Table 1.

Summary of Empirical Studies on the Effects of Language Proficiency and Interpreter Use on Psychiatric Assessment and Diagnosis, Treatment, and Patient-Provider Interaction

Author (Year) Location Setting	N, Sample	Interpreter Type	Method	Results	Comments
Marcos et al. (1973a) New York Inpatient (28)	10 Spanish-speaking patients with schizophrenia who were conversant in English	None	Closed circuit television recordings of standardized evaluations in Spanish and English rated by Spanish or English-speaking psychiatrists	Patients received higher psychopathology ratings in English than Spanish (BPRS scores of 93 versus 65, respectively). BPRS scales affected most by language were “depressed mood”, “anxiety”, “tension”, “hostility”, “emotional withdrawal”, and “somatic concern” with patients scoring higher in English than Spanish.	Sample is the same as in Marcos et al., 1973b. Interview language and rating psychiatrists’ language were confounded.
Marcos et al. (1973b) New York Inpatient (29)	10 Spanish- speaking patients with schizophrenia who were conversant in English	None	Closed circuit television recordings of standardized evaluations in Spanish and English rated by Spanish or English-speaking psychiatrists	Patients gave short replies (6 words or less) significantly more often and spoke slower in English compared to Spanish. Their responses contained more speech disturbances (such as grammatical problems, repetition, stutter, etc.) Patients gave different responses to the same question in English and Spanish interviews. Patients spoke in the present tense more often in English than Spanish. Patients appeared more uncooperative and guarded when evaluated in English compared to Spanish.	Sample is the same as in Marcos et al., 1973a
Price (1975) Australia Setting not specified (34)	Unspecified number of Hindustani-speaking patients evaluated by 1 of 3 psychiatrists	Ad hoc - 2 hospital orderlies with differing levels of experience and 1 educated man with schizophrenia in remission	Audiotapes of evaluations	Errors in interpretations of patients’ utterances were twice as common as errors in interpretations of psychiatrists’ utterances Omissions were likely to occur when patients provided long or rambling answers. More interpreting errors were made in utterances by acutely psychotic patients compared to patients without active thought disorders. The hospital orderlies had lower English proficiency and made more errors in interpreting than the lay interpreter. The experienced hospital orderly made addition errors that appeared related to his prior experience with history-taking, whereas the less experienced orderly distorted questions into leading questions.	2 psychiatrists were bilingual.
Marcos (1979) New York Setting not specified (33)	8 Chinese- speaking or Spanish- speaking patients	Ad hoc – psychiatric nurse, nurse’s aid, patients’ relatives	Audiotaped encounters	Errors in interpretation were related to inadequate language proficiency of interpreters, interpreters’ attempts to”normalize” disordered thought process, minimization or amplification of pathology by relatives, or interpreters responding without asking the patient.
Kline Los Angeles Outpatient (41)	a. 61 Spanish- speaking patients (21 with limited English proficiency and 40 with some English proficiency) b. 16 psychiatric residents	No interpreter or professional interpreters (training not specified)	Questionnaires assessing effectiveness of communication at psychotherapy intake appointments from patient and provider perspectives	a. Compared to patients seen without interpreters, patients seen with interpreters felt more helped (76% versus 40%) and gained more self-understanding (90% versus 53%). Most patients seen with interpreters wanted return visits (76%). b. Few psychiatric residents believed that patients seen with interpreters were helped (0%) or wanted to return to the clinic (31%). One resident (6%) felt comfortable providing ongoing care to patients who need interpreters.	Patients were selected by surname. Patients’ English proficiency was not assessed. Use of an interpreter was based on patient request, which may have contributed to favorable ratings of interpreter-assisted evaluations.
Price & Cuellar (1981) San Antonio Inpatient (30)	32 bilingual Mexican- American patients with schizophrenia (31/32 patients were native Spanish- speakers)	None	Videotaped standardized psychiatric evaluation conducted in both English and Spanish rated by bilingual Masters-level mental health professionals	Patients were judged to have greater psychopathology on the BPRS during Spanish-language interviews than English- language interviews. Patients had greater self-disclosure during interviews in Spanish. Verbal fluency (in English) predicted the difference in detected psychopathology between languages.
Dodd (1984) Saudi Arabia Primary Care (38)	Unspecified number of patients treated by 10 Arabic- speaking physicians or 10 non-Arabic- speaking physicians	Professional and ad hoc (bilingual nurses) interpreters	Retrospective medical record review determining the number of new diagnoses of ICD mental disorders and “signs, symptoms and ill-defined conditions”	Arabic-speaking and non-Arabic speaking physicians diagnosed similar proportions of patients with mental disorders and similar proportions of patients with signs, symptoms and ill-defined conditions	Patients’ language proficiency not specified.
Farooq et al. (1997) United Kingdom Setting Not Specified (35)	20 patients (10 English- proficient, 7 Mirpuri- speaking and 3 Punjabi- speaking patients with limited English proficiency)	Professional interpreter	Audiotaped encounters of patients interviewed via checklists by bilingual psychiatrist and English-speaking psychiatrist via interpreter	Common interpreting errors included omission, conceptual substitutions, condensations, miscommunication due to interpreter’s lack of language skills and subtle changes in phrasing. No significant differences between the psychiatrists on scoring of mental status items.	Interpreter was not blinded to study aims. Bilingual psychiatrist conducted interviews and coded transcripts.
Haasen et al. (2000) Germany Inpatient (32)	150 patients with psychotic symptoms (100 Turkish patients with good or bad German proficiency and 50 German patients)	None	Comparison of clinical diagnosis with diagnoses from structured interviews conducted by 2 psychiatric residents (1 bilingual in Turkish and German; 1 monolingual German-speaking)	Low agreement between clinical and both research diagnoses for Turkish but not German patients and lower agreement between research diagnoses for Turkish than German patients. Diagnostic disagreement occurred in 4% of German patients and 19% of Turkish patients. Turkish patients’ level of proficiency in German did not affect rate of diagnostic disagreement.	Method of determining patients’ German proficiency was not described. Method of determining clinical diagnosis (including use of interpreters) was not specified. Authors did not specify whether the standardized instrument has been validated in German and Turkish.
Drennan and Swartz (2002) South Africa Inpatient (31)	Unspecified number of Xhosa- speaking patients	Professional, ad hoc (e.g., bilingual nurses and staff), and no interpreter	Mixed qualitative study using retrospective chart review, semi-structured interviews with interpreters and staff and notes from direct observation of encounters	When language abilities not accounted for, clinicians were likely to conclude that intellectual impairment or thought disorders (e.g., impoverished thought) were present. Interpreters offered opinions on patients’ intelligence and motivation. Reasons patients refused interpreters included pride in their English-speaking abilities, paranoia about the interpreter, and wanting to eliminate third-party interference.
Eytan et al. (2002) Switzerland Primary Care (36)	319 asylum- seekers entering Switzerland from Kosovo	Professional interpreters, ad hoc interpreters (family and friends), and no interpreter (when patients were bilingual or when an interpreter was needed but not available)	Medical record review of standard health screening questionnaires administered to asylum- seekers by nurses	Nurses rated more communication as “poor” or “fair” in 84% of evaluations without interpreters and 72% of evaluations with ad hoc interpreters, but “good” in 94% of evaluations with professional interpreters. Use of professional interpreters was associated with increased disclosure of traumatic events (77% of patients seen with professional interpreters, 46% with ad hoc interpreters, 55% without interpreters) and psychological symptoms (33% of patients seen with professional interpreters, 14% with ad hoc interpreters, 12% without interpreters) but not physical symptoms (25% of patients seen with professional interpreters, 22% with ad hoc interpreters, 14% without interpreters) Use of professional interpreters was associated with greater referral for psychiatric care (15% of patients seen with professional interpreters, 3% with ad hoc interpreters, 4% without interpreters). When age, gender, severity of symptoms and prior trauma exposure were controlled in multivariate analyses, the effects of professional interpreter use on referral disappeared.	Because professional interpreter use was associated with greater disclosure of trauma and of symptoms, this may be an important pathway by which referrals for psychiatric care came about. It is unclear why the authors chose to control for these variables rather than assess whether these variables mediated the effects of professional interpreters on referral rates.
Bischoff et al. (2003) Switzerland Primary Care (37)	723 asylum- seekers entering Switzerland, primarily from Balkan states	Professional interpreters, and ad hoc interpreters (family and friends), and no interpreter (when patients and providers shared a common language or when an interpreter was needed but not available)	Medical record review of standard health screening questionnaires administered to asylum- seekers by nurses	Lack of any interpreter was associated with low reporting of physical and psychological symptoms (18% and 18%, respectively). Use of professional interpreters led to higher reports of both types of symptoms (25% and 32%, respectively). Use of ad hoc interpreters led to higher reports of physical symptoms (26%) and low reports of psychological symptoms (16%). When nurses believed communication was good (versus poor), patients reported more physical symptoms (OR 2.1, 95%CI 1.2–3.6), psychological symptoms (OR 3.7, 95% CI 1.5–4.8), and exposure to trauma (OR 4.7, 95% CI 3.0–7.5). Referral for psychological care but not medical care was more likely when language concordance was adequate (bilingual provider or professional interpreter) compared to no concordance (OR 3.2, 95% CI 1.2–8.6). Partial concordance (ad hoc interpreter or nurse with some language skills) did not improve referral rate compared to no concordance (OR 1.6, 95% CI 0.5–4.9).	Sample includes participants in the study by Eytan and colleagues (2002).
Zayas et al. (2007) USA Outpatient (40)	98 Latino patients (with good or limited English proficiency or bilingual) evaluated by Latino (n = 16) or non-Latino (n = 33) clinicians of varying backgrounds	Ad hoc - 2 bilingual mental health professionals, one untrained and one with some training in interpreting. Interpretation was in-person for English-speaking clinicians (n = 6) or interpretation of videotapes for interviews conducted by Spanish-speaking clinicians (n = 65).	Comparison of encounters with interpreters (n = 71) to those without (n = 27). Questionnaire of clinicians’ assessment of the interpreting process	Clinicians believed they were getting accurate translations and reported that use of the interpreter improved their confidence in diagnosis. Clinicians believed that the presence of the interpreter led to assessments of equal or greater severity of psychopathology and impairment of functioning.	In the majority of encounters assessing the effects of interpreters (65/71), the clinician was Spanish-speaking and the interpreter was not present during the encounter, thus differing from encounters between English-speaking providers and patients with limited English proficiency seen with in-person interpreters. The assessments of Spanish- speaking clinicians who conducted the interviews were not compared to the impressions of the English-speaking clinicians reviewing videotapes. No direct assessment of the accuracy of interpretation was conducted to validate clinicians’ impressions.
Gilmer et al. (2009) San Diego Outpatient (39)	7784 patients with schizophrenia (5695 English- proficient non- Latino whites, 1196 English- proficient Latinos, 523 Latinos with limited English proficiency, 298 English- proficient Asians, and162 Asians with limited English proficiency)	Unknown	Review of administrative data including antipsychotic prescription renewal rate and hospitalization rate	Among Latinos, limited English proficiency was associated with greater medication adherence (41% versus 36% of English-proficient Latinos) and less excess prescription filling (15% versus 20% of English-proficient Latinos). Among Asians, limited English proficiency was associated with lower medication adherence (40% versus 45% of English-proficient Asians) and excess prescription filling (13% versus 17% of English-proficient Asians) and greater nonadherence (29% versus 22% of English-proficient Asians) Controlling for adherence, limited English proficiency was associated with lower hospitalization rates.	Language of care was used as a proxy for English language proficiency.