If You Don’t Ask, They Don’t Tell: The Cultural Formulation Interview and Patient Perceptions of the Clinical Relationship

The public health crises of structural racism (1) and COVID-19 (2) have highlighted the health disparities experienced by BIPOC (Black, Indigenous, and people of color) patients. Even after studies control for age, gender, and income, minoritized racial-ethnic status predicts premature discontinuation from psychotherapy for anxiety (3), depression (4, 5), and trauma- and stressor-related disorders (6). Disparities may be partly due to patients’ mistrust of providers. Mental health clinicians can exhibit implicit biases when making diagnoses, conducting risk assessments, formulating guideline-concordant treatments, and suggesting psychotherapy referrals, thereby exacerbating this mistrust (7). Mistrust may lead to decreased treatment initiation, participation, and continuation (8), resulting in disengagement from services before patients’ symptoms, functioning, or quality of life have improved (9). Mistrust can worsen disparities in access to care, if BIPOC patients avoid mental health services, and in the type of care that is received, if implicit biases worsen clinical outcomes.

Evidence suggests that improving patient-clinician communication could overcome mistrust (8). Many BIPOC patients believe that non-BIPOC clinicians cannot appreciate crucial aspects of their lives without discussing the patient’s racial-ethnic identity (10). BIPOC patients may feel less satisfied with care if clinicians ignore how racial-ethnic background affects their health care experience, such as through the patient-clinician relationship or instances of discrimination (11). This dissatisfaction can lead patients to end treatment if they think that clinicians are avoiding these topics because of a lack of empathy or interest (12). However, clinicians may avoid these topics because of a lack of skills—not a lack of interest; most training and supervision programs do not teach clinicians how to discuss racial-ethnic backgrounds with patients (3).

To help clinicians consider how differences in racial-ethnic identity could affect the clinical encounter, DSM-IV included the Outline for Cultural Formulation (OCF) in 1994 (13). The OCF encourages clinicians to consider patients’ experiences across four domains (cultural identity of the individual, cultural explanations of illness, cultural levels of psychosocial support and functioning, and cultural elements of the patient-clinician relationship) and to consider information influencing diagnosis and treatment as a fifth domain (13). As part of the DSM revision process, the DSM-5 cross-cultural issues subgroup (DCCIS) reviewed all studies on the OCF since 1994 and identified two themes regarding the patient-clinician relationship (14). First, despite clinicians’ best efforts, the medical encounter can be influenced by stereotyping, discrimination, racism, and implicit bias, so clinicians should routinely engage in self-introspection (14). Second, to build trust and anticipate treatment barriers, clinicians should ask patients during the initial encounter about negative experiences with the health system and potential miscommunication (14).

In response to these recommendations and clinicians’ suggestions for more implementation guidelines on the OCF, the DCCIS created the Cultural Formulation Interview (CFI), a 16-item, semistructured protocol based on a literature review, a field trial in six countries with 321 patients and 75 clinicians, and expert consensus (15). Patients and clinicians who participated in the field trial found the CFI feasible, acceptable, and clinically useful (16), and its revision appeared in DSM-5 (17). During the DSM-5 field trial, some clinicians contended that the CFI’s question on identity differences affecting the patient-clinician relationship would not be feasible or acceptable because patients could refuse to answer the question or may want to avoid offending their intake clinicians (18). This study aimed to explore how patients responded to this CFI question on the patient-clinician relationship in a pilot study to implement the CFI among lower/middle-income BIPOC outpatients, and to analyze patient responses in accordance with a widely known framework from psychotherapy for understanding cross-cultural patient-clinician relationships.

Methods

Study Setting

The CFI was implemented at Flushing Hospital Medical Center (FHMC), a community hospital in New York City’s multicultural borough of Queens. Of approximately 1,500 active outpatients, about 30% (N=450) identified as non-Latinx White, 30% (N=450) as Latinx, 32% (N=480) as non-Latinx Asian, and 8% (N=120) as non-Latinx Black. To initiate treatment at FHMC, all patients attend a 60-minute intake session with a social worker (session 1), a 45-minute appointment with a psychiatrist (session 2), and a 30-minute psychotherapy session (session 3) within 1 month. Administrators assured clinicians that a CFI training session would fulfill the hospital’s continuing education requirements on cultural competence and that all sessions were billable.

FHMC’s clinic is referral based, and intake coordinators introduced the study when scheduling patients’ intake session, emphasizing that participation was voluntary and would not affect service eligibility. Each new patient received a flyer on the study during registration, and interested patients met with a separate research team to give consent.

Clinicians were recruited through monthly staff meetings. Interested clinicians approached the research team to discuss the study. Flyers describing the project were placed in all clinicians’ mailboxes. The institutional review boards at the New York State Psychiatric Institute and FHMC approved the study. All participants provided written informed consent.

Participants

Eligible patients were from any racial-ethnic background, ages 18–80, spoke English (including English as a secondary language to match their clinician’s language fluency), and presented to establish care. Individuals were excluded if they presented with acute suicidality, intoxication, substance withdrawal, or conditions such as dementia, intellectual disability, or florid psychosis that might interfere with answering CFI questions.

FHMC clinicians who could practice with an independent license were eligible for participation. Clinicians attended a 2-hour session that replicated the DSM-5 CFI field trial training protocol, which consisted of reading the CFI aloud, watching videos illustrating use of the CFI with a patient (e.g., 19, 20), behavioral simulations to practice questions, and a question-and-answer session with researchers (18, 21).

Study Design

Patients and clinicians completed demographic surveys before the interview. All clinicians agreed to start their intake sessions with the CFI, followed by their standard diagnostic assessment. A research assistant audio-recorded the interviews with patient and clinician consent. A copy of the CFI was provided to each clinician. Another copy was available for patients upon request. All participating patients received $30 in compensation for their time.

Data Source: Open-Ended Questionnaires

This study analyzed all patient responses to the 16th and final question of the DSM-5 CFI on the patient-clinician relationship. Clinicians asked patients the question, “Sometimes doctors and patients misunderstand each other because they come from different backgrounds or have different expectations. Have you been concerned about this and is there anything that we can do to provide you with the care you need?” (17).

The CFI’s instructions to clinicians were to

Data Analysis

Data collection ran from April 2017 to December 2018. Frequencies and percentages were used to characterize the demographic identities of patients and clinicians. The analytical team comprised two research psychiatrists, both trained in clinically applied medical anthropology. The first author transcribed the CFI from each audiotaped session and checked each transcription for accuracy and spelling or grammatical mistakes by reading it while listening to the recording.

Data analysis followed procedures for deductive content analysis, whereby previous theories are tested against new data through a theoretical framework that produces codes and categories that can be compiled into codebooks (22). To analyze responses to the CFI question, we used Comas-Díaz and Jacobsen’s typology of reactions when patients and clinicians had different identities (23). This typology classifies how patients and clinicians may relate to each other on the basis of whether they have similar or different cultural identities (23). Experts have used this framework to recommend that clinicians ask patients about their negative experiences with the health system associated with cultural background while completing the OCF (24) and the CFI (14).

In nearly all dyads in this study, the patients and clinicians were from different backgrounds. Therefore, the first author created a codebook listing all possible reactions and their definitions from Comas-Díaz and Jacobsen’s classification of reactions for patients and clinicians with different cultural identities (23). These themes included unsureness (the patient was not sure whether the clinician’s identity would affect the therapeutic relationship or the response was unclear and the clinician did not probe for details), overfriendliness (the patient wanted to be considered “a good patient” to avoid negative reactions from the clinician, who was perceived to be more powerful), denial of cultural factors (the patient worked through a discussion of their identity or the clinician’s but decided that identity factors would not affect the relationship), mistrust (the patient did not believe that the clinician could understand the patient and doubted the clinician’s motivations), and ambivalence (the patient struggled to work through possible negative feelings toward the clinician but simultaneously wanted a positive emotional attachment). We included an additional theme, called “no concern,” which was not in Comas-Díaz and Jacobsen’s classification of reactions, and defined it as the patient not voicing concerns about the clinician’s identity affecting the relationship.

Responses where patients expressed concerns were further coded by using three mutually exclusive themes: concerns only about past clinicians, concerns only about current clinicians, and concerns about past and present clinicians. The first and second authors coded each CFI response with one code. Both team members met weekly to discuss codes and coded each response with 100% concordance.

The first author uploaded all transcriptions into NVivo and clustered codes into categories to derive themes. This process involved generating queries and reports on codes in NVivo, exploring patterns, and drafting analytical memos on themes and subthemes. For analytical rigor and validity, the first author used an audit trail of memos and meeting notes, triangulation of narrative data, and peer debriefing sessions and checked the coding done by other team members.

Results

Participant Characteristics

Twenty-seven patients enrolled in this study. Eleven identified as male and 16 identified as female. Their ages ranged from 18 to 67 years (mean±SD=32.8±14.0). Two identified as non-Latinx White, 14 as Latinx, eight as non-Latinx Black, and three as non-Latinx Asian. Their primary languages were English (N=21), Spanish (N=5), and Polish (N=1). Their primary sources of income were employment (N=12), family assistance (N=5), disability (N=3), public assistance (N=3), and pension (N=1); three patients did not answer. One patient earned $120,000 in annual income, and the rest earned $60,000 or less annually. Their primary intake diagnoses by class were depressive (N=11), anxiety (N=9), bipolar (N=3), adjustment (N=2), psychotic (N=1), and somatic symptom (N=1) disorders.

Two clinicians, both social workers, enrolled in the study. One identified as male, the other as female, and both as non-Latinx White. The male therapist was a native citizen of the United States, and the female therapist was a naturalized citizen born in Central Asia. Neither had any experience with the CFI before the study.

Themes on the Cross-Cultural Patient-Clinician Relationship

Table 1 presents themes by patient-clinician dyad. The most prevalent themes were mistrust (N=12), no concern (N=10), ambivalence (N=3), and unsureness (N=2). Representative quotations are presented below.

Mistrust.

Mistrust was the most common theme and was expressed in dyads where patients and clinicians differed in demographic identity. Patient concerns were about past clinicians (N=4), present clinicians (N=4), and both past and present clinicians (N=4).

Mistrust manifested as concerns that immigrant clinicians do not understand native-born patients. A 51-year-old Black man described prior treatment experiences to a White female therapist:

Mistrust appeared when patients perceived that clinicians had different religious identities, as in this exchange between a 67-year-old White Latina and a White female social worker:

Mistrust also emerged through racial-ethnic differences. A 62-year-old Latino told his White male therapist: “Because there is a different culture, sometimes you won’t understand what I’m going through or relate with it because it’s not your culture. And the best thing is to see if I can see a Latin therapist. If not, I’m flexible.”

Mistrust also took the form of gender differences. A 45-year-old Latina and her White male social worker had this exchange:

Finally, mistrust appeared from socioeconomic differences. A 28-year-old Black Latina told her White male social worker, “I think that’s why I stopped therapy before. Because I had an Asian therapist and she—I don’t think she understood about my poverty.”

No concern.

No concern was the second most prevalent theme. Eight patients simply said “no” in response to the final CFI question, and their clinician proceeded to the rest of the intake evaluation. Two patients explained that they had no concerns about cultural differences. Representative of both explanations, a 56-year-old Black woman told her White male social worker, “You know what? If this is for me to get help, I don’t give a damn what color you are, what sex you are. If you can help me, I’m going to put it out there, and maybe you can help me.”

Ambivalence.

Ambivalence appeared in three instances where patients verbally worked through the possibility of negative feelings toward their clinicians but wanted to develop positive emotional attachments. Ambivalence appeared only in relation to present clinicians, not past clinicians.

A 24-year-old Black Latina responded to her White female social worker in a representative manner:

Unsureness.

This theme appeared in two patient responses, one referring only to a past clinician and one referring only to current clinicians. In both instances, patients said, “I don’t know” but did not explain further. An exchange between a 30-year-old Black woman and her 50-year-old White male social worker was representative. The patient said “I don’t know” twice and used the word “probably” to think through a possible misunderstanding.

Discussion

This article presents data on the CFI’s question regarding identity differences in the patient-clinician relationship. A plurality (N=12, 44%) of new patients discussed their mistrust of clinicians. A minority (N=3, 11%) expressed ambivalence. Hence, over half of this sample verbalized apprehensions about treatment providers during their intake session. This finding merits research into how patients perceive expectations of care and providers within treatment.

Slightly over a third (N=10, 37%) of patients expressed no concerns about being treated by a clinician with a different identity. In our study, patients were asked about potential misunderstandings with clinicians from different backgrounds only once, so we could not determine whether patients did not have such concerns or whether they felt uncomfortable articulating them. Future work could analyze sessions longitudinally to elucidate whether interpretations of the patient-clinician relationship evolve over time, in affectively positive or negative directions, along with reasons.

Consistent with the CFI’s approach to helping patients co-construct a narrative about their illness and treatment experiences (17), clinicians can broach this topic during the initial session to encourage solutions and anticipate problems. An objection could be that negative reactions should be explored only after trust is established. However, studies repeatedly show that patients discontinue treatment prematurely when they think identity differences with clinicians are unbridgeable, even after one session (9). Delaying discussions of identity differences may reinforce patient mistrust by communicating that therapists do not care about their experiences.

Our study had several limitations. First, our data came from one site and may not be generalizable. Second, only social workers enrolled, not other types of clinicians. Studies are needed across more clinician disciplines. Third, six out of the 27 patients did not speak English as a primary language, which might have affected their ability to fully express themselves, especially if they had concerns that their intake clinician would misunderstand them. Fourth, our study focused only on patients’ perceptions of the relationship. A CFI supplementary module on clinicians’ perceptions of the patient-clinician relationship could be used in conjunction with the CFI (23).

Conclusions

This study suggests that the CFI can elicit patient perceptions about past and present clinicians. Clinicians can use the CFI to allow BIPOC patients to describe their experiences within the health care system and to anticipate treatment barriers. Future work can determine whether the CFI helps alleviate disparities in care by fostering communication and trust so that BIPOC patients continue treatment to improve symptoms, quality of life, and functioning.