How to be Successful in Healthcare Analytics (Part 1)
The healthcare industry has evolved to be data-reliant, collecting volumes of unstructured data on a multitude of patients’ health and financial reports. With the progressive wave of healthcare analytics, a data-driven enterprise is the ideal standard for all health systems. But with inexperienced vendors unable to ground a strong framework to meet this standard, their haphazard adoption of data analysis poses a threat to a hospital’s clinical efficiency and cost. Thankfully, we have experts like Health Catalyst who have dirtied their hands to make vendors’ lives easier.
Health Catalyst is a company that provides data analytic services to healthcare organizations. Their client success stories range from optimizing patient satisfaction, to improving surgical services for successful financial and clinical outcomes. They currently serve over 250 hospitals and 3000 clinics —including Stanford Health Care.
With their understanding of the problems that arise with using analytics, they developed the Healthcare Analytics Adoption Model, which serves as a guide for organizations to be more data-informed and to implement a systematic framework. The model provides nine levels to ensure high quality medical delivery, however, in this article, we will be going through the first five. A second article will be released soon for the rest of the levels. If you’re interested in knowing more about Health Catalyst or its model, make sure to check them out at the links posted!
Before we dig into the adoption model, it’s important to keep this in mind before going any further:
big da·ta
/ bigˈdādə / Noun
Datasets of both unstructured and structured data that are so large in volume and are being generated at such a high rate, that it is highly difficult to analyze using conventional data management methods.
To summarize briefly: big data is a large component in healthcare, with hospitals managing data ranging from biometric and machine sensor data, to human-generated data such as physician hand-written notes and prescriptions. It is vital for this big data to be cleaned, vetted, and organized into a conceptual framework. If working with faulty data, then what data analytics does to help clinical operations run smoother would instead cause adverse effects — such as wrong-patient orders and not keeping up with patient demand — so shout out to the adoption model for helping us avoid that!
Level 0: Fragmented point solutions
“Inefficient, Inconsistent”
Healthcare organizations are novices at this point. The list of solutions they have compiled to improve medical care is analytically limited. They are “fragmented point solutions” — solutions which are unnecessarily labor-intensive, facilitates a lack of coordination, produces inconsistent reports, and causes an inefficient allocation of resources. There is no proper data integration and the analytical applications are isolated to a single department (finance, pharmacy, etc.), creating data silos and causing the mentioned problems to occur. But with the desire to use their data in an efficient manner, healthcare organizations go onto the next step.
Level 1: Enterprise Data warehouse
“ Collecting and integrating data”
This is where vendors grow out of using fragmented solutions to developing a proper data governance function: a collection of methods that manages the “availability, usability, integrity, and security” of the given data. Relying on data governance, Master Data Management (MDM) is used to link identity data (such as patients and providers) and reference data (common linkable vocabulary) together to create singular points that identify a piece of data, such as diagnosis codes or patients. Using these methods, the data is stored in an Enterprise Data Warehouse (EDW), a database essential to business intelligence, that reports and analyzes the data inputted. Its data content covers several domains, such as financial, clinical, material and supplies, patient experience, and if possible, insurance claims. By using data integration, EDW increases the data’s quality, provides a searchable metadata repository to the entirety of the enterprise, and furthers helps the organization make appropriate, data-driven decisions.
Level 2: Standardized Vocabulary and Patient Registries
“Relating and organizing data”
Now that we have all the data organized and stored in a EDW, it’s time for the data to be given a name. Reference data are further identified in this level, given industry-standard vocabularies, and are standardized across the EDW. Patient registries rely on the billing codes identified in the EDW to evaluate affordability and care procedures.
Level 3: Automated Internal Reporting
“Consistent production of reports and availability”
Analytics are focused on making basic operations of the healthcare system more efficient. Internal reporting ensures that consistent, accurate reports are readily available to the management of the organization, and helps facilitate coordination between all units of healthcare.
Level 4: Automated External Reporting
“Consistent production of reports and adaptability”
The reports here are focused on external needs, such as maintaining health registries, funds and payments, and societal databases. Data conformed to industry-standard vocabularies (done in level 2) from the EDW is inputted in MDM, and data such as patient records are available to anyone by just a quick search. The data content in the EDW has now extended to include patients’ clinical notes and reports.
Now that we’ve gone through different levels within this data-driven adoption model, let’s see how this would fit together in a single use case. A hospital starts off with solutions that have no data integration and uses inefficient analytical applications. Realizing how fragmented their solutions are, a hospital’s next step is to ensure data integration by developing a proper data governance function. After collecting all the data and using methods which ensure management and organization of the data, the data is then inputted in a EDW which analyzes the data and makes it accessible throughout the enterprise. Now that all the hospital’s data is stored somewhere, the data is labeled and identified. Using the identified data, the hospitals can start developing registries, i.e. patient registries. The EDW makes sure to provide consistent, accurate reports of the hospital’s functionings to management, and reports of the external workings — such as health registries and funds — of the hospital.
Reiterating a previous statement, Health Catalyst has success stories ranging from ensuring patient satisfaction in cost and care, to improving surgical services. An example of a health system which had several success stories with Health Catalyst is Community Health Network (CHNw), in Bell Gardens, California. One of their stories concerns pregnant patients struggling with substance abuse. CHNw saw that their rate of maternal substance use disorder was higher than the national rate. To address this using the analytical services offered by Health Catalyst, they implemented a treatment program offering medical assistance and therapy to women undergoing substance abuse, and developed the motto of “ focusing on compassion, not punishment”. CHNw saved over $760K in savings by reducing variable cost per case across the health system during the past few years, and overall had lower readmission rates and shorter stays. We will be evaluating more of Health Catalyst’s success stories in part 2, but this is one example of the benefits a health system reaps using Health Catalyst’s services. To check out more of their success stories, check them out at link posted and catch you in the next article!
This is the first article of a health series, meant to provide a very simple and broad overview of the Health Care Adoption Model, a model that has aided the progression in healthcare analytics. Stay tuned for part 2!
