It takes two to tango: A dyadic approach to understanding the medication dialogue in patient-provider relationships

Abstract

Objective

To describe typologies of dyadic communication exchanges between primary care providers and their hypertensive patients about prescribed antihypertensive medications.

Methods

Qualitative analysis of 94 audiotaped patient-provider encounters, using grounded theory methodology.

Results

Four types of dyadic exchanges were identified: Interactive (53% of interactions), divergent-traditional (24% of interactions), convergent-traditional (17% of interactions) and disconnected (6% of interactions). In the interactive and convergent-traditional types, providers adopted a patient-centered approach and used communication behaviors to engage patients in the relationship. Patients in these interactions adopted either an active role in the visit (interactive), or a passive role (convergent-traditional). The divergent-traditional type was characterized by provider verbal dominance, which inhibited patients' ability to ask questions, seek information, or check understanding of information. In the disconnected types, providers used mainly closed-ended questions and terse directives to gather and convey information, which was often disregarded by patients who instead diverted the conversation to psychosocial issues.

Conclusions

This study identified interdependent patient-provider communication styles that can either facilitate or hinder discussions about prescribed medications.

Practice Implications

Examining the processes that underlie dyadic communication in patient-provider interactions is an essential first step to developing interventions that can improve the patient-provider relationship and patient health behaviors.

1. Introduction

Collaborative patient-provider communication [including patient-centeredness and shared decision-making], plays an important role in helping patients with chronic diseases develop an accurate understanding of their disease and adhere to their treatment plan.[1–4] Despite the fact that it is dyadic, communication research has typically focused on either the provider (e.g., types of questions asked) or patient (e.g., type of information given) during clinical interactions. [5, 6] However, in a conversation patients and providers influence each other’s thoughts, attitudes, emotions, and behaviors.[7] Conversation analysis views patient-provider communication as a dyadic process that gives equal consideration to both partners.[8, 9] This approach emphasizes the “co-construction of narratives” (i.e., stories jointly constructed) that are influenced by speaker and listener characteristics (e.g., age, gender), based on prior conversations, contextually-bound, and continuously being refined.[10–12]. In this study we sought to empirically develop a typology of dyadic exchanges that describes how primary care providers and their hypertensive patients discuss antihypertensive medications. This process was modeled after social science research that has identified typologies of dyadic interactions.[8, 10, 13, 14]

2. Methods

Data were collected as part of an observational study conducted in three ambulatory care centers in New York City[15]. Eligible patients: 1) self-identified as non-Hispanic African American or White; 2) were diagnosed with hypertension and taking ≥1 antihypertensive medication; 3) were ≥18 years of age and; 4) had at least one prior visit with the participating provider. All MDs/DOs or nurse practitioners providing care to enrolled patients were eligible. The Institutional Review Board of New York University Langone Health approved the study.

We audiotaped clinical interactions between 29 providers and 94 hypertensive patients (mean 3.44 patients per provider). After obtaining informed consent, research assistants started the tape recorder in the exam room and then left. Providers and patients were able to turn off the recorder at any time; 3% of patients requested this option. Patient and provider demographic data were collected prior to the audiotaped interaction. Patient-level data included age, sex, race/ethnicity, employment status, educational and income level, insurance status, and length of relationship with their provider. Data on antihypertensive medications and blood pressure were abstracted from patients’ medical record. Provider-level data included age, sex, race/ethnicity, and duration of practice at the site. In addition, data on length of the interaction (i.e., total time in minutes from the first spoken utterance of the conversation to the final utterance) and talk-time (i.e., total number of utterances spoken by patients vs. providers) were derived from the audiotaped interactions. Utterances were coded for each independent and every non-restrictive dependent clause of a sentence spoken by the patient or provider.

The audiotaped interactions were transcribed verbatim and analyzed in two steps using the grounded theory constant comparison method. [16, 17] In the first step, two members of the research team with expertise in behavioral medicine and ethnographic research reviewed the transcripts to identify overarching themes related to patient-provider communication behaviors (e.g., question-asking) and discussions about medication-taking (e.g., side effects). The coders iteratively developed a codebook to maintain coding consistency and transparency in coding decisions.[18] Eighteen (20%) transcripts were coded by a third reviewer to verify the themes. Discrepancies in coding were resolved through an interactive process of re-reading and discussing the transcripts until consensus was reached; the codebook was updated to reflect any changes. In the second step, two other members of the team reviewed the coded transcripts to identify dyadic communication behaviors that patients and providers would use in response to their partner’s behavior (e.g., content of information given in response to provider questioning). As in step 1, a subset of transcripts was double-coded by an independent reviewer to verify the analysis. For both steps, assessments of inter-rater agreement were calculated using Krippendorff’s alpha to ensure an acceptable level of agreement was reached (>0.80) between the coders.[19]

3. Results

Table 1 shows participant and visit characteristics for the 94 interactions. Below we describe the typologies and the reciprocal patient and provider communication behaviors characteristic of each type.

Table 1.

Comparison of participant and visit characteristics by typology of patient-provider communication

	Total Sample (N=94)	Interactive (n=54)	Convergent- Traditional (n=15)	Divergent- Traditional (n=25)	Disconnected(n=7)	P1
Patient Characteristics
Female Gender: n (%)	58 (57.4)	34 (63)	14 (46.7)	7 (56)	3 (42.9)	0.16
Mean Age (SD)	59.4 (10.8)	57.8 (11.5)	64.2 (8.4)	60.8 (10.9)	59.4 (10.8)	0.18
African American: n (%)	61 (60.4)	30 (55.6)	9 (60.0)	17 (68.0)	5 (71.4)	0.38
Income <$40,000: n (%)	67 (66.4)	37 (68.5)	8 (53.3)	17 (68.0)	5 (71.4)	0.59
Insurance Status: n(%)						0.90
None	19 (18.8)	9 (16.7)	3 (20.0)	5 (20.0)	2 (28.6)
Medicaid	41 (40.6)	25 (46.3)	4 (26.7)	7 (28.0)	1 (14.3)
Medicare	26 (25.7)	11 (20.3)	5 (33.4)	11 (44.0)	3 (42.9)
Private	15 (14.9)	9 (16.7)	3 (20.0)	2 (8.0)	1 (14.3)
High School and above: n(%)	90 (89.0)	49 (90.7)	12 (80)	23 (92)	5 (85.7)	0.74
Unemployed: n (%)	68 (67.3)	16 (70.4)	5 (66.7)	10 (60)	2 (71.4)	0.79
Mean SBP and DBP (SD)	132.2	133.5	123.5	132.2	141.2	0.19
	(16.6)/77.0 (12.1)	(16.8)/79.5 (12.4)	(16.5)/74.5 (12.1)	(16.5)/74.2 (9.1)	(11.0)/74.2 (17.7)	0.35
Mean # Antihypertensive	2.2 (1.2)	2.4 (1.4)	2.6 (1.0)	1.8 (0.8)	1.7 (0.8)	0.08
Medications (SD)
Mean Years with HTN (SD)	13.8 (11.3)	13.9 (11.4)	19.5 (13.3)	10.4 (8.9)	12.3 (9.8)	0.10
Diabetes: n (%)	57 (56.4)	31 (57.4)	7 (46.7)	16 (64)	3 (42.9)	0.41
Stroke: n (%)	15 (14.9)	7 (13.0)	3 (20.0)	4 (16.0)	1 (14.3)	0.62
Kidney Disease: n (%)	9 (8.9)	7 (13.0)	1 (11.1)	1 (11.1)	0 (0)	0.17
≥1 Years with PCP: n (%)	65 (64.4)	35 (64.8)	10 (66.7)	14 (56.0)	6 (85.7)	0.50
PCP Characteristics
Female Gender: n (%)	18 (66.7)	9 (62.9)	5 (62.5)	2 (66.7)	2 (66.7)	0.89
Mean Age (SD)	36.2 (6.0)	33.4 (5.7)	35.7 (4.0)	42.1 (3.7)	33.3 (3.1)	0.004
Race/Ethnicity: n (%)
White	15 (55.6)	8 (61.5)	1 (33.3)	5 (62.5)	1 (33.3)	0.19
African American	5 (18.5)	2 (15.4)	0 (0)	2 (25.0)	1 (33.3)
Latino	2 (7.4)	2 (15.4)	0 (0)	0 (0)	0 (0)
Asian	5 (18.5)	1 (7.7)	2 (66.7)	1 (12.5)	1 (33.3)
Mean Years at Clinic (SD)	5.8 (4.7)	4.3 (3.8)	6.0 (4.4)	9.0 (5.8)	3.3 (2.1)	0.12
Visit Characteristics
Mean Visit Length: minute (SD)	24.8 (10.2)	25.8 (10.6)	21.3 (7.1)	22.1 (9.7)	31.3 (7.1)	0.08
Mean Talk-time Ratio2 (SD)	1.20 (1.13)	1.05 (1.03)	1.18 (1.31)	1.43 (1.32)	0.93 (1.01)	0.44

SD: Standard Deviation; SBP: Systolic blood pressure; DBP: Diastolic blood pressure; HTN: Hypertension; PCP: Primary care provider

¹comparisons across dyad types using a χ2 or standard t test

²Ratio of utterances spoken by patients vs. providers during the visit; scores >1 is equivalent to greater provider talk

3.1. Description of the patient-provider communication typologies

Four typologies of patient-provider communication were empirically derived from the analysis: interactive, divergent-traditional, traditional-convergent, and disconnected. Table 2 contains selected quotes that exemplify patient-provider exchanges for each typology, as they occurred across different parts of the medical interview (e.g., when discussing medications). In the interactive and convergent-traditional exchanges, providers adopted a patient-centered approach by collaboratively setting the agenda with patients, using a high frequency of open-ended questions, assessing patients’ understanding of their medications and problem-solving when patients admitted non-adherence. Patients in these interactions adopted either an active role in the visit equally sharing information and asking questions (interactive), or a passive role (convergent-traditional). For example, patients in the interactive exchanges often asked for clarification when information was unclear, asked informed questions about their medications, and were knowledgeable of their regimen. Alternatively, patients in convergent-traditional exchanges were often unable to provide accurate information about their medication regimen, tended to only provide information in response to provider questioning, and deferred decision-making to the provider. However, these patients would be able to describe their specific medication-taking routine (e.g., “Every morning between 9:00 and 10:30”) and the aids (e.g., pill boxes) they used to ensure adherence.

Table 2.

Sample Quotes across the typologies of patient-provider communication

Part of the Medical Interview	Interactive Typology (53% of interactions)
Information delivery	Provider: I think just like the blood pressure it’s not a sprint; it’s a marathon. We don’t just cure everything all at once like we do with antibiotics for an infection. These are chronic medicines that you’re gonna continue to take for a long time every day. That’s the bad news. The good news is these pills work very well for these problems. They give your body the help that your body needs. Does that make sense?
	Patient: Yes.
	Provider: That’s great news.
	Patient: Cuz without them [the medications] I might catch a stroke or a heart attack.
	Provider: Exactly. Strokes are a big deal. Heart attacks are a big deal. We’re all risk for them. These are some of the best ways we can reduce those risks is taking the blood pressure pills, doing the checkups like we’re doing and healthy living. All that stuff. You’re doing great, so that’s the long story short.
	Patient: Thanks.
Medication discussion	Provider: Mm-hmm. With the weight.
	Patient: With the weight. Cuz I feel like if I come down with the weight, then my pressure comes down.
	Provider: You’re exactly right. They’re tied together.
	Patient: Well, I might not need the pressure pills.
	Provider: That would be the goal.
	Patient: My goal too [laughs].
	Provider: [Laughs] Exactly. All right, so let’s do this. We’re gonna make a plan. You said that you’re gonna try to increase your walking a couple times a week more.
Decision-making	Provider: If you would prefer to try to cut down on smoking rather than increase the Lisinopril dose, I am totally fine with that. If, however, you say, “I really wanna get this blood pressure down,” then I see very little harm in increasing from 10 to 20.
	Patient: Well, when you say you don’t see the harm, let me ask you—all right, so let’s say that I would start tomorrow.
	Provider: Start what?
	Patient: Doubling my dosage on the Lisinopril. I’m sorry; go to 20. I’m concerned about—well, obviously, about having—being hypo, feeling as down as I am to begin with. Is there not a synergy or an association there?
	Provider: Do you mean between these blood pressure medicines and your mood?
	Patient: Yeah.
	Provider: It’s a great question. There are some that are more commonly known to be associated with depression.
	Provider: Usually, no, the Lisinopril itself is not associated with any mood disturbances, though it’s a great question because there are some agents that are.
	Patient: Well, let’s try it.
	Provider: Okay, sounds good. I will write that down.
	Divergent-Traditional Typology (24% of interactions)
Information delivery	Provider: You know, for people who do have problems with circulation, they can have what are called claudications which are basically, you know, pain with walking a stereotype distance. So if you did have significant compromise of your circulation, you know, walking on a level surface, you would only be able to walk, two or three blocks before you had to stop because of this pain. But given that you can kind of, you know, walk unlimited and, you know, and you only, you know, kind of feel the, the pain after, you know, kind of walking up, you know, fairly, you know…I mean, four or five flights is not, is not an insignificant amount of steps to do that at a long time.
	Patient: It's not?
	Provider: No. I mean, it's, it's, it's actually, you know…I would, I would bet that lots of my patients, you know, wouldn’t, wouldn’t be able to do that. So, you know, it's not unusual for you to have pain like that after walking that many steps. Now, you know, if you were to tell me that you felt the pain after walking just three or four steps as opposed to four or five flights of steps then I would be a little bit more concerned. But, you know, given what you’ve told me and given that I really do feel good pulses, you know, I think we can … I don’t think we need to do any further workup right now. During…but, but you did have, have some, some leg pain during the stress test, it sounds like
	Patient: Okay.
	Provider: Okay. All right. Well, you know, I can definitely discuss with doctor, with Dr. you know, to see if we should think about maybe, looking at the circulation in your legs a little bit more formally with basically an ultrasound, which it's called an ankle brachial index. And it, it basically helps…it gives us a way to kind of quantify how your circulation is in your legs compared to your arms. And we would get a ratio and depending on what that, that ratio is, you know, it tells us whether or not we need to be concerned about circulation.
	Patient: Okay.
Medication discussion	Provider: So…well, I guess that proves that you need your Amlodipine, correct?
	Patient: My who?
	Provider: The blood pressure medication.
	Patient: Yes.
	Provider: You take it?
	Patient: Yes, I do, yes.
	Provider: 'Cause if you…
	Patient: I do.
	Provider: …don’t take it, it goes high
Decision-making	Provider: I think we need to add something to what you're taking for blood pressure 'cause it's a little bit elevated today. But, you know, before I decide for sure, let me take a look at what it's…yeah. It was high in February and April too. So let me just take another peek at…okay. So…yeah. I did write a note to myself the last time I saw you that I wanted to add another blood pressure medicine if it continued to be elevated. So I think we can do that today. It's called Amlodipine.
	Patient: Okay.
	Provider: And your insurance should pay for that without any problem. It's available in a generic form. It's a once-a-day medication.
	Patient: Mm hmm.
	Provider: Okay. And so you can take it at the same time you take the Lisinopril.
	Convergent-Traditional Typology (17% of interactions)
Information delivery	Provider: What's new?
	Patient: Nothing much.
	Provider: Nothing much?
	Patient: No.
	Provider: Let me remind myself. I saw you 6 months ago. How is everything with your blood pressure?
	Patient: Fine.
	Provider: Yeah, any problems or concerns?
	Patient: None, unless you think there is one.
Medication discussion	Provider: For your blood pressure, you're taking a few medicines, correct?
	Patient: Yes.
	Provider: What medicines are those? Do you remember the names?
	Patient: No.
	Provider: You're taking one medicine here called Lisinopril?
	Patient: Yes.
	Provider: How many pills of that are you taking, do you know?
	Patient: One.
	Provider: One?
	Patient: Yeah.
	Provider: It says on here you should be taking two tablets of that.
	Patient: Ah.
Decision- making	Provider: I've thought about whether I should stop it. And I thought, “Oh, he's controlled on this medicine.”
	Patient: Well, if you wanna change it, we’ll change it.
	Provider: I think we should change it.
	Patient: Okay. We’ll change it to whatever you wanna change it to.
	Provider: All right.
	Patient: You know, me-…you…like I said, doc, you're the boss.
	Disconnected Typology (6% of interactions)
Information delivery	Patient: I felt a little lightheaded, like, my blood pressure might have been up or something.
	Provider: Mm hmm.
	Patient: But I attributed that to the alcohol.
	Provider: Mm hmm.
	Patient: I had about…a little vodka…
	Provider: Mm hmm.
	Patient : …with some juice…
	Provider: Okay.
	Patient: …and we were smoking weed.
	Provider: Okay.
Medication discussion	Provider: We’ll see if we need to do any adjustment in the medication because today I don’t know—
	Patient: I’ll eat a banana and I’ll be okay.
	Provider: Yeah. It should—Lisinopril, Amlodipine, calcium, asprin, and Crestor.
	Patient: Well, calcium you don’t really have to write them for because I buy it anyway—
	Provider: Yeah. You need—you can—whatever. Okay.
	Patient: And I have so much—so many pills.
	Provider: That one is not gonna—Amlodipine, asprin, and Crestor.
Decision- making	Provider: Does that make sense? Okay. How far you're walking?
	Patient: Well, I don’t get out that much because…I guess I could walk, like, a half a mile when I walk real slow and easy.
	Provider: Okay.
	Patient: But I really haven't been traveling so much. I'm having a lot of problems with the police and the…you know, personal problems.
	Provider: Okay.
	Patient: And I'm, I'm afraid. I'm living in fear.
	Provider: Okay.
	Patient: You know, I'm trying to put together a lawsuit…
	Provider: Okay.

In the divergent-traditional exchange, providers were verbally dominant throughout the visit and controlled the flow of the conversation through the high use of medical terminology, monologues, and directives thereby inhibiting patients’ ability to ask questions, seek information, or check understanding of information. Providers also questioned patients’ adherence with a series of close-ended questions most frequently as part of medication reconciliation, and rarely engaged patients in problem-solving when barriers to adherence were mentioned. As a result, patients spent much of the encounter repeating the same information and asking multiple questions in an attempt to clarify provider medical language rather than add new information. The importance of adherence was also discussed from the provider’s perspective rather than being initiated by the patient. Finally, providers in disconnected exchanges used mainly closed-ended questions, terse directives, and monologues to gather and convey information. In response, patients would move the conversation away from their hypertension to speak about psychosocial challenges (e.g., drug use) after which the provider would respond with a simple utterance (e.g., “Okay”). When discussing adherence, providers relied on leading and close-ended questions and patients in turn provided vague responses to these inquiries (e.g., “Mmmhmm”). Patients’ discussions about medications primarily focused on their perceived doubts about their effectiveness and concerns about side effects, which were often not addressed by the provider.

4. Discussion and Conclusion

We used a dyadic approach to develop four typologies of patient-provider interactions that characterize the interdependence of patient and provider communication: interactive, divergent-traditional, convergent-traditional and disconnected. Bi-directional communication has been studied in a variety of ways across diverse fields. The actor-partner interdependence model and interdependence theory are analytical frameworks for quantitatively examining dyadic relationships such as caregiver relationships, parent-child interactions, romantic couples, and siblings.[20, 21] Recently, these analyses have been use to examine shared decision-making in patient-health care provider interactions.[22, 23]. In the patient-provider literature, empirical analysis of audiotaped interactions has identified patterns of provider communication, and the communication behaviors that distinguish each pattern.[24–27] Conversation analysis has also been used to develop typologies of patient-provider interactions related to turn-taking, achieving understanding, patient problem presentation, and opening and concluding a clinical visit among others.[9, 11, 12, 28] Together, these studies tell us that the development of a shared understanding between patients and providers is fundamental for improving patients’ health knowledge,[29] shaping health-promoting behaviors including medication adherence,[30] increasing health-related self-efficacy,[31] and enhancing patient satisfaction.[24, 32] The present study advances this work by refining our understanding of how patients and providers co-construct conversations in the clinical encounter to discuss patients’ medication regimen.

Several limitations should be noted: First, our study was comprised of predominately low-income African-American and White patients thus; the findings may not generalize to other patient groups. Second, patient-provider communication was examined at one time point providing a narrow view of a longitudinal relationship. Third, since encounters were audiotaped, we could not examine non-verbal behaviors, which can provide cues into patients’ cognitive status and emotional state that can go undetected if providers rely solely on verbal behaviors.[33]

4.1. Conclusion

We identified interdependent patient-provider communication styles, which may lead to a deeper understanding of how patients and providers interact to influence health behaviors. There are several areas of research that could stem from this work, guided by evidence-based frameworks from other disciplines (e.g.,[34, 35]) Research is needed to examine the contextual, situational and individual-level factors that could affect the typologies. This may include examination of patient and provider personality traits, health literacy, and gender and race/ethnic concordance [36–39]). Contextual and situational factors that may affect the typologies include the interactional preferences of the partners[40], conversation topic (e.g., routine check-up vs. disclosure of new diagnosis) or context the conversation is occurring within (e.g., rushed appointment). Future exploration of the identified typologies might reveal useful ways to improve both care processes and outcomes at the patient, provider, and relationship-level, especially for chronic diseases such as hypertension.

Highlights.

• We identified four typologies of interdependent patient-provider communication

• The four typologies were labeled: interactive, divergent-traditional, convergent-traditional, and disconnetected

• Patient-provider discussions about medications differed across the typologies