CDI Industry Challenges

  • Dr. Terrance Govender
  • May 16, 2019

With the upcoming 2019 ACDIS conference on the horizon, it is an opportune time to reflect on our industry: What have we accomplished over the years? What can we do better? What does the future hold? And what skills do we need to develop to bring the vision of CDI we have over the next few years to fruition? This post is the result of me exploring these very questions, and I thought it would be worthwhile to highlight a few challenges that have plagued the industry for a while and have prevented organizations from optimizing their CDI performance.

By no means is this list complete and I am sure that you could add to the list based on your personal experiences over the years. This list does, however, include topics which I believe are universal, based on the many discussions I have had with individuals in varying CDI roles, as well as my own CDI experiences.

What Should My CMI Be?

I have addressed this topic in a separate blog post, which talks about the challenge of benchmarking your CMI and the pitfalls of using CMI as an indicator of CDI performance, especially in the short term. Maybe you too have experienced the scenario of being held accountable for a drop in CMI. In that blog post, I described how there are factors that influence the fluctuations in CMI that are beyond the control of CDI and HIM – patient mix being one of them.

CDI has no control over the types of patients that happen to walk through the door and get admitted. They do, however, have a role to play in the accurate reporting of severity on those patients. A typical example would be if your neurosurgery department increased the number of procedures performed for a specified time period. This increase in higher weighted procedural DRGs would cause an increase in the overall CMI, regardless of the appropriate reporting of MCCs and CCs.

ClinIntell provides its clients with metrics to tease out the severity component of CMI and to assess the richness of MCCs and CCs in each month’s actual patient mix (more on that later on). In short, CDI should be able to immediately claim or dispute the responsibility of fluctuations in CMI through access to advanced analytics.

Identifying True Overall Opportunity And Performance

Frequently, CFOs are approached by vendors claiming that their facilities have a “CMI gap” and that it is due to under-reporting of severity. Further inquiry will reveal that the CMI gap was most probably identified through benchmarking methodologies.

In a previous post (“How MCC/CC Capture Rates Could Be Curbing Your True CDI Potential”), I discussed the pros and cons of benchmarking. Essentially, the biggest flaw as it pertains to CDI is that it does not take your unique patient mix and other facility-specific variables into account. Assessing progress towards a CMI goal based on benchmarking alone can be misleading, since it assumes a static patient mix, which is not the case.

Back to our scenario: If the CFO is curious enough to ask the vendor, “Where is that gap coming from?”, they are frequently met with a response that sounds similar to this: “Well, we can show you, but we will have to do a chart review.” Here are the limitations of using chart review to justify the overall CMI gap or opportunity at your facility:

  • It is costly and resource intensive.
  • You will never be able to review a statistically significant number of charts in order to breakdown the opportunity by service line, let alone physician.
  • The charts may be “cherry-picked” by the vendor, increasing the probability of finding severity reporting opportunities, e.g. longer lengths of stay cases. Unless properly projected when building up to the overall hospital opportunity, such non-random approaches can overstate the opportunity.
  • It is auditor dependent: Several individuals can review a chart and come up with different opinions on what should be reported. This, to a large degree, depends on the skillset and experience of the auditor.
  • Quantifiable opportunities can only be identified based on information that already exists in the chart. No clinical clues in the documentation means that no compliant query can be submitted.
  • The vendor has an incentive to justify the initial benchmarking-based opportunity estimate that may or may not be accurate, potentially adding bias to the chart review process.

We all agree that chart review by the CDI team on a daily, concurrent basis is of extreme value – where would CDI be without it?! However, using that methodology to quantify the overall opportunity and identify where it is coming from is not the most reliable and cost-effective approach to adopt. ClinIntell’s data analytic approach does not rely on the chart as a unit of analysis, but rather the population level using the hospital’s actual claims.

Aggregated Data And Trends/DRG Group Analyses

CDI metrics/data monitored by either healthcare leaders or CDI managers frequently include looking at trends over time only. This may include monitoring your “mini-CMIs” and/or MCC/CC capture rates for high-volume DRG groups and identifying those that need “attention”. There are limitations to this traditional, widely adopted approach.

Firstly, analyzing DRG groups mean that you will almost always be limited by the samples in each monthly datapoint, even for the facility’s highest volume DRG group(s). The power of such analyses stems from the ability to detect an effect when there is one to be detected, and a small sample size will increase the likelihood of mistaking noise for effect. Hence, a decline in a DRG group MCC/CC capture rate could potentially be misleading, causing you to make unnecessary decisions, like more staffing needs and increased chart review.

Another issue with this approach is that is does not provide you with any actionable data on which you can capitalize on. Try informing a physician, after identifying a decline in performance on the aggregated data and trends, that he/she needs to do better in a specific DRG group. That’s not very helpful, not to the physician anyway. An important question to ask when reviewing what CDI data you have access to or are planning on purchasing would be to determine if the data you are looking at periodically can be presented to you by an internal resource with good Excel skills.

In other words, are you buying analyses or a data presentation dashboard? Additionally, what kind of actionable insight does the product provide (e.g., focusing provider education content, decisions around CDI resource allocation, concurrent chart review selection) that will, in the end, likely result in more complete external reporting of patient acuity?

Actionable Data

Briefly mentioned in the previous paragraph was how CDI historically lacks the ability to share actionable data with physicians. In an environment where we are compelled to increase our physician engagement efforts, having actionable data to share with physicians is crucial. Sharing query response and/or agreement rates, CMI, or MCC/CC capture rates very often means little to a physician, especially if that physician has the desire to support the CDI efforts and improve their performance metrics.

ClinIntell’s clinical condition algorithms allow providers to not only have insight into their performance on specific conditions, but organizations can also adopt a focused system or facility-wide strategy to improve capture of diagnoses that have gaps versus the expected rates. The recent ACDIS newsletter correctly points out the value of the CDI educator role, as well as the value of physician engagement in CDI.

The traditional lack of actionable provider-level data had limited the ability to engage physicians and change their ongoing documentation practices. If you want to deploy a focused strategy that engages physicians and aids educators in developing data to drive content to the most appropriate providers or groups of providers, this is an absolute must!

No “Expected” Metrics In CDI!

I saved the best for last. CDI’s historical lack of an expected performance target (not based on benchmarking) is applicable to all of the above-mentioned challenges. Whether monitoring your CMI, MCC/CC capture rates, or clinical condition performance, we, as an industry, have not been bullish about having an expected value in CDI. We see extreme value in having access to expected levels of performance in length of stay and mortality rates, but are okay with not having that same expectation in CDI.

Having access to ClinIntell’s validated expected levels of severity performance, which does not rely on benchmarking, helps organizations know how close they are getting to expected levels of severity reporting performance at all levels, from system-/facility-wide, down to the individual physician and corresponding clinical conditions. Historically, access to sophisticated CDI analytics has been limited. I am, however, seeing shifts in the industry, where certain organizations are asking the right questions, acknowledging their challenges, and looking for viable solutions.

A steep decline in CMI can attract inappropriate attention to CDI. However, focusing on the gap between the actual CMI and ClinIntell’s “clinically expected” CMI will reveal the true documentation quality.

For example, there was a dramatic decline in CMI during 2Q18. However, the gap between the actual CMI and “clinically expected” CMI was 0.08, which is smaller compared to the gap of 0.10 during the CMI increase observed in 3Q18.

ClinIntell’s condition-level “expected” value will identify gaps in clinical condition reporting, even when accurate DRG assignments are achieved. This helps a provider document the full depth and breadth of their patient mix, and not just achieve any appropriate DRG shifts. Having access to monitoring the condition performance over time is also crucial to an organization’s efforts on provider education and CDI resources management.

As CDI continues to evolve, be curious, do your homework, ask the right questions, and identify the limitations that prevent you from optimizing performance. The challenges outlined here, while not all-inclusive, should pose as opportunities for us to grow, improve our knowledge, and actually make a dent in the industry.

“Without data, you’re just another person with an opinion.”

-W. Edwards Deming


Dr. Terrance Govender
VP of Medical Affairs, ClinIntell, Inc.