Clinical priority setting

BMJ 2008; 337 doi: (Published 09 October 2008) Cite this as: BMJ 2008;337:a1846
  1. Ole Frithjof Norheim, professor12
  1. 1Department of Thoracic Medicine, Haukeland University Hospital, 5020 Bergen, Norway
  2. 2Department of Public Health and Primary Care, University of Bergen, Kalfarveien 31, 5018 Bergen, Norway
  1. Correspondence to: O F Norheim ole.norheim{at}

    Twelve years ago (BMJ 1996;312:1553-4) the BMJ argued that health systems needed to be explicit about rationing and published articles describing different ways of rationing fairly. Here a clinician, two ethicists (doi:10.1136/bmj.a1850), and four health economists (doi:10.1136/bmj.a1872) discuss how their ideas have developed—and been put into practice—since then

    There are no simple solutions to resource allocation in health care,1 2 but the principles guiding fair priority setting are quite straightforward.3 4 5 This article explains the key principles and criteria for fair and clinically relevant priority setting. Clinicians should know these basic principles and be active in improving priority setting at all levels of the healthcare system.

    Clinical priorities

    Clinical priority setting means choosing whom to investigate and what diagnostic tests to perform; sorting the flow of patients so some are diagnosed or treated before others; allocating patients to surgery, medical treatment, or watchful observation; and selecting or excluding patients for a given treatment.6 7 Justice requires a fair distribution of the benefits and burdens of priority setting.8

    Clinicians’ decisions rest on two types of information: the patient’s severity of disease (prognosis without the intervention) and the expected outcome (prognosis with the intervention). The evidence underlying their judgment is also important.9 Clinicians can help support priority setting by stopping procedures that have little evidence of effectiveness.

    Priority setting is an integral part of daily practice in many clinical specialties. Cardiology is but one example. Firm evidence supports sorting patients with risk of a cardiovascular event into prognostic priority groups.10 European clinical guidelines, for instance, distinguish between three prognostic groups (graded from high to low priority): patients with established cardiovascular disease; asymptomatic patients with a 10 year risk of cardiovascular death ≥5%; asymptomatic patients with a 10 year risk of cardiovascular death ≤5%.11 Patients in the first two groups almost always get preventive treatment, but those in the third group may not because the evidence of effectiveness is weaker (but not absent) and the risk is low. There is therefore a lively debate about the marginal value of preventive treatment for this group.

    Non-clinical considerations

    Non-medical patient characteristics are also important when making decisions about priorities. All ethical theories of fair priority setting require impartiality in the consideration of patient characteristics.4 Only those characteristics that everyone can agree are relevant should affect the ranking of patients into prognostic priority groups. This implies that the characteristics such as area of residence, sex, religion, ethnicity, sexual orientation, educational level, social status, mental and physical disability, and (in most healthcare systems) ability to pay are irrelevant. It goes without saying that non-relevant criteria are often used in practice.

    Other characteristics are contested and should be considered and discussed carefully. Age may be relevant if clinicians must choose between life extending procedures (such as heart transplantations when there is scarcity of organs).12 13 Age is not relevant for palliative care or care that improves activities of daily living. Comorbidity may be relevant if it clearly reduces effectiveness of treatment. Personal responsibility may justify taxes on cigarettes but should not affect priorities for those already ill.14

    Cross cutting impartiality

    Clinical priorities must also be consistent across patient groups and clinical specialties. If a patient with a given severity of disease and expected outcome is assigned high priority within one specialty, another patient with a disease with the same characteristics in a different specialty should be treated similarly. Such decisions may seem out of the hands of clinicians, but fair minded clinicians should not always fight for more resources for their patients if this leads to lower priority for other patient groups with stronger claims. Narrow minded clinical autonomy and professional interest can hamper fair priority setting.15

    To improve priority setting across patient groups, clinical leaders must engage with others and make transparent choices based on two key principles of just resource allocation. There is wide agreement that the healthcare system should aim to maximise average healthy life expectancy and distribute health fairly across patient groups.8 The two key principles of resource allocation often point in the same direction, but sometimes one intervention will maximise health and increase inequality.16 If so, value choices have to be made on the appropriate trade-offs.17 Such value choices go beyond the domain of clinical priority setting and must also include the public through transparent deliberation.18

    The clinician’s concern for expected outcomes is grounded in the goal of maximising health through efficient allocation of resources. The concern for severity of disease is grounded in the goal of distributing health fairly. The patient with the poorest prognosis, if untreated, will be the worst off from a lifetime perspective. A patient with juvenile rheumatoid arthritis will have fewer healthy life years over her lifetime compared with an otherwise healthy patient who develops angina at 65 and dies of a myocardial infarction at the age of 75. If we assume that the outcomes of treatment are of comparable magnitude, fairness requires that higher priority should be assigned to the patient with juvenile rheumatoid arthritis, who is worst off from a lifetime perspective. In many healthcare systems, priorities often go in the opposite direction.

    Cost effectiveness analysis can be used to inform decisions on which priority strategies maximise health. It is more difficult to quantify fairness in the distribution of health, and this goal tends to be ignored in analysis of clinical priority strategies. However, tools to measure inequality, such as the Gini index, can be used (box).19

    Gini impact analysis of dialysis for end stage renal disease in Thailand

    Health authorities in Thailand have decided to include peritoneal dialysis or haemodialysis for patients with end stage renal disease in their tax based universal health insurance scheme. Although the national policy aims at offering dialysis to everyone in need, it is costly and requires health staff who are not yet available. The hospital will in the foreseeable future have limited capacity to treat all, so it must define priority groups. The hospital’s nephrologist consults a study estimating the gain in life expectancy from peritoneal dialysis for different age groups.20 He realises that if he treats two 60 year old patients and a 70 year old patient, the aggregated gain in life years is the same as if he treats a 40 and a 50 year old patient. The nephrologist outlines three possible priority strategies: priority to the youngest (<60); priority to the oldest (≥60); and first come, first served as long as capacity permits. All the priority strategies yield the same sum of additional life years (table).

    However, the nephrologist is also interested in the distribution of life years between patients. His moral intuitions tell him that priority to the youngest is fairer and he wants to test this quantitatively. He therefore calculates life expectancy separately for each age group and compares the Gini index for each priority strategy (table 1). The figure shows the Gini impact of each strategy. Priority to the youngest gives the largest reduction (54%) of inequality in the distribution of life years. So this is the fairest priority strategy.

    Outcomes from three priority strategies to decide who should get dialysis

    View this table:

    Gini impact from distribution of life years according to three priority strategies for dialysis. The Gini impact is calculated as percentage reduction of Gini index compared with no intervention

    Although many accept the relevance of the principles and criteria outlined here, clinical priority setting has limitations. Fairness in the distribution of health is largely associated with unfair distribution of the social determinants of health.21 Priority setting also takes place at higher levels of the system. Clinical leadership in setting priorities requires political support for hard choices.22 This is often not present. Sometimes there is lack of evidence.23 Institutions devoted to priority setting, such as the National Institute for Health and Clinical Excellence (NICE) in the UK, have come a long way in implementing evidence based, clinically relevant, and publicly informed priority setting.24 25 However, it should rethink its heavy reliance on quality adjusted life years (QALYs) and cost effectiveness.26 27 More work needs to be done on methods for quantifying the distribution of QALYs, and on how NICE guidance can be better integrated with clinical decision making.28 However, clinical priority setting remains the cornerstone of any healthcare system. We, as clinicians, should take the lead and improve our thinking, our methods, and our choices.


    Cite this as: BMJ 2008;337:a1846


    • Contributors and sources: OFN is chair of the International Society for Priority Setting in Health Care and his current research interest concerns measuring fairness in the distribution of healthy life years.

    • Competing interests: None declared.

    • Provenance and peer review: Commissioned; externally peer reviewed.