The Reply

Abstract

We read the valuable comment written by Ayubi and Safiri, who addressed methodological issues about our recently published article in The American Journal of Medicine.1Kwon O.C. Hong S. Ghang B. Kim Y.-G. Lee C.-K. Yoo B. Risk of colchicine-associated myopathy in gout: influence of concomitant use of statin.Am J Med. 2017; 130: 583-587Abstract Full Text Full Text PDF PubMed Scopus (27) Google Scholar The authors pointed out that the wide confidence interval (CI) can be indicative of a problem in the data, otherwise known as sparse data bias,2Greenland S. Mansournia M.A. Penalization, bias reduction, and default priors in logistic and related categorical and survival regressions.Stat Med. 2015; 34: 3133-3143Crossref PubMed Scopus (173) Google Scholar, 3Greenland S. Mansournia M.A. Altman D.G. Sparse data bias: a problem hiding in plain sight.BMJ. 2016; 352: i1981Crossref PubMed Scopus (482) Google Scholar especially in the univariable and multivariable analyses results of colchicine dose. In our study, the number of studied events was 12, which is relatively sparse. Thus, as Ayubi and Safiri pointed out, there is a possibility that data sparsity might have exaggerated the results in the univariable and multivariable analyses. There are various statistical methods developed to reduce or eliminate the sparse data bias,2Greenland S. Mansournia M.A. Penalization, bias reduction, and default priors in logistic and related categorical and survival regressions.Stat Med. 2015; 34: 3133-3143Crossref PubMed Scopus (173) Google Scholar, 3Greenland S. Mansournia M.A. Altman D.G. Sparse data bias: a problem hiding in plain sight.BMJ. 2016; 352: i1981Crossref PubMed Scopus (482) Google Scholar and we performed the “penalized estimation via data augmentation” as Ayubi and Safiri suggested. After “penalized estimation via data augmentation,” the following factors were still shown to be significantly associated with an increased risk of myopathy as in the original article: chronic kidney disease (adjusted hazard ratio [HR], 6.445; 95% CI, 2.233-13.973), liver cirrhosis (adjusted HR, 5.892; 95% CI, 1.392-15.906), higher colchicine dose (adjusted HR, 6.908; 95% CI, 1.861-18.933), and concomitant use of a CYP3A4 inhibitor (adjusted HR, 6.137; 95% CI, 1.872-14.695). The risk of myopathy was not increased in concomitant statin use with colchicine (adjusted HR, 1.248; 95% CI, 0.425-2.815) (Table). On the basis of these reanalyzed data, we presume that the statistically significant association between colchicine dose and myopathy risk in the multivariable model in our original article is not likely to have been the result of sparse data bias.TableAnalysis of Clinical Factors Associated with Myopathy in Patients with Gout Treated with ColchicineMultivariable Analysis (Penalized Estimation via Data Augmentation)Adjusted HR95% CIChronic kidney disease6.4452.233-13.973Liver cirrhosis5.8921.392-15.906Colchicine dose6.9081.861-18.933CYP3A4 inhibitor6.1371.872-14.695Statin1.2480.425-2.815CI = confidence interval; HR = hazard ratio. Open table in a new tab CI = confidence interval; HR = hazard ratio. Also, as Ayubi and Safiri pointed out, in the Cox proportional hazard regression model, the proportional hazard assumption is one of the most vital assumptions.4Kleinbaum D.G. Klein M. Survival Analysis: A Self-Learning Text. Berlin/Heidelberg, Germany: Springer Science & Business Media.2006Google Scholar We agree that if this assumption is violated, the results of Cox proportional hazard regression model may be misleading. Thus, we examined the proportional hazard assumption using log (-log [survival]) curves and the Schoenfeld residual test, and observed no relevant violations, which validates the use of the Cox proportional hazard regression model in our data. Risk of Colchicine-Associated Myopathy in Gout: Influence of Concomitant Use of Statin; Methodologic IssuesThe American Journal of MedicineVol. 130Issue 10PreviewWe read with great interest the valuable article by Kwon et al1 published in The American Journal of Medicine in 2017. The authors aimed to examine the risk of myopathy when statins are co-administered with colchicine in patients with gout. As one of the main findings, they found that higher colchicine dose was not associated with myopathy risk in the univariable analysis (hazard ratio [HR] 7.068; 95% confidence interval [CI], 0.622-80.343; P = .115), whereas it was found to be associated in the multivariable analysis (HR 20.960; 95% CI, 1.835-239.481; P = .014), which is problematic. Full-Text PDF