Exploring the Impact of Language Concordance on Cancer Communication

PURPOSE:

Patients with cancer who have limited English proficiency are more likely to experience inequities in cancer knowledge, timely care, and access to clinical trials. Matching patients with language-concordant clinicians and working with professional interpreters can effectively reduce language-related disparities, but little data are available regarding the impact of language-concordant interactions in oncology care. This study aimed to assess the use of the Roter Interaction Analysis System (RIAS) in language-concordant and -discordant interactions for patients with non-English language preference presenting for an initial oncology visit at four New York City hospitals.

METHODS:

We used the RIAS, a validated tool for qualitative coding and quantitative analysis, to evaluate interactions between 34 patients and 16 clinicians. The pairings were stratified into dyads: English language-concordant (n = 12); professionally interpreted (n = 11); partially language-concordant (n = 4, partially bilingual clinicians who communicated in Spanish and/or used ad hoc interpreters); and Spanish language-concordant (n = 7). A trained Spanish-speaking coder analyzed the recordings using established RIAS codes.

RESULTS:

Spanish language-concordant clinicians had almost two-fold greater number of statements about biomedical information than English language-concordant clinicians. Spanish language-concordant patients had a higher tendency to engage in positive talk such as expressing agreement. The number of partnership/facilitation-related statements was equivalent for English and Spanish language-concordant groups but lower in professionally interpreted and partially language-concordant dyads.

CONCLUSION:

Language concordance may facilitate more effective biomedical counseling and therapeutic relationships between oncology clinicians and patients. Future research should further explore the impact of language concordance on cancer-specific health outcomes.

INTRODUCTION

Effective communication between physicians and patients is a therapeutic component of the doctor-patient relationship that contributes to better health outcomes.¹ Approximately 66 million people in the United States speak a language other than English. About 38.8% of them report limited English proficiency (LEP), defined by the US Census Bureau as speaking English less than very well.² Several studies have shown that LEP patients experience more challenges in attaining quality care and have worse outcomes because of communication barriers. Moreover, even patients who may not meet LEP definitions in everyday contexts may most effectively communicate about health issues in a language other than English.³ Language discordance between clinicians and patients with LEP during a health care visit has been associated with decreased patient access to preventative services, satisfaction with care, understanding of goals of care, and adherence to treatments.^4-6

In oncology, clinicians and patients often engage in sensitive, complex conversations that may include the communication of serious news, discussions about prognosis, and jargon-heavy medical information related to treatment options. Therefore, adequate communication is a critical component of cancer care. When language discordance exists between an oncology patient and their cancer care provider, the therapeutic relationship can be negatively affected, and the patient might not receive complete or accurate information regarding their care. Research has shown that LEP patients are more likely to have insufficient knowledge of their cancer diagnosis, receive delayed cancer treatment, and be excluded from participating in clinical trials,^7-12 and oncologists feel they provide less patient-centered treatment discussions.¹³

Evidence suggests that matching patients with a language-concordant clinician and working with professional interpreters effectively reduces health care disparities for LEP populations.^6,14,15 One study found that interpreter use resulted in better patient-reported understanding of their diagnoses and treatment plans, higher satisfaction ratings, and improved health outcomes, including higher cancer screening rates, better access to care, and improved medication adherence.¹⁴ However, studies show that certified interpreters are not always used appropriately, and clinicians report using their partial non-English language skills and/or ad hoc interpreters (ie, family members, friends, or bilingual staff not trained in interpretation) to communicate with LEP patients.¹⁵ Health care organizations' inconsistent and conditional compliance with the federal regulations established by Section 1557 of the Affordable Care Act and guidelines from the National Standards for Culturally and Linguistically Appropriate Services in Health Care (CLAS Standards) exacerbate language-related disparities.^16-18 A systematic review found that professional interpreters were consistently associated with fewer errors, better clinical outcomes, fewer disparities in utilization of resources, and improved satisfaction with care compared with ad hoc interpreters.¹⁹

Although communication is an essential aspect of cancer care, no studies have assessed the quality of communication in LEP populations with cancer nor have any used direct observations of clinical encounters in this setting. The Roter Interaction Analysis System (RIAS) is a validated tool for evaluating clinical interactions using qualitative coding and quantitative analysis.²⁰ The RIAS has previously been applied in English-concordant oncology encounters^21-26 and LEP language-concordant and interpreted primary care visits,²⁶ but it has not been used to evaluate communication in LEP patient-clinician interactions in cancer care. Our goal was to generate preliminary hypotheses regarding differences in quality of communication between language-concordant and language-discordant patient-clinician pairings in cancer care.

METHODS

Sample and Procedure

The study evaluated the interaction between clinicians and patients presenting for an initial oncology visit at four oncology clinics in New York City, three in a large public health care system, and one National Cancer Institute–designated Comprehensive Cancer Center. Our study sample comprises 50 participants: 34 patients and 16 clinicians. Eligible patients were Spanish- or English-speaking adults being seen for their first visit with an oncology clinician. Bilingual research staff approached the patients, explained the study, obtained informed consent, and administered an intake survey in the patients' preferred language, which included demographic and health-related questions, and asked patients to identify their primary oncology clinician. Clinicians self-assessed their Spanish language proficiency levels using the validated Interagency Language Roundtable scale for health care settings.²⁷ We also asked clinicians questions about their demographics, work environments, and non-English language communication experiences.

Patient-clinician pairings were categorized into four different dyads on the basis of the level of language concordance in their interactions, per clinicians' answers to the Interagency Language Roundtable survey. Dyad categories included English language concordant (n = 12, clinician and patient fully communicated in English); professionally interpreted (n = 11, clinicians using professional interpreters); partially language concordant (n = 4, partially bilingual clinicians who communicated partly in Spanish and/or used ad hoc interpreters); and Spanish language concordant (n = 7, clinician and patient fully communicated in Spanish). The dyad categorization for the partially language-concordant and professionally interpreted groups was finalized upon analysis of the clinician's behavior in the encounter (ie, whether the clinician used a professional interpreter, ad hoc interpreters, and/or partial Spanish communication). This study was approved by the Memorial Sloan Kettering Cancer Center Institutional Review Board.

Measures

Communication appraisal through RIAS.

An experienced, bilingual (Spanish and English) RIAS coder coded all study visit audio recordings. The visit recordings were analyzed using established RIAS codes and speaker-specific code composites developed for use with interpreters in a previous primary care study.²⁷

The RIAS data are reported as the mean number of statements conveying a complete thought (ie, word(s), sentence(s), or a clause of a compound sentence). Each statement is assigned to one of 37 mutually exclusive codes with a speaker designation (eg, patient, clinician, or interpreter). Following conventions in RIAS analysis and used in the primary care study,²⁷ individual codes were combined into six composites: medical information, medical questions, psychosocial/lifestyle information, psychosocial/lifestyle questions, emotional statements, and partnership/facilitation statements. Medical information included codes pertaining to biomedical counseling statements and therapeutic regimen information. Medical questions integrated open and closed-ended questions about medical conditions and treatments. The psychosocial/lifestyle information and questions composite code comprise counseling and open and closed-ended questions related to psychosocial topics, respectively. The emotional domain includes statements of concern, reassurance, empathy, legitimation, partnering, and self-disclosure. Finally, the partnership/facilitation composite code incorporated statements asking for understanding or reassurance, cues of interest, checking for understanding, asking for opinion, and asking for permission to proceed. Additional coded domains included positive talk (laughing, direct approval, compliments, showing agreement, or understanding) and negative talk (disagreement or general or direct criticism).

Analyses.

We summarized descriptive statistics using frequency and percent for patient and clinician demographics, as well as clinician survey responses. The RIAS data are reported as the mean number of statements conveyed by each speaker (ie, clinician or patient). For encounters involving interpreters, we report the estimated mean difference in the number of composite statements expressed directly by the clinician or patient versus the interpreter. Because of the small sizes of our subgroups, we did not have sufficient power to detect statistical significance between the groups.

RESULTS

Participants

Patients.

A total of 34 patients participated in the study. As displayed in Table 1, the cohort included English- (n = 12) and Spanish-speaking (n = 22) patients, most of whom were female (n = 27, 79.4%) and had a breast cancer diagnosis (n = 23, 67.6%); the mean age was 56 years (range 26-80). All Spanish-speaking patients identified as Hispanic individuals.

TABLE 1.

Patient Demographic Information

Clinicians.

Table 2 describes the clinicians' demographics and non-English language communication practices for the 16 participating clinicians. A majority of clinicians were medical oncologists (n = 11, 68.8%), and most (n = 15, 93.8%) reported spending over 70% of their time working in the clinical setting. English was the first language for 10 clinicians (62.5%). Of the clinicians who reported communicating with patients in a non-English language, 75% (n = 9) reported seeing LEP patients who required interaction in a non-English language at least two to three times per week. Half of clinicians who report using non-English languages with patients rated their ability to communicate clearly and accurately with patients in that language as excellent (n = 3) or very good (n = 3). Furthermore, 16.5% (n = 2) reported always using interpreters in clinical encounters with LEP patients.

TABLE 2.

Clinician Information

Quality of Communication

General communication-specific outcomes—RIAS codes.

The average duration of all encounters was 35 minutes (range 11-88 minutes). The average visit time by dyad was 36 minutes for English language-concordant (range 11-69 minutes), 38 minutes for professionally interpreted (range 13-88 minutes), 24 minutes for partially language-concordant (range 15-42 minutes), and 32 minutes for Spanish language-concordant (range 19-48 minutes). Table 3 displays the mean number of coded statements for clinician and patient speakers. Spanish language-concordant clinicians had almost two-fold greater mean number of statements about biomedical information than English language-concordant clinicians. Language-concordant clinicians (both English and Spanish) had higher mean number of statements related to psychosocial/lifestyle information than the partially language-concordant or professionally interpreted groups. Spanish language-concordant clinicians had a greater mean number of statements in the emotional talk category. On the other hand, patients across all dyads engaged in a similar frequency of emotional talk, but Spanish language-concordant patients had the highest mean number of statements of positive talk. Language-concordant dyads showed similar rates of partnership/facilitation statements. These were five-fold and 1.5-fold lower in the professionally interpreted and partially language-concordant categories, respectively.

TABLE 3.

Mean Number of Statements for Various Communication-Specific Roter Interaction Analysis System Codes

Involving interpreters.

For professionally interpreted and partial language-concordant encounters, we evaluated the fidelity of interpretation with respect to differences in the mean number of statements made by clinicians and conveyed by the interpreter and the mean number of statements made by patients and conveyed by the interpreter (Appendix Table A1, online only). Overall, professional interpreters effectively communicated clinician speech related to medical and psychosocial information, medical and psychosocial questions, and emotional statements. By contrast, professional interpreters conveyed less of the patients' speech related to medical and psychosocial information: Professional interpreters conveyed 105% of clinician statements (signaling a potential addition of more statements in their interpretation to comprehensively convey information in Spanish) and 65.2% of patient statements related to medical information, 100% of clinician statements and 61.6% of patient statements related to psychosocial information. On the other hand, ad hoc interpreters consistently had lower mean numbers of conveyed statements for either speaker (the mean conveyed clinician statements was 25.1% [0%-48.6%]; the mean conveyed patient statements was 22.0% [0%-66.4%]). Both professional and ad hoc interpreters communicated fewer partnership/facilitation statements when delivering information to and from both clinicians and patients (professional interpreters conveyed 55.7% of clinician statements and 0% of patient statements versus ad hoc interpreters who communicated 6.0% of clinician statements and 0% of patient statements).

DISCUSSION

Our study demonstrates that the RIAS can be used to assess the quality of communication between oncology patients and clinicians in both language-concordant and -discordant encounters. Our data reveal that clinicians who can provide care in Spanish to Spanish-speaking patients deliver a high number of biomedical statements, which could indicate that they may deliver more comprehensive biomedical counseling. We found that ad hoc interpreters do not effectively convey speech in this setting, and their use is consequently inadequate and potentially dangerous in health-related conversations, as shown in other studies.^28-30 Furthermore, our data suggest that although professional interpreters are effective mediators of communication in the context of language discordance, they interpreted fewer partnership/facilitation statements made by clinicians and patients in initial oncology visits.

Although we were not powered to detect statistically significant differences between dyads because of our small sample size, we found that both Spanish and English language-concordant clinicians engaged in more dialog related to psychosocial/lifestyle information compared with the partially language-concordant and professionally interpreted dyads. This finding is consistent with a study by Butow et al,³¹ which reported that interpreter involvement in the oncology setting leads to less conversation time focused on psychosocial issues.

Several studies have demonstrated the impact of social determinants of health in cancer stage at diagnosis and patient outcomes.^32-34 Lack of education, poverty, immigration status, lack of social support, and low socioeconomic status have been among the psychosocial issues and social determinants associated with poor health outcomes in cancer care.^35,36 As noted on Healthy People 2030, language is an important factor related to access to and quality of care.³⁷ Thus, this poses potential implications for oncology patients with LEP who lack a language-concordant clinician and may experience disparities due to communication barriers.

We also noted important differences between the language-concordant pairings. First, Spanish language-concordant clinicians spent more time providing biomedical information to patients. Although the reason for this pattern is unclear, it could potentially be associated with LEP patients requiring more medical counseling because of a potential deficit of medical knowledge stemming from a history of insufficient explanations in the patient's preferred language regarding therapeutic regimens, comorbidity management, and other biomedical information. Additionally, although patients across all dyads engaged in emotional talk with similar frequency, Spanish language-concordant clinicians had a larger number of emotional statements compared with clinicians in other groups. Interestingly, Spanish language-concordant patients also seemed to have a higher tendency to engage in positive talk, such as expressing agreement, compliments, or laughing. A possible explanation for this pattern could be that a clinician-led establishment of emotional rapport leads to enhanced patient openness, optimism, and satisfaction with care, as has been shown by others.^38-40 Clinician empathy is also associated with increased patient compliance, quality of life, and lower emotional distress.^40-44 Additionally, a recent randomized trial demonstrated improvements in several domains of patient satisfaction for LEP patients receiving care from a language-concordant radiation oncology physician.⁴⁵ Studies exploring the impact of language concordance in the care of patients who speak languages other than Spanish have reported similar findings, citing higher satisfaction with care, ratings of interpersonal care, comprehension of medical situations, and understanding of their diagnoses.^46-49 Thus, this finding has important implications in supporting the role of language concordance in establishing a more effective doctor-patient relationship and potentially improving outcomes for LEP patients.

Our study also explored aspects of communication involving interpretation. Consistent with the findings of Fagan et al, visits with professional interpreters in our study did not increase the duration of the medical encounter.^50,51 However, interpretation-associated time pressures because of schedule limitations during busy clinical practice may contribute to challenges in establishing rapport between patients and clinicians. Although we found professional interpreters to be effective communicators of verbal information, the lower number of statements in emotional talk and partnering activation in these encounters suggest that they may interfere with the facilitation of partnership between patients and clinicians. This finding is not exclusive to Spanish communication; a similar assessment was reported in a cross-sectional study focused on evaluating communication and quality of care for Asian-American patients with LEP.⁵² Several studies have shown that interpreter mediation undermines the ability of clinicians and patients to identify emotional cues, offer empathetic responses, and successfully establish partnership.^53-55 Although interactions within the partially language-concordant group sometimes involved the mediation of a third-party, ad hoc interpreter, these clinicians also communicated directly using their partial Spanish skills at times. This nuance could potentially explain why their ability to engage in emotional talk and partnering activation was greater than clinicians in the professionally interpreted dyad. Importantly, our results reinforce the inadequacy of using untrained interpreters because of their inability to convey information accurately and thoroughly. This is especially critical in oncology care discussions where family-mediated interpretation has been shown to complicate ethical clinician communication by omitting or altering information regarding end-of-life care.^5,14,56 These data highlight the need for more comprehensive policies that promote trained professional interpreters' involvement and active participation in oncology practices when language-concordant clinicians are unavailable, along with adequate scheduled time for professionally interpreted visits and clinician training on working with interpreters.

Our study has limitations. First, participants' awareness of being recorded during clinical interactions might have affected their behavior, which we attempted to address by having nonobtrusive audio equipment. Additionally, because of our small sample size and unequal number of pairings within each dyad, our results could have been influenced by specific interactions or clinicians. We intended to minimize these behavioral effects by focusing on first-time oncology visits, where the content of discussions might be more extensive and focused on treatment discussions. Our study sample was also limited to one geographic location (New York City) in an urban setting. Although this may make our results less generalizable outside of an urban setting, we believe that our inclusion of four study sites, each with different community demographics, aids in the generalizability of our data. Additionally, we lacked data on the training level and/or certification status of the interpreters involved in the professionally interpreted encounters. Similarly, we do not know if the clinicians in this study received any training in working with professional interpreters in the past. It is plausible that if the interpreter is certified and the clinician is trained to effectively work with interpreters, that communication could be improved.

Our study contributes to the growing body of literature supporting the need for linguistically appropriate care and standards to guide clinician use of non-English language skills and interpreter services in oncology settings. Addressing the limited diversity of the oncology workforce is crucial for the development of patient-centered health services that are sensitive to the linguistic needs of the populations we serve. Diversity and inclusion efforts should also extend to the research workforce as LEP patients are less likely to be recruited into research studies compared with English-speaking patients.⁵⁷ Solutions include employing research staff who are reflective of the community (eg, bilingual and bicultural staff), making forms and materials available in the target group's language(s), and incorporating multimedia tools in the informed consent process.^58-60 Other strategies that oncology clinicians can implement to improve cancer care for patients who communicate in non-English languages include integrating certified professional interpreters within their practices, accounting for additional necessary time for interpretation, facilitating patients' access to multilingual health education materials, seeking educational opportunities on how to work effectively with interpreters and accurately self-assess their non-English language skills, and participating in language education/certification programs. Future research should investigate the hypotheses generated in this study and explore the impact of language and cultural concordance on cancer-specific health outcomes, including patient satisfaction, treatment completion, and mortality.

APPENDIX

TABLE A1.

Differences in Categories of Clinician and Patient Expressed and Interpreter Conveyed Statements

AUTHOR CONTRIBUTIONS

Conception and design: Debra Roter, Lisa C. Diamond

Collection and assembly of data: Jackie Finik, Lisa C. Diamond

Data analysis and interpretation: All authors

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

Exploring the Impact of Language Concordance on Cancer Communication

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated unless otherwise noted. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/op/authors/author-center.

Open Payments is a public database containing information reported by companies about payments made to US-licensed physicians (Open Payments).