|Home | About | Journals | Submit | Contact Us | Français|
Many older adults who die by suicide have had recent contact with a primary care physician. As the risk-assessment and referral process for suicide is not readily comparable to procedures for other high-risk behaviors, it is important to identify areas in need of quality improvement (QI).
Identify patterns in physician-patient communication regarding suicide to inform QI interventions.
Qualitative thematic analysis of video-taped clinical encounters in which suicide was discussed.
Adult primary care patients (n=385) 65 years and older and their primary care physicians.
Mental health was discussed in 22% of encounters (n=85), with suicide content found in less than 2% (n=6). Three patterns of conversation were characterized: (1) Arguing that “Life’s Not That Bad.” In this scenario, the physician strives to convince the patient that suicide is unwarranted, which results in mutual fatigue and discouragement. (2) “Engaging in Chitchat.” Here the physician addresses psychosocial matters in a seemingly aimless manner with no clear therapeutic goal. This results in a superficial and misleading connection that buries meaningful risk assessment amidst small talk. (3) “Identify, assess, and…?” This pattern is characterized by acknowledging distress, communicating concern, eliciting information, and making treatment suggestions, but lacks clearly articulated treatment planning or structured follow-up.
The physicians in this sample recognized and implicitly acknowledged suicide risk in their older patients, but all seemed unable to go beyond mere assessment. The absence of clearly articulated treatment plans may reflect a lack of a coherent framework for managing suicide risk, insufficient clinical skills, and availability of mental health specialty support required to address suicide risk effectively. To respond to suicide’s numerous challenges to the primary care delivery system, QI strategies will require changes to physician education and may require enhancing practice support.
Suicide is a stigmatized behavior accounting for more than 30,000 deaths1 and more than 300,000 self-harm-related emergency department visits2 per year in the United States. These rates have remained relatively stable despite widespread prevention efforts, including significant increases in antidepressant use.3,4 Ten percent of primary care (PC) visits involve mental health issues,5 and PC is the most common venue for depression care in the US,6 with over 20 million depression-related visits annually.7 People who die by suicide are 2.5 times more likely to have seen a primary care provider than a mental health specialist in the month preceding their death.8,9 Primary care physicians (PCPs) who see older adults will inevitably be confronted with addressing suicide risk in a greater proportion of their patients, given the aging of the population and the elevated rates of suicide in older adults.
Physicians have difficulty discussing suicide,10–12 a topic heavily laden with psychosocial stress and rarely considered within the purview of primary care. Very little is known about the factors that inhibit suicide-related communication in primary care settings, despite numerous studies on other emotionally laden and stigmatized health issues. For example, Sugg and Inui13 found that physicians were reluctant to explore domestic violence because of concerns of discomfort, fear of offending, powerlessness, loss of control, and time constraints. Conversely, patients were more likely to disclose being victimized if they perceived their physician to be caring, protective, and easy to engage, or the physician offered a follow-up visit.14 All of these factors, both inhibiters and facilitators, would appear to be pertinent for suicide-related communication.
Inquiries regarding suicide are rare in PC, even when indicated.15–18 Feldman et al.19 analyzed data from a standardized patient study involving 298 visits to 152 PCPs from a variety of health care delivery systems and found that even when inquiry to suicide ideation was clearly indicated for diagnostic clarity, it occurred in only 36% of the visits. Using a micro-analytic approach to studying inquires from the same standardized patient20 data set, Vannoy et al.21 reported that the PCPs who inquired about suicide did so in a manner consistent with best practices regarding question formulation.22–24 Specifically, physicians frequently initiated the dialogue from within a psychosocial context and followed their inquiries with contextually relevant material. Unfortunately, both studies were limited by the use of standardized patients who all denied the presence of suicidal behavior; hence, only the initial aspect of suicide risk management, namely inquiry, could be analyzed. In a psychological autopsy study of late-life suicides in Norway, Kjolseth, Ekeberg, and Steihaugh 12 concluded that older adults and their health care providers are aware of communication difficulties related to suicide risk. Their investigation revealed that older patients may be reluctant to discuss suicide for fear of losing autonomy and general distrust of health services, while providers were concerned about their ability to help and the interest of older adults in being helped. Davidsen18 used a retrospective approach to investigate the impact of patient suicide on physicians and found that while rare, patient suicide has intense emotional impact on physicians that lasts for many years. However, Davidsen found no relationship between emotional impact and future propensity to discuss suicide with patients, and that uncertainty about how to discuss the topic was an inhibitor.
Identifying a health problem is necessary but not sufficient for achieving a desired outcome. Detailed assessment, developing a treatment plan, and executing that plan must follow case identification. This is the first investigation of the ensuing conversation following identification of suicide ideation in primary care encounters involving actual patients. In this study, we conduct a thematic analysis in order to examine variations in the conversation related to suicide inquiries in late-life primary care encounters. Because very little is known about suicide risk management in primary care, we chose a qualitative methodology aimed at generating hypotheses for future research. We focus both on the immediate reaction to identification of suicide ideation, and the subsequent dialogue and to what extent the physician facilitated an environment conducive to patient engagement and provision of effective treatment. Informed by the interpersonal factors that influence conversation regarding domestic violence,13,14,25,26 we attempt to characterize physician behaviors that may interfere with elucidation of suicide risk factors, engagement in treatment or referral to mental health specialty.
Videotapes were selected from a sample of patient office visits with their usual source of care.27 The medical practices included an academic medical group in the Southwest, a private managed care group in a Midwest suburb, and a number of fee-for-service solo practitioners in a Midwestern inner city. Physicians and patients were informed that the videotapes would be used to study and improve patient-physician interaction, and that the videotapes would be archived for future research. All relevant institutional review boards approved the research protocol.
The sample used for this study included 385 visits to 35 physicians, all of whom had completed their training (14 internal medicine, 8 family medicine, 13 other) at the time of the study. To be eligible, patients had to be at least 65 years of age, identify the participating physician as their usual source of care, and provide informed consent. Eligible patients were identified by office managers. They were invited to participate when they came to the clinic for a visit, regardless of the nature of the visit (e.g., acute upper respiratory infection, or for routine checkup for diabetes or hypertension). If they expressed willingness to participate, informed consent was obtained, and the encounter with the physician was videotaped. Research assistants approached patients whom they believed would be willing to allow the taping of additional visits, based on their perception of positive feedback from some patients. Physicians were not asked to recommend patients for multiple taping. Patient participation rates ranged from 61% to 65% at the three sites. Mental health issues were addressed in 84 of 385 videotaped visits.28
Suicide was discussed in 6 of the 84 encounters in which mental health was addressed. Physician Adam1 is a 34-year-old, White male internist with 6 years of practice experience along with geriatric training during residency working in a managed care group. He was involved in one encounter with a female patient (Patient Alice). Physician Becky is a 50-year-old African-American, female internist with 6 years of practice experience, including a geriatric specialty focus, working in an academic medical center. She was involved in two encounters, one with a male patient (Patient Bob) and one with a female patient (Patient Betty). Physician Curt is a 48-year-old White, male family practice physician with 16 years of practice experience, and no geriatric specialty training, working in an academic medical center. He was involved in three encounters, with two males (Patient Carl) and (Patient Charles) and one female (Patient Carmel). Patient characteristics are presented in Table 1.
Five investigators, two clinical psychologists with research backgrounds in suicidology, a board-certified family physician researcher and two health services researchers with experience in qualitative research methods applied to primary care encounters, performed a qualitative analysis on the six videotapes.
We utilized a thematic analysis approach to examine our data.29 We began with each investigator watching a pair of encounters with two physicians, followed by a series of telephone conferences in which we reviewed each team member’s field notes and looked for emerging themes related to the manner in which physicians responded to the indication of suicide risk. The team discussed observations from each pair, noting similarities, differences, and overall impressions. We examined pairs of encounters to highlight contrasting approaches as much as congruency. During our team meetings we reviewed transcripts and video segments to assist in recall, clarification, and thematic identification. The process was repeated for the remaining four physician-patient encounters. Following the dyadic reviews, our impressions were refined and applied to all encounters via additional review of transcripts and video segments based on group discussion. Differences in interpretation were discussed during phone conferences and via e-mail. All decisions regarding identification of relevant themes were achieved via consensus. After our last debriefing from viewing all segments, we initiated a series of meetings in which we prioritized topics identified as being of interest and further refined our analytic focus.
There was significant range in visit length (minimum=1,146 s, maximum=2,630 s), yet all were above the average length of visits in which suicide was not discussed (mean=1,141 s). The three physicians also had considerable variation in their respective average encounter length (Table 2).
Commonalities and differences were readily apparent in these encounters. Across all six encounters we were impressed with the level of general caring displayed by all three physicians with each patient. At least some attention was given to psychosocial factors in patients’ lives, and depression in particular was discussed and treated in one form or another in all but one encounter. We reached consensus on the notion that the physicians identified life stress as a topic of concern and that they appeared willing to address suicide within the office visit.
Despite general similarities we observed distinct communication patterns. The first metaphor that emerged from our team discussions was that of a doorway leading to treatment being opened or shut. We repeatedly saw therapeutic opportunities knocking on the door (e.g., the patient bringing up suicidal thoughts), which led the physician to open the door a crack (e.g., the physician asking questions about such thoughts), and then, all too often, shutting the door precipitously (i.e., no attempt to address suicidal thoughts using established therapeutic methods or an articulated plan to follow up on this topic). We present patterns of communication below with excerpts from transcripts of the visits. See Table 3 for a summary of four exemplary visits, including the relevant communication pattern and a breakdown of key components of our analysis.
Physician Adam demonstrated what we came to view as an argumentative approach. His response to Patient Alice’s expressed feelings, that her life was not worth living and that she had nothing to offer her family except an inheritance, was to argue that he did not agree with her and that he thought her family would not agree with her either. Interestingly, this encounter resulted in the longest sequence of exchanges related to suicide. In response to the lack of overt empathic validation, Patient Alice continued to plead her case. Physician Adam resorted to self-disclosure indicating that he felt bad about living far from his own parents and that this prevents them from spending time with their grandchildren. The disclosure shifts the topic off Patient Alice’s thoughts of suicide. She makes one more attempt to justify her thoughts that ending her life is a rational decision and then rests her case. Physician Adam eventually informs her that he is worried she is depressed. Patient Alice responds by indicating that she’s always felt that way. No treatment is offered. The suggested follow-up is 6 weeks with the stated purpose of checking on her blood pressure.
A metaphor that emerged in this encounter was maneuvering around the topic, in which two parties attempt to get their points across with limited success, if any. Even their physical positioning evoked a power differential, as Patient Alice sat on the exam table the entire time with Physician Adam standing in front (over) her or pacing around her. With respect to treatment engagement, despite efforts to be heard, this patient finds the door is being held shut (summarized in the first column of Table 3).
The second column of Table 3 presents the first of two encounters in which Physician Becky addressed suicide with her patients; she is seeing “Patient Bob,” who presents with his spouse. Physician Becky initiates the visit with a sequence of compliments and statements that clearly indicate affinity towards and care about the patient’s well-being. Whereas Physician Adam stands throughout the encounter while Patient Alice sits on the exam table, Physician Becky uses her wheeled stool to move in close to Patient Bob, who was seated in a chair, and occasionally touches him caringly on the arm or leg.
Opportunities for providing therapeutic empathy are missed,11 yet Physician Becky performs a more thorough assessment of depression symptoms than Physician Adam, and she discusses treating depression with medications (Zoloft). Surprisingly, she also suggests admission to an inpatient mental health unit as a treatment option. This extreme polarity of options, either an increase in the dose of Zoloft, or admission to an inpatient unit, may reflect limited mental health services resources or limited knowledge about options for treating depression. Regardless of underlying causes for limited options, the patient does not show any interest in entering an inpatient unit. Rather than explicitly rejecting that option, however, he states that giving his wife a hug is the best thing he can do to make himself feel better. This serves to transition to a new topic, and Physician Becky does not make any additional attempt to motivate Patient Bob to seek inpatient mental health treatment. They then begin engaging in idle chitchat with Physician Becky asking where the couple has been eating out lately. The topic of suicide does not come up again. Prior to ending the visit, Physician Becky does tell the patient she wishes to increase the dose of his antidepressant medication. There is no clear follow-up appointment, just a general indication that the patient will be seen in the clinic again in the future.
The ebb and flow of this encounter reflect a strong pull for the patient to reassure the physician that he is okay. It is remarkable how the topic flips from statements of bio-psycho-social stress to exclamations of how happy the patient is and how good life is now. One can imagine how the physician’s strong expressions of affiliation make it difficult for the patient to be “not okay.”
In summary, this encounter presents a very different pattern from the “life’s not that bad” pattern of conversation. Here, suicide is acknowledged, and care is expressed but there is a quick move to chitchat in an apparently aimless manner with no obvious therapeutic goal. This results in a superficial and misleading connection with rambling conversation that buries the topic of suicide amidst small talk about restaurants. By avoiding dialogue directly focused on the suicide ideation, the patient may be interpreting the conversation as an indication that while the physician cares about him, she does not want to hear about such emotionally distressing thoughts and feelings. The door is open, but it doesn’t lead to a desired place.
The four remaining encounters can be characterized by the physicians initially addressing the issue of suicide and co-morbid mood disorder, and offering some potential courses of action, only to drop the subject precipitously with no clear sense of closure or treatment plan. This pattern was displayed between Physician Becky and Patient Betty, with whom the physician appeared to have a relationship that differed qualitatively from the relationship she had with Patient Bob (as evidenced by no physical contact and less shared social chitchat). From the excerpt in the third column of Table 3, we see a pragmatic response to Patient Betty’s disclosure of suicide ideation, with the physician asking, “and do you feel that way now?” In response to the patient’s cagey denial of suicide ideation “at this moment,” the physician moves to a discussion of depression treatment. The conversation does not return to suicide.
Physician Curt consistently takes a proactive approach to assessing his patient’s emotional functioning. He probes further when his patients make references to suicide, e.g., “when did you start feeling that way,” and he validates frequently by reflecting back the patient’s statements. For example, Physician Curt said to Patient Carmel, “I think I hear you saying it's hard growing old. It's hard being 88.” Physician Curt’s use of caring reflection and validation invites his patients to express the psychosocial distress they are feeling.
In his interactions with Patient Carl, Physician Curt does not hesitate when the patient explains that he has a gun and ammunition, and intends to use it if he reaches a critical point. Instead, Physician Curt immediately validates the patient by acknowledging his thoughts about using the gun, and he communicates that he needs to know that the patient is safe. He does so by eliciting a no-harm commitment from the patient, agreement to an antidepressant medication, and agreement to a referral to a psychiatrist.
Unfortunately, in all of the encounters in which the physician clearly acknowledges the presence of suicide ideation and probes for further information, we were struck by the abrupt transitions away from the topic of suicide. It was as if the physicians ran out of tools to address the topic and, worn down, had to resort to changing the subject. The door is open, but the patient is let into a hallway that leads to somewhere, but where?
Geriatric patients present in primary care with suicide ideation in the context of complex bio-psycho-social challenges. Some primary care physicians are able to recognize when suicide is to be explored and readily respond by expressing concern, in some form. Yet a fundamental expression of concern for patient well-being will not address this complex issue in all (or even in most) instances. We stress that in all or our cases, it appears that physicians are genuinely concerned about their patient’s well being as evidenced by their giving substantial attention to psychosocial factors. It does appear that discussing suicide is likely to extend the visit time above an average length; yet two out of three of the physicians who did discuss suicide had overall average session times that were longer than the average sample session regardless of whether suicide was discussed or not. Hence, it may be that physicians who take more time with their patients are more likely to discover suicide ideation and address it.
Three broad patterns of conversation illustrate our main conclusion that expressing concern is necessary but insufficient: (1) Argumentative: Life’s Not That Bad; (2) Superficial: Engaging in Chitchat; and (3) Insufficient: Identify, Assess, and…?
To avoid drawing the patient into the argumentative pattern, physicians can make more empathic responses (as evidenced by Physician Curt) by employing emotional validation (e.g., “you are really having a hard time these days…”) and active listening (e.g., “so what I hear you saying is…”).
However, as illustrated by the superficial and insufficient patterns, caring alone is not inevitably reassuring, nor will it instill a sense of hope or lead to the development of an action plan.
Our study has significant limitations. Given the small data set, we clearly have not identified all patterns of physician-patient conversation when discussing suicide, nor do we know how representative these patterns are in routine practice. The extent to which suicide was a topic of discussion between physician and patient prior to study initiation is unknown. However, as can be seen from the scant risk assessment and treatment planning, we would encourage more attention to these aspects of suicide risk management regardless of prior discussions given the dynamic aspect of suicide-related behaviors. Our intention was to generate and elucidate what may be prototypical patterns leading to undesirable encounters. Furthermore, we have applied common axioms in clinical communication theory to these encounters in the absence of direct patient reactions to their visit. It is possible that, contrary to our interpretation of the data, patients did not actually feel as if the door to treatment was not open. However, our observations regarding the lack of treatment planning and structured follow-up are relatively objective and represent opportunities for quality improvement. Future research is needed to determine the prevalence of these patterns and their impact, both on longer-term patient outcomes as well as physician practice.18
Suicide is an inherently psychosocial issue with existential connotations for many. It is not readily comparable to diseases for which standard medical procedures exist. Nonetheless, suicide risk assessment is analogous to other difficult, stigmatized issues, such as domestic violence, that primary care physicians have addressed effectively. Older males, while at highest risk for suicide, may be particularly reluctant to initiate conversation related to suicide.12 Getting past barriers to initiating conversation about suicide is critical but insufficient. Adapting a core element of managing chronic illness in primary care, namely longitudinal monitoring and structured follow-up, along with timely referral to specialty care may result in better patient engagement and ultimately save lives.
Role of Funders and Data Funding for this study was provided in part by the National Institute on Aging (AG15737) and the National Institute of Mental Health (MH01935, KL2RR025015, K24MH072712). The design, conduct, data collection, analysis, and interpretation of the results of this study were performed independently of the funders. The funding agencies also played no role in review or approval of the manuscript.
Conflict of Interest Statement None disclosed
1To protect privacy, we use pseudonyms when referring to the patients and physicians.