Charlson index for dataset that contains both ICD9 and ICD10 for different years

[Femi Olowu]

Hello,

I am working on a hospital administrative dataset combined for multiple years between 2004-2018 the cases in the earlier years made use of ICD9, while 2016 upward made use of ICD10. I am trying to calculate the Charlson Co-morbidity index for each case in the combined data set. The available code index only accommodates either ICD9 or ICD 10, is there a way I can combine both. The code I used is pasted below.

charlson DX1-DX30 I10_DX1-I10_DX40 , index(c) assign0 wtchrl cmorb *for ICD 9* or charlson DX1-DX30 I10_DX1-I10_DX40 , index(10) assign0 wtchrl cmorb *for ICD 10*

I look forward to your help with cheerful optimism.


Olowu

[Ken Chui]

I have not used this before, but from the command -help charlson-, you can see the syntax chart:

charlson varlist [if exp] [in range] , index(string) [idvar(varname) diagprfx(string) assign0 wtchrl cmorb noshow]

and verify that it takes "if". So, theoretically, you can try:

charlson DX1-DX30 I10_DX1-I10_DX40 if year < 2016, index(c) assign0 wtchrl cmorb

and see if you can just process part of it, and then apply the same to year >= 2016 (Notice that you may have variable overwriting error, so be prepare to rename some variable before processing the ICD10 part.). If you got no luck in that, then try to save your data into two chunks, split by 2016. Process them separately, and then use -append- to combine them back.

[Femi Olowu]

Thank you. Hd to split and append!

Cheers
Olowu