January 28, 2012
The CEO's Guide to Medical Inflation: The Case for Measurement, Part 2
by Michael Rowe
This is the second article in a three-part series written to education Property & Casualty executive leadership about how to discern whether and to what extent medical cost inflation is affecting their company.
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.