Hospitals are facing tremendous pressure to remain profitable. A recent study from the Congressional Budget Office suggests that absent productivity growth, 51—60 percent of hospitals could have negative margins by 2025. With the rising U.S. healthcare costs, $765 billion a year is estimated in waste, and there’s uncertainty about how much risk healthcare systems can and should take on. The solutions are clear: Organizations must drive revenue through consolidation, improve coding and clinical documentation to promote accurate reimbursement and lower costs by reducing preventable events and avoidable care—while learning about the value-based care world.
The healthcare industry is applying data analytics to address the need for revenue generation, gain insight into waste and care variation, and measure the impact of various quality improvement initiatives. Given current cost pressures, organizations need data and actionable information to effectively measure the total cost of care and allocate scarce resources to areas that have the highest probability of meeting outcome and financial goals.
"The proliferation of quality measures has made profiling and identifying potential quality problems increasingly complex"
Big Data in Value-based Care:
Applying artificial intelligence (AI) and other tools to a longitudinal set of patient records and accessing a comprehensive view of the patient based on clinical, claims, patient-generated, and socio-economic data can accelerate our journey toward value-based care and precision medicine. Unfortunately, most organizations have limited data, expertise, and resources to harness Big Data. Organizations are now hiring data scientists to manage the complexity of data quality, unstructured data, and lack of a single standard or tools (like risk stratification and attribution). Many have heard of the “4Vs of Big Data” (velocity, volume, variety, and veracity), but most overlook the importance of the fourth ‘V’ which represents the most difficult aspect given the heterogeneity, diversity, and complexity of healthcare language and data.
Too Many Measures
The proliferation of quality measures endorsed by organizations such as the National Quality Forum has made profiling and identifying potential quality problems increasingly complex. For payment applications, diverse quality measures must be combined and incorporated into systems to determine the net impact on payment. For most quality measures, there is no direct link to the cost or payment associated with the quality aspect being measured. Many of the quality measures are process measures, such as adherence to sterile techniques in the OR. The challenge of physician burnout increases with the ever-expanding documentation requirements for these measures.
The significant issues with data quality and documenting care variation within Electronic health records(EHR) systems are due to varying physician preferences and documentation styles, lack of standardized clinical information models for diseases, and overall usability challenges with the current EHRs. For example, patients with CHF or COPD do not have consistent information documented in the EHR, despite the many clinical guidelines for care. EHR companies have not adopted a standardized template with common data elements, which would help address the documentation burden on physicians.
Once the data is cleaned, aggregated, and linked to individual patients for longitudinal analyses, it must be organized or enriched to allow for the development of risk-adjusted quality outcomes across facilities and allow for evaluation of cost analysis or payment models. Without such enrichment, we cannot perform an apples-to-apples data comparison to accurately attribute costs and outcomes. Measuring and incentivizing payment for performance or outcome requires that physicians who take care of the sickest patients are not penalized when compared to those who take care of healthier patients. In addition, patients must be grouped to allow segment management One commonly used regression model produces a clear picture of the individuals who are utilizing care at a higher rate, but it does not allow for patient groupings that clinical teams can use to manage segments of a population and track improvement over time.
Big Data Needs a Framework for Transformation into Smart Data
Transforming Big Data into actionable insights requires philosophy and framework for value-based care. Without a clear understanding of what we want to learn and how best to measure our data, Big Data’s potential will not be realized. To understand the interdependency between the length of stay and complications or avoidable events, for example, the data must be linked and enriched to allow correlation analysis of root causes for potential action. Figure 1 proposes one such framework for consideration.
What does all this mean for integrated healthcare systems? Cost reduction is crucial for all CEOs, and this means identifying variation or waste in the system, such as potentially preventable readmission or complications that could lower costs and increase quality and outcomes. It requires a consistent set of outcome measures within a larger analytic framework to drive the transformational changes needed to achieve these goals. This means first and foremost that HCOs need the Big Data along with data enrichment (risk stratification, benchmark, attribution, and more) down to the individual service line and physician level. Adequate risk adjustment ensures an accurate determination of the relative rate of negative outcomes and establishes fair rewards and penalties. HCOs must either acquire the skill sets to tackle the process of data normalization and enrichment or outsource the work, without underestimating the level of complexity and the resources needed.
In conclusion, healthcare organizations should consider the following goals in capitalizing on Big Data to meet the growing demands of value-based care:
• Healthcare organizations must continue to focus, manage, and lower costs while optimizing revenue through:
1) Consolidation/partnership with payers or other providers
2) Focus on waste reduction and variation
• Health IT companies should support HCOs with Smart Data, not just Big Data, and focus not only on aggregating, cleaning, and normalizing data but also on categorization and enrichment of the data with risk stratification and benchmarking for improved performance management.
• Payers and the Government should focus on measurable outcomes tied to significant cost savings with an appropriate and transparent incentive program to make healthcare more affordable and sustainable.
• Apply Smart Data built on an analytic framework based on a philosophical foundation of quality outcomes.