Countertransference towards suicidal patients: a systematic review

Countertransference towards suicidal patients may blur healthcare professionals’ clinical judgment and lead to suboptimal decision-making. We conducted a systematic review of the quantitative studies on this topic. Following PRISMA guidelines, various databases were searched for studies measuring countertransference in healthcare professionals treating suicidal patients. Two authors independently performed screening and the quality of included studies was formally assessed. Ten studies were identified (3/5/2 of low/intermediate/high quality, respectively). Cross-sectional studies showed evidence for specific and adverse countertransference (e.g., disinterest, anxiety, overwhelming, rejection, helplessness or distress) towards suicidal patients. Furthermore, countertransference was prospectively associated with suicidal behavior and ideation in studies that explored this issue, but the meaning of this association remains to be clarified. Healthcare professionals’ characteristics (e.g. professional background, gender, personality traits) influenced countertransference. Suicidal patients elicit adverse countertransference, which should be addressed in clinical practice and through dedicated training.


Introduction
Suicidal patients are highly challenging for healthcare professionals: their wish to die strongly contrasts with the healingperspectives of clinicians and may thus trigger strong emotional reactions (Waern, Kaiser, & Renberg, 2016). Maltsberger described these reactions as "countertransference hate" in a seminal article (Maltsberger & Buie, 1974). Countertransference was originally described by Freud as the "result of the patient's influence on [the therapist's] unconscious feelings" (Freud, 1957;Heiman, 1950). This definition was later broadened (Gabbard, 2001) to encompass the conscious and unconscious emotional, cognitive, and behavioral reactions by therapists towards patients.
To our knowledge, no systematic review has yet examined quantitative studies on countertransference with suicidal patients. This systematic review was thus designed to address three primary research questions: Céline Bourquin and Stéphane Richard-Devantoy contributed equally to this work.
(i) Do suicidal patients elicit specific countertransference from healthcare professionals? (ii) Is countertransference towards suicidal patients related to future suicidal outcomes (i.e. behaviors or ideation)? (iii) Which characteristics of healthcare professionals (e.g. professional background, gender, age) and settings (e.g. outpatient treatment, psychiatric hospital, emergency ward) impact countertransference towards suicidal patients?

Protocol and Registration
Our study followed the PRISMA statements for systematic reviews Moher, Liberati, Tetzlaff, Altman,, & Group, 2009) and was registered on PROSPERO with the number CRD42017078725.

Information Source
Medline, Pubmed, EMBASE, Web of Sciences, PsycINFO and CINAHL databases were searched from their respective first records to 30 September 2017. All the identified articles were evaluated for inclusion, as were references from five relevant PhD dissertations. The search was later updated to September 30, 2018.

Search Strategy
The literature search was conducted for articles in English. A first search combined the largest relevant subject headings related to suicide across all the databases except for Web of sciences (Medline: medical subject heading Self-injuringbehavior; Embase Emtree: Suicidal Behavior; Psychinfo thesaurus: Suicide; CINAHL heading: Suicide) and the heading "countertransference". A second search was performed, based on the finding that our definition of countertransference reactions did not neatly correspond with available keywords but rather encompassed a broader range of concepts. We thus combined the title/abstract [TIAB] terms Suicid* OR selfharm OR automutilation* with the TIAB terms countertransference OR cognitive response* OR cognitive reaction* OR behavioral response* OR behavioral reaction* OR emotional response* OR emotional reaction*, and with tools measuring countertransference (Feeling Word Checklist OR Checklist of feelings OR Gegenübertragungs-Rating OR Impact Message Inventory OR Rating of Emotional Attitudes to Client by Therapists OR Circle OR Psychotherapy Process Q-Set). These tools were identified on the basis of three recent systematic reviews on countertransference (Hayes, Gelso, & Hummel, 2011;Kächele, Erhardt, Seybert, & Buchholz, 2015;Machado et al., 2014).

Inclusion and Exclusion Criteria
Included studies had to meet all of the following criteria: (i) original research papers reporting encounters between healthcare professionals and suicidal patients; (ii) written in English; (iii) employed quantitative measures of countertransference; and (iv) published in a peer-reviewed journal.
Measures of countertransference had to be based on a specific clinical encounter, or a specific set of clinical encounters, with the same patient. Studies measuring general perceptions of, or attitudes towards, self-harming or suicidal patients, or exploring reactions after suicide were excluded.

Study Selection
After excluding duplicates, two of the authors (LM and SRD) independently screened the titles and abstract then full-texts of potential studies for inclusion, and resolved differences in opinion by a consensus process.

Quality Assessment
As no suitable existing assessment tool was identified, we designed one based upon the work of Kirtley (Kirtley, O'Carroll, &O'Connor (O'Connor, Ferguson, Green, O'Carroll, & (Table 1). Total scores ranged from 0 to 14 and quality was graded as low (0-4), intermediate (5-9) or high (10-14). Initial quality assessment was conducted by the first author and further validated by the last author. Disagreements were resolved by discussion until a consensus was reached.
• Patient characteristics explained much more of the variance in negative feelings than in positive Diagnostic gathered only information about the presence/absence of psychosis  (Beck et al., 1979) and MARIS. Comparison between 2 groups for prospective analysis: 53 patients without SA and 6 with SA, as determined by administration of CSSr-S.
One subscale of MARIS was TRQ-SF (see Table 4) Rated just after meeting patient TRQ-SF is a likert-type scale score on a 4 point scale with 10 items. Specifically designed to record countertransference with suicidal patients and based on five items of countertransference questionnaire (Betan et al., 2005), two of Working Alliance Inventory (Horvath & Greenberg, 1989) and three developed de novo by researchers; see Table 4 for further details.
• In cross-sectional analyzes, MARIS total score correlated with lifetime SA and SI and SI in past month but specific correlations between countertransference measured by TRQ-SF and SI/SA past history was not detailed.
• In prospective analysis, countertransference measured by the TRQ-SF independently captured differences between attempters and non-attempters. Cross-sectional analyzes, • Total TRQ-SF, subscales and all items of TRQ-SF were correlated with risk assessment by the clinician. Total TRQ-SF, affiliation subscale and items 1,2,6, and 10 were correlated with lifetime and recent SB and distress and hope subscales as well as items 5 and 8 only with lifetime SB.

Measures of Suicidality
Suicidality was most often measured in a non-formalized way through suicidal ideation, intent to die, past history, or a combination of these factors. However, three recent studies (Barzilay et al., 2018;Hawes et al., 2017;Yaseen et al., 2017) used standardized scales: the Beck Scale for Suicidal Ideation (Beck, Kovacs, & Weissman, 1979) and the Columbia Suicide Severity Rating (Posner et al., 2011).

Measures of Countertransference
Three studies used non-validated tools consisting of items rated on unipolar or bipolar Likert scales that ranged from 2

Do Suicidal Patients Elicit Specific Countertransference?
Four studies (1/2/1 of low/intermediate/high quality; 1/2/1 inpatients/outpatients/emergency patients) found significant cross-sectional associations between suicidality and countertransference (Barzilay et al., 2018;Dressler et al., 1975;Rossberg & Friis, 2003;Soulié et al., 2018), one study obtained mixed results depending on the healthcare professionals' profession (Colson et al., 1986), and one study showed no association (Gillig et al., 1990). Suicidality was associated with more anxiety, confusion, and anger among psychiatric residents assessing suicide attempters (Dressler et al., 1975). Among psychiatric nurses treating suicidal inpatients, it was associated with feelings of being rejected, on guard, bored, overwhelmed, and inadequate (Rossberg & Friis, 2003), and in therapists treating suicidal outpatients, with an "ambivalent combination" of entrapment/rejection and fulfillment/ engagement (Soulié et al., 2018). In addition, a study employing the TRQ-SF found that lifetime suicidal behavior of outpatients was associated with low scores on affiliation and hope subscales and a high score on the distress subscale, whereas recent suicidal behavior/ideation was only associated with the affiliation subscale (Barzilay et al., 2018). Two studies (one with inpatients, one with ward emergency patients) of low quality found mixed or negative results. Colson et al. (Colson et al., 1986) found associations between suicidality of inpatients and countertransference only for nurses and social workers but not psychiatrists. In terms of negative studies, Gillig and al aimed to identify specific patient characteristics or emergency ward clinicians' countertransference as predictors of psychiatric hospitalizations (Gillig et al., 1990). In this study, suicidal ideation did not generate a specific countertransference.

Is Countertransference towards Suicidal Patients Related to Future Suicidal Outcomes?
Five high quality studies (Barzilay et al., 2018;Hawes et al., 2017;Perry et al., 2013;Yaseen et al., 2013;Yaseen et al., 2017) examined potential associations between countertransference towards suicidal patients and future suicidal outcomes, all indeed finding associations (four with outpatients, one with inpatients). The same research team in New-York conducted four of these studies (Barzilay et al., 2018;Hawes et al., 2017;Yaseen et al., 2013;Yaseen et al., 2017). In their first (retrospective) study on psychotherapists, patients who subsequently died by suicide generated more overwhelming and disorganized countertransference, as well as a so-called paradoxical combination of high hope in the treatment with "negative" feelings (avoidance of the patient and discomfort in the treatment), as compared to those subsequently dying from another cause (Yaseen et al., 2013). Following these results, this team developed the previously mentioned tool designed specifically to study countertransference towards suicidal patients (TRQ-SF  see Table 4). This tool includes three subscales: hope, distress and affiliation. Lower hope, higher distress and lower affiliation generate higher subscores, and these last are summed for a total score. In two subsequent studies on psychiatric inpatients, the total TRQ-SF score has been shown to distinguish between future attempters and non-attempters (Hawes et al., 2017), and a specific combination of distress and hopefulness subscores predicted post-discharge suicide behaviors and ideation . In their most recent study, total score and hope subscores were associated with suicidal ideation and behavior in the month following discharge, while distress and affiliation subscores only predicted suicidal ideation (Barzilay et al., 2018). Finally, another team made a longitudinal study (Perry et al., 2013) on long-term psychodynamic psychotherapists and showed that suicidal patients elicited a "negative" countertransference (externally judged and expressed through dogmatic attitudes, rationalizations, inadequate confrontations), which was predictive of worsening suicidality (measured through the level of suicidal ideation and frequency of suicide attempts).

Which Characteristics of Healthcare Professionals and Setting Impact Countertransference towards Suicidal Patients?
Characteristics of Healthcare Professionals Although characteristics of healthcare professionals were rarely taken into account in the examined studies, some associations have been identified. One high quality study found that amongst professionals treating outpatients, psychiatrists experienced a less fulfilled/engaging countertransference than psychotherapists (Soulié et al., 2018). A low quality study showed that suicidal behaviors of inpatients elicited specific countertransference in social workers (e.g. angriness, helplessness) and in nurses (e.g. positive and protective but fearful), but not in psychiatrists (Colson et al., 1986). In the same study, female clinicians were found to be more fulfilled/engaged and less aroused/ reactive than male counterparts (Colson et al., 1986). Finally, one study showed the state anxiety of psychiatry residents to be associated with negative countertransference towards their suicidal outpatients (Barzilay et al., 2018).
Setting Of the two studies where healthcare professionals evaluated suicide attempters on emergency wards, both of low quality, one found psychiatric residents to be anxious, confused, and angry (Dressler et al., 1975), whereas the other showed no influence of suicidality on social workers' countertransference (Gillig et al., 1990). Regarding studies on hospital staff, three (1/2 of intermediate/high quality) had positive results, either on cross-sectional associations (Rossberg & Friis, 2003) or on countertransference as predictors of suicidal outcomes (Yaseen et al., 2013;Yaseen et al., 2017), and one low quality study found results depending on the health professionals' profession (Colson et al., 1986). The four studies, all of high quality, on therapists conducting outpatient treatments had positive results, either on cross-sectional associations (Barzilay et al., 2018;Soulié et al., 2018) or on countertransference as predictor of suicidal outcomes (Barzilay et al., 2018;Perry et al., 2013;Yaseen et al., 2013).

Discussion
The following discussion addresses the available evidence on countertransference of healthcare professionals encountering suicidal patients based on the systematic review of ten studies. Despite important limitations, moderate and high-quality data demonstrates (i) the existence of specific and mostly adverse countertransference towards suicidal patients which depends on the level of current or recent suicidal ideation and suicidal behaviors, and (ii) associations between this countertransference and suicidal outcomes, with the caveat that causality requires further exploration. Such associations were (iii) identified across various settings, and (iv) were influenced by characteristics of the healthcare professionals (e.g. profession, gender), though evidence remains scarce on this last point.

Specific Countertransference towards Suicidal Patients
Our results show that suicidal patients elicit disinterest, angriness, anxiety, confusion, overwhelming, entrapment, rejection, inadequacy, helplessness or distress -but also engagement and fulfillment -among healthcare professionals, which suggests a specific and mostly adverse suicidal-related countertransference. Furthermore, current suicidal ideation appears to be involved in eliciting such a countertransference (Barzilay et al., 2018;Dressler et al., 1975), while the evidence is inconclusive on recent or past suicidal behavior (Barzilay et al., 2018;Colson et al., 1986;Gillig et al., 1990;Yaseen et al., 2017). This may reflect the clinical observation that current suicidality has a stronger dynamic influence on countertransference than past suicidality. Moreover, acutely suicidal patients experience very strong emotions (e.g. hopelessness, self-hate, irritation, entrapment) (Joiner, Simpson, Rogers, Stanley, & Galynker, 2018;Rogers, Galynker, Yaseen, DeFazio, & Joiner, 2017) and countertransference may represent an "emotional contagion" of the healthcare professionals by the negative emotions of the patients, which can be understood from biological (Joiner & Stanley, 2016) and psychodynamic (Richards, 2000) perspectives. This result emphasizes that the "here and now" state of suicidal patients is crucial in suicide risk assessment. Clinicians should take care to avoid overestimating the importance of self-harm history in short-term risk assessment (Carter & Spittal, 2018), and researchers should continue to study acute suicidal states (Joiner et al., 2018;Rogers et al., 2017).
Aggressive behaviors of patients (Lion & Pasternak, 1973;Maier & Van Rybroek, 1995) and the presence of a personality disorder (Betan, Heim, Zittel Conklin, & Westen, 2005;Antonello Colli & Ferri, 2015;A. Colli, Tanzilli, Dimaggio, & Lingiardi, 2014) may also influence countertransference and have to be considered when examining our results. Two of the reviewed studies (Gillig et al., 1990;Rossberg & Friis, 2003) accounted for countertransference related to aggressive behaviors and one (Soulié et al., 2018) for personality disorder, and they confirmed that both are associated with adverse countertransference. These factors were not considered explicitly in the other studies, and their respective influence remains unclear. Future research is warranted: one study recently found that personality disorders play a role in explaining countertransference towards suicidal patients (Michaud et al., 2020) and another article showed that negative affects impact suicidal patients more when they suffer from a personality disorder (Mou et al., 2018), and such an interaction is probable regarding countertransference.

Countertransference Related Suicidal Outcomes
Results consistent across five high quality studies show that specific patterns of countertransference may be associated with future suicidal behaviors (Barzilay et al., 2018;Hawes et al., 2017;Perry et al., 2013;Yaseen et al., 2013;Yaseen et al., 2017) and ideation (Barzilay et al., 2018;Perry et al., 2013;Yaseen et al., 2017). This relationship could be understood in several manners. As the team conducting these studies has argued, countertransference might be a marker of patients' states, in that those who have a higher future suicidal risk elicit specific countertransference reactions. Alternatively, this association may be related to the crosssectional link discussed above between countertransference and concurrent suicidal variables (e.g. suicidal ideation or behaviors). In other words, if (i) highly suicidal patients elicit adverse countertransference and (ii) a high suicidality is a marker in terms of future suicidal behaviors and ideation, then adverse countertransference is likely to be associated with a worse prognosis. Indeed, the core components of countertransference identified in these studies share some similarities with those observed in cross-sectional studies: mostly adverse reactions but sometimes paradoxically combined with "positive" reactions (fulfillment/engagement or higher hope). Finally, this prospective association may be due to the influence of healthcare professionals' countertransference on future outcomes, as was acknowledged in several clinical studies decades ago (Bloom, 1967;Modestin, 1987). While these alternative explanations were briefly evoked in the studies (Barzilay et al., 2018;Hawes et al., 2017;Yaseen et al., 2013;Yaseen et al., 2017), many questions remain. Countertransference may be a consequence of a patient's suicidality, thereby serving as an indicator of suicidal risk, but it may also havedepending on how it is managed or acted out by cliniciansa positive or negative impact on outcomes. To explore these questions, future research should aim to make longitudinal, prospective measurements of countertransference such that their predictive power can be clarified. Quality of care measures, or qualitative methods that include patient perspectives could better characterize how countertransference can influence health outcomes and affect both healthcare providers and patients. Lastly, results of these studies call for replication as the samples, especially regarding the "positive" suicidal outcomes (e.g. suicidal ideation and behaviors) are small.

Impact on Countertransference of the Health professional's Characteristics and of the Setting
Our results also shed light on the importance of healthcare professionals' characteristics. Firstly, professional background seems to have an importance but results were contradictory at times (Colson et al., 1986;Soulié et al., 2018) and inconclusive overall. Further research on this topic is warranted. On the one hand, health professional's training may have an influence on the way they moderate their reactions, e.g. psychotherapists or psychiatrists experiencing psychotherapy as a patient gaining insight on their own countertransference. On the other hand, compared to other mental health professionals, physicians usually take more responsibility for legal (e.g. compulsory admission) and therapeutic (e.g. discharge from hospital, choice of an appropriate treatment) decisions, and have to fulfill multiple tasks (e.g. psychiatric assessment and building therapeutic alliance (K. Michel et al., 2002)). Qualitative research shows that it represents a significant burden (Waern et al., 2016), and thus may engender a specific countertransference. Secondly, differences regarding gender of the health professional (Colson et al., 1986) also require confirmation. Thirdly, the only study measuring variables of the healthcare professionals' personality showed that both state and trait anxiety of healthcare professionals may be associated with negative countertransference (Barzilay et al., 2018). This is in line with theoretical work on countertransference with suicidal patients (Maltsberger & Buie, 1974). Altogether, our review revealed a lack of research regarding the influence of healthcare professional's characteristics on countertransference towards suicidal patients. Indeed, studies on this question show that an important part of the countertransference variance is attributable to the healthcare professionals, though exactly how and in what way is not known (Barzilay et al., 2018;Lindqvist et al., 2017).
Regarding setting, our results were strong for psychiatric hospitals and outpatient clinics, but scarce for emergency wards. Studies on other types of patients in emergency wards found an instant countertransference, defined as an "instant, spontaneous set of feelings that form towards patients, even in the shortest of clinical interactions" (Moukaddam et al., 2019;Moukaddam, Tucci, Galwankar, & Shah, 2016). Further research on such countertransference towards suicidal patients would be of utmost value. On the one hand, "positive" or "negative" countertransference could influence major decisions such as discharge or hospital admission. On the other hand, the presence of adverse countertransference may trigger negative attitudes towards suicidal patients in emergency wards in a variety of hospital staff (Saunders et al., 2012). As part of a recent study in an emergency ward setting (not included in this review), we found evidence of such an instant countertransference towards suicidal patients (Michaud et al., 2020).
Of note, the two identified studies also measuring countertransference after a single clinical encounter (Barzilay et al., 2018;Hawes et al., 2017) towards outpatients outside of an emergency context found it to be predictive of suicidal outcomes. In summary, our results on this point are aligned with theoretical (Jacoby, 2004;Maltsberger & Buie, 1974) and empirical (Richards, 2000) work that focused on countertransference in psychotherapeutic settings, but confirmed that countertransference is also an actual and important consideration in every clinical encounter with patients, even if it is a once-off meeting.

Implications for Practice
Altogether, our results raise the question of how to best manage and make use of countertransference towards suicidal patients. Psychodynamic-oriented training emphasizes the value of personal psychotherapy and individual supervision, and this would ideally be promoted for all mental health professionals, albeit at significant cost. Alternatively, specific trainings on countertransference, for example through case-based learning with simulated patients (McLean, 2016), can be implemented in undergraduate and postgraduate mental health training, in clinical settings such as psychiatric emergency departments, and in training programs on suicide risk assessment. Such trainings should help healthcare professionals increase their capacity to recognize and manage their countertransference (Linn-Walton & Pardasani, 2014) and allow better management of suicidal patients. In addition, we believe that an approach to suicidal patients that is more patient-experience-based, collaborative, and compassionate, with less emphasis on clinician experience-based risk assessment (Hawgood & De Leo, 2016;Konrad Michel, Valach, & Gysin-Maillart, 2017) may reduce the development of adverse countertransference by promoting a more horizontal relationship between healthcare professionals and patients (Konrad Michel et al., 2017). In a recent qualitative study (Petrik, Gutierrez, Berlin, & Saunders, 2015), suicidal risk assessment was indeed shown to greatly depend on the quality of healthcare professionals' communication with patients.

Limitations
Our conclusions must be considered with several limitations. First, study designs, settings, and measurement methods of countertransference and suicidality greatly varied across included studies, hampering any quantitative synthesis of results. Secondly, as no existing tool to assess quality was well-suited to our sample of studies, we developed our own. We, however, formalized this tool based on existing ones (Kirtley et al., 2016;O'Connor et al., 2016). Third, our literature search, although as inclusive as possible, excluded publications not available in English, as well as dissertational theses.

Conclusions
While a recent meta-synthesis attested to the complexity of encountering patients with suicidal ideation or behavior (O'Connor & Glover, 2017), our review demonstrates that suicidal patients may elicit significantly adverse countertransference in healthcare professionals working in various settings, and that this countertransference may be related to suicidal outcomespotentially in a causally bidirectional manner. Clinicians should be aware of the risks related to such adverse reactions, and aim to identify and manage their own countertransference in order to offer the best care possible to suicidal patients.
Sponsor's Role None.
Funding Open Access funding provided by Université de Lausanne.
Data Availability Data available on request from the authors.

Declarations
Ethical Approval Statement All procedures performed in reviewed studies involving human participants were in accordance with the ethical standards of specific Ethics Committees mentioned in each study, and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.
Informed Consent Informed consent was obtained from all individual participants included in the reviewed studies.
Conflict of Interest All authors have no conflict of interest with this text.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.