Matter over mind: How mental health symptom presentations shape diagnostic outcomes

Abstract

Mental health disorders face less stigma today than in the past, yet they continue to be misdiagnosed and at times improperly treated. One account for this problem is that physicians rely exclusively on a verbal interview of patients for diagnosis. Because this diagnostic method is likely to be shaped by the way patients present their symptoms, it is critical that we examine whether and how patients’ communication practices shape diagnostic and treatment outcomes. This study examines a sample of 14 encounters involving mental health-related symptoms from a dataset of adult primary care visits. Using conversation analytic methods, I show that when patients present mental health symptoms by simply describing the symptoms, primary care physicians exhibit a preference for providing a physical health diagnosis. Conversely, when patients provide a concrete link between their symptoms and the way the symptoms are disrupting their everyday lives, primary care physicians typically provide a mental health diagnosis.

Introduction

Studies have focused on the stigmatization of mental illness since the mid-twentieth century (Goffman, 1963; Link and Phelan, 2001; Martin et al., 2007). The stigma associated with mental illness has remained largely unchanged despite its increasing awareness due to more research, public campaigns, and mental health education (Phelan and Link, 1998). Attitudes toward mental illness continue to engender negative stereotypes and discrimination (Phelan, 2005). However, evidence points to some improvement in perceptions of mental illness. The public’s orientation to mental illness has become more tolerant (Phelan et al., 2000). Furthermore, those with mental health symptoms are seeking treatment more frequently (Kessler et al., 1999). Research has reported that when patients seek care for mental health issues, they do so largely from their primary care provider (PCP) (Wang et al., 2006).

Despite this increase in help-seeking for mental health concerns, there still appears to be bias against diagnosing mental health disorders as current estimates suggest that the majority of those with mental health issues remain untreated (Boyer and Lutfey, 2010). Conversely, there is another complication to the treatment of mental health disorders which is a concern that they are actually over-pathologized and over-diagnosed (Kutchins and Kirk, 1997). Why are these medical issues the subject of such fraught treatment perspectives? This article suggests that the answer to this question lies in the moment-to-moment interaction between PCPs and patients. Specifically, the way patients characterize their mental health symptoms to their PCPs can condition how PCPs take up and treat those symptoms as indicative of a mental health disorder or not. At the same time, the way that patients talk about their symptoms with their PCP will likely be complicated by the stigma associated with mental health disorders.

This pilot study investigates the ways patients present their mental health-related symptoms and the relationship between patient symptom presentations and physician orientation to those symptoms in diagnosis. I show that PCPs are biased against mental health diagnoses and treat mental health symptoms only in contexts where patients go beyond a basic presentation of symptoms to show that their lifeworld (Mishler, 1984) has been disrupted. This is true even when the symptoms being presented are essentially the same.

Background

Diagnosing mental health problems in primary care

Longitudinal data suggest that most people experiencing mental health issues are not receiving treatment for their symptoms despite an increase in help-seeking for such problems (Kessler et al., 2005). Most individuals in the USA who pursue help do so in primary care rather than seeking out a specialist for treatment. This is thought to be because patients feel more comfortable communicating to a provider with whom they have a relationship (Glazier et al., 2015). However, there may also be structural bases for this, such as insurance requirements that primary care physicians serve as gatekeepers to specialty care including mental healthcare (Forrest, 2003). Overall, the literature suggests that mental illness remains undertreated despite increased visits for related symptoms, which demonstrates evidence for an institutional bias against diagnosing and treating mental illness.

Yet, studies document both underdiagnosis and overdiagnosis of mental illnesses. Kutchins and Kirk (1997) argue that changing requirements of treatments reimbursable by health insurance companies have caused physicians to inappropriately diagnose their patients with a mental illness for compensation. Others argue that mental health diagnosis categories are too broad, therefore physicians too frequently diagnose mental illness and overprescribe (Dowrick and Frances, 2013). Many critics blame the drug industry for promoting the medicalization of emotions and thus the overdiagnosis of mental illness (Horwitz and Wakefield, 2007; Karp, 2009). Overdiagnosis is particularly problematic because medications indicated for treatment of mental health problems can have adverse consequences including drug dependency, more severe mental health symptoms, and suicide (Bezchlibnyk-Butler et al., 2013; Karp, 2009).

However, underdiagnosis is also problematic leading to improper treatment of depressive and anxiety disorders (Bet et al., 2013). Researchers have called for interventions to improve detection and treatment in specific medical settings (Gilbody et al., 2003), and underscore that while major depression is often properly diagnosed and treated, moderate depression or anxiety are underdiagnosed and under treated (Kroenke et al., 2007; Thompson et al., 2001).

The complexity of diagnosing mental health problems

One account for how mental health symptoms can be both over and under-diagnosed is that the mental health diagnostic process is particularly multifaceted with opaque boundaries and contingencies. This complexity contradicts the straightforward guidelines for primary care treatment of problems like depression, which follow a biomedical treatment model comprised of psychotropic drug prescription and referral to a mental health specialist (Gelenberg, 2010). Some have argued that this “one size fits all” approach is inappropriate for primary care patients in particular, who present with an array of symptoms that do not necessarily fit within the biomedical treatment model (Nutting et al., 2002). Different from other medical issues which include clear signs that physicians can see, hear, or feel, the detection and treatment of mental health problems are entirely reliant on information patients provide, and this information is conditioned by the physicians’ diagnostic interview. This context may make for a ripe environment for doctor–patient inter- action in shaping diagnostic outcomes.

Previous research suggests that patients’ behavior in the medical encounter can shape both the diagnosis and treatment of medical issues. In studying pediatric acute care, Stivers (2002) found that the ways parents presented their children’s symptoms had direct implications for physician uptake of those symptoms—what Stivers terms symptoms-only and candidate diagnosis presentations. While the former lists just the symptoms (e.g. “She has a scratchy throat”), the latter includes a suspected diagnosis (e.g. “We think she has strep because she has a fever and a scratchy throat”). When examining physician uptake of each presentation type, Stivers established that following a symptoms-only presentation, physicians simply moved to investigate the problem further, while following a candidate diagnosis presentation, physicians moved to either confirm—or disconfirm—the existence of the diagnosis proposed by the family.

Other research also suggests that how patients communicate can shape outcomes. A randomized controlled trial conducted by Kravitz et al. (2005) using standardized patients found that the ways in which patients with mental health symptoms requested psychotropic drugs (no request for medication vs general request for medication vs specific brand request) shaped how physicians treated their symptoms. Patients who made either a general request or a brand-specific request for medication received a psychotropic drug prescription and/or referral to a mental health specialist significantly more frequently than those who provided just their symptoms but no medication request. An exploratory investigation into decision-making for depression treatment in primary care found that patients who presented their symptoms in line with diagnostic criteria received prescriptions for antidepressants, while patients who presented their symptoms as the result of a situational problem were unlikely to receive such a remedy (Karasz et al., 2012)

Studies point to several reasons why presented symptoms may not be properly addressed in the physician-patient encounter. Some findings demonstrate that in contrast with a clear symptom presentation at the outset of the primary care visit, failure to topicalize all symptoms related to all medical concerns early in the visit is associated with the symptoms being less likely to be properly taken up (see Robinson, 2003). This has implications for proper diagnosis and treatment of the problem. Moreover, some believe that primary care physicians are over diagnosing mental health problems while others believe there is an underdiagnosis problem. Others point to constraining diagnostic guidelines which shift PCP’s orientations toward diagnosing and treating depressive disorders involving a particular set of symptoms (Karasz et al., 2012). Despite work thus far, it is still unclear what in the interactional machinery of these visits leads some studies to arrive at different conclusions about the outcome of primary care visits that treat mental health-related symptoms. I propose a conversation analytic (CA) pilot investigation of the moment-to-moment interactional practices of US patients and their PCPs, specifically exploring how patients present mental health-related symptoms and whether this impacts subsequent diagnoses.

Methods

The data used for this investigation are video-recorded US adult acute primary care visits from clinics in Southern California collected from 1997 to 2004. Study procedures were approved by the institutional review board (IRB). All participants provided written informed consent. A corpus of 240 encounters was examined for inclusion of mental health-related symptoms in the encounter. Surprisingly, just fourteen cases were identified in which the patient discussed mental health-related symptoms. Thus, this study acts as a pilot investigation of the phenomenon. Inclusion criteria were determined based on the symptoms indicated in the Mental Health Inventory (Hays, 1994). The Mental Health Inventory (MHI) is a 38-item survey from core measures of emotional well-being from the Medical Outcomes Study. Patients who mentioned one or more of the adverse symptoms as indicated on the MHI during their visit met inclusion criteria.

Each visit was manually transcribed by the author using standard CA conventions and notations as detailed in Hepburn and Bolden (2012). These notations are crucial for understanding not only what participants said, but how they said it. Data analysis was guided by CA practices (see Heritage, 1984; Sidnell and Stivers, 2012). The CA approach to studying medical interaction hinges on the idea that the medical encounter involves commonsense reasoning (see Garfinkel, 1967) and a distinct interaction order (see Goffman, 1983). CA offers a novel and important lens through which the interaction order plays out in medicine and how physicians and patients work to co-construct a medical visit (Heritage and Maynard, 2006). Through CA’s implementation, patterns in clinical interactions are evaluated turn-by-turn which can then be shown to be ordered and systematic in usage within and across medical contexts, known as a conversational practice. Such conversational practices are meaningful on an institutional level because interactions are couched in larger activities and end-goals to which the institutional figures are oriented (Drew and Heritage, 1992).

Findings

Patient presentations of mental health concerns

I am concerned with the ways in which patients characterize symptoms that may indicate a mental health disorder. Drawing on Stivers’ aforementioned work on problem presentations, I reviewed data for any consistent pattern in symptom presentation. What emerged was a two-way distinction between “symptoms-only” presentations as Stivers identified and presentations that provided an additional causal link between mental health-related symptoms and an everyday life disruption (e.g. “I can’t do household chores because I’m so anxious”). I identify the latter as “Lifeworld Disruption” presentations. Although the actual symptoms presented are nearly identical in both presentation types, the different presentation formats lead to distinct diagnostic outcomes. I examine how these different problem presentations affect the way physicians diagnose and treat this information in clinical interactions.

Symptoms-only problem presentations

In 57% (n = 8) of cases in the sample, patients presented symptoms related to a mental health concern and nothing more. An instance of this is shown in Extract 1. At the beginning of the visit, this patient believes something is not right with him. At the outset of his turn in Extract 1a, it is unclear whether he is moving into the territory of mental health as he speaks about his family’s diabetic history and his own symptoms, which he thinks may be related. He discusses family members who have had complications with diabetes and that he feels like “something is going do:wn” on him (line 04). He analogizes this feeling to how he feels on an empty stomach (line 06):

(1a)C3408
01 PAT: Some days I feel like umhh .tch my whole family has diabetes my
02    sister my brother my- (0.4) grandma: di:ed my mother passed away 
03    from that. .hhh and sometimes I feel like uhm (0.4) you know like
04    like ah something is going do:wn on me .hhh you know like 
05    ah- (.) you know (.) like ah when I haven’t e:aten nothing (.)
06    you know like I would- empty stomach I’d=have, and than I- and I
07    said you must be s- ah that I need something to e:at you know.

After more narrative about a feeling that he cannot quite describe, the patient becomes more concrete in Extract 1b by referencing a feeling similar to tension or nerves (lines 01–02):

(1b)C3408
01 PAT: It’s it’s: (.) kind of uhm like uh (0.2) when you’re tense or
02    ner- nerv- in my nerves yiknow (0.2) uhm hh (1.0) en- en (.) I
03    never felt that tru- like um (.) like uh (.) days like this
04    depress me yiknow so >I’m like< why do I feel depression if I
05    gotta nice family, .hh >y’see I-< (.) I feel uncomfortable mayn
06    it’s something weird that I’ve never felt befo:re.

The patient equates his feelings with tension or nervousness in lines 01–02, “It’s it’s: (.) kind of uhm like uh (0.2) when you’re tense or ner- nerv- in my nerves” which signals a relationship to a mental health issue. The patient’s first mention of depression in lines 03–04 seems to be related to the weather outside, as the patient points to the window when saying “days like this depress me.” Next the patient mitigates his prior assertion of depression by questioning his feelings in lines 04–05, “why do I feel depression if I gotta nice family,.” He continues to provide evidence of discomfort (lines 05–06): “I feel uncomfortable mayn it’s something weird that I’ve never felt befo:re.” which is a nebulous description of his symptoms. In this presentation, he outlines symptoms of nervousness, tension, discomfort, and depression on certain days. However, he does not link these symptoms to disrupting his everyday life.

In this example, we see that one way patients present mental health problems is to describe just symptoms. These symptoms-only presentations allow for the possibility that there could be a physical health diagnosis or that the symptoms are not in need of immediate treatment. This problem presentation type contrasts with one that includes the provision of a causal link between mental health symptoms and a disruption to the patient’s lifeworld.

Lifeworld disruption problem presentations

Here, I examine problem presentations involving mental health symptoms that are causally linked to concrete lifeworld disruptions. When patients do this, they treat the problem as in need of immediate treatment. In Extract 2, the interaction begins with an understanding by both physician and patient that something bad happened the day before. What that is, however, we do not initially see. Instead, the patient discusses stress and provides situational evidence as causation in Extract 2a:

(2a)C25–06
01 DOC: How are you doing?
02 PAT: I’m much better toda[:y than yesterday,
03 DOC:              [M:uch better today (.) good seeing you eheh,
04 PAT: Hheheh [you too,
05 DOC:       [Wha- .hh what happened yes[terday.
((8 lines omitted about camera))
14 AT:  I- I- I’m under a lot of stress lately.=
15 OC:  =Oh:.[kay?
16 PAT:    [And I think it’s just building up.
17 DOC: What kind of stress,
18 PAT: .hh my m- we went to New York: uh- about three weeks ago to
19     visit the family .h[hh
20 DOC:             [Uh huh,=
21 PAT: =An:d I found out that my mother had had a stroke.it wasn’t a bad
22     stroke buts- but she is very nervous en thinks she’s gonna di:e
23     any minute .hhh
24 DOC: How old is she,
25 PAT: Um:eighty-one (.) n- this month (.) tomorrow she’ll be eighty-
26     one.hhhh and ↑ the doctor says she’s okay:↑ .tch but she’s very
27     afraid and that’s made me very upset.

The mutual understanding between physician and patient that something has happened is indicated in the first exchange. When the physician walks in and asks, “How are you doing?” (line 01) he indexes knowledge of some already known medical issue (Coupland et al., 1994). It is clear from the patient’s response, “I’m much better toda:y than yesterday,” (line 02) that she orients to this question as a reference to medical business to be covered (Heritage and Robinson, 2006). Next the physician asks, “Wha- .hh what happened yesterday” (line 05). Rather than responding to this, the patient indicates that she has difficulty falling asleep, attributes this to her experiencing stress, and discusses situational instances– her mother’s health– that she interprets as causing her stress. At this point, the presentation is a symptoms-only presentation.

Subsequently, the physician asks whether this occasioned her recent visit to the emergency room (line 01), thus requesting confirmation of an implicit causal link between the mental health symptoms and emergency room visit. The patient does not provide this link initially:

(2b)C2506
01 DOC: So you went to emergency room: then?
02 PAT: I went.
03 DOC: And then what happened,
04 PAT: .hhh (0.2) I was there five hours. You wanna know what my
05     pressure was befo[re I went in?
06 DOC:            [Mhm: mhm,=
07 PAT: =cuz I wrote everything down.
08    ((Pat pulls out a piece of paper from purse and points at it))
09 PAT: .hh ok this is what it was at four A.M. it just kept getting
10     higher (0.4) and I was panicking.

In lines 04–05, the patient offers a pre-announcement (Terasaki, 2004) to the doctor about the nature of her high blood pressure, and in lines 09–10 this is expanded, specifically that the blood pressure continued to increase which caused her to panic. So far, this patient has described her mental health-related symptoms and indicated that she visited the emergency room but she has not made an explicit link between her symptoms and the lifeworld disruption. This comes in Extract 2c:

(2c)C2506
01 DOC: So it was one seventy ni:::ne? over one oh eight.
02 PAT: At four A.M. then i- then it went up? then it wen- kept going 
03      ↑up.hhhh en I had called the: the emergency number and then the
04      on call didn’t call me back,
05 DOC: Mhm,
06 PAT: So after a half hour I called again en I- en I held on fer about
07      five minutes en they got him (.) en I told him that I’m coming
08      right in=he said yes you should come in. (0.2) [cuz I was just
09 DOC:                             [the-
10 PAT: too afraid.

In lines 03, 08, and 10, the patient causally links her symptoms to a lifeworld disruption, that she called her physician’s emergency number because she was “just too afraid” about her increasing blood pressure.

While symptoms-only presentations include symptoms of feeling unusual, depressed, or down without provision of a causal link to a lifeworld disruption, these problem presentation types contrast with those that do provide the addition of a causal link between mental health symptoms and a disruption to the patient’s lifeworld. These latter presentations reflect a patient orientation toward a problem in need of treatment, which contrasts with how symptoms only presentations are made even though actual symptoms are effectively the same. The question to which I now turn is whether the design of the problem presentation conditions physician response.

Physician uptake of mental health concerns

The two presentation practices employed by patients–symptoms-only and lifeworld disruption— have consequences for physician response. In this section, I show that questions pursued by physicians are implicitly biased and move toward different diagnostic trajectories depending on the problem presentation. Specifically, following symptoms-only presentations physicians primarily pursue physical health diagnoses, while in the context of lifeworld disruption presentations physicians primarily pursue mental health diagnoses. This result suggests that amid nearly identical symptoms, physicians generally show bias toward physical health diagnoses when only mental health symptoms are presented, but once patients provide the addition of a causal link between their symptoms and a lifeworld disruption, physicians orient to the symptoms as indicative of a mental health problem.

The symptoms-only context: setting aside a mental health diagnosis

A primary way that physicians show a bias toward a physical health diagnosis in the symptoms-only context is to directly acknowledge a patient’s mental health issue but then set it aside in favor of the physical health issue. This is illustrated in Extract 3. The patient presents with cold symptoms and then suggests that there may be a connection between depression and the colds she has been getting (lines 01–02). This presentation is symptoms-only:

(3a)C24–08
01 PAT: .hh en you know what too I feel like um:. to know (.) can 
02     depression: also: cause .hh like- sickness or co:lds:?
03     (0.4)
04 PAT: Um I don’t know if I have it I’ve just been feeling like under
05     the weather but I’m thinking it’s just everything like .hh from
06     take care of family having children working and like what’s
07     going on in: the world that’s causing me to feel a little down
08     .hh um,=
09 DOC: =You don’t look depressed b[ut d’you feel sad all the time or,=
10 PAT:                 [uh okay.

Following a question that receives no uptake by the physician in lines 01–02, the patient provides an account for her question (lines 04–07). Here, she collateralizes her suspicion of depression symptoms with situational evidence. The core of this interaction unfolds in line 09, where the physician makes an assessment about the patient’s appearance, “= You don’t look depressed but you feel sad all the time or,=.” In telling the patient that she does not physically look depressed, this physician counters the patient’s implied diagnosis. However, in the same turn, this physician asks a polar question which is biased toward a mental health problem (see Heritage, 2010): “but d’you feel sad all the time or,=“ (line 09). Yet, because the physician in his professional opinion has communicated a clear stance that the patient does not look depressed prior to asking about her feeling sad, based on his assessment we might assume that she would minimize any response about feeling sad moving forward in her symptom narrative.

Following this sequence, the patient continues with opaque symptoms of feeling “blue” and tired in the mornings, which get no uptake. After the patient has completed her narrative, the physician offers a still stronger counter-diagnosis of no depression:

(3b)C24–08
01 DOC: Well no. (.) You’re not getting sick because you’re dep- you- you 
02     don’t have depression (.) no=
03 PAT: =Oh[kay,
04 DOC:    [No Tina but I mean if that continues where you’re feeling
05     really sad en [.hh don’t wanna do anything fu:n en .hh um:.
06 PAT:           [Mhm,
07 DOC: Cryin all the ti:me decrease in energy: can’t slee:p,
08 PAT: O:[h,
09 DOC:   [Yeah don’t wanna go out with your frie:nds you wanna try to
10     stay ho:me,
11 PAT: Oh ok[ay yeah,
12 DOC:      [All those things are the symptoms of depression.
13 PAT: Okay no I’m pretty much up for doing anything .hh if something
14     comes up,
15 DOC: Good.
16 PAT: Yeah: so=
17 DOC: =I would just get some more exerci:se then.

In this extract, the physician shuts down potentially more discussion about the patient’s symptoms as related to depression. In lines 01–02, his diagnosis that the patient is not depressed is done under the guise of answering the patient’s question from (3a): whether depression can cause colds. In the next few lines of talk, he states that should she present with sadness, no desire to have fun, the inability to sleep, and frequent crying— in that case, he would consider a depression diagnosis. When, in lines 13–14, she says “Okay no I’m pretty much up for doing anything .hh if something comes up,” he responds with an assessment, “Good” in line 17 and then discusses treatment of her fatigue and cold symptoms.

This physician sets aside the issue of depression because he briefly inquires about the depression in line 09 of Extract 3a following her initial suggestion that depression may be the cause of her colds, “= You don’t look depressed but d’you feel sad all the time or,=,” yet pursues a physical health diagnosis in line 17 when he recommends more exercise. The patient could have further pushed the issue of her symptoms or been prompted to be clearer had the physician moved toward a mental health concern. Instead, she did not orient to her mental health concerns as problematic because her physician dismissed those concerns. This physician moves toward a physical health diagnosis, yet here he acknowledges and sets the potential mental health diagnosis aside in favor of a physical health one. Like in Extract 1 of the previous section, we see depression symptoms being described but without a concrete provision linking these symptoms to disrupting the patient’s lifeworld. Notably, the patient provides information about her lifeworld as a potential cause for these symptoms but does not causally link these symptoms to a disruption in her lifeworld.

In this case shown, the patient uses a symptoms-only mental health problem presentation. In this type of presentation, once patients provide mental health-related symptoms that could be taken up by physicians as indicative of a mental health issue, physicians in these data instead move toward physical health diagnoses by setting aside mental health- related diagnoses. This is done via tilting questions to a physical health diagnosis by being problem-attentive (Stivers, 2007) toward a physical ailment. Although this is a small dataset, this pattern is consistent.

The lifeworld disruption context: physician uptake of mental health symptoms

When patients include a causal link to a lifeworld disruption, physicians are more likely to orient to the problem as a mental health one. Specifically, in the context of lifeworld disruption problem presentations, physicians pursue a mental health diagnosis instead of a physical health one. In Extract 4, the patient presents with situational depression due to stress from her job and personal issues. She has noted that she went to urgent care because she was likely having a panic attack. She continues to indicate that her symptoms are due to stress and includes a concrete link between her symptoms and an everyday life disruption in lines 05–06:

(4)C1402
01 PAT: [.hhh en I- now I am still te:nse (.) because yiknow I’m
02     still trying to- fight with my son [you know,
03 DOC:                     [mhmm
04 PAT: but I have to go to court on the twenty-sixth and I’m just- not
05     sleeping? (0.2) I’m so depressed depressed .hhh >lemme tell you<
06    (0.6) I start doing things around the house (0.4) I ↑can’t.
07 DOC: Okay.
08 PAT: Very depressed.
09 DOC: Are you crying easily? °Or-°
10 PAT: hhhhh yes yes,.

The physician responds to the patient’s example of a disruption with the question, “are you crying easily? °Or-°” (line 09). The problem-attentive nature of this question is in marked contrast to the prior cases insofar as it is biased in favor of a mental health problem. Here, the physician orients to this disruption as problematic and provides an opportunity for the patient to further expand her symptoms. Notably, it is only when the patient makes the causal connection between symptoms of sleeplessness, headaches, tension, and the inability to do jobs around the house (a lifeworld disruption) that the physician actively pursues additional mental health-related symptoms. As a result of the physician’s active search for additional symptoms, the patient’s illness presentation becomes more granular (Schegloff, 2000).

A pattern in these data is apparent: physicians pursue questioning biased toward a physical diagnosis when patients present symptoms-only mental health concerns and physicians pursue questioning biased toward a mental health-related diagnosis when patients provide a causal connection between a concrete lifeworld disruption and their symptoms. This pattern holds for 12 out of the 14 total cases. However, two cases do not fit this pattern. In what follows, I show that the ways in which a departure from the pattern nonetheless supports the broader analysis.

A deviant case

How do we account for a symptoms-only mental health presentation that results in the pursuit of a mental health diagnosis? Such an instance arose in the dataset as shown in Extract 5. The patient in this case presents with persistent fatigue. Her problem presentation begins below:

(5a)C3106
01 PAT: =For a long while= 
02 DOC: =Mhm.
03 PAT: I felt- I’ve been feeling really fatigued.
04 DOC: Mkay=
05 PAT: =As- as long as I can remember actually.

The patient explains that she had visited this office twice before with the same complaint and was given a pregnancy test during one visit and a thyroid test during another. Next, the physician reassures her about her thyroid and continues with diagnosis:

(5b)C3106
01 DOC: =So (.) your- your thyroid’s all right and I was looking back at
02     your chart (.) .hh I was thinking could you be anemic if they
03     checked you in September=
04 PAT: =Right.=
05 DOC: =It certainly wouldn’t hurt to do that again but I’ll tell you
06     what I think’s going on. .hh (.) When you tell me this history
07     (.) is- (.) when you’ve been in fatigue as long as you can
08     remember (.) and your blood counts are all right (.) and your
09     thyroid’s all right you’re not pregnant it makes think (.) that
10     you have a neurochemical imbalance an imbalance of chemicals in
11     your [central nervous system.

The physician in this case also references these past visits and the physical tests conducted in lines 01–03 and again in lines 08–09. At these past visits where this patient initially presented with the same symptoms, the physician did as the analysis in this article would predict: pursued a physical health diagnosis. During this visit, however, the physician pursues a mental health-related diagnosis, a “neurochemical imbalance,” in line 10.

On the surface, this visit does not fit with the rest: the patient used a symptoms-only presentation yet a mental health diagnosis follows. Note, however, that the physician makes clear that he is only pursuing this diagnosis in the context of previous physical diagnoses that he is now able to rule out: “When you’ve been in fatigue as long as you can remember (.) and your blood counts are all right (.) and your thyroid’s all right you’re not pregnant it makes think (.) that you have a neurochemical imbalance” (lines 07–10). Purely because the two physical health diagnoses did not solve the symptoms of fatigue does this physician now move toward a mental health diagnosis. This provides further support that physicians are biased against a mental health diagnosis initially. The physician continues:

(5c)C3106
01 DOC: Because we haven’t been able to find something else.=.hhh and
02     usually this (.) diagnosis of a neurochemical imbalance is
03     generally (.) a diagnosis of exclusion (0.2) in other words
04     y’try you check thi:s you check this you check this and
05     everything’s fine [but you still have the complaints oka:y.
06 PAT:             [Right.

This physician’s account further demonstrates that he is arriving at this mental health-related diagnosis in the context of physical diagnoses that have still left the patient with the complaint (lines 01–05). He labels the neurochemical imbalance diagnosis a “diagnosis of exclusion” (line 03), which effectively concludes how he arrived at this particular diagnosis without a concrete lifeworld disruption. This outcome therefore can be explained because it was a diagnosis made following two failed physical-health diagnoses. Furthermore, we have evidence for the physician’s behavior on two prior visits that the default orientation was to a physical health diagnosis before a mental health one. It took extra work—a third visit from the patient—in order to secure a mental health diagnosis.

Why does the addition of causal link make the difference for physicians’ treatment orientations? One explanation for this could be how we understand biographical disruption in chronic illness (Bury, 1982). In studying patients with rheumatoid arthritis, Bury investigates the way chronic illness is conceptualized—as a disruptive event. He writes,

… illness, and especially chronic illness, is precisely that kind of experience where the structures of everyday life and the forms of knowledge which underpin them are disrupted … it brings individuals, their families, and wider social networks face to face with the character of their relationships in stark form, disrupting normal rules of reciprocity and mutual support. (169)

Chronic illness, therefore, represents a disruption of our taken-for-granted lifeworld (see Schutz, 1967). Barker (2009) argues that because our sense of self is rooted in our bodies’ daily functioning, an interruption of that functioning which restricts our everyday life performances threatens our fundamental selves. It is perhaps our orientation to this threat of our selves before illness which makes lifeworld disruption salient to us– and consequently to physicians– who take up these issues as indicators of an issue beyond a physical ailment. Once symptoms are presented, the addition of a lifeworld disruption tips the scales.

Discussion

This article outlines two approaches patients take to present their mental health– related symptoms in the primary care setting. Taken one step further, this dichotomy is matched to two distinct physician responses and thus holds consequences for diagnosis. When patients use a symptoms-only problem presentation to articulate their mental health symptoms, physicians orient to such complaints as indicative of a physical health problem and therefore pursue questioning biased in favor of a physical health diagnosis. In contrast, when patients provide the addition of a causal link between their mental health symptoms and a lifeworld disruption in their problem presentation, termed lifeworld disruption presentations, physicians align to the symptoms as indicating a mental health issue and subsequently pursue questioning biased in favor of a mental health diagnosis. This pattern holds in the majority of cases in the sample. While two of the cases in the sample deviated from this initial pattern, they still provided evidence for the pattern: that without the addition of a causal link between mental health symptoms and an everyday life disruption, physicians will demonstrate bias toward a physical health problem even when they initially topicalize a potential mental health problem.

In the medical encounter, physicians typically initiate actions while patients typically respond, resulting in an interactional asymmetry (Robinson, 2003). What underlies a medical diagnosis is not just information that patients provide to their physicians; it is also the ways in which physicians can shape patients’ symptoms through the interactional sequence. This can affect how patients explain their illness and whether physicians adequately hear patients’ experience of illness (Gill and Maynard, 2006). While this analysis primarily examines how patients’ descriptions of symptoms conditions physicians’ diagnoses, it is important to also note that the ways in which physicians guide patients through the symptom presentation conditions how those symptoms get explained in the first place.

Mental illness continues to be a fundamental problem in society today. It is estimated that approximately half of American adults will experience at least one mental health disorder in their lifetime and 25% will have an ongoing mental health disorder at any given time (Kessler et al., 2005). When compared with other illness categories including cancer and heart disease, the World Health Organization (WHO) concludes that mental health disorders account for the highest incidence of disability in developed nations (WHO, 2004). Moreover, mental health disorders are associated with a reduction in quality of life, social functioning (Saarni et al., 2007), and impairment in carrying out daily tasks (Knudsen et al., 2013). These disorders also play a significant role in morbidity and mortality in the United States and are positively associated with the incidence of other chronic health problems (Chapman et al., 2005).

As Parsons (1951) theorized, illness is a form of social deviance, and by nature this deviance disrupts the normal, everyday functioning of those who are sick. Furthermore, the responsibility of remediating this deviance lies in the physician’s domain. As far as the data presented in this article is concerned, we can see evidence for today’s physicians closely following a Parsonian script in the pursuit of mental health diagnoses. If patients are not showing functional incapacity, their physicians are not taking their symptoms up as indicative of a mental health issue and instead are orienting the diagnosis to a physical problem, echoing prior work which suggests that PCPs frequently address medically ambiguous and/or complex symptoms with somatic interventions (Ring et al., 2005). These data further show that patients must not only strongly assert their mental health symptoms but assert them as unquestionably tied to the inability to carry out everyday life tasks. If patients query, mitigate, or collateralize these symptoms, they are consistently treated as insufficient evidence to indicate a potential mental health problem and instead get taken up as indicative of a physical health issue.

Conclusion

The concern from many in the medical community regarding the misdiagnosis of mental health issues is unsurprising given what this pilot study suggests. Current literature is split on whether this problem of misdiagnosis could be one of underdiagnosis or over diagnosis. A possible explanation for these diagnostic errors can be deduced from the data presented. Potentially, problems of underdiagnosis can be associated with patients who provide a symptoms-only problem presentation but who also have underlying mental health issues. Based on the data in this article, such patients will not get treatment for those underlying mental health issues because they have not demonstrated a disruption to their lifeworld. Importantly, when mental health symptoms are provided as just symptoms, there appears to be bias in the medical community toward providing a physical health before a mental health one.

This preliminary analysis may also help account for the problems of overdiagnosis. Overdiagnosis of mental health problems could be correlated with patients who do not actually have underlying mental health issues, but who over-catastrophize their symptoms by indexing their symptoms as causally linked to a lifeworld disruption. As is shown by the data analyzed, when patients do the extra work to provide a causal link between mental health symptoms and a lifeworld disruption, whether it is indicative of a mental health issue, a mental health diagnosis is often the result. Thus, there is suggestive evidence of an association between the way patients present their mental health symptoms and the misdiagnosis of underlying mental health problems in primary care.

Limitations and future research

Related studies have found that patients who include situational problems in their symptom narrative are actually less likely to receive a psychotropic drug prescription (Chew-Graham et al., 2002; Karasz et al., 2012). Yet in these studies, patients causally linked their lifeworld to their symptoms, rather than causally linking their symptoms to disrupting their life- world, as this study has found. An overlapping concern among these investigations is what happens to patients with these complex symptom narratives, who go improperly diagnosed or leave without a diagnosis, after they exit the clinic. Further research on a larger scale could follow patients with mental health symptoms longitudinally from diagnosis to follow-up. This could help us begin to uncover the medically complicated root cause of their symptoms through following treatment outcomes over time.

Patient and physician self-reported demographic variables, such as gender, race/ethnicity, socioeconomic status, and education level are not available for analysis in this study. This is a limitation of the study and would undoubtedly offer an important level of explanatory power for the findings. As we know, patients from underrepresented groups receive inadequate treatment for mental healthcare (Alegría et al., 2008) which can be attributed to a variety of interrelated social forces (Link and Phelan, 1995). For instance, research has shown that ethnicity and gender both impact how pain gets perceived by clinicians (Al-Hashimi et al., 2015; Norman, 2018). This study hopes to lay groundwork on which health disparities research can be built, where patterns in physician uptake of patient-presented mental health symptoms could be explored for diagnostic biases and differential treatment of patients from marginalized groups.

Author biography

Alexandra Tate is a Postdoctoral Fellow in the Department of Medicine at The University of Chicago. She received her MA and PhD in Sociology from the University of California, Los Angeles, and her BA in Sociology from the Northwestern University. Her research explores the US healthcare system and engages theories of ethnography and conversation analysis to inform her findings. Her interests lie in the complexities of doctor–patient interaction and implications for patient care, focusing primarily on the physician-patient relationship in oncology, primary care, and palliative care settings.