隔壁那位MD, 你认真看人家paper了吗? For observational studies, confounding and bias are always concerns. 你觉得文章作者和杂志评审们会留这么个大破绽让人诟病吗? 麻烦你好好看看Henry Ford的文章, 尤其是Table 3, Table 4 & Fig 2. 人家知道differences in baseline and clinical characteristics between subjects with HCQ and those without 是个问题, 所以人家做propensity score matching了, 1:1 match了190对病人 (HCQ vs No HCQ). 这些matched的病人, 他们baseline和clinical characteristics是一样的,包括 steriod的使用 (44% vs 44%). Match之后再做Cox, hazard ratio = 0.487 (p=0.009), 就是说用HCQ的病人和不用HCQ的病人比, 医院内死亡风险降低51.3%. 麻烦MD们评论人家文章前, 先看懂数据 https://www.ijidonline.com/article/S1201-9712(20)30534-8/fulltext
Why are prospective studies better than retrospective? Prospective studies usually have fewer potential sources of bias and confounding than retrospective studies. A retrospective study looks backwards and examines exposures to suspected risk or protection factors in relation to an outcome that is established at the start of the study. 请楼上的几位不要再丢人现眼了。谢谢！
Why are prospective studies better than retrospective? Prospective studies usually have fewer potential sources of bias and confounding than retrospective studies. A retrospective study looks backwards and examines exposures to suspected risk or protection factors in relation to an outcome that is established at the start of the study. 请楼上的几位不要再丢人现眼了。谢谢！ Goodendoodle 发表于 2020-08-01 22:16
Why are prospective studies better than retrospective? Prospective studies usually have fewer potential sources of bias and confounding than retrospective studies. A retrospective study looks backwards and examines exposures to suspected risk or protection factors in relation to an outcome that is established at the start of the study. 请楼上的几位不要再丢人现眼了。谢谢！ Goodendoodle 发表于 2020-08-01 22:16
What's your point? 你不会把propensity看成prospective了吧? "丢人现眼" 原装奉还!
本身就是一个observational study, 作者从来没否认过有limitations. 按常规, 在文章最后专门有一段写limitations. 而且作者也没要求把其结果当RCT来看 - 结论里写了: "prospective trials are needed to examine this impact". 'flaw'? you could also find flaws in a double-blinded, placebo controlled randomized clinical trial, right?
回复 35楼medstat的帖子 No, not obviously like this one! You can have unknown or unclear cofounding variables involved and claim them as limitations. Not when something can so obviously biased the results. I don’t know what is your training. If you have done works using structure or reduced form, this should not be a question.
回复 35楼medstat的帖子 No, not obviously like this one! You can have unknown or unclear cofounding variables involved and claim them as limitations. Not when something can so obviously biased the results. I don’t know what is your training. If you have done works using structure or reduced form, this should not be a question. 心平气和2020 发表于 2020-08-02 00:45
Sorry if i misunderstood, and i apologize for what I just said. I think what you are referring is much deeper than what that MD meant. It is true what the analysis of this study is not perfect but we cannot deny that they did make efforts to control for bias. IMO, their analysis was solid (not perfect though). I think what people ask from this paper is just a FAIR look of the potential benefit of HCQ on covid.
麻烦MD们评论人家文章前, 先看懂数据
https://www.ijidonline.com/article/S1201-9712(20)30534-8/fulltext
估计是隔壁那个“反智”楼主littleman。
他原帖里面benefit risk 的论据也不对，不过貌似他后来又改过了。
请楼上的几位不要再丢人现眼了。谢谢！
https://www.click2houston.com/news/investigates/2020/08/01/houston-area-doctor-agrees-with-use-of-controversial-drug-to-treat-covid-19/
一个本来要招2000人的multi-site trial最后只招了20人就给匆匆关了，然后天天在那说有效果的都不是golden standard study，也不心虚
是的 你说的在理 这个Henry Ford的结论也是recommend去做双盲trial
然而到了媒体就变味道了 变成了打架的工具 哎 可悲！
你有没有看懂楼主的问题？没看懂就不要谈了，连丢人现眼都说得出来……
我今天已经在好几个地方说了
氯喹是SARS绝迹之后实验室里面体外试验发现有效的 并不是SARS治疗的经验
03年都是用激素的。
如果你有03年有人用氯喹治疗SARS的记录，欢迎贴出来！
所以对这几个MD的credibility 表示怀疑，哈哈
你是不是哪里没看懂？那个md最大的质疑是这篇paper没有讨论他们是如果处理corticosteroids 的问题。这是非常明显的flaw。
都是考过Bar的MD，也别要求太高了！LOL
是那个”MD “ 没看懂，这个是基本的流行病的方法balance了，楼主写的很明白了。
+1，好好的针对轻症和预防的双盲实验硬是不肯做，然后搞一堆的伪装数据的，重症的，overdose的，细胞实验等等来试图证明没用甚至有巨大的副作用。然后还给质疑结果的人扣上反智的大帽子。
华人大妈虽然不都是医生，受过系统科学训练的phd比例却不低，看懂医学文献也不是大问题，不是那么好骗的，真不知道是谁在反智。
你是不是对experimental design 有什么误解。这类的design,除了对病人的characteristics 有做exact matching,更主要的是treatment 的control。这篇paper，无法处理关于corticosteroids产生的confounding issue。如果非要用这些病人做实验，这两组做cox，是得不出严谨的结论了。至少要再根据treatment ,分出另外一组，做ddd。可是，这篇paper没有尝试做robust check。虽然是peer reviewed的，但还是有flaw。
What's your point? 你不会把propensity看成prospective了吧? "丢人现眼" 原装奉还!
那个MD的原话见下, 不知道谁没看懂?
"最近被炒作的火热的Henry Ford Hospital Study, 称服用了 HCQ的病人死亡的hazard ratio降低了51%
https://www.ijidonline.com/article/S1201-9712(20)30534-8/fulltext
仔细看的话，发现使用了HCQ的病人绝大部分都同时使用了激素，而没有使用HCQ的病人只有少数给了激素。激素已经通过randomized controlled trial被证明对新冠病人有用，所以使用激素就是明显的confounding，我们不知道这些病人是因为服了激素好的还是服了HCQ好的。还有就是服用了HCQ的病人上呼吸机的更多，假设两个病人都病得很重，一个愿意上呼吸机并且服用了HCQ活了过来，另一个不想上呼吸机也没有服用HCQ没有活过来,怎么证明病人是HCQ救活的？ 类似这样的study还有挺多，相似点就是都是retrospective analysis,就是后知后觉，知道了病人结果以后再去对data做analysis,这种方式得出的结论我们肯定是不能相信的。"
激素已经通过randomized controlled trial被证明对新冠病人有用，所以使用激素就是明显的confounding，我们不知道这些病人是因为服了激素好的还是服了HCQ好的。
你是如何解读这句话的？这篇paper是如何address 这个问题的？
我知道华人网站有的MD是冒充的，自称美国医生，但是说话漏洞百出,经不起推敲。
哈哈哈哈哈，LOL
本身就是一个observational study, 作者从来没否认过有limitations. 按常规, 在文章最后专门有一段写limitations. 而且作者也没要求把其结果当RCT来看 - 结论里写了: "prospective trials are needed to examine this impact".
'flaw'? you could also find flaws in a double-blinded, placebo controlled randomized clinical trial, right?
原来你是个外行,根本不懂... 这样, 你先去补统计101, 然后回来再试试能不能看懂一点儿
No, not obviously like this one! You can have unknown or unclear cofounding variables involved and claim them as limitations. Not when something can so obviously biased the results. I don’t know what is your training. If you have done works using structure or reduced form, this should not be a question.
不好意思，我教过很多年统计课了，也带了很多学生。你的知识盲点是是在数学系之外。
Sorry if i misunderstood, and i apologize for what I just said. I think what you are referring is much deeper than what that MD meant. It is true what the analysis of this study is not perfect but we cannot deny that they did make efforts to control for bias. IMO, their analysis was solid (not perfect though). I think what people ask from this paper is just a FAIR look of the potential benefit of HCQ on covid.
哈哈哈， 考过BAR 的MD， 华人的独特风景