Table 1.
The Impact of Professional Interpreters on Clinical Care (N = 21)
| Author (Year), Country | N | Comparison Groups | Interpreter Type (Professional Only/ Professional versus Ad Hoc) Professionals Trained? (Yes/No/ Unclear) | Control for Confounders (Yes/No) or Qualitative Methods | Outcome Related to Interpreters | Results Related to Interpreters (Statistical Analysis/Test) |
|---|---|---|---|---|---|---|
| Comprehension & Errors | ||||||
| Prince and Nelson (1995), United States | 34 | LEP w/professional interpreter versus LEP w/partial language concordance and ad hoc interpreter | In-person professional (Trained: unclear) versus ad hoc—family & friends | Qualitative—audio recordings of directly observed clinical encounters | Type and frequency of errors after ED residents attended 45-hour Spanish course | Major errors occurred during six visits, five of these were with ad hoc interpreters. Professional interpreters present at 46% of visits; no major errors during these visits. (Descriptive statistics only) |
| Hornberger et al. (1996), United States | 49 | Pediatric visits randomized to type of interpreter | In-person (Trained: unclear) versus Remote Simultaneous Medical Interpreter (Trained: yes) | No (Randomized-controlled trial) | Number of utterances and errors by modality of interpretation | 10% more physician and 28% more maternal utterances (p < 0.05), and 13% fewer errors with Remote Simultaneous Medical Interpreter (p > 0.05) (t-test) |
| Farooq, Fear, and Oyebode (1997), United Kingdom | 20 | Interview of LEP patients by language concordant* psychiatrists versus interview of the same patients via interpreter | In-person single interpreter (Trained: unclear) | No | Mental Status Exam score Family history items | No significant differences for Mental Status Exam or family history items. (95% CI for mean differences in MSE all cross 1; percent agreement for FH items 90–100%) |
| Chan and Woodruff (1999), Australia | 13 | LEP interpreted versus LEP noninterpreted | Telephonic professional (Trained: unclear) | No | Comprehension of terminal diagnosis | Complete comprehension of diagnosis more likely when interpreter used (67% versus 30%), but not significant (p = 0.5; χ2) |
| Flores et al. (2003), United States | 13 | Professional versus ad hoc interpreters | In-person professional (Trained: unclear) versus ad hoc—family & staff | Qualitative—audio recordings of directly observed clinical encounters | Clinical significance of interpretation errors | Professional interpreters made fewer clinically significant errors than adhoc interpreters (53% versus 77%; p < 0.0001; χ2) |
| Utilization | ||||||
| Enguidanos and Rosen (1997), United States | 48 | LEP interpreted versus English speakers | In-person professional (Trained: unclear) | No | Adherence to follow-up from ED | Equal adherence rates among LEP and English speakers (71% versus 63%; p = 0.76; χ2) |
| Tocher and Larson (1998)†, United States | 622 | LEP interpreted versus English speakers | In-person professional (Trained: unclear) | Yes | HbA1C frequency Nutrition referral 12-month ED visit 12-month admit rate Total charges | LEP patients with same or better utilization rates on all measures and equal 12 month charges as English speakers (t-test; χ2; multiple linear/logistic regression)§ |
| Tocher and Larson (1999), United States | 166 | LEP interpreted versus English speakers | In-person professional (Trained: unclear) | Yes | Time spent with physician Physician perception of time spent and needed. | LEP and English speakers spent same amount of time with physicians (26 minutes; p > 0.05); physicians wanted more time with LEP patients. (t-test; multiple linear regression) |
| Kravitz et al. (2000), United States | 275 | LEP interpreted versus language concordant* | In person professional (Trained: yes) versus ad hoc—family & friends | Yes | Time spent with MD | LEP patients spent more time with physicians overall; the effect was confined to follow-up visits with residents for chronic condition when professional interpreters were present (16 additional minutes for Spanish speakers p = 0.005; 10 additional for Russian speakers p = 0.013) (Zellner's seemingly unrelated regression) |
| Bischoff et al. (2003a,b), Switzerland | 723 | LEP with interpreter or full language concordant (combined group) versus LEP with ad hoc interpreter or partial language concordant (combined group) versus LEP with no interpretation and no language concordance | In-person professional (Trained: yes) versus ad hoc | Yes | Referral for psychological care at the time of application for asylum | Odds of referral for psychological care higher (OR 3.2; CI 1.2-8.6) for those with adequate language concordance (nurse fluent in patient's language or trained interpreter) compared to those with inadequate concordance (no interpreter). No difference in referral rate for those with partial concordance (nurse with some ability or ad hoc interpreter) compared to those with inadequate concordance. (multiple logistic regression) |
| Fagan et al. (2003), United States | 613 | LEP interpreted versus language concordant | In-person professional (Trained: yes), and telephonic professional (Trained: unclear) versus ad hoc—family & friends | Yes | Time spent with clinician (provider time) & total time spent in clinic from check-in to check-out (clinic time) | Patients using a telephone interpreter or an ad hoc interpreter had longer mean provider times (8.3 & 4.6 more minutes, respectively) and those using in-person interpreters had equal mean provider time compared to language concordant patients. (multiple linear regression)§ |
| Jacobs et al. (2001), United States | 4,380 | LEP interpreted versus LEP noninterpreted versus language concordant | In-person professional and telephonic (Trained: yes) | Yes | Preventive service rates Office visit rates Prescription rates | After health plan institution of professional interpreter services, existing differences decreased significantly in all three measures, but least for preventive services. (paired t-tests)§ |
| Bernstein et al. (2002), United States | 500 | LEP interpreted versus LEP noninterpreted versus language concordant | In-person professional (Trained: yes) | No | Intensity of ED services ED return rate Clinic referral rate 30-day charges | LEP with interpreters had lowest ED return rate and highest referral rate; intensity of ED services and 30-day charges closer to language concordant than noninterpreted LEP group. (p < 0.05 all comparisons; ANOVA. Kruskal-Wallis tests for charges) |
| Hampers and McNulty (2002), United States | 4,146 | LEP interpreted versus LEP noninterpreted versus language concordant | In-person professional (Trained: yes) | Yes | Mean test charge IV fluid use Length of ED visit Rate of admission | LEP interpreted patients had utilization closer to language concordant patients on all measures. (multiple logistic/linear regression)§ |
| Clinical Outcomes | ||||||
| Parsons and Day (1992), United Kingdom | 3,781 | LEP interpreted versus LEP noninterpreted versus language concordant | In-person health advocates/professional interpreter (Trained: yes) | No | Instrumental delivery rate Cesarean section rate | After institution of interpreters, LEP patients had lower rates of instrumental delivery (14% versus 7%) and Cesarean section (11% versus 9%). Both types of deliveries increased over the same time period at a control hospital. (p < 0.001; χ2) |
| Tocher and Larson (1998)†, United States | 622 | LEP interpreted versus English speakers | In-person professiona (Trained: unclear) | Yes | HbA1c, lipid, and creatinine values | LEP patients had HbA1c (8.5% versus 8.4%), LDL (132 versus 122 mg/dL) and creatinine (1.1 versus 1.2 mg/dL) values equal to those of English speakers. (p > 0.5 all comparisons; t-test; multiple linear regression) |
| Satisfaction | ||||||
| Hornberger, Itakura, and Wilson (1997), United States | 301 | Type of interpreter: trained professional versus untrained medical staff versus patient's family and friends | In-person professional (Trained: yes), and telephonic professional (Trained: unclear) versus ad hoc—staff, family & friends | No | Clinician satisfaction with quality of interpretation | Clinicians most satisfied with the quality of interpretation when using in-person trained professional interpreters versus ad hoc medical staff or family and friends (p < 0.001; t-test) |
| Kuo and Fagan (1999), United States | 149 patients; 51 docs | LEP interpreted patients versus resident physicians | In person and telephonic professional (Trained: unclear) versus ad hoc—staff, family & friends | No | Clinician and patient satisfaction with different types of interpreters | Patients and physicians had high satisfaction with professional in-person interpreters (92% versus 98%; p = 0.17), and low satisfaction with ad hoc staff (40% versus 44%; p = 0.05). Patients more satisfied than physicians when family and friends interpreted (85% versus 62%; p < 0.01) and less satisfied than physicians with professional telephone interpreters (54% versus 75%; p < 0.01) (Wilcoxon Rank-Sum) |
| Lee et al. (2002) | 536 | LEP interpreted versus language concordant | Telephonic professional (Trained: unclear) versus ad hoc staff, family and friends | Yes | By method of interpretation, patient: Overall satisfaction with visit Satisfaction with seven provider communication characteristics | Patients using professional telephonic interpreters as satisfied as language concordant patients (77% versus 77%p = 0.57); LEP patients using ad hoc staff (54%; p < 0.01) and family and friends (49%; p = 0.007) much less satisfied than language concordant patients both overall and on multiple communication characteristics. (χ2; multiple logistic regression) |
| Bischoff et al. (2003a,b#2), Switzerland | 1,016 | Allophones (non-French speakers) versus language concordant | In-person professional (Trained: yes) | Yes | Satisfaction w/communication before and after clinician training in working with interpreters | Satisfaction with the consultation process and with the doctor's respectfulness increased significantly for allophones after the intervention. (p = 0.04 and 0.001 respectively; multiple linear regression) |
| Ngo-Metzger et al. (2003), United States | 122 | Professional interpreters versus ad hoc interpreters | In-person professional (Trained: yes) versus ad hoc—family | Qualitative—patient focus groups | Preference for type of interpreter | Patients preferred trained professional interpreters over family members due to issues of accuracy and family dynamics. (Grounded Theory analysis) |
| Karliner, Perez-Stable, and Gildengorin (2004), United States | 158 | Clinicians with prior interpreter training versus clinicians w/o prior training | In person professional (Trained: yes) versus ad hoc—family, staff | Yes | Satisfaction with medical care provided during most recent encounter using an interpreter | Clinicians reporting prior training on interpreter use had higher odds of being very satisfied or satisfied with their care provided than those with no prior training. (OR 2.6; CI 1.1–6.6). Clinicians with prior training also has higher odds of using a professional (rather than ad hoc) interpreter (OR 3.2; CI 1.4–7.5) (multiple logistic regression) |
*
Language concordant refers to encounters in which both the patient and the clinician speak the same language; this is most often English, but may be in another language in a non-English speaking country (e.g., Arabic in Saudia Arabia), or in a non-majority language (e.g., Spanish or Chinese in the United States).
†
This study by Tocher et al. appears twice in this table because it addressed the impact of professional interpreters on both utilization and clinical outcomes.
§
p-Values not listed because there are too many to summarize succinctly.