Factors Associated With Surgical Outcomes in Congenital Ptosis: A 10-Year Study of 319 Cases.
Factors Associated With Surgical Outcomes in Congenital Ptosis: A 10-Year Study of 319 Cases.
复制标题
DOI：
10.1016/j.ajo.2017.04.027
复制
发表时间：
2017-09
影响因子：
4.2
通讯作者：
E. Ayubi;S. Safiri
中科院分区：
文献类型：
--
作者：
E. Ayubi;S. Safiri
我们饶有兴趣地阅读了 Ho 及其同事撰写的发表在《美国眼科杂志》上的论文。 1 作者旨在检查先天性上睑下垂手术的 10 年结果，并确定与良好结果和复发相关的因素。他们发现，伤口感染、术后睑内翻、结膜脱垂、额肌悬吊治疗、术前边缘反射距离较差、年龄小于1岁等7个因素与复发独立相关。 1 虽然本研究对该领域做出了有价值的贡献，但需要注意一些方法论点。作者比较了复发组和未复发组之间的伤口感染，发现与替代组相比，伤口感染患者的复发几率增加了 20 倍（感染伤口复发 [n= 6]，感染伤口未复发 [n= 2]，未感染伤口复发 [n= 49]，未感染伤口未复发 [n= 333]；比值比 (OR)= 20.38；95% 置信区间 (CI)：4.00-103.86)。科学家们认为，大效应估计并不一定能证明大效应，可能是由于自变量和因变量组合中的数据不足而导致的。 2, 3 大 CI 和大效应估计证实了数据稀疏性。 2 此问题称为稀疏数据偏差。 Ho等人的研究表4显示，“感染伤口没有经过处理，仅存在2例”。
WE HAVE READ WITH GREAT INTEREST THE PAPER AUTH-ored by Ho and associates that was published in the American Journal of Ophthalmology. 1 The authors aimed to examine the 10-year results of surgery for congenital ptosis and determine factors associated with excellent outcomes and recurrence. They found that 7 factors, including wound infection, postoperative entropion, conjunctival prolapse, treatment with frontalis suspension, poorer preoperative marginal reflex distance, and age younger than 1 year old, were independently associated with recurrence. 1 Although the present study makes a valuable contribution to the area, some methodological points need to be noticed.The authors compared the wound infection between the groups with and without recurrence and found that the odds of recurrence increase by 20 times in patients with wound infection compared to the alternative group (infected wound with recurrence [n= 6], infected wound without recurrence [n= 2], uninfected wound with recurrence [n= 49], uninfected wound without recurrence [n= 333]; odds ratio (OR)= 20.38; 95% confidence interval (CI): 4.00-103.86). The scientists have argued that the large effect estimates do not necessarily demonstrate the large effect and may have resulted because of the insufficient data in the combinations of the independent and dependent variables. 2, 3 The large CI along with large effect estimate confirms the data sparsity. 2 This problem is known as sparse data bias. As shown in Table 4 of the study conducted by Ho and associates, only 2 cases exist in the ‘‘infected wound without