求助关于统计的问题（COX回归）

给BJH投稿了一篇文章，大致是抗核抗体ANA对疾病结局的影响。文章选择了ANA滴度1:160为界值进行分组讨论，分为ANA滴度≥1:160组和ANA滴度<1:160两组，经过COX多因素回归分析，结果显示：ANA滴度≥1:160组进展为AID的风险更高。

BJH经过32天under review，回来的意见中，统计审稿人给出意见：

This paper looks at ANA titre in a group of children with ITP. The variable for ANA titre level is the maximum over the course of follow-up - this is of course confounded by time, and this can be seen in the different length of follow-up in those with a measurement over 160 and those without a measurement over 160. This inherent bias needs to be fixed, and the nature of when the titre changes dealt with - as the time to high titre is likely to be important as well as how long the titre is maintained above that level. A single sporadic measurement after a long time is likely to represent a different patient path than a high titre at all times; but this is not taken into account in the analyses.

Likewise the measurement of all titres <160 needs to allow for the number of times this was measured. The HR for those whose titre drops has a wide confidence interval which appears consistent with the other groups.

求助解决办法，审稿人的意见有什么好的解决办法么？或者改成重复测量数据基础上进行COX回归？但是这样会增加很大工作量，而且许多化验并非同时做的，存在缺失数据问题。。。。