Health Insurance: Coverage of Autologous Bone Marrow Transplantation for Breast Cancer

September 1, 1996

Rapid advances in biomedical research and technology are producing a continuous stream of new, and often expensive, medical devices, drugs, and therapies. Health insurers' decisions about whether and when to provide coverage for these new

United States
GAO General Accounting Office
Washington, D.C. 20548
Health, Education, and Human Services Division

April 24, 1996

The Honorable Ron Wyden
United States Senate

Dear Senator Wyden:

Rapid advances in biomedical research and technology are producinga continuous stream of new, and often expensive, medical devices,drugs, and therapies. Health insurers' decisions about whetherand when to provide coverage for these new medical products andtreatments play a pivotal role in determining their availabilityfor use in general medical practice. In recent years, conflictover insurers' coverage decisions of new medical treatments hasled to litigation and to a variety of federal and state legislationand regulations that mandates insurance coverage of some new medicaltreatments.

Some of the most visible and contentious coverage decisions haveinvolved the treatment of breast cancer with high-dose chemotherapysupported by autologous bone marrow transplantation (ABMT). Inthis procedure, bone marrow or stem cells from the blood are takenfrom the patient and then reinfused after high doses of chemotherapyhave been administered. The high-dose chemotherapy is toxic tothe bone marrow which produces the blood cells that fight infections.ABMT following the chemotherapy treatment helps restore the patient'sability to produce the blood cells that fight infection.

Most experts say that more research is needed before definitiveconclusions can be reached about the treatment's effectivenesscompared with conventional chemotherapy. Proponents of insurancecoverage of the procedure say it provides breast cancer patientswith access to a promising, potentially life-saving treatment.Critics say that the public is not well-served by the proliferationof an unproven treatment that is costly and possibly harmful,and that such proliferation hinders clinical research to determineif the treatment is effective.

To illustrate the issues raised when demand grows for coverageof a new treatment in advanced clinical trials, you asked thatwe provide you with information regarding insurance coverage ofABMT for breast cancer. Specifically, you asked that we address(1) the factors that have influenced insurers in deciding whetherto cover the treatment, (2) the status of the research on ABMTfor breast cancer and the consensus on what is known about itseffectiveness, and (3) the consequences of the increased use andinsurance coverage of the treatment while it is still being evaluatedin clinical trials.

To develop this information, we conducted structured interviewswith officials responsible for medical coverage decisions at 12health insurance companies, including some of the nation's largestinsurers.[1] These companies also included a mix of indemnityand managed care plans. We also obtained information from researchersand oncologists at major research centers, large urban hospitals,and community hospitals. Others we obtained information from includedthe American Society of Clinical Oncology; the National Associationof Insurance Commissioners; patient and women's health advocates,including the National Breast Cancer Coalition; state health officials;technology assessment organizations; and the National Cancer Institute(NCI). We also reviewed state and federal legislation and regulationsregarding insurance coverage of ABMT, as well as relevant scientificliterature, and visited a large, private transplant center. TheNational Institutes of Health (NIH) and the Office of PersonnelManagement (OPM) provided formal comments on a draft of this report.[2]We did our fieldwork and analysis from April to December 1995in accordance with generally accepted government auditing standards.

As agreed with your office, unless you release its contents earlier,we plan no further distribution of this report for 30 days. Atthat time, we will send copies to other congressional committeesand members with an interest in this matter, the Secretary ofHealth and Human Services; the Director, NIH; and the Director,OPM.

This report was prepared by William Reis, Assistant Director;Joan Mahagan; and Jason Bromberg under the direction of Mark Nadel,Associate Director. Please contact me on (202) 512-7119 or Mr.Reis on (617) 565-7488 if you or your staff have any questionson this report.

Sincerely yours,

Sarah F. Jaggar
Director, Health Financing
and Public Health Issues

[The text of the report follows. Ed.]

Results in Brief

Although it is widely considered an experimental therapy, manyhealth insurers are covering ABMT following high-dose chemotherapyfor breast cancer. The 12 insurers we spoke with said they basedtheir decision to cover the treatment on the preliminary clinicalevidence, but also on factors like fear of litigation and adversepublic relations.

The use of ABMT for breast cancer has increased rapidly in recentyears, from an estimated 522 patients in 1989 to an estimated4,000 in 1994. At least seven states now require insurers to coverABMT for breast cancer, and other states have such legislationunder consideration. Medicaid covers the treatment in some states,and OPM has required that all beneficiaries of the Federal EmployeesHealth Benefits Program be covered.

Despite its increased coverage and use, most experts say theydo not yet know whether ABMT for breast cancer is effective, andfor which patients, compared with conventional therapy. Randomizedclinical trials sponsored by NCI are expected to provide the mostdefinitive answers, but these will not be completed for severalyears. In the meantime, there have been sharp disagreements amongresearchers, physicians, NCI, insurers, and patients about theappropriate use of the treatment before definitive research resultsare available. At one end are those who argue that the preliminaryevidence supports a policy of widespread use and universal insurancecoverage of the treatment. At the other end are those who feelthat the treatment should largely be restricted to patients enrolledin randomized clinical trials until the treatment's effectivenesshas been clinically proven.

The NCI-sponsored trials have been slow to accrue patients. Manyexperts expressed concern to us that the wide availability ofABMT has impeded the ability to complete these randomized clinicaltrials, which require a control group of patients who receiveconventional therapy. There is also concern that a substantialportion of patients receiving ABMT are doing so outside of anyresearch setting, which may further slow down the effort to learnwhether the treatment is effective.

If ABMT is ultimately shown to be preferable to conventional therapyfor some groups of breast cancer patients, then those patientswill have benefited from the early diffusion of this technology.If it is not, however, then the widespread availability of thetreatment before its effectiveness has been established will meanthat many patients may have been unnecessarily subjected to anaggressive and toxic treatment. The diffusion of the treatmentalso has implications for health care costs: ABMT typically costsanywhere from $80,000 to over $150,000, compared with approximately$15,000 to $40,000 for conventional chemotherapy.


Breast cancer is the second leading cause of cancer deaths amongAmerican women. The American Cancer Society estimates that therewill be 184,300 new cases of breast cancer diagnosed in US womenin 1996 and that 44,300 women will die from the disease. One ineight women will develop breast cancer during her lifetime.

Breast cancer is generally classified into four main stages basedon the size of the tumor and the spread of the cancer at the timeof diagnosis. Mortality rates are strongly related to the stageof the disease at the time of detection. Stage I patients havean excellent chance of long-term survival, while stage IV (metastatic)breast cancer is usually fatal. A wide variety of treatments existsfor breast cancer patients, including surgery, chemotherapy, radiationtherapy, and hormone therapy. The particular treatments used dependon the stage and characteristics of the cancer and other aspectsof the patient and her health.

ABMT is a therapy that allows a patient to receive much higherdosages of chemotherapy than is ordinarily possible. Because high-dosechemotherapy is toxic to the bone marrow (which supports the immunesystem), methods have been developed for restoring the bone marrowby reinfusing stem cells (the bone marrow cells that mature intoblood cells) taken from the patient before chemotherapy. Stemcells are removed from the patient's blood or bone marrow, thenconcentrated, frozen, and sometimes purged in an attempt to removeany cancerous cells. The patient then undergoes chemotherapy atdosages 2 to 10 times the standard dosage. To restore the abilityto produce normal blood cells and fight infections, the patient'sconcentrated stem cells are thawed and reinfused after chemotherapy.When the transplant is done from the blood rather than the bonemarrow, the procedure is often referred to as peripheral bloodstem cell transplantation.[3]

ABMT is an expensive treatment although the cost per patient hasbeen falling in recent years. Aside from financial costs, thetreatment is usually very unpleasant for the patient and may posesignificant risks. The high doses of chemotherapy are very toxic,leading to treatment-related morbidity and mortality rates that,while declining, are still higher than for conventional chemotherapy.There may also be problems in restoring the patient's abilityto produce normal blood cells and thereby fight infections. ABMTis being evaluated in the treatment of a number of types of cancerother than breast cancer and is considered standard therapy fortreating certain types of leukemia and lymphoma under certainconditions.

Many clinical trials have been conducted to assess ABMT for breastcancer, but most of these studies have been phase I and phaseII trials, which most experts agree have been of limited use infirmly establishing the effectiveness of ABMT compared with conventionaltherapy.[4] NCI is currently sponsoring three randomized clinicaltrials that seek to determine whether ABMT is better than currentstandard therapy in comparable breast cancer patients. These trialsseek to ultimately involve a total of about 2,000 women at morethan 70 institutions around the country.

Use of ABMT for Breast Cancer Has Become Widespread Even ThoughIts Effectiveness Is Uncertain

Although most experts believe the clinical research has not yetestablished that ABMT is superior to conventional therapy, andfor which patients, insurance coverage of the treatment has becomerelatively common and use of the treatment is diffusing rapidly.

Use and Coverage of the Treatment Has Become Relativey Common

According to the Autologous Blood and Marrow Transplant Registry-NorthAmerica, the number of breast cancer patients receiving ABMT hasincreased rapidly, growing from an estimated 522 in 1989 to anestimated 4,000 in 1994.[5] About one-third of all

ABMTs reported to the Registry in 1992 were for breast cancer,making it the most common cancer being treated with this therapy.The Registry reports that although the treatment is most commonlyused in women with advanced disease, there is a growing trendto use it more frequently on patients with earlier stages of breastcancer. There has also been a dramatic increase in the numberof patients undergoing this treatment in Europe.

Many insurers, including some of the nation's largest, now routinelycover ABMT for breast cancer both inside and outside of clinicaltrials, although some still deny coverage for the treatment becausethey consider it experimental. One study looked at 533 breastcancer patients in clinical trials who requested coverage forABMT from 1989 through 1992. It found that 77% of them receivedapproval for cover-age of the treatment after their initial request.[6]

Treatment's Effectiveness Is Still Unknown

We reviewed the current medical literature and spoke with severalleading oncologists and technology assessment experts regardingABMT for breast cancer.[7] While there were differences of opinion,the consensus of most of the experts and the literature was thatcurrent data indicate ABMT may be beneficial for some breast cancerpatients but that there is not yet enough information to establishthat it is more effective than standard chemotherapy.

The medical literature includes several studies showing longerperiods before relapse and improved survival for some poor prognosis,high-risk breast cancer patients receiving ABMT rather than conventionaltherapy.[8] However, it is unclear whether the superior outcomesof patients receiving ABMT in these studies were the result ofthe treatment itself or the result of bias caused by the selectionof patients chosen to receive the treatment. Most of the medicalliterature and nearly all of the experts we spoke with said thatthe current data are not yet sufficient to make definitive conclusionsabout the effectiveness of ABMT and about which groups of breastcancer patients would be most likely to benefit. Although thereare wide differences of opinion about the appropriate use of ABMT,nearly all sides of the debate agree that the results of randomizedclinical trials are needed to provide definitive data on the treatment'seffectiveness.

Several studies have reviewed and analyzed the extensive medicalliterature related to ABMT for breast cancer. In 1995, ECRI, anindependent, nonprofit technology assessment organization, publishedan analysis stating that the weight of the evidence in the medicalliterature did not indicate greater overall survival for metastaticbreast cancer patients receiving ABMT compared with conventionaltherapy.[9] The Blue Cross and Blue Shield Association's TechnologyEvaluation Center, after reviewing the available data in 1994,concluded that the evidence was not yet sufficient to draw conclusionsabout the effectiveness of ABMT compared with conventional therapyfor breast cancer patients.[10] Similarly, NCI, at a congressionalhearing, said that while ABMT has shown promise in some clinicalstudies, the results of the NCI randomized clinical trials wereneeded before conclusions could be reached about whether and forwhom the treatment is more beneficial than conventional therapy.[11]

Insurers Cite Promising but Inconclusive Research and OutsidePressures in Covering ABMT

We interviewed the medical director, or another official who makescoverage decisions, at 12 US health insurance companies. We discussedthe insurer's coverage policies and the factors that influencedtheir coverage policy with regard to ABMT for breast cancer. Theinsurers' coverage policies regarding ABMT for breast cancer reflectedsome incongruity. In general, the insurers said they did not normallycover experimental or unproven treatments and that they believedABMT for breast cancer fell into this category. Yet, with somerestrictions, all 12 insurers nonetheless covered ABMT for breastcancer with only one requiring that patients enroll in clinicaltrials.

In explaining this, most cited as the primary influence the factthat although until recently the treatment had not been testedin randomized trials, it has become widely used and that the existingresearch suggests it may be beneficial to certain patients. Butinsurers told us that a variety of nonclinical factors also stronglyinfluenced their coverage policy, such as the threat of litigation,public relations concerns, and government mandates.

Insurers' Technology Assessment and Coverage Decision Process

All health insurers must decide whether and when they will covera new or experimental treatment. To do this, they engage in someform of technology assessment, a process that seeks to assessthe safety and effectiveness of a medical technology based onthe best available information. For the most part, health insurersdo not gather primary data but, rather, rely heavily on peer-reviewedmedical literature and on the assessment of experts inside andoutside of their companies.

Some large health insurers have elaborate technology assessmentunits. One example is the Technology Evaluation Center, a collaborationof the Blue Cross and Blue Shield Association and Kaiser Permanente.The Center's staff includes physicians, research scientists, andother experts who review and synthesize existing scientific evidenceto assess the safety and efficacy of specific medical technologies.The Center has published assessments for over 200 technologiessince 1985, including several for ABMT for breast cancer. Otherlarge insurers, including Aetna and Prudential, also have specialprograms that do formal assessments of specific technologies.Smaller insurers also do technology assessment, but on a smallerscale; for instance, they may have a small office that does literaturesearches or reviews the findings of larger technology assessmentorganizations.

Using their assessments, insurers then decide whether they willcover a particular treatment and under what conditions. Whateverthe overall policy, coverage of costly and complicated proceduresmay require special preapproval before they are covered. Amongthe insurers we spoke with, preapproval for ABMT was generallyrequired by the office of the medical director or some other officethat reviews claims for medical appropriateness. They said theywanted to ensure that a case meets any coverage restrictions andthat ABMT is medically appropriate for that particular patient.For certain difficult cases, some insurers also use an outsidepanel of experts, serving as a mediation service, to determinewhether ABMT is the appropriate treatment.

Promising Research, Widespread Use Cited for Coverage Policy

Seven of the 12 insurers we spoke with explicitly characterizedABMT for breast cancer as experimental. Four others did not specificallyterm the treatment "experimental" but nonetheless saidthat ABMT for breast cancer should not yet be considered standardtherapy since its effectiveness over conventional therapy hadnot yet been proven. One insurer did not express an opinion onthe issue.

Yet while the insurers said they typically do not cover experimentaltherapies, many said that in this case there was enough preliminaryevidence that ABMT may be effective to justify covering it. Sevenof the 12 insurers cited the clinical evidence as one of the primaryreasons that they decided to cover ABMT. These insurers said thatthe existing data indicate that ABMT may hold promise for certainbreast cancer patients and that flexibility was needed in payingfor experimental treatments for seriously or terminally ill patients.

Two insurers also said that they cover ABMT for breast cancersince, although its efficacy has not been established, it hasbecome generally accepted medical practice in that it has becomea common treatment for breast cancer throughout the United Statesand is covered by many other insurers. They said they would receivepressure from their beneficiaries if they were to deny coveragefor a treatment that other insurers cover.

Court Decisions and the Threat of Litigation Have InfluencedMany Insurers

While the medical evidence was an important factor in the coveragepolicy of a majority of the insurers, other factors were alsoclearly at work, with the threat of litigation being among themost important. When an insurer refuses to pay for a treatmentrequested by the patient or the patient's physician, coveragemay ultimately be decided in the court system. Over the past severalyears, many breast cancer patients have sued their insurers afterbeing denied coverage for ABMT.

Nine of the 12 insurers that we spoke with specifically mentionedlitigation, or the threat of litigation, as a factor in theirABMT coverage policy. For five of these insurers, legal concernswere characterized as among the most important reasons for choosingto cover ABMT for breast cancer. Before changing their policiesto cover ABMT for breast cancer, six of the insurers we spokewith had been sued after denying coverage for the treatment.

Overall, the insurers had not been very successful in these casesand had often either settled before judgment was rendered or hada judgment rendered against them. The insurers who had been suedon the issue said the financial costs of legal fees, settlements,and damages were high. For the most part, the insurers said theyfound different courts to be widely inconsistent in ruling whetherABMT is experimental and should be covered, a point also madein reviews of case law on the issue. In addition to the financialcosts, insurers said the lawsuits were harmful to their publicrelations. Publicity of their coverage policy led to the impressionthat they were denying a gravely ill patient a beneficial therapyfor economic reasons. The insurers we spoke with no longer facemany lawsuits on the issue since they now generally cover ABMT.

Court decisions on health insurance coverage disputes have usuallyturned on the language of the insurance contracts, which generallybar coverage for experimental treatments but are often ambiguouswith regard to what is defined as "experimental." Arecent review of such litigation noted that state courts havetended to favor policyholders in these coverage disputes, althoughfederal courts, where disputes for self-insured companies areoften decided, have been split on whether insurers must coverABMT for breast cancer.[12]

The courts, in ruling whether an insurer must provide coveragefor ABMT for breast cancer, have based their decisions on a numberof factors. These have included whether ABMT is generally acceptedin the medical community for the treatment of breast cancer, whether"experimental treatment" is defined clearly in the insurancepolicy, whether the treatment was intended primarily to benefitthe patient or to further medical research, and whether the insurer'sdenial of coverage was influenced by its own economic self-interest.This last argument was the focus of Fox v. Health Net of California,a highly publicized case in which a California jury awarded $89million in damages to a policyholder whose deceased wife had beendenied coverage of ABMT for breast cancer.[13] Plaintiffs in anumber of recent cases have alleged that denial of coverage forABMT constitutes discrimination against women in violation ofcivil rights laws or discrimination against a specific diseasein violation of the Americans With Disabilities Act. Most of thesecases are still pending.

Insurers have had some success in court as well. Some state courtshave ruled that ABMT is still widely considered to be experimentaland that the health insurance contract clearly precluded coverageof experimental treatments. Courts in at least three federal circuitshave also upheld insurers' coverage denials for ABMT to treatbreast cancer. Courts in many of these cases permitted insurerswide discretion in making coverage decisions as long as the decisionswere not arbitrary or capricious.

ABMT Coverage Is Mandated by At Least Seven States

The controversy over access to ABMT for breast cancer patientshas led several states to propose or enact legislation regardinginsurance coverage of the treatment. As of June 1995, at leastseven states had enacted legislation that, under certain parameters,requires that insurers provide coverage for ABMT for breast cancer.At least seven additional states have similar legislation pending.[14]Some of these laws are mandates requiring that coverage of ABMTfor breast cancer be part of any basic package of health insurance.Other laws simply require that the treatment be made availableas a coverage option, at perhaps a higher premium. The laws insix of the states require coverage whether or not the patientis enrolled in a clinical trial, while one state requires patientswith certain types of breast cancer to join well-designed randomizedor nonrandomized trials. Three of the 12 insurers we spoke withsaid they were required by a state mandate to cover ABMT for breastcancer for most of their beneficiaries. One of these three saidit would not cover the treatment if it were not for the mandate.

Those who advocate passage of the state laws argue that they arenecessary to make a promising therapy available to breast cancerpatients. Among the arguments used is that insurers classifiedABMT for breast cancer as "experimental" as much foreconomic as medical reasons because ABMT is an expensive treatment.Insurers respond that ABMT for breast cancer is an experimentaltreatment still being evaluated in clinical trials and they shouldnot be in the business of paying for research. Furthermore, insurerssay that legislation mandating coverage of specific treatmentsis a poor way to make medical policy and that it distorts themarket because self-funded plans are exempt from state mandates.

The National Association of Insurance Commissioners (NAIC) isconsidering a model act for states that would set minimum standardsof coverage for health insurers. The model act, which has notyet been approved by the full NAIC membership, would require insurersto cover an experimental treatment if the peer-reviewed medicalliterature has established that the treatment is an effectivealternative to conventional treatment. A representative from NAICtold us that in a state that passed such an act, insurers wouldnormally be required to cover ABMT for breast cancer if the treatingphysician considered it the medically appropriate treatment.

ABMT Coverage by Federally Funded Health Insurance Programs

Programs such as Medicaid, Medicare, and the Civilian Health andMedical Program of the Uniformed Services (CHAMPUS) have varyingpolicies regarding coverage of ABMT for breast cancer. Coveragecriteria for Medicaid, a jointly financed federal and state programthat provides medical care to the poor, varies by state, but somestates' Medicaid programs will cover ABMT for breast cancer underat least some circumstances. Of nine state Medicaid programs wecontacted, five provided coverage for ABMT for breast cancer.[15]The Medicare program, which provides health coverage primarilyfor the elderly, specifically excludes ABMT coverage for solidtumors such as breast cancer because the Health Care FinancingAdministration, which administers the Medicare program, considersthe treatment experimental. The practical impact of the Medicarepolicy is limited since the elderly are not normally appropriatecandidates for ABMT treatment. CHAMPUS, the Department of Defense'shealth care program for active duty and retired military personnel,and their dependents and survivors, considers ABMT for breastcancer experimental but provides coverage through a demonstrationproject in which beneficiaries may receive ABMT by enrolling inone of three NCI randomized clinical trials.

The Federal Employees Health Benefits Program (FEHBP), run byOPM, provides health insurance coverage for over 9 million federalemployees, retirees, and dependents through over 300 independenthealth plans. In September 1994, OPM imposed a requirement thatparticipating health insurers must cover ABMT for breast cancerfor all FEHBP beneficiaries both in and outside of clinical trials.OPM acknowledged to us that the evidence is mixed on the effectivenessof ABMT for breast cancer. They said they decided to mandate coveragelargely because so many insurers were already covering the procedureand they wanted to make the benefit uniform across all of theircarriers.

Insurers we spoke with said they complied with the OPM mandate,although they criticized the mandate as a political rather thanclinical decision. Two of the 12 insurers we spoke with specificallymentioned the OPM decision as having influenced their own coveragepolicy, largely because it brought so much publicity to the issue.

Physicians, Insurers, and Advocates Raise Concerns About WidespreadDiffusion of ABMT

Medical experts, insurers, and others have debated whether ABMThas become too widely used before there is convincing evidenceof its efficacy. While the medical community seeks to learn whetherABMT is more effective for some breast cancer patients than conventionalchemotherapy, the number of patients receiving the treatment andthe number of facilities providing it continue to grow. If ABMTwere a new drug, it would be restricted mostly to patients onclinical trials until its efficacy were established and the Foodand Drug Administration (FDA) had approved its use in generalmedical practice. Yet because ABMT is a procedure, rather thana drug, it does not require approval from FDA, making it easierfor it to be widely used while its effectiveness is still beingtested in clinical trials. The rapid diffusion of ABMT for breastcancer has implications for patient care, health care costs, andresearch.

Implications for Patient Care

There is debate over whether patients benefit from the rapid diffusionof a new technology that is still being tested in clinical trials.In the case of ABMT, the high doses of chemotherapy administeredin conjunction with the treatment can make it a particularly difficulttreatment for patients. This is evidenced both by the extremesickness and side effects that patients may experience and bythe higher rate of treatment mortality for ABMT than for conventionalchemotherapy. If the clinical research ultimately shows ABMT tobe preferable than conventional therapy for some groups of patients,then some of those patients will have benefited from the earlydiffusion of this technology. If it is shown not to be more effective,however, or if it is shown to be effective for a much smallersubset of patients than are currently being treated with the therapy,then many patients will have been unnecessarily subjected to anaggressive treatment that can be risky and produce many severeside effects.

In addition, while ABMT formerly was available only at a selectnumber of cancer research centers across the country, it is nowbeing performed by a rapidly growing number of smaller hospitalsand bone marrow transplant centers. Many physicians we talkedwith, including researchers and insurance company medical directors,expressed concerns that there may be some facilities that performtoo few transplants to ensure sufficient staff expertise or thatdo not have the infrastructure needed to support this complicatedprocedure. Partly to address these concerns, several medical societieshave developed guidelines that set out specific criteria for facilitiesthat perform bone marrow transplants.

Implications for Health Care Costs

ABMT is an expensive treatment, costing anywhere from $80,000to over $150,000 per treatment, depending on the drugs used, anymedical complications, and the length of hospital stay required.Conventional chemotherapy, by contrast, typically costs betweenabout $15,000 and $40,000. The cost of ABMT has been decreasingover the years and is expected to decrease further as the technologyis refined and becomes more common. Some medical centers havealready been able to reduce the cost of the procedure by offeringthe treatment on more of an outpatient basis.

While the cost per individual treatment is likely to decrease,total spending nationwide on the procedure is likely to increase.More patients in different stages of breast cancer are being treatedwith ABMT, a trend that is expected to continue. The fact thatABMT can be a highly profitable procedure for the institutionthat performs it, many experts say, has created further incentivefor the diffusion of the treatment.

Virtually all sides of the debate agree that ABMT is worth thecost if it is shown to be the best available treatment. But someworry that the research has not yet established which breast cancerpatients, if any, are likely to benefit from ABMT and that therapid diffusion of this costly treatment outside of research settingsbefore its effectiveness has been proven may not be the best useof health care resources.

Implications for Clinical Research

There is clear consensus among the scientific community that,if possible, the best way to compare the effectiveness of a newtreatment with conventional treatment is through randomized clinicaltrials. A randomized trial assigns patients either to a controlgroup receiving conventional treatment or to one or more experimentalgroups receiving the treatment being tested. Random allocationhelps ensure that differences in the outcome of the groups canbe attributed to differences in the treatment and not differencesin patient characteristics. In the case of ABMT, some expertshave argued that early research showing favorable results forABMT may have been due to the fact that the breast cancer patientsreceiving ABMT had more favorable characteristics than those whowere not receiving the treatment. NCI has three large-scale randomizedclinical trials ongoing to compare ABMT with conventional therapyfor breast cancer. These trials randomly assign patients who fitcertain criteria either to an experimental group that receivesABMT or to a control group that instead receives a more conventionalform of therapy.

NCI has had difficulty accruing enough patients to its randomizedtrials. Two of the three ongoing NCI trials are accruing patientsat about half the rate researchers originally anticipated, anda fourth trial was closed because of low enrollment. NCI expandedthe enrollment goal of the third trial to improve the statisticalpower of the results, and results from all three trials are notexpected until nearly the turn of the century. NCI says patientaccrual to the trials, although slow, appears to be progressingadequately, but many experts we spoke with questioned whetherthe NCI trials will ever be completed as planned.

Many medical experts believe that the wide availability of thetreatment is one reason researchers are having problems accruingpatients to the randomized trials. ABMT is now widely availableto many breast cancer patients either through other clinical trialsor outside of a research trial. Under most circumstances, insurersthat cover ABMT do not require that the patient enter a randomizedtrial, and many patients are reluctant to do so. Patients whobelieve ABMT is their best hope for survival may not be willingto enter a trial where they may be randomly assigned to a groupreceiving conventional chemotherapy. The ABMT Registry estimatesthat only about 5 percent of all breast cancer patients receivingABMT are enrolled in the randomized clinical trials.

Proponents of ABMT that we spoke with pointed out that most proceduresin common medical practice today have not been subjected to thestrict scrutiny of randomized trials and that this potentiallylifesaving therapy should not be withheld until the NCI trialsare completed many years from now. Other medical experts, insurers,and patient advocates we spoke with said that ABMT for breastcancer should only be available to patients enrolled in clinicaltrials, possibly only randomized trials. They argued that theproliferation of ABMT outside of randomized trials or outsideof any research setting at all-is making it difficult to gatherthe data necessary to assess whether and for whom ABMT may bea beneficial treatment.

A large number of clinical trials are being conducted on ABMTfor breast cancer apart from the NCI randomized trials. Many majorcancer research centers are conducting non randomized trials,and numerous clinical trials are also under way at smaller hospitalsand private transplant centers. Yet some experts have argued thatmany of these trials will contribute little useful informationbecause the study population is too small, the trial is not sufficientlywell-designed, or because the results will not be published. Theseexperts are concerned that the proliferation of smaller clinicaltrials may be diverting patients from larger clinical trials,including the NCI randomized clinical trials, that are more likelyto yield meaningful results about the effectiveness of ABMT forbreast cancer.

The controversy over ABMT has also highlighted the issue of theextent to which health insurers should pay for the costs of clinicalresearch. Clinical research in the United States has been financedprimarily by the federal government, private research institutions,the pharmaceutical industry, and insurers. Insurers have oftenpaid the patient care costs for certain clinical trials. But givenfederal funding constraints and other economic pressures, manyresearchers and other experts we spoke with believe that healthinsurers should assume the costs of more clinical trials, especiallythe patient care costs of well-designed trials that offer promisingtreatments in an advanced stage of testing. They say the insurerswould have to pay for patient care costs even if the patient werenot in a trial and that the trials will ultimately benefit everyoneby helping identify effective treatments. The insurance industry'sposition has been that insurers should pay only for standard medicalcare and that insurers should not be in the business of financingresearch. But insurers have made exceptions, especially for clinicaltrials involving promising treatments for patients with terminalillnesses. Many insurance industry officials we spoke with saidthey would be open to paying the costs of some clinical trialsfor promising treatments, as long as the costs were to be spreadequitably among all insurers and health providers, and as longas there were strict standards to ensure that the research beingfunded was of high quality.


The controversy over insurance coverage of ABMT for breast cancerillustrates several issues related to the dissemination and insurancecoverage of new technologies. The rapid diffusion of new, oftenexpensive, medical technologies puts in conflict several goalsof the US health care system: access to the best available care,the ability to control health care costs, and the ability to conductresearch adequate to asess the efficacy of a new treatment.

Specifically, the ABMT controversy illustrates the challenge healthinsurers in the United States face in determining whether andwhen to provide coverage for a new technology of unknown efficacy,given the decentralized process for assessing new medical technologies.Insurers have less clear direction regarding coverage of medicalprocedures than they do for drugs because of FDA's role in drugapproval. Insurers thus have wide discretion, and little nationwideguidance, in determining whether and when a medical procedureshould no longer be considered "experimental" and shouldbe covered. The result can be great disparity in the coveragepolicies of insurers, with coverage decisions being influencednot just by the medical data and clinical judgments, but alsoby factors such as lawsuits and public relations concerns.

Furthermore, the lack of a systematic process for the disseminationof new technologies in the United States raises issues for thehealth care system. Those who advocate widespread access to experimentaltechnologies argue that patients should not be denied access topromising therapies, especially when clinical trials for thosetherapies may take many years. Those who advocate restrictingaccess to new technologies argue that the rapid diffusion of anew treatment before its effectiveness has been definitively provenis not ultimately beneficial to patient care, may waste resources,and may impede controlled research on the treatment.

Agency Comments

NIH provided us with comments on a draft of this report. Theyagreed with the conclusions and stated that the report presenteda balanced, thoughtful discussion of the controversial issues.NIH also noted that in the past, many insurers provided coverageonly in the context of clinical trials, but this became untenablebecause of the factors discussed in the report, particularly theOPM decision to require FEHBP coverage of the treatment both insideand outside of clinical trials. NIH also recommended some technicalchanges, which we incorporated in the report where appropriate.(See page 1348 for a copy of the NIH comments.)

OPM also reviewed the draft report and provided comments regardingthe decision to require that all FEHBP health insurance plansprovide coverage for ABMT for breast cancer. Their comments reemphasizedthat (1) many FEHBP plans were already providing this coverage;(2) the OPM decision was based on a desire to broaden coverageto all FEHBP enrollees; and (3) each plan retains the flexibilityto determine when and how the treatment will be covered, but plansthat limit coverage to patients enrolled in clinical trials haveto offer coverage in nonrandomized as well as randomized trials.(See page 1354 for a copy of OPM'S comments.)


1. The 12 health insurers were Aetna Health Plans, Anthem HealthPlan of Florida, Blue Cross and Blue Shield of Oregon, CNA Insurance,Harvard Pilgrim Health Care, HealthGuard of Lancaster, HealthPartners,Kaiser Permanente, Mutual of Omaha, Prudential HealthCare Group,United HealthCare (formerly Meta Health), and United HealthCareof Ohio.

2. Also, Martin S. Tallman, md, Assistant Professor of Medicineat Northwestern University Medical School, assisted us by providinga technical review of a draft of this report.

3. In this report we refer to autologous bone marrow transplantation(ABMT), which is commonly used to refer to all autologous stemcell rescue, whether the transplant is of the bone marrow or theperipheral stem cells of the blood. "Autologous" transplantsrefer to transplants through which patients receive their ownmarrow or peripheral stem cells. "Allogeneic" transplantsrefer to transplants through which patients receive marrow donatedby another person. This report addresses only autologous transplantsfor breast cancer.

4. A clinical trial is a medical experiment in which proceduresor drugs are tested on human subjects to assess their safety oreffectiveness. Phase I trials are designed to determine the dosethat can be given with an acceptable level of toxicity. PhaseII trials seek to evaluate the response in specific tumor types.Phase III trials seek to assess a treatment's effectiveness bycomparing patients receiving the experimental treatment with patientsreceiving a conventional treatment In a randomized phase III trial,patients are randomly assigned either to a control group receivingstandard treatment or to one or more experimental groups receivingthe treatment being tested.

5. The Autologous Blood and Marrow Transplant Registry-North America,begun in 1991, collects treatment information on ABMT recipientsfrom 128 participating centers, primarily in North and South America.The database is used to help identify trends in the use and outcomesof ABMT for several types of cancer, including breast cancer.The estimates of ABMT use given here extrapolate from Registrydata, which the Registry estimates represent about half of allbreast cancer ABMTs in the United States.

6. WP Peters, MC Rogers: Variation in approval by insurance companiesof coverage for autologous bone marrow transplantation for breastcancer, N Engl J Med 330(7):473-477, 1994.

7. Some of the oncology experts we spoke with included Karen Antman,md, Columbia University; Lois Ayash, md, the Dana Farber CancerInstitute; Craig Henderson, md, the University of California,San Francisco; Roy Jones, md, the University of Colorado CancerCenter, William Peters, md, the Karmanos Cancer Institute; EdwardStadtmauer, md, the University of Pennsylvania; and James Vredenburgh,md, Duke University Medical Center. We also spoke with researchersat other large and small cancer centers around the country. Technologyassessment experts we consulted included Naomi Aronson, phd, theBlue Cross and Blue Shield Association's Technology EvaluationCenter Jeffrey Lerner, phd, ECRI; and William McGivney, phd, Aetna.

8. For example, WP Peters, M Ross, JJ Vredenburgh, and others:High-dose chemotherapy and autologous bone marrow support as consolidationafter standard-dose adjuvant therapy for high-risk primary breastcancer. J Clin Oncol 11(6):1132-1143, 1993.

9. ECRI, High-Dose Chemotherapy With Autologous Bone Marrow Transplantationand/or Blood Cell Transplantation for the Treatment of MetastaticBreast Cancer (Plymouth Meeting, Pa.: Health Technology AssessmentService, ECRI 1995).

10. Blue Cross and Blue Shield Association Technology EvaluationCenter, High-Dose Chemotherapy With Autologous Stem Cell Supportin the Treatment of Breast Cancer, TEC Assessments, Vol. 9, No.33 (Chicago: Blue Cross and Blue Shield Association TechnologyEvaluation Center, Assessment Program, Nov. 1994).

11. Autologous Bone Marrow Transplantation as a Treatment forBreast Cancer, statement of Dr. Bruce Cheson, Head of the MedicineSection, Clinical Investigations Branch, Division of Cancer Treatment,NCI, NIH, Department of Health and Human Services, before theHouse of Representatives, Committee on Post Office and Civil Service,Subcommittee on Compensation and Employee Benefits (Aug. 11, 1994).

12. Denise S. Wolf, "Who Should Pay for 'Experimental' Treatments?Breast Cancer Patients v. Their Insurers," American UniversityLaw Review, Vol. 44, No. 5 (June l995), pp. 2029-2107.

13. Case No. 219692 (California Superior Court, Dec. 28. 1993).

14. The states with legislation enacted were Florida, Georgia,Massachusetts, Minnesota, New Hampshire, Rhode Island, and Virginia.The states where similar legislation is known to have been introducedare California, Connecticut, Louisiana, Missouri, New Jersey,New York, and Ohio.

15. Florida, Georgia, Massachusetts, New Jersey, and Texas providedcoverage, while Minnesota, New Hampshire, Tennessee, and Virginiadid not.