The Dutch physicist Gorter predicted in 1936, on theoretical grounds, the existence of the magnetic resonance (MR) phenomenon. This was ten years before Felix Bloch and Edward Purcell, independently of each other in different experimental settings, demonstrated it. Current medical and biological applications of magnetic resonance imaging and spectroscopy are directly derived from in vitro tissue studies performed in the 1960s and 1970s. Although Odelbiad and co-workers published as early as 1956 an MR study of human red blood cells, this and other early work was limited by the unavailability of suitable MR instrumentation.

The physical phenomenon of nuclear magnetic resonance (NMR) was first characterized almost forty years ago in 1946 by the simultaneous but independent experimental successes of American scientists Felix Bloch and Edward Purcell. Their discoveries prompted development of conventional NMR spectroscopy. a technique used to describe the molecular composition and behavior of chemical compounds. Twenty-five years later, in 1971, Damadian used NMR to demonstrate differences in the behavior of water in malignant and benign tissues, and he suggested that NMR possessed “many of the desirable features of an external probe for the detection of internal cancer” (7). In the same year, Lauterbur produced the first two-dimensional NMR image, a cross-sectional portrait of two tubes of water (25). The potential utility of this technique to medical imaging was obvious, and soon afterwards multiple researchers began development of clinical NMR imaging systems. The first human whole-body NMR scan was accomplished by 1977. Improvements in the scanning process and image quality continue with, as yet, no limits in sight. In this clinical context, NMR techniques have experienced a name change to the current prevailing appellation, magnetic resonance imaging (MRI).

Magnetic resonance (MR) imaging is an exciting new diagnostic modality that has created tremendous interest in the medical profession. Although not unparalleled, the excitement engendered by MR imaging conjures up memories of the “CAT fever” induced by introduction of X-ray computed tomography (CT) scanners in 1973 (19).

Magnetic resonance imaging (MRI) is a rapidly evolving medical technology which has become widely diffused before comprehensive assessment of its safety, efficacy, cost-effectiveness, and relative cost-effectiveness compared with diagnostic modalities that it seems most likely to replace could be made. In early 1982 there were fewer than ten units worldwide. In August 1984, there were about 150 units and the figure will be closer to 300 by the end of 1985.

Any new technology that promises significant costs as well as potential benefits generates considerable interest. Such innovations bring into bold relief the fundamental problem of achieving maximal medical benefits while efficiently and equitably allocating scarce resources (II). Two tools that decision makers can use to assess the benefits and costs of implementing new technologies are cost- benefit analysis (CBA) and cost-effectiveness analysis (CEA). One technology that seems especially appropriate for the application of CBA and CEA is nuclear magnetic resonance (NMR) imaging.

That new health technologies often diffuse into the health care systems of developed countries without adequate evaluation has long been a cause of concern (I). In addition, where clinical or economic assessments have been carried out, they often contain methodological weaknesses which reduce their usefulness to health policy makers or clinical practitioners (5,6).

The purpose of this paper is to outline some of the key methodologic issues in designing clinical trials to assess the cost-effectiveness of clinical magnetic resonance imaging (MRI). Because the possible objectives of such trials are so diverse, and the appropriate methodologies correspondingly numerous, no attempt is made to identify a single objective or methodology as appropriate. Rather, this paper should be read as an “options paper,” in which alternative formulations of objectives, outcome measures, study design, and protocol instrumentation are reviewed, with critical discussion of their theoretical strengths and weaknesses, as well as their practical implementability.

Magnetic resonance (MR) undoubtedly constitutes a major technological advance, but as yet we do not know whether this invention also represents a breakthrough from the clinician's point of view.

Guyatt and Drummond suggest that the clinical value of MR is tested by means of a randomized clinical trial (RCT) in a well-defined and limited clinical area, and I fully agree with the authors that the use of such trials should not be confined to the testing of new treatments. In this particular field, however, I miss a discussion of alternative options. With few exceptions, new drugs can only be tested satisfactorily by means of an RCT, but the superiority of one imaging technique over another can sometimes be proved in other ways. Imagine, for instance, that the CT-scan is normal but that MR reveals a space-occupying lesion, the presence of which is confirmed by surgical operation or an autopsy. In that case the diagnostic superiority of MR has been proved in that particular patient. If such cases are seen frequently, and if it is known from controlled therapeutic trials that they are susceptible to treatment, then the value of MR has been proved without doing an RCT.

The first successful demonstration of the phenomenon of nuclear magnetic resonance (NMR), or nuclear induction in solids and liquids, was published almost simultaneously in 1946 by Bloch, Hansen, and Packard (7) working at Stanford University and Purcell, Torrey, and Pound (75) working at Harvard University. The immediate impact of their work was in physics and chemistry, but the applications have steadily widened and recently the application of NMR in medicine has become very exciting.

These new developments show the value of MR in studying a phase of metabolism hitherto unavailable, namely that of “preventive” study prior to significant lactate formation and surely prior to the cell damage edema and cell death so characteristic of stroke and heart disease. We believe that these approaches can be developed appropriately for clinical application and will thereby open a new type of diagnosis and therapy.

Magnetic resonance spectroscopy (MRS), especially of 31phosphorus, seeks to collect biochemical and metabolic information related to energy metabolism, intracellular pH, and oxygen delivery to the kidney (1,2). Earlier applications in clinical medicine have been established in diagnosis of metabolic myopathy (3,4), infantile cerebral anoxia (5), stroke (6,7), and in vascular insufficiency of the lower limb (8).

Nuclear magnetic resonance (NMR) imaging has recently become an accepted technique in the medical practitioner's armory (38). NMR spectroscopy (44) is a subtly different application of the same physical principles underlying NMR imaging, but the clinical potential for this modality is currently still under evaluation. The most important application of clinical NMR spectroscopy is for the nonin-vasive monitoring of changes in metabolite levels and intracellular pH of intact tissues during physiological stress or in response to pharmacological agents or disease. The 31phosphorus (31P) nucleus has been the most commonly investigated in muscle disease (39) but the applications of proton (1H), (4,5,8) and 13carbon (13C), (2,7) are currently being explored.

In a standard whole-body NMR-scanning machine, the static magnetic field is generated by an electric current driven through large solenoid coils. Dynamic magnetic gradient fields are generated by electric current pulses in coils located at various orientations, thus producing magnetic gradients in x, y, and z directions. The Rf (radiofrequency) radiation is transmitted through a specially shaped coil which also serves as an antenna receiving the NMR signals.