As the elderly population rapidly expands, retina specialists are serving an increasing volume of patients in their clinics. In an effort to improve clinical outcomes and increase efficiency in retina clinics, physicians and companies are integrating the use of big data into practices. There is an immense volume of clinical and business data in a retina clinic, and the challenge is organizing these data so that clinicians can use it meaningfully to meet patient care goals. Historical data such as claims data takes time to collect, organize, and analyze. With big data, data are available in real time, but challenges include information overload, location of data in multiple systems, and organization of the data to create reports. Retina specialists and practices are looking to implement new technologies and cloud-based programs that can use real-time analytics of big data to improve efficiency in retina clinics.

DEFINITION OF BIG DATA AND ANALYTICS

Big data is a term used to describe extensive data sets, characterized by the 3 V’s: volume (large amounts of data), variety (data heterogeneity), and velocity (speed of access and analysis). The size of the data exceeds what traditional database software is able to store, process, and analyze. This massive amount of data may be structured or unstructured and can be analyzed to solve future problems. Analytics is the systematic processing and manipulation of data to discover trends, relationships, and patterns.¹

ANALYTIC TOOLS WE USE

Our practice uses a multitude of analytic tools including our electronic health record (EHR) and practice management (PM) programs. We use Nextech IntellechartPro, which is a cloud-based EHR, for clinic visit documentation. Along with clinical documentation, our EHR also captures all patient communication, coding of clinical encounters, and imaging data. This immense repository of data is then coupled with our PM system, which then contributes claims data, denials, reimbursements, patient demographic data, and referring physicians. This meets the criteria for big data: volume, variety, and velocity.

Physicians and staff usually access EHR and PM data through the software front end, where they can interact with various screens to look at individual data elements. They can generate reports that summarize a section of the data present within the database. What is not possible with these tools is the generation of cross-platform reports that combine disparate data sets. In our practice, we have paid to have our EHR and PM vendors provide database-level access, so we can feed the rapidly changing large data sets into tools that are specialized in analyzing big data. This allows us to answer questions that would otherwise be impossible to answer using traditional methods. As an extreme example, let us imagine a practice that would like to see how inclement weather impacts appointment cancellations by regional office. Utilizing these tools, we can actually answer this question.

We use the analytics platform Tableau, which allows the connection of separate data sets to perform analyses. In our hypothetical example, a ZIP code level data set of daily snow accumulation can be obtained from a publicly available government database. This ZIP code level can then be matched to each patient’s home ZIP code, which can generate a list of the snowfall for each of their appointments, and the patient data can then be tied to the office location using the PM appointment data. Tableau can then select and visualize the percentage of patients who reside in each zip code who cancel based on different levels of snowfall. The analysis can then be furthered by integrating EMR data to assign risk scores for vision loss based on clinical parameters, and also used to determine how that impacts the chance a patient will cancel their appointment or not.

At present, such nuanced analysis is typically reserved for larger technology companies. However, the ability to perform these analyses is available for businesses of all sizes. It may seem like questions such as these serve only to satisfy curiosity and do not provide any practical benefit. The true power of big data, however, is in using the data to make predictions. The tools described above can be used to make practice decisions. With the above analysis, let us imagine the results reveal that if a patient lives greater than 10 miles away from the office, they have a 90% chance of canceling when snowfall exceeds 4 inches. The exception might be a patient who had a decline in vision due to a missed injection appointment within the past 6 months; in this case, they never cancel. Tableau can create a real-time dashboard and provide practice managers with an estimate of how many patients will cancel for an office based on current weather patterns, the status of the patient’s retinal disease, and the distance the patient lives from the office. Dashboards such as these can be created once, and automatically update their models based on historical data, and function in perpetuity allowing practices to make data-driven decisions.

ALTERNATIVE ANALYTIC TOOLS

In addition to Tableau, alternative analytic tools include SAS, SPSS, and Python. SAS is a closed-source, advanced analytics software that specializes in data management, statistical analysis, predictive analytics, and business intelligence. Operational challenges such as workforce planning, performance management, and cost analysis can be addressed with SAS.^2,3 IBM SPSS is a statistics software with capabilities that include data preparation, descriptive statistics, linear regression, and visual graphing. SPSS is used to create tables and visual graphs to depict relationships and to make predictions.² Python is an open-source programming language used to build machine learning algorithms. Managing patients and scheduling appointments can be a challenge in a clinic with limited staff. Python applications may help with appointment scheduling, medication refills, answering frequently asked questions, or updating health information.⁴ Our office has elected to use Tableau because we have found it to be the most user friendly of the analytic packages, allowing incredibly complicated analyses to be performed using graphical interfaces and drag-and-drop functionality. SAS and SPSS are capable of doing equivalent analytic work, however greater expertise is needed to perform these functions. Python has an immense benefit of being free, but to perform analytics requires software development experience.

BENEFITS AND VALUE OF ANALYTICS IN SOLVING PROBLEMS

Many retina clinics are continuously seeking ways to increase productivity, decrease costs, reduce chances for medical error, and increase efficiency. Common challenges for clinic flow include decreasing patient wait time, increasing the ratio of face time spent with patients to visit duration, increasing physician productivity, increasing patient capacity, and decreasing physician workload. Operational challenges include clinic staffing, staffing schedules, and inventory management. Real-time analytics may help us to solve these problems with automation. The main advantage of real-time analytics is that no staff time is required to generate up-to-date reports. The software produces informative measures and metrics quickly, and these can then be used to make efficient, data-driven decisions.

DATA ANALYTICS ON A NATIONAL LEVEL

The IRIS Registry (Intelligent Research in Sight), established in 2014, is one of the largest specialty society clinical data registries in medicine, with deidentified data from more than 75 million patients over the course of approximately 10 years.⁵ It is also a tool that can analyze data to generate benchmark reports. Verana Health is a health care analytics company that manages clinical data registries and has partnered with the American Academy of Ophthalmology on IRIS Registry analytics since 2017.⁶ Verana Health can integrate the data in various EHRs to provide practices with meaningful performance feedback, quality reporting, and Merit-based Incentive Payment System reporting, and it may identify patients who qualify for clinical trials. EHRs at ophthalmology practices will soon be integrated with the IRIS Registry by way of Verana Health.

Data-driven decisions should not be the exclusive domain of practices. With national-level data, analytics can help inform public policy decisions. Recently, our group collaborated with SamaCare, a company that provides prior authorization assistance for Medicare Part B drugs. With their large national data set, we were able to analyze more than 33,000 prior authorization requests to find that there were widespread delays in care due to anti-VEGF prior authorization requirements. Additionally, we were able to identify that retina specialists had very few denials (<3%). With data analytic tools, we can provide legislators with clear evidence of the negative impact these policies are having on quality patient care.

DATA ANALYTICS ON A PRACTICE LEVEL

On a practice level, data analytics can improve retina clinic efficiency, productivity, and clinical outcomes. For example, geographic data from our patients and referral sources can be used to compare different office locations and may offer insights into the strategic opening or closing of office locations. In addition, data on patient volume per physician may be analyzed to potentially optimize the distribution of new patients or overbooks within a practice. Comparison of staff productivity levels may help to identify team members who need additional training or may incentivize staff to optimize their performance. Furthermore, patient wait time data can be trended against time of day to both improve patient satisfaction and optimize physician schedules. Patient wait times may also be broken down into time spent in each phase of the visit, and these metrics can be analyzed to identify bottlenecks in clinic flow. Implementation of a data simulation model reduced mean total waiting time in an ophthalmology clinic by 21%,⁷ while application of Lean and Six Sigma process flow maps decreased patient flow time by 18%.⁸ Decreased patient wait time is correlated with increased patient satisfaction.⁹ As reimbursements decrease and practice costs increase, practices are looking to increase efficiency in their clinic flow, and this would in turn increase revenue.

Further insights into the organization of big data and the use of generated reports to implement meaningful changes in retina clinics may improve productivity, efficiency, and outcomes. Retina specialists have an opportunity to embrace this new technology to improve the care of their patients. RP

REFERENCES

1. Executive Office of the President. Big data: seizing opportunities, preserving values. 2014. Accessed March 31, 2023. https://obamawhitehouse.archives.gov/sites/default/files/docs/big_data_privacy_report_may_1_2014.pdf
2. IBM Corporation. Propel research and analysis with a comprehensive statistical software solution. May 2021. Accessed March 31, 2023. https://www.ibm.com/downloads/cas/ERYRD6G5
3. SAS. Health care operations. October 26, 2022. Accessed March 31, 2023. https://www.sas.com/en_us/industry/health-care/solution/operations.html
4. Python. Using Python to develop a patient health portal. Accessed March 31, 2023. https://www.python.org/success-stories/using-python-to-develop-a-patient-health-portal/
5. American Academy of Ophthalmology. IRIS Registry data analysis. February 13, 2017. Accessed March 31, 2023. https://www.aao.org/iris-registry/data-analysis/requirements
6. American Academy of Ophthalmology. MIPS 2022—Verana Health and the IRIS Registry. May 3, 2022. Accessed March 31, 2023. https://www.aao.org/eyenet/article/mips-2022-verana-health-iris-registry
7. Kern C, König A, Fu DJ, et al. Big data simulations for capacity improvement in a general ophthalmology clinic. Graefes Arch Clin Exp Ophthalmol. 2021;259(5):1289-1296. doi:10.1007/s00417-020-05040-9
8. Ciulla TA, Tatikonda MV, ElMaraghi YA, et al. Lean Six Sigma techniques to improve ophthalmology clinic efficiency. Retina. 2018;38(9):1688-1698. doi:10.1097/IAE.0000000000001761
9. McMullen M, Netland PA. Wait time as a driver of overall patient satisfaction in an ophthalmology clinic. Clin Ophthalmol. 2013;7:1655-1660. doi:10.2147/OPTH.S49382