This case, filed under the Employee Retirement Income Security Act ("ERISA"), 29 U.S.C. Sec. 1001 et seq., involves the denial of a claim for health care benefits for the treatment of cancer.

David A. Hendricks of Greenville County, South Carolina, was diagnosed with small cell lung cancer in November 1993. Small cell lung cancer is an aggressive carcinoma of the lung and is almost always fatal within three to six months if left untreated. Upon the recommendation of his doctor, Hendricks elected to participate in a clinical trial of a multiphase treatment, known as "high-dose chemotherapy with peripheral stem cell rescue," which his doctor believed was his best chance for survival.

As part of his employer's welfare benefit plan, Hendricks received health care coverage under a policy issued by Central Reserve Life Insurance Company. When Hendricks presented this proposed treatment to Central Reserve for preapproval, Central Reserve advised him that the terms of the policy did not cover the proposed treatment because it was "experimental/investigative."

Hendricks filed suit against Central Reserve under ERISA, seeking an injunction ordering Central Reserve to provide benefits for this cancer treatment. Following a bench trial, the district court made findings of fact and conclusions of law that the proposed treatment was not covered by the Central Reserve policy. From the judgment entered in favor of Central Reserve, Hendricks appealed.

Hendricks assigns three errors. First, he asserts that the district court should not have relied upon detailed definitions of the terms "experimental" and "investigative" contained in the official employee welfare benefit plan documents on file with Central Reserve, but instead should have relied upon generally accepted definitions of those terms, because the summary plan description provided to him did not contain any definitions of "experimental" or "investigative." Second, he asserts that the district court improperly ruled that the proposed cancer treatment at issue is "investigative" or "experimental" and that we should review the court's findings on this issue de novo. Finally, he argues that even if we were to find that some phases of the proposed treatment were excluded from coverage of the Central Reserve policy, coverage should nevertheless have been provided for other phases of the multiphase treatment under our holding in Doe v. Group Hospitalization & Medical Services, 3 F.3d 80 (4th Cir.1993).

For the reasons that follow, we affirm.

* When David Hendricks, at age 41, was diagnosed with small cell lung cancer, he was given standard-dose chemotherapy in conjunction with localized radiation therapy, which proved effective and resulted in a complete response. Central Reserve provided benefits under its policy for this treatment. Dr. Jeffrey K. Giguere, Hendricks' treating physician, however, stated that notwithstanding the immediate success of the standard-dose chemotherapy in the treatment of this disease, two-year survivors are few, and only three to eight percent of the persons with the disease survive more than five years. Based on an April 1993 study relating to the results of high-dose chemotherapy, Dr. Giguere believed that Hendricks' survival potential could be improved with a multiphase treatment involving high-dose chemotherapy, combined with peripheral stem cell rescue. Dr. Giguere recommended that Hendricks partake in a clinical trial being conducted by Dr. Giguere's firm, Response Technologies, Inc.

Clinical trials or investigations are regulated by the Food and Drug Administration and the Department of Health and Human Services. When one is conducted, a protocol document is required to standardize the procedure. The protocol document outlines in detail the treatment planned and the manner in which data is to be collected for the investigation. The regulations require that the investigators advise each patient participating in a clinical trial of the nature of the treatment and the risks.

The objective of the clinical trial proposed by Dr. Giguere for the treatment of small cell lung cancer with high-dose chemotherapy was "[t]o determine the impact of [the treatment] on duration of complete response and overall survival in patients with limited stage small cell carcinoma of the lung in complete remission or near-complete remission following standard [chemotherapy]." In accordance with the regulations governing clinical investigations, Hendricks executed an informed consent form in January 1994 which advised him of the nature of the treatment and the risks. In the consent form, he agreed that the treatment is "being explored as a new treatment" and that:

Although it is hoped that this treatment and study will be of benefit or that it will help others, it cannot be said that it will help me directly. I may expect to benefit from this project to the extent that any treatment I receive helps to manage my condition. There is no assurance, however, that participating will result in any improvement. I may also receive personal satisfaction from the knowledge that my participation in this project has contributed to the advancement of science.

The protocol for the proposed treatment utilizing high-dose chemotherapy outlines a number of phases. The first is the "induction phase" which qualifies the patient for the proposed high-dose chemotherapy and consists of standard-dose chemotherapy to determine whether the patient responds positively. Only a patient's positive response to standard-dose chemotherapy could indicate any success with respect to high-dose chemotherapy. In the second phase, "mobilization," the dosage level of chemotherapy is raised in order to mobilize "stem cells" from the bone marrow into peripheral blood. Stem cells are generally found in the bone marrow, and they can develop into white blood cells, which are the body's mechanisms for fighting infection. Mobilization is done in preparation for leukapheresis, during which blood is taken out of the body and centrifuged to separate and "rescue" the stem cells, and the remainder of the blood is returned to the body. The stem cells are maintained in a cold environment so that they remain viable for later return to the bloodstream after high-dose chemotherapy.

The next phase, the actual high-dose chemotherapy, is a radical attack on the lung cancer itself. Mobilization and leukapheresis are necessary precursors to high-dose chemotherapy since the high level of chemotherapy kills not only the cancer but also the body's white blood cells, leaving the body largely defenseless against infections. Following the high-dose chemotherapy, the preserved stem cells are reinfused into the bloodstream with the hope that they will grow into healthy white blood cells.

The final phase consists of a two-week hospitalization period, during which the patient is monitored for inevitable infections while the stem cells are growing into white blood cells.

In December 1993, Hendricks' doctors wrote Central Reserve for preauthorization of this proposed multiphase high-dose chemotherapy treatment and a determination of benefits under the policy issued by it. In February 1994, Central Reserve wrote Hendricks' doctors and advised them that the proposed treatment is considered "experimental/investigational" and that the Central Reserve policy expressly excluded coverage for treatments that are experimental or investigational in nature. After further urging from Hendricks' doctors and his attorney, Central Reserve conducted a second administrative review of the denial of Hendricks' request for preapproval and concluded by reaffirming its decision to deny coverage for the proposed procedure. Central Reserve observed that "[b]oth the protocol and the informed-consent form signed by Mr. Hendricks state that he is going to be treated as a patient in a research study" and that "the treatment may not benefit him." The letter from Central Reserve concluded:

It appears that the primary focus of the studies is to determine whether High-Dose Chemotherapy will make patients live longer. Importantly, the only randomized study which compared High-Dose therapy with standard therapy which you sent to us indicates that the researchers found no statistically significant difference in long-term survival for patients undergoing High-Dose Chemotherapy over those that received standard treatment. High-Dose Chemotherapy has simply not been proven to be a safe and effective treatment for small-cell lung cancer.

While the language of the Central Reserve policy provides benefits for radiation therapy and chemotherapy treatment, the policy excludes any charges:

For treatment or services experimental or investigational in nature....

Which are not necessary to the care or treatment of an illness....

For treatment or services which are not generally accepted medical practices in the United States for a given illness....

Hendricks filed suit in April 1994 and an expedited bench trial was held on May 2, 1994. After hearing all the evidence, the district court ruled:

The court finds that the evidence establishes that HDC/PSCR [high-dose chemotherapy with peripheral stem cell rescue] for small cell lung cancer is an experimental or investigational rather than an accepted standard of practice. The protocol, the consent form, the medical evidence received each described the procedure in terms which emphasized its experimental or investigational nature. The purpose for which the plaintiff is being treated is to either cure or give him relief from his condition or prolong his life, and in utilizing HDC/PSCR, there is no evidence to this point that that will be accomplished.

The court also found that the treatment was not medically necessary as defined by the policy since the "efficacy of this procedure is undetermined, and it is not possible to know whether the procedure could be omitted without adversely affecting plaintiff's condition or the quality of medical care."

Hendricks first contends that the district court, in ruling on coverage for his treatment, improperly relied upon definitions of the terms "experimental" and "investigative" in the official plan document maintained in Central Reserve's files and did not rest its decision on the terms of the summary plan description circulated to employees. The summary plan description provides no definition for "experimental" or "investigative," whereas the plan itself defines "experimental treatment" as "treatment that is performed under controlled conditions in order to discover an unknown effect or to test a theory," and "investigative treatment" as "treatment performed when the service, procedure, drug or treatment has limited human application but has not achieved general acceptance in medicine."

A summary plan description must be provided to participants and beneficiaries of the plan and must be "written in a manner calculated to be understood by the average plan participant, and shall be sufficiently accurate and comprehensive to reasonably apprise such participants and beneficiaries of their rights and obligations under the plan." 29 U.S.C. Sec. 1022(a)(1). Because the summary plan description in this case contains no definitions of "experimental" or "investigative," Hendricks argues that the district court should have relied upon the common understandings of these terms. If it had done so, he maintains, the court would have been more likely to rule in his favor and provide coverage for his treatment.

In Aiken v. Policy Management Systems Corp., 13 F.3d 138 (4th Cir.1993), we held that " 'if there was a conflict between the complexities of the plan's language and the simple language of the [summary plan description], the latter would control' " if the participant relied on the summary plan description or was prejudiced by it. Id. at 140-41 (quoting Pierce v. Security Trust Life Ins. Co., 979 F.2d 23, 27 (4th Cir.1992)). See also Fuller v. FMC Corp., 4 F.3d 255, 262 (4th Cir.1993). The reason for this principle is drawn from the reality that the summary plan description is " 'the statutorily established means of informing participants of the terms of the plan and its benefits,' and the 'employee's primary source of information regarding employment benefits.' " Aiken, 13 F.3d at 140 (quoting Pierce, 979 F.2d at 27). In Aiken, for example, the summary plan description allowed a retiring employee to obtain a vested interest in a pension account if he had completed 20 years of service with the firm, while the official plan document required 20 years of service and the attainment of age 60. With this type of conflict, we ruled that the summary plan description controls. 13 F.3d at 142. Where, however, the summary plan description and the plan itself do not conflict, our cases provide no prohibition against review of the official plan itself for a fuller understanding of the plan's terms. Indeed, in those circumstances the plan is the controlling document for determining the scope of benefits provided.

Thus, in this case, if we were to find that the generally accepted definitions of "experimental" and "investigative" as used in the summary plan description differed substantially from the definitions of those terms given in the official plan document, the resulting conflict might require us to apply our Aiken line of cases. But, in consulting the official plan document to determine whether its definitions allow for Hendricks' proposed treatment, the district court concluded that the plan's definitions are not in conflict with commonly understood meanings for the terms:

Whether the court utilizes the plan itself or just the summary, it makes no difference, because if only the summary is used, the ordinary ... meaning of those terms would be utilized, and they are not inconsistent with the plan document itself.

* * * * * *

[T]he definition in the plan document is the same as the ordinary meaning of those two terms which are under review.

These conclusions by the district court are confirmed by comparing the definitions given in the plan document with definitions obtained from any current dictionary. The plan defines "experimental treatment" as a treatment performed "under controlled conditions in order to discover an unknown effect or to test a theory." This is substantively identical to Webster's Ninth New College Dictionary 's definition of "experiment," i.e., "an operation carried out under controlled conditions in order to discover an unknown effect or law, to test or establish a hypothesis, or to illustrate a known law." Webster's Ninth New College Dictionary, 437 (1988). See also Random House Dictionary of the English Language, 681 (2d ed. unabridged 1987). The plan defines "investigative treatment" as that performed "when the service, procedure, drug or treatment has limited human application but has not achieved general acceptance in medicine." The Random House Dictionary states similarly that an "investigative new drug" is "a regulatory classification assigned by the U.S. Food and Drug Administration to an unproven drug, allowing its use in approved studies with human patients." Id. at 1004.

We agree with the district court that the definitions of "experimental" and "investigative" given in the official plan document are not so different from common understandings of the terms that they can be thought to conflict with those understandings. Accordingly, we find no error in the district court's reliance on the definitions given in the plan document, even though these definitions were not provided in the summary plan description.

Hendricks next contends that the district court erred in concluding that high-dose chemotherapy with peripheral stem cell rescue for small cell lung cancer was experimental or investigative and thus not covered by the Central Reserve policy. Whether a form of treatment is experimental or investigative can be answered on at least two levels. When resolution of the issue depends on an interpretation of plan documents, we review the district court's conclusions de novo since contract interpretation is a question of law. Where a case turns simply upon a reading of the document itself, there is no reason to believe that a district court is in any better position to decide the issue than is an appellate court. When, however, a ruling depends upon an evaluation of evidence extrinsic to the contract, the standard of review is more deferential, and the appellate court reviews only for clear error. See United States v. Pollard, 959 F.2d 1011, 1023 n. 6 (D.C.Cir.), cert. denied, --- U.S. ----, 113 S. Ct. 322, 121 L. Ed. 2d 242 (1992); Guidry v. Halliburton Geophysical Services, Inc., 976 F.2d 938, 940 (5th Cir.1992). Thus, when the district court bases its determination of whether a medical procedure meets a contract's definition on facts presented and the opinions of experts, the question is one of fact and appellate courts will not reverse the findings unless they are found to be clearly erroneous. Federal Rule of Civil Procedure 52(a) provides that findings of fact by the district court "shall not be set aside unless clearly erroneous, and due regard shall be given to the opportunity of the trial court to judge of the credibility of the witnesses." See also 9 Charles A. Wright & Arthur R. Miller, Federal Practice and Procedure Sec. 2588 (1971 & Supp.1994).

In this case the district court considered whether Hendricks' proposed treatment was experimental or investigative both by construing the terms of the Central Reserve policy and by determining whether, as a factual matter, the proposed treatment of Hendricks fell within those terms. As already discussed in Part II, the legal definitions relied on by the district court were derived from both the official plan document and the commonly accepted meanings of the terms. Thus, the plan excludes a treatment if it "is performed under controlled conditions in order to discover an unknown effect" or if the treatment "has not achieved general acceptance in medicine." Having found no error in the district court's conclusions in this regard, we proceed to address whether its factual findings were clearly erroneous.

At trial, the court was presented with conflicting expert testimony as to the results of a number of complex medical studies and as to the likely effect of high-dose chemotherapy upon small cell lung cancer. Evaluating the credibility of experts and the value of their opinions is a function best committed to the district courts, and one to which appellate courts must defer. An appellate court should be especially reluctant to set aside a finding based on the trial court's evaluation of conflicting expert testimony. See id. at Sec. 2586.

The studies evaluated by the district court, while expressing some optimism about the future prospects of high-dose chemotherapy for attacking small cell lung cancer, were nevertheless decidedly guarded in recommending the procedure. The most relevant study, one conducted by Dr. Yves Humblet, et al., and reported in 1987, concluded that "overall survival was not significantly improved" by high-dose chemotherapy. While trial testimony from the plaintiff's experts highlighted the increase in short term response noted in the studies and established that high-dose chemotherapy is at least as effective as standard-dose chemotherapy, expert testimony for the defendant supported the conclusion that the proposed treatment was still investigative. In particular, Dr. William Gradishar, a professor of medicine at Northwestern University, testified that based upon the trial protocol, the medical literature, and his own knowledge, high-dose chemotherapy for Hendricks would be considered investigational. In his deposition, Dr. Gradishar summarized the medical literature on the subject as follows:

[T]here's been variability in the results, but the common theme is that high dose chemotherapy, whatever your means of rescue, peripheral stem cell, harvested stem cells or autologous bone marrow that's given back as a rescue, you can get higher response rates in some cases using higher doses of chemotherapy but ... the vast majority [of patients] follow the same time course of their disease, follow[ ] the same time course as patients treated with conventional therapy. The bottom line is that there hasn't been a survival advantage that's been demonstrated by using this approach.

(Emphasis added).

The view taken by the parties toward the treatment is also revealing. Dr. Giguere proposed the multiphase high-dose chemotherapy treatment of Hendricks' cancer as a "clinical trial," which required, under federal regulations, a protocol and an informed consent describing the experimental nature of the treatment. The purpose of the clinical study, as stated in the protocol, was to determine whether high-dose chemotherapy, described as a "new treatment," improved the duration of a complete response to standard chemotherapy treatment and increased survival rates. In the consent form for the new treatment, Hendricks acknowledged that the treatment may not help him and that his only benefit may be contributing to the "advancement of science." It also appeared that Hendricks would be the first patient in South Carolina with small cell lung cancer to receive the high-dose chemotherapy treatment.

While the components of the treatment proposed are fairly well known and data demonstrate that high-dose chemotherapy generally has a tendency to improve a patient's response to cancer, the data failed to demonstrate that small cell lung cancer would be more effectively treated with high-dose chemotherapy than with standard chemotherapy.

These facts tend to show that the treatment is performed "under controlled conditions in order to discover an unknown effect" and to support the conclusion that the treatment "has not achieved general acceptance in medicine." In these circumstances, we cannot conclude that the district court was clearly erroneous in its factual determination that the treatment proposed for Hendricks was experimental or investigative.

Finally, Hendricks contends that even if high-dose chemotherapy has not been medically accepted for the treatment of small cell lung cancer, he should be entitled to proceed with it and have the policy construed to cover other phases of the proposed treatment, separate and apart from the high-dose chemotherapy. He argues that each of the other phases of the procedure, considered independently, are standard medical practices which unquestionably would be covered in other contexts. For instance, Hendricks argues, the "mobilization" phase is not significantly different from standard-dose chemotherapy, which the parties agree is covered under the policy, and there can be no doubt but that the policy covers the two-week hospitalization phase to monitor infection for a person with a dramatically low white blood cell count. The district court, however, rejected Hendricks' arguments, focusing on the treatment as a whole. The court stated:

The flaw in Hendricks' analysis is in looking at his treatment as phases as opposed to looking at it as a whole. While splitting the treatment into phases may aid the doctors administering the treatment or gathering data from it, it does not assist the court in determining whether benefits exist under the policy. Hendricks' entire treatment is investigative. Therefore, the entire treatment is excluded from coverage under the policy. Those elements of the treatment that would otherwise be covered are, in this case, being incorporated into an investigative procedure. Each of Hendricks' four phases of treatment is an integral part of the investigative HDC/PSCR treatment.

Following this analysis, the court denied coverage of the treatment as a whole, declining to rule that coverage existed for certain phases of the treatment. We agree with the district court's analysis.

To fragment the phases of treatment and consider each in light of the policy language produces an unrealistic and distorted analysis. Thus, while it is true that the policy would provide benefits to a person who requires hospitalization for a low white blood cell count for the hospital expenses incurred, we note that, but for the high-dose chemotherapy, Hendricks would not have required hospitalization for that reason. All the evidence points to the existence of an otherwise normal white blood cell count in Hendricks. He would require such a hospital stay only if he proceeded with the high-dose chemotherapy with peripheral stem cell rescue, a process that destroys white blood cells. In that context, the low white blood cell count that Hendricks would experience would be a complication resulting from treatment not covered by the policy, i.e. high-dose chemotherapy for small cell lung cancer. Similarly, although the policy might independently cover the higher level of chemotherapy called for in the mobilization phase, that level would be necessary only for the purpose of harvesting stem cells in preparation for the disputed treatment. In the absence of the disputed treatment, the policy would exclude mobilization as a charge "not necessary" to the treatment of an illness. Since other phases of the multiphase treatment would not be medically necessary in the absence of high-dose chemotherapy which is not a medically accepted treatment for small cell lung cancer, the other phases would be excluded from coverage either as complications from a non-covered treatment or as medically unnecessary. Thus, even if a fragmented analysis of the treatment were undertaken, the phases other than high-dose chemotherapy for which Hendricks seeks benefits would still fall outside the coverage of the policy because of several of the policy's exclusions.

A fragmented analysis, however, is not appropriate here because according to the policy the scope of the exclusion is defined by the scope of the excluded treatment. Thus if the experimental treatment is high-dose chemotherapy with peripheral stem cell rescue for small cell lung cancer, then it is that entire treatment which is excluded, not just certain phases or aspects of it. If we were to accept the logic of a fragmented analysis, virtually any treatment, no matter how inconsistent with good medical practice, could be broken down into component parts, with a great majority of them covered under a policy like the one at issue here. We have no reason to conclude that the parties to the policy or the employee welfare benefit plan intended this approach.

In pressing his point on appeal, Hendricks relies heavily on our decision in Doe v. Group Hospitalization & Medical Services, 3 F.3d 80 (4th Cir.1993). In Doe, we addressed the question of whether the policy in question covered any or all of the costs associated with high-dose chemotherapy treatment. Plaintiff was stricken with multiple myeloma, a form of blood cancer, and his doctor recommended high-dose chemotherapy in conjunction with an autologous bone marrow transplant. The policy specifically covered "chemotherapy for treatment of a malignant condition," but it also excluded coverage for autologous bone marrow transplants to attack this condition, along with "services or supplies for or related to" such transplants. 3 F.3d at 87, 88. We concluded that the exclusion of bone marrow transplant and related services should not mandate exclusion of the chemotherapy, as the chemotherapy was specifically covered by the policy.

The holding of Doe does not apply here to provide coverage for phases of the high-dose chemotherapy treatment other than the high dose chemotherapy phase itself because, unlike Doe, the exclusion in this policy is defined by the entire treatment and not by a particular phase of it. In Doe, only the autologous bone marrow transplant procedure and the services and supplies relating thereto were excluded by the language of the policy. High-dose chemotherapy was not an excluded treatment. In the case before us, the exclusion applies to any treatment that is "experimental" or "investigative." Since the treatment which is experimental or investigative includes the high-dose chemotherapy as well as the preparational and recovery phases of the treatment, the scope of the exclusion here is broader.

For the foregoing reasons, the judgment of the district court is

AFFIRMED.