Study sample

Multi-specialty doctors who care for seriously ill patients in two large training hospitals in California (Stanford Hospital and Clinics and the VA Palo Alto) participated at the end of the clinical training year just before graduation in the academic years of 2010 and 2011 (data were collected from calendar years 2010 through 2012). We specifically did purposive sampling of trainees from various sub-specialties just before graduation as these doctors have very high practice volumes and care for numerous seriously ill persons from various ethnic backgrounds. Of the 1234 eligible participants, 1040 participated (84.3% response rate).

Data collection

The questionnaire was administered one time online and no personal health identifiers were collected in an effort to promote participant confidentiality and honest responses without concerns about individual scrutiny. There was no repeated contact of participants. The Stanford Institutional Review Board approval was obtained to analyze the data presented in this paper.

Data analysis

Qualitative data analyses of development cohort to identify key codes.

The data of the participants who reported any barriers to effective EOL conversations was sorted based on age group (20–29 years, 30–39 years. and 40 years or older). 29 participants were 40 to 49 years old, and this cohort was designated to be the development cohort (see Fig 1) for purposes of qualitative analyses and code development. The open ended responses of the 29 development cohort participants were transcribed, coded and analyzed using Microsoft Access and NVivo 7 (QSR International Pty Ltd., Melbourne Australia). Grounded theory techniques[19–20] of intense open and axial coding were used to identify key barriers to conducting effective EOL conversations with ethnic minority patients. First, using an open coding approach, one of the authors (VSP) and a trained research assistant independently coded the transcripts of the development cohort. During open coding, recurrent barriers were identified and characterized. During axial coding, the barrier categories were further defined, and the relationships between them were characterized. Next, the coders compared their assignment of codes and a third investigator (HCK) mediated any discrepancies between codes assigned until agreement was reached. Efforts to maintain the validity of the qualitative data included the following: to evaluate inter-coder reliability, the exact assignment of codes was assessed for 100% of the development cohort transcripts; agreement was found to be greater than 90%, which was deemed to be comparable with previous work. After 19 transcripts had been coded, no new codes emerged from the next five transcripts (data saturation had been reached). To confirm data saturation, the last five transcripts in the development cohort were coded, with no new code labels emerging. At the end of this process, 6 primary barriers were identified and the development cohort was discarded and not used in subsequent analyses. Actual words and phrases that participants used were used to name the barriers as feasible (see Table 1 for barriers with exemplars).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. STUDY FLOW: The study subjects were divided into code development and validation cohorts.

After identifying the 6 primary barriers using the development cohort, these participants were not used in subsequent analyses. The flow of study participants is shown in the figure.

https://doi.org/10.1371/journal.pone.0122321.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Top six barriers (with exemplars) to effective end-of-life conversations with ethnic minorities according to multi-ethnic, multi-specialty doctors.

https://doi.org/10.1371/journal.pone.0122321.t001

Code validation.

The validation cohort consisted of 996 participants age range of from 20–39 years (see Table 2 for demographics). Of the 996 transcripts, 50 were drawn randomly and coded independently by the coders (VSP and RS) for the presence or absence of each of the six primary barriers identified, as well for any new codes. The two coders then met and compared codes for each transcript and determined that no new recurrent codes emerged and that the six primary barriers identified earlier were sufficient to appropriately code the entire validation dataset. (See Fig 1 for study flow.)

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Demographic characteristics showing validation cohort participant gender, age, race, ethnicity and subspecialty (n = 996).

https://doi.org/10.1371/journal.pone.0122321.t002

Barrier 1: Language and medical interpretation issues

Medical jargon is difficult to translate into other languages (as equivalent words may not exist) and approximate translations do not convey the true meaning and may lead to misunderstandings and miscommunication. Doctors felt that the technical medical terms used in EOL discussions (cardio-pulmonary resuscitation, ventilators) or vague phrases like "heroic measures" made it very challenging to have these discussions with persons of Limited English Proficiency (LEP). The doctors in our study identified medical interpreters (MI) services as a barrier to effective EOL conversations for the following reasons: (i) logistical difficulties encountered with in-person/phone/video MIs, (ii) conversations involving MIs were time intensive, (iii) using MIs as communication intermediaries psychologically distanced the doctors from their patients, (iv) the variability in quality and accuracy of the MIs, (v) sometimes MIs interfered with the diplomacy of the doctor-patient encounter by giving advice to the patient or by assuming the role of a patient advocate.

As the US population is becoming more diverse, it is imperative that we train doctors and medical interpreters to work together respectfully and effectively following standard guidelines[27–28]. Furthermore, system triggers need to be built to automatically identify patients with LEP and schedule adequate clinical encounter time for doctors to work with MIs and these patients. Additionally system changes need to be made so that all doctors will be able to use special billing codes like the interactive complexity add-on Current Procedural Terminology[29] codes when working with MIs and be adequately reimbursed for these time-intensive and complex encounters.

Barrier 2: Patient/ family religious and spiritual beliefs about death and dying

Doctors reported that religious and spiritual beliefs greatly influenced how ethnic patients perceived EOL issues ranging from (i) an unwillingness to discuss EOL issues or plan ahead due to their beliefs that the timing and nature of a person's death should be determined “by the will of God” and humans should not tamper with the process (ii) religious taboos about withholding/withdrawing high intensity interventions at the EOL (iii) as families were praying for miracles and thus refused to engage in EOL discussions and planning even in the face of impending death; and (iv) patient/family beliefs that the specific circumstances around the death impacts the patient’s afterlife and they were reluctant to participate in actions that may in some way alter the timing and course of death. Doctors need to be trained to work effectively with patients and families from diverse religio-spiritual backgrounds and to consult and partner with chaplains and community spiritual leaders in providing necessary support to seriously ill patients and families to facilitate quality EOL decisions.

Barrier 3: Doctors’ ignorance of patients’ cultural beliefs, values and practices

The doctors felt that their ignorance about the diverse cultural values around death and dying was the third biggest barrier to effective EOL conversations with seriously ill patients and their families from diverse cultures. This lack of knowledge led to the doctors' (i) inability to empathize with the ethnic patients' cultural values that influence EOL decisions, and (ii) committing cultural faux pas by discussing taboo topics which inadvertently offended the patient/family and undermined the therapeutic relationship. Doctors need to be trained[15–17] on how to effectively conduct EOL discussions in a culturally sensitive and competent manner with seriously ill persons from various cultural and ethnic backgrounds. Such training could be incorporated as a longitudinal curricular thread across the continuum of medical education ranging from undergraduate to postgraduate training and as an essential component of continuing medical education.

Barrier 4: Cultural differences in truth handling and decision making

Doctors stated that in some cultures, the patient/family believed that even speaking about death and dying would invoke death sooner and therefore refused to engage in EOL discussions. In others, the family wanted the diagnostic and prognostic information withheld from the patients due to a belief that (the patient) knowing the truth about the illness would cause the patient to lose hope. Sometimes, the patients recused themselves from decision making, especially common in women in certain cultures. Instead, they relegated decision-making to their family who may or may not use the principles of substituted judgment or, worse still, may act to promote their own personal interests over those of the patient's. This was particularly challenging where large extended families were involved, the hierarchy and process of decision-making was unclear or group decision-making was preferred. These situations typically triggered multiple time-intensive family meetings about EOL issues, a source of great stress for doctors. Faced with these complex situations, doctors who may not have the time and/or training to engage patients and families in a constructive manner may procrastinate or even avoid vital EOL conversations with their seriously ill patients.

Barrier 5: Patient/family's limited health literacy

National data[30] show that limited health literacy disproportionately affects ethnic minorities, especially the poor, the uninsured and older adults. Research also shows that patients forget 40–80%[31] of medical information provided to them and half of what they do remember is incorrect[32]. Patients with limited health literacy may lack the ability to comprehend complex medical concepts like resuscitation preferences. Alternatively, they may have unrealistic expectations that medical technology is capable of curing any type of terminal illness or at least is effective in keeping the patient alive indefinitely with technological support.

Furthermore, it is very important to remember that even commonly used words can mean very different things to different people. For example, oncologists often use the word “cure” to indicate five years of cancer-free survival. However, to many cancer patients and families, the word “cure” means eradication of cancer and restoration of normal health. Another example is the English word 'hospice" which sounds like "hospiscio" in Spanish, meaning poorhouse. It is possible that when a doctor raises the option of referral to hospice care, an impoverished Spanish-speaking patient may misunderstand that the doctor is withholding expensive interventions and referring them to the poorhouse and thus be deeply offended. Such perceptions, whether merited or not, can create a deep mistrust in patients who may not vocalize these concerns but may resist end-of-life conversations with their doctors. The health literacy level of each patient needs to be assessed and clearly documented in the electronic medical records. Doctors need to be trained to avoid medical jargon when communicating with patients and use techniques like “teach back” [33] to assess for the patient’s understanding of information imparted.

Barrier 6: Patient/family’s mistrust of doctors and the health care system

Because of historical events that have occurred between the medical establishment and certain ethnic communities, patients may not believe that doctors have their best interests at heart [34–36] and may fear abandonment or being subjected to poor quality care by self-interested medical professionals. Patient’s family may also believe that high-intensity interventions are being withheld or withdrawn from the seriously ill patient because of racism. Thus, any notions of withholding high-intensity medical interventions and instituting hospice care may be misinterpreted as deliberate provision of sub-standard care, leading the families to be deeply suspicious of EOL discussions. In situations where there is conflict about EOL decision-making, doctors (fearing litigation, family mistrust) automatically provide high-intensity interventions that may be ineffective and burdensome to terminally ill patients. Honest and open communication in simple language and shared decision making[37–38] will help foster trusting therapeutic alliances and thereby better quality EOL decision making. Additionally, public engagement campaigns are needed to educate Americans that high-intensity treatments on terminally ill patients do not necessarily result in prolonging life with quality but may only prolong the dying process while increasing the patient’s suffering.

To our knowledge, ours is the first and largest study of its kind using an innovative mixed methods approach to characterizing doctor-reported barriers to effective EOL conversations with diverse Americans. We were able to qualitatively identify the 6 main themes using a development cohort, empirically validate these themes by analyses of 50 randomly identified transcripts from the validation cohort and then finally use the results of the qualitative analysis to quantitatively analyze the validation cohort data. We believe that through this novel approach, we have demonstrated a strategy to employ rigorous quantitative research methods to assess the magnitude and frequency of constructs and rigorous qualitative research to explore the meaning and understanding of the constructs[39]. Furthermore we have merged quantitative and qualitative data to develop a more complete understanding of a complex and important problem[39]—in this case the barriers doctors face in conducting effective EOL conversations.

Our study is limited by the fact that the study participants were doctors from two hospitals in one geographic area. However, our institutions recruit and employ doctors from throughout the US and the ethnic and gender background of our doctors is reflective of the recent national trends of increasing women and diverse doctors reported by the Accreditation Council for Graduate Medical Education in 2013 (White 65.1%, Asian 21.2%, Black 6.3% and Hispanic American 6.3%). In fact, our diversity in doctors and the patients we serve was the main reason that made this study possible. Also, our study, is in part, a qualitative study and thus methodologically challenging to conduct nationally. However, it is to be noted that we have followed the COREQ[40] guidelines in conducting our study and in reporting our qualitative findings.