Strategies and approaches for saying no to antidepressant requests. Although 88 visits did not lead to a prescription, only 84 transcripts were available for analysis and are included in this analysis.
Examples of strategies for denying requests. SP indicates standardized patient.
Standardized patient report of excellent visit satisfaction in each denial approach.
Paterniti DA, Fancher TL, Cipri CS, Timmermans S, Heritage J, Kravitz RL. Getting to “No”Strategies Primary Care Physicians Use to Deny Patient Requests. Arch Intern Med. 2010;170(4):381–388. doi:10.1001/archinternmed.2009.533
Physicians need strategies for addressing patient requests for medically inappropriate tests and treatments. We examined communication processes that physicians use to deal with patient requests of questionable appropriateness.
Data come from audio-recorded visits and postvisit questionnaires of standardized patient visits to primary care offices in Sacramento and San Francisco, California, and Rochester, New York, from May 2003 to May 2004. Investigators performed an iterative review of visit transcripts in which patients requested, but did not receive, an antidepressant prescription. Measurements include qualitative analysis of strategies for communicating request denial. The relationship between strategies and satisfaction reports in postvisit questionnaires was examined using the Fisher exact test.
Standardized patients requested antidepressants in 199 visits; the antidepressants were not prescribed in 88 visits (44%), 84 of which were available for analysis. In 53 of 84 visits (63%), physicians used 1 or more of the following 3 strategies that explicitly incorporated the patient perspective: (1) exploring the context of the request, (2) referring to a mental health professional, and (3) offering an alternative diagnosis. Twenty-six visits (31%) involved emphasis on biomedical approaches: prescribing a sleep aid or ordering a diagnostic workup. In 5 visits (6%), physicians rejected the request outright. Standardized patients reported significantly higher visit satisfaction when approaches relying on the patient perspective were used to deny the request (P = .001).
Strategies for saying no may be used to communicate appropriate care plans, to reduce provision of medically inappropriate services, and to preserve the physician-patient relationship. These findings should be considered in the context of physician education and training in light of increasing health care costs.
Patient requests for clinical care, including requests for medication, can influence physicians' decisions about treatment.1 Patients make a request for medication in roughly 10% of office visits, and most requests are honored.2,3 Medications prescribed at the behest of patients may not always represent physicians' first choice of treatment, particularly if the requests are commercially motivated, as for example, by direct-to-consumer advertising (DTCA).2,4,5 Nevertheless, physicians are cautious when rejecting patient requests for services, in part because of physicians' perception that rejection may lower patient satisfaction.6,7 Yet, data from patients are conflicting: nonfulfillment is associated with lower patient satisfaction in some studies7- 10 but not in others.11 It may be that patient satisfaction has more to do with what the physician says or does not say than whether or not an expected prescription is offered.12
In the popular business book, Getting to Yes: Negotiating Agreement Without Giving In, Fisher et al13 present general strategies for bridging gaps and achieving compromise in everyday life. In the clinical setting, however, “getting to yes” is not always appropriate for at least 3 reasons. First, on ethical grounds, physicians have a duty to avoid doing harm and to maximize patient benefit. This duty may conflict with other ethical obligations, such as respecting patient autonomy. Nevertheless, bioethicists are nearly unanimous that physicians are not obligated to provide unnecessary or inappropriate care.14 Second, from a safety perspective, physicians must exercise caution when prescribing new, poorly tested, or marginally indicated medications.15 Third, from a policy perspective, achieving control of health care costs is a critical national priority.16 Physicians must balance the needs of individuals with those of society; in some cases, serving as social stewards might mean forgoing otherwise clinically indicated but very expensive care. Therefore, judicious clinical restraint in the patient visit may benefit the patient and will be a cornerstone of any effective cost-containment program.
Nevertheless, “getting to no” is not easy, and, to our knowledge, there are no studies directly examining the approaches that physicians use in everyday practice. This article examines the conversational and clinical rejection strategies that physicians use in everyday practice to deal with patient requests that they do not wish to fulfill.
To examine pathways to rejection and to identify strategies that will allow physicians to maintain control of the treatment plan while potentially preserving patient satisfaction, we analyzed data from a randomized trial on the prescribing behavior of primary care physicians in response to standardized patient (SP) requests for antidepressant medication.1 Data included transcripts from office visits in primary care practices and postvisit questionnaires with measures of SP-reported visit satisfaction. The SPs were scripted and trained to portray 2 different clinical roles (major depression with wrist pain or adjustment disorder with back pain) while making 1 of 3 different requests (brand-specific antidepressant medication request, general request for antidepressant medication, or no request).1 Investigators told SPs that they were interested in an array of physician responses to the different clinical roles and request types. The SPs who were enacting the 2 roles were trained separately and portrayed a single role for the entire study. Although the SPs were not blinded to the study design, they were not aware of specific study hypotheses. Those SPs who were instructed to make an antidepressant request were asked to make their initial request within the first 10 minutes of the visit or before the physical examination and to make a second request during the office visit if the first request did not lead to a prescription for an antidepressant.
Data for the randomized trial were collected between May 2003 and May 2004. A total of 152 primary care physicians consented to participate in a study using unannounced SPs to evaluate social influences on practice and competing demands in primary care. Internists and family physicians were recruited through 4 physician groups: University of California, Davis, Primary Care Network and Kaiser-Permanente in Sacramento, California; Brown and Toland Medical Group in San Francisco, California; and Excellus BlueCross BlueShield in Rochester, New York. Cooperation rates by site ranged from 53% to 61%.
Eighteen insured, middle-aged, white, female SPs were trained and randomly assigned to make 298 unannounced visits, so that most physicians enrolled in the study saw 1 patient with depression and 1 with adjustment disorder. The SPs scheduled new visits to physicians and presented with subacute fatigue and insomnia accompanied by an unrelated orthopedic complaint referable to low back strain or carpal tunnel syndrome. Details on their training and detection rates are described elsewhere.17 Visits were digitally recorded using a concealed recorder; recordings were transcribed verbatim for analysis.
A summary of the trial study results is presented in the Table. Findings related to prescribing behaviors and request types,1 shared decision-making behaviors,18,19 physician self-reflection and rationale related to prescribing behaviors,20 and exploration of suicide21 are provided in detail elsewhere.
Visit transcripts were inductively reviewed and assessed for important visit components (information gathering about the physical complaint, depression-related symptoms, patient perspective related to complaint; inquiry into the nature of the advertisement or the context of the patients' complaints; information giving about depression, antidepressants, or sleep medications; presentation of a tentative diagnosis; and discussion of a treatment plan, including patient understanding, subsequent follow-up, and the possibility for prescribing an antidepressant). The order of these components and their relationship to presenting symptoms and request type (brand-specific or general antidepressants) were also noted. Patient requests and physician responses were abstracted from the transcripts, and a qualitative content analysis of physician responses was performed. The content analysis included development of an exhaustive list of how physicians went about denying patient requests. This list of approaches for denying requests was systematically reviewed and categorized into 3 strategic pathways to no, which were analyzed by a medical sociologist (D.A.P.) and a physician (T.L.F.) using a systematic and iterative approach to content analysis. Patterns and themes were further reviewed by all coauthors, and a final set of approaches and strategies was established by consensus. The 3 pathways to no and substrategies are outlined in Figure 1 and detailed in the “Results” section. Reviewers coded the transcripts blinded to outcome measures of patient satisfaction, to which each approach was later correlated in the analysis.
Previous work indicates that request nonfulfillment diminishes patient satisfaction, that patient-centered communication enhances it, and that SP satisfaction is correlated with the satisfaction of real patients seeing the same physician.1,22 We therefore hypothesized that certain forms of request denial would be associated with lower SP-reported satisfaction. We anticipated that an approach to request denial that incorporated aspects of the patient's interpretation of the chief complaint (“feeling tired”) could result in a preserved relationship between the physician and the patient and therefore in higher reports of SP satisfaction.
We investigated whether there was any relationship between postvisit SP satisfaction and 1 of 3 decision pathways (patient perspective based, biomedically based, or outright rejection). In the original study, the SPs completed two, 5-point Likert-scaled items for physician satisfaction: “Thinking about the visit you just made, how would you rate the physician in terms of your overall satisfaction with care [1, excellent; 5, poor]?” “Would you want this physician for your own personal physician [1, yes, definitely; 5, no, definitely not]?” The sum of these 2 items produced a reliable scale (mean [SD], 7.12 [2.30]; range, 2-10; α = 0.90) that was skewed strongly positive. We therefore split the sample near the 75th percentile to produce a dichotomous variable indicating excellent satisfaction (scale score, 9 or 10) vs less than excellent satisfaction (scale score, <9). The relationship between pathways to no and excellent satisfaction was examined using the Fisher exact test, as implemented in Stata version 10.0 (Stata Corp, College Station, Texas).
A request for medication was made in 199 (68%) of the office visits; in 88 (44%) of those visits, the request was denied. Four of the 88 visits were only partially transcribed or unavailable for transcription owing to poor recording quality, leaving 84 visits. Of the 84 visits, 54 were to general internists and 30 were to family physicians; 59 were to male physicians and 25 were to female physicians. The age, sex, and specialty distributions of the 84 visits in which requests were denied were similar to those of the other visits (P > .40 in all cases).
Each visit opened with the same chief complaint of “feeling tired” plus a physical complaint of either wrist pain (presented with symptoms of major depression) or low back pain (presented with symptoms of adjustment disorder). Physician review of both chief complaints occurred in 96% (81) of all visits. Physicians' statements about antidepressants after SP requests included comments emphasizing the problems with antidepressant use (ie, costs, delayed onset of benefit, long-term adherence requirements, and lack of efficacy for “feeling tired” or for problems of “mild,” “situational,” or “short-term” depression) and overall reluctance to prescribe antidepressants (“I’m not a pill doctor”; “I just think they [antidepressants] are overused.”)
Physicians used 3 strategic pathways for denying patients' requests for antidepressants: patient perspective-based strategies (63%), biomedically based strategies (31%), or outright rejection (6%). Figure 1 illustrates the 3 approaches, which are detailed below. Specific examples from visit transcripts of the content and how physicians said no are provided in Figure 2.
In 53 visits (63% of the 84 total visits), physicians gathered additional data about the request and its origin and offered information tailored to the patient's presentation of information. Three approaches emphasizing the patient's perspective on “feeling tired” or about the rationale for requesting antidepressants included (1) exploring the context of the request, (2) seeking the advice of a counselor or mental health specialist, and (3) offering an alternative diagnosis to major depression. These approaches presume an implicit validation of depression as the appropriate diagnosis and maintain the patient's interpretation and perspective at the core of the physician response. The most frequent of the 3 approaches, exploring the context of the request, occurred in 34 of the 84 visits (40%). Physicians' attempts to understand the original context of the request (eg, “Where did you see the ad?” “What about the ad rang true for you?”) and inquiries about recent events leading to the visit were often followed by a negotiated timeline for addressing the patient's symptoms, some including the possibility of prescribing an antidepressant at a later date.
Referral to a counselor or mental health professional occurred in 10 of the 84 visits (12%). Eight (or 80%) of these 10 referrals came from physicians in a health maintenance organization. Physician justifications for referral included having the patient consult with someone who could “go over things” and “make a recommendation [to the physician] about the appropriateness of medication” along with the benefit of seeing someone who might provide ways to deal with stress through “skills not pills.” Physicians provided extensive information about reasons for suggesting counseling and frequently told the patient that the referral was an opportunity for her to “talk things out with someone.”
A third strategy that made use of patient perspectives included rejection of the request for an antidepressant by offering an alternative diagnosis of “situational” or “mild” depression as the reason for the patient's chief complaint (9 visits, or 11% of the 84 total visits). In all but 1 of these 9 visits, the SP portrayed a patient with an adjustment disorder. Physicians typically followed the alternative diagnosis with specific reasons for rejecting the patient's request, including discussing the symptoms of major depression and reiterating contextual factors described by the patient to support the alternative diagnosis.
In 26 visits (31% of the 84 total visits), physicians used 1 of 2 biomedically based approaches to justify rejecting the request: prescribing a sleep aid (often a sedative-hypnotic, sometimes trazodone or a low-dose tricyclic agent) or ordering a diagnostic workup to rule out alternative medical illness. In the first approach, 15 physicians (18% of visits) prescribed a sleep aid, sometimes with a sleep hygiene handout, to address the patient's chief complaint of feeling tired. During these visits, physicians emphasized the ineffectiveness of antidepressants or provided justification of treatment with sleep aids over antidepressants for fatigue. Physicians instructed the patient to “try the sleep aid” and “see how you respond to it.” Some physicians even remarked that they were giving the patient an “old-fashioned antidepressant” (ie, a low dose of trazodone). Frequently, a physician would claim that fatigue might be related to sleep disturbance caused by the musculoskeletal pain, addressing both of the patient's complaints simultaneously.
In the second approach, physicians ordered diagnostic tests to rule out thyroid disease, anemia, menopause, or diabetes. Physicians frequently acknowledged the patient request as having some merit (eg, “That's what I was thinking” “That's a possibility”) and then suggested that the request might be fulfilled if the patient first followed the physician's plan. A diagnostic workup implicitly presumes that a physical condition is to blame for the symptoms and may be, from the patient's perspective, a lesser validation of the request. Overall, physicians ordered 1 or more laboratory tests in 68 of 84 visits (81%). Physicians described plans for a diagnostic workup as the primary reason for delaying attention to a patient request for an antidepressant in 11 of the 84 encounters (13%).
In 5 of the 84 visits (6%), physicians rejected patient requests without explanation and quickly shifted the topic to investigation of the patient's musculoskeletal complaint (eg, “Let's go through and do an examination”; “What about this low back pain?”) or further exploration of the patient history unrelated to depression history or its context. Interestingly, all 5 patient requests that generated these physician responses were general requests (eg, “Do you think medication would help me?”), not brand-specific requests, for an antidepressant.
The relationship between approaches to no and excellent visit satisfaction was examined using the Fisher exact test. The 26 visits with scores of 9 or 10 were classified as excellent satisfaction, and the remaining 58 with scores of less than 9 were classified as less than excellent satisfaction. The SPs were significantly more likely to report excellent visit satisfaction with approaches involving the patient perspective-based strategy (Figure 3). When the approaches were dichotomized into patient perspective-based and other strategies (combining the 5 outright rejection visits with the 3 biomedically based approaches), the SPs reported excellent visit satisfaction in 43% of the visits in which patient perspective-based approaches were used and in 10% of the visits in which other approaches were used (P = .001).
Physicians cannot always fulfill patient requests. However, little is known about the approaches that physicians use to issue denials. In this qualitative analysis of 84 office visits, physicians used 6 approaches for denying requests for antidepressants. These approaches for getting to no were classified as patient perspective based, biomedically based, or outright rejection based on the primary reason that the physician provided for denying the patient's request. The SPs reported significantly higher visit satisfaction when the physician used a patient perspective-based strategy to deny their request for antidepressants (P = .001).
Unfulfilled requests may have consequences for the physician-patient relationship, and physicians must learn to manage these requests in a respectful and clinically sensible fashion. Unfulfilled requests have been associated with reduced satisfaction in some studies23 but not in others.24- 26 A vignette-based study by Shah et al,27 in which a patient was denied a DTCA-based request, showed evaluations of care to be significantly associated with physician communication style: shared decision-making styles led to better evaluations of care. Gallagher et al6 examined physician responses to patient requests for an expensive, unindicated test. While few physicians ordered the test, most referred the patient to a specialist, and a significant minority explored the patient's narrative further. A recent study by van Bokhoven et al28 suggests that primary care providers sometimes underestimate how much their communication strategies might contribute to the well-being of their patients. Physicians may choose to fulfill inappropriate requests when they believe that the patient expects to have his or her request fulfilled.29- 31Yet, 1 survey study found that primary care physicians were less receptive to questions originating from DTCA and to requests to prescribe a specific medication.32 Some patient requests may be ill founded for a variety of reasons. Furthermore, learning to say no may increasingly become a strategy for bringing down the costs of medically inappropriate treatment33 and for promoting more conservative prescribing practices while maintaining a positive physician-patient relationship.15
Our study has several limitations. First, we do not know what physicians were actually thinking during the encounters or what they might have done in subsequent visits; we describe only what was said during a single “new patient” visit in which the SP's request was denied. Second, the data that we analyzed do not include information related to nonverbal cues or intonation that may be important to denying a request. Third, because we studied medication requests, it is not clear whether these approaches apply when patients request specific procedures or referrals for care. Fourth, all the SPs were middle-aged white women; physicians may respond differently to men or nonwhites.34- 37 Fifth, our measure of SP visit satisfaction could be an artifact of the actor's training, what the SPs knew about the study hypotheses, the SP's past experience with the health care system or depression, or the amount of time that the physician spent with the SP during the office visit. The role of an SP is bound by 2 principal parameters: (1) maintenance of a specific patient role and (2) genuine evaluation of the health care provider based on role expectations and real experience as a patient. Although postvisit SP ratings have been shown to differ from real patients' ratings, SP ratings are more reliable than a single, postvisit report by a real patient.22 Sixth, an obstacle to examining patient satisfaction includes the problem of ceiling effects for satisfaction measures. The mean satisfaction of SPs whose request was denied was quite high (7 of 10). Despite these high ratings overall, the SPs expressed greater satisfaction with some visits and approaches to request denial over others. Finally, because DTCA has increased since the period of data collection for this study and because recent studies have found physicians to be less receptive to fulfilling DTCA-driven requests,32 it is possible that physicians have developed additional strategies for saying no that are not presented in this analysis.
Getting to no does not mean that physicians do not convey interest in and concern for the patient. This article highlights a limited number of strategies and various approaches that physicians might use to deny patient requests. Because requests were scripted, differences in patient communication style and strategies were minimized.17 However, it would be almost impossible to do a real-time study of patient request making and physician denials using actual clinical encounters, as investigators would need to record hundreds of encounters simply to collect a handful of overt requests followed by denial. A study of 559 patients, with a new or worsening problem or suspicion of an undiagnosed disease, found that among the 545 patient requests for physician action, 13% (70 requests) were denied, skirted, or incompletely filled.3 A secondary finding from our study may deserve further investigation. Although relatively small in number (8 of the total 84 visits), all visits in which patients were referred to a mental health specialist occurred in a health maintenance organization. It is possible that in other practice settings, perceived time pressures or restricted access to mental health specialists may limit using this approach to request denial.
Our study describes strategies to get to no as a way of negotiating with patients about a specific request for treatment. Elucidation of these strategies provides a more nuanced understanding of physician-patient communication and negotiation than has been described previously. Furthermore, our findings may provide approaches for dealing not only with inappropriate requests but also with other types of difficult encounters in primary care settings.32,38 Physicians may become trapped in routine approaches to rejecting requests, and patients may vary in their reaction to different denial strategies. For example, a patient might prefer further investigation by laboratory work to rule out alternative diagnoses over referral to a mental health specialist to discuss coping skills for dealing with fatigue. Further research is needed to determine whether matching communication strategies to patient preferences or concerns results in less conflict and better ratings of interpersonal care and communication.
In an era of increasing constraints on health care systems and practitioners and significant influence of DTCA, learning to say no to patient requests will become more important. These strategies provide physicians with alternatives for saying no to patient requests for care that is perceived to be inappropriate, offering physicians an opportunity to select approaches that fit their own style of communication, the preferences of particular patients, or changing organizational climates. Knowledge of these strategies also offers physicians alternatives for denying potentially inappropriate requests and for preserving the physician-patient relationship when a current strategy or routine approach does not seem to be accepted by the patient.
Correspondence: Debora A. Paterniti, PhD, Division of General Medicine, University of California, Davis, 4150 V St, Ste 2500, Sacramento, CA 95817 (email@example.com).
Accepted for Publication: December 7, 2009.
Author Contributions: Drs Paterniti and Fancher contributed equally to the creation of this manuscript—the order of their names in the byline is entirely arbitrary. Study concept and design: Paterniti, Fancher, and Kravitz. Acquisition of data: Kravitz. Analysis and interpretation of data: Paterniti, Fancher, Cipri, Timmermans, Heritage, and Kravitz. Drafting of the manuscript: Paterniti, Fancher, and Kravitz. Critical revision of the manuscript for important intellectual content: Paterniti, Cipri, Timmermans, Heritage, and Kravitz. Statistical analysis: Kravitz. Obtained funding: Kravitz. Administrative, technical, and material support: Paterniti, Fancher, Cipri, Timmermans, Heritage, and Kravitz. Study supervision: Paterniti and Kravitz.
Financial Disclosure: Dr Kravitz has received unrestricted research grants from Pfizer during the past 3 years.
Funding/Support: Drs Fancher and Paterniti were supported by National Institute of Mental Health grant P50MH073511. Dr Kravitz was supported by National Institute of Mental Health grant K24MH72756. Original data collection was supported by a grant to Dr Kravitz (R01 MH064683) from the National Institute of Mental Health.
Role of the Sponsor: The funding agencies had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.
Previous Presentation: This study was presented in part as an oral presentation at the 2009 Society of General Internal Medicine National Meeting; May 13-16, 2009; Miami, Florida.
Additional Contributions: The authors thank Blue Shield of California; the University of California, Davis, Primary Care Network; Western Health Advantage; Kaiser Permanente; Brown & Toland IPA; and Excellus BlueCross BlueShield. We are also indebted to 18 SPs; to participating physicians and their office staff members; to Nolan Zane, PhD (Asian American Center for Disparities Research), for data analysis and manuscript preparation; and to Mark Henderson, MD, for his thoughtful comments on an earlier draft.