The Recognition of Depression: The Primary Care Clinician’s Perspective

Sampling and the Sample

Grounded theory relies on a purposive sampling technique (eg, theoretical sampling) that is driven by ongoing analysis. Sampling decisions are made to facilitate the constant comparison of cases and to identify sources of variation. In this study, sources of influence and variation in clinicians’ decision making regarding depression recognition were identified from interviews with clinicians and from published research. Sampling was determined to enhance analysis of whether and how these factors influenced the recognition of depression by the clinicians interviewed for the study. In response to published literature, theoretical sampling was initially done to explore the impact of time available for patient encounters. As the analysis proceeded, it became clear that other factors—familiarity with patients, extent of clinical experience in general, and the patients’ recognition that depression might be the source of the symptoms, for example—were also influential in both the likelihood that clinicians would recognize depression and speed with which they might do so. Theoretical sampling was therefore done to examine the impact of these factors, leading to a final sample that included variation in time pressures, general clinical experience of the clinician, and the patients’ identification of depression as a possibility.

The final study sample included 8 clinicians from the 3 groups: family physicians, general internists, and nurse practitioners. We reached theoretical saturation on the conditions that influence the sorting processes in which clinicians engage; that is, time available, years of experience, and familiarity with the patient continued to explain variations in the symptom-sorting process and in the likelihood of recognizing depression. This finding was consistent across the clinicians, regardless of training or specialty.

Data Collection

Using a letter approved by the University of Wisconsin Institutional Review Board as an invitation, in-depth, face-to-face interviews were conducted with 8 primary care clinicians in 3 primary care settings. Early interviews were deliberately broad and open, using unstructured questions, such as, “Tell me about your work here,” and “Tell me about seeing patients here.” The data from these 2 interviews were then used to develop conceptual categories and guide ongoing data collection. Several questions were asked in every interview (Table 1⇓), whereas other questions evolved to fill in theoretical gaps as the analysis progressed. Drafts of ongoing interview questions were developed and discussed in 2 grounded theory research groups. Interview probes were used as needed throughout the interviews to clarify, expand, and compare what was being talked about. For example, when familiarity with a patient emerged as a condition for recognizing depression, clinicians in subsequent interviews were asked to compare how their approach to familiar and unfamiliar patients differed.

Data Organization and Analysis

All interviews were audiotaped and transcribed. Grounded theory coding schemes—open, axial, and selective—were used to analyze the data. To promote reliable interpretation of the data, we used 2 experienced grounded theory research groups. Group analysis increased sensitivity to and awareness of assumptions that the researchers might have brought into the analysis and opened the analysis for interpretation across a broad range of perspectives. In the same vein, we also talked openly about our preconceptions and expectations related to primary care clinicians’ care of depression. For example, one preconception was that the conditions under which a clinician works (eg, time) would influence his or her care for patients with depression. After an analysis of each interview, data matrixes were developed to organize categories with indexes to supporting quotes and succinct descriptions of relationships among the emerging theoretical categories.

Ruling Out

When patients complained of symptoms that might indicate depression but did not identify themselves as depressed, clinicians responded by thoroughly sorting through the patient’s symptoms. Even when clinicians suspected depression, they considered a broad range of possibilities in the initial sorting of symptoms. This sorting began with a ruling out of organic causes. Ruling out generally began with fitting symptoms into a medical-physical (vs mental health) diagnosis. This medical explanation gave clinicians a framework for understanding symptoms—they could fit the symptoms into a medical diagnosis.

A lot of times people who come in with fatigue … I’m tired, I’m just tired all the time. Well, immediately you start thinking this could be cardiac, it could be heart failure, it could be lung failure, it could be a metabolic problem.… So it just becomes part of a differential diagnosis. [Nurse practitioner (NP)1]

Two characteristics of depression—the subjective nature of the symptoms combined with an absence of corroborating physical evidence and the stigma associated with the diagnosis—often prolonged the ruling out process, delaying the diagnosis of depression. The societal stigma of mental illness and current and more subjective methods to diagnose depression that rely on what the patient says (rather than on a biological or objective marker) led clinicians to rule out other possible medical causes before shifting their thinking to the possibility of a mental illness, such as depression.

I think it’s a bit threatening to the patient to suggest that they may have a mental illness.… So you don’t want to jump to that conclusion quickly. You may want to … give it some thought. [General internist (GI)1]

I won’t force them [patients] into accepting it [depression], I mean you CAN’T! The patient has to agree, because depression is a little different in that there’s so much subjectivity to it. You know, like diabetes, if I say … and I’ve had patients say, “I don’t think I have diabetes.” And I’ll say, “YES! YOU DO, because your blood sugar would not be like that unless you had diabetes. You might not feel like it, but you do.” But depression, it’s not like that. Because if I say, “You are depressed,” and you say, “No I’m not,” the only thing I can say is, “Well, here’s how you feel, and you’re the one who knows how you feel.” There is no real objective measure for that. It’s a subjective feeling of your emotional state. So it’s not like I can prove to somebody that they’re depressed. I can tell them that it’s my OPINION that they have a lot of symptoms of depression and that they may be depressed, but they have to accept it. [Family physician (FP)2]

Familiarity with the patient was found to be an important condition. When clinicians were not familiar with the patient, the subjective nature of the symptoms and the social stigma associated with depression increased the reluctance of the clinician to diagnose depression. In addition, when clinicians were not familiar with the patient, they acknowledged that the patient was less likely to share personal information, making diagnosis even more difficult. The general experience of clinicians also influenced the process. Clinicians’ reluctance to suggest depression early in the visit increased when clinicians had less general experience. Lack of familiarity and less general experience led clinicians to exhaust medical diagnoses before considering or suggesting a mental health diagnosis.

Extensive ruling out was used to avoid missing a medical condition or upsetting the patient. As a consequence, recognizing depression was often delayed. In contrast, when clinicians were familiar with the patient, the ruling out process was considerably shortened. The clinician often explicitly compared the familiar person she or he knows with the person in the room, highlighting differences that might suggest depression. While this process is effective and results in much quicker identification of depression, it can be used only when the clinician is familiar with the patient.

When somebody comes in new, I really don’t have any information on them at all from previous appointments, and it’s kind of like starting from scratch. Because sometimes things don’t come out in their history.… “Do you have any history of health problems?” and they’ll say, “No.” Whereas they might have a lot of things that just aren’t at the point where they want to divulge it. I spend more time kind of getting to the bottom of it. [NP2]

Opening the Door

Opening the door was defined as allowing the patient to talk about whatever was bothersome, which most experienced clinicians saw as inviting patients to share the real reasons for the visit. Diagnosing a mental health condition such as depression requires the patient to share highly personal information, and gathering this personal information is crucial to diagnosing depression. When patients do not share information, the diagnosis is delayed. Such sharing is less likely with unfamiliar patients. By opening the door, these clinicians often could delve into personal information relatively early in patient encounters. Clinician experience also influences this process. Less experienced clinicians were more likely to express discomfort with opening the door and talking to patients about mental health issues. Clinicians with more experience (14 to 28 years), however, described how they gradually became more comfortable with opening the door to a conversation about depression and listening to clues about the real reason for the visit.

Experienced clinicians were more willing to use this strategy, especially with unfamiliar patients. Experienced clinicians believed that unfamiliar patients were much less likely to divulge personal information, so they deliberately open the door to patients early in the encounter to gather personal information. Experienced clinicians also described how they learned to recognize a patient’s verbal as well as nonverbal cues, whether the real issues have surfaced, and whether the patient is prepared to discuss them.

I do a lot of early work with patients now that I didn’t do 20 years ago … to say, “What are you worried it might be?” “Why did you come here today?” So I start to get at some of the things I think are real reasons why they came. [FP1]

They give you some nonverbal cues or they give you verbal cues, such as they’ll say, “BY THE WAY, there’s something else bothering me.” Or when they’re all finished, they’ll look like there’s still something else going on. They won’t say, “Thank you,” and get up and leave.… They will often switch the discussion if they don’t want to discuss it. [FP3]

Time availability also influenced a clinician’s willingness to open the door. If clinicians felt rushed and were catching up when they encountered such a patient, they were unlikely to open the door. Even when these clinicians were comfortable discussing emotions and considered depression a likely diagnosis, they anticipated that such discussions would be time consuming and avoided opening the door.

Recognizing the Person

Clinicians described recognizing the person as comparing the person in front of them with the person they know. Recognizing the person appeared to be an alternative (and often much quicker) process to ruling out or opening the door for expediting the depression-recognition process with familiar patients. When clinicians were familiar with patients, their perception of who the patient is compared with who the patient appears to be offered a clue to diagnosing depression. Although the clinicians could not fully elaborate what “knowing the patient” meant, it allowed them to read the patient’s behavioral language better, such as facial affect and body posture, when they were familiar with the patient. Clinicians also stated that, when they know the patient, they could recognize change early in the encounter, often within a few seconds to a minute. Once a level of suspicion has been established with familiar patients, clinicians could raise the possibility of a mental health diagnosis by asking the few screening questions they have found to be useful rather than going through the DSM list of symptoms. The 2 symptoms most often used to screen for depression were sleep problems and low energy level.

I had an experience almost 20 years ago with a very experienced general practitioner in Wales, in Britain. I watched him for a little while. One time he says to me, “Do you watch how I work? What I do is I pay complete attention for the first 15 seconds the patient is in the room. I use all of my senses, and I pay attention.” And he says, “I can get a feeling of what’s going on in the first 15 seconds.” You know, I thought this guy’s crazy. But actually—especially for people you know—he knew all of those patients, those were old patients of his. And he’d walk in and he’d say, “What’s the trouble?” And they’d say, “Why do you think there’s something that’s trouble?” And he says, “I know you, you look different, you seem different, you act different.” And it is true with people you know. Now think of people you work with regularly, and you walk in and you see them, and they look different or they seem different. Then you know something’s different. And whether they say, “No, I feel fine,” or not, reality is something different. Now whether it’s good or bad is not the issue, it’s different. So with my patients, it’s the same way. I actually … I see it in their faces, I see it in their bodies, I see it in how they make eye contact with me when I come in, I hear it in their voices. I mean, this old man, Dr. Williams, was right. It’s in the first 15 seconds, you KNOW something is different. [FP1]

Recognizing depression through recognizing the person seems to occur simultaneously with knowing the patient and being comfortable dealing with emotions. First, to recognize change in the patient, the clinician must know the patient and be willing to reflect this possibility verbally. More importantly, however, time availability overrides these 2 conditions. Even if the clinician recognizes change and feels comfortable dealing with emotions, time constraints often prevent clinicians from opening the door to patients.