 shows data from the 2011 study by Biswas P, Sengupta S, Choudhary R, et al for the 18-24–month but not the 12...

  Show More

  To the Editor,
  Intravitreal antivascular endothelial growth factor (VEGF) agents undeniably have many clinical applications and we read with great interest the recent meta-analysis published in your journal by Low et al1 comparing the effectiveness and harms of these agents in three retinal disorders.
  We would first like to thank the authors for their exhaustive review and synthesis of the evidence in this area. The conclusions they reached served to confirm what many of us had already suspected.2 Nevertheless, the article features some important methodological flaws and inadequate reporting of data that we would like to highlight to ensure that readers are in a position to interpret the findings of the meta-analysis correctly.
  In relation to reporting issues, we were surprised to see that Table 1, which is quite creative and unique in terms of systematic review tables, does not include a list of the studies analyzed for each section. The authors, for example, state that they included two clinical trials comparing aflibercept and ranibizumab, but they do not specify which ones. This detracts from the transparency of the study and makes it difficult to review the findings. We also noticed a lack of uniformity within the figures, as some of the studies are listed by author name and others by author name and year of publication. In addition, Figure 3 shows data from the 2011 study by Biswas P, Sengupta S, Choudhary R, et al for the 18-24–month but not the 12-month period, and there is no explanation for this omission in the text.
  We also detected some methodological issues that depart from international recommendations.3,4 On replicating the meta-analysis using the data provided by Low et al1 in the STATA statistical package, we did not find the same results as those reported. In Figure 2, the mean difference (95% CI) given for months in the GEFAL study is 2.36 (-0.72 to 5.44), whereas we obtained a difference of 1.90 (95 CI%: -0.73 to 4.53). This modifies the magnitude of the overall measure (in our case, -0.25; 95% CI: -1.39 to 0.89), but not its direction or interpretation. We observed a similar finding for the 18-24 month period for the CATT study, as Figure 2B shows a difference of -1.00 (95% CI: -3.54 to 1.54), whereas we observed an overall difference of -0.51 (9 5CI% :-1.91 to 0.88).
  Another aspect that caught our attention is that some of the meta-analysis subgroup analyses were underpowered (<50%), but this is not mentioned in the results or discussed as an important limitation. The authors also make no mention of publication bias, which is normally evaluated using a funnel plot or Egger’s plot, as per reporting guidelines.3,4 Finally, Low et al1 reported that they analyzed the quality of the individual studies included in their systematic review and show the corresponding results in Table 1. However, they did not report on any sensitivity analyses to identify lower-quality studies that should have possibly been excluded or studies that might have significantly affected results.
  The authors clearly attempted to cover many aspects in their systematic review, from the effectiveness and harms to the cost and cost-effectiveness of three VGEF agents in three conditions: neovascular age-related macular degeneration, diabetic macular oedema (DME), and branch retinal vein occlusion. It is not our place to comment on the wisdom or not of analyzing so many aspects, but we would like to point out that just one study was analyzed to assess the evidence on the cost-effectiveness of bevacizumab versus aflibercept versus ranibizumab in DME. Despite the commendable efforts of the authors, they failed to report on some key methodological aspects from the PRISMA (Preferred Reported Items for Systematic Review and Meta-Analyses) checklist3 and the Cochrane Handbook for Systematic Reviews of Interventions.4
  REFERENCES
  1. Low A, Faridi A, Bhavsar KV, Cockerham GC, Freeman M, Fu R, et al. Comparative effectiveness and harms of intravitreal antivascular endothelial growth factor agents for three retinal conditions: a systematic review and meta-analysis. Br J Ophthalmol. 8 de noviembre de 2018;
  2. Cai S, Bressler NM. Aflibercept, bevacizumab or ranibizumab for diabetic macular oedema: recent clinically relevant findings from DRCR.net Protocol T. Curr Opin Ophthalmol. noviembre de 2017;28(6):636-43.
  3. Moher D, Liberati A, Tetzlaff J, Altman DG. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. BMJ. 21 de julio de 2009;339:b2535.
  4. Higgins JP, Green, Sally. Cochrane Handbook for Systematic Reviews of Interventions [Internet]. The Cochrane Collaboration. Disponible en: https://handbook-5-1.cochrane.org/

  Show Less

  Conflict of Interest:
  None declared.

  • Back to top