ICD-9 & ICD-10 Codes Equivalency

[Joshua Tobias]

Good morning everyone! I was wondering if I could get some advice on ICD-9 and ICD-10 codes. For my dissertation, I use roughly 11 panels of the Medical Expenditure Panel Survey (MEPS) by the AHRQ. However, I am having problems figuring out the best way to ensure that I am capturing the codes correctly. In one version, I created a dummy called "Has Diagnosis" that flags cases if they have a substance abuse problem using the primary ICD-10 diagnosis codes and ICD-9 codes. I merged the panel, FYC, and condition files in another version to create a "master" panel dataset. But the issue is that when I tried to make the 2015/2016 panel, I encountered the problem of having ICD-9 and ICD-10 diagnosis codes. Upon realizing that, I decided to take a break (it's been bothering me for the last two weeks). I would like to know anyone's thoughts if whether or not I am complicating my process or does just "flagging" records is the best way to work with this data.

[Leonardo Guizzetti]

There would have been a cut-off date when diagnosis codes and the like were switched from ICD-9 to -10. This should be documented somewhere with your dataset documentation. So you'll need to process that transitional year twice: once for the data under ICD-9 and again for the data under ICD-10. As for flagging, it seems reasonably straightforward enough, if you only need to know when the code appeared.
You may also want to verify that the diagnosis codes being used under both schemes are equivalent from a clinical perspective. That is, the codes under both schemes are being used the same way to capture the same sorts of diagnoses for defining your analysis population.