A 2008 report by the Insurance Research Council (IRC) provides empirical evidence that since 2000, a significant decrease in auto claim frequency has been mysteriously offset by an equally significant rise in severity, leaving loss costs essentially unchanged. Of particular note, the Insurance Research Council report points to a study by the National Safety Institute showing that improved automobile safety engineering has been lowering the frequency of serious injuries and deaths. This makes the rise in bodily injury and personal injury protection claim severity particularly vexing.

The graph below is taken from the Insurance Research Council report and it depicts the described trends:

This paper is the result of research that was aimed at discerning the root causes of the three simultaneous trends and whether their relationships are coincidental or causal.

In summary, the findings were as follows:

1. The decline in claim frequency was (is) primarily causally related to lower per-capita driving, which in turn is causally related to the combination of a significant rise in gasoline prices coupled with rising unemployment over the same period.
2. The rise in property damage severity is largely causally related to an increase in vehicle repair and replacement costs that is in turn driven by the efforts to make vehicles safer, such as airbags.
The rise in bodily injury and personal injury protection severity is the result of a significant shift of the total cost of national health cost burden from the government and private health insurers to the Property & Casualty insurance industry.

The Root Cause Of Medical Cost Shifting
The decline in claim frequency and the rise in property damage severity were considered transparent enough not to warrant further elaboration in this article. This turned our focus to unraveling the seemingly mysterious rise in bodily injury and personal injury protection severity, especially in light of the decline in the rate of severe injuries that can be causally related to the property damage severity rise.

Our findings in this regard are a confluence of causally related phenomena as follows:

• The government has been tightening payment controls and cutting reimbursement rates for medical providers under programs like Medicare and Medicaid.
• Private health coverage payments to providers have declined as fewer employers provide coverage; for those who do, the deductibles and co-payments have grown considerably.
• Hospitals and other significant medical entities have responded by appointing financially savvy administrators and implementing electronic medical record systems that have morphed from being efficiently focused to being revenue enhancement driven.
• These savvy administrators and their new software programs are defeating the government and private software tools designed to vet electronic billing submissions. They accomplish this via "diagnostic upcoding," a means of reporting an injury or illness as being more severe than in actuality.
• Though these tactics are aimed primarily at the government and private health plans, they have worked equally well at overcoming Property & Casualty company claim deterrents, driving up the cost of what were once considered small soft-tissue personal injury protection and bodily injury claims.
• While the tactics are effective against the government and private health insurers, those entities have mitigated their effects by lowering medical procedure reimbursement rates. Because the Property & Casualty industry lacks the means to do the same, it is absorbing an ever increasing share of total national health care costs, a phenomenon referred to as "cost shifting."

Understanding Diagnostic Upcoding
The rise in bodily injury and personal injury protection severity is the manifestation of the growing cost shift from hospitals and other medical providers that are passing a progressively greater percentage of the total cost of national medical costs to the Property & Casualty industry.

In many ways, this shift is an unintended consequence of a financial stalemate between the government and private health insurers on one side and hospitals and other medical providers on the other. Hospitals and other providers turned increasingly to upcoding to offset the steadily decreasing reimbursements from the government and private health. But the Property & Casualty industry, which accounts for only about 10% of hospital utilization, was subjected to those same programmed upcoding schemes, seemingly without awareness, or at least without taking mitigating actions.

The following are direct quotes from the IRC (emphasis added):

"We use the following indicators and more: extent of disability, rate of hospitalization, days unable to perform duties. And we see that injuries are not becoming more serious nor have they changed much. But, it is still evident that something is driving the rise in severity. Since injuries do not appear more serious, medical usage and treatment costs are driving the increase in medical care expenses."

"Low reimbursements from public health insurance programs, such as Medicare and Medicaid, have prompted hospitals to shift costs to automobile insurance companies — raising auto injury claim costs. Cost shifting in 2007 resulted in $1.2 billion in excess hospital charges. The full impact of hospital cost shifting, including that occurring in other insurance coverage, is likely much greater."

Medical Coding 101
The medical profession has a well-established regimen of coding to describe both the nature of injuries and illnesses as well as the therapies utilized to treat them. The coding classification that defines the nature of pathology, the diagnosis, is referred to as "ICD-9 codes." For therapies, the codes are referred to as "CPT and HCPCS codes."

ICD-9 codes form the basis for CPT and HCPCS codes because the diagnosis precedes the selection and introduction of the appropriate therapy or therapies. So if in the same instance the ICD-9 code indicated a bruised hand but the CPT/HCPCS code reflected a coronary bypass, software programs employed by the government, private health care providers and the Property & Casualty industry can detect the mismatch and prevent the payment.

The government and private health insurers tie their reimbursement rates to the CPT/HCPCS codes. When the therapy code is correct for a particular diagnostic code, the software can discern the appropriate payment amount based on a programmed fee schedule. With such a system, it is easy for the government to institute an across the board fee reduction of 10% or to apply the reduction in a variable way and accurately estimate its aggregate impact.

The Property & Casualty insurance industry is able to follow government fee schedules in some instances (mainly Workers' Compensation) but in most states relies on "usual and customary" charges for auto, the aggregate average of what medical providers bill for each CPT/HCPCS code. There has been significant controversy and litigation related to the sources of the "usual and customary" data.

Overall the government, private health care and the Property & Casualty industry all use software programs to vet medical billing that start by insuring that the CPT/HCPCS coding for each therapy match the ICD-9 diagnostic code and then, checking the amount charged for each therapy against a fee schedule or, for the Property & Casualty industry, the "usual and customary" charge.

From "Patient Centered" to "Profit Center"
The belief that doctors would not falsely inflate a diagnosis, outside of outright fraud, was common at the outset of the electronic vetting system, and with good reason. But the confluence of government fee reductions and the decline of private health care created significant financial duress for hospitals and providers. Something had to give, and turning away the uninsured was not an option, although more and more providers are opting out of Medicare and Medicaid.

That answer came in the form of savvy hospital administrators and slick new software that added efficiency but even more so, drove up revenues. It did so by making certain that the highest legitimate diagnostic code was being selected. It also provided upfront audits to spot instances such as a therapy that overshot the initial diagnosis so that "coding corrections" could be made before the billing process was invoked. Increasingly, such corrective activity was administered by support personnel who lacked medical knowledge and sought only to satisfy the systems requirements.

The Impact Of Diagnostic Upcoding
The medical coding system is logical and it lends itself well to automation, which vastly increases efficiency for everyone. But its cornerstone is the ICD-9 diagnostic code and while expert systems can aid a doctor in making a diagnosis, the ultimate decision (at least for now) still rests with the doctor's judgment.

With enough information, a system could flag potential upcoding in certain instances. Gradually such a system will evolve and become increasingly effective, but for now, expert human intervention is required to decide whether the reported ICD-9 diagnostic code is reasonable in light of numerous data points derived from a forensic examination of the mechanism of injury.

This chart reflects a huge shift in ICD-9 supported CPT coding from 2002 to 2008:

Most Doctors Are Victims Not Perpetrators
It should be noted that hospitals and other providers were likely losing legitimate revenues by not paying sufficient attention to billing practices in the past. Therefore, some of the change in the above chart is likely justified. But like a pendulum that may have swung too far in one direction in the past, there is ample evidence of over-compensation on its way back.

A Few Recent Headlines

4-18-2011: CtW Investment Group Calls for Board Accountability

7-21-2011: Prime Healthcare Accused of Medicare Fraud

8-31-2011: Alleged Fraud Uncovered During EMR Training

9-12-2011: Dallas Hospitals To Pay $1.4M, Settle Upcoding Investigation

9-21-2011: Doctors Protest CEO at Knapp Medical Center

This is not meant to suggest that all hospitals and providers, or even the majority, cross the line, but a systemic change has occurred and one manifestation is the alarmingly frequent allegations of billing related maleficence.

Examples Of Recent Articles From Hospital Industry Publications:

The Property & Casualty Industry Response
Had cost shifting not occurred in parallel with the significant decline in accident frequency, loss costs would have dropped and companies would have utilized the capital windfall to lower their prices and gain market share. That would have led to a significant decline in premiums accompanied by expense reductions significant enough to maintain underwriting and claim expense ratios.

Many of the financial incentives in the industry are determined as a percentage of premiums. Shrinking premiums produce shrinking companies, lower cash flow, weakening balance sheets and endangered financial ratings. None of those are desirable outcomes for carriers. Independent Agents and Brokers work on commission, a percentage of written premiums, so if they earn 15% on an $800 auto policy that without cost shifting may have shrunk to a $600 policy, they would have taken a 25% revenue cut.

There may be Nash equilibrium in play, with everyone being happy under the current circumstances. But as the Insurance Research Council points out, if accident frequency returned, first loss costs, and then premiums would spike. The longer and larger the cost shift is allowed to grow, the greater the potential market disruption becomes when the day of reckoning arrives. But for anyone who is not betting on a near term economic revival, and/or the significantly lowered gasoline prices that could fuel increased frequency, this may not be a burning issue.

Is Anybody Hurt By Cost Shift?
Auto insurance consumers are financing the cost shift through artificially high insurance premiums. To the extent that some of those same individuals consume more medical services than are paid for, there seems to be fairness. But for those who shoulder the full burden (or more) of their medical expenditures, "the affluent," this resembles income redistribution.

What Does This Portend For Claims?
As things stand, most claim department leaders either don't realize what has happened, or believe that their particular policies and practices have prevented this for their company. A benchmark comparison of company loss costs against like competitors would force one to choose to either disbelieve the phenomenon entirely or accept that they are as vulnerable as everyone else.

The standard industry claim deterrents for medical cost containment are two software programs, medical bill repricing and automated bodily injury evaluation. Both are highly susceptible to diagnostic upcoding, but few claim leaders understand the algorithms that drive these systems and imbue them with capabilities that they don't have. Those claim people far enough into the details to know better are not empowered to act.Efforts to inform claim leadership typically meet with stiff resistance, because they genuinely believe that they "have it covered." And even if they knew they did not, why would they want to have such significant leakage called out when loss ratios are stable, and nobody is pointing a finger in their direction? But acknowledging and addressing the issue would create more work on top of what already feels insurmountable.

Closing Thoughts
This article was written for the purpose of calling out the significance of cost shifting to the industry. I do not believe that the industry is purposely allowing the cost shift, but when a bad situation creates comfort, a sense of urgency is elusive. The more entrenched the problem becomes, the more difficult the exit strategy.

The Property & Casualty industry will continue to represent a relatively small percent of hospital utilization, but it is continuing to pay a growing disproportionate share of the total cost. Even if driving never perks up again, as the cost shift grows a tipping point nears where premiums become too high.

Finally, there has been too much focus on automating claims over the past twenty years or more and far too little focus on basic claim handling. This is one example of numerous significant pockets of leakage that arose as a result.

Electronic records and fund transfers between the industries have not even reached a nascent state and beyond that huge opportunity is the even bigger prospect of Property and Casualty aligning with government and healthcare payers in a common effort to contain medical costs.

Left Standing At The Station
The government mandate of electronic health records (EHR) leading to electronic data interchange (EDI) for Medicare and Medicaid payments, combined with the increasing adoption of both by private health insurers, is driving the healthcare industry toward the efficiencies of increased connectivity. But the Property and Casualty industry has not been part of that effort, probably because it does not think of itself as a healthcare payer.

The Property and Casualty industry makes multiple manual payments for each claim while government, and increasingly private health insurers, pays once on the basis of the initial diagnosis. Traditional payers operate under the theory that making a fair diagnosis based upfront payment encourages providers to be as efficient as possible. The Property and Casualty approach requires that providers earn their fee treatment by treatment, which can encourage overutilization.

The contrast between traditional payers and the Property and Casualty industry is stark; the traditionals incur one electronic data interchange transaction while the Property and Casualty industry processes multiple bills per claim manually and issues multiple checks per claim. The traditional payer approach saves significant processing costs for both the providers and payers.

Additional interconnectivity opportunities include benefits coordination, the avoidance of double-dipping, and subrogation.

An Ounce Of Prevention
With most healthcare payers making diagnosis based upfront payments, providers place greater emphasis on diagnostic code selection, and this has led to the phenomenon of Diagnostic "Upcoding," the intentional substitution of a more severe diagnostic code than is justified to obtain a higher fee.

In response to diagnostic upcoding problem traditional healthcare payers are making increasing use of predictive modeling tools. Such tools analyze and compare the historic coding patterns of providers to discern those with chronically above average severity coding. In contrast, the Property and Casualty industry continues to rely on tools that offer no defense against diagnostic upcoding.

As traditional payers tighten their controls, billions in additional medical costs are being shifted to the Property Casualty industry, as has been reported by the Insurance Research Council.

A Long Way To Go
It would be disingenuous to suggest that traditional healthcare payers have achieved anything close to perfection related to preventing hard or soft fraud, but they are far ahead of the Property and Casualty industry.

By combining forces with traditional payers not only could the Property and Casualty industry vastly reduce transaction costs and begin to deter excessive treatment, but it could also contribute to the overall payer effort, speeding the evolution of increasingly effective solutions.

But instead, Property and Casualty efforts underway in this space are largely fragmented, not only by carrier, but often by line of business within carriers. These multiple divergent efforts all aim at covering the same ground that traditional payers have already navigated.

If You Can't Beat Them, Join Them
Moving to a universal healthcare platform would open up additional doors, including expanded Preferred Provider Organization (PPO) networks, medical management expertise, and shared efforts at identifying and shutting down professional fraudsters. The larger the sample, the more statistical significance, and the easier it becomes to develop reliable and actionable data.

What should be most obvious is that managing medical costs is the core competency of traditional healthcare payers, not the Property and Casualty industry. In the competition to contain medical costs, traditional payers are winning to the point where billions in costs are annually shifted to the Property and Casualty industry.

The Property and Casualty industry should begin moving in a direction that might even lead them to outsource their medical payment process to advanced healthcare payers. Incenting traditional payers to effectively lower Property and Casualty medical costs would produce a substantial win/win. The biggest winners of all might be Property and Casualty customers who have unwittingly been picking up the tab.

Appraisal Genre Versus Individual Resource Driven
The term "genre" is used to describe the traditional high-level appraisal resource categories of Direct Repair Program, Desk Review, Staff, and Independent Appraiser. Within the industry the typical order of priority for each genre is:

1. Direct Repair Program
2. Desk Estimate (if within parameters)
3. Staff Appraiser
4. Independent Appraiser

The implicit assumption is that the mean performance within each genre is highest for Direct Repair Programs and lowest for Independent Appraisers, making independent appraisers the choice of last resort. Even if these assumptions were accurate, basic statistics suggest that more than one third of the assignments from such a system will be sub-optimal.

The graph below represents a statistical normal distribution which exists for most populations of similar phenomena. As an example, the height of all human beings would form a similar pattern as would the batting averages of all major league players for the last 100 years. For most populations, the majority of individuals fall around the mean (average) range and then tail off gradually in both directions.

Assuming a statistically significant sample size of Direct Repair Programs, desk estimators, staff, or independent appraisers, and reliable performance measures, each would produce a similar bell curve distribution. Companies that do a better job of managing their appraisal resources should have more compact distributions, but they would still form a bell curve, albeit taller and narrower.

The main point is that, while on average the current hierarchy of appraisal resource utilization might be accurate, using available information it is possible to do much better than that. As a specific example, a highly performing independent appraisal firm could be a better option than a low performing Direct Repair Program.

Taking this to its logical conclusion, the old paradigm about an ordered genre based assignment priority should be eliminated in favor of a best performing resource approach. Over a large population of assignments a best performing resource approach produces a much better financial outcome than does the genre based approach.

Beating a Dead Horse
The industry currently operates on a genre based approach to valuing appraisal resources, meaning that it lumps together all staff, all Direct Repair Programs, all Independent Appraisers, etc. as if each resource within each genre performed at a consistent level. Let us be clear that we are speaking about the variables of service, severity and expense and combining them into a single weighted composite performance score. When data was scarce, a genre based priority was the "Best Practice" approach, meaning that it represented the most optimized appraisal distribution possible given that state of technology. But with the advent of the information age, trying to force more efficiency out of the old way is not as effective as evolving a tool that leverages the new possibilities.

Building a Better Mousetrap
As a baseline, the assumption is that carriers already have the data they need to construct composite performance ratings for all of their appraisal resources. The standard measures around service, severity and expense typically form the basis for composite performance ratings, but when it comes to questions like how much additional expense or severity am I willing to incur to avoid service degradation, each company has its own tolerance ranges and the construction of individual resource composite performance ratings should reflect those preferences. Individual company preferences are essentially their "secret sauce" that allows them to tune the system to facilitate their competitive strategy. In an extreme example, a non-standard carrier may think differently about the weight they would attach to service versus severity than a standard carrier. Such distinctions are already reflected in the carrier appraisal protocols that vary from company to company. Though the secret sauce may vary, the overall process structure for composite performance rating could be static.

Composite Ratings
It makes good sense to think about this from the perspective of software requirements because the end game is to design a program that looks at the assignment criteria and matches it to the best available appraisal resource, regardless of genre. This simply means that we want to express the result of qualitative analysis in machine readable code.

A fairly straightforward approach would be to utilize a 1, 2 or 3 rating for each of the three critical performance components of service, severity and expense and then combine the three into a composite rating between 3 and 9. An appraisal resource receiving the best score of 3 for each of the three performance components would earn a composite performance rating of 9, while a resource receiving the lowest rating of 1 for each would be rated a 3. By capturing the assignment criteria in a system, which happens routinely today, the program could identify potential appraisal resources and choose the highest rated resource.

This leaves the question of how to generate the ratings? The best answer would be to allow the program to analyze recent historical data for service and quality and add a static expense table. As part of their "secret sauce" each carrier could decide what result levels would trigger their ratings. As an example: cycle time of 7 days or more earns a 1 rating, from over 5 days to under 7, a 2 rating and 5 days or less a 3 rating. The approach to severity would be similar, but with an added level of complexity because the various quality measures would first need to be composited. For expenses, a static rating table could be incorporated where, as an example, a Direct Repair Program would likely get a 3, Desk Review a 2 and Staff or Independent Appraiser a 1.

Clearly a carrier could opt to expand the rating scale and once a carrier established its parameters, they could be programmed allowing the resulting system to run the same algorithms for each new assignment. The result would be that the best possible resource would be matched to each assignment. If a particular resource began to falter its assignments would automatically be curtailed.

Bootstrapping Success
The system contemplated would produce valuable measures of its own success and the key performance indicator would be a trend-line of the average composite resource rating per assignment. For the scale example suggested above, the highest possible average composite resource rating would be 9 and the lowest 3. If, in the initial stages of the system deployment, the baseline result was 6 and as the system ran it increased to 7, 7.5, 8, etc., that would be empirical evidence of improvement that could be financially quantified. The most important job of appraisal managers would be to identify and add as many new quality appraisal resources as possible. The system would provide detailed enough data to indicate exactly what genre of such resources were needed where. The Holy Grail is drawing causal correlations to improvements in customer service surveys, reductions in loss costs and reductions in expenses.

Desk Estimating As The Corner Stone?
Used properly, desk estimating can provide a significant advantage by quickly and cost effectively resolving routine damages making more capacity for appraisal resources better suited to more complex damages. Though dollar amounts tend to drive desk adjustment decisions, in a perfect world the nature of the damages would also be considered. The proliferation of smart phones has opened the door to pre-assignment vehicle damage photos, a potential boon to the triage process, especially related to desk estimating. The very term "desk estimate" is heading for obsolescence as software programs increasingly interact to complete more assignments without human intervention squeezing cycle times into seconds, ensuring high levels of accuracy, and further driving down expense. As long as desk estimates (or eventually system estimates) are properly targeted during assignment they will produce consistently high average composite performance scores. Finally, allowing customers to take their settlement checks and shop for the best repair option brings the benefits of the free market to bear, a better alternative to directing them to average or lower performing Direct Repair Program shops.

Size Matters
The system described becomes exponentially profitable as it increases in scale. The one pitfall compared to the legacy genre approach is that within each genre there existed a sufficient population of data to provide statistical significance. Drilling down to an individual appraisal resource level will reveal that about 20% of appraisal resources lack sufficient data samples for valid ratings, typically such resources are in low concentration rural areas. Given the programmed approach described above, desk estimating would be the likely system selection whenever the damages are of an appropriate nature and that seems logical. Outside of that, the program could default to the legacy genre approach. But there is an even better answer to the scale issue.

Coopetition
According to Wikipedia:

Coopetition occurs when companies work together for parts of their business where they do not believe they have competitive advantage and where they believe they can share common costs. For instance, the arrangement between PSA Peugeot Citroën and Toyota to share components for a new city car — simultaneously sold as the Peugeot 107, the Toyota Aygo, and the Citroën C1 — qualifies as coopetition. In this case, companies save money on shared costs while remaining fiercely competitive in other areas.

There is no question that customer service and loss cost management impact competitiveness, but price and brand largely drive the insurance buying public's decisions. Given the fact that scale can increase the financial and customer service benefits of an appraisal program, at the very least, small to medium size carriers should take note. By entering into a "coopetition" style approach to automobile appraisal that invokes the methodologies defined in this article, such carriers could greatly improve their competitive positions vis-à-vis the larger industry carriers, which should be their primary focus.

A collective appraisal program would allow the partner carriers to judge the performance of each appraisal resource based on the collective results produced for all partner companies. As an example, if 50 companies participated in a collective program and a particular appraisal resource had been handling an average of one assignment per month for each carrier, performance for that resource could be judged on the basis of the collective population of 50 appraisals and that resource would understand that all of the assignments would be at risk going forward if performance measures were not maintained.

More importantly, where sufficient collective claim concentrations emerged drive-in claim facilities and Direct Repair Programs leveraged to their full extent driving down costs and speeding up service. Similarly, higher assignment volumes to desk estimating and independent appraisal firms create more loyalty and better rate structures.

In the claim business it is not untypical for claim leaders from each company to believe that their particular practices excel beyond those of all of their competitors, but going back to the normal distribution from statistics, it's just not reality. Some companies in the population are at the high-end but even for them, the structural limitations presented by low concentration imposes hard physical limitations on what they can accomplish, and even if they were to ostensibly lose some of the potency of their employed acumen, with sufficient scale, it would be compensated for many times over.

The optimum approach combines significant scale, performance based assignments, and Best Practice APD oversight, which can only be attained through coopetition.

If you're not one of the top 15 automobile writers in the US, you should really start to concentrate.

What Has Held Claims Back?
In every domain where statistical rigor has been brought to bear, old paradigms steeped in years of certainty have crumbled overnight. Claims is one of the few remaining significant institutions that have yet to undergo this transformation. In fact, it is the perpetuation of the myth that claims is different and that introducing statistical rigor is dangerous, that has kept claims from experiencing the benefits of significant investments in technology.

There certainly are dangers inherent in claim handling, especially bad faith actions because they can create painful financial and reputational damage. Even general litigation can be extremely expensive and outcomes are never certain. But the job of the astute claim leader is to find the sweet spot where legitimate claims are quickly honored and inflated claims identified and resisted. Litigation does not happen out of thin air — a claim adjuster and then several others in the chain of command have made conscious decisions not to accede to what is perceived to be an excessive settlement demand. The litigation process does not occur swiftly so there are many opportunities to settle cases in advance of an actual trial, and again, any decision not to should be well thought out. Plaintiff attorneys typically cite bad faith as a way of ratcheting up the pressure, but astute claim professionals understand what does and does not constitute bad faith.

This current upcoding problem is an example of where many claim leaders will cite bad faith potential as a reason not to offer resistance. But when a qualified physician undertakes an unbiased evaluation of a peer and arrives at a different conclusion that is strongly supported by empirical evidence, it does not constitute bad faith. If a claim adjuster introduces such an evaluation in the course of settlement negotiations, allowing the claimant to provide evidence to rebut it, and giving it due consideration, that also does not constitute bad faith. The objective should be to get as close as to the truth as possible.

This really boils down to the question of whether an adjuster armed with empirical evidence of upcoding will generally achieve a more accurate settlement outcome than without one and whether the cost benefit of that evidence is justified by its impact on the settlement. Anyone who really thinks about this, and certainly anyone who tests and measures it, will find that the answer to both questions is, absolutely. But for those who rely on traditional high-level claim measures and see in them what they want, there is no impetus to engage in continuous improvement.

The problem is really one of perspective because claim leadership has far greater exposure to outlier cases than to those that flow seamlessly through the system. Having been immersed in outliers for many years and lacking measures that would reveal them as outliers, decisions are made on the basis of the 5% population of outliers rather than the 95% population of more typical claims. Claim leaders experience the world paradoxically as if outliers somehow represented the majority and hyper-caution is a logical consequence, but at the cost of operational stagnation and significant leakage.

Conclusion
CEOs and most other industry executives cannot hide from poor results very easily, but for years claims has been a black box only understood by claim people. Most CEOs have solid fundamental understanding of underwriting, finance and marketing, but few emerge from claim backgrounds. When encountering persons with significant knowledge in fields less familiar to us, we tend to see them as experts.

But raw knowledge of any field absent empirical measures is pseudo-expertise. That does not mean the experience and expertise is not valuable, it just means that it has not undergone the transformative effect of the information age, hence it lacks empirical coherence. Significant decisions remain largely the domain of gut instincts or acquired preferences, reinforcing and perpetuating instead of testing long held assumptions about reality.

If one were to ask who the top ten carriers are from a loss or expense ratio standpoint, or with regard to premium growth or premium to surplus ratio, that information is easy to obtain. But if asked who the top ten claim departments are from a loss cost management standpoint, nobody knows. This is another example of where claims is increasingly becoming the exception to the rule as inter-industry and even intra-industry comparisons between similar functions have become increasingly relevant as the information age pulls back the curtain to reveal the key performance indicators that matter most and specify common measurement methodologies.

When standardized claim key performance indicators do ultimately arise, you can expect to see the emergence of a standard distribution of claim department performance with most lumped into the middle, a few on the high end, a few on the low end and the rest scattered in between the middle and extremes on both ends. How do you or any other CEO know what position your claim department occupies? If 100 CEOs were asked, 90 would likely answer "above average" or better, but that's just not possible.

The types of measures discussed in this series are universal and could form the basis of an industry-wide system of potential benchmarks from which company claim department results could be compared. In the meantime, as CEO, you own the responsibility for bringing your claim department into the information age. You may not be a claim expert but you surely have sufficient measurement acumen to demand empirical measurements that can be universally understood in support of abstract reasoning.

This issue of medical cost shift should become the rallying cry for bringing claims into the information age. This is one of those things that will happen on its own in time, even if no one were to take up the cause. But some company(ies) will take up the cause and gain a huge competitive lift.

Ten percent of total claim costs are a conservative estimate of leakage for the "average" claim department. Without certainty about where your claim department ranks among its peers, average is a good bet but the fact that you lack the means to be certain is the true cause for concern.

Looking Beyond Standard Claim Measures
The medical industry uses ICD-9, "diagnostic codes" to represent the nature and severity of injuries within their electronic medical records systems. The parameters for each code are spelled out clearly to promote consistency and they form the baseline for treatment plans. Unique circumstances, such as preexisting conditions do influence medical treatment but, theoretically, if all else were equal among patients, the same diagnostic code should lead to similar treatment regimens and outcomes. This is another example of how seemingly immeasurable phenomena have given way to measurement.

Hospital administrators seeking to optimize treatment facilities and Medicare, Medicaid and private health insurers, seeking to vet electronic billing, all rely on diagnostic codes as their baseline. Things are not so dissimilar in the Property & Casualty industry where medical bill repricing and bodily injury evaluation software also rely on diagnostic codes as a baseline. The designers of all of these systems employ diagnostic codes as common machine language to efficiently convey the nature and severity of injuries on an interindustry basis and as the foundational data element for financial computations.

Property & Casualty claim departments all rely on medical bill re-pricing software to programmatically compare the accumulating CPT (treatment codes) from medical bill data to the claimant's diagnostic code(s) to discern whether the treatment type is reasonable for the diagnosis. A second set of algorithms compares accepted treatment codes against a table of reasonable and customary charges to vet pricing. Nearly every claim department can easily access the data from the medical bill repricing software even when the service is provided by an external carrier.

Seeing the Forest
As a data set, diagnostic codes are subject to the law of large numbers so when aggregated and profiled with basic statistical tools, they reveal valuable insights. A simple example would be a chart of the distribution of diagnostic codes that points to the frequency with which each code appears in the overall population. Comparing such a distribution over several consecutive years reveals whether any significant changes to the distribution are occurring. With the nature of injuries remaining static or, arguably, diminishing, the more significant the evolving migration toward higher severity diagnostic codes within each yearly distribution, the more certain it is that diagnostic coding is the means by which claim deterrents are being defeated.

For simplicity, the above example condenses numerous diagnostic codes into five classification ranges based on diagnosis severity from low (A) to high (E). While the data is fictitious, the author believes that actual data would reflect a change of the magnitude depicted, resolving any uncertainty about whether diagnostic upcoding was a causal contributor to the cost shift.

This is a very simple and inexpensive data analysis exercise and even if it demonstrated that diagnostic upcoding is not the driver of the cost shift, ruling it out would advance understanding.

How Much Of The Cost Shift Is Diagnostic Code Driven?
Medical bill re-pricing software captures the accumulated amounts billed and allowed for each claim in addition to diagnostic and treatment codes. That makes it possible to discern the average amount of total billing for each diagnostic code (or condensed category as in the above exhibit) as well as the aggregate total allowance for that distribution.

By reconfiguring the 2010 data into the 2005 distribution, a measurement of the aggregate total allowance for the remodel distribution is discernible. That amount reflects how much less the medical audit process would have allowed in 2010 had the distribution of diagnostic codes from 2005 remained static. The most important metric from this analysis is the percentage increase in allowable billing between the two distributions because it is the basis for a final calculation.

Bodily injury evaluations and settlements have a significant non-economic (pain and suffering) component that is not captured by medical bill repricing software but is directly related to the severity of the injury, hence driven by the diagnosis. In order to measure the full extent of the increase in bodily injury loss costs arising from diagnostic upcoding, it would be necessary to discern the total paid bodily injury losses for the current year (for this example 2010) and multiply them by the estimated percentage increase in medical billing.

From Causal Correlation To Correction
Profiling data in the ways described not only produces insight into what happened but it also guides corrective action. By ordering the diagnostic codes by their ratio of period to period growth it is easy to identify those with a combination of high relative growth and high rates of frequency. Such codes likely contain the majority of upcoding events; hence become the most effective point for preventative action. A focused granular level review of new cases where the suspect codes arise would be the most productive intervention. Targeted physician diagnostic peer reviews for subjective injuries and radiological film reviews for objective injuries would easily detect upcoding, preventing it from expanding medical bill re-pricing allowances and bodily injury settlements.

Measuring the success of such an intervention is also easy and critical to honing the effort. Tracking the number of cases identified, reviewed, and corrected as well as the ratio of identified cases reviewed and reviewed cases corrected, provides valuable process measures. By adding financial measures such as the administrative and physician costs associated with the effort and comparing them to the recorded savings resulting from code corrections, both the rate of return and net financial impact of the project can be reported.

It is of vital importance that physician reviews be viewed as another tool in an adjuster's belt and that it be understood that the goal is to utilize the information to achieve more accurate settlements, not to create increases in litigation. This is all about improving negotiation skills by advancing medical knowledge to bring about the right settlement. Careful measurements can provide an early warning system of the potential for increased litigation. Two of the most effective measures are the average age of pending claims and pending claim counts. A trend reflecting an increase in the aggregate average age of open bodily injury claims, outside of normal variation, leads to growing pending claim counts and is a warning sign of a potential for increased litigation long before it actually arises. Nothing in this process prevents an adjuster from negotiating a settlement but with the right measures, it is absolutely possible to significantly reduce the artificial costs of medical inflation without triggering litigation.

This series only begins to tell the story of what is possible if claim departments join the information age. Other examples of the data referenced in this series include the ability to profile individual provider diagnostic code distributions, to cull out and focus on those with the most skewed distributions. Knowing which providers are playing it straight could curtail unnecessary effort. Both actions would push up the ROI of the effort. Ongoing software analysis of the data could provide regular updates on provider trends keeping the effort focused as needed. Persistent patterns of significant upcoding by the most problematic provider's should be referred to the Special Investigations Unit. Some offenders might rehabilitate for your company if they know you are on to them.

The Insurance Research Council reports a huge unexplained increase in medical costs. Claim leaders are all confident that it is not happening in their shops, yet it is happening to the industry to the tune of billions.

The objective of this series is to:

1. Continue to create awareness of the significant change in the complexion of loss costs resulting from artificial medical cost inflation
2. Educate Property & Casualty executive leadership about how to discern whether and to what extent this issue is affecting their company
3. Demonstrate how measurement and data analysis should be used by claim departments to vet this issue and many others like it

A Summary Of The Problem
As the Insurance Research Council reported in 2008, in spite of a significant decline in automobile accident frequency and incidences of serious injury, loss costs have not only held steady, but have increased. The Insurance Research Council has discerned that medical costs associated with minor to moderate injuries have undergone substantial inflation and pointed to "hospital cost shifting" as the source. That phrase doesn't necessarily mean that hospitals are intentionally making up for government fee reductions and declining private health enrollment by deliberately charging more for auto accidents. But the effect is the same because as hospitals struggle with the major players (government and private insurers) the Property & Casualty industry is incurring collateral damage as it stands on the sidelines.

This cost shift is occurring virtually undetected by Property & Casualty claim departments by evading recognition by the principle software programs designed to vet medical bills, medical bill repricing software. There is very good reason to believe that this is occurring via a methodology known as "diagnostic upcoding" and for the purposes of this series, I will assume this to be the case.

A Lopsided Competition
Hospital administrators have become financially sophisticated and technologically savvy while claim departments have traded technical acumen for cost efficiency. The widest performance gap between the new generation hospital administrators and the current claim leaders exists in the data analytics space, where the hospital administrators are light years ahead. Hospital administrators have scored a coup as billions in artificial costs are pouring into the Property & Casualty industry without claim leadership being aware.

This series will identify available untapped claim data and explain how it can be cultivated and leveraged to achieve a level of strategic deterrence sufficient to detect, arrest, and reverse the upcoding problem cost. It will also demonstrate the importance of bringing far more statistical rigor imbued with requisite technical acumen to a claim profession in decline, in spite (or because) of, a multi-year technology spending binge.

Measuring The Immeasurable
In a recent book, How To Measure Anything, author Douglas Hubbard makes the critical point that the inability to measure a variable with exactness does not make it immeasurable or unworthy of measurement. Foregoing measurement when faced with subjectivity ensures sub-optimal decisions because an informed inexact measure is better than no measure at all, and uncertainty in any variable is reducible by measurement.

Decisions vary in magnitude and complexity driven by multiple interacting variables. Each decision rides on multiple variables with each variable holding higher or lower levels of significance. The greater the significance of a variable, the more valuable a reduction in its uncertainty becomes. Variables of minor significance don't warrant as much, if any, investment in measurement. The combination of the magnitude of a decision, and the significance of a given variable pertaining to that decision, determine the "value of information". If a decision involved 10 million dollars and rested on one particular highly uncertain variable, the value of information that could reduce that uncertainty would support a considerable measurement investment.

Hubbard has formulas for calculating the value of information but in most cases, it is apparent. Typically the problem is less about the cost of making a measurement and more about the belief that subjectivity portends immeasurability. This belief is driven by a dearth of thought leadership about measuring seemingly subjective phenomena.

Success requires that leaders:

1. Assume that uncertainty can always be reduced by measurement
2. Identify the key variables in a decision
3. Target the variables where uncertainty reduction is most valuable
4. Be creative in identifying measures

This series will provide a real world working example of this process.

How Measurement Led The Insurance Research Council To The Problem
The Insurance Research Council's finding that much lower accident frequency and fewer incidents of serious injury did not result in decreasing loss costs, was itself a measurement exercise. While accident and serious injury frequency data is readily available and objective, measuring what loss costs should have been except for the cost shift involves some subjectivity. But that inexact measurement reduced uncertainty and established an order of magnitude. Once the Insurance Research Council realized that a few billion in loss costs reduction had somehow been offset, they began a more granular series of measurements that revealed increases in medical treatment well beyond inflation and without any increase in the nature of injuries being sustained. With vehicles having become far safer over the period studied, even routine injuries should have become less frequent and severe.

Is My Company Being Impacted?
If you are a company CEO you might start by asking the question; is the spike in medical treatment costs impacting my company, and if so, to what extent? A pretty clear path to discerning that answer would be to mimic the Insurance Research Council approach. If your company has experienced the same pattern of lower accident frequency and fewer incidents of serious injury without offsetting declines in loss costs, that is a strong indication.

If you find that your company does follow the pattern described by the Insurance Research Council, you need answers from your claim leader. Paid losses should not have escalated so significantly without them being aware of it and bringing it to your attention. If your company has absorbed significant loss costs from inflated medical costs and alarms did not ring in the claim department, your claim deterrents are likely inadequate.

This scenario points directly to the dearth in data analytical acumen within Property & Casualty claim departments. Many of the metrics being monitored are too high-level. For instance, your claim leader might reassure you by indicating that your company's average paid bodily injury (BI) trend is rising but within reason considering normal inflation and, the data might reflect that. But they have not factored the significant decline in the rate of serious injury that should have offset reasonable medical inflation.

What is needed is a more granular analysis by the nature of the injury; however, data captured by most claim systems is notoriously bad for such analysis because in large part it is captured at the beginning of the claim, is not fully developed, and often does not get updated.

But below is an example of a very quick and easy rudimentary means to obtain an order of magnitude measure of the change in injury cost patterns.

The settlement ranges are arbitrary but will work. The idea is to compare the most current full year of bodily injury settlements, e.g. 2010 against a chosen baseline year, say 2005. In the baseline and current cells are input the percentage of the total paid and closed bodily injury cases that fall inside of each range. The final column simply compares the change in those ratios from year to year. We know that bodily injury settlements tend to increase with typical inflation and we also know that the rate of serious injuries have declined, so we would expect to see evidence of both in the final column. But if the final column reflects increases well beyond what reasonable medical inflation would suggest, you can be more certain that your company is assuming its share of the cost shift.

In their early days Special Investigations Units were often integrated with and working in close proximately and partnership with the general claim population. Their presence and influence helped keep a universal focus on fraud. In many instances the general claim adjuster retained the claim and the Special Investigations Unit acted as a partner, an excellent training mechanism.

The formula was SIU + informed general claim adjusters = fraud deterrent

The former situation was not perfect because adjuster ability to identify and refer suspect cases varied. In time, rules-based intelligence and predictive modeling tools evolved to strengthen detection and referral. Early on, the new tools alerted the general adjuster who decided whether to refer. But over time alerts were increasingly elevated to the Special Investigations Unit directly because the tools increased referral activity and Special Investigations Unit managers needed a mechanism to control the volume and ensure that the cases with the highest potential were retained.

As the paths of the Special Investigations Unit and the general claim department increasingly diverged, the latter was also becoming increasingly specialized and downsized as automation and production became the top priority. The "need for speed" took an even greater toll on general claim fraud focus and today Special Investigations Units and general claims live in different worlds.

The Soft Fraud Deficit
The Special Investigations Unit has its focus largely on hard fraud and does its best to reach out to general claims about pervasive soft fraud, but the general claim department no longer has the capacity to effectively respond. As a result, a significant soft fraud deterrent deficit opened up across the industry. It's most notable manifestation is the significant rise in routine automobile injury medical costs.

This rise of routine medical costs involves both hard and soft fraud. Special Investigations Units have dedicated much time and attention to the hard fraud aspect and achieved many notable successes. But as the “churn and burn” automated claim processes of the general claim department gained speed, soft fraud went undetected and undeterred.

The irony is that the widely reported successes of highly professional Special Investigations Units leads carriers to believe that they have their fraud problems fully under control, hence scant awareness of the pervasive soft fraud exists. As predictive modeling tools continue to advance, the focus remains on the Special Investigations Unit, and while advances such as the identification of social networks are great and necessary, where are the tools for the general adjuster aimed at detecting soft fraud?

Who Owns Soft Fraud?
Most internal and external constituents continue to think of the Special Investigations Unit as the owner of fraud issues but Special Investigations Unit leaders are increasingly frustrated with their inability to influence change in the general claim departments.

While different companies may have different philosophies, it seems clear that the Special Investigations Unit is not synonymous with fraud. A wall has been erected, physically and psychologically, between the Special Investigations Unit and the general claim department that has orphaned soft fraud.

The Financial Deficit
The Insurance Research Council estimates that a hospital cost shift has been hitting the Property & Casualty industry to the tune of billions annually. This would likely be much worse if not for the work of Special Investigations Unit professionals pushing back on hard fraud. But unless the detection and prevention of soft fraud begins to get the attention it deserves, substantial fraud will continue unabated and consumers will continue to face needlessly excessive premiums.

In spite of such trends, loss costs were unyielding, sparking curiosity until further investigation identified that the cost of treating routine injuries had risen fast enough to eclipse what should have been substantial loss cost reductions. What is particularly vexing is that the Insurance Research Council also demonstrated that the nature of routine injuries has not changed, just the cost.

Medical Cost Shift
As the battle rages between medical providers and the government over Medicare and Medicaid reimbursement rates and private health insurance continues to decline, Property & Casualty claim departments have become the path of least resistance and are absorbing an increasing percentage of national healthcare costs. The Insurance Research Council estimates that the annual shift had reached 1.2 billion per year conservatively as of 2006. Inpatient emergency room billing practices reflect empirical evidence of a substantial upward rise in evaluation and management (E/M) billing levels occurring between 2002 and 2008.

The government and private health insurers have been far more effective at combatting these trends through negotiated fee arrangements than Property & Casualty claim departments who are over-reliant on automated systems that cannot detect upcoding. The billions in medical cost shift being absorbed by the Property & Casualty industry is akin to collateral damage to a bystander caught in the crossfire.

Diagnostic Upcoding
Property & Casualty claim departments rely on medical bill repricing software to keep medical costs in check through a combination of fee schedules, negotiated agreements, and where lacking, usual and customary charges. But there is one glitch — the software uses the providers; reported diagnostic code(s) as a baseline for vetting billing levels. Providers have learned to push back against government fee reductions by inflating diagnostic codes, and evidence of such activity is ubiquitous. Not only does diagnostic code inflation drive up medical bill repricing allowances, but the knowledge that it does so without any repercussion has many hospitals acutely aware of the opportunity presented by auto accident patients. A whole series of exploiting machinations have evolved to fully leverage the opportunity. This type of thing was once the domain of a few dishonest chiropractors, but the financial pressures heaped onto legitimate providers has caused it to go viral.

The following is an excerpt from a McKinsey and Company health care study:

Of the $2.1 trillion the United States spends on health care, nearly $650 billion is above expected, even when adjusting for the relative wealth of the U.S. economy

Outpatient care, which includes same-day hospital visits and is by far the largest and fastest-growing part of the U.S. health system, accounts for $436 billion, or two-thirds of spending above expected. Fueling this growth are a number of supply- and demand-related factors, including (1) provider capacity growth in response to high outpatient margins; (2) the judgment-based nature of physician care; (3) technological innovation that drives prices higher rather than lower; (4) demand growth that appears to be due to greater availability of supply; and (5) relatively price-insensitive patients with limited out-of-pocket costs.¹

Can Anything Be Done?
As the saying goes, "necessity is the mother of invention" and a number of new services are evolving that are aimed at preventing diagnostic upcoding, but claim leaders are reticent to try them. Recent history reveals several instances where efforts to curb medical abuse in claims have resulted in class action lawsuits and disruptive collection processes. In many states, claim leaders have decided that the cures are worse than the disease, but medical cost inflation is now so far out of control that a reassessment may be forced.

What seems to have been lost in the rush to drive down claim expenses over the past several years is the art of negotiation. Most providers are amenable to reasonable resolution when offered the opportunity versus simply having the terms of resolution dictated in a form letter. Claims don’t go to trial unless a carrier makes a conscious decision that it has no reasonable alternative, so pointing to legal outcomes to justify inaction is dubious. In states with Personal Injury Protection and relatively low limits, claim leaders feel even more compelled to just process the bills as quickly and inexpensively as possible but they totally miss the impact of diagnostic upcoding on bodily injury claims, multiplied by non-economic (pain and suffering) damages.

Who Gets Stuck With The Tab?
Generally speaking, the market remains highly competitive for automobile insurance so it could be argued that no one is adversely affected. But there are actually two cost shifts underway, one from the health care industry to Property & Casualty carriers and the other from Property & Casualty carriers to their customers. While auto insurance consumers don’t feel the cost pinch yet due to competitive factors, if gasoline prices continue to moderate and driving picks up, frequency will rise, but neither diagnostic codes nor the cost shift they have wrought will recede. In fact, they will both continue to gain momentum until either prices rise to the point of consumer rebellion or claim departments return to their fundamental roots of investigation, evaluation, and negotiation.

Great claim departments were once recognized as a tremendous competitive advantage, but the age of expense reduction and automation has lowered claim deterrents and even the desire to deter to historically low levels.

An iceberg has been struck, the ship is taking on water, but it’s still warm and dry up on the deck.

¹ McKinsey & Company, "Accounting for the cost of US health care: A new look at why Americans spend more." Dec. 2008, 9 Jan. 2012, http://www.mckinsey.com/Insights/MGI/Research/Americas/Accounting_for_the_cost_of_US_health_care