Geographical variations in health services that do not correspond to demographical or epidemiological differences are ethically challenging as they express unwarranted differences in care. As indicated in the introduction, this can be differences in the provision of care, access to care, use of health services, in quality and standard of care, in outcomes of care etc.—all infringing the principle of justice . Some patients will get more, and some will get less, depending on where they live or are served. As pointed out in the introduction, there are socioeconomic, racial, ethnic, risk- and health-related differences in the provision, access, use, and quality of care [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59] that are profoundly unjust.
As already pointed out, globally healthcare needs are higher among the poor, while access to health care and healthcare use is higher among the rich [43,44,45,46,47]. Locally health services are not placed where the needs are highest [48, 49].
Unwarranted geographical variations express variation in resource use , inefficient health services management , overuse, e.g., in terms of unnecessary surgery , but also of underuse. Overuse of low-value care represents an opportunity cost, reducing the access to efficient care for others. Low-value care is defined as “an intervention in which evidence suggest it confers not or very little benefit for patients, or risk of harm exceeds probable benefit or, more broadly, the added costs of the intervention do not provide proportional added benefits” .
Correspondingly, geographical variations raise issues of equity for a wide range of health services [2, 6, 43, 44, 46, 47, 51, 59, 79, 80] and pose challenges with respect to balancing concerns for equity with concerns for efficiency. As pointed out by Rice and colleagues, there exists a trade-off “between equity and efficiency in the allocation of health care resources. The pursuit of efficiency, in the sense of equalising the marginal cost of a unit of health gain, must be moderated by equity considerations” .
The main point ethically is that geographical variations in health services (that do not stem from morally relevant differences) are challenging, as they result in unwarranted discrimination, inequity, and injustice. While the causes for these geographical variations are well-known, e.g., structural, socioeconomic, and racial/ethnic differences, biases, stigmatization, as well as differences in living conditions, climate and power structures, the solutions are difficult to obtain exactly due to these ingrained causes. The important lesson is that while some of the causes of and solutions for geographical variations in the health services can be found within the services (see below) many of them have to be found outside the services.
Hence, unwarranted geographical variations in the health services infringes the principle of justice and thereby poses a substantial problem to health policy makers, health care managers, health professionals, and patients.
Another ethical challenge with geographical variations in the health services is that they are indicators of harm. There are two main forms of harm related to such variations: harm from underuse and harms from overuse.
Harm from underuse is increased morbidity and mortality due to late or lack of provision, access to, or use of health services. A patient who could have been diagnosed and successfully treated for cancer (at an earlier stage), but who suffers more as the result of the delay or neglect, is a typical example. The consequence of underuse is poor health outcomes in terms of welfare and health shortfall . Underuse has been documented in a wide range of health services . The underuse of each of these services has a range of specific harms, that should be addressed in each specific case.
For overuse, there is a plethora of harms. The harms of diagnostic overuse include false positive test results (anxiety and follow up), incidental findings of little or no significance (incidentalomas), overdiagnosis (diagnosis of a condition that would never harm the person) , side effects of diagnostics (e.g., reactions to contrast media or ionizing radiation in imaging), unnecessary subsequent diagnostics (to verify findings), and unnecessary subsequent treatment (overtreatment). Harms from therapeutic overuse include direct (expected) harms, adverse side-effects, and complications from treatments. Overuse has been documented for a wide range of health services globally [62, 83]. The overuse of each of these services has a range of specific harms, that ought to be addressed in each specific case.
Hence, unexplained geographical variations are ethically problematic as they indicate specific harms from underuse and overuse. Each of these harms need to be addressed specifically when investigating and tackling geographical variations.
Another key challenge with geographical variations is that it expresses variation in quality of care . Geographical differences in the quality of health services can result in worse outcomes for patients depending on where they live (or are served) and not on their health condition or characteristics [13, 16, 35, 38]. This goes for both benefits and harms (safety, as discussed above) and may infringe the principle of beneficence, as the geographical variations indicate that certain services are not of best benefit to patients.
One must also ask whether the same goes for those getting better care than others? It can be argued that too much of services with a benefit-to-harm-ratio > 1 cannot be bad. For example, to provide an effective treatment that is above the cost-effectiveness threshold  will benefit the patients getting the treatment. Hence, extra use (“overuse”) of such treatments is good (according to the argument). However, as pointed out above, providing (cost) effective services may generate injustice due to the opportunity cost (infringing the principle of justice).
Correspondingly, getting less of a service with a benefit-to-harm-ratio < 1 than others is good from the perspective of the principle of beneficence. Again, other principles, such as the principle of justice, kicks in to differentiate the cases.
Another interesting and relevant issue is when services have other effects than intended, making them beneficial in some other (sometimes undocumented) way, and that this beneficence is acknowledged and applied differently geographically. One example of this is the anxiolytic effects of radiological services. Many radiological investigations are not used because the health professionals believe that patients have specific pathologies (i.e., the pre-test-probability is low), but because the examination will have a reassuring or relaxing effect on the patient or in order to do something (ut aliquid fiat) [86, 87]. If the acknowledgement and application of such anxiolytic (or other beneficial) effects are unevenly distributed, it can be argued that geographical variations are good, or at least not bad. One problem, however, is that many such effects are uncertain and hard to document . Hence, the benefit is unclear. Additionally, we encounter problems with the principle of justice, discussed above.
Thus, geographical variations in the provision, access to, and use of health services represent differences in benefit to patients, and thereby violate the principle of beneficence. While having more of a beneficent service or having less of a non-beneficent service can be considered a good, this may infringe other principles, such as justice.
Respect for autonomy
When it comes to the principle of respect for autonomy, one can differentiate between variation in the experience and execution of patient autonomy and variations in professionals’ respect of patients’ autonomy. Moreover, one may distinguish between variation in autonomy due to variations in the provision, access, and use of health services and due to variations in the premises for executing autonomy, e.g., provided information. However, such distinctions are not clear.
Geographical variations in the execution of patients’ autonomy have been documented [89, 90], e.g., due to socioeconomic differences. Patient autonomy is also considered by health professionals to be a driver of unnecessary care, such as low-value radiological examinations , and as far as patient demand and professional integrity have geographical variations, this can cause geographical variations in the execution of patients’ autonomy.
Correspondingly, there are geographical variations due to differences in the respect for patients’ autonomy, e.g., in withdrawal and limitation of life-sustaining treatments for children . Such variations can be due to personal professional preferences .
Yet another way that geographical variations can be challenging considering the principle of respect for autonomy is that core criteria for making autonomous choices are not met. While geographical variations in the capacity to consent can warrant geographical variations in autonomy, variations in information (disclosure) and voluntariness may not. As professional preferences are shown to differ from and influence patients’ decision-making , geographical variations in the provision and use of health services can be due to differences in information and voluntariness.
As geographical variations unfairly affect low income, vulnerable, and underprivileged groups, they will depend on charity and are subject to another type of paternalism, as the few “lucky persons” who are offered “extra services” have little or no choice (i.e., lack voluntariness) and are not well informed.
This is not the place to enter a detailed discussion on the drivers of differences in respect for patients’ autonomy in health care. The point here is that geographical variations due to different levels of paternalism are ethically problematic. Hence, when addressing geographical variations, we should address the issue of paternalism.
From geographical variations to right care
As referred to in the introduction, there are many reports of geographical variations of health services, and it is generally assumed that such variations are ethically problematic, as they represent underuse and overuse. However, one thing is to describe geographical variations and to assume underuse and overuse. Quite another thing is to decide what is too little and too much, i.e., what is the right level of care. Defining the standard of care is challenging: “A further major lacuna in seeking to make ethical policies operational has been the reluctance to specify what constitutes a “standard” package of health care” . In general, effective care gives indication of “right care” .
Geographical variations have been reported for decades  without there been significant changes . One reason for this may be the difficult move from the description of geographical variations to the prescription of the right level and quality of care.
Many reports and atlases (e.g., in Table 1) present medians or mean for the various types of geographical variations, and thereby implicitly infer about under- and overuse based on such measures. However, this approach is by far justified, as it mixes quantity and quality and implies reasoning from is to ought.
To address such issues a wide range of professional and national bodies have established Clinical Care Standards  and National Safety and Quality Health Service Standards . Guidelines, checklists, and norms for right care are elaborated. See for example https://www.england.nhs.uk/rightcare/. In radiology an evidence- and consensus-based set of appropriateness criteria has been developed . While these approaches are powerful tools to reduce unwarranted geographical variations in health services, they raise ethical issues themselves. How do we know and decide what is appropriate?
In the Lancet series on this issue, right care is defined in terms of the four ethical principles: “What is right care? In its simplest definition it is care that weighs up benefits and harms, is patient-centred (taking individual circumstances, values, and wishes into account), and is informed by evidence, including cost-effectiveness” . To make this practically applicable two important questions raise: (1) Which outcome measures should be used to assess good or appropriate care? (2) Where do we set the limits for good or appropriate care (for the selected outcomes)? For example, when deciding on appropriate imaging, which outcome do we apply? Do we use subsequent improved health, changed subsequent health management (diagnostic or therapeutic path), reduced number of negative tests, reduced number of false (negative or positive) test results etc? If we decide to measure outcome in terms of changed subsequent health management, the question becomes how much this has to be changed in order to be good care. Where to set the limit very much depends on the context (local, national, global). Hence, the first question hinges on the value base of health care [100,101,102,103] and the second on the line-drawing problem in medicine [104, 105]. Both questions are ethically relevant.
Hence, unwarranted geographical variations of health services pose the core question “what is right care?” which is an ethically relevant issue. While professionally developed standards, guidelines, and checklists can be helpful for deciding on “right care” in specific contexts  there is empirical evidence that much more work needs to be done in order to reduce geographical variations. In particular, defining good care more explicitly is a crucial task which requires ethical reflection.
The point here is that paying attention to the ethical issues with geographical variations can make valuable contributions. In particular, addressing the injustice and potential harm that geographical variations entail, the lack of beneficence that they imply, and the paternalism that they indicate can help adjusting care. Knowing why geographical variations are wrong can help mitigating them.
As geographical variations may be contingent on the type of health service, let me briefly address this issue explicitly.
Geographical variations contingent on type of service
According to one of the pioneers in the studies of practice variations, Albert Mulley, “[i]f all practice variation were bad, then it would be an easy problem to solve” . Some variations are good, and some are bad, and the type of service may be relevant for discerning between them.
Health services are frequently divided into necessary services, preference-sensitive services, and supply-driven services . Necessary services are services that are documented to be crucial to people’s health and/or are strongly recommended in guidelines or standards of care, while preference-sensitive care are “those interventions in which there is a choice between at least two treatments that have different risks and benefits” , and supply-sensitive services are directed by the capacity or the interests of the local health care system and not by patients’ needs or preferences.
Accordingly, it may be argued that variations in preference-sensitive services are less ethically problematic than in necessary services. Certainly, variations in preference-sensitive services may not be ethically challenging as they can express variations in people’s preferences. However, if there are great geographical variations in preference-sensitive services in homogenous populations, this warrants attention as the variations then do not stem from patients’ preferences but can be iatrogenic or systemic (and thereby threaten the respect for autonomy).
Conversely, geographical variations in necessary services appear ethically challenging, as this indicates unwarranted differences in access or quality of care. However, there are not always agreements on what are necessary services, and there may be (inter)national differences of what is standard of care. Again, if there are geographical variations in necessary services in professionally and demographically homogeneous areas, this can be ethically problematic, especially with respect to differences in access to and quality of care. This poses challenges with respect to beneficence and non-maleficence.
When it comes to supply-sensitive services, it has been argued that these are ethically challenging as such, as they are (a) independent of patients’ preferences and (b) are unnecessary services [60, 84]. Accordingly, these can cause problems with patient autonomy and justice (opportunity cost, see below).
Hence, although there may be reasons to believe that geographical variations in necessary services are more problematic than in preference-sensitive services, and that they are always challenging in supply-sensitive services, geographical variations can pose ethical problems in all types of services. Moreover, geographical variations in necessary services may be more pronounced in low-income and vulnerable groups posing challenges with beneficence and justice, but also with non-maleficence and paternalism.