The world of workers' compensation has been complicated in recent years, due to the financial woes of the federal government. The idea is simple enough, strive to have those responsible for the cost of an injury or disease bear that cost. Before the turn of the century, there was a pattern, perceived at least, of employer/carriers settling workers' compensation cases with injured workers, followed by the medical care for that worker becoming a burden on Medicare or Medicaid. Whether those settlements included money for future medical care (and workers received a windfall when the government paid for the care) or did not include that care money because of Medicare (and the payer received a windfall by shifting cost to Medicare), or some combination of the two is perhaps debatable in any particular case.
The federal government implemented legislation and began requiring those settling a workers' compensation case to take Medicare's interests into account. This required both the injured worker (payee) and the employer/carrier (payer) to consider the cost of medical care that would be likely following settlement, and to make arrangements for that cost to be born by the settlement itself. The process has had its critics. Some believe that this process is slow and cumbersome. Some doubt that the financial impact on the federal government is meaningful.
The Centers for Medicare and Medicaid Services describe the process of a "Workers’ Compensation Medicare Set-Aside Arrangement (WCMSA)." This is "a financial agreement that allocates a portion of a workers’ compensation settlement to pay for future medical services." These funds are sequestered and are used to pay for medical care that is related to the "workers’ compensation injury, illness, or disease." These funds are, of course, based upon an estimate of likely future expenses, determined at the time of settlement. Predicting the future can be challenging under the best of circumstances.
However, the variables in each particular case are reviewed and the future care cost is predicted. That estimated amount "must be depleted before Medicare will pay for treatment related to the workers’ compensation injury." That, in itself, may be difficult because there are complications and co-morbidities involved often in work injuries. What is or is not "related" may be the subject of disagreement. The responsibility for estimating or predicting the future care amount is shared by "all parties in a workers’ compensation case." The responsibility is imposed by the Medicare Secondary Payer (MSP) laws, with a focus on protecting Medicare’s interests.
That laudable goal and shared responsibility make sense. However, the process and time involved can be significant. Each WCMSA "is determined on a case-by-case basis." That means that individual data is accumulated by people, compiled by people, analyzed by people, and then shipped to the government where it is all analyzed by other people. Some estimate that a minimum time frame for approval is 60 days, and in some cases, the process can take much longer.
I can get an offer on a plane ticket in minutes on the Internet. I can rapidly and conveniently buy virtually anything, often with multiple vendors competing for my business. It may be a simple purchase such as a book, or it could be a quote for a 30-year mortgage. Often, my Internet search for one product (plane ticket to a particular destination) leads to prompts or advertisements for another (car rental or hotel there). I have been told of people receiving prompts for re-order of purchases around the time some computer predicts their last order of a product should run out.
Programs have been developed and deployed for all manner of agreements and interactions. Google and others are watching our consumption, predicting our interests in services and products, and reacting to our decisions. These programs are analyzing digital data. The day is coming when all the workers' compensation information, diagnosis, treatment frequency, costs, etc. on a worker will be digitized anyway. Some are there now, but it is not universal. Those that have not digitized yet will do so. It is cost-effective and makes financial sense.
Why can't the Centers write a simple computer program that estimates future medical care costs? This could be constructed for user input or automated processes. It could be as simple as a website where an adjuster, attorney, or others could input the data regarding diagnosis(es), procedures that have been performed, and medications prescribed, along with their duration, strength, and prognosis. Or, for those carriers and servicing agents that have digitized records, they could write a computer program to "harvest" the relevant data from their records, align and organize it in a method consistent with the Centers, and submit the data in a further automated process.
The Centers could easily construct a database that measures past history. I suspect that a patient diagnosed with a rotator cuff tear will travel a path similar to others' with that diagnosis. I suspect that a patient prescribed any particular medication will have similarities in duration of use with others who use that medication. Averages can be determined. There are literally millions of people who are treated for various maladies, and insurance carriers are already using computer programs in this proposed manner for "predictive analytics."
Those companies are collecting a premium, investing the money, and then using the proceeds to pay claims when required. To remain in business, those companies must make valid and accurate predictions about the costs for which they will likely be responsible. And, they must make those predictions about populations of people before they even become ill or suffer injury. Those companies are not guessing, they are using data. And, increasingly, they are becoming more sophisticated with data, using "predictive analytics."
Data exists to study the path traveled by medication prices. Whether the price of other medications will increase or decrease can be predicted on those past experiences. The inflation rates faced in medicine are predictable. The cost of medication today can be readily adjusted for inflation in the future. Whether complications will occur can be predicted. Some percentage of people taking a particular medication will need some other medication or modality as a result. That risk can be measured, valued, and predicted.
The performance of certain procedure codes could be similarly used to extrapolate the probability of future procedures. A patient undergoing a particular surgery will be at risk of requiring some additional surgery in the future. That probability is predictable, measurable, and based on the data at hand. Life expectancy is similar. Statistically, it should not be difficult to document a probable treatment path based on injury, age, degree of perceived recovery from the procedure, and perhaps a few other criteria.
Proof of all of this is also at hand. In fact, this prediction is occurring right now, at the Centers. When a Workers' Compensation Medicare Set Aside Arrangement (WCMSA) is approved, it is because a human being somewhere has made these predictions. That human being has accumulated data, categorized it, organized it, and submitted it. Some other human beings at the Centers have reviewed it, recalculated it, and reanalyzed it. And, as with all human-intensive processes, time has been invested. That is the real issue, the passage of time. In some cases, it can require months. In some instances, much of that work is invested in a "what if" process, trying to figure out what a WCMSA "would" require "if" a settlement of the workers' compensation case could be reached. It may be hard to settle a case if the payer does not know how much it will pay and the payee cannot be told how much she/he will have in pocket afterward.
The assumptions, predictions, and analytics already exist. With all of this programmed, a payer (insurance company or employer) should be able to input the raw data from adjusting files, and the computer program should be able to produce a reasonably accurate MSA amount in a matter of minutes (at most). And, barring some misrepresentation (innocent or intentional) in that input process, the government should be willing to live with the calculation that results. After all, the government would oversee the programming, the methodology, which the machine would apply to that historical data in reaching its output, its prediction.
Certainly, the issue with statistical analysis is always that there are outliers. If you are above or below average, then "your mileage may vary," and everyone understands that. The program would not produce "the" singular, absolute, "correct" prediction of future medical expense for this patient. However, it could produce a reasonably accurate prediction that is appropriate to use for this patient. Certainly, there would be a margin for error. Sometimes future medical costs would be over-predicted or under-predicted. But, it is naive to believe that is not happening today.
Currently, there are a great many humans who are toiling to make similar predictions. They are just as apt to rely on averages, predictions, and projections. They are just as apt to overstate for those who are actually below the mean and understate for those who are above. But, overall, statistically, the result would be similar using the current process or the programming I propose. However, the programming I propose could be simplified to allow any payer to input the required data and immediately get a stated value. That value of "future medical" could then be understood and applied in the negotiation of settlement.
And, the government should not necessarily shoulder that expense burden. Instead of a payer (employer/insurance carrier) paying a vendor for this predictive expertise (today), let the government charge the payer a fee to prepare the estimate using this computer program that is replete with that expertise. The same database could persistently monitor its own performance. It could consider whether its estimate for a particular person was accurate; if it predicted use of a medication would continue for two years, then in two years the program could note being correct or not. It could then adjust its predictions in future instances of that medication. The computer program could learn through actual outcomes to better predict future outcomes.
In this method, the government recoups the cost of programming and development through a user fee. The government can downsize a significant workforce that is engaged today in human review and approval of these human predictions. The payer can pay a small processing fee to the government instead of paying the commercial processor to have humans accumulate and package the data for human consideration.
In the end, It is probable that an automated process would cost the payer less and expedite the settlement to the benefit of both the worker and payer. The predictions would be quicker, more consistent, and less costly. To further simplify the process, alleviate the management of that set-aside money. Today, the "set-aside" from the settlement sits in a bank and the injured worker pays for care from that until it runs out. Then Medicare begins to pay for care.
To simplify the process, simply require the payment of the "set aside" directly and immediately to the government. If the program predicts future care will be $50,000, then upon settlement the worker gets her/his money and the $50,000 is paid to the government just like child support arrearage is paid today. Medicare could begin to cover care immediately because the $50,000 is in the coffers of the Medicare trust fund (you remember, the Al Gore "lock box") immediately. That way, the entity taking the risk (if the expenses exceed estimated, Medicare will pay) is also entitled to the benefit that results if the expenses are actually less than predicted. In the modern world of American socialized medicine, this is a logical consequence.
The further benefit of this "immediate pay" process is that the injured worker will immediately receive medical care from Medicare. And, those bills will be paid by Medicare based on diagnosis codes (ICD-10) and treatments (CPT) which would be documented. The program that estimated the likely future cost could immediately begin to analyze the actual future cost. That would further "educate" the database as to the "actual" expense compared to the predicted expense in any particular case. Predictions versus actual performance on cases today could be used to adjust assumptions and predictions on future cases. The program could "learn" as it goes.
Perhaps most importantly, the process would be simpler and faster. It would be as accurate and effective as what is being done today by humans, and probably far more so. The benefits of over-prediction would be enjoyed by the same entity as currently faces the risk of under-prediction. Injured workers would not have to administer those set-asides. And, overall, the cost to individuals and the system would decrease.
And, perhaps those are all the best reasons to leave things just the way they are. Perhaps such a program would simply be too logical, efficient, and fast? Or, perhaps someone out there will read this and write to explain how I have completely missed the point?
