Introduction

The integral role of professional interpreters to the provision of safe and quality healthcare for patients with limited English proficiency (LEP) is well-established. Literature shows that providing professional interpreting services to LEP patients in the hospital setting improves communication, decreases communication errors of clinical consequence, increases both patient and clinician satisfaction with healthcare, increases appropriate healthcare utilisation and improves clinical outcomes [1,2,3,4,5,6,7]. However, the impact of professional interpreting provision in Australasian hospitals on patient length of stay (LOS), readmission rates and reducing costs of care is still largely undetermined.

An audit of all public hospital Emergency Departments (ED) in Queensland, Australia, showed that a language spoken at home other than English was shown to be significantly associated with a longer LOS in the ED [8]. Similarly, patients with English as second language (ESL) across 41 Australian sites enrolled in the CONCORDANCE acute coronary syndromes study were shown to have a longer length of overall hospital stay compared to patients with English as a first language [9]. A large retrospective audit in Toronto, Canada showed that patients with LEP experienced longer stays in hospital, ranging from 0.7 to 4.3 days longer based on type of medical condition [10].

It has been hypothesised that at least part of the increased LOS for LEP patients may be the result of communication deficiencies [10, 11]. However, investigation of the impact of the provision of professional interpreting services to LEP patients on their LOS has yielded conflicting results. A small prospective study of Spanish speaking patients in a large urban public hospital in the United States of America (USA), showed no significant difference in LOS in patients provided with professional interpreting services [12]. Another larger study in an USA academic hospital found that that the provision of professional interpreting services was associated with an increased LOS [13]. A third USA hospital study was restricted to patients who had received at least one occasion of interpreting and this study showed a significant decrease in LOS for patients receiving professional interpreting services who were provided these services on the day of admission [11]. Patients who did not have an interpreter present on both admission and discharge days were in hospital about 1.5 days longer than patients who had interpreters on both days [11].

The published findings on the association between professional interpreting engagement and odds of readmission to hospital are similarly mixed. One study found patients receiving professional interpreting services at both admission and discharge were significantly less likely to be readmitted within 30 days—[11], while two other studies found no association between overall professional interpreting provision during hospital stay and readmission [12, 13].

To our knowledge there are no published studies of the relationship between professional interpreting provision and LOS and 30 day readmission in the Australian hospital setting. The aim of this study was to investigate the impact of the provision of professional interpreters during the ED stay and ward stay on the overall LOS and 30-day readmission of LEP patients admitted through the ED of a tertiary referral hospital in Brisbane, Australia.

Methods

Approval to undertake this audit study was granted by the relevant hospital Human Research Ethics Committee.

A retrospective audit of all LEP patients admitted as inpatients to one public tertiary referral adult hospital through the ED was conducted for two calendar years (January 1st 2013–December 31st 2014).

Patients requiring an interpreter who were admitted to the hospital were identified using the Emergency Department Information System (EDIS) database. The “interpreting required” field of EDIS is based on the combination of: (1) Patients’ existing electronic administrative hospital record (iPatient manager (iPM)) that includes three interpreter fields (a tick box for interpreter required (based on self-reported language preference), an interpreter alert flag and language required) and (2) A question asked verbally at ED admission about the patient’s need for an interpreter. For all admissions of patients identified as LEP (based on “interpreter required” field), demographics (gender, age, language spoken), admission and discharge date and time for both ED and ward, LOS and readmission within 30 days was obtained from the hospital iPM database using patient identification number (PIN) and date/time of admission for data matching.

All interpreting occasions at the hospital are recorded in an external interpreter services database, “Interpreter Service Information Service (ISIS)”, managed by the State of Queensland Health Department. Using PINs, admission and discharge dates, and patient name, interpreting records were manually matched to all LEP patient admission occasions. Date and time of each interpreting request, language, duration and mode (telephone or face-to-face) of interpreting were recorded for each admission. All three data sources were merged for each LEP admission onto an Excel spreadsheet.

There were a total of 20,563 admissions of patients through the ED to the hospital over the 2 years. After using this method described there were 448 patients (146 male) identified that required an interpreter. Provision of an interpreter was variable for patients on subsequent visits; therefore the statistical analysis was restricted to the first visit for each of the 448 patients. The effect of provision of an interpreter, age and gender on LOS were screened in univariate analysis and modelled using negative binomial regression. Results are reported as incidence rate ratios (IRR) with 95% confidence intervals, and summary statistics in the original units (hours) as means with Wald 95% confidence intervals (mean, 95% CI) derived from the models.

Re-admission within 30 days of discharge was recorded as a dichotomous variable, with patients admitted more than once recorded as ‘re-admitted’. The effect of provision of interpreter, age and gender on readmission was modelled using logistic regression. Results are reported as Odd Ratios (OR) with 95% confidence intervals. The data was analysed using IBM SPSS Statistics 22 [14].

Results

During the 2 year period of data collection, of a total of 20,563 patient admissions to the hospital through the ED, 582 (2.8%) were identified as LEP and requiring an interpreter. There were 448 individual patients (146 male) that had 582 total admissions during the time period. Three hundred and seventy-two patients had only one visit, while 76 patients re-attended on additional occasions. Overall, interpreter services were provided to 37.2% of admissions identified as patients with LEP. Details about the patterns of provision of professional interpreters during patients’ initial hospital stay and demographics are shown in Table 1.

Table 1 Patient characteristics and presence of professional interpreter at ED and inpatient ward (448 patients)
Full size table

Mean age was 58.7 (SD 23.1) years, 57.1 (SD 23.4) years for males and 59.5 (SD 22.9) years for females. Patients spoke a total of 64 different languages. Greek, Vietnamese, Mandarin, Farsi/Persian and Spanish comprised 31.7% of admissions. Mean LOS overall was 52.01 (SD 72.9, median 24, IQR 11–68.5) hours.

Ninety-three patients (21%) were provided an interpreter in the ED. Provision of interpreter in the ED reduced LOS (IRR 0.58, 0.46–0.74, p < 00001) and age increased LOS (IRR 1.01, 1.01–1.02, p < 0.0001). Mean LOS (adjusted for age) was 53.7 h (95% CI 46.6–62.0) versus 31.3 h (95% CI 24.1–40.7) in those not provided and provided an interpreter in the ED, respectively. Thus mean LOS, adjusted for age, was 22.4 (95% CI 10.9–33.9) hours shorter in those provided an interpreter at ED. Table 2 shows the IRRs for different patterns of interpreting provision. Patients provided an interpreter in ED but not in the inpatient ward had a significantly shorter length of stay compared to patients not provided an interpreter at either ED or ward [IRR 0.41 (0.31–0.55, p < 0.0001)].

Table 2 Mean predicted length of stay (hours) with 95% confidence interval (*adjusted for age), in those who received no professional interpreting, professional interpreting in ED only, professional interpreting in the inpatient ward only and professional interpreting in both ED and inpatient ward
Full size table

One hundred and sixteen patients (26%) were provided an interpreter in the ward. Mean LOS was longer in those provided an interpreter (IRR 2.2, 1.8–2.3, p < 0.0001) and increased with age (IRR 1.02, 1.01–1.02, p < 0.0001). Adjusted for age, mean LOS was 37.6 h (95% CI 33.6–41.6) versus 81.9 h (95% CI 67.3–96.5) in those without and with an interpreter respectively. Table 2 shows that for patients provided only an interpreter in the ward (no provision in ED), their mean LOS was significantly longer at 100.2 h (49.0-204.1) compared to 19.3 h (9.2–40.3) for patients provided only an interpreter in the ED.

Sixty-eight patients (15%) were re-admitted at least once within 30 days of discharge. Five patients were re-admitted twice. There was no association between provision of an interpreter in the ED and re-admission (OR 0.89, 0.45–1.77, p = 0.74) nor any association between age and re-admission (OR 1.0, 1.0–1.02, p = 0.1). Table 3 shows no association between any pattern of interpreter provision in either ED or the inpatient ward and readmission rate.

Table 3 30-day readmission and odds ratios (ORs) associated with professional interpreting provision in the ED and inpatient wards
Full size table

The mode of interpreting in the ED for first occasion provided across all admissions over the 2 years of the audit was predominantly phone, with 75.4% of admissions (86 of 114) provided interpreting in this way. In the inpatient ward, 36.8% (57 of 155) of first occasions of interpreting were provided by phone and the remaining 63.2% (98 of 155) provided onsite.

Discussion

This is the first Australian study to investigate the rates of interpreting provision for hospital patients with LEP, and more specifically examining the effect of engaging professional interpreters in the ED and ward and the impact this practice has on LOS and readmission rates.

The overall rate of engagement of professional interpreters through the hospital journey is low with only 38.6% of patients receiving interpreting services at any point, 20.7% and 25.9% receiving interpreting services at ED and in the ward respectively. These rates are however comparable to rates of hospital interpreting published internationally that vary from 34 to 39% overall [1, 4, 13, 15, 16] and 4 to 36 per cent in the ED [17,18,19,20].

The findings of this study are complex, showing that the provision of professional interpreters in ED is significantly associated with a reduced LOS whilst provision of professional interpreters at the ward is associated with an increased LOS. The finding that access to professional interpretation at the ED has a marked impact on LOS supports the previous work showing that a patient’s access to professional interpretation on the day of admission has the greatest effect on decreasing patient LOS [11].

The importance of professional interpreter engagement in the ED has been highlighted in the literature [21, 22]. Oral communication between the clinician and patient in ED has been shown to be critical to eliciting sufficient patient history and making a timely diagnosis to enable effective and safe clinical decisions [21, 22]. For patients with LEP, engagement of professional interpreting services is integral to effective communication with a clinician [2, 3].

These results support the importance of provision of professional interpreting services early in a patient’s hospital admission for the provision of effective and efficient hospital care. Lindholm and others showed that provision of professional interpreting on both the day of admission and day of discharge significantly decreased patient length of stay [11] but that interpreting engagement at admission had the biggest impact.

The current study shows that engaging professional interpreting services in the ED saves time and money across the patients’ entire hospital stay, with an average reduced LOS of 22.4 h. This result of overall cost and time saving brought about by interpreter engagement provides an important evidence-base to counter the resistance to engaging professional interpreters as a consequence of perceived costs of provision [23, 24]. The cost of provision of 30 min of professional interpreting (phone or in person) is much lower than the cost of an extra day in hospital. As others have shown, although initial costs of providing professional interpreting services is higher than using adhoc or no interpreting, the escalation of long-term costs are prevented [25], and in addition costs are saved through the prevention of well documented health and safety risks [2, 3, 21, 22].

In contrast to a LOS associated with provision of professional interpreting at the beginning of the hospital journey and consistent with Lopez et al. the provision of professional interpreting services throughout the hospital ward admission was associated with a longer LOS [13]. Lopez and others postulated the reason for their finding was possible clinician selective use of interpreters for more medically complicated situations and patients with higher rates of comorbidities [13]. A limitation of the current study was that data on patient comorbidity or seriousness of condition was not included. It was therefore unclear whether there was any causal relationship between interpreter provision on the ward and increased LOS and 30-day readmission rate. However, as shown in Table 1, patients receiving professional interpreting at the ward were older (62.4 years compared to 51.6 years) and more likely to be experiencing more complex health conditions associated with older age. Longer length of stay of these patients may both be a result of more serious medical conditions, for example, requiring interpreting for consent for medical procedures. Patients experiencing longer length of stay in the ward may have increased the opportunity for staff to provide interpreting. This may represent a patient group that is more medically complex and as a result, increase their need of and likelihood of receiving interpreting services.

Consistent with the findings of other studies [12, 13], our study showed no relationship between professional interpreting provision in ED and 30-day readmission. In contrast, Lindholm et al. and others found that patients who did not have access to a professional  interpreter at admission or discharge were more likely to be readmitted within 30 days. A limitation of the comparison with the Lindholm study was that it had a focus on patterns of interpreting and, in contrast to the current study, did not include any patients who did not receive any interpreting services. All LEP patients included had at least one recorded interpretation event during their hospital stay [11]. Patients provided with professional interpreting services in ED were younger and as a result less likely to have longer-term medical conditions reducing the likelihood of readmission. In the context of the ED compared to the ward, it is less likely that complexity of medical condition would be a driver of the decision to engage an interpreter for the purpose of history taking and diagnosis in contrast to consent for procedures once a patient has been admitted to a ward.

Based on the data available on mode of interpreting, quality of interpreting can also not explain the decreased length of stay for those patients provided interpreting services in the ED. A greater proportion of ED first interpreting occasions were phone compared to inpatient ward interpreting occasions (75.4% compared to 36.8%).

Our study has important implications for hospital policy and practice of engaging professional interpreting services. Although the ED environment presents more barriers to professional interpreter engagement than other areas of the hospital [22, 26], we have shown that it also presents the most opportunity for benefit for patient health, safety, patient satisfaction and cost saving through reduced LOS. Providing clinicians education on the evidence-base supporting the impact of interpreter provision, removing structural barriers and increasing positive role modelling are important for change at the level of the individual. Clear organisational norms and structural changes as part of a quality improvement effort is needed at the level of the hospital environment. Quality improvement interventions to increase the use of telephone interpreting and introduce new affordable internet-based technologies for videoconferencing interpreters show promise [13, 27, 28].

There are a number of limitations of the current study that need to be considered. Firstly is its retrospective design meant that data were analysed from the patient records and only the encounters with professional interpreters were extracted. The influence of adhoc interpreters such as family members or bilingual medical staff on both professional interpreter engagement and LOS cannot be determined. The study was conducted in one hospital in Brisbane, Australia and the findings may not be generalizable to other settings. Aside from provision of interpreting in ED compared to the inpatient wards, data on timing of interpreting provision during the inpatient journey (for example, on the day of discharge) was not available. Data on diagnosis or medical complexity was also not available so we can only hypothesize that the increased length of stay in the wards was partly a result of a patient group with more serious, long-term medical conditions associated with older age.

More qualitative research is needed to better understand clinician decision making for engagement of professional interpreting for the different environments of the ED and wards. For example, for what reasons are interpreters engaged at the ward and who engages them? Are nurses more likely to “get by” without engaging interpreters than doctors?

Conclusion

The provision of professional interpreters in ED significantly reduced overall patient LOS, but not 30 day readmission rate. In contrast, provision of interpreting in the ward was significantly associated with an increased LOS. There needs to be further investigation of the association with increased LOS and interpreter provision in the ward environment taking into account severity of patient medical condition, timing and reason for provision (e.g. consent for medical procedure). Research is needed, with a particular focus on understanding clinician decision making for increased engagement of professional interpreting in the ED and early in the hospital journey.

New Contributions to the Literature

  • Engagement of professional interpreters in the Emergency Department can decrease hospital length of stay.

  • Access to professional interpreters in an Australian hospital setting is low but comparable to published rates of access internationally.

  • The association between interpreter use on the ward and length of stay needs further research to determine causality.

  • More research is needed to better understand clinician decision making for engagement of professional interpreters.