z**********i 发帖数: 12276 | 1 搞到了全套的part A, B claims 数据。想把Part A的claims, 与part B Claims 连接
,计算总的cost,然后算出每个医院的平均cost,比如想比一下heart failure.
但无从下嘴,请大家给点建议吧。
我怎么知道该用哪个data table?
正在试图从别的公司要code,他们计算了mspb.
多谢!! |
s**********l 发帖数: 395 | 2 你只想计算医院的平均成本吗?据我所知,Part B 是用来支付是用来支付看医生的费
用,而且大部分医生有自己的私人诊所的。所以,你应该先搞清哪一个CLAIM是用于支
付医院的。 |
g*********r 发帖数: 2847 | 3 Assuming you have provider ID field in the data, I suggest take a look at
this field to get an idea that how many provider IDs in Part B data can be
mapped to Part A. usually, a Part B provider ID represents either the
physician or the facility.
If most can be mapped, then you are done. If not (or maybe you don't even
have this field in Part B data), try linking each member to Part A at claim/
line level or by date of service, depending on your data. then you may need
to make more assumptions on hospital transfer etc.
Let me know your data source if it's OK. If I happen to have experience on
it, which is very possible, I may have better suggestions. |
z**********i 发帖数: 12276 | 4 我们是bene level的claims,病人在住院的费用是part A, 在这个期间或随后如果看了
医生,产生的part B,也算是这次入院产生的费用。如果是在这次入院时间以外产生的
physician visit,不计入费用。
这样算出cost,然后计算在这个医院的病人的平均费用。
多谢你的回复!!
【在 s**********l 的大作中提到】 : 你只想计算医院的平均成本吗?据我所知,Part B 是用来支付是用来支付看医生的费 : 用,而且大部分医生有自己的私人诊所的。所以,你应该先搞清哪一个CLAIM是用于支 : 付医院的。
|
z**********i 发帖数: 12276 | 5 我们从cms拿了个一年的小课题。
claims data, 非常大的数据。因为要做病人level的merge,同一个病人可能有几个
unique ID,并且,part B的时间与part A的时间match,才把part b的费用加入到part A
的费用中,否则,part B的费用就不用管了。
多谢回复!!
claim/
need
【在 g*********r 的大作中提到】 : Assuming you have provider ID field in the data, I suggest take a look at : this field to get an idea that how many provider IDs in Part B data can be : mapped to Part A. usually, a Part B provider ID represents either the : physician or the facility. : If most can be mapped, then you are done. If not (or maybe you don't even : have this field in Part B data), try linking each member to Part A at claim/ : line level or by date of service, depending on your data. then you may need : to make more assumptions on hospital transfer etc. : Let me know your data source if it's OK. If I happen to have experience on : it, which is very possible, I may have better suggestions.
|
g*********r 发帖数: 2847 | 6
A
for CMS data, you can easily use HIC # (and together with historical HIC #,
however, I don't think they trace the change on time) to link claims
incurred in different parts. however, claims/line # usually doesn't work as
a good map among different parts. So I think you have to take a look at the
from and thru dates.
【在 z**********i 的大作中提到】 : 我们从cms拿了个一年的小课题。 : claims data, 非常大的数据。因为要做病人level的merge,同一个病人可能有几个 : unique ID,并且,part B的时间与part A的时间match,才把part b的费用加入到part A : 的费用中,否则,part B的费用就不用管了。 : 多谢回复!! : : claim/ : need
|
z**********i 发帖数: 12276 | 7 You are right. The episode is using the from and through date to determine
the inhospital period.
Thanks!
,
as
the
【在 g*********r 的大作中提到】 : : A : for CMS data, you can easily use HIC # (and together with historical HIC #, : however, I don't think they trace the change on time) to link claims : incurred in different parts. however, claims/line # usually doesn't work as : a good map among different parts. So I think you have to take a look at the : from and thru dates.
|
z**********i 发帖数: 12276 | 8 多谢回复!
每人一个双簧包。
,
as
the
【在 g*********r 的大作中提到】 : : A : for CMS data, you can easily use HIC # (and together with historical HIC #, : however, I don't think they trace the change on time) to link claims : incurred in different parts. however, claims/line # usually doesn't work as : a good map among different parts. So I think you have to take a look at the : from and thru dates.
|
s**********e 发帖数: 368 | 9 which database are these claims from, HAJI or CCW? do you have bene id to
link the claims? you can define one stay from part a claim first using the
claim from and thru date. |
z**********i 发帖数: 12276 | 10 我们都有。
并且要link part A to part B.这个link比较难些。
你在qio?
【在 s**********e 的大作中提到】 : which database are these claims from, HAJI or CCW? do you have bene id to : link the claims? you can define one stay from part a claim first using the : claim from and thru date.
|
|
|
m*******g 发帖数: 3044 | 11 分两步做,先把PART A的算出来,找出每个病人,能不能把病人都ASSIGN给一个UNIQUE ID
?你是怎么知道同一个病人可能有几 unique ID的?根柜名字?应该有办法把他们的
MUTIPLE ID 换成一个UNIQUE ID 吧? 能弄出来吗,然后我再说下一步
A
【在 z**********i 的大作中提到】 : 我们从cms拿了个一年的小课题。 : claims data, 非常大的数据。因为要做病人level的merge,同一个病人可能有几个 : unique ID,并且,part B的时间与part A的时间match,才把part b的费用加入到part A : 的费用中,否则,part B的费用就不用管了。 : 多谢回复!! : : claim/ : need
|
z**********i 发帖数: 12276 | 12 我是道听途说的,但有个reference cross talk table,里面有个bene_clm_num_curr
is the most current identifier of bene.
多谢回复!
ID
【在 m*******g 的大作中提到】 : 分两步做,先把PART A的算出来,找出每个病人,能不能把病人都ASSIGN给一个UNIQUE ID : ?你是怎么知道同一个病人可能有几 unique ID的?根柜名字?应该有办法把他们的 : MUTIPLE ID 换成一个UNIQUE ID 吧? 能弄出来吗,然后我再说下一步 : : A
|
g*********r 发帖数: 2847 | 13
curr
just want to point out, as I mentioned before, CMS does track HIC # change,
but not on time. So don't trust the transition date in the historical HIC #
cross walk table.
【在 z**********i 的大作中提到】 : 我是道听途说的,但有个reference cross talk table,里面有个bene_clm_num_curr : is the most current identifier of bene. : 多谢回复! : : ID
|
z**********i 发帖数: 12276 | 14 实际上,我的问题是如何可以了解这些data tables.
是否有一些图示,说明这些table的关系。
因为,part A, B, 里面有很多的table。
有没有什么类似data dictionary 的东西,让我可以参照的?
多谢大家的回复!!
,
#
【在 g*********r 的大作中提到】 : : curr : just want to point out, as I mentioned before, CMS does track HIC # change, : but not on time. So don't trust the transition date in the historical HIC # : cross walk table.
|
z**********i 发帖数: 12276 | 15 刚才找出来part A, B standard Non-institutional claims data dictionary.
准备静心看看。
因为,她们让我算一下,mammography screeening rate, 不得不自己研究一下了。
多谢大家热心的回复!
【在 z**********i 的大作中提到】 : 搞到了全套的part A, B claims 数据。想把Part A的claims, 与part B Claims 连接 : ,计算总的cost,然后算出每个医院的平均cost,比如想比一下heart failure. : 但无从下嘴,请大家给点建议吧。 : 我怎么知道该用哪个data table? : 正在试图从别的公司要code,他们计算了mspb. : 多谢!!
|
g*********r 发帖数: 2847 | 16
use HCPCS/CPT codes
【在 z**********i 的大作中提到】 : 刚才找出来part A, B standard Non-institutional claims data dictionary. : 准备静心看看。 : 因为,她们让我算一下,mammography screeening rate, 不得不自己研究一下了。 : 多谢大家热心的回复!
|
z**********i 发帖数: 12276 | 17 还有的用了icd-9.
这些code倒是相对容易理解,spec manual里都具体列出来了。
难的还是对part A, B结构的理解和运用。
多谢回复!
【在 g*********r 的大作中提到】 : : use HCPCS/CPT codes
|
g*********r 发帖数: 2847 | 18
Basicly, ICD9 is for facility use (Part A) only. But usually revenue codes
are the first choice here for mapping (or you can use the combination).
【在 z**********i 的大作中提到】 : 还有的用了icd-9. : 这些code倒是相对容易理解,spec manual里都具体列出来了。 : 难的还是对part A, B结构的理解和运用。 : 多谢回复!
|
o********s 发帖数: 853 | 19 yes, bene_clm_num_curr is the most recent bene_num, and there is the xtalk
table which links to previous bene_num(s)
curr
【在 z**********i 的大作中提到】 : 我是道听途说的,但有个reference cross talk table,里面有个bene_clm_num_curr : is the most current identifier of bene. : 多谢回复! : : ID
|
z**********i 发帖数: 12276 | 20 嗯,记到小本本上。
多谢大家的回复!
【在 o********s 的大作中提到】 : yes, bene_clm_num_curr is the most recent bene_num, and there is the xtalk : table which links to previous bene_num(s) : : curr
|