Guiding Choice — Ethically Influencing Referrals in ACOs

As the accountable care organization (ACO) model spreads in the United States, early experiences are being used to improve operations. One aspect of the model receiving substantial attention is the organization’s influence on referrals. ACOs are accountable for all their patients’ expenditures, whether incurred within or outside their organization, and many patients receive specialty care outside their ACOs.¹ Influencing where patients receive care may be a mechanism for assuring quality and controlling cost by reducing duplicative, unnecessary, or high-priced care or by increasing the use of high-quality care — and may therefore be critical to achieving ACOs’ cost and quality targets.

Some ACOs are already creating incentives, such as lower co-payments, for patients to seek specialty care inside their organization. Others use preferred referral lists to encourage physicians to refer within the ACO or to specialists selected for cost and quality reasons. Some proposals for modifying the Medicare ACO program focus on ACOs’ need to control referrals in order to provide high-value care.^2,3 ACOs could influence referrals in a number of ways (see table).

Mechanisms for ACOs to Influence Referrals.^*

Method	Examples
Informational	Distributing lists of preferred providers for referrals, selected on the basis of cost, quality, and other relevant criteria, such as patient-centered outcome
	Developing decision aids to encourage shared decision making about referrals
	Designing electronic referral processes to support referrals to high-value providers (e.g., through default referral to preferred providers)
Nonfinancial	Offering positive reinforcement (e.g., thank-you cards) to patients who obtain care within the ACO
	Providing individual feedback on referral performance to both referring physicians and specialists, either confidentially or in peer groups
	Recognizing physicians who excel at referring patients to high-value providers, either within the organization or publicly
Financial	Lowering patient copayments for care obtained within the ACO
	Giving financial bonuses to physicians and other referring clinicians (e.g., through portions of shared-savings distributions) on the basis of their referral patterns

^*Examples are subject to regulatory permissibility.

But influencing referrals raises ethical concerns regarding physicians’ autonomy to recommend what they believe is best for their patients and patients’ autonomy to choose their own physicians. If ACOs must influence referrals in order to achieve their goals, can they do so while respecting the ethical values of autonomy and duty of fidelity to patients?

Although choice is a cherished American value, influencing patients’ choice of health care providers may not be inherently unethical. Physicians may sometimes know, from past experience, the best specialist for a patient’s condition. But they may lack objective cost and quality information on specialists in general, and their overall referral practices may be idiosyncratic and not necessarily focused on high-value care.³ Patients who can choose their own specialists might have better outcomes, be more satisfied, or at least appreciate that freedom, but if patients tend to follow their primary care physicians’ recommendations, it’s unclear how much choice they truly have and whether current referral practices are unequivocally in their best interest.

Whether influencing referrals is ethical therefore depends on how it’s accomplished. Similar issues arose in the managed care era of the 1990s,⁴ and ACOs should apply the lessons learned from that experience. We believe that ACOs can influence referrals in an ethical manner that simultaneously enhances choice and improves patient outcomes if they consider three basic issues: transparency, appropriate metrics, and the right incentives.

Given the complexity of various ACOs’ arrangements with physicians, as well as the shadow cast by managed care “gag rules” restricting the information physicians could provide to patients (e.g., prohibiting discussions of treatment options not offered by the plan, disclosures of incentives to provide less care, or referrals to outside specialists), transparency about how referrals are influenced is arguably paramount. From an ethical standpoint, neither physicians nor patients can exercise autonomy without adequate information. For instance, it would be inappropriate for an ACO to preprogram computerized referral systems with preferred clinicians without telling physicians or patients it was doing so. Transparency requires informing physicians and patients about the use of preferred referral lists, whether they affect physicians’ recommendations or patients’ choices. Transparency also requires disclosure of the rationale behind such lists, their expected consequences for patients (both benefits and costs), and if applicable, the ways in which the organization influences referrals (e.g., through incentives).

Transparency could be achieved in several ways. Many ACOs, including those in the Medicare ACO program, have public-notification requirements or maintain public websites, which provide means for general disclosure of their use of preferred referrals and the rationale behind it. At the time of referral, organizations have the opportunity to make their referral practices clear to individual patients; for example, referral paperwork or electronic communications could include information on why the ACO prefers the listed physicians.

Second, physicians’ duty of fidelity requires that organizations base their influence over referrals on appropriate metrics. It would be problematic to influence referrals solely on the basis of cost, for instance, without attention to patients’ health outcomes or experiences of care. Physicians will want to be certain that preferred referral lists do not conflict with their commitment to patients’ well-being, and patients will want to know whether the referral process incorporates metrics of interest to them.

Systems for selecting preferred providers should therefore incorporate more than cost and hard medical outcomes (such as mortality and readmission rates). ACOs could draw on patient-centered outcomes research to identify the outcomes most relevant to patients. For example, an ACO creating a preferred referral list for cardiovascular disease could incorporate such outcomes as validated measures of quality of life or symptoms in addition to cost or disease-specific metrics such as statin-prescribing rates. Referral decisions could also legitimately be guided by evaluation systems that consider patients’ broader concerns and values, such as scheduling convenience, location, racial or cultural concordance, communication skills, or the existence of an established therapeutic relationship. Such a system might necessitate the inclusion of a diverse provider selection; it might also require allowances for exceptions or an appeals process to permit some referrals to providers outside the ACO without patient or physician penalties.

Some physician or patient preferences might well conflict with an ACO’s cost and quality goals. Such tensions may be irresolvable, but the organization could mitigate them by engaging physicians and patients in the process of choosing metrics for preferred providers, which could improve patient-centered care, ensure buy-in, and engender trust in the ACO. Engaging the wider provider community in the process could help ensure their buy-in, clarifying expectations and increasing their commitment to the goals of measuring and encouraging high-value referrals.

Finally, there’s the question of whether and how to incentivize physicians and patients to use and follow new referral practices. Influencing the process financially, by paying physicians bonuses or promising them a portion of the ACO’s shared savings, may be controversial and might rekindle the concerns about financial gatekeeping that plagued managed care. But financial incentives are not the only way to influence behavior.⁵ A more ethical approach might be to provide physicians and patients with data on specialists’ performance on the relevant metrics, perhaps in the form of a point-of-care decision aid, and allow them to jointly determine the best course of action; their mutual interest in choosing high-value care may be incentive enough for engaging in the desired referral practices.

If information alone proves insufficient but physicians and patients retain control over referral decisions, additional incentives may be necessary to influence referrals. Nonfinancial incentives, such as organizational recognition of “high-value referrers,” may be a logical next step. If that approach is ineffective, carefully applied financial incentives, such as bonuses linked to high-value referral practices, may be necessary. For the use of such incentives to be ethical, however, patients must be informed of their existence, both by the ACO and by the referring physician.⁴ Whatever incentives are used, patients’ best interests should remain primary, and the incentives should not inappropriately influence medical decision making.

If ACOs influence referral patterns in the right way, they could ensure the provision of high-value care. And influencing referrals by helping patients knowingly and intentionally choose the most appropriate physicians could actually be more respectful of patient choice and physicians’ duty of fidelity than current referral practices are.

Whether the approach we envision is feasible in all circumstances remains to be seen, but we believe that ethical considerations should guide referral practices. Research is needed to elucidate ACOs’ current referral practices and their conformity to and effects on core ethical values. As health care organizations assume increasing financial risk, the need to influence referrals will probably grow, and today’s ACOs have an opportunity to develop and disseminate models for doing so ethically.