Coders in many countries found the transition to ICD-10 diagnosis coding to be a challenge. In the US, we are attempting something no one has done yet – the implementation of ICD-10 in addition to ICD-10 procedure coding, at the same time. It’s important to note that the US version of ICD-10 is more extensive than the international version, which makes it likely that this situation will undoubtedly prove even more challenging.

The good news from the international implementations is that after a few years, the quality of the coded diagnosis data was as good as or better than the previous ICD-9 data. Januel et. al. reported on their own study of Swiss hospital coding, as well as numerous other studies that showed the quality of ICD-10 coded data as comparable, or slightly better than the ICD-9 data that preceded it after the initial acclimation period. They also confirmed that the quality of the coded data increases slightly over time.

Walker. et al. studied eight years of coded hospital discharge data in Canada, and determined that the number of comorbidities did not change significantly. This suggests that the impact of better documentation may produce better codes, but not more of them. One of the weaknesses of the study is that in at least one Canadian province, coders were instructed to focus on common comorbidities like diabetes and hypertension as a way of dealing with the impact that ICD-10 had on productivity.

Lastly, the productivity difference is important to consider. As we’ve noted, coder productivity in one Canadian hospital fell by approximately 50 percent immediately after transitioning to ICD-10. Over the next year, coding productivity improved, but never improved past a 20 percent decrease from ICD-9 levels. The authors of the study noted that “it was at least three to six months post-implementation before there was any appreciable improvement in the decreased productivity and almost a year before productivity levels approached pre-ICD-10 levels.” This result was confirmed by additional international studies and by two US-based studies – one by AHIMA and another by WEDI.

Given that a 50 percent decrease in coding productivity requires a doubling of the coding staff to retain previous coding levels, this particular impact is likely to be the most significant. If you’re interested in minimizing the impacts of decreased coding productivity, review our series detailing strategies for minimizing the impacts of ICD-10 on coding productivity.