Off Script: As the Quantification of Cath Lab Quality Gets an Upgrade, Frustrations Abound

NEW ORLEANS, LA—As modern healthcare systems struggle to contain skyrocketing costs, it is inevitable that we move towards pay-for-performance or value-based purchasing reimbursement models that reward providers for meeting quality measures. Despite the superficial idea that quality can simply be measured by outcomes after the procedure, the actual pursuit of truth is far more complex, as systems of measurement have to account for how sick patients are, so as to not unfairly punish institutions who taking on patients most in need.

Of course, this system of reporting has, in some instances, produced the opposite, unintended outcome: sicker patients do poorly because physicians avoid treating them lest their stats suffer. As a cath lab director and cardiovascular operations director at my health network, I have had to struggle with the quantification of quality and the effort to improve metrics for many years. My task has been confounded by the fact that I work at a tertiary referral center, in which we treat sicker patients. As a consequence, some of our metrics of performance consistently fall outside the “normal” range. Whether these variances are fairly categorized is a constant source of debate amongst physicians and quality officers.

Quantifying quality for cath lab outcomes in the United States falls to the Cath Lab PCI Registry administered under the American College of Cardiology (ACC)’s ever-growing suite of studies grouped within the National Cardiovascular Data Registry (NCDR). The registry was started as a joint partnership between the Society for Cardiovascular Angiography and Interventions (SCAI) and the ACC in 1997 and recently, after 9 years of stability, saw a significant upgrade from version 4.4 to version 5.0.

This upgrade has drawn widespread criticism from the cardiovascular community fearful of the increased burden of data collection and the elimination of the clause that allowed data extractors to enter a default value for a field in which the specific data element could not be found. As a result, a large number of cases are now classified as “insufficiently complete” due to the huge variability, lack of standardization in procedure notes, and the presence of data fields that are not usually clarified in the course of standard dictation. Within my healthcare system, the difficulty in completing data collection led to the forced introduction of an additional structured form to supplement the standard procedure note, adding to the bureaucratic load of busy physicians.

Despite the widespread initial dissatisfaction with the 5.0 upgrade, there are many benefits built into this version. For the first time, data will be collected to allow for risk stratification and improved definition of cardiac arrest patients. Numbers of patients in cardiogenic shock treated with mechanical circulatory support such as Impella percutaneous LVAD, Tandem Heart, and ECMO devices are being added, which can be expected to improve the inevitably high, “definitional” bleeding rates associated with the use of these otherwise lifesaving devices. This problem has long plagued quaternary centers who use these devices in the rescue of cardiogenic shock patients.

Importantly, the upgrade should improve the ability of the database to categorize the appropriateness of the procedure. A full 40% of cases in version 4.4 were previously deemed “unclassified” because the extracted data could not account for all the scenario permutations that lead up to the coronary intervention. Long overdue, data on intracoronary physiology studies—fractional flow reserve (FFR) and iFR (instantaneous wave-free ratio) and the intracoronary imaging modality of optical coherence tomography (OCT)—are now being recorded to help justify procedures. Instances in which surgical treatment with CABG is an option but the surgeons or the patient turn down the operation are also finally being captured, instances that have been shown to confer increased risk. Optional fields have been added to capture patient reported quality-of-life metrics such as the Seattle Angina Questionnaire or the Rose Dyspnea scale.

Part of the drive to upgrade was a need to reduce data collection burden. To achieve this, more than a hundred research-only or data elements not necessary for supporting nonadjusted outcomes, appropriate use criteria, or quality metrics were eliminated. This reduction was offset however by a large increase in data points for demographic, episode of care, history, risk factors, discharge, intra- and postprocedure events as well as an optional section for follow-up; resulting in a total data element count increase from 262 to 345. On closer examination, the large increase represents to some degree an overestimation. Certain elements are only required if triggered by an appropriate entry and other data points are optional, such that the final minimum tally is 266. Coupled, however, with the loss of default value entry as described above, the net effect for end users has been a significant increase in data entry time per case.

In an effort to further understand the sweeping changes and devise local strategies to improve the efficiency of our participation in NCDR Cath PCI, I attended the pre-ACC 2019 Quality Conference here in New Orleans. The most striking feature of the conference was extremely low physician presence amongst the 1,500 meeting attendees. Almost all the attendees were nurses or CV techs who have been employed by their healthcare systems as data abstractors. From numerous discussions over the 3 days, one thing became abundantly clear: relationships between nurses or quality champions and many of their institutional physicians are often adversarial and confrontational.  This seems to stem from the different lenses by which the quality challenge is being approached by the many important stakeholders and is further aggravated by the logistical structure we have built ourselves.

The key steps in the quality-quantification production line look something like this:

1.      The cardiologist performing the procedure produces a cath lab report describing the procedure.

2.      A nurse data abstractor, hired by the local healthcare system, reads notes remotely after the case has concluded, and abstracts a very large amount of complicated data points required by the NCDR CathPCI registry electronically.

3.      The NCDR processes the data, compares multiple outcomes for the particular site with all other site participants using complicated statistical modeling then generates quarterly reports several months after the data is submitted.

This production line frustrates both physicians and data abstractors since a large amount is lost in translation and efficiency between step 1 and 2, compounded by the rigorous demands of rigidity in data collection and continued physician reluctance to participate in structured reporting. NCDR has been cautiously moving in the direction of producing a standardized EHR cath report template that could act as a direct source of registry data partially, but the slow pace of change is due to the difficulty in working with the behemoths of EHR who continue to ignore the importance of frontline healthcare worker interface satisfaction and efficiency.

Despite the ACC publication of a 2014 health policy statement addressing the dire need for structured cath lab reporting, we appear to have made little actual progress on the ground 5 years on. As expected in the vacuum of an unmet need, multiple start-up companies have rushed in to fill the gap with a variety of cloud-based, electronic data solution platforms. Local implementation has not, however, been uniform due to the challenges of funding yet another duct tape third party solution to our EHR systems.

We have a mountain to climb in what should be an obvious, singular, unified cause: the delivery and achievement of quality for our patients. We need to be more agile in adapting to and harnessing technology. We need to merge the silos. We need to unify as aligned physician and nurse champions.  

In the meantime, we should all look forward to the release of the first report from the version 5.0 risk-adjusted data next month and hope this will represent progress.