Diskography: Science and the Ad Hoc Hypothesis

Foreword

Two editorials were sought on the subject of diskography based on the paper by Schellhas et al in this issue of the AJNR. The suspicion was that there would be divergent opinions concerning the value of diskography in general and on its value at the C2–C3 level in particular. The reliance that many spine surgeons place on diskography should make us aware of the potential value of diskography in certain well-defined clinical situations. The journal encourages the submission of papers on this subject, particularly those that are evidence-based and deal with clinical outcomes.

Robert M. Quencer, MD,Editor-in-Chief

In this issue of the AJNR (page 269), Schellhass et al prompt this reader to reflect on the nature of science, and the concept of the hypothesis. In many ways, one has to admire this article as a technical tour de force. Nevertheless, it builds upon a hypothesis that has never been accepted. Even proponents of diskography will admit that instead of being evaluated and proven under strict scientific guidelines, such as those that apply to new drugs, diskography was popularized and adopted before its validity was determined (1).

Any proposed explanation in science is regarded initially as a hypothesis that is more or less probable on the basis of the available facts and relevant evidence. As a hypothesis, the question of its truth or falsehood is open to scrutiny, and there should be a continual search for more and more evidence to decide the question. The hypothesis in this case is the authors' stated claim that “cervical diskography is a clinically useful test.” It is certainly likely that the disk, under certain conditions of derangement, is a source of a patient's pain. This is supported by the observation that certain patients have reproduction of their symptom complex after disk stimulation. The question that arises, however, is whether this information is enough to support the entire hypothesis. In particular, whether the information available supports diskography as a “clinically useful test.”

Unfortunately, this question of clinical usefulness is mired in controversy with both pro and con forces disparaging the scientific methods of the opposing camp. To date, both the critics and proponents of diskography have attempted to advance their positions by methodologically flawed scientific studies or editorial position papers. These publications have biased entry criteria and cohorts, and lack reasonable control groups. The end results are therefore difficult, if not impossible, to generalize.

Why is a diagnostic test performed in the first place? The purpose of a diagnostic test is twofold; it must provide reliable information about a patient's condition and influence the physician's plan for managing the patient

Concerning the first purpose, a diagnostic test, the following is known: 1) the morphologic characteristics of a disk, as revealed with diskography, is essentially irrelevant (1); 2) the negative studies by Holt (2) are sufficiently flawed as to be of no scientific value; and 3) the study by Walsh (3) suggests that the false-positive rate for a positive painful response to stimulation of the disk is very low. Normal disks have been shown not to produce pain with stimulation. Abnormal disks will produce pain with stimulation; however, not all abnormal disks are painful when stimulated. To employ a diagnostic test successfully, one needs to define the clinical problem in such a way that we understand the characterization of the patient population, the natural history of the disease process, including prevalence, incidence, and behavior over time, as well as the various diagnostic therapeutic options available. This underscores the complex morass that diskography seeks to inhabit; ie, the concept of “internal disk disruption” as a symptom-producing complex. Understanding the disease or symptom-producing complex purported to be the stimulus for diskography is a daunting challenge. We do not understand the natural history of this putative disorder, whether it always results in pain, whether it is one entity or many, or whether it represents a surgical objective.

There is tremendous pressure on caregivers to provide an explanation for patient symptoms, because patients and physicians are more comfortable with clear-cut relationships between symptoms and disease, cause and effect. There is also a sense that the more clearly defined the pathophysiologic process of the disease is, the better use is made of the available treatment options. It is contended that diskography is an informational tool only, a test designed to obtain information about the source of a patient's pain. How that information is used or abused is not the responsibility of the proponents of the test. Accepting diskography as a valid diagnostic tool, however, may lull surgeons into considering surgical treatment for “discogenic pain” and potentially may lead to inappropriate surgery. This brings us to the second purpose of a diagnostic test—to influence the physician's plan for managing the patient.

A diagnostic test that has no impact on treatment choice is unlikely to benefit the patient, except through the reassurance of the physician. Others would contend that knowledge about a disease, even in the absence of effective treatment options, has been described as an important part of the healing process (4). In the case of diskography, however, these assertions clearly need to have some type of utility assessment before they can be accepted.

Let us return to the most important question inherent in accepting the hypothesis that diskography is clinically useful. Does it affect the choice of treatment so that it has a positive impact on patient outcome? Proponents would argue that patient management may be improved by excluding invasive therapy such as surgery if one finds multiple painful disks or obtains indeterminate results. This argument presupposes that the surgical intervention that is held in the balance has a proven efficacy, which is not the case.

Given that 1) the symptom complex and source of disease are poorly understood; 2) the treatment techniques vary widely in their theoretical approach and efficacy; and 3) the diagnostic test that serves as a cornerstone in the decision-making process has not been subjected to well-controlled studies, one must conclude that the use of diskography is predicated on flawed logic and science, and vulnerable to abuse and misuse. I do not believe it is worthwhile to pursue diskography as an “informational tool” for the purposes of establishing a diagnosis for which there is no proven therapy. In fact, any use of diskography as an informational tool (outside of well-controlled and designed trials to establish basic efficacy) must be seriously questioned from an ethical standpoint.

Diskography is not going to go away, despite wishful thinking. Clearly, prospective investigations of diskography are warranted. Although it seems likely that disk stimulation can enable identification of a symptomatic disk, it remains to be shown that this information has prognostic value. This must be done. What might these clinical trials look like? The following is one example. A patient population could be stratified based on symptoms/working diagnosis and identification of patients likely to benefit from diskography (probably those with chronic pain). These patients could be randomly placed into groups that are tested with either MR imaging or MR imaging plus diskography, and followed thereafter. The surgical rates, costs, number of days worked, complications, patient anxiety, and sense of well-being could be tabulated, with the primary endpoint being functional status. These two groups could then be compared to determine if diskography had a positive predictive value or a positive influence on therapeutic thinking.

The beliefs that a false test is useful or that a useful test is false are equally reprehensible. There is a need for continued research into the pathophysiologic mechanisms of disk stimulation. The real challenge lies in identifying a patient who would benefit from this test. Until these decisions can be based on well-controlled clinical trials, there is no basis for the performance of diskography in clinical medicine.

The authors' final statement is perhaps a finer use of the English language than they intended: “Provocative cervical diskography, including [diskography of] C2–3 if possible, can be employed to evaluate head and neck pain of suspected cervical discogenic origin.” The question is, however, not can it be employed, but rather should it be employed.

We do no justice to the test or our patients by refusing to follow good clinical science. If diskography is a valuable tool, let's prove it. If it is not, let's discard it quickly once and for all.