I grew up in an area of the country where it rarely snowed, but when it did, it was difficult to predict. Years of anticipation was followed by disappointment, and occasionally a few unexpected and ill-prepared surprises. Thanks to advancements in meteorological technology, weather forecasting has improved over the years allowing us to better predict inclement weather. Similarly, so has the idea of using data forecasting to identify impactful and cost-efficient use of claims and clinical resources, to help eliminate unnecessary clinical expense while delivering the highest impact.
Identifying risk is not new to the industry and has become a pre-requisite for effectively managing claims. But analysis is only as strong as the data collection, analytic capabilities and, most importantly, what actions you take. To get actionable insights, payers need clean data, which requires data aggregation, normalization, standardization of terminology and elimination of duplicate information, just to start. Building a risk identification program moves the traditional approach of transactional interventions, to a data-driven model that allows organizations to identify at risk claims and apply the right resources, at the right time.
According to the National Oceanic and Atmospheric Administration (NOAA), there are six sources of data used to predict the weather. NOAA's Weather and Climate Operational Supercomputer System (WCOSS) compiles information from weather satellites, weather balloons, buoys, and surface stations from around the world. Using a computer processing system called AWIPS (Advanced Weather Information Processing System) all the data is combined into graphical detail. It is this information that meteorologists use to analyze data and predict the forecast. In the workers' compensation industry, the various sources of data include claims, medical utilization, pharmacy utilization, case management, utilization review and overall outcomes. For decades, payers have struggled to aggregate the various sources of data as it was segregated across multiple systems and various claim partners. Aggregating and analyzing all the data allows payers to accurately identify risk throughout the life of the claim and, even more critical, demonstrate total program performance.
Risk models can forecast claims that may become problematic based upon historical data, allowing “pre-emptive” or early clinical intervention. The goal being, put your best resources on it early to circumvent potential adverse outcomes. The best predictive risk models work with historical data and analytics to identify those claim attributes that typically result in less than desired outcomes. In other words, if rain is coming, grab your umbrella and stay dry. However, not all claims are that easy to predict. Many times, claims start out with no indicators of potential risk, but over time, become complex. There is much discussion these days about taking a holistic view of the individual when managing a claim. The logic makes total sense, the more we know about a person, the more targeted and timelier the intervention. Combining claim, medical utilization, and pharmacy data, and integrating that with treatment and return-to-work guidelines builds an accurate picture of the individual as the claim matures and allows us to identify risk throughout the life of the claim. Identifying conditions that may impact recovery, like co-morbid conditions, may further indicate the need for intervention. What about the nuances not contained in a diagnosis or procedure code? Use of computerized coding algorithms opens the door to large-scale analysis of narrative text to help identify the non-coded but documented granular insights about an injured employee. This is particularly helpful in identifying Social Determinants of Health (SDoH). Examples include poor support systems, access to healthy food, living conditions and health literacy.
After the storm, it is clear if the weather was forecasted with any degree of accuracy. Within our industry, it is a bit more complex, requiring a comprehensive analysis of your total program outcomes. To do so, we must answer the following questions:
- Did we target the right claims for intervention?
- Were overall claims costs reduced?
- Did claims close faster?
- Were we able to reduce the rate of litigation?
- How were return-to-work outcomes?
Disparate data from numerous resources all demonstrate claim-level outcomes with estimated savings based upon a point in time when a service was discounted, or the frequency or duration was reduced. Certainly, a “temperature” check that your overall program outcomes will be favorable. However, creating an outcome-focused claim management program requires a consolidation of all the data and turning it in to actionable information, not only to proactively identify the need for intervention, but to evaluate the fiscal impact of the program and gain new knowledge of current trends that may require attention.