10.1177/0272989X05284110MEDICALFORRESTDECISIONPHYSICIANJAN–FEB ANDDECISION MAKING SPECIALTY OTHERS IN MAKING/JAN–FEB CLINICAL REFERRAL PRACTICE DECISION 2006 MAKING DECISION MAKING IN CLINICAL PRACTICE Primary Care Physician Specialty Referral Decision Making: Patient, Physician, and Health Care System Determinants Christopher B. Forrest, MD, PhD, Paul A. Nutting, MD, MSPH, Sarah von Schrader, MA, Charles Rohde, PhD, Barbara Starfield, MD, MPH Purpose. To examine the effects of patient, physician, and creased risk of referral included PCPs with less tolerance of health care system characteristics on primary care physi- uncertainty, larger practice size, health plans with gate- cians’ (PCPs’) specialty referral decision making. Methods. keeping arrangements, and practices with high levels of man- Physicians (n = 142) and their practices (n =83)locatedin30 aged care. The risk of a referral being made for discretionary states completed background questionnaires and collected reasons was increased by capitated primary care payment, survey data for all patient visits (n = 34,069) made during 15 internal medicine specialty of the PCP, high concentration of consecutive workdays. The authors modeled the occurrence specialists in the community, and higher levels of managed of any specialty referral, which occurred during 5.2% of vis- care in the practice. Conclusions. PCPs’ referral decisions are its, as a function of patient, physician, and health care system influenced by a complex mix of patient, physician, and structural characteristics. A subanalysis was done to exam- health care system structural characteristics. Factors associ- ine determinants of referrals made for discretionary indica- ated with more discretionary referrals may lower PCPs’ tions (17% of referrals), operationalized as problems com- thresholds for referring problems that could have been man- monly managed by PCPs, high level of diagnostic and aged in their entirety within primary care settings. Key therapeutic certainty, low urgency for specialist involvement, words: referral-consultation; primary care; managed care; and cognitive assistance only requested from the specialist. medical decision making; uninsured. (Med Decis Making Results. Patient characteristics had the largest effects in the 2006;26:76–85) any-referral model. Other variables associated with an in- ompared with other types of clinical decisions, rates show a high year-to-year correlation2,3 and Cmaking a referral is an uncommon event in pri- marked interphysician variability.2,4,5 Thus, the thresh- mary care practice. Approximately 5% of primary care old that each PCP has for making a referral appears to visits include a referral, 71% have a medication pre- be a stable practice style trait, although one that is scribed, 36% a laboratory test, and 13% an imaging variable among physicians. study ordered.1 Primary care physician (PCP) referral able. Financial support for this study was provided by grant no. R01 Received 24 November 2004 from the Department of Health Policy HS09377 from the Agency for Healthcare Research and Quality. The and Management (CBF, BS) and Department of Biostatistics (CR), funding agreement ensured the authors’ independence in designing Johns Hopkins Bloomberg School of Public Health, Baltimore, MD; the the study, interpreting the data, and writing and publishing the report. Center for Research Strategies and the Department of Family Medi- The authors report no conflicts of interest. Revision accepted for publi- cine, University of Colorado, Denver (PAN); and the Department of Psy- cation 12 June 2005. chological and Quantitative Foundations, University of Iowa, Iowa City (SvS). Laurie Vorel and Jim Werner provided technical assistance with Address correspondence and reprint requests to Christopher B. data collection and project implementation. We are grateful to the phy- Forrest, MD, PhD, Department of Health Policy and Management, 624 sicians and office coordinators who donated their time and provided N. Broadway, Room 689, Baltimore, MD 21205; e-mail: cforrest@ their energy to make this study possible. Their involvement was invalu- jhsph.edu. DOI: 10.1177/0272989X05284110 76 • MEDICAL DECISION MAKING/JAN–FEB 2006 Downloaded from mdm.sagepub.com at JOHNS HOPKINS UNIV on September 9, 2010 PHYSICIAN SPECIALTY REFERRAL DECISION MAKING The most important inputs to the referral decision Elucidating the determinants of referral decision are the clinical characteristics associated with the pa- making can inform policy directed at management of tient’s presenting health problem.4,6,7 The majority of the interface between primary and specialty care phy- explained variation in referral decision making ap- sicians. Our objective in this article is to examine the si- pears to be related to diagnostic or case-specific multaneous contributions of patient, physician, and factors.4,6,7 The frequency with which a condition pre- health care system effects on PCPs’ specialty referral sents to primary care physicians explains about two decision making. This study adds to the literature on thirds of the variation in condition-specific referral physician referral decision making in several ways. rates.8 In a qualitative study on general practitioners’ Virtually all prior studies have examined just a few de- referral decision making, Dowie found that differing terminants at a time. Although the National Ambula- perceptions of disease severity and its potential impact tory Medical Care Survey (NAMCS) has been used to on future health were important considerations in re- study multiple influences on referral decision making,8,24 9 ferral decision making. concordance on whether a referral decision occurred is Getting advice on diagnosis or treatment is a com- only moderately high between the survey instrument 10,11 mon reason for referral, and physicians’ responses and direct observation of physician encounters.25 The 9,12 to clinical uncertainty are not uniform. However, NAMCS collects information on a variety of processes prior research has not found psychological responses of care—for example, medications prescribed, coun- to uncertainty to be important determinants of refer- seling done, tests ordered—which may lead to incom- 13 13,14 rals. Fear of malpractice, risk attitudes, and toler- plete capture of information on referrals. Our study 13,15 ance of uncertainty have nil or weak associations methodology was designed specifically to study refer- with referral. Physicians’ psychological reactions to ral decision making and its determinants. Other re- uncertainty may not play a role in all types of referral, search has used medical claims of specialist use as a perhaps just those for which there is incomplete proxy for referral decision making4,13,19; however, much knowledge of referral efficacy. specialist use results from ongoing referrals or patient Supply of specialists is an important system deter- self-referral, neither of which requires active PCP deci- minant of referrals. Greater availability facilitates entry sion making on need for specialty referral. Lastly, the into specialty care, heightens patients’ expectations for study links detailed information about patient morbid- referral, and increases demand for specialty care. More ity and other case-mix factors, health plan characteris- specialists in a community are associated with signifi- tics, and physician psychological variables for a large cant increases in rates of referral.16–19 US referral rates 20 sample of visits made to 142 physicians who practice are twice those of patients in the United Kingdom, a in 30 states. The multilevel perspective of this study, its finding best explained by the larger supply of special- large sample size, and its explicit focus on referral deci- ists in the United States. sion making strengthen the importance of our findings The high costs of tests and procedures that result to the policy dialogue regarding the boundaries and from specialist care have motivated health care organi- connections between primary care physicians and zations to manage physician and patient demands for specialists. specialty referrals. A variety of financial and organiza- tional constraints on decision making have been em- ployed to limit demand for specialists.21 These include METHODS referral performance assessment that may be tied to fi- nancial incentives such as bonus payments or spe- Physician Recruitment cialty withholds. At the patient level, plans have used gate-keeping arrangements to impede direct access to Physician recruitment was directed to physician specialists and have managed specialist supply with members of the Ambulatory Sentinel Practice Network selective provider contracting. In one of the few com- (ASPN), physicians affiliated with Medical Group prehensive assessments of the effects of managed health Management Association, other practice-based re- plan controls on specialty referrals, Grembowski and search networks (Minnesota Academy of Family Phy- colleagues found that among patients with common sicians Research Network, the Wisconsin Research pain problems, a specialty withhold was associated Network, and the Dartmouth COOP), and the larger with lower likelihood of referral,22 and among patients community of physicians. The study was publicized with depressive symptoms, this finding held true for via direct mailings to physicians, articles in research low-income patients.23 network newsletters, notices in journals, and presenta- DECISION MAKING IN CLINICAL PRACTICE 77 Downloaded from