The "Colossus" Claims Program, and How It Affects You and Your Patients
By Arthur Croft, DC, MS, MPH, FACO"This is the voice of world control. Obey me and live, or disobey and die." So went the words of the out-of-control renegade computer Colossus in the 1970 Universal Pictures feature, The Forbin Project. In the movie, both Americans and Soviets agreed (at least) on one point: human fingers hovering over launch buttons threatened the utter destruction of the human race. A computer, they reasoned, would provide the logical safety net for the mutually assured, destruction-capable nuclear arsenal, while still allowing each country its highly explosive security blanket.
Unfortunately, Dr. Forbin inadvertently built his computer with an inherent flaw - a piece of code that resembled human DNA, perhaps - and once the computer had the Russian and American launch codes for their respective nuclear arsenals, it went rogue in a big way, blowing New York of the face of the map as a gesture of its evil intent for world domination. But that was only a movie. ...
The original Colossus of Rhodes was built in about 282 BC of bronze and stood 110 feet in height. It was a statue of the Greek god Helios, so remarkable in its construction that it was considered one of the "Seven Wonders of the World." However, it lasted only 58 years, crashing to the ground during an earthquake.
Today, the name "Colossus" is most often associated with a computer program developed originally in Australia for the purpose of personal-injury claims review and valuation. Subsequently, the program was acquired by Computer Sciences Corporation, which has cornered the market on cost-containment software, with nearly $10 billion in revenue in 2000.1 Colossus is now implemented by 70 or more of the major insurers in the U.S. Its use is quite simple: The claims adjustor sits in front of the computer, inputs case data, and Colossus spits out a settlement offer in dollars and cents. (What the insurer does with that figure is another issue.)
Colossus code is based on what is referred to as artificial intelligence (AI), meaning that the program code continuously evolves based on user input. For AI programs, such as those that predict weather patterns, this is an important feature; there is no substitute for experience. But in any AI-based system, evolving code can only be as reliable and useful as the data it has digested; bad data makes for dubious code. When that happens, "AI" becomes something of a misnomer. (Perhaps it should more realistically stand for adulterative imbroglio, but I digress.)
Herein lies one of the dubious features of the AI of Colossus, and also a fact physicians should come to appreciate. This is probably the most important point I can make: Virtually all of the major influences of the Colossus output are based on the physician's daily notes and other documents. Moreover, what you write in your notes - and often, the specific words or phrases you use - can have a significant impact on the value Colossus attaches to the claim. Again, in plain English: The doctor's records are the chief data source for Colossus.
Is this likely to influence you or your patients? Yes, for a number of reasons: Suppose your patient became injured as a result of the insured's negligence, and suppose that patient expected a monetary settlement of some kind to cover his or her medical and other expenses, inconveniences, etc. Now, suppose Colossus decided the claim was worth - say - $2,334.24 (which is what the insurer will probably offer). If this amount wouldn't even cover the medical bills, it likely will not mollify your patient, making a lawsuit more likely. Conversely, if Colossus valued the claim higher, that offer might well be acceptable to your patient, making life easier for you.
There are newer and better programs that serve this claims-assessment purpose, but the majority of insurers in the U.S. use Colossus. Another program physicians should know about it ISO's ClaimSearch, which serves as a clearinghouse of information about hundreds of millions of liability, property and bodily-injury claims. It is a cooperative effort of more than 3,500 insurance companies, representing more than 90 percent of all policies written. It allows them to search through virtual mountains of digital data, extracting vital information about claimants, such as their previous health profiles, prior claims, etc. What most physicians and plaintiff lawyers don't realize, however, is that they, too, are profiled by these computer Goliaths - not just their medical profiles, but also their practice profiles.
When a physician treats patients who make claims against insurers for injuries suffered in automobile collisions, the (Colossus-derived) value of the "cases" they treat are recorded and associated with the physician's federal tax ID number or Social Security number. Because Colossus follows the AI operational behavior, it means that if Colossus devalued most of your cases last year because you took poor notes; failed to refer for second opinions; had illegible handwriting; or used the "wrong" terminology (more about that later), the offers it comes up with for claims you manage this year will be correspondingly lower. But that's not all: The attorney on the case also is tracked and (depending on his or her track record) can also affect future claims offers. Moreover, the physicians are linked (again, by the federal tax ID or SS number) to the attorneys. So, if you happen to treat a substantial number of patients referred to you by plaintiff attorneys who routinely settle claims for very little and rarely (or never) set foot in court, you become digitally linked and branded in such a way that the settlement amounts offered by Colossus for your patients' claims may be lower than average.
When the plaintiff does not deign to accept an offer recommended by Colossus, he or she often files a lawsuit and take the claim to the next level. In some states, the original offer becomes the benchmark for the case, because if the case goes to trial and the jury returns a defense verdict, or offers less money than the original settlement offer, the plaintiff is then obligated to pay the defense's legal costs, in addition to his or her own. On the other hand, if the jury awarded a six-figure amount, the original offer by Colossus would be all Colossus would ever know about the case, thus ensuring that the program does not evolve in a direction detrimental to the entities using it. After all, it simply wouldn't do to enter the six-figure amount in the "settlement" field, so that Colossus would place a higher value on a similar claim next time.
The closer one examines Colossus, the more one comes to realize it is consistently manipulated to downgrade claims over time, rather than evolving, like most true AI programs, on the basis of newer, more accurate, unbiased information. For example, Allstate uses a program called the Medical Bill Review Service (MBRS) to determine how much treatment was reasonable and necessary. This amount, rather than the actual physician's bill, is entered into Colossus under "physician fees." For MBRS, however, chiropractic treatment is cut off arbitrarily at eight weeks or 22 visits, regardless of the severity of injury or other complicating variables.2 Some insurers also arbitrarily shave an additional 20 percent off physicians' fees.
In other cases, extrinsic factors may exert more influence. For example, some insurers cap all settlement offers at $3,000 when the patient's vehicle property damage is below $1,000, regardless of other circumstances. And because the adjustor will enter a code for liability in every case (e.g., "clear," "unclear," or "contested"), there is another potential for the adjustor to prompt reductions in settlement offers when, for example, liability is incorrectly listed as unclear or contested. This practice occurs in states that have legal liability division, allowing reductions in liability based on comparative negligence. According to DeShaw,2 Farmers Insurance has instituted a nationwide goal of finding an average of 15 percent contributory liability. Similarly, severity of impact is input as "light," "moderate," or "major." One might well ask how impact severity can be judged fairly in light of the fact that most passenger vehicles can withstand impacts of up to 8-12 mph without sustaining property damage, while the risk for injury for many individuals is quite high below this range.
There's also something along the lines of "having one's cake and eating it too," when it comes to Colossus. Some companies arbitrarily shave 20 percent off the top of what Colossus comes up with (or they may decide to reject the number altogether), and frequently, data that would clearly elevate the value of the claim is withheld purposely from Colossus. For example, Colossus reasons that in any case in which a second opinion is sought, the claim should be worth more. Therefore, referrals are a value driver. So, if an insurer sends the patient out for a defense medical examination, it may effectively inflate the case's value. Not surprisingly, many insurers withhold this information from Colossus. (By the way, DCs count as second opinions, although DC-treated cases are valued lower by Colossus than those treated by MDs.)
Earlier, I noted that certain words or phrases were particularly influential for Colossus. At present, its code is based on approximately 10,400 value drivers. (This is the fuel that drives the value of a claim.) However, Colossus does not always value findings in a way that is concordant with general medical or chiropractic thinking. For example, in the last several years, there have been several reports in the literature that initial losses of cervical spine range of motion are important prognostic features of whiplash injury. Individuals with greater initial limitations generally don't fare as well as those with minimal initial limitations. Similarly, the American Medical Association and the American Academy of Orthopedic Surgeons have long considered permanent loss of range of motion an important factor in terms of impairment. On the other hand, muscle spasm, as a finding, is fairly nonspecific and offers little in the way of reliable or meaningful outcome forecasting. Nevertheless, Colossus apparently attaches at least as much significance to muscle spasm as it does to loss of range of motion. In fact, if the physician notes three different points of muscle spasm, the condition is that much more severe in the view of Colossus.
Another example is contusions. In and of themselves, they are usually little more than an inconvenience to the patient and a cosmetic "nonissue" to physicians (except in special cases, such as the "seat-belt sign"), but these are once again significant to Colossus. Again, the more contusions present, the greater the claim value is likely to be in Colossus' digitally glaucomatous vision. Many of us might not normally document each and every contusion or point of muscle spasm and would unknowingly deflate the claim's value. Needless to say, physicians who write illegibly or fail to document other important findings deprive Colossus of the very substrate claims adjustors put into it to assess the claim.
Another important - but often undocumented - value driver is the patient's participation in his or her normal gainful employment while still in pain. Referred to as "duties under duress," Colossus considers this a serious issue, so when it is present, it should be noted accordingly. I also should point out that much of what drives Colossus remains sub rosa, or hidden. Most of what we now know comes from disaffected claims adjustors, former defense attorneys, and evidence leaked from bad-faith trials that have transmogrified into the public domain. (See "Recommended Reading" and "Recommended Software" on page 39.)
What do the words "herniations," "extrusion," "prolapse," "protrusion," "bulge," and the term "strain/sprain" have in common? It all depends on your programming. While most radiologists have their own preferred lexicon when it comes to describing disc pathology, we generally consider extrusion, herniation, prolapse, and protrusion to signify a degree of disc pathology capable of producing significant amounts of pain and disability - potentially on a more-or-less permanent basis. Alternatively, they may require surgical remediation.
"Bulge" is a term we generally consider less ominous and often inconsequential, and "strain/sprain" is another entity entirely. Colossus, however, recognizes only the words "bulge" and "herniation" under the heading of "disc lesion." According to DeShaw,2 a high-level Farmers memo suggested to its adjustors that they enter diagnoses named "protrusion," "prolapse," and even "extrusion," as "bulges," if radiologists did not use the specific term, "herniation." Bulges are automatically downcoded by Colossus to a strain/sprain diagnosis. In a similar way, "subluxation," in a chiropractic case, will be coded as a strain/sprain diagnosis. So, if there is a neurological component, it should be diagnosed separately.
To say that Colossus is indelibly and irretrievably flawed, from a scientific or medical basis, is probably too charitable. It represents the apogee of antiscience: its yardsticks are poorly conceived and hardly calibrated. And while the entering of medical data provides the verisimilitude of scientific credibility and reliability, closer examination reveals a rather serious lack of scientific underpinning and validity, compounded with a very unreliable data-entry methodology. It never had a firm epidemiological or physiological basis, and has subsequently become so polluted, through faulty data input and arbitrary manipulation by insurers, that it is difficult to say what it really is at this point. Not only do the insurers make up the rules of the Colossus game, they also decide when (and if) they will follow them. However, for the most part, insurers are private or public businesses and can jolly well do whatever they like when it comes to managing their bottom line. Like it or not, Colossus and other programs mentioned here will be with us for the long haul, so it is helpful for physicians to understand some of the inner workings of these programs so that their patients can be dealt with as fairly as possible by the insurers who use them.
Concerning the issue of impairment or disability, lamentably, Colossus will not give credence to reports from chiropractors. They must come from a medical doctor. (DeShaw believes this general bias against chiropractic may stem from the fact that the original program was developed in Australia during a time when chiropractors were not fully recognized by the government. 2) In light of this, some are advocating the sandwiching of chiropractic between medical providers, such that the initial visit is with a medical practitioner who recommends chiropractic care, then sees the patient at the end of that care to provide any final impairment assessment needed.
As always, I advise physicians to say what they mean and mean what they say. I certainly have never recommended or condoned the use of wording, diagnoses or terms that were not truthful or accurate. However, if use of terms otherwise medically synonymous to physicians (e.g., extrusion, protrusion, prolapse, or herniation) may actually result in a diagnostic error based on the idiosyncrasies of the Colossus code, physicians should act accordingly and translate terms in such a way that Colossus will not misinterpret their significance. Physicians also should be sure to make their daily SOAP notes complete, legible and understandable, and should avoid the use of nonstandard shorthand. After all, your SOAP notes are one of the chief data sources for Colossus.
Patients should probably also now be concerned that property damage to their vehicles is assessed fairly and accurately. Many body shops participate in an affiliate cooperative arrangement where incentive discounts are offered to insurers. This often results in repairs being made with aftermarket or reconditioned parts, or with parts being repaired that should rightfully be replaced. We also are starting to see more egregious violations being reported, such as shops replacing airbags with used or stolen bags, or even replacing the bags with rags or other inert filler material. Independent repair estimates are a good idea, for this reason and because property damage may not always be visible to the untrained observer. Some shops have laser devices that can even detect subtle frame distortions.
I strongly encourage physicians and attorneys who treat or represent the injured to obtain a copy of the book by Aaron DeShaw, DC, JD,2 which served as an important resource for this article, and contains much more important information than I can convey here. This insight can greatly assist you, your patient, the patient's counsel, and, yes, even the claims person handling the claim. Dr. DeShaw can be reached at .
James Mathis was in charge of the defense department for a large insurer for many years. He become highly acquainted with the workings of Colossus in that capacity, and has since left the insurance side of personal injury. This resulted in a number of lawsuits: he sued the insurer, it sued him; he was sued by the software owner, who he then countersued; and then there were the bad-faith claims in which Colossus played a key part. Mathis now teaches seminars on the subject, and is considered to be the leading authority on Colossus. I spoke with him, at length, several days before writing this article.
Mathis, whose Web site is www.sequoiavisions.com, has developed two software programs: Demand Expert, intended to assist attorneys with the complexities or vagaries of Colossus; and Medical Report Expert, which is intended for physicians of all types and provides the same help and guidance to them. If the attorney representing the physician's patient uses Demand Expert, the reports created by Medical Report Expert will be importable directly. Otherwise, Medical Report Writer takes the physician or office worker through the important details of the case and prepares a report Colossus will understand, leaving no significant stone unturned. I have tinkered with Medical Report Writer, and it appears to do the job quite well, although it will not, of course, take the place of the final narrative report.
I would like to point out that I do not receive any incentive or financial gain in any way from the products recommended here, and none of the organizations or corporations, public or private, that I am affiliated with gain any benefit or sponsorship from sales of these products. I do not have any business or personal associations with Dr. DeShaw or Mr. Mathis, or their respective companies, and recommend the products only so they may assist practitioners and their patients.
Arthur Croft, DC, MS, MPH, FACO, FACFE
Click here for more information about Arthur Croft, DC, MS, MPH, FACO.