Last month, we posted a poll on our LinkedIn page asking our ClinIntell LinkedIn Community what severity reporting performance data/metrics they share with their physicians. 53% of participants said CMI, followed by CC/MCC Capture Rates with 23%, Query Response Rates with 20% and None with 3%. We talk more in detail about the 3 severity-reporting performance metrics below.

CMI

Most organizations that share CMI with their providers are most likely doing so with the intent of providing them with insights on their severity documentation performance. However, one of the pitfalls of using CMI as a severity documentation performance metric for physicians or even physician groups is that all DRGs have varying relative weight values at baseline (without CC/MCC). This explains why surgeons usually report a higher CMI than medical service lines. It is not necessarily that they are better documenters of severity, but rather, surgical DRGs have a higher relative weight even without any documentation of pertinent comorbidities. Furthermore, because of this very same reason, medical providers in the same service line and even at the same facility will ultimately have variations in their CMI, which is driven mostly by the base value of the DRG mix of the patients they cared for versus the CC/MCC severity documentation component.

If the goal is to give providers insight on how well they are doing on severity documentation, CMI can be very misleading, since it can fluctuate due to reasons that have no link to severity documentation performance.

CC/MCC Capture Rates

CC/MCC Capture rates is the second metric showing up in our poll. This may be because the industry  acknowledges  that this metric, theoretically, is dependent upon the severity documentation performance of providers. There are, however, challenges with associating this metric directly to physicians’ severity documentation performance. Firstly, every physician, even if in the same service line and at the same hospital will have a unique set of DRGs reported. Since capture rates are influenced by the DRG itself (amongst other variables), it is not an apples-to-apples comparison amongst physicians. A low capture rate for a physician may not necessarily imply that that physician is a poor documenter of severity since low capture rates may be driven by the specific mix of DRGs attributed to that physician during that time. One solution to control for the DRG mix is to measure and compare the capture rates for a specific DRG at the provider level, however, by doing so, you are now faced with a sample size limitation. Essentially, any perceived insight you get from this approach will be riddled with “noise” in the data.

In summary, capture rates, especially at the group or provider level are not a statistically sound metric. Furthermore, it only offers a binary response (either a CC/MCC is present or not) and does not consider that not all CCs and MCCs are of equal relative weight value. Capture rates also, on its own and without additional actionable data, does not engage physicians on a clinical level. One cannot communicate to physicians to get better at documenting CCs and MCCs, since there are too many conditions that fall into that category.

Query Response Rates

In the third place we have “Query Response Rates” which also isn’t necessarily an indicator of physician performance on severity documentation. The response rate metric does not incentivize physicians to do it right the first time, and it is dependent on variables outside of the physicians’ control. For example, a physician with no queries for the month may imply that the physician appropriately documented severity, or that the reviewer was not astute enough to identify opportunities to generate a compliant query. Query Response Rates also run into the same issue as Capture Rates: small sample size at the physician level. As we know, queries are dependent on a CDS team member initiating it and the team may be bias as to which providers receive a query or not since query response rates oftentimes can be used as a performance metric for CDSs, so they may skew their queries towards physicians that are more likely to engage and respond. A higher query response rate does not necessarily mean severity documentation is improving. If query response rates are used as a performance metric for physicians, they may not focus on documenting it right the first time since they may be encouraged to wait for the query and then respond, hence improving their response rate. In summary, query response rate is not and should not be an indicator of physician documentation performance, nor should it be used as an indicator of physician engagement with the documentation initiative.

Lastly, a small percentage of participants said they do not share any of these metrics with their physicians. There are two sides to this response. They either don’t share these metrics because they have an appreciation that the top three aren’t impactful (and may be sharing other metrics) or it means your program isn’t at a stage where they can hold physicians accountable for engagement in CDI. Either way, as the scope for CDI expands to touch several different initiatives ranging from finance to quality, provider level performance metrics and actionable data will have to take center stage eventually. Relying on the query process alone is not a LEAN approach, and the last thing you want to do is increase the number of queries that physicians receive in the midst of physician burnout and other competing initiatives.

ClinIntell’s advanced patient population severity reporting metric, CMI Documentation Score, is the only metric in the industry that analyzes the unique patient population of every provider and shows you based on that unique patient population, what percentage of available MS-DRG related severity is being reported. This is essentially an O:E percentage ratio, which means that you can compare this performance metrics amongst providers since the E is dependent on the population for each provider. When this is paired with the industry’s most advanced patient to provider attribution methodology, as well as clinical condition actionable data specific to each provider, the game has truly changed and the results speak for themselves.