When do doctors follow patients’ orders? Organizational mechanisms of physician influence

Social Science Research 48 (2014) 171–184

Contents lists available at ScienceDirect

Social Science Research

journal homepage: www.elsevier .com/locate /ssresearch

When do doctors follow patients’ orders? Organizationalmechanisms of physician influence

http://dx.doi.org/10.1016/j.ssresearch.2014.05.0120049-089X/� 2014 Elsevier Inc. All rights reserved.

⇑ Corresponding author.E-mail addresses: [email protected] (D.A. Menchik), [email protected] (L. Jin).

1 See CTS Physician Survey (1997–1999) and Leicht and Fennell (2001). The most recent study of Chicago lawyers finds that solo practitioners andsmall firms constituted only 28% of the profession, down from 42% 20 years earlier (Heinz et al., 2005).

Daniel A. Menchik a,⇑, Lei Jin b

a Lyman Briggs College and Department of Sociology, Michigan State University, 509 E. Circle Dr., Rm 316, East Lansing, MI 48824, United Statesb Department of Sociology and School of Public Health, RM 431, Sino Building, The Chinese University of Hong Kong, Shatin, Hong Kong SpecialAdministrative Region

a r t i c l e i n f o

Article history:Received 26 February 2013Revised 8 May 2014Accepted 30 May 2014Available online 20 June 2014

Keywords:PhysiciansOrganizationsPatientsNegotiationsHospitalsInfluence

a b s t r a c t

Physicians, like other professionals, are expected to draw from specialized knowledgewhile remaining receptive to clients’ requests. Using nationally representative U.S. surveydata from the Community Tracking Study, this paper examines the degree to which physi-cians are influenced by patients’ requests, and how physicians’ workplaces may mediateacquiescence rates through three mechanisms: constraints, protection, and incentives.We find that, based on physicians’ reports of their responses to patients’ suggestions,patient influence is rare. This influence is least likely to be felt in large workplaces, suchas large private practices, hospitals, and medical schools. We find that the protectionand incentives mechanisms mediate the relationship between workplace types and physi-cian acquiescence but more prescriptive measures such as guidelines and formularies donot affect acquiescence. We discuss these findings in light of the ongoing changes in thestructure of medicine.

� 2014 Elsevier Inc. All rights reserved.

1. Introduction

All professionals encounter clients who seek influence. The professional musician is pressured into playing the clichédsongs that patrons request (Becker, 1951; Grazian, 2003). Teachers are pushed to calibrate portions of their curriculum tothe needs of children with overzealous parents (Horvat et al., 2003). Even the highest-status members of a field are heldto account; for example, the corporate lawyer frequently acts as a handmaiden to the interests of the firms she represents(Abbott, 1981; Heinz and Laumann, 1982). Studies in this tradition frequently center on professionals’ efforts to ‘‘keep theirdistance in order to not let any one client interfere with one’s own ongoing program of work and leisure’’ (Hughes, 1952).This literature’s underpinning theme is that much of one’s occupational life is spent negotiating one’s independence.

But, given the significant changes to the structure of medicine in the last two decades, how much space for resistingpatients’ influence do physicians really have? Paralleling the trend in the legal profession, the percentage of medical profes-sionals in large bureaucracies has grown, expanding from 10% in 1965 to about 33% of the U.S. physician population in 1999.1

This trend noted by Parsons (1963a) has continued, to the point that in 2008 more than 50% of doctors worked in practicesowned by hospitals or large organizations known as ‘‘integrated delivery systems’’ (Kocher and Sahni, 2011).

those in

http://crossmark.crossref.org/dialog/?doi=10.1016/j.ssresearch.2014.05.012&domain=pdf

http://dx.doi.org/10.1016/j.ssresearch.2014.05.012

mailto:[email protected]

mailto:[email protected]

http://dx.doi.org/10.1016/j.ssresearch.2014.05.012

http://www.sciencedirect.com/science/journal/0049089X

http://www.elsevier.com/locate/ssresearch

172 D.A. Menchik, L. Jin / Social Science Research 48 (2014) 171–184

In many ways, these workplace venues now may strengthen or weaken physicians’ capacity to exercise independentjudgment, according to how they recontextualize state mandates, mediate the access of local populations to health care,and establish payments for physicians. Yet this literature remains incomplete because it has not connected such workplacechanges to a key unit of analysis in medicine: professional influence. In this paper we study the relationship between variousworkplace controls and the way that physicians respond to patients’ requests. Earlier research focused on workplaces thatwere different—particularly those that were much smaller and more insulated from the market and the state. Because themarket and the state were minimally important before the 1960s and 1970s, when the social organization of medicineunderwent transformations, the key social influences on physicians’ decisions could be traced to their relationships with col-leagues and the community. Alongside influential studies of doctor–patient interactions (e.g., Maynard, 1991; Stivers, 2007)as well as direct-to-consumer advertising (Peyrot et al., 1998; Bell et al., 1999; Feldman et al., 2006), focus has been placedprimarily on the relationship between referral and collegial ties and the content of the consultation (Hall, 1946; Freidson,1960, 1970:89–90; Coleman et al., 1966).2

This paper builds on this research by accounting for the massive shift in the forms taken by medical workplaces, testingthe mechanisms through which these venues constrain or facilitate the influence of patients. And as the number of doctorsworking in large workplaces rises, we might expect—theoretically, at least—that we need to study not only ties to outsidecolleagues, but also systems of social control inside organizations. This paper, then, is about how workplaces control thedegree to which physicians can maintain influence, with a particular focus on acquiescence to patient requests. Consideringchanges in the way workplaces arrange physicians, it asks: in the exercise of their influence, when do physicians decidewhether or not to acquiesce to their patients’ requests, and what features of the social organization of medicine influencetheir decision? However, rather than focusing on the individuals who mediate client crises—the janitor’s wife who protectsher husband from untimely tenant requests, the receptionist who ensures the physician can control his or her time (Hughes,1951:66)—we demonstrate how acquiescence operates when a workplace mediates the relationship between clients andprofessionals. In addition to the benefits of moving beyond characterizing professional action in role-based terms, under-standing this process is valuable because, in practical terms, when a doctor does not influence a patient and instead acquies-ces to a request, it can have durable effects on a community (e.g., treating viruses with antibiotics, where inappropriate useleads to new forms of resistance and threatens the effectiveness of antibiotics in treating bacteria that cause pneumonia,strep throat, and ear infections [Stivers, 2007:4]). Consequently, the outcomes of physicians’ abilities to negotiate spacein which to exercise influence can be as important for studies of individual and public health as they are for studies ofthe social organization of professional work. In evaluating the mechanisms underpinning physician acquiescence, then, thispaper contributes to work on physician decision making (Feldman et al., 2006; May, 2007; Lutfey et al., 2012; Lin, 2014) andon medical and other types of professional bureaucracies (Dobbin and Kelly, 2007; Dixon-Woods et al., 2009; Currie et al.,2010; Waring and Bishop, 2011; Chiarello, 2013). And, by describing the differentiated landscape of medical work and thecrucial traits of professional action and its contingencies, this study can begin to counterbalance analyses of changes in themedical profession that engage in what Stevens (2001) referred to as oversimplified ‘‘theories of decline and fall.’’

2. Three workplace-based mechanisms influencing physician acquiescence

Since Simmel, a central focus of sociological theorizing has been the relationship between social forms and their contents.The workplace changes we have outlined above would open new possibilities for patients to influence doctors, and so thedoctors would feel pressure to acquiesce. We will study how influence is organized by looking at the relationship betweenphysicians’ experiences of being influenced and the features of the work setting. We can identify three kinds of mechanismsthrough which doctors might be influenced to acquiesce. First, these new workplace forms—which housed a threefoldgreater proportion of all physicians in 1999 than in 1965 (Leicht and Fennell, 2001)—require new kinds of bureaucratic con-straints on physicians’ ability to make decisions including acquiescence. Second, new protections provided in the workplaceagainst the demands of the market and competition may provide a mechanism that allows doctors to selectively respond topatients’ demands. Finally, an organizational emphasis on reputation and positive evaluation for services rendered mayprovide a mechanism that directs physicians’ attention toward or against patients. We call these mechanisms: constraints,protection, and incentives.

2.1. Constraints

The first mechanism is that of constraints that are imposed on the discretion of physicians to order certain tests andprocedures.

The success of an organization is largely dependent on control of participants. Yet control is especially problematic in aprofessional bureaucracy whose success depends on high-status employees. Organizations want participants to internalizetheir obligations and voluntarily carry out their assignments, but must contend with the fact that employees who are

2 Although Freidson (1975) later studied a medical group of about 50 physicians, he chose that site precisely for its atypicality. In Profession of Medicine,Freidson (1970) was so conscious of the importance of state-led changes that he pointed out that his analysis was relevant to the institutional landscape ofAmerican medicine only in what he called its pre-1965 ‘‘Golden Age.’’

D.A. Menchik, L. Jin / Social Science Research 48 (2014) 171–184 173

professionals will look to one another and external colleagues for guidance. Prior studies suggest that control of work oftenthen becomes divided between officials and informal leaders, with subordinates resenting the incursion of bureaucratic pri-orities into their work matters. The most extreme version of this model is that enacted in institutions such as prisons andmental hospitals and might involve the threat to use physical sanctions (Etzioni, 1961). Its more general feature, however,is alienation on the part of subordinates.

The nature of tensions in what Goss (1961) called ‘‘advisory bureaucracies’’ suggests that because physicians work inlarge organizations, the professional mode of social control will collide with bureaucratic orders. For example, bureaucraticmeasures may supplant technical competence and professional autonomy (Burdi and Baker, 1999; Freidson, 2001; Leichtand Fennell, 2001).

In American medicine, one way workplaces constrain physicians is through formularies. Formularies are lists of selectedmedications and therapies intended to constrain treatment costs. Such measures are widely prevalent, especially amongphysicians who work with group-model HMO [Health Management Organizations] (Neumann, 2004). Friction in contestingadherence to a list may arise from a patient request (a specific drug or test may not be covered) or insurer (the same requestmay be reimbursed at a lower rate by one insurer than another). For all these reasons, unusual requests can beget a greatdeal of administrative labor and expenses likely to be sanctioned by superordinates sensitive to the details of the formulary.Given the work required to deviate from approved treatments, we propose physicians will be less likely to satisfy requestswhen a high percentage of their patients hold insurance policies with a formulary.

Social scientists across cultures have long described the scrutiny of professionals by ‘‘impure’’ authorities. Organizationsmay use numbers, graphs, and formulas to transform qualities of work into numerically measurable units, and they expectprofessionals to reach performance goals (Porter, 1995; Power, 1997). Accordingly, thousands of guidelines are developedeach year, recommending benchmarks for conditions such as diabetes, heart failure, and hypertension. Like formularies, theseguidelines are influenced by clinical trials that indicate the relative benefit of one treatment protocol or another. Unlike for-mularies, however, they diffuse through professional organs such as journals, having been developed by committees of high-status practitioners. Although widespread neglect of these guidelines is considered unacceptable enough that it warrantsjournal articles and debates (e.g., Al-Khatib et al., 2011), their impact on medical work has been uneven (Timmermans andEpstein, 2010). Nonetheless, insurance reimbursement may be tied to clinicians’ adherence. Consequently, we propose thatthe more physicians feel compelled to incorporate guidelines into practice, the less likely they will acquiesce.

2.2. Protection

The second mechanism is that of the protection that a workplace offers physicians through reducing uncertainties thatrelate to matters of payment.

Professional specialization may be complemented by the resources and management a large organization can provide.Because they can focus on the core of their practice, professionals may benefit from working in bureaucratic organizations.To accomplish the respective interdependence, organizations must protect their core technologies from fluctuations in mar-ket demand. In industries such as steel, organizations stockpile materials acquired in an irregular market and steadily insertthese into production (Thompson, 1967). In contrast to sites described by Thompson, resources (e.g., sick patients) in medicalorganizations cannot be stockpiled, but workplaces can protect physicians with temporarily low patient volumes by poolingreimbursement-based income into fixed salaries. Goss (1961) shows the efficacy of this approach, demonstrating that phy-sicians do not resent such management as long as they retain full autonomy in their prescribing behaviors. Conflict may thenoccur only when workplaces impede core practices.

Because medicine is a service profession, in which physicians are often paid according to the number of people they canprocess, anything that disturbs this system of payment will be a problem. At the same time, patients are medicine’s cost cen-ter, a high expense for workplaces if they require too much time or other resources. Workplaces can reduce the degree doc-tors feel responsible for attracting and retaining patients. For instance, a large professional bureaucracy may attract patientson the basis of its prestige in an area, as well as its connections to insurers. We argue, then, that doctors who feel protectedfrom market forces—thereby sensing less competition—will feel less pressure to please their patients, simply in order toretain them.

Doctors in professional bureaucracies play a role much closer to employee than entrepreneur. And as the size of organi-zations has shifted, the proportion of doctors with flat pay rates and bonuses has also increased (see Casalino et al., 2008).These physicians commonly have a discontinuous relationship with their patients, who will see a different doctor dependingon the day they come to the office. In many ways this rotating custody resembles the ‘‘bureaucratized’’ nature of nursingwork (see Zerubavel, 1979:51). (Such rotations will likely be more common among doctors in specialties such as ob/gyn.)Like nurses, too, these physicians are salaried, an arrangement that—in contrast to those paid on a fee-for-service basis—relaxes the financial incentive to attract and keep patients. Of course, these salaried doctors may still have long-termrelationships with particular patients. However, we propose they will be less likely to acquiesce to their requests.

2.3. Incentives

The third mechanism is that of incentives provided by a principal when an agent performs tasks satisfactorily for thebenefit of the organization.


Principal–agent relationships involve the problem of relying on other individuals to act on one’s behalf. The usual chal-lenge is that the principal cannot perfectly monitor the agent’s action and information. This creates problems of inducementand enforcement (Pratt and Zeckhauser, 1991). When principals seek out experts, when principals are one-shotters andagents repeat players, or when principals are unexpectedly foisted into a new role with no time or life experience to formu-late preferences, let alone a contract or monitoring strategy (e.g., the new parents of a critically ill newborn [Heimer andStaffen, 1998]), the asymmetry of power shifts from the principal to the agent (Shapiro, 2005).

Medicine is saturated with principal–agent relationships. The classic example is that the patient hands over control ofcare to a presumably more knowledgeable physician. The scenarios discussed in this paper involve a further complicationbeyond that of the patient–physician relationship, however: administrators want doctors to act according to the best inter-ests of the workplace as a whole. Administrators in the role of principal also have a difficult time specifying an agency con-tract because they may not know what expert services are required or how much of them, what procedures ought to befollowed, or what criteria are appropriate to limit agent discretion. They also have difficulty evaluating the quality of servicebecause of the intrinsic indeterminacy in highly specialized tasks (Sharma, 1997:771). Finally, the interests of the organiza-tion may differ from those of the patient (with respect to, for instance, the amount of time allocated per visit). Given theseuncertainties, the agent ultimately wields control.

Incentive-based payment schemes are ways of structuring the agency situation. White (1991) argues that the role of com-pensation in the principal–agent relationship is less important in giving the agent incentives to act in the principal’s interestthan it is in informing agents what their responsibility is and how it changes (198). But how are relevant signals established?Workplace administrators have a problem because, unlike work in Adam Smith’s pin factory, medicine involves a compen-sation system not clearly linked to productivity. As a result, they attempt to align incentives by using commissions, piecerates, stock options, and deductibles. Bonuses are a classic approach and have shown some success in influencing whetherphysicians make referrals and conduct examinations and tests consistent with standards set by insurers (Cunningham,2004). Physicians who acquiesce to patients’ requests will require more time in the consultation (Hoff, 2010:29), and suchtemporal constraints have been shown to influence their decisions (Stepanikova, 2012). Doctors rewarded for attracting andmaintaining a high patient load may be concerned about both retaining these patients and allocating too much time to nego-tiating in the consultation. For this reason, we propose that doctors whose pay is weighted according to their level of clinicalproductivity will be more likely to acquiesce to requests.

Finally, administrators in professional bureaucracies want their clients to be happy with their service. If using perfor-mance-driven payment schemes reflects an explicit attempt to align incentives, another way administrators will seek todirect attention toward patients is by carrying out satisfaction surveys and linking physicians’ pay with their results. Thesesurveys are usually oriented toward patients’ general sentiments; they are as likely to address traits of the service relation-ship (e.g., office waiting time) as those of the patient–physician interaction itself. (Importantly, then, their results do not nec-essary reflect whether physicians acquiesce to requests.) We propose that the greater the degree to which such surveys areconducted—and, especially, the degree to which their results affect physician compensation—the more likely physicians willacquiesce to patients.

3. Data, measurements, and analytical strategy

3.1. Data

We examine physicians’ perceptions of their acquiescence using the 2000–2001 Community Tracking Study (CTS), anationally representative survey conducted by the Center for Studying Health System Change (see Metcalf et al., 1996).3 Thistelephone survey utilized stratified random sampling and interviewed about 12,000 physicians from 60 metropolitan and non-metropolitan health care markets who spent 20 h or more per week in direct patient care. We excluded physicians in surgicalspecialties from our sample because previous studies have suggested that the patterns of interaction between surgeons andtheir patients are fundamentally different from those for doctors in other specialties (Schwarze et al., 2010). Among the10,999 respondents of non-surgical specialties, 982 have missing values on at least one of the variables used in our analysisand therefore are excluded. Our final analytical sample includes 10,017 respondents, who represented 276,523 U.S. non-surgeonphysicians.

3.2. Measurements

3.2.1. Dependent variablePhysicians’ responses to the following question were used to measure the extent of patients’ influence on their clinical

decision making: ‘‘During the last month, for what percentage of your patients did you order tests, procedures, or prescrip-tions suggested by patients that you would not otherwise have ordered?’’

3 We use the 2000–2001 CTS physician survey because this data set includes all the variables we need. Later waves (2004–2005 and 2008) did not askphysicians about patient influence over their clinical decision-making.


Since we use survey data, it is difficult to get at the precise nature of a patient’s suggestion. How, then, do we know thatwhen physicians made these orders they were swayed by the patient? Because we cannot be sure how doctors might haveinterpreted the survey question, it is necessary to turn to the literature on patient–physician interaction for some guidance.

Recent research reinforces the notion that patient influence is at work when physicians report that they ordered some-thing ‘‘that [they] would not otherwise have ordered.’’ Studies of patient–physician interactions indicate that these arehighly stylized exchanges (Maynard, 1991) and share a common format across contexts; doctors are taught this format inmedical school, patients are trained in it from childhood, and both parties closely adhere to its boundaries (e.g., Sleathet al., 1997; Stevenson et al., 2000; Robinson and Heritage, 2006). To change the original plan of a doctor is not trivial; onlyby disrupting the usual sequence of interaction can patients override a doctor’s script. For example, patients may have tooffer their own candidate diagnosis (Stivers, 2002); overtly or tacitly suggest a causal relationship between their symptomsand a hypothesized diagnosis (Gill and Maynard, 2006); or propose that the physician consider recommendations made byanother practitioner, a process Gill (2005) calls ‘‘praising the predecessor.’’

A physician’s decision to accept a patient’s request thus indicates a willingness to break from usual behavior and is anunexpected act in the situation. This is also consistent with Parsons’s definition of influence as a means of persuasion. Here‘‘ego is able to bring about a decision on alter’s part to act in a certain way because it is felt to be a ‘good thing’ for him, on onehand independently of contingent or otherwise imposed changes in his situation, on the other hand for positive reasons, notbecause of the obligations he would violate through noncompliance’’ (Parsons, 1963b:48, italics in original).4 Thus, the pro-cess does not have to be coercive; the key element is that the physician realizes he has been convinced to act on the patient’sbehalf. Our survey question features precisely this element. Consequently, it represents an improvement over questions aboutphysician acquiescence that, because they are more direct, are more vulnerable to social desirability bias (Campbell et al., 2013).

The question’s wording can also capture a variety of types of requests and reasons why a physician might have been will-ing to grant a patient’s request. The request might have been routine, as when a patient requests a prescription for a ‘‘life-style’’ drug that prevents hair loss, assists with sleeplessness, or improves sexual function. Alternatively, the patient’squestions might have reminded the doctor of reasonable alternative treatments and procedures that were overlooked.Finally, a patient might have made a request involving topics on which the medical literature was silent—what Wennberget al. (2004) refer to as ‘‘supply-sensitive services’’—such as the use of imaging and other diagnostic tests for patients withchronic illnesses. By linking our dependent variable to features of physicians’ practice organizations, we can identify howphysicians’ immediate practice environment influences whether they acquiesce to patients.

3.2.2. Independent variablesOur independent variables center on physicians’ workplace organizations and the mechanisms of constraints, protection

and incentives related to these organizations.Physicians’ workplaces. Following other sociologists of the medical profession, we assess the size and form of physicians’

workplaces using a detailed seven-category typology: solo or two-physician practices,5 small medical groups (three to ninephysicians), medium and large medical groups (P10 physicians), medical school affiliates, staff- or group-model HMOs, hospi-tals, and other types of organizations and practice settings (mainly including governmental clinics and large and emergent orga-nizational forms6) (Leicht and Fennell, 2001).

Constraints. The constraints mechanism manifests in the measures through which organizations seek to guide physicianbehavior. We include two indicators in our analysis. In the CTS survey, physicians were asked, ‘‘How large an effect does youruse of formal, written practice guidelines such as those generated by physician organizations, insurance companies or HMOs,or government agencies have on your practice of medicine?’’ Responses ranged from ‘‘no effect’’ (0) to a ‘‘very large’’ effect(5). Respondents were also asked what percentage of their patients had prescription coverage that included the use of aformulary.

Protection. To assess the protection mechanism, we first examine physicians’ perceptions of market competition faced bytheir practices. The respondents were asked whether their practices operated in an environment that was ‘‘not at all com-petitive,’’ ‘‘competitive,’’ or ‘‘very competitive.’’ We also examine whether the respondents were paid a fixed salary.7

Incentives. We assess three measures of how workplaces attempt to align physicians’ motivations with the goals of theworkplaces. First, we examine whether physicians’ compensation is weighted by their clinical productivity, which refersto the amount of revenue generated, the number of relative value units produced, the number of patient visits they provided,

4 Although not wholly successful in our view and that of others (e.g., Janowitz, 1975), Parsons identifies some of the central theoretical issues that must beaddressed in any serious discussion of influence as a mechanism of social control (Sandefur and Laumann, 1998), and formulates the concept in a wayconsistent with other authoritative treatments (e.g., Coleman, 1990).

5 We combined these categories for four reasons. First, there were no differences in results when they were used independently. Second, combining themallows us to compare findings with the multiple other studies that use the CTS and other data sets to study physician behavior in organizations (e.g., Landonet al., 2001). Third, there are a limited number of two-person practices in the United States. Fourth, following Simmel (1950), we can expect the move from two-to three-physician practices to shift internal group dynamics most meaningfully.

6 The category of ‘‘large and emergent organizational forms’’ encompasses a diverse set of experimental workplaces. These tend to be large organizationsmotivated by the logic of managed care and market competition. Although some analysts consider them to represent the future of medical organizations(Robinson, 1999), the proportion of all physicians practicing in them is quite small (Casalino, 2004).

7 Full owners of solo practices were not asked these questions, and it was assumed that they were not paid a salary. For the dummy indicating whether arespondent was paid a fixed salary, the value is 0 for full owners of solo practices.


or the size of the enrollee panel. Second, we evaluate whether compensation is influenced by the results of patient satisfac-tion surveys. We also evaluate the extent to which physicians feel the surveys affect their practice. Respondents were asked,‘‘How large an effect does feedback from patient satisfaction surveys have on your practice of medicine?’’ Responses rangedfrom ‘‘no effect’’ (0) to a ‘‘very large’’ effect (5). Examining both measures is important because satisfaction surveys can affectdoctors’ behavior through pathways other than their compensation. For example, the surveys may be not be used to sanctionphysicians, but may simply be presented as signals of valued behavior in the workplace.

3.2.3. Control variablesWe control for a number of physicians’ background characteristics that may determine their willingness to acquiesce to

patient requests, which include their gender, their race/ethnicity, the number of years they have practiced medicine, theirspecialty, the ownership status of their workplace, and whether they have been trained as DOs or as MDs. We also evaluatesome relevant attitudinal factors. First, we examine physicians’ concerns about individual responsibility over decision mak-ing by examining the answer to the following question: ‘‘Using any number from one to ten, where ‘1’ is not important, and‘10’ is very important, tell me how important control over your clinical decisions is.’’ We also assess the extent to which phy-sicians factor patient input into their decisions, examining the answer to the question: ‘‘On balance, what do you think is theeffect of medical information obtained by your patients from sources other than you on your ability to provide high qualitycare?’’ The response categories are positive, neutral, or negative. We also control for whether physicians perceived that theyhad enough time to spend with their patients during clinical encounters.

Finally, we control for the percentage of practice revenues from patients with Medicare or Medicaid insurance. To controlfor propensity of patients to make requests, we hold constant the percentage of physicians’ patients who bring up self-collected medical information during the clinical encounter.

3.3. Analytical strategy

We define our measure of patient influence to be the proportion of a physician’s patients for whom he or she responded toa request in the month prior to the survey by ordering tests, procedures, or prescriptions that he or she would not haveotherwise ordered. This variable is bounded between zero and one and takes the value of zero for a relatively large numberof the cases in the sample. We therefore use Tobit analysis to model the dependent variable and unstandardized coefficientsfrom Tobit regressions are presented. Sampling weights and strata are taken into consideration to account for the stratifiedrandom sampling strategies employed by the CTS survey and to obtain nationally representative parameter estimates. Toassess protection, constraints, and incentives as potential workplace-based mechanisms for increasing acquiescence, we firstexamine whether the effect of measures of these mechanisms varies with different types of workplaces. We then assess therelationship between the dependent variable and measures of protection, constraints, and incentives and the changes in therelationship between physician acquiescence and organizational types after these measures are included in the analysis.

4. Results

4.1. Patient requests and fulfillment by physicians

Table 1 describes the dependent variable of our analysis. On average, the physicians’ likelihood of fulfilling patientrequests is quite low. They reported that during the month prior to the survey, they ordered tests, procedures, or prescrip-tions on the basis of a request for only 4.6% of patients. The tendency to fulfill patient requests was not evenly distributed.For about 7% of the physicians, making clinical decisions based on their patients’ requests was relatively common; they didso for more than 10% of their patients. On the other hand, about a third of the physicians fulfilling requests did so for lessthan 1% of their patients. The majority of the physicians (56%) fell between these two extremes.

Table 1Descriptive information on the dependent variable.

During the last month, for what percentage of your patients did you order tests, procedures, or prescriptions suggested by patients that you would nototherwise have ordered?

Mean (SD) 4.6 (7.8)Interquartile range (IQR) 5Distribution (%)<1% 341–5% 476–10% 11>10% 7.1

Note: Based on data from 10,017 physicians in non-surgical specialties, who represented 276,523 non-surgeon physicians in the United States.


4.2. Descriptive statistics of independent variables

Table 2 displays the descriptive statistics of independent and control variables. Here we briefly describe the characteris-tics of the main independent variables. First, physicians practice in a diverse range of workplaces. A third of surveyed phy-sicians worked in solo or two-physician groups. Eighteen percent worked in small groups with three to nine physicians, andanother 11% were in medium and large groups (P10 physicians). About 8% were affiliated with medical schools, and 14%worked in community hospitals. Around 4% practiced in staff- or group-model HMOs. The remaining 12% worked in govern-ment-funded clinics, other large organizations such as physician-hospital organizations, and other types of work settings.

Among indicators of workplace constraints, on average, physicians reported that 61% of their patients had prescriptioncoverage that included the use of a formulary. Clinical practice guidelines exerted large or very large effects on the practiceof 23% of the physicians and had no effect for 7% of them. In terms of the protection mechanism, about half of physicians(53%) received their compensations in salary. About 35% of physicians did not consider their practices to face a competitivesituation, whereas around two-thirds reported that their practices faced a competitive (46%) or very competitive (19%) sit-uation. Among the indicators of the incentive mechanism, 53% of physicians reported that their compensation was weightedby their clinical productivity. Around 20% of physicians reported that their compensation was affected by patient satisfactionsurveys. About 34% of physicians indicated that the effect of feedback from patient satisfaction surveys on their practice waseither large or very large, whereas about 13% reported that the feedback had no effect.

4.3. Workplaces and patient influence

We first examine the effects of the workplace on the tendency of physicians to fulfill patient requests. As model 1 inTable 3 shows, workplace form strongly predicts physician acquiescence. Among private physician groups, size is stronglycorrelated with physicians’ inclination to fulfill patient requests; doctors in larger practices are less likely to fulfill requeststhan are those in small practices. For example, the rate of acquiescence was about 1.29% lower in medium and large groupsthan in solo or two-physician practices (p < 0.01). Doctors in other large and complex organizations, such as hospitals, werealso significantly less likely to fulfill requests than were those in solo or two-physician settings. Among all types of practicesettings, doctors affiliated with medical schools were the least likely to fulfill patient requests; on average, the acquiescencerate is 2.88% lower than those in solo or two-physician settings (p < 0.001).

4.4. Variation in the social organization of organizational mechanisms across workplace types

We proposed that workplaces may influence physician acquiescence through three mechanisms: constraints, protection,and incentives. That is, the presence of these mechanisms should vary across different workplace types, and this variationshould account for the different rates of perceived acquiescence across workplaces. As Table 4 shows, among all types ofworkplaces, the mechanism of constraints was the most prominent for physicians in staff- or group-model HMOs; 73%reported experiencing moderate to very large effects of clinical practice guidelines, and 87% of their patients had formularies.In contrast, the corresponding percentages for those in solo or two-physician practices were 56% and 55%.

In terms of the protection mechanism, 17% of physicians in solo or two-physician practices were salaried; around 53% ofphysicians in small groups and in medium and large groups were salaried; and more than 89% of physicians in hospitals andmedical schools were salaried. On the other hand, the perception of competition was the most acute in solo or two-physicianpractices; 24% of physicians in these workplaces perceived that their practices faced a very competitive situation. In smallgroups and in medium and large groups, hospitals, and medical schools, around 17–19% of physicians perceived themselvesto face a very competitive situation. Interestingly, staff- or group-model HMOs did not seem to have shielded their memberphysicians from the feelings of external competition; about 23% of these physicians felt they faced a very competitive situation.

Physicians also face different incentives across workplaces. Only 20% of physicians in solo or two-physician practicesreported that their compensation was weighted by their clinical productivity, whereas three-quarters of those in smallgroups and in medium and large groups reported that their pay was so weighted. The percentage was 69% in medical schoolsand 66% in both hospitals and staff- or group-model HMOs. Similarly, physicians in larger organizations were also morelikely to have their pay affected by results from patient satisfaction surveys. In particular, 66% of physicians in staff- orgroup-model HMOs reported that their compensation was affected by results from satisfaction surveys, as compared to5% in solo or two-physician practices. The percentage of physicians in group practices and medical schools whose paywas affected by results from patient satisfaction survey ranges from 16% (small groups) to 23% (medium and large groups).The effect of patient satisfaction surveys on physicians’ practice was also more prominent in larger and more complex orga-nizations, with 77% of physicians in staff- or group-model HMOs, 64% in hospitals, 64% in small groups, and 54% in solo ortwo-physician practices reporting moderate to large effects of these surveys.

4.5. How well do organizational mechanisms predict acquiescence?

Having found that workplace is important in explaining acquiescence, and given that workplaces vary according to thepresence of potential mechanisms for patient influence, we investigate further whether these features explain acquiescencerates.

Table 2Descriptive statistics of independent and control variables.

Variables % or Mean (SD)

Workplace typesSolo or two-physician practice 33Small groups (3–9 physicians) 19Medium/large groups (P10 physicians) 11Medical schools 8Hospitals 13Group- or staff-model HMOs 4Other workplace types 11

Organizational mechanismsConstraintsExtent to which clinical practice guidelines affected physician practice (%)

No effect 7Small/very small 33Moderate 36Large/very large 23% Patients who had prescription coverage that included the use of a formulary (Mean (SD)) 61 (26)

ProtectionPaid salary (%) 53Perceived competitive facing practices (%)

Not competitive 35Competitive 46Very competitive 19

IncentivesCompensation is affected by clinical productivity 53Compensation is affected by patient satisfaction surveys 20Extent to which feedback of patient satisfaction survey affected physician practice (%)

No effect 13Small/very small 24Moderate 29Large/very large 34

Control variablesYears of practice (Mean (SD)) 15 (10)Osteopathic physicians (%) 8Women physicians (%) 27Ownership status (%)

Owners 31Partial owners 22Nonowners 47

Physician specialty (%)Internal medicine 18Family/general practice 22Pediatrics 9Medical specialties 35Psychiatry 7Ob/Gyn 8

Physician race (%)Caucasian 75African American 4Asian 13Hispanic 6Other 2

Having enough time with patient during clinic visit (%)Disagree strongly 13Disagree 23Neither agree nor disagree 2.2Agree 34Agree strongly 27

The effect of independently obtained information on quality of care (%)Negative 15Neutral 35Positive 50

The importance of controlling clinical decision making (Mean (SD)) 9.6 (0.9)% Practice revenues from patients with managed care insurance (Mean (SD)) 47 (28)% Revenue from patients with Medicaid (Mean (SD)) 16 (18)% Practice revenues from patients on Medicare (Mean (SD)) 29 (23)% Patients who brought up independently-obtained medical information during clinical consultation in previous month (Mean (SD)) 18 (18)

Note: Based on data from 10,017 physicians in non-surgical specialties, who represented 276,523 non-surgeon physicians in the United States.


Table 3Workplace types, organizational mechanisms, and physician acquiescence.

Fulfilled patients’ requests

Model 1 Model 2 Model 3 Model 4 Model 5

Workplace types (reference: solo or two-physician practices)Small groups (3–9 physicians) �0.61 �0.58 �0.45 �1.14* �0.90*

Medium/large groups (P10 physicians) �1.29** �1.26** �1.17** �1.84*** �1.66**

Medical schools �2.88*** �2.77*** �2.56*** �3.33*** �2.91***

Hospitals �1.34** �1.32** �1.11* �1.77** �1.48**

Group- or staff-model HMOs 0.09 �0.02 0.38 �0.35 �0.13Other workplace types �0.67 �0.64 �0.41 �0.86 �0.54

Organizational mechanismsConstraintsThe effect of clinical practice guidelines on

physician practice– 0.08 – – 0.01

% Patients with formularies – 0.01 – – 0.01

ProtectionSalaried – – �0.80** – �0.92**

Perception of competition (reference: Not competitive)Competitive – – 0.27 – 0.11Very competitive – – 0.90* – 0.71

IncentivesCompensation is affected by clinical

productivity– – – 1.18** 1.16**

Compensation is affected by patientsatisfaction surveys

– – – 0.26 0.33

The effect of feedback of patientsatisfaction surveys on physicianpractice

– – – 0.24** 0.23*

F statistics F(30,2492) = 38.7 F(32,2492) = 36.6 F(33,2491) = 35.29 F(33,2491) = 37.71 F(38,2486) = 33.45P < 0.001 P < 0.001 P < 0.001 P < 0.001 P < 0.001

* p < .05. ** p < .01. *** p < .001. Two-tailed tests.Note: Results are unstandardized Tobit regression coefficients. Based on data from 10,017 physicians in non-surgical specialties, representing 276,523 non-surgeon physicians in the United States. The dependent variable is the response to the question: ‘‘During the last month, for what percentage of yourpatients did you order tests, procedures, or prescriptions suggested by patients that you would not otherwise have ordered?’’ All models also control foryears of practice, gender, whether the respondent was an osteopathic physician, ownership status, specialty, race and ethnicity, having enough time withpatients during clinical visits, physicians’ evaluation of the effects of independently-obtained medical information on the quality of patient care, theimportance of controlling clinical decision making, the percentages of patients with managed care, Medicare or Medicaid insurance, and% patients whobrought up independently-obtained medical information in the previous month. The results for the control variables are in Appendix A.

Table 4Organizational mechanisms in different types of workplaces.

Coercion Protection Incentives

% Moderateto largeeffects ofguidelines

Average %patientswithformularies

% Paidsalary

% Perceivingverycompetitivesituation faced bypractices

% Compensationaffected byclinicalproductivity

% Compensationaffected bysatisfactionsurveys

% Moderate tolarge effects ofpatientsatisfactionsurveys

Workplace typesSolo or two-physician practices 56 55 17 24 20 5 54Small groups (3–9 physicians) 57 58 53 18 74 16 64Medium/large groups (P10

physicians)60 56 53 19 75 23 69

Medical schools 61 60 97 17 69 20 64Hospitals 60 58 89 17 66 33 64Group- or staff-model HMOs 73 87 89 23 66 66 77Other workplace types 61 58 76 12 57 29 67

P-value 0.003a <0.001b <0.001a <0.001a <0.001a <0.001a <0.001a

Note: Based on data from 10,017 physicians in non-surgical specialties, representing 276,523 non-surgeon physicians in the United States.a From Pearson Chi-square tests.b From one-way ANOVAs.



Constraints. As model 2 in Table 3 shows, in contrast to our expectations, no indicator of the constraints mechanism issignificantly correlated with the level of physician acquiescence. Neither the extent to which physicians’ practices wereaffected by clinical practice guideline nor the percent of physicians’ patients with drug formularies were significant predic-tors of physician acquiescence with patient requests.

Protection. As model 3 in Table 3 shows, as expected, salaried physicians were less likely than the nonsalaried to do whattheir patients requested. The acquiescence rate of salaried physicians was 0.8% lower than that of the nonsalaried. A percep-tion of strong competition was associated with a higher acquiescence rate. Compared with those who perceived the situationfaced by their practices as noncompetitive, the acquiescence rate of physicians who perceived the situation as very compet-itive was 0.9% higher.

Incentives. As model 4 in Table 3 shows, compared with those whose compensation was not weighted by clinical produc-tivity, the acquiescence rate of physicians whose compensation was so weighted was 1.18% higher. The feedback frompatient satisfaction surveys also has a significant effect on physician acquiescence. Physicians who reported that their prac-tice was under heavier influence of the feedback of patient satisfaction surveys were more likely to do what their patientsasked (p < 0.01). Linking physicians’ pay with results from patient satisfaction surveys, however, does not significantly influ-ence the acquiescence rate.

Model 5 in Table 3 includes variables indexing all three organizational mechanisms. Although these indicators differamong workplace types and some are predictive of physicians’ tendency to comply with patient requests, the workplace dif-ferences in the rate of physician acquiescence still remain after these variables are included. Results from model 5 thus showthat workplace types remain significantly associated with the tendency to fulfill patient requests. We return to this subject inthe discussion.

4.6. Control variables and physician acquiescence

Appendix A displays the effects of control variables on physician acquiescence. We note several interesting findings here.First, female physicians were less likely than their male counterparts to fulfill patient requests. This resistance possibly arosebecause they anticipated that patients were less likely to respect their authority and thus asserted their position in the clin-ical encounter more forcefully. We also found that doctors with a more patient-centered disposition were not more likely tofulfill patient requests. That is, those who thought that independently-obtained information by the patient has positiveeffects on quality of care were no more likely to fulfill patient requests than were those who were negatively disposed(see also Peräkylä, 2006). Curiously too, racial characteristics of providers were also important factors; both Asian and His-panic doctors were more likely than white doctors to do what their patients asked. Past research has shown that physicians’racial/ethnic background influences their interaction dynamics with patients (e.g., Johnson et al., 2004), which may partiallycontribute to the observed differences in the acquiescence rate. But the exact mechanisms underlying these differencesawait further investigation. Interestingly, specialists were less likely than family physicians or general internists to fulfillpatients’ requests. This is likely a simple matter of specialization; fewer patients can garner the expertise necessary for asuccessful challenge. At the same time, the nature of the doctor–patient relationship differs once the client has progressedto meet those with the most specialized expertise. At this stage, specialists may be consulted for only a brief period with theconsultation focusing on specific issues. There the patients will have less opportunity to make requests. To control for phy-sicians’ individual inclination to control clinical decision making, we also include physicians’ self-report of how important itwas for them to control clinical decision making. Indeed, physicians who valued controlling clinical decision makingreported a 1.2% lower likelihood of fulfilling patient requests. Finally, we also controlled for the percent of patients whobrought up independently obtained medical information during clinical consultation in previous month as a proxy to theextent to which patients were making requests. Not surprisingly, this variable is highly significant; a percent rise in thepatients who brought up such information increases the acquiescence rate by 0.12%.

5. Discussion

Even if the social organization of health care has shifted, the dominant direction of influence in the clinic has not. Neworganizational arrangements have paralleled patients’ very successful expansion of legal rights in the past four decades. Fol-lowing the Patient Self-Determination Act in 1990 arrived living wills, durable power of attorney, and other advance healthcare directives. Combined with changes in access to medical information, it might be reasonably thought that this movementwould expand patient influence in the medical encounter. Indeed, it is common to assert that changes in the availability ofexpert and lay information on medical care have expanded patients’ control over their encounters with doctors (e.g., May,2007). Although we do not have longitudinal data to test this claim, the meager number of occasions in which patients haveinfluence is nonetheless striking. This finding complements ethnographic work reviewed by Bloor (2001: 178), who observed‘‘early medical sociology texts probably overemphasized the extent to which many routine medical consultations assumed anegotiated character: in the great majority of primary care consultations (such as in British General Practice) it is unlikely tobe a relevant pursuit for patients to exercise overt or covert influence on their consultation outcome.’’

Yet a key origin of influence today can be tied to features of the workplace. More specifically, we show that only certainelements of the workplace—the protection- and incentive-based mechanisms, in particular—structure whether acquiescence


is perceived. As Burawoy (1979) found in the Allied Corporation factory, the sole use of sanctions ‘‘from above’’ is largelyineffective in influencing work practices. Instead, the use of salaries and other administrative systems that protect physiciansfrom competitive pressures are key factors that influence professional practices. Physicians respond to these incentives intheir local practice environment—much more so than to attempts of professional associations to influence how individualphysicians apply their knowledge base. Workplace context, then, may be one key source of pressure that sways doctors fromthe ambivalence that Merton (1975) argued would emerge from their conflicting professional roles.

Although the differences in workplace size and form are responsible for variations in acquiescence, they cannot be fullyaccounted for by the two key mechanisms we identify. It is thus necessary to consider other factors varying across settings.One involves differences in collegial pressure across workplaces. Surveys are limited in evaluating interactions that establishand enforce norms on medicine’s ‘‘shop floor,’’ but better-suited studies allow us to speculate. They suggest that as work-place size grows, acquiescence should decline as resources dedicated to communication expand (see Kasarda, 1974). Physi-cians who practice in large organizations are pressured by colleagues to maintain normative standards (Stanton andSchwartz, 1954) or at least to ‘‘diffuse dissent’’ (Anspach, 1993). There, physicians ask one another about past experienceswith patients, information on the latest scientific research that might be applied to diagnosis and treatment, and adviceon interpreting these findings (Montgomery, 2005). There, colleagues may become less likely to fill requests, having spokento one another and solidified opinions about ‘‘best practices’’ (see Burt, 2005). A second explanation might be that those inlarger settings are better connected to those outside their practice. They are more likely to participate in clinical research, forinstance, a practice that deeply shape clinical work (e.g., Petty and Heimer, 2011). They may also place more value on statuswith outside colleagues, and thus be reluctant to acquiesce to patient-suggested practices rather than those capitalizing ontheir comparative advantage with colleagues.

While we should be wary about inferring too much about interaction processes from an organization’s size, at the veryleast the effects of the workplace imply that the same physician would behave differently in two workplace contexts,meeting the requests of patients more readily in a small practice than elsewhere. The pattern is familiar. Recent studies haveshown that the execution of some of American medicine’s unfunded mandates—such as federal limits on the workload oftrainees—are ultimately moderated by features of organizations (Szymczak et al., 2010; Kellogg, 2011).8

It is worth noting some limitations of the data. First, we used data from a cross-sectional survey, which limited our abil-ities to make causal inferences or assess the trends in the way physician acquiescence has evolved with the social organi-zation of medicine. In addition to historical and ethnographic work on medical practice, the optimal data would involvesurvey data collected over time. Although our data enabled us to understand some key features of acquiescence rates today,longitudinal data would inform how they have fluctuated. Second, it would be informative to know about some features ofpatients’ conditions, such as acuity. Perceived urgency of treatment may influence how doctors express their authority in thetime and place of the encounter. Third, our data are not new. However, because a primary purpose of this paper is to developand test general theoretical arguments about the relationship between organizations and acquiescence, we would argue thatits importance does not hinge on the use of the most recent data. Fourth, physicians’ reports of the number of times theyacquiesced to patients’ requests may be subject to recall bias. Although it is difficult to fully account for this issue, we arereassured by previous work that validates instances of physicians’ longer-term recollections by using a subset of data thatasks about their practices occurring in the last week, and finds little substantial variation between the validation and the fullsamples (Weissman et al., 2004). Furthermore, because they are non-normative, reported cases of acquiescence are likely tobe more easily recalled than many other pieces of information.

Ultimately, cross contextual differences suggest that it may not be most useful to describe changes to the profession usingterms that approximate the distance from the ideal-typical model of professional activity described by Freidson and scholarsof medical education. Instead, to better understand these changes in the profession, we should take our clue from a concep-tual shift in recent research on medical training (e.g., Brooks and Bosk, 2012): look at evolving work tasks and the kinds ofdoctors they produce. From our findings on patient satisfaction surveys, payment systems that reward a high patient volume,and salaries, we conclude that a substantial proportion of physician acquiescence is a result of both workplace-based pro-grams that generate feedback on physicians’ performances, and these venues’ ties to parent organizations that create policiesand provide funding. Consequently, it may be useful to study doctors less in relation to a specific professional standard intowhich they are ‘‘socialized,’’ as much literature has put it, and more in relation to the differentiated contexts they traverseand occupy in a career (see also Menchik, 2014: 103).

If physicians’ acquiescence depends on features of their workplaces, it suggests that expertise is institutionalized in notonly in professional realms but also in bureaucratic ones. To evaluate the matter of profession-wide change more generally,then, future studies might move from the study of individual doctors to confront how they are collectively organized. Thisshift to a study of collective organization will go far to address the limits of claims about the nature of a larger group basedon aggregated data from individual physicians. As Mead (1938) stressed, a common perspective does give rise to individualperspectives, but the common perspective cannot be said to be built up out of individual perspectives. Therefore, in order tounderstand how physicians come to decide whose knowledge informs clinical practices, we will need to know more about

8 Another symptom of the increased salience of the workplace in medicine is recent research finding solely within-organization and not between-organization processes of innovation diffusion (Huesch, 2011).


recurrent interactions and other ties among peers, industry, the state, and other institutional stakeholders which organizethe practices of medicine.

Acknowledgments

We would like to thank Larry Casalino, Tomas Farchi, Edward Laumann, Linda Lee, Sida Liu, Karen Lutfey, Aaron McCrightand Lin Tao for helpful comments. This research was supported by the University of Chicago Center on Demography andEconomics of Aging, the Oxford Institute of Population Ageing, and the Foundation for Informed Medical Decision Making.This article is dedicated to the University of Oxford’s Nuffield College librarians.

Appendix A. The effects of control variables on physician acquiescence

Fulfilled patientrequests

Control variables
Years of practice �0.24 Osteopathic physicians 0.51 Women physicians �0.76*
Ownership status (reference: full owner)
Partial owner 0.30 Nonowner 0.73
Physician specialty (reference: General Internal Medicine)
Family/general practice 0.45 Pediatrics �2.61***
Medical specialties
�3.17***
Psychiatry
�4.05***
Ob/Gyn
�1.50*
Physician race (reference: Caucasian)
African American �1.20 Asian 1.16*
Hispanic
1.88*
Other
�0.69 Having enough time with patients during clinical visits �0.55***
The effect of independently obtained information on quality of care (reference: negative)
Neutral 0.20 Positive 0.54
The importance of controlling clinical decision making
�0.63***
% Practice revenues from patients with managed care insurance
�0.02 % Practice revenues from patients on Medicare 0.14 % Revenue from patients with Medicaid 0.13 % Patients who brought up independently-obtained medical information during clinical
consultation in the previous month
0.12***
* p < .05. *** p < .001. Two-tailed tests.Note: Results are unstandardized Tobit regression coefficients. They are based on data from 10,017 physicians in non-surgical specialties, representing276,523 non-surgeon physicians in the United States. The dependent variable is the response to the question: ‘‘During the last month, for what percentageof your patients did you order tests, procedures, or prescriptions suggested by patients that you would not otherwise have ordered?’’ Other independentvariables in this model include: the types of physicians workplaces, the effect of clinical practice guidelines on physicians’ practice, % patients withformularies, whether a physicians was paid a salary, physicians’ perceptive of competition, the effect of feedback of patient satisfaction surveys on physicianpractice, whether physicians’ compensation is affected by results from patient satisfaction surveys and whether physicians’ compensation is affected byclinical productivity. The results for the other independent variables are shown in Table 3.

References

Abbott, Andrew, 1981. Status and status strain in the professions. Am. J. Sociol. 86, 819–835.Al-Khatib, Sana M., Hellkamp, Anne, Curtis, Jeptha, Mark, Daniel, Peterson, Eric, Sanders, Gillian D., Heidenreich, Paul A., Hernandez, Adrian F., Curtis, Lesley

H., Hammill, Stephen, 2011. Non-evidence-based ICD implantations in the United States. JAMA 305, 43–49.Anspach, Renée R., 1993. Deciding Who Lives: Fateful Choices in the Intensive Care Nursery. University of California Press, Berkeley, CA.Becker, Howard S., 1951. The professional dance musician and his audience. Am. J. Sociol. 57, 136–144.Bell, Robert A., Wilkes, Michael S., Kravitz, Richard L., 1999. Advertisement-induced prescription drug requests: patients’ anticipated reactions to a

physician who refuses. J. Fam. Pract. 48, 446–452.

http://refhub.elsevier.com/S0049-089X(14)00121-5/h0005








Bloor, Michael, 2001. The ethnography of health and medicine. In: Atkinson, P., Coffey, A., Delamont, S., Lofland, J., Lofland, L. (Eds.), The Handbook ofEthnography. SAGE, London, pp. 177–188.

Brooks, Joanna Veazey, Bosk, Charles L., 2012. Remaking surgical socialization: work hour restrictions, rites of passage, and occupational identity. Soc. Sci.Med. 75, 1625–1632.

Burawoy, Michael, 1979. Manufacturing Consent: Changes in the Labor Process under Monopoly Capitalism. University of Chicago Press, Chicago, IL.Burdi, Michael D., Baker, Laurence C., 1999. Physicians’ perceptions of autonomy and satisfaction in California. Health Aff. 18, 134–145.Burt, Ronald S., 2005. Brokerage and Closure: An Introduction to Social Capital. Oxford University Press, Oxford.Campbell, E.G., Pham-Kanter, G., Vogeli, C., Iezzoni, L.I., 2013. Physician acquiescence to patient demands for brand-name drugs: results of a national survey

of physicians. JAMA Internal Med. 173, 237–239.Casalino, Lawrence, 2004. Physicians and corporations: a corporate transformation of American medicine? J. Health Polit. Policy Law 29, 869–885.Casalino, Lawrence, November, Elizabeth A., Berenson, Robert A., Pham, Hoangmai H., 2008. Hospital–physician relations: two tracks and the decline of the

voluntary medical staff model. Health Aff. 27, 1305–1314.Chiarello, Elizabeth, 2013. How organizational context affects bioethical decision-making: pharmacists’ management of gatekeeping processes in retail and

hospital settings. Soc. Sci. Med. 98, 319–329.Coleman, James S., 1990. Foundations of Social Theory. Harvard University Press, Cambridge.Coleman, James S., Elihu, Katz, Herbert, Menzel, 1966. Medical Innovations. Bobbs-Merrill, Indianapolis, IN.Cunningham, Robert, 2004. Professionalism reconsidered: physician payment in a small-practice environment. Health Aff. 23, 36–47.Currie, Graeme, Finn, Rachael, Martin, Graham, 2010. Role transition and the interaction of relational and social identity: new nursing roles in the English

NHS. Organ. Stud. 31, 941–961.Dixon-Woods, Mary, Suokas, Anu, Pitchforth, Emma, Tarrant, Carolyn, 2009. An ethnographic study of classifying and accounting for risk at the sharp end of

medical wards. Soc. Sci. Med. 69, 362–369.Dobbin, Frank, Kelly, Erin L., 2007. How to stop harassment: professional construction of legal compliance in organizations. Am. J. Sociol. 112, 1203–1243.Etzioni, Amitai, 1961. A Comparative Analysis of Complex Organizations: On Power, Involvement, and Their Correlates. Free Press, Glencoe, IL.Feldman, Mitchell D., Franks, Peter, Epstein, Ronald M., Franz, Carol E., Kravitz, Richard L., 2006. Do patient requests for antidepressants enhance or hinder

physicians’ evaluation of depression? A randomized controlled trial. Med. Care 44, 1107–1113.Freidson, Eliot, 1960. Client control and medical practice. Am. J. Sociol. 65, 374–382.Freidson, Eliot, 1970. Profession of Medicine. University of Chicago Press, Chicago, IL.Freidson, Eliot, 1975. Doctoring Together: A Study of Professional Social Control. University of Chicago Press, Chicago, IL.Freidson, Eliot, 2001. Professionalism: The Third Logic. University of Chicago Press, Chicago, IL.Gill, Virginia T., 2005. Patient ‘demand’ for medical interventions: exerting pressure for an offer in a primary care clinic visit. Res. Lang. Soc. Interact. 38,

451–479.Gill, Virginia T, Maynard, D.W., 2006. Explaining illness; patients’ proposals and physicians’ responses. In: Heritage, J., Maynard, D. (Eds.), Communication in

Medical Care: Interactions between Primary Care Physicians and Patients. Cambridge University Press, Cambridge, pp. 115–150.Goss, Mary E.W., 1961. Influence and authority among physicians in an outpatient clinic. Am. Sociol. Rev. 26, 39–50.Grazian, David, 2003. Blue Chicago: The Search for Authenticity in Urban Blues Clubs. University of Chicago Press, Chicago, IL.Hall, Oswald, 1946. The informal organization of the medical profession. Can. J. Econ. Polit. Sci. 12, 30–44.Heimer, Carol, Staffen, Lisa, 1998. For the Sake of the Children: The Social Organization of Responsibility in the Hospital and the Home. University of Chicago

Press, Chicago, IL.Heinz, John P., Laumann, Edward O., 1982. Chicago Lawyers: The Social Structure of the Bar. Northwestern University Press, Evanston, IL.Heinz, John P., Nelson, Robert L., Sandefur, Rebecca L., Laumann, Edward O., 2005. Urban Lawyers: The New Social Structure of the Bar. University of Chicago

Press, Chicago, IL.Hoff, Timothy, 2010. Practice Under Pressure: Primary Care Physicians and Their Medicine in the Twenty-first Century. Rutgers University Press, Piscataway,

NJ.Horvat, Erin Mc Namara, Weininger, Elliot B., Lareau, Annette, 2003. From social ties to social capital: class differences in the relations between schools and

parent networks. Am. Educ. Res. J. 40, 319–351.Huesch, Marco D., 2011. Is blood thicker than water? Peer effects in stent utilization among floridian cardiologists. Soc. Sci. Med. 73, 1756–1765.Hughes, Everett Cherrington, 1951. Work and self. In: Coser, L. (Ed.), On Work, Race and the Sociological Imagination. University of Chicago Press, Chicago,

IL, pp. 57–66.Hughes, Everett Cherrington, 1952. The sociological study of work: an editorial foreword. Am. J. Sociol. 57, 423.Janowitz, Morris, 1975. Sociological theory and social control. Am. J. Sociol. 81, 82–108.Johnson, Rachel, Roter, Debra, Powe, Neil R., Cooper, Lisa A., 2004. Patient race/ethnicity and quality of patient–physician communication during medical

visits. Am. J. Public Health 94, 2084–2090.Kasarda, John D., 1974. The structural implications of social system size: a three-level analysis. Am. Sociol. Rev. 39, 19–28.Kellogg, Katherine C., 2011. Challenging Operations: Medical Reform and Resistance in Surgery. University of Chicago Press, Chicago, IL.Kocher, Robert, Sahni, Nikhil R., 2011. Hospitals’ race to employ physicians—the logic behind a money-losing proposition. N. Engl. J. Med. 364, 1790–1793.Landon, Bruce.E., Reschovsky, James, Reed, Marie, Blumenthal, David, 2001. Personal, organizational, and market level influences on physicians’ practice

patterns: results of a national survey of primary care physicians. Med. Care 39, 889–905.Leicht, Kevin, Fennell, Mary, 2001. Professional Work: A Sociological Approach. Blackwell, Oxford.Lin, Katherine Y., 2014. Physicians’ perceptions of autonomy across practice types: is autonomy in solo practice a myth? Soc. Sci. Med. 100, 21–29.Lutfey, K.E., Campbell, S., Marceau, L., Roland, M., McKinlay, J., 2012. Influences of organizational features of healthcare settings on clinical decision making:

qualitative results from a cross-national factorial experiment. Health 16, 40–56.May, Carl, 2007. The clinical encounter and the problem of context. Sociology 41, 29–45.Maynard, Douglas W., 1991. Interaction and asymmetry in clinical discourse. Am. J. Sociol. 97, 448–495.Mead, George H., 1938. The Philosophy of the Act. University of Chicago Press, Chicago, IL.Menchik, Daniel A., 2014. Simmel’s dynamic social medicine: new questions for studying medical institutions? Soc. Sci. Med. 107, 100–104.Merton, Robert, 1975. Sociological Ambivalence and Other Essays. Basic Books, New York.Metcalf, Charles E., Kemper, Peter, Kohn, Linda T., Pickreign, Jeremy D., 1996. Site Definition and Sample Design for the Community Tracking Study. Center

for Studying Health System Change, Washington, DC.Montgomery, Kathryn, 2005. How Doctors Think. Oxford University Press, Oxford.Neumann, Peter J., 2004. Evidence-based and value-based formulary guidelines. Health Aff. 23, 124–134.Parsons, Talcott, 1963a. Social change and medical organization in the United States: a sociological perspective. Ann. Am. Acad. Polit. Soc. Sci. 346, 21–33.Parsons, Talcott, 1963b. On the concept of influence. Public Opin. Quart. 27, 37–62.Peräkylä, Anssi, 2006. Communicating and responding to diagnosis. In: Heritage, J., Maynard, D. (Eds.), Communication in Medical Care: Interactions

between Primary Care Physicians and Patients. Cambridge University Press, Cambridge, pp. 214–247.Petty, Juleigh, Heimer, Carol A., 2011. Extending the rails: how research reshapes clinics. Soc. Stud. Sci. 41, 1–24.Peyrot, Mark, Alperstein, Neil M., Van Doren, Doris, Poli, Laurence G., 1998. Direct-to-consumer ads can influence behavior. Mark. Health Serv. 18, 2.Porter, Theodore M., 1995. Trust in Numbers: The Pursuit of Objectivity in Science and Public Life. Princeton University Press, Princeton, NJ.Power, Michael, 1997. The Audit Society: Rituals of Verification. Oxford University Press, Oxford.
















































































Pratt, John W., Zeckhauser, Richard J., 1991. Principals and agents: an overview. In: Pratt, John W., Zeckhauser, Richard J. (Eds.), Principals and Agents: TheStructure of Business. Harvard Business School Press, Boston, MA, pp. 1–35.

Robinson, James, 1999. The Corporate Practice of Medicine. University of California Press, Berkeley, CA.Robinson, Jeffrey D., Heritage, John, 2006. Physicians’ opening questions and patients’ satisfaction. Patient Educ. Couns. 60, 279–285.Sandefur, Rebecca L., Laumann, Edward O., 1998. A paradigm for social capital. Ration. Soc. 10, 481–501.Schwarze, Margaret L., Bradley, Ciaran T., Brasel, Karen J., 2010. Surgical ‘buy-in’: the contractual relationship between surgeons and patients that influences

decisions regarding life-supporting therapy. Crit. Care Med. 38, 843–848.Shapiro, Susan P., 2005. Agency theory. Annu. Rev. Sociol. 31, 263–284.Sharma, Anurag, 1997. Professional as agent: knowledge asymmetry in agency exchange. Acad. Manage. Rev. 22, 758–798.Simmel, Georg, 1950. The triad. In: Wolff, Kurt (Ed.), The Sociology of Georg Simmel, Glencoe. Free Press, IL, pp. 145–170.Sleath, Betsy, Svarstad, Bonnie, Roter, Debra, 1997. Physician vs. patient initiation of psychotropic prescribing in primary care settings: a content analysis of

audiotapes. Soc. Sci. Med. 44, 541–548.Stanton, Alfred, Schwartz, Morris, 1954. The Mental Hospital: A Study of Institutional Participation in Psychiatric Illness and Treatment. Basic Books, New

York.Stepanikova, Irena, 2012. Racial-ethnic biases, time pressure, and medical decisions. J. Health Soc. Behav. 53, 329–343.Stevens, Rosemary A., 2001. Public roles for the medical profession in the United States: beyond theories of decline and fall. Milbank Quart. 79, 327–353.Stevenson, Fiona A., Barry, Christine A., Britten, Nicky, Barber, Nick, Bradley, Colin P., 2000. Doctor–patient communication about drugs: the evidence for

shared decision making. Soc. Sci. Med. 50, 829–840.Stivers, Tanya, 2002. Participating in decisions about treatment: overt parent pressure for antibiotic medication in pediatric encounters. Soc. Sci. Med. 54,

111–130.Stivers, Tanya, 2007. Prescribing under Pressure: Parent–Physician Conversations and Antibiotics. Oxford University Press, Oxford.Szymczak, Julia E., Brooks, Joanna Veazey, Volpp, Kevin G., Bosk, Charles L., 2010. To leave or to lie? Are concerns about a shift-work mentality and eroding

professionalism as a result of duty-hour rules justified? Milbank Quart. 88, 350–381.Thompson, James D., 1967. Organizations in Action: Social Science Bases of Administrative Theory. McGraw-Hill, New York.Timmermans, Stefan., Epstein, Steven., 2010. A world of standards but not a standard world: toward a sociology of standards and standardization. Annu.

Rev. Sociol. 36, 69–89.Waring, Justin, Bishop, Simon, 2011. Healthcare identities at the crossroads of service modernisation: the transfer of NHS clinicians to the independent

sector. Sociol. Health Illn. 33, 661–675.Weissman, J.S., Blumenthal, D., Silk, A.J., Newman, M., Zapert, K., Leitman, R., Feibelmann, S., 2004. Physicians report on patient encounters involving direct-

to-consumer advertising. Health Aff. 23, W4-219–W4-233.Wennberg, John E., Fisher, Elliott S., Skinner, Jonathan S., 2004. Geography and the debate over medicare reform. Health Aff. 23, W96–W114.White, Harrison C., 1991. Agency as control. In: Pratt, John W., Zeckhauser, Richard J. (Eds.), Principals and Agents: The Structure of Business. Harvard

Business School Press, Boston, MA, pp. 187–221.Zerubavel, Eviatar, 1979. Private time and public time: the temporal structure of social accessibility and professional commitments. Soc. Forces 58, 38–58.



































When do doctors follow patients’ orders? Organizational mechanisms of physician influence

Documents