The specialty of otolaryngology is committing increasing resources toward cultivating large clinical datasets. The promise of using “big data” to revolutionize healthcare has captivated the minds of clinicians, researchers, and myriad other stakeholders. During the past five years, a group of cochlear implant surgeons and audiologists have been methodically collecting clinical data surrounding cochlear implant care. Our experience with this database has highlighted important pearls—both inspiring and cautionary—that can hopefully inform future data science efforts within otolaryngology. A few of these pearls are highlighted below.
What Is a Registry?
A registry is a collection of information about individuals, usually focused around a specific diagnosis or condition. There are many types of clinical data registries. For example, some registries are very focused and contain information about specific diseases, conditions, or interventions. Other registries recruit patients with broad health conditions that can later be used to address a specific disease. Registries can also be used in multiple ways, including clinical research, quality improvement, cost effectiveness, policy making, or government-mandated quality reporting. Each type of registry has specific considerations that should be weighed before embarking on data collection. Some of these considerations include:
- What is the purpose of data collection?
- How will the data be collected?
- Who is going to benefit from the registry data?
- Who will pay to build and maintain the registry?
- Is the financial model sustainable?
- Who will be able to access the data?
The value that can be derived from a specific registry will largely depend on the answers to the above questions. Our experience with cochlear implant data collection has taught us that the devil is truly in the details. Proper planning can make the difference between powerful clinical insights and frustrating failure. The following lessons are just a few of the insights we have learned.
Lesson 1: There Is Power in Numbers
Otolaryngology is a diverse and creative specialty with dozens of hyperspecialized areas of clinical interest. While this makes for a fascinating career, it also means that generating studies with large sample sizes can be challenging. In a world where everything (clinical practice, policy, etc.) depends on good data, the curse of small, single-institution studies can threaten progress for niche areas of otolaryngology. This is certainly true with cochlear implants. By combining data from multiple institutions, we have been able to investigate clinical scenarios with statistical power that has previously not been possible.
For example, one group recently utilized our database to explore how cochlear implant testing conditions (i.e., using backgrounds of noise versus quiet) influence patient selection and outcomes. By combining data from many institutions, this study evaluated the experience of more than 2,000 patients. Additionally, the data used in this study came from real-world clinical practices. Without such a registry, data from these community-based practices would likely never be found in research studies.
Otolaryngology has entered an exciting era where data science can revolutionize our specialty. —Jedidiah J. Grisel, MD
Another group used our registry to explore how the time between cochlear implant candidacy evaluation and surgery varies for children with private insurance compared to those with state-funded insurance. These clinicians were able to clearly demonstrate that deaf children with state-funded insurance suffered significant delays in obtaining access to sound. These arguments were very persuasive when advocating for policy changes in the cochlear implant authorization process.
Clinical registries are a powerful way in which otolaryngologists can unite to advocate for our patients and our specialty.
Lesson 2: Balance Is Key
The value of a registry depends entirely on the data that feeds it. In our case, much of the data surrounding cochlear implant care simply does not exist in an extractable format within the electronic health record (EHR). For this reason, our group created its own web-based, HIPAA-secure clinical application to capture data. This process allowed us to customize exactly which data elements to include in data collection, making for robust datasets.
Customized data entry forms, of course, have disadvantages. These forms must be manually completed, which represents a tremendous hurdle in achieving a sustainable, long-term registry. Even the most committed clinicians struggle to allocate resources to consistently and accurately enter data. Manual data entry also increases the cost of maintaining a registry. Clinicians must either shoulder the cost of data entry themselves, or funds must be available to support this process. We have utilized both methods with our registry, and both methods can limit sustainability.
The ideal registry would identify strategic data elements that are readily available for passive data extraction, yet would still offer valuable clinical insight. Passively extracting poor quality data is not helpful. Manually entering data strains the resources of busy clinicians. Finding the proper balance is the key in achieving an effective clinical registry.
Lesson 3: Registries Are Expensive
There is no easy way to put it—registries cost money. Effective registries will likely cost more than most of us expect before they start generating a return on investment. Hyperspecialized registries require manual data entry, which is expensive. Passive data extraction requires sophisticated software. Both types of registries require secure data storage. Biostatisticians, data scientists, and clinical researchers are needed to make sense of these mountains of data. The costs associated with effective registries are not insignificant. Without proper planning, these costs can threaten a registry’s survival before it matures into a revenue-generating asset.
Many clinicians cringe at the concept of monetizing large datasets. The reality is that a poor or unsustainable financial model is still the most likely cause of a registry’s demise. The challenge for our specialty moving forward will be how to ethically create sustainable financial models surrounding registries.
Otolaryngology has entered an exciting era where data science can revolutionize our specialty. While the sirens’ call of “big data” is tantalizing, realizing this revolution can also seem mockingly just beyond our reach. Successful registries can build on the lessons learned from our early efforts, ushering in an era of care that is truly data-driven in the most noble sense of the word.
Dr. Grisel is a co-founder of the Auditory Implant Initiative, a nonprofit organization that developed and supports a cochlear implant registry. He is also with the Texoma Hearing Institute in Wichita Falls, Texas.