Hi,
I am using hospital diagnosis data (ICD-10) to understand the co-morbidities of a group of patients. We wanted to use both the Charlson and Elixhauser to do this but I have a few questions which I am hoping that someone with more knowledge and experience of how these work on STATA might be able to answer:
1. When I first ran the Charlson, it showed that none of the patients had diabetes (completely incorrect) so after I looked at the code that was available online, it appeared that the ICD-10 codes were used without the dots (eg E11.9 changed to E119). I made these changes and the program identified diabetes. Is this correct that the dots need to be removed? If so, do you know why it was programmed like this rather than in how the ICD-10 codes are normally seen?
2. When I ran the Elixhauser with the same data and in the same way (no dots), comparison with the Charlson results showed differences in the frequency counts of some of the diagnoses. For example, eg Diabetes no complications: Charlson 12,063; Elixhauser 10,722. The diabetes with complications was also different so I thought maybe it was because they had been coded slightly differently but the total number of diabetes identified was also different: Charlson diabetes with and without complications 34,787 and Elixhauser 34,921. Does anyone have any insight into the reasons for this? How have you dealt with it?
Any feedback would be really helpful.
Joanna
I am using hospital diagnosis data (ICD-10) to understand the co-morbidities of a group of patients. We wanted to use both the Charlson and Elixhauser to do this but I have a few questions which I am hoping that someone with more knowledge and experience of how these work on STATA might be able to answer:
1. When I first ran the Charlson, it showed that none of the patients had diabetes (completely incorrect) so after I looked at the code that was available online, it appeared that the ICD-10 codes were used without the dots (eg E11.9 changed to E119). I made these changes and the program identified diabetes. Is this correct that the dots need to be removed? If so, do you know why it was programmed like this rather than in how the ICD-10 codes are normally seen?
2. When I ran the Elixhauser with the same data and in the same way (no dots), comparison with the Charlson results showed differences in the frequency counts of some of the diagnoses. For example, eg Diabetes no complications: Charlson 12,063; Elixhauser 10,722. The diabetes with complications was also different so I thought maybe it was because they had been coded slightly differently but the total number of diabetes identified was also different: Charlson diabetes with and without complications 34,787 and Elixhauser 34,921. Does anyone have any insight into the reasons for this? How have you dealt with it?
Any feedback would be really helpful.
Joanna
Comment