The Academy’s IRIS® Registry has recorded more than 230 million patient visits, and its value to practicing ophthalmologists and researchers is growing. During Mid-Year Forum 2019, the registry was the focus of a session moderated by Michael Chiang, MD, Academy trustee-at-large and chair of the IRIS Registry Analytics Committee.
Abstract
The IRIS Registry is the largest specialty clinical database in medicine, featuring over 230 million patient encounters and 50 million unique patients. The value of the IRIS Registry to the profession is over $186 million collectively in avoided penalties in 2019 and bonuses on top of this for the vast majority of participants. Now, there are technology solutions to benefit Academy members and to bring value to their practice in other ways, including tools for more efficient clinical trial recruitment, and for looking at different patient populations and their outcomes.
Background Information
The Academy initiated the IRIS Registry on March 24, 2014 for the purposes of quality improvement and providing a method for ophthalmologists to benchmark their performance and to track patient outcomes. Over the years, the IRIS Registry has documented increases in performance over time, with improved patient outcomes and better adherence to standard processes of care, as well as helping ophthalmologists to meet federal government quality reporting requirements.
Summary of Comments From Guest Speakers
Michael F. Chiang, MD, trustee-at-large, Academy Board of Trustees
Associate Director, Department of Ophthalmology, Oregon Health and Sciences University
Dr. Chiang discussed the cycle of continuous quality improvement, starting from biomedical research to standards to creation of guidelines to information decision support. This is actually coming to fruition because of the incorporation of real-world data from sources like the IRIS Registry.
The IRIS Registry is the nation’s first comprehensive eye disease clinical database. It was designed to improve care delivery and patient outcomes, to provide individual feedback on performance and comparison to benchmarks, and to help practices meet Merit-based Incentive Payment System. As of April 1, 2019, there are 234 million patient visits, representing 53 million patients. There are 14,793 physicians from 2,937 practices contracted for EHR integration, which includes other eligible clinicians practicing with the ophthalmologists.
Anne L. Coleman, MD, PhD, president-elect, Academy Board of Trustees
Fran and Ray Stark Professor of Ophthalmology at the Jules Stein Institute, David Geffen School of Medicine and professor of epidemiology at the Fielding School of Public Health
Dr. Coleman discussed the definition of big data, which can be characterized by the following features: high volume, a variety of sources, high velocity in terms of accessing or acquiring data quickly and accuracy of the data, or veracity. It is important to assure that the data is accurate.
There are different types of big data, including traditional medical data such as electronic health record data, OMICs or large-scale datasets in the biological field like genomics and microbiomics. Data sets also come from social media, including the internet, mobile applications, sensor devices and other technology like wearables that hold patient-generated data.
Wearable computing device information could potentially be uploaded to the electronic health record to see the level of physical activity of patients. With this patient-generated information, we as physicians can better understand diseases, risk factors and how to prevent poor outcomes.
Over the past decade, we started with big data through the analysis of large claims databases. Claims data can be used to assess our own practices and performance, compare to peers, understand risk factors, estimate adherence to therapy and evaluate utilization.
One of my early studies was to evaluate the rate of gonioscopy through claims data and because the rate was lower than expected, it was critical to understand the reasons in the real world setting for the lack of documentation. The benefits of claims data are that the potential sample sizes are much larger, patients can be followed up for outcomes and costs, modeling can be performed to account for potential confounding factors, data on nonocular conditions can be added to the ocular information, and studies can be much less costly than traditional clinical trials.
Dr. Coleman explained why big data is necessary with an example of endophthalmitis occurring after cataract surgery because the rate is so low with a background rate of 0.14%. To detect a 50% increase in the rate of endophthalmitis, the sample size would need to be 58,786 patients in each group. To detect a 25% increase in the rate of endophthalmitis, the sample size would need to be 207,182 patients in each group. To detect a 10% increase in the rate of endophthalmitis, the sample size would need to be 1,175,948 in each group.
The seminal claims data analysis article in ophthalmology was published in 1991 on a 50% sample of Medicare patients. The study reported on 338,141 cataract surgeries, with a rate of 58% of extracapsular extractions and 30% intracapsular cataract extractions.
Regarding the rate of endophthalmitis following cataract surgery, patients who had intracapsular cataract extractions had a 0.17% rate compared with a 0.12% rate in patients with extracapsular cataract extractions or phacoemulsification, a finding that was statistically significant. This finding, along with other factors, ultimately led to the predominance of extracapsular cataract surgery over intracapsular cataract surgery. However, if the number of patients were lower in the sample, it would not have been possible for the study authors to detect this rate of difference.
In a recent study of glaucoma surgery in patients receiving corneal transplants, based on a 2010-2013 5% random Medicare claims dataset, the total sample size was 3,098 patients, including 1919 EK, 1012 PK, 46 ALK, 32 Kpro and 89 both PK and EK. The rates of glaucoma surgery ranged from 6.1–9.4%. But a few of the different transplant procedure groups were few in number, underscoring the fact that the 5% Medicare claims database is too small for some analyses.
The limitations of claims and EHR data are that there are potential misclassifications, with inaccurate coding, missing data and incomplete data. Claims data lack clinical information such as visual acuity, intraocular pressures and medications. Confounding by clinical indication is another limitation of both claims and EHR data because this is the art of medicine, and this could create a bias. This bias is addressed through randomization in a prospective clinical study.
Dr. Coleman described the David E. I. Pyott Glaucoma Education Center on the Academy’s Ophthalmic Network for Education. Based on the IRIS Registry database for 2017, there were 3.92 million individuals with glaucoma, with the largest group at 2.45 million patients with the diagnosis of glaucoma suspect. There were 145,600 total glaucoma surgeries, with laser trabeculoplasty comprising about 55% of the total surgeries performed.
Dr. Coleman concluded that claims data ushered the era of big data. EHR data advantages include the presence of clinical observations and measurements such as visual acuity and intraocular pressure. We are in the midst of an information wave, with the amount of digital data in the world being more than 40 zettabytes. Big data brings with it the capacity to continuously improve outcomes and the quality of care we provide.
Aaron Y. Lee, MD, assistant professor of ophthalmology, University of Washington
Drs. Chiang and Lee demonstrated the member value tools that Verana Health has created now. These tools include a outcomes tracker module and a treatment tracker module that help practices to look at their patient populations in different ways to review patient outcomes, and a clinical study module to help practices that are interested to learn about ongoing clinical studies and to identify patients that might meet the inclusion criteria for the study.
Summary of Audience Comments
- How can patient-generated data be incorporated? To address patient-generated data, informed consent would be required for patients. Patients have been coming in and showing their wearable device data to physicians. This data is useful to provide meaningful advice for personalized care. For wearable data, patients could directly contribute and approve their data for use of research purposes.
- It would be good to address costs with big data and could be very informative. This would need to be carefully considered because of all the ramifications. People care a great deal about their vision, and vision always appears to be No. 1 or No. 2 in their valuation of their well-being and their future. If we move towards more a value-based purchasing mode, this would be an interesting approach.
- We would be interested in using these tools for our own clinic. These tools are going to be made available for Academy members participating in the IRIS Registry in the next few months.
- The sole quality metric in health care appears to be patient satisfaction. We should brand these as doing all these activities for the satisfaction of patients.
- Has the IRIS Registry been discussed with Food and Drug Administration (FDA) about usage for post-approval marketing studies and justification for new indications for treatment? Phase IV post-marketing approval studies could probably be easily performed using the IRIS Registry. There appears to be interest from the regulatory agencies in an independent registry for performing these studies. There is one post-approval study now planned using the IRIS Registry database. IRIS Registry could also be used as a platform for prospective, clinical trials. IRIS Registry could be a very good source for determining disease prevalence, particularly for rare diseases.
- The idea of incorporating best practices such as patients taking hydroxychloroquine and who haven’t received a visual field test and reminding physicians of best practices could be considered. It could also be good for the PPP guidelines to be encoded in the future to help elevate how practitioners provide care.
- Topol recently tweeted that a great deal of medical innovation has been in the area of ophthalmology with artificial intelligence. Dr. Chiang asked the audience what else would be useful for practices in the member value tools. One member noted that feedback about how our practices are doing would be helpful. Another member noted that with respect to amblyopia, it would be good to see treatment outcomes and give us a concept of how we are doing.
Pediatric ophthalmologists do not have to participate in MIPS program but there is a significant volume of data in the IRIS Registry largely because pediatric ophthalmologists are part of participating group practices. There are newer quality measures for treatment outcomes, i.e., amblyopia, pediatric strabismus and adult strabismus. Another member noted that it might be helpful for physicians to note when they changed their practice patterns so that they can view differences for improving patient care.
- The IRIS Registry would be very interesting for ocular oncology and rare diseases. Patients are seen in children’s hospital and oncology centers which are not integrated with the IRIS Registry but it would be helpful to capture the continuum of care for these patients. Currently, there is no way to capture this information outside of the ophthalmology practice.
High Priority Objectives
To enhance the member value tools by continuing to collate member input and feedback
Read the full Mid-Year Forum 2019 Report: View as webpages or as a PDF (360 KB).