Primary care providers are able to abide mainly by deontological principles. However, some program directors and most health administrators are obliged also to apply utilitarian principles. It is when resources are limited that these 2 perfectly legitimate ethical approaches clash — and give rise to the difficult decisions involved in health care rationing.

To help with such decisions, the movement for evidence-based medicine has promoted the writing of clinical practice guidelines. In this issue (page 634), Mita Giacomini and colleagues analyze the psychosocial content of a broad set of guidelines used in cardiology, rightly surmising that the selection criteria set out in these guidelines might unwittingly hide value judgements that could unjustly limit an individual patient's access to treatment.¹

They identify the following 6 psychosocial factors deemed to be relevant to the eligibility of patients for cardiac treatment:

• behavioural and psychological makeup (e.g., lifestyle, mental competence or stability, capacity to comply with treatment)
• availability of family and other support systems, when these are important to a favourable outcome
• financial resources
• occupation and social roles
• "environmental" factors such as proximity to a treatment centre
• a mixture of psychosocial and biomedical characteristics, including age, disability and lifestyle

The authors demonstrate competently the merits of scrutinizing clinical practice guidelines for their psychosocial biases, and this is the strength of their report. However, their discussion does not make explicit the implications of bringing selection criteria to bear on procedures that are both very costly and in limited supply (as in the case of heart transplantation) as opposed to selecting candidates for procedures that are available to all without the need for rationing. Severe material limitations (financial or biological) need not limit entitlement for patients in need of procedures that are reasonably accessible, such as pacemaker implantation, cardiac catheterization, angioplasty, coronary artery stent insertion, echocardiography or even coronary artery bypass surgery. In such cases, psychosocial selection criteria remain within the field of deontology. That is, in considering factors such as the patient's competence, family or other support systems, occupation-related considerations or even transport from remote locations, the ethical drive is still to maximize benefit for that patient. Weighing psychosocial factors in this context is evidence of a deontological commitment to nonmaleficence, not of a utilitarian commitment to collective outcomes.

However, in selecting patients to receive a heart transplant, the circumstances and competing needs of all those who would benefit from a transplant must be considered, in the knowledge that only a minority of patients will be placed on the waiting list and only a minority of those on the waiting list will eventually be given a new heart. Here, the ethical context is inevitably utilitarian: one must seek to maximize the benefit for the whole group of potential transplant recipients by weighing each case against the others. This makes choices much more difficult and, therefore, presents a greater risk that clinical practice guidelines may cloak unclear bias and occult, and possibly flawed, value judgements. As Giacomini and colleagues point out, psychosocial and biomedical factors may be used to select patients who will provide good outcome results for a particular program. The ethics of this are seriously flawed, except perhaps when a program is just starting, ongoing funding is not yet assured, and the use of this criterion is publicly acknowledged and the duration of its application is stated from the start.

In drawing attention to the values inherent in the selection criteria of practice guidelines, especially where psychosocial factors are concerned, Giacomini and colleagues alert us to the essential ethical nature of such constructs and help to clarify their often obscure and sometimes poorly justified rationale. Giacomini and coworkers have published their own framework for understanding the ethical imperatives of their present report elsewhere.² Rightly, they identify this as an important area for future research.

Clinical practice guidelines will probably play an increasingly important role in guiding decision-making by front-line health care providers in the selection of recipients for special services, including services that are rationed. If this is to happen, then guidelines will need to be legitimated by means other than simply relying on the expertise of physician "gatekeepers." This aspect, which is not part of the report by Giacomini and colleagues, is nevertheless a foreseeable outcome of future research in guideline development. The question of what constitutes full legitimization is addressed by Norheim,³ who insists that, to be legitimate and authentic, clinical practice guidelines must pass public as well as widespread stakeholder standards of acceptability based on such features as the following:

• public and stakeholder participation
• explicit information on values used in reaching recommended conclusions and the rationale used in applying these values
• transparency of the process to all those affected by it
• evidence of impartiality

Norheim also asks the following questions: "Are the inclusion or exclusion criteria discussed and justified with reference to:

• Medical criteria?
• Costs and opportunity costs?
• Non-medical criteria such as age, productivity, social status, gender?"

These considerations would be even more important for clinical practice guidelines developed by, or on behalf of, for-profit institutions.