Leveraging Data Science to Better Manage Individual Workers’ Compensation Claims

Graphic Case Complexity
Case Complexity

All Workers’ Compensation claims are not created equal. There are numerous factors that determine the complexity of a claim. Understanding the complexity of a claim is critical to understanding how to manage the claim. The complexity of the claim should guide the continuum of care for the injured worker, determine the amount of money to reserve, inform utilization audits, determine if a nurse case manager should be assigned, and more. 

Applying data science to these factors can enable you to more accurately calculate the complexity of the claim and determine the impact this complexity will have on the cost and duration of the claim. 

How can self-insured companies and insurers leverage clinical data and data science to calculate a personalized complexity score for an individual claim? The following will discuss how to use this complexity score benchmark as a measurable guide for clinical practice as well as utilization review and medical claims management. 

What is data science and how can it help manage Workers’ Compensation claims? 

In the simplest terms, data science refers to the extraction of actionable insights from raw data. In the world of Workers’ Compensation claims, the goal of using data science to better manage claims is to identify cases at risk of becoming outliers – those with a likelihood of a higher cost and longer Return-to-Work. Using this data, we want to prevent claims from becoming an outlier. Following a common axiom in healthcare, roughly 20 percent of Workers’ Compensation claims account for 80 percent of total costs. 

Applying regression analysis to scores of data points available for a claim, data scientists can identify the variables and combinations of variables that impact the length, cost, and duration of a claim. They can then build models that can be applied to new claims to predict the length, cost, and duration of the claim and predict the likelihood of the claim becoming an outlier. This information can help adjusters know which claims to focus on to prevent them from becoming an outlier. They can also use this information to more accurately manage reserves. 

For a specific diagnosis, the Bardavon Health Innovations team calculates the mean number of visits, length, and cost of the Physical or Occupational Therapy case. We then calculate the standard deviation and define outliers as more than one standard deviation above the mean. 

“We want to predict and continuously refine expected utilization and outcomes of active cases to streamline therapy, identify high performing physical therapists, reduce total care and indemnity costs, and speed return to full-duty for employees suffering on-duty injuries,” says Doug Dickerson, senior vice president of Software Products and Data Science for Bardavon. 

Calculating case complexity with data: What data has the greatest potential to generate actionable insights for Workers’ Compensation claims? 

The data needed to determine the complexity of a case goes well beyond the medical diagnosis. Data is needed about the injured worker’s history, demographics, overall health, occupation, and more. It is important to put a face on the injured worker through the use of clinical data. 

“Our diagnostic index specifies the expected number of PT/OT visits and duration (treatment time period) for a workers comp injury. The expected values are derived by ICD10 code, body part, and whether the case was surgical or not,” Dickerson says. 

Dickerson and the Bardavon team calculate a complexity score for every case, based on the additional data they collect about the case. 

“Our own clinical data, collected by Bardavon Provider Partners using bNOTES®, incorporates an individual’s history, injury, and occupation to calculate a case complexity score. The complexity score benchmark is then used as a measurable guideline for clinical treatment plans, as well as optimizing utilization and care management.  As treatment progresses, behavioral factors (such as cancelling appointments) are used to update the complexity” 

The team is now working to combine the case complexity score with the diagnostic index to calculate a personalized patient index. This index will specify the expected number of PT/OT visits and duration for a specific case, accounting for the case complexity score and the diagnosis. 

Using bNOTES, Bardavon Network Clinicians collect specific data about each patient. Bardavon has identified theses data points as ones that influence the complexity of the case. Bardavon’s complexity models identify the impact of these variables individually as well as the impact of combinations of these variables. 

For example: females over age 50 with high BMI that require shoulder surgery have high likelihood of high case complexity, long case durations, and high overall case cost. 

Bardavon’s proprietary approach to case complexity unites Providers, adjusters, nurse case managers, and Payors by improving functional outcomes for the injured worker and reducing excessive or unnecessary utilization of medical services. 

How does case complexity impact the cost and duration of a claim? 

Case complexity is not simply diagnosis-based. The more complex a case, the longer it lasts and the more clinical visits, including PT/OT, will be required, which directly relates to days out of work. Additionally, the physical demands of the injured worker’s job also contributes significantly to case duration. The combination of factors can increase the complexity even more. 

For example, the Bardavon data science team looked at the combination of age and the physical demands of the job, and identified that, while both have an impact on the duration of a case, when combined, the impact can be significant. They looked at the difference in case duration across all of their employer clients and determined that: 

  • Patients older than 50 were in therapy 17% longer than those under 50 
  • Patients with very heavy physical job demands were in therapy 6% longer 
  • Patients over 50 with a very heavy physical job demand were in therapy 30% longer than those under 50 and in a less physically demanding job 

Identifying outliers like these with a high likelihood of high cost and long return-to-duty times early in the claim process can improve claim predictability while reducing overall costs and return-to-work times. 

Managing claims 

Bardavon’s automated claims assessment process runs every weekend. From the data generated and analyzed through automated regression algorithms, our Quality Team audits and analyzes every single case we manage to identify outliers based on complexity variables. Based on their analysis they may dive deeper into the therapists’ notes to identify plateaus, then contact the therapist directly to discuss ways to better manage the individual to improve outcomes and shrink the timeline to return to full-duty. High risk cases require numerous touch points per month from our team in coordination with the patient’s physical therapist in order to prevent the case from becoming an outlier. 

Based on data collected through managing physical therapy activities for Workers’ Compensation claims or transaction data that shows a Physical Therapy element in a claim, we find the following metrics stand out: 

  • percentage of Workers’ Compensation claims that involve physical therapy is close to 20% 
  • those 20 percent of claims account for 80% of overall spend on Workers’ Compensation  

This is because, other than death claims, in claims of significant duration, there will be some type of physical therapy. This is logical because musculoskeletal (MSK) injuries are, by far, the most expensive injuries related to Workers’ Compensation claims. They too frequently represent claims that are longer and more expensive than they could be. 

If you are able to examine available information about a particular claim, and from that predict what overall costs will be, including medical and indemnity costs, you can set reserves correctly. You can also start to realize these claims need tighter management. You are then enabled to act quickly on the potential outliers, and ensure the adjuster is working closely on cases at high risk of becoming outliers based on behavioral variables and overall complexity.  

For example, Bardavon has found that if a claimant cancels 20 percent of their care or physical therapy appointments, it is likely to be or become a problem case. 

In practice you can see how these data science applications works for Bardavon clients. Working closely with one of our large customers, the Bardavon team provides weekly updates to adjusters, identifying cases at risk of becoming an outlier. Within the first 4 months, the percentage of surgical cases becoming outliers was reduced from 17 percent to 12 percent. As Bardavon optimizes this complexity modeling, we strive to reduce this further. 

Let’s talk 

If you would like to know more about how our risk assessment system helps risk managers and Workers’ Compensation claims adjudicators improve outcomes including reduced time to return to full-duty and therapy overall costs, contact businessdevelopment@bardavon.com. 

Doug Dickerson

Doug Dickerson is Senior Vice President of Data Science and Product Management at Bardavon Health Innovations. Doug has 30+ years of technology leadership experience. He has worked in both large corporations and start-ups including IBM, Sprint, Informix, Digital Archaeology, Handmark, and OneLouder. He has held various technical, design, product management, and executive positions. He played a critical role in leading two start-ups from idea through successful acquisition and is working to do it again at Bardavon.

Share This
Doug Dickerson

Doug Dickerson

Doug Dickerson is Senior Vice President of Data Science and Product Management at Bardavon Health Innovations. Doug has 30+ years of technology leadership experience. He has worked in both large corporations and start-ups including IBM, Sprint, Informix, Digital Archaeology, Handmark, and OneLouder. He has held various technical, design, product management, and executive positions. He played a critical role in leading two start-ups from idea through successful acquisition and is working to do it again at Bardavon.