The Good (Gatekeeper), the Bad (Gatekeeper), and the Ugly (Situation)

Control of the “referral gate” to specialty care often is a lose-lose proposition for primary care physicians in managed care arrangements. If primary care physicians approve referrals to specialists too freely, they risk a reduction in their capitation rate or depletion of the fund reserved for specialty care and procedures. Also, specialists may complain about referrals for conditions that they think the primary care provider should have managed alone. Conversely, when primary care physicians infrequently open the gate to specialty care, patients may believe their physicians are denying them necessary care. Some plans reduce the primary care physician’s capitation rate when patients are dissatisfied. These opposing forces squeeze the primary care physician who is trying to meet patient needs at a reasonable cost.

Two articles in this issue present data that illustrate some of these phenomena. Kerr and colleagues surveyed more than 10,000 patients in 88 practices participating in a large, network-model HMO in California.¹ They found that patients in practices with policies that limited access to specialists were less satisfied than patients in practices with less-stringent policies. Dissatisfaction was especially high when patients reported an unmet desire to see a specialist.

Donahue and colleagues investigated 222 referrals from generalists to specialists in a single academic center.² Specialists rated nearly one third of the referrals as potentially or definitely inappropriate. Generalists and specialists disagreed about the necessity of more than one third of the referrals.

We commend these investigators for tackling some of the thorny issues related to specialty referral. The flow of patients between generalists and specialists is not an easy process to study. The strengths of Kerr’s study include the large sample size and carefully collected data on actual utilization review policies at study practices. The simultaneous study of both generalist and specialist physicians’ perceptions related to actual referral episodes is a strong point of Donahue’s study.

Although these studies advance our understanding of specialty referral, neither one helps us decide how to use specialists appropriately. First, both studies may have limited relevance to other health care settings. The data for the Kerr study come from only one managed care system in California in 1993, and the Donahue study involves providers in only one institution in 1995. Second, potential respondent bias weakens both studies. More than half the patients whom Kerr and colleagues attempted to contact for the telephone survey had wrong or disconnected numbers, while only 65% of specialists in the Donahue study returned questionnaires. Third, and most importantly, neither study measured clinical outcomes or cost-effectiveness.

Some findings in both studies may stem from problems in doctor-patient and doctor-doctor communication. An intriguing contradiction in the Kerr study was the observation that dissatisfaction with care was associated with “desiring” to see a specialist who was not seen, but not with having a referral actually denied. Factors that the investigators did not consider that might help us understand this peculiar situation include the duration of the patient–primary care physician relationship, patient confidence in the primary care provider’s clinical skills, and patient reluctance to ask directly about seeing a specialist. For example, unhappy patients may not have voiced discontent to their primary physicians, thereby missing an opportunity to learn from their physicians why the referral was not necessary. In Donahue’s study, nearly one third of referred patients did not show up for the referral appointment. In these cases, perhaps the generalist did not communicate clearly to the patient or the specialist the reason for the referral.

What have we learned about gatekeeping so far? There is ample evidence that restricted access to specialists and technology results in lower costs and utilization.^3–7 We know also that patients guard access to specialists jealously. Of 115 patients seen in an academic dermatology clinic, 90% described direct access to the specialists in the clinic as “very important” to their health care. Only about one fifth were satisfied with the care they received before coming to the clinic.⁸ In a previous analysis, Kerr demonstrated a relation between dissatisfaction with access to specialists and a desire to leave a managed care plan.⁹ Access to specialists is also an issue outside the United States. A survey of 2,734 Israeli patients found that one third would disenroll from an insurance plan that did not provide direct access to specialists.¹⁰ Kerr, Donahue, and their colleagues now provide evidence for what we have long suspected—the perception of restricted access to specialists breeds patient discontent, and doctors often disagree on the appropriateness of a given referral. At the same time, however, we know that for conditions such as major depression,¹¹ rheumatoid arthritis,¹² and asthma,¹³ specialists can improve the use of drugs, the patient’s clinical status, and the cost-effectiveness of care. Specialty care may improve survival for persons with rare conditions such as ovarian cancer.¹⁴ Yet, the primary provider alone can manage many problems well, and there is evidence that having a primary care physician can improve the efficiency of health care delivery and reduce unnecessary emergency department visits.^15,16 It is now imperative that we learn whether policies that restrict access to specialty care improve or worsen the effectiveness and cost-effectiveness of care.

Patient satisfaction is clearly important, but we might be willing to tolerate some dissatisfaction associated with restrictive referral policies if these policies also resulted in excellent clinical outcomes at reasonable costs. Despite the Kerr study, we should not cave in to every patient who wants to see a specialist. Doing so would send us back to the dark days of poorly coordinated care by multiple providers at high cost. (Does this sound like fee-for-service?) Conversely, despite the Donahue study, we should not restrict access to specialists simply because the referral might be perceived as avoidable or inappropriate. Donahue suggests reducing formal consultations through generalist training and telephone consultation with specialists, but there are clearly clinical situations in which education and telephone consultations cannot replace specialty expertise. Donahue defines appropriateness and avoidability from the perspectives of the referring and consulting physicians. Yet, we cannot really know whether referrals are appropriate or avoidable until we study the actual outcomes associated with referral patterns.

More than a decade ago, Eisenberg warned that gatekeeping would be a challenging role for primary care physicians.¹⁷ Since then we have had ample opportunity to experience firsthand just how thankless a role this can be. Current studies of gatekeeping focus on utilization and costs. But gatekeepers are still searching for the research that will help them decide how to use specialty resources to the patient’s best advantage. Without solid evidence to guide us in how often and easily we should open the referral gate, someone will always be berating us for opening it too much or too little.