Renal damage was determined by the Systemic Lupus International Collaborating Clinics/American College of Rheumatology Damage Index (50% reduction in glomerular filtration rate, proteinuria ≥3.5 gm, or end-stage renal disease.) LUMINA = LUpus in MInorities, NAture versus Nurture; 95% CI = 95% confidence interval.
Article first published online: 29 OCT 2009
Copyright © 2009 by the American College of Rheumatology
Arthritis Care & Research
Volume 61, Issue 11, pages 1615–1616, 15 November 2009
How to Cite
Alarcón, G. S., Pons-Estel, G., McGwin, G., Danila, M. I., Zang, J., Bastian, H. M. and Vilá, L. M. (2009), Reply. Arthritis & Rheumatism, 61: 1615–1616. doi: 10.1002/art.24896
- Issue published online: 29 OCT 2009
- Article first published online: 29 OCT 2009
To the Editors:
We sincerely appreciate the concerns raised by Drs. Vlad and Vinet et al regarding our recently published study on the protective effect of HCQ on renal damage in patients with lupus nephritis. Dr. Vlad raises 3 concerns: confounding, immortal bias, and the magnitude of the HRs; whereas Vinet et al refer to the concerns of the immortal bias and the magnitude of the HRs. In their view, these concerns make our study potentially biased and its conclusions questionable. We certainly do not agree with these statements and offer the following comments in reply.
No matter what method is used, such as the propensity score that we used (1–3), or adjusting for all variables that differ between those who did and did not take HCQ (4, 5), there is always the possibility of unmeasured confounding. Furthermore, Dr. Vlad is correct in that the variables adjusted for in the analyses presented were baseline variables. To address this concern, and the concern related to immortal bias, we have now conducted additional analyses using a longitudinal approach, i.e., instead of modeling patients or person-time, we have modeled visits and used a repeated-measures logistic regression. Therefore, each patient contributed visits from the time of enrollment (T0) until they either developed renal damage or were seen for their last visit. The HCQ variable and several confounders can be viewed as “time-dependent” since they came from each visit (therefore, anything that changes over time will be captured). The results are consistent with the data presented in our study, although the effect is not as strong. Table 1 depicts crude and adjusted odds ratios for renal damage, and Table 2 shows the same measures for renal damage minus proteinuria.
|Model||Odds ratio||95% CI||P|
|Model||Odds ratio||95% CI||P|
Regarding the magnitude of the estimated HR being inconsistent with the clinical experience (as Dr. Vlad points out), or not observed even with cyclophosphamide in the lupus nephritis trials (as Vinet et al indicate), we must point out that they have misinterpreted this estimate as a reduction (71%; 95% confidence interval [95% CI] 32–87) in the risk of renal damage occurrence among those receiving HCQ as compared with those not receiving HCQ. Since this is a time-to-event analysis, what the data indicate are that patients with lupus nephritis who are receiving HCQ experience a prolonged time to the occurrence of renal damage (anywhere from 32–87% as per the 95% CI) in comparison with those who do not take HCQ. So, it is not that they will not develop renal damage, but it is that the time to this occurrence will be delayed as compared with those who do not take HCQ, which is still very important from the clinical point of view (the maintenance of HCQ treatment even when serious disease manifestations occur).
In short, the time-dependent analyses support the protective effect of HCQ in retarding the development of renal damage in lupus patients with renal involvement.
- 1Effect of hydroxychloroquine in the survival of patients with systemic lupus erythematosus: data from LUMINA, a multiethnic US cohort (LUMINA L). Ann Rheum Dis 2007; 66: 1168–72., , , , , , et al.
- 2The role of aggressive corticosteroid therapy in patients with juvenile dermatomyositis: a propensity score analysis. Arthritis Rheum 2008; 59: 989–95., , , , .
- 3The benefits of early treatment in rheumatoid arthritis: confounding by indication, and the issue of timing [editorial]. Arthritis Rheum 2003; 48: 1–5..
- 4Evidence from nonrandomized studies: a case study on the estimation of causal effects. Am J Epidemiol 2008; 167: 1120–9., , .
- 5A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods. J Clin Epidemiol 2006; 59: 437–47., , , , , .
Graciela S. Alarcón MD, MPH*, Guillermo Pons-Estel MD*, Gerald McGwin Jr. PhD*, Maria I. Danila MD, MSc*, Jie Zang MPH*, Holly M. Bastian MD, MSPH*, Luis M. Vilá MD, * University of Alabama at Birmingham, University of Puerto Rico Medical Sciences Campus, San Juan, Puerto Rico.