Charlson Comorbidity Index or Something Better?



I am doing a study on comparing the healthcare utilization of two types of epilepsy patients. I’m using Cox Proportional Hazard models and I’d like to include comorbidity in the covariates. I’ve seen other papers similar to our study use Charlson Comorbidity Index, but I’ve read that Elixhauser’s is better. I assume there are other useful indices out there as well.

My dataset already has the comorbidities used in the Charlson Comorbidity Index. However, we have access to the patient data if we needed to find other comorbidities. Although it would take some time and effort.

So my question is: are there other indices that are so much better than the Charlson Comorbidity Index at predicting comorbidity burden that I should try to convince the PI of my study to dig back into patient data to find other comorbidities?


The Charlson index has an arithmetic error in its derivation. The authors added hazard ratios to obtain the index, whereas the log hazard ratios should have been added. Even despite this, the Elixhauser index is far better. There are more comprehensive indexes than Elixhauser’s including work by George Stukenborg. See also this.


Interesting, thank you very much!