Assessing the Cost of Individual Radiologists’ Reporting Habits

Twitter icon
Facebook icon
LinkedIn icon
e-mail icon
Google icon
Medical billing systems are database files managed using programs that perform accounting functions. These are written to accommodate a daunting number of rules imposed by health-insurance companies and state/federal programs. They contain edit functions that screen all relevant acquired demographic and clinical information for completeness and adherence to claim-filing requirements. While this screening prevents submission of claims with incorrect field data, it cannot anticipate rules pertaining to physician credentialing, exam preauthorizations, patient eligibility, and so on. The system is capable, however, of providing valuable insight by cataloging insurance rejections based on payors’ rule sets. Such analysis provides useful information when these rejections are organized in various ways. There may be a preponderance of a certain rejection by ordering physician, reading physician, imaging site, or type of exam, for example. This information can be acted on to reduce rejections in the future. A system should provide reports that enable practices to judge short- and long-term progress in collecting patient-services income. The reports’ extent depends on file structure and data-mining capacity. Today’s systems can retain and catalog an unlimited amount of transaction detail, which can be rapidly compiled due to ever-increasing processing speeds. The degree of detail is only limited by the imagination of the practice administrators. If information is captured, it can be organized to suit the needs of the user. The most advanced systems are providing access to receivables reports via Internet, where practice members can log in and either automatically print standardized reports or use some data-mining options that enable them to isolate information specific to a practice segment. Transaction detail is generally offloaded to a separate server, either daily or weekly, so that this data manipulation does not interfere with operational processing (such as posting debits and credits or generating claims/statements). The Role of the RVU Many practices compile work RVUs per physician. It is easy to populate a library file with work RVUs per CPT code, using this library to compute monthly and year-to-date statistics based on exam volumes. It is possible to show the data as units posted to the receivables system in the processing month or to reorganize the statistics by original month of service (MOS). The MOS data, however, will be slightly inaccurate because a practice is required to use the date that the patient had the exam, not the date that the dictation was performed. In about 80% of cases, the two will be the same, but exams performed after 3 PM might not be read until the following day. Work RVUs can also be used to construct a time valuation, albeit an imperfect one. The work units are closely linked to the time that it takes to read and dictate a case. The entire fee schedule can be converted to time elements by dividing all work RVUs by the single-view chest RVU, then establishing an agreed-upon average time for handling a single-view chest exam, yielding a time conversion factor. Multiplying this conversion factor by all other work RVUs provides a library table of approximate reading times per exam. It then becomes possible to approximate the reading time per month, per physician. Although these data are readily available, there is limited evidence that they are used as a basis for income distribution. There is universal concern about the impact of volume-driven incentive plans on group chemistry. An important element in group cohesion is the fair distribution of departmental coverage, where some time slots carry lower reading volumes. There is also the opposite issue, where some members, being concerned about volume-driven income, take on inappropriately large caseloads that trigger reading errors and general burnout. Deriving the work RVU statistics per member is valuable as a way of benchmarking group size against regional/national criteria. Groups should share these data with the ACR, which occasionally surveys its members, so that all member practices have access to current information. The last work RVU survey included many practices that did not have the advantage of a full PACS environment, causing such large deviations that it made the statistics difficult to use as planning tools. It has been proven that PACS materially improves productivity. New, post-PACS survey data would be useful. Billing systems also compile actual cash revenues per physician. This is a bad idea because the information is useless in an environment where members are obligated to serve patients regardless of coverage. The availability of this information sometimes causes divisiveness in very large groups that are highly subspecialized, prompting divisions to seek to structure their benefit packages based upon their own income systems, which might happen to generate high revenues that they do not wish to share with divisions that produce less revenue. Enforcing Fair Play Accounts-receivable reports offer information mostly on effect, not cause. It is not a simple matter to standardize reporting on how well practice members comply with carrier-based rules in a manner that ensures the highest legitimate incomes. It may be useful to examine how a receivables system can be a tool to help a practice successfully play by the rules. The primary source of information generated by a radiologist is the dictated finding. This text file is part of the RIS that virtually all hospital departments have as a production system. The RIS sequence begins with an order by the referring physician, generally through electronic or telephoned scheduling. Eventually, all hospital departments will be equipped with electronic order-entry systems having appropriateness algorithms that accept requests only for clinically necessary exams. The ordered exam is part of the header of the report. It is in narrative, rather than CPT® code. Regulations prohibit using the ordered exam as the basis for assigning procedure codes. Only the dictated findings can be used as the source of the appropriate codes to be included on a submitted claim. The report is an HL7 electronic record in which the ordered exams are a unique field that can be stored by the receivables system. Ideally, the ordered exams can be compared to those derived from the dictation. The outcome of the comparison would stage the billing step; matches would be billed immediately, but deficiencies (fewer, or differently coded, procedures) would have to be checked for coder error. If the deficiency is valid, it will be returned to the reading physician, who either approves his or her findings or dictates an addendum. Dictations reflecting more exams than the header description are billed, and a file of the dictation and code variances is maintained for each reader. The dominant players in coding via natural language processing do have add-on subroutines that perform match-up comparisons of the ordered exams versus the evidence reflected within the findings text. Anomalies are automatically isolated. This is a very powerful tool for measuring dictation patterns.
imageTable 1. Statistics on Ultrasound Dictation Patterns
Assuming that coverage assignment is uniform, there should be patterns of dictation that follow recognizable trends. Table 1 shows an example of procedure grouping that can isolate issues requiring action by a practice. The groupings of codes pertain to abdominal and retroperitoneal ultrasound exams. The complete and limited studies carry different pricing and work RVUs. Dr E’s pattern is clearly at variance with the other four. The compilation at the bottom shows the statistics with and without Dr E to highlight the degree to which the dictation patterns of the other four are so uniform (as measured by their very low standard deviations). The economic implications can be easily computed with these statistics.
imageTable 2. Statistical & Economic Variances of Dr. E Dictation Patterns
In Table 2, Dr E’s dictation patterns are reconstituted to match the other four. Although both charge and work-RVU changes are shown here, the change in work RVUs is more relevant because it is the basis for insurance reimbursement. Dr E’s dictation pattern for these specific exam combinations might indicate poor documentation, triggering the more complete exam that was ordered to default to a more limited study. The multiorgan abdominal ultrasound is a study encompassing eight organ systems. If the physician looks at the other organs and finds them unremarkable, but comments in the report only on the pathology in the spleen, then according to coding guidelines, that study must be coded as a limited study. It is still statistically possible that Dr E is correct and the other four are wrong (the probability that Dr E’s caseload fits within this normalized population is less than 3%). The benefit of the compilation is to bring this pattern to light, then review Dr E’s cases to determine whether there is, in fact, a probable loss of revenues. A practice can construct any number of exam configurations that it considers meaningful. Procedures that have codes for two versus three views can be isolated. If a reader fails to cite the number of views, the coder must default to the lowest number. Some complex coding configurations require more sophistication. For example, the same tracking for Dr A through Dr E could pertain to myocardial perfusion imaging, where incremental coding for wall motion and ejection fraction require their presence in the dictation. Another example would be differing patterns of CT angiography exams versus CT with contrast (which would be the default if the reader fails to mention reconstruction/reformatting). Summary Receivables systems can track virtually unlimited trends pertaining to the billing and collection of professional charges. The only impediment is the imagination of the user. Spotting poor dictation patterns that cost a practice real income is more challenging, but not impossible. The use of the ordered exam as a benchmark against the dictated finding can identify readers who do not document their work well. Subsequent organization of monthly exam patterns by physician can also identify a reader who is careless in not citing all relevant views/exam components.