Six lessons about intended ... and unintended ... results.

In the first part of this article, we shared six design and implementation lessons that were learned in the preliminary stages of our new ambulatory utilization review (AUR) program. In this follow-up article, we discuss some of the lessons that have been learned from the results of the program.

AUR Lesson #7: How to Measure Savings. Most efforts to measure savings in managed care and related programs have proved to be at least as much art as science. AUR savings measurement is a bit more straightforward than most such efforts, because AUR entails extensive integration of both the administrative processes and the data associated with claims payment and review.

Integration of the administrative processes was essential because of the need to review a high volume of ambulatory episodes of care without interfering with claim turnaround time objectives. Similarly, the use of claims data during the review process provided an electronic environment amenable to careful collection of detailed data on the review actions taken linked to each claim reviewed.

While the linked claims-review database permitted direct calculation of the relationship between AUR actions and claims, the nature of the linkage of claims and review processes did uncover one slight complication. AUR helps determine whether or not a claim should be deemed "eligible for payment," which itself precedes calculation of plan benefits. For example:

* AUR might determine that a service is not medically necessary and that therefore it should not be considered "eligible for payment." While such an action clearly entails some "savings," it is not at all clear how to quantify the amount in terms of claims payments.

* The fee charged may have exceeded the amount that would have been allowable under the benefit plan.

* The patient may not have reached the benefit plan deductible, so, even had the charge been considered eligible for payment, there would not have been a payment made for that service.

* We might have been the secondary payer under coordination of benefit (COB) provisions, so the amount payable for the charge, had it been eligible for payment, would have been reduced.

* The benefit plan might have only paid 80 percent for the service (of course, unless the patient's out-of-pocket maximum under the benefit plan were reached). These and other benefit-plan-dependent factors need to be taken into account in order to measure AUR program savings. For the savings estimates presented here, this was done in a four-step process.

First, a nonrandom sample of 25 large employer groups, each with a different benefit plan design and with mature AUR experience, was selected for study. This provided not only a representative sample of benefit plan designs, but also a large sample of claims: $348,217,734 of covered expense for claims paid from October 1989 to May 1990.

Second, the net reduction in expenses "eligible for payment" (also known as "covered expenses") was calculated directly from the claims records.

The third step in calculating savings was to determine the relationship between total expenses "eligible for payment" and total benefits actually paid. In this case, the amount of benefits paid was 81.11 percent of covered expenses, or $282,438,018.

Finally, this percentage was used to reduce the savings directly measured with respect to eligible or covered expenses in order to estimate the savings in terms of benefit payments. How these savings estimates fell yielded several additional lessons.

Expected Savings from the AUR Program

We implemented our AUR program in 1986 in conjunction with preferred provider programs. At that time, virtually all utilization review focused on inpatient hospital care, and benefit plan designs often had incentives to encourage ambulatory care in lieu of inpatient care. However, there was some concern that the health cost pendulum might have swung too far in the ambulatory direction.

The same incentives that promote ambulatory care as an alternative to inpatient care also provide incentives to overutilize ambulatory care. For example, some network-based benefit plans had zero or significantly reduced employee coinsurance or copayments for doctors' office visits. There was a valid concern that it would be tempting for some physicians to "recoup" revenues lost from discounting via increased frequency of service, and employees would have little incentive to discourage this utilization.

AUR Lesson #8: Identification of Inappropriate Utilization. The largest share, 72 percent, of the savings from the AUR program in 1990 came from application of the Patterns of Treatment[R]. (Patterns of Treatment is licensed for use by Health Payment Review, Inc., Boston, MA. See Part 1, Sept.-Oct., 1993, issue of Physician Executive, for description.) The case studies included as boxes with this article provide examples of situations where services "fail" the Patterns:

* Services provided with excessive frequency (Cases 2, 3, and 4).

* Services that are inappropriate for a given diagnosis (Cases 1 and 3).

* Services that are investigational or not generally accepted medical practice (Cases 1 and 3). Review of the cases also serves well to introduce the second set of lessons learned from AUR--the unintended results.

Unintended Benefits of the AUR Program

Ambulatory care is often a fairly unstructured set of services provided by multiple providers to the same patient over extended periods. In contrast, the AUR process comprises comparisons of such episodes of ambulatory care to a very structured set of criteria, the Patterns of Treatment. This comparison identifies any deviation from the Patterns standards.

One example of this phenomenon was demonstrated in Lesson #7. Not only were services found to be provided in excess of standard frequencies, but services or codes not included in the Patterns--services inappropriate for a given diagnosis--were also identified. This "exception reporting" inherent in the AUR process resulted in several additional types of program savings, the first of which derived from focusing review on "exceptional" physicians.

AUR Lesson #9: Focused Provider Review. One initial lesson learned from AUR was that a very small percentage of all physicians accounts for most "problem" claims: less than 1 percent of physicians account for more than 80 percent of all AUR-declined charges. A logical result of this phenomenon is to focus review on that small percentage of providers associated with most of the question able claims.

Physicians on whom AUR is focused often turn out to be submitting claims that are "questionable" for a variety of reasons beyond their variance from the Patterns of Treatment. These aberrant practices would be rather hard to find in the absence of AUR--for example, changing tax identification number (TIN) in an effort to resume previously uncovered aberrant practices; changing diagnoses to seek payment for uncovered services; and uncovering claims missed by AUR because a Pattern had not yet been established for a new technique or because CPT-4 procedure codes did not exist.

Cases 1, 3, and 4 illustrate uses of focused AUR that make up 13.3 percent of the total direct AUR savings. The final category of unexpected savings, 14.7 percent of the total savings, came from "other" sources, primarily identification of aberrant billing practices.

AUR Lesson #10: Identification of Aberrant Billing Practices. Part I of this article recounted a lesson critical to AUR implementation, the close cooperation and the physical proximity of claims and medical management personnel. This aspect of AUR implementation resulted in improved detection of creative billing practices, which, in turn, reduced claims payouts.

The most common example of creative billing is coding fragmentation, also called "unbundling." With this type of billing, practitioners attempt to increase revenues by charging for individual components of a service that should be billed under one, less expensive and all-encompassing code (see Case 2).

More esoteric examples of creative billing may arise before a claim is entered into the claim system--for example, a new procedure or therapy that looks unfamiliar to a claim approver or one for which no procedure code exists. In such circumstances, claims personnel confer with AUR staff and together investigate the claim.

Because creative billing practices are constantly evolving, individual claim situations that arise usually result in revised claim payment guidelines. For example, as a result of the AUR program, we developed and implemented an electronic Coding Fragmentation Review Program (CFRP). While this new, knowledge-based technology may decrease some of the savings directly attributable to AUR, overall cost-effectiveness will be enhanced. CFRP not only can be handled largely by the computer, but also reviews inpatient surgical procedure coding not specifically addressed by AUR. Improved coding due to CFRP may increase other areas of AUR savings.

Savings Summary

AUR reduced a "covered expense" base of $350.7 million by $2,514,000, or about 0.7 percent. Recalling that claims payments in the study reported here were about 81 percent of covered or eligible expenses, the AUR savings may also be expressed as a $2,039,000 reduction in claims paid (AUR Lesson #7).

The significance of the AUR savings may be appreciated in two senses. First, the savings outweigh the costs of the program by a ratio of more than 4.84:1. That is, for every dollar of program cost, there is a return of $4.84 in directly measured claims payment reductions. If sentinel effect savings were included, the ratio would be considerably higher. (The program's sentinel effect is the change in physicians' behavior that results from their knowledge that they are being reviewed. Such a reduction in aberrant practices yields lower claims expense compared to levels expected in the absence of AUR, but does not affect directly measured savings.) Second, there are two types of indirect benefits of AUR that need to be considered: its contributions to selection of network providers and quality assurance in managed care.

AUR Lesson #11: AUR Provider Profiling. Early in the AUR program, we thought that the Patterns of Treatment, used in statistical analysis, could help profile physicians in advance of their consideration for participation in our managed care networks. Because of benefit plan considerations and certain limitations of claims data, initial attempts at profiling physicians by merely passing historical data through the Patterns failed.

Subsequently. however, the results of medical professional review were svstematically captured and reported, and the desired profiles did become useful. The table above shows excerpts from a monthly report we produce for each three-digit Zip Code area in the United States. These reports are used for three purposes:

* To identify physicians who may not be appropriate for invitation to be network physicians--for example, Doctor 12345 has a very high (52 percent) percentage of claim dollars that were declined.

* To identify physicians already under contract who may need some counseling and remedial attention--for example, Doctor 12346 had only 5.7 percent of his charges declined, but most "network" physicians have zero percent of charges declined. Doctor 12346 may need to be contacted in the hope that some education will help change his or her behavior.

* To complement previously developed inpatient profiling capabilities to provide a complete picture of physicians' practices.

While the "savings" from this use of information are somewhat intangible in the short run, this process will, over time, result in networks consisting of physicians with much more efficient practice patterns than those of "nonnetwork" practitioners.

AUR Lesson #12: Quality Assurance. The previous AUR lessons showed that physicians who are "outliers" in one aspect of their performance are often outliers in other respects. Inappropriate utilization and "creative billing," for example, seem to go hand in hand. Physicians found to be aberrant by AUR in these respects sometimes also practice poor quality medicine: providing medically unnecessary care is providing poor quality care!

AUR's attention to the appropriateness of care generally helps efforts to ensure high quality. More specifically, it helps identify unnecessary exposure of patients to potentially significant discomfort or risks. The case examples illustrate this. AUR can help patients avoid unnecessary injections (Case 1) or unnecessary intravenous therapy (Case 3). And one must wonder about the quality of care when vitamin and nutritional therapy are employed by a physician for treatment of a condition ordinarily requiring, hospitalization (Case 4).

While most of the focus of AUR is on excessive or unnecessary care, there is another potential application: identifying gaps in care. Preventive services, such as periodic health screening, lend themselves to AUR, particularly in a primary care physician-based network environment.

Summary

In 1986, when we implemented AUR, health care cost containment was on the minds of most employers. AUR's initial objective, therefore, was to achieve cost-containment by identifying unnecessary ambulatory utilization. This it did ... and much more.

Unanticipated benefits of AUR and the advent of managed care have heightened the importance of this information technology. Physicians who are aberrant in one respect often fail to pass muster in others. Inappropriate ambulatory practice patterns may be evident in clinical practice or billing, but they are certainly inappropriate in managed care.

Historically, physicians' guardianship of medical information has limited the ability of patients or benefit plan sponsors to discriminate on the basis of value when seeking or purchasing care. Patients rely on their physicians to exercise patient care judgments. AUR and related information technologies employed in managed care do not seek to supplant the trust between a patient and his or her physician. They do seek to augment employers' and other benefit plan sponsors' efforts to achieve improved value. This value may be achieved, in managed care, by restricting access to carefully selected networks of high-quality providers. AUR is one good tool for this purpose.

Ambulatory Utilization Review, Case 1

A claim was submitted for a 41-year-old male with a diagnosis of chronic obstructive pulmonary disease. The claim was for the following services, all rendered during one visit to a physician.

  CPT-4 Code      Description                         Charge
  90020           Office Visit, New, Comprehensive     $185
  94010           Breathing Capacity Test                70
  99070           Special Supplies                       65
  64405           Injection for Nerve Block             170
  93000           Electrocardiogram, Complete            60
  Claim total                                          $550

This claim was suspended by AUR technology for two reasons. It failed the criteria for inappropriate utilization, and the provider involved was previously identified for focused review. This claim was referred for medical review because the procedure "injection for nerve block" is inappropriate for the diagnosis of chronic obstructive pulmonary disease. This would have caused the claim to suspend for review even if the provider were not identified for focused review. Further investigation into previous claims submitted by this physician revealed that the procedure "injection for nerve block" was a combination of medications injected in order to diminish the effects of nicotine addiction. The services rendered were actually for a "smoking cessation program" that was not a covered expense.

This case illustrates the quality implications of ambulatory review on several levels. First, detection of services that are inappropriate for the diagnosis frequently leads to detection of services that are not eligible for coverage but have been billed in such a way that benefit payments will be made. The AUR process assists in detection of these claims and the providers who submit them, and feedback is given to the claims department so that the quality of overall review is increased.

On another level this case focuses review on the actual services rendered. Not only is the therapy unproven for the treatment of chronic obstructive pulmonary disease or nicotine addiction, but injections expose the patient to a variety of risks. The medications used in this case include atropine, epinephrine, lidocaine, and scopolamine, which carry a high risk when not used appropriately.

Ambulatory Utilization Review, Case 2

A 41-year-old female submitted a claim with a diagnosis of gastric ulcer and the following charges:

CPT-4 Code       Description                                      Charge
90080            Office Visit, Established Comprehensive          $60
71010            Radiologic Examination, Chest-Single
                 View, Frontal                                     40
71020            Radiologic Examination, Chest-Two
                 Views, Frontal and Lateral                        60
71030            Radiologic Examination, Chest-Complete,
                 Minimum of Four Views                             60
91299            Unlisted Diagnostic Gastroenterology Procedure   400
Claim total                                                       $620

In this case, services have been "unbundled," which is the practice of coding separately for individual components of a test that should be billed together under one code. In this case, there are three x-ray tests billed. Code 71030 should have been billed alone, because it included the services for 71020 and 71010. In addition, service 91299 was suspended for review because it is an unlisted procedure. This suspension for review would result in an inquiry to the physician about the specific procedure being rendered, which in this case was "gastric analysis" and was declined as medically unnecessary. Here, AUR technology assisted both with the determination of medical necessity and with the review of appropriate billing practices.

Ambulatory Utilization Review, Case 3

A provider routinely submits a claim for the following services for the diagnosis of arteriosclerosis.

  CPT-4 Code      Description                  Charge
  90060           Office Visit, Intermediate    $100
  90780           TV Infusion                    175
  Claim total                                   $275

This claim was suspended by AUR for both office visit frequency (the service is usually rendered weekly) and inappropriateness of the infusion for the diagnosis given. The physician in this case was rendering intravenous "chelation" therapy, typically used as a treatment for lead poisoning but in this case used to remove plaque associated with arteriosclerosis. Initially, services were denied as not generally accepted medical therapy for the given diagnosis. However, the doctor changed his billing practice and started using the diagnosis of lead toxicity, which would have allowed the service to be paid were the physician not placed under focused AUR. The doctor screened for lead poisoning by doing "hair analysis," a test known to have an unacceptably high false positive rate to "establish" a diagnosis of lead toxicity. Standard serum lead levels were never done. Patients of this doctor with "lead toxicity" were then subjected to three to four intravenous therapy sessions per month, each session lasting several hours. The potentially toxic medications, such as EDTA, used in this treatment were not indicated for the diagnosis of arteriosclerosis. To subject patients to "chelation" therapy for the diagnosis of arteriosclerosis is not considered safe, efficacious, or generally accepted. The patient was also exposed to the discomfort and risks associated with prolonged TV therapy.

Ambulatory Utilization Review, Case 4

A physician routinely submitted an identical work-up on all patients on the initial visit for the diagnosis of disseminated candidiasis and fatigue. The services were billed as follows:

  CPT4 Code         Description                     Charge
  90020             Office Visit, New                $150
  86008             Antibody, Quant., First            15
  86009             Antibody, Quant., Additional      435
  90784             Injection  IV)                    120
  DRx               Take Home Rx                      121
  85021             Automated Hemogram                 25
  80062             Cardiac Evaluation Panel           40
  85650             RBC Sedimentation Rate             10
  86357             Lymphocytes, T&B, Diff.           200
  87081             Bacteria Culture Screen            35
  87081             Bacteria Culture Screen            35
  87081             Bacteria Culture Screen            35
  87081             Bacteria Culture Screen            90
  86421             Radioallergosorbent Test           30
  87081             Bacteria Culture Screen            35
  90784             Injection (IV)                    120
Claim total                                      $1,496

The services listed here were suspended by AUR for both frequency and inappropriateness. The diagnosis of disseminated candidiasis, if valid, would indicate admission to a hospital. Follow-up investigation and request of office records indicated that the diagnosis was not verified. After the initial work-up, the physician would routinely start the patient on a course of weekly vitamin therapy and nutritional supplements, the necessity of which was not evident based on review of the office records.

The initial investigation of this provider resulted in placement of the provider on "focused review," in which all claims submitted were reviewed. The focused review revealed that nearly all claims were identical for diagnosis, work-up, and treatment. Testing appeared to be done for the purpose of validation of the physician's treatment of nutritional support, rather than for diagnostic need based on individual history.