Search tips
Search criteria 


Logo of jgimedspringer.comThis journalToc AlertsSubmit OnlineOpen Choice
J Gen Intern Med. 2010 March; 25(3): 186–193.
Published online 2009 December 22. doi:  10.1007/s11606-009-1178-3
PMCID: PMC2839333

“We’ll Do this Together”: The Role of the First Person Plural in Fostering Partnership in Patient-physician Relationships

Helen Kinsman, BS,1 Debra Roter, DrPH,1 Gail Berkenblit, MD, PhD,1 Somnath Saha, MD, MPH,2 P. Todd Korthuis, MD,2 Ira Wilson, MD,3 Susan Eggly, PhD,4 Andrea Sankar, PhD,4 Victoria Sharp, MD,5 Jonathon Cohn, MD,4 Richard D. Moore, MD, MHS,1 and Mary Catherine Beach, MD, MPHcorresponding author1



Partnership is integral to therapeutic relationships, yet few studies have examined partnership-fostering communication behaviors in the clinic setting. We conducted this study to better understand how statements in which physicians use the first person plural might foster partnership between patient and provider.


We audio-recorded encounters between 45 HIV providers and 418 patients in the Enhancing Communication and HIV Outcomes (ECHO) Study. We used the Roter Interaction Analysis System (RIAS) to code for statements made by the physician that used the first person plural to refer to themselves and their patient. Using multiple logistic regression, we examined the associations between the occurrence of one or more first person plural statements with patient ratings of provider communication. To better understand the meaning of first person plural statements, we conducted a qualitative analysis.

Main Results

Providers were mostly white (69%) and Asian (24%); 57% were female. Patients were black (60%), white (25%), and Hispanic (15%); 33% were female. One or more first person plural statements occurred in 92/418 (22%) of encounters. In adjusted analyses, encounters with first person plural statements were associated with younger patient age (OR 0.97, 95% CI 0.94–0.99), higher patient depression scores (highest tertile compared to lowest tertile: OR 1.89, 95% CI 1.01–3.51), the patient not being on anti-retroviral therapy (OR 0.53, 95% CI 0.29–0.93), and older provider age (OR 1.05, 95% CI 1.00–1.09). After adjustment, patients were less likely to highly rate their provider’s communication style if first person plural statements were used (AOR 0.57, 95% CI 0.33–0.96). There were 167 first person plural statements made by physicians in the 418 encounters. Qualitative analysis revealed that many first person plural features had at least one negative feature such as being overtly persuasive (“That’s going to be our goal”), indirect (“What can we do to improve your diet?”), or ambiguous (“Let’s see what we can do”), although there were also positive statements that involved patients in the health-care process, contributed to a mutual understanding, and addressed the patients’ goals.


Contrary to our hypotheses, use of first person plural was not associated with higher ratings of provider communication, probably because some of these statements were overtly persuasive, indirect, or ambiguous. Physicians should become aware of benefits and pitfalls of using the first person plural with patients. Further research is needed to determine the most effective methods through which providers can build alliances with patients.

KEY WORDS: patient-provider relations, patient-provider communication, patient-centered care, partnership, HIV/AIDS

In lieu of a cancer-free bill, but with a good deal less of his old caution, he repeated that we were in far better shape than we might have been. I noted his first time use of the we and began comprehending the stake he’d had in my case from the start…

Reynolds Price, A Whole New Life1

Partnership, generally defined as the physician’s alliance with the patient,2 is part of any patient-centered interaction. As described by Mead and Bower, patient-centered encounters have at their core a sharing of power and responsibility between the doctor and the patient, and the building of a therapeutic alliance.3 Through partnership, providers can both share power and build a therapeutic alliance with their patients. As illustrated in the above quote, physicians’ use of the first person plural to refer to themselves in alliance with the patient may be one way in which partnership is fostered.

Although we may strive for true partnerships between health-care providers and patients, this may not be the natural state of these traditionally paternalistic relationships. A partnership must be cultivated by both parties in order to be effective. And yet with respect to patient-provider communication, one study found that it is mainly the provider who utilizes the word ‘we’ to reference the patient-provider relationship, while patients never used the first person plural to reference their relationship with their physician.4 Thus, it is predominantly the provider who uses this sort of partnership language in any given encounter.

To better understand how providers’ use of first person plural statements might be related to achieving partnership, we conducted both quantitative and qualitative analyses. We identified instances in which the first person plural was used, and then attempted to understand the underlying meaning and context. We hypothesized that patients would have higher ratings of care in the setting of first person plural statements, signifying the perception of an alliance.


Study Design, Subjects, and Setting

We conducted a cross-sectional qualitative and quantitative analysis of data collected as part of the Enhancing Communication and HIV Outcomes (ECHO) Study. Study subjects were 45 providers and 418 of their HIV-infected patients at four sites in the US (Baltimore, Detroit, New York, and Portland). The study received IRB approval from each of the four sites. Eligible providers were physicians, nurse practitioners, or physician assistants who provided primary care to HIV-infected patients. Overall, 82% of all providers across the four sites participated. Eligible patients were HIV-infected, older than 18, English-speaking, and had had at least one prior visit with their provider. Overall, 73% of approached eligible patients participated. The most common reasons for patient refusal were that they did not have time to complete the interview (65%), that they were not feeling well (13%), and that they were not interested in studies (8%).

Data Collection Methods

HIV providers who agreed to participate gave informed consent and completed a baseline questionnaire. Research assistants approached patients of participating providers in the waiting rooms, with the goal of enrolling ten patients per provider. After patients gave informed consent, research assistants audiotaped their encounter with their provider and conducted a 1-h post-visit interview to assess demographics, social, behavioral and clinical characteristics, and patient ratings of care.

Main Measures

Our main study measures are derived from audio-taped analysis of patient-provider encounters, provider baseline questionnaires, and post-encounter patient interview. Audiotapes collected as part of the ECHO study were analyzed using the Roter Interaction Analysis System (RIAS), which categorizes each statement made by the patient or provider into one of 34 mutually exclusive and exhaustive categories and has been well validated.2,58 The RIAS routinely and explicitly codes for the first person plural made by the provider under the category of ‘partnership’ statements.2 Statements containing variations on the first-person plural, such as ‘we,’ ‘us,’ ‘our,’ and ‘you and I,’ to refer to the patient and provider fell into this category. On the baseline questionnaires, providers’ self-reported basic demographic information (age, sex, type of training) as well as underlying characteristics such as empathic tendency were measured using a validated 14-item scale.9 Examples of empathy items are “I often have tender, concerned feelings for people less fortunate than me” and “Before criticizing someone, I try to imagine how I would feel if I were in their place;” possible responses are measured on a 5-point Likert scale and are anchored between ‘describes me very well’ and ‘does not describe me very well.’ In post-visit interviews, patients’ self-reported basic demographic information (age, sex, race/ethnicity, employment, and education), social support10, illicit drug use11, depressive symptoms12, use of antiretroviral therapy, self-efficacy in managing medication, and experience of care (ratings of their providers’ communication style,13 their providers’ participatory decision-making style (PDM)14, and their overall satisfaction).

Quantitative Analysis of Encounters with First Person Plural Statements

First, we used descriptive statistics to explore the characteristics of our study sample. We then compared visits in which one or more vs. no first person plural statements were made by the provider (‘first person plural encounters’) with respect to patient characteristics [self-reported age, sex, race/ethnicity, employment status, education, social support in tertiles, illicit drug use, depressive symptoms in tertiles, use of antiretroviral therapy (ART), and medication self-efficacy], provider characteristics (age, sex, type of training, and self-reported empathic tendency in tertiles), and relationship characteristics (length, age and sex concordance, and cultural dissimilarity). For all these analyses, we used logistic regression, adjusting for site and accounting for clustering of patients within providers using generalized estimating equations, with first person plural encounter as the dependent variable and patient, provider, and relationship characteristics as independent variables.

Next, we assessed the association of the first person plural with patient experience of care. Based on skewed response distributions, ratings of provider communication style and overall satisfaction were dichotomized at the highest value; ratings of PDM style were dichotomized at the median score. We used logistic regression with each patient-rated experience as a dependent variable and the first person plural encounter as the independent variable. Both bivariate and multivariate associations adjusted for site and accounted for clustering of patients within providers using generalized estimating equations. In multivariate analyses, we further adjusted for any patient, provider, or relationship characteristics associated with the use of the first person plural at the p < 0.20 level. Covariables that did not substantively alter the results and were not significant in multivariate analyses were removed from the final models. All analyses were conducted using Stata, Version 8.1.

Finally, after our qualitative analysis was complete, we conducted an exploratory analysis to assess the trustworthiness of our qualitative coding. We created a partnership score for each first person plural encounter based on the balance of positive or negative features defined in the qualitative analysis of the first occurance of a first person plural statement. Each positive feature contributed 1 point to the score, while each negative feature contributed (−1) point. Because there were four catgegories of positive features and three categories of negative features described in our qualitative analysis, the range of possible parternship scores was from −3 (representing statements that had three negative features and no positive features) to +4 (representing statements that had four positive features and no negative features). We repeated our logistic regression analysis, restricting our sample to first person plural encounters, with patient ratings of provider communication as the dependent variable and the partnership ‘score’ as the independent variable.

Qualitative Analysis of First Person Plural Statements

After we identified first person plural statements using the RIAS, we listened to each encounter in its entirety and transcribed each statement along with associated contextual elements. Two investigators (HK, MCB) then met to devise more specific coding categories for these statements based upon recurring themes in the dialogue. Categories were not mutually exclusive, so a particular statement could have more than one category assigned to it (including both positive and negative features). All statements were coded by both investigators, and disagreements were resolved by discussion and consensus. We developed a model to describe hypothesized themes suggested by our qualitative findings.


Study Sample and Prevalence of First Person Plural Statements

Patient, provider, and relationship characteristics of the entire study sample are shown in the first column of Table 1. The majority (326/418, 78%) of encounters did not contain any first person plural statements. In the remaining 92 encounters, there were 157 statements using the first person plural.

Table 1
Patient, Provider, and Relationship Characteristics Associated with First Person Plural Statements Made by Providers

Associations of Patient and Provider Characteristics with First Person Plural Statements

The unadjusted and adjusted associations of patient, provider, and relationship characteristics with the use of the first person plural are shown in Table 1. The use of the first person plural was not significantly associated with patient race/ethnicity, education, social support, or illicit drug use. The first person plural was used less frequently by providers in encounters with male patients, with older patients, with patients who reported high vs. low levels of depression, and with patients on anti-retroviral therapy. In unadjusted analyses, providers who used the first person plural were older, less likely to be physicians (compared to nurse practitioners or physician assistants), and more likely to be in the low vs. high tertile of empathic tendency. There were no associations between the use of the first person plural and length of patient-provider relationship, patient-provider age or gender concordance, or patient-provider cultural dissimilarity.

After adjustment for site and potentially confounding patient and provider characteristics, encounters with first person plural statements remained independently associated with younger patient age (OR 0.97, 95% CI 0.94–0.99), higher patient depression scores (highest tertile compared to lowest tertile: OR 1.89, 95% CI 1.01–3.51), not being on ART (OR 0.53, 95% CI 0.29–0.93), and older provider age (OR 1.05, 95% CI 1.00–1.09).

Association of First Person Plural Statements with Patient Experience of Care

Among the entire study sample, 40% of patients gave their physician the highest possible rating on a scale of communication quality, 49% rated their physician’s PDM style above the median, and 62% rated their overall satisfaction as excellent. The unadjusted and adjusted associations between first person plural statements and patient experience of care are shown in Table 2. Patients had lower odds of highly rating their provider’s communication style following encounters in which the provider made a first person plural statement (OR 0.57, 95% CI 0.33–0.96). First person plural statements were not significantly related to patients’ ratings their provider’s PDM style or their overall satisfaction.

Table 2
Association of First Person Plural Statements with Patient Experience of Carea

Characteristics of First-person Plural Statements

Characteristics of first person plural statements identified in qualitative analyses that may or may not foster partnership are summarized in Table 3 and described below. In Figure 1, we present a hypothesized model in which first-person plural statements contain positive features that seemed to foster partnership between the provider and the patient, and negative features that did not.

Table 3
Features of First Person Plural Statements
Figure 1
Model for first person plural dialogue. Patient-physician relationships occur within the context of an inherent power imbalance. Positive first person plural statements may shift the balance of power towards the patient, whereas negative first person ...

Features of First Person Plural Statements that May Foster Partnership

Involves Patient in Health-care Process In some instances (n = 61), the provider actively involved the patient in the health-care process, as in the following example: “And then we can look at the results together when you return. And then we can make a decision about what to do next.” We noted some visits in which the physician went to great lengths to ensure that the patient played a part in the decision-making process:

Provider: So I think that this is a decision between you, Dr. _, and me…to say, what do we think is the right way?

Patient: Right.

Provider: If he recommends treatment, then I think the four of us really need to discuss it. We need to talk about what the treatment involves. And what’ll happen. OK?

Creates a Common Understanding In some instances (n = 42), the first person plural was used by the provider to come to a common understanding with the patient, ensuring that they are on the ‘same page.’ Statements in which a common understanding was created include: “I’m going to reread the note to you so that we’re both on the same page” and “I want you to write all of your medicines in, and let’s go over them.”

Addresses the Patient’s Goals There were a smaller number of instances (n = 24) in which the provider purposefully addressed a goal that was identified as explicitly important to the patient. These statements differ significantly from statements in which providers steered the topic of conversation towards goals of their own. An example of this type of statement was coded when one doctor, noticing that her patient was under stress because of a recent move, decided to postpone talks about smoking until a later date: “Alright, well let’s take that off the table for today. And let’s get you moved.”

Reflection/Discussion of Shared Past The final group of statements found to facilitate partnership were those that referenced the relationship between the provider and the patient with regards to shared past experiences. Although this dialogue was infrequent (n = 7), its power to expose a sense of partnership between provider and patient was meaningful. In one instance, the physician and patient were discussing some difficulties in communication they were able to overcome:

Provider: I may have been a little curt, and if I was, that was not my intention. You know that.

Patient: I know, you are a good person, and you are a good provider, and you do talk…I can share things with you. I can tell you about every type of thing I’m going through.

Provider: Yeah! And we worked on it, like any relationship, it’s a process, right?

Patient: Yeah, and that’s why I was, you know—it happened a couple of time before. But I thought maybe she’s busy…but I understand now…

In another example, the doctor and patient were discussing how the patient had been able to turn his life around and was now adhering to his medication regimen:

Provider: But you’ve come around. And I think we work well as a team now.

Patient: Well you always was a team player. It was just me.

Provider: Well now we’ve got it together. That’s good.

Features of Partnership Statements that May Not Foster Partnership

Persuasion One way in which providers did not foster partnership with their patients was to use these statements as a method of persuasion. A substantial number of statements (n = 36) were found to be overtly persuasive in nature by advancing provider goals that were in conflict with patient goals. Often these statements were made when providers were trying to convince patients to change their behavior with regard to smoking, diet, and adherence to medication, and are in contrast to partnership statements where the provider aligns directly with the patients’ goals. In one example, the provider spent a significant amount of time during the visit trying to convince the patient to address her addiction to alcohol. The patient, however, had already been enrolled in Alcoholics Anonymous and was not interested in being in another treatment program. At the end of the interview the provider used the following language to convince to her seek treatment:

D: Now remember, OK? The most important thing…let’s solve the alcohol issue, OK? That’s our priority now. OK?

P: OK [reluctantly]

In another encounter, the provider attempted to persuade the patient to adhere to her medication by stating, “So that’s what we’re going to focus on right now,” yet there was never any indication made by the patient in the visit that she believed that ought to be their focus.

Indirect Communication In some instances, we found that providers used the first person plural in a manner that was indirect and inaccurate (n = 18). This included dialogue in which providers used the words ‘we’ and ‘our’ when they actually meant ‘you’ or ‘I.’ By misusing the first person plural, providers could make requests or demands of patients without being direct and explicit and without holding the patient accountable for their actions. A phrase like “What can we do to improve your diet?” is indirect in that it involves the doctor in an aspect of the patient’s daily life—diet—of which he is not a part. In another example, the provider tries to convince the patient to make a New Year’s resolution to quit smoking:

Provider: I think the next thing we really need to figure out is two things: the exercise and the smoking.

Patient: I knew you were gonna say that.

Provider: If we can make a New Year’s resolution by January that you’re going to stop…

Ambiguous Use We also found that the first person plural was sometimes used ambiguously (n = 49), when the use of key words such as ‘we’ and ‘our’ was often unclear and confusing. An ambiguous use of the word ‘we’ is exemplified in such statements as “Just so we can follow up on some of these issues.” and “Let’s see what we can do.” In these instances it is not made clear whether the word ‘we’ refers to the doctor and patient, the health-care team, or the doctor or patient alone. Therefore, the providers may use the first person plural in these instances to diffuse or avoid responsibility.

Overall Summary of Positive and Negative Features of First Person Plural Statements

We found that 104/157 (66%) statements had at least one positive feature, while 77/157 (49%) had at least one negative feature. Because positive and negative features were not mutually exclusive, there was a total of 42/157 statements that were coded as having both a positive and negative feature. Figure 2 presents the range and frequency of partnership scores for the first occurance of the first person plural in each of the 92 first person plural encounters. There was a significantly increased odds of the patient rating the provider’s communication skills highly with each one-point increase in the partnership score (OR 1.46, 95% CI 1.13–1.90).

Figure 2
Distribution of partnership scores in 92 encounters with first person plural statements.


The use of the first person plural often does not lead to more effective patient-provider communication. To our surprise, in encounters where the first person plural statements were used, patients were less likely to rate their provider’s communication skills highly. The use of these types of statements did not positively or negatively affect patient ratings of provider’s PDM style or their overall satisfaction. Although we hypothesized that this type of dialogue would foster the sense of an alliance between the provider and the patient, our analyses suggest otherwise.

These surprising findings might be explained by our qualitative analysis, which delineated features of first person plural dialogue that may hinder the sense of true partnership between patient and provider. We found that first person plural statements were sometimes used by providers in a manner that appeared to be persuasive, indirect, patronizing or condescending. Although we found that first person plural dialogue has both positive and negative features, it was the negative features that were most surprising to us, and which we believe help explain the negative association with patients’ experience of care.

As depicted in Figure 1, persuasive statements seemed to hinder communication by moving the focus of care from the patient’s wishes and values to those of the provider. We also considered instances of persuasion and indirect communication, in which providers made implicit requests of the patient, to be disempowering for the patient. Within the contexts of medication adherence, weight loss, smoking, and substance abuse, first person plural dialogue did not encourage an appropriate amount of responsibility and self-efficacy from the patient. The use of the word “we” falsely involved the provider in activities that are practically the responsibility of the patient. Under these circumstances, it appears that more direct language is desirable, language that honestly acknowledges the patient’s authority to take control of the situation.

In other instances we identified first person plural statements that appeared to foster a true sense of alliance. Statements that occurred in the context of purposefully addressing the patient’s goals, sharing medical decision-making, and legitimizing the patient-provider relationship seemed to contribute to a sense of equality between the two parties. Many of these conversations represented instances during which providers shared power with patients, giving them autonomy and control. These sorts of efforts on the part of physicians are inherently patient-centered and truly involve patients in their own care. When analyzed in the context of an overall partnership score that weighs both positive and negative statements, it appears that positive uses of the first person plural increase the patient’s odds of highly rating provider communication skills. These types of statements exemplify how a provider can better communicate with patients in order to forge a therapeutic alliances.

In our quantitative analysis, we also found that first person plural statements were used most frequently by older providers, and with patients who were younger, were more depressed, and had less social support. Extrapolating from these findings, we emphasize that first person plural statements often occur in the context of greater power inequalities between patients and providers. Surprisingly, we also noted that providers who used first person plural statements reported lower empathic tendencies. These quantitative findings are more understandable in light of our qualitative analysis. Often in these visits, the patients seemed disengaged, while the provider was attempting to engage them in some goal, without seeming to understand that the patient wanted something else. Given how important, yet dysfunctional, this dialogue seems, we suggest that providers be offered training in communication with patients regarding difficult behavior changes. In addition to training in more effective behavior change counseling strategies, medical educators may wish to consider explicitely noting the negative and positive uses of the first person plural with their students.

Several limitations should be considered when interpreting these findings. First, we sometimes found it difficult to determine the intentions or impact of particular first person plural statements. We attempted to address this issue by listening to the entire encounter, and transcribing dialogue preceding and following each statement in an effort to establish its context. However, we recognize that these statements are made in the context of an ongoing relationship with a history that is largely unknown to us. Also, the generalizability of our findings may be reduced by our HIV-specific sample of providers and patients. We wonder if this particular group of subjects may have unique tendencies based on patient demographics, physician training, and frequency of clinic visits. Although we consider this to be a possibility, we feel that most our findings are applicable outside of the realm of HIV care.

We consider our study to have important implications for the complex task of achieving partnership in patient care. Although physicians may attempt to foster an alliances with patients through the use of first person plural statements, it appears that this strategy is not always successful. Despite the sense of alliance that Reynolds Price describes in response to his physician’s use of the first person plural,1 his experience may not be universal. Provider’s use of the word “we” does not automatically create a sense of partnership between patient and provider. It is ultimately the intentions of the provider and his or her willingness to share power with a patient that forms the foundation for a true healthcare alliance. In order to empower patients, especially among vulnerable populations, providers must actively and explicitly cultivate a positive sense of alliance and shared power. By prioritizing the sharing of power and minimizing persuasive, dominating dialogue, providers can create a more patient-centered environment.


This work was supported by the Johns Hopkins University Osler Center for Clinical Excellence and by AHRQ contract 290-01-0012. In addition, Dr. Beach’s effort was supported by K08 HS013903-05, and Ms. Kinsman’s effort was supported by the Predoctoral Clinical Research Training Program at Johns Hopkins (UL1-RR025005).

Conflict of Interest None disclosed.


1. Price R. A Whole New Life: An Illness and a Healing. New York: Scribner; 1982.
2. Roter DL. Roter Interaction Analysis Manual. Johns Hopkins University Bloomberg School of Public Health; 1999.
3. Mead N, Bower P. Patient-centredness: a conceptual framework and review of the empirical literature. Soc Sci Med. 2000;51:1087–1110. doi: 10.1016/S0277-9536(00)00098-8. [PubMed] [Cross Ref]
4. Skelton JR, Wearn AM, Hobbs FD. "I" and "we": a concordancing analysis of how doctors and patients use first person pronouns in primary care consultations. Fam Pract. 2002;19:484–488. doi: 10.1093/fampra/19.5.484. [PubMed] [Cross Ref]
5. Roter DL, Hall JA, Katz NR. Relations between physicians’ behaviors and analogue patients’ satisfaction, recall, and impressions. Med Care. 1987;25:437–451. doi: 10.1097/00005650-198705000-00007. [PubMed] [Cross Ref]
6. Wissow LS, Roter DL, Wilson ME. Pediatrician interview style and mothers’ disclosure of psychosocial issues. Pediatrics. 1994;93:289–295. [PubMed]
7. Wissow LS, Roter D, Bauman LJ, et al. Patient-provider communication during the emergency department care of children with asthma. The National Cooperative Inner-City Asthma Study, National Institute of Allergy and Infectious Diseases, NIH, Bethesda, MD. Med Care. 1998;36:1439–1450. doi: 10.1097/00005650-199810000-00002. [PubMed] [Cross Ref]
8. Roter DL, Stewart M, Putnam SM, Lipkin M, Jr, Stiles W, Inui TS. Communication patterns of primary care physicians. JAMA. 1997;277:350–356. doi: 10.1001/jama.277.4.350. [PubMed] [Cross Ref]
9. Turknett Leadership Group, Emotional Intelligence Quiz. 2009.
10. HCSUS Baseline Questionnaire. Module 9: Social support and coping. Available at; accessed October 29, 2009.
11. Cacciola JS, Alterman AI, McLellan AT, Lin YT, Lynch KG. Initial evidence for the reliability and validity of a "Lite" version of the Addiction Severity Index. Drug Alcohol Depend. 2007;87:297–302. doi: 10.1016/j.drugalcdep.2006.09.002. [PubMed] [Cross Ref]
12. Andresen EM, Malmgren JA, Carter WB, Patrick DL. Screening for depression in well older adults: evaluation of a short form of the CES-D (Center for Epidemiologic Studies Depression Scale) Am J Prev Med. 1994;10:77–84. [PubMed]
13. Stewart AL, Napoles-Springer A, Perez-Stable EJ. Interpersonal processes of care in diverse populations. Milbank Q. 1999;77:305–339. doi: 10.1111/1468-0009.00138. [PubMed] [Cross Ref]
14. Heisler M, Bouknight RR, Hayward RA, Smith DM, Kerr EA. The relative importance of physician communication, participatory decision making, and patient understanding in diabetes self-management. J Gen Intern Med. 2002;17:243–252. doi: 10.1046/j.1525-1497.2002.10905.x. [PMC free article] [PubMed] [Cross Ref]

Articles from Journal of General Internal Medicine are provided here courtesy of Society of General Internal Medicine