Speaking together: Findings from a hospital disparities collaborative

Background:  To improve the availability and quality of health care language services for patients with limited English proficiency (LEP), the University of Michigan Hospitals and Health Centers (UMHHC) participated in the Speaking Together Collaborative which began in November 2006.  Our objectives were to 1) assess language capture and accuracy; 2) improve the percent of LEP patients receiving interpreter services; 3) evaluate whether the quality of care is similar for LEP and English speaking patients; and 4) enhance staff awareness, competence, and delivery of care in patients’ preferred language.  UMHHC is an academic health system that consists of three hospitals, 30 health centers and 120 outpatient clinics.  In 2007 over 11,000 UMHHC employees assisted in the provision of care for over 1.6 million outpatient visits and 43,000 hospital discharges.  The population in Ann Arbor, Michigan is 72% white, 16% Asian, 7% African American, and 3% Hispanic and top four non-English languages include Spanish, Chinese, Japanese and Arabic.  Providing language specific care is challenging in our institution as limited English proficient (LEP) patients comprise only 2-3% of our population while generating 32,000 interpretation requests in over 40 languages at a cost of $1.6 million annually.  To improve language services a Speaking Together task force was assembled which included the Director of Interpreter Services, Associate Medical Director, Ambulatory Care Services (ACS), Cardiovascular Nursing Champion/Educator, Medical Director Outcomes Management and a data analyst.  During the collaborative, the Director of Admissions, Director of Registration and an ACS project manager were added.

We performed a prospective observational study between November 2006 and May 2008 evaluating the effectiveness of language services using two subsets of our patient population; outpatients with diabetes mellitus and inpatients with congestive heart failure (CHF) admitted to the cardiovascular unit.  UMHHC “active” outpatients with diabetes were defined as those having had two outpatient visits within the past two years and one within the past 13 months.  Over the study period approximately 3,670 patients with diabetes were seen per month with 117 of the patients being LEP (3%).  For CHF, we identified patients with a principal diagnosis of CHF at discharge from the cardiovascular service.   Language field completion was assessed in UMHHC administrative data; accuracy was validated by confirming the "preferred" language with the patient.   Interpreter presence for a LEP patient visit was identified through the outpatient scheduling system, the use of over the phone interpreting, or a bilingual provider.  The primary performance measures included 1) screening for preferred language 2) patients receiving language services (LS) from qualified LS providers 3) HEDIS quality measures were compared for LEP and English.  Collecting monthly data on performance measures provided insight on how well we were meeting the needs of LEP patients in these two groups.

Innovation/Success:  Among all outpatient visits, the proportion of patients screened for preferred language increased from 59% to 84% between November 2006 and May 2008 with targeted inventions such as capturing language at registration, appointment scheduling or check-in or correcting prior inaccurate language data entry.  Several innovative initiatives were implemented in order to improve language capture.  A workflow task was added to the outpatient check-in system to remind staff to collect language information for patients whose language fields are blank on arrival.  When a patient with an empty language field arrives for their appointment, a message appears during check-in that states “Language field blank, please go to patient demographics and complete.”  Staff now complete blank language fields at the time of scheduling an appointment. We also developed daily inpatient and outpatient reports showing patients with missing language field data prompting our managers to investigate areas in which the language field is not being captured consistently (e.g., emergency department admits, direct admit from operating room, labor and delivery).

During the study, the proportion of LEP outpatients with a documented interpreter increased from 19% to 68% through the use of staff, contract interpreters, over the phone interpreting (OPI) and bilingual providers.  During this time period, OPI phones were purchased and training was initiated throughout the health system.  Also, a survey was conducted to identify bilingual faculty who were interested in providing care in their non-English language.  A daily inpatient and outpatient LEP census report was generated to allow staff to verify a LEP patient’s language field and assure interpreting was provided. In addition, contract and staff interpreters use the daily inpatient LEP census report to proactively round on LEP inpatients.  In the outpatient arena, a list of LEP patients with upcoming appointments is printed a day in advance in each clinic to assure an interpreter is scheduled, OPI is available, or the patient is seeing a bilingual provider.  Contracted interpreter visits are linked to the patient’s visit through our Enterprise Wide Scheduling system, which allows the interpreter to be cancelled or rescheduled concurrently with the patient’s visit.  Notification of any patient appointment changes are sent to the interpreter through e-mail and paging.

When investigating the accuracy of language data, we discovered that 20% of patients noted to be LEP actually spoke English well and did not require interpreter services.  We found that during the registration process when asked for “preferred or primary language” many English speaking patients replied with their country of origin as opposed to their preferred language for health care.  The health system implemented a training program to help employees understand why collecting accurate language data is important.  Staff are now trained to ask “what language do you prefer to speak to your doctor or nurse.”  To better capture the population in need of an interpreter, the wording of the registration field was changed from “Primary Language” to “Language for Health Care.”  This change in the method of data collection has improved our health systems ability to correctly identify patients in need of interpreter services.  Lastly, in regards to HEDIS quality measures, there were no differences between LEP and English speaking outpatients with diabetes.

Lessons learned/Challenges:  UMHHC is a large organization and it was important to have the right people on the task force who could help implement change.  Adding a project manager who could meet individually with managers across UMHHC and assist with data collection/analysis was invaluable to our success.  Significant gaps remain in accurately capturing patient preferred language for healthcare and in providing qualified LS services for LEP patients, despite a significant health system expense.   We need to better identify those patients who truly need interpreter services through accurate completion of the language field and provide high quality, cost effective care.

Collaborative Team
            Michelle Harris, BA, MBA, Project Leader
            Connie Standiford, MD, Physician Champion
            Elizabeth Nolan, MS, RN, Nursing Champion
            Pat Warner, MPH, Executive Sponsor
            Steven Bernstein, MD, MPH, Data Analysis Leader
            Paul Paliani, MS, MBA, Project Manager
            Katie Young, Data Analyst
            Pam Chapelle, Director Admissions
            Mary Ann Ryan, Director Registration
            Iris Lagrou, MS, RN, Inpatient Cardiology Champion
            Cathy Kendrick, BSN, RN, Clinical Nurse Manager