Big data analysis can benefit ophthalmic practice and bump up the bottom line
The evaluation and analysis of large sets of data from registries, health care companies and government can improve patient treatments and cut down costs for ophthalmology practices.
Ophthalmology is “one of the most data intensive” communities in the health care industry, but the analysis of big data has not yet taken a step past a “cursory look” of how a patient is doing at follow-up, OSN Cataract Surgery Section Editor John A. Hovanesian, MD, FACS, said.
“We are constantly measuring the eye, whether it’s an OCT, corneal topography or biometry. We collect data all the time, but we don’t use it beyond a very cursory look at how the patient is doing,” Hovanesian said. “In fact, the HIPAA laws make it difficult to share data because you have to deidentify data in order to do research on it and share it outside the covered entity that is taking care of the patient.”
Electronic health records should make the analysis and accessibility of big data easy for physicians, but many of the systems are too cumbersome and store data in ways that are not compatible from one system to the next. The systems do not “talk” or interact with each other, and there is no incentive for doctors to share their data with colleagues, Hovanesian said.
There is tremendous potential for big data to transform aspects of health care and medicine, but many fields are still trying to figure out how to use the information to improve patient care and change practices, Harry Glorikian, author of MoneyBall Medicine, said.
Big data in medical fields and health care describe “large data sets — instead of megabytes, think terabytes, petabytes, etc,” Glorikian said.
“I think the term also encompasses the data analytics behind the data itself since having the data isn’t very helpful unless you’re going to organize it, analyze it and hope to glean some insights from it,” he said.
Oncology, as a collective field of medicine, has utilized the analysis of big data most effectively to improve patient treatments. By using the large amounts of data generated from clinical trials and genomics, patient care has improved extraordinarily over the past two to three decades, Glorikian said.
“Doctors aren’t treating cancer patients using the exact same treatments for each cancer type anymore. We’ve learned that cancer isn’t just one disease. It’s many different types, even within a category like breast cancer or colorectal cancer. Researchers have identified numerous biomarkers, drivers of cancers, and we’re now taking that data to give patients more personalized cancer treatments based on those biomarkers,” he said.
The potential for big data in medicine is there, but most organizations are still working out infrastructure issues to get at the information or are developing pilot projects to test out ideas and processes before extending it systemwide, Glorikian said.
Embracing big data
Most fields of medicine have started to embrace big data analysis in some capacity. Radiology uses big data and advanced analytic techniques, such as cognitive computing and deep learning, to create algorithms that can scan patient images. If an area of concern is identified in these images, the technology can alert doctors or help narrow down unusual diagnoses, Glorikian said.
“Helping physicians get to the right diagnosis more quickly can mean the patient gets treatment faster or might reduce complications from misdiagnosis. Drug design and development are being informed by molecular modeling programs built on pharmacokinetic databases. Big data is also impacting regulatory decisions. The recent Keytruda approval is a case in point. Identifying the specific patient population that could benefit from the treatment across a variety of cancers would have been improbable without the insights derived from genomic studies,” he said.
The FDA granted accelerated approval in May to Keytruda (pembrolizumab, Merck) for the treatment of patients whose cancers have a specific biomarker, the first time the agency has approved a cancer treatment based on a biomarker rather than the location in the body where the tumor began, according to an FDA press release.
But while some specialties have embraced the use of big data analysis, ophthalmology lags behind, Hovanesian said.
EHRs can collect data
Ophthalmologists who use EHR systems have massive amounts of data at their fingertips, but the systems are not designed to report on the data they take in to help improve outcomes or clinical efficiencies, OSN Cataract Surgery Board Member Cynthia A. Matossian, MD, FACS, said.
Matossian said that her practice, Matossian Eye Associates, found a solution to the problem by hiring a third-party company to delve into the data generated by its EHR system. The company generates reports to be used by the clinic to proactively improve patient experiences and help guide treatments in some cases.
The third-party company uses the EHR server to collect data with the permission of the medical records company and through signed mandatory confidentiality agreements.
“There’s a lot of information out there that we regularly look at. Any company that wants to stay successful, that wants to stay on course for whatever their goals are, needs to know the data. Without your coordinates, how can a ship get from point A to point B? You’re just drifting. Our coordinates, our data, are what help guide us there,” she said.
The reports generate data points such as the number of new patients seen in the clinic per month, the number of surgeries, surgical outcomes, complications during surgery and the number of emergency patients served per day. These data reports help the clinic proactively accommodate the daily needs of their patients, Matossian said.
“For example, one thing that we do, we survey our patients after every single encounter. Once they leave, we survey them to find out what their experience was at Matossian Eye. Whether they’re a patient for the very first time or an established patient, their experience may be different than what we think we’re delivering to our patients. Getting feedback from our patients is critically important. Then, having a mechanism to address the feedback is even more important. ... You can get all the feedback you want, but if you don’t have a means of categorizing, delegating it, following up with it, it’s lost information,” she said.
Similar big data tools
Hovanesian created a similar tool for his practice and colleagues, MDbackline, which is designed to complete contact between the doctor and patient by reaching out to the patient after he has been seen in the office to find out how he is doing with the treatment.
The software reaches out to every patient after undergoing surgery, Hovanesian said. It queries the patient’s satisfaction rate, symptoms, adverse events and outcomes, and logs the information into the clinic’s records.
Patients who are happy with their outcomes are given the option to share their reviews and experiences on social media sites such as Yelp or Healthgrades, he said.
For patients who are not happy with their outcomes, the information is sent back to the practice for doctors or staff members to review and decide if the patient needs to be brought back for a follow-up appointment or if a phone call checkup is necessary, he said.
“Then, of course, the big data part of this is we have an aggregation of information from hundreds of patients, thousands of patients who have been treated, so in my practice we can publish data on what our satisfaction rates are. We have collected all kinds of quotes and information from individuals that make the case of why people would want to come to see us. We are using this data to negotiate higher payer contracts — typically, 1% to 2% better pay from commercial contracts. We can tell them better what our value is beyond just performance and service. It also helps to meet the rules for MACRA-MIPS, which is part of the government’s pay-for-performance criteria,” he said.
In addition to third-party generated data reports, Matossian said her clinic participates fully in the IRIS Registry, the American Academy of Ophthalmology’s comprehensive eye disease clinical registry.
Participating in the registry has helped Matossian Eye Associates reach Physician Quality Reporting System and Meaningful Use requirements, but the influx of data from the registry should have a greater impact on the field in the future, she said.
“It takes time for us to participate in it, but there is a value add to the practice and to us. How will it impact us in the future? A lot of this is uncharted territory. But I’m sure that more value and emphasis and scrutiny are going to be placed on available data for practices or individual surgeons. It could be looking at complication rates for doctors, what percentage of the time they are encountering a situation where they’re doing an unplanned vitrectomy — things like that may really come into play, potentially,” she said.
Right now, most U.S. ophthalmologists are participating in the IRIS Registry, which makes it a valuable tool, Michael F. Chiang, MD, of Oregon Health and Science University, said.
The value of the registry lies in its vast amount of data and its potential to identify associations among ophthalmologic diseases, understanding how ophthalmologists around the country treat certain diseases and what may happen during rare events, he said.
“We can do that now on a scale that has never been practical before. We can look at relationships, trends, what bad outcomes are associated with what complications. For new devices, from a post-market surveillance perspective, what are the complications associated with them? That’s something that is unprecedented right now, to be able to do that. We have a lot of public health and public policy issues that can be examined with this big data,” he said.
The future of registries
Chiang pointed to the results of the Framingham Heart Study, the groundbreaking study that began in 1948 with 5,209 patients from Framingham, Massachusetts. The study’s findings were the first ever published on what increases a patient’s cardiovascular risk.
The first enrolled patients were studied and followed for a decade, with new generations of patients still being enrolled and evaluated today. However, with the massive amounts of data now available, such as with the ophthalmology registries, the role of such long and expensive studies may be evolving. In addition, registries can help accelerate recruitment of patients for clinical trials involving promising treatments, Chiang said.
“Hopefully with big data we’re going to move into an era where it doesn’t always take decades and tens of millions of dollars to identify those types of associations for eye disease. Hopefully because of things like the IRIS Registry we’ll be able to identify those associations much faster because we have these large-scale data repositories. There will always be a role for clinical trials, and as a field maybe we can identify those associations faster and test the more promising ones using clinical trials. I think that’s going to be one of the things, the rate of discovering knowledge, that will hopefully increase a lot faster. It’s what makes things really exciting in the field of ophthalmology overall,” he said.
In addition, the possibility of combining ophthalmology registries with registries from other specialties to better understand how eye disease relates to systemic disease is an exciting opportunity, he said.
Several studies have been published showing the benefits of the ophthalmology registries. During the Edward Jackson Memorial Lecture at the 2015 annual meeting of the American Academy of Ophthalmology, Anne L. Coleman, MD, noted that big data drawing on Medicare claims and the IRIS Registry can help ophthalmologists identify areas for quality improvement. By evaluating outcomes and complication rates in these big data sets, Coleman said they can provide better guidance and benchmarks to improve the quality of cataract surgery and other eye care procedures.
For example, a review of the IRIS Registry from 2013 to 2014 showed 18.3% of eyes 1 month after cataract surgery had visual acuity worse than 20/40. By analyzing these patients and noting similar trends and complications, ophthalmologists can learn from these outcomes and improve the quality of visual acuity in patients undergoing cataract surgery in the future.
Medical devices gather data
Ophthalmology medical device representatives are also designing devices to gather data from procedures and help guide surgeons with difficult cases. Carl Zeiss Meditec has more than 1 billion images captured from its imaging devices, all of which can be used to establish normative glaucoma and retina data, Andrew I. Chang, head of global sales, ophthalmic devices, at Carl Zeiss Meditec, said. Chang recently spoke on the topic of big data at the OCTANe Ophthalmology Technology Summit in Newport Beach, California.
This data collection system has been built into Zeiss software for its devices, such as OCT or a Humphrey field analyzer, and when linked up to the Zeiss work station, called Forum, it gives the surgeon the ability to monitor disease management and disease progression, Chang said.
“If you want to look at a patient’s first visit from the first time of their treatment to the last few visits, or over a span of 10 months or 10 years, that’s possible. Whatever span of time you want to gather, you can see. That’s the beauty of it. I think that’s so important because it can lead to better care in the future. We know we have this huge portion of patients coming in worldwide that needs diabetic screening and diabetic treatments. We don’t do the treatment, but we can certainly help the clinicians determine whether they should have an intervention of some sort and manage the disease,” he said.
The platform for the Zeiss imaging devices provides the diagnostics to identify, detect and monitor disease progression in patients. The devices gather data throughout the continuation of treatment and create more data sets, more in depth scans and readings to help clinicians make decisions faster, Chang said.
One aspiration is to be able to utilize the data to help guide treatment in remote locations where Zeiss imaging devices are used for ocular screenings. The information taken from these screenings could be sent to the cloud, and a clinician in a central office could read the data and help identify whether the patient needs an additional point of care, screenings or scans, he said.
Ethical questions for big data
The future of big data could even include such everyday actions as internet searches to help with diagnoses, Hovanesian said.
For example, a study published in 2015 in the Journal of Oncology Practice found patterns in patient search logs can predict a future diagnosis of pancreatic cancer. Researchers found that certain search patterns and the time of these searches emerged for patients looking up information about specific symptoms such as indigestion and back pain, he said.
“They looked at what people searched for in the weeks and months previous to their diagnosis, and they found a very clear pattern. First, the same group of people that ended up having pancreatic cancer looked up information on indigestion and then information on how to treat abdominal pain. The progression from early indigestion and abdominal pain, it predicted quite specifically and quite sensitively the development of pancreatic cancer. In other words, are Google search terms a valid diagnostic test for pancreatic cancer?” Hovanesian said.
It raises the ethical question of when or if an internet search engine has the duty to log the data from those searches and inform the searcher that he should see a doctor, Hovanesian said.
“Does Google have a duty to report to people who ask those questions in that order? ‘Hey, you may want to get in touch with your doctor because this could be a symptom of pancreatic cancer?’ While we want to have our privacy respected, it’s not just a poke of the needle in your arm to draw blood to see whether you may just have the disease. It may be the color of your skin today or what you searched for on the web yesterday that can lead to a diagnosis in the future,” Hovanesian said.
At this point, no practice should resist the move toward big data and leaning on data reports to help guide practice and treatment and increase clinical efficiencies, Matossian said.
“I think that if your practice is doing everything right, doing the best you know how for your patient, you have good surgical skill sets and diagnostic and treatment modalities, there should be no fear of big data. It will only underscore how well you are doing. Some practices do worry about it, that it’s an invasion of privacy or things that they do within their practices will get aired, but if there’s nothing to fear and you’re proud of what you’re doing, it’s only going to help you succeed further,” she said.
- Coleman AL. Am J Ophthalmol. 2015;doi:10.1016/j.ajo.2015.09.028.
- FDA approves first cancer treatment for any solid tumor with a specific genetic feature. https://www.fda.gov/newsevents/newsroom/pressannouncements/ucm560167.htm. Published May 23, 2017. Accessed on Aug. 1, 2017.
- Glorikian H, Branca MA. MoneyBall Medicine: Thriving in the New Data-Driven Healthcare Market. New York: Productivity Press; 2017.
- Mahmood SS, et al. Lancet. 2014;doi:10.1016/S0140-6736(13)61752-3.
- Paparrizos J, et al. J Oncol Pract. 2016;doi:10.1200/JOP.2015.010504.
- For more information:
- Andrew I. Chang can be reached at Carl Zeiss Meditec, 5160 Hacienda Driver, Dublin, CA 94568; email: firstname.lastname@example.org.
- Michael F. Chiang, MD, can be reached at Oregon Health and Science University, Casey Eye Institute, 3375 SW Terwilliger Boulevard, Portland, OR 97239; email: email@example.com.
- Harry Glorikian can be reached at firstname.lastname@example.org.
- John A. Hovanesian, MD, FACS, can be reached at Harvard Eye Associates, 24401 Calle De La Louisa, Suite 300, Laguna Hills, CA 92653; email: email@example.com.
- Cynthia A. Matossian, MD, FACS, can be reached at firstname.lastname@example.org.
Disclosures: Chang reports he is an employee of Carl Zeiss Meditec. Chiang reports he is an unpaid member of the scientific advisory board for Clarity Medical Systems and a consultant for Novartis and receives grant support from the National Institutes of Health and National Science Foundation. Hovanesian reports he is a consultant to Veracity, Alcon and Carl Zeiss Meditec and is an equity holder in MDbackline. Glorikian and Matossian report no relevant financial disclosures.
Click here to read the , "What are the advantages and pitfalls of using big data to inform clinical practice?"