Are Language Barriers Associated With Serious Medical Events in Hospitalized Pediatric Patients
http://www.100md.com
《小儿科》
Child Health Institute, Department of Pediatrics, University of Washington, and Children's Hospital and Regional Medical Center, Seattle, Washington
Office of Genomics, Centers for Disease Control and Prevention, Atlanta, Georgia
ABSTRACT
Objective. Language barriers may lead to medical errors by impeding patient-provider communication. The objective of this study was to determine whether hospitalized pediatric patients whose families have language barriers are more likely to incur serious medical errors than patients whose families do not have language barriers.
Methods. A case-control study was conducted in a large, academic, regional children's hospital in the Pacific Northwest. Case patients (n = 97) included all hospitalizations of patients who were younger than 21 years and had a reported serious medical event from January 1, 1998, to December 31, 2003. Control patients (n = 475) were chosen from hospitalizations without a reported serious medical event and were matched with case patients on age, admitting service, admission to intensive care, and date of admission. The main exposure was a language barrier defined by self- or provider-reported need for an interpreter. Serious medical events were defined as events that led to unintended or potentially adverse outcomes identified by the hospital's quality improvement staff.
Results. Fourteen (14.4%) of the case patients and 53 (11.2%) of the control patients were assigned an interpreter during their hospitalization. Overall, we found no increased risk for serious medical events in patients and families who requested an interpreter compared with patients and families who did not request an interpreter (odds ratio: 1.36; 95% confidence interval: 0.73–2.55). Spanish-speaking patients who requested an interpreter comprised 11 (11.3%) of the case patients and 26 (5.5%) of the control patients. This subgroup had a twofold increased risk for serious medical events compared with patients who did not request an interpreter (odds ratio: 2.26; 95% confidence interval: 1.06–4.81).
Conclusions. Spanish-speaking patients whose families have a language barrier seem to have a significantly increased risk for serious medical events during pediatric hospitalization compared with patients whose families do not have a language barrier.
Key Words: language barriers medical errors
Abbreviations: OR, odds ratio CI, confidence interval SES, socioeconomic status
Medical errors are a serious and often preventable problem in hospitals in the United States. Adverse events as a result of medical error occur in 3% to 4% of adult hospitalizations1,2 and in 1% of pediatric hospitalizations.3 The Institute of Medicine has called for broad improvements in the quality of health care in the United States,4 in part by decreasing preventable medical errors.5 Medication errors6,7 and events in which patient safety is compromised8 are common in hospitalized pediatric populations and occur at similar or increased rates to adults. Hospitalized children are at high risk for medical errors because of patient-provider communication barriers and dependence on parents or guardians who cannot continuously oversee their care.
Language barriers may contribute to medical errors by impeding patient-provider communication. One particularly vulnerable group is immigrant children and the children of immigrant parents, the fastest growing segment of the US child population.9 Several barriers may hamper the delivery of optimal care to children of immigrant parents, and foremost among these barriers may be language.10 More than 25% of Hispanic, Asian, and Pacific Islander families are linguistically isolated, meaning that no member of the family who is older than 14 years speaks English well.11 Limited English proficiency in pediatric patients and their families is associated with increased use of diagnostic tests in the emergency department12 and increased rates of hospitalization.12,13 In outpatient adult clinics, patients with a primary language other than English or Spanish were at increased risk for reporting prescription drug complications.14
Very few if any previous studies in children or adults have examined the effect of language barriers on inpatient medical errors. The objective of this study was to determine whether hospitalized pediatric patients whose families have language barriers are more likely to incur serious medical events than patients whose families do not have language barriers.
METHODS
Study Design and Setting
We conducted a case-control study at a large, academic, regional children's hospital in the Pacific Northwest from January 1, 1998, to December 31, 2003. Data on case patients and control patients were obtained primarily from hospital admission and billing databases. Electronic and paper patient charts were reviewed to gather additional data, when data were missing. For both case patients and control patients, we gathered information on the patient's age (<1, 1–3, 4–8, 9–13, or >13 years), gender, race and ethnicity (white, Hispanic, black, Asian American or Pacific Islander, American Indian or Alaskan Native, other, or not identified), need for an interpreter, language of interpreter requested, admitting service (surgical or medical), admission to intensive care (admission to ICU or admission to general medical/surgical ward), insurance (employer-based, Medicaid, or self-pay), and date of admission.
Race and ethnicity were assessed because of the potential correlation among language barriers, race, and ethnicity. On admission to the hospital, a patient representative asked the patient or patient's caregiver the race and ethnicity of the patient with options defined by the hospital admitting system as listed above. At the time of this study, the study hospital offered patients a limited number of choices for race and ethnicity that did not allow for Latino individuals to identify with both Hispanic ethnicity and another race. For the purposes of this study, patients who were identified as Hispanic in the hospital database are referred to as Latino. References to subjects in other studies that identified subjects as Hispanic are referred to as Hispanic.
This study was approved by the institutional review board of Children's Hospital and Regional Medical Center (Seattle, WA). Informed consent was not needed because we analyzed only de-identified, retrospective data.
Case Patients
Case patients included all hospitalizations of patients who were younger than 21 years and had a reported serious medical event during the study period. We included patients up to 21 years of age because the study hospital admits adolescents and young adults up to that age. Serious medical events were defined as events that led to unintended or potentially adverse outcomes in a patient and were attributable to deviation from the hospital's usual practice or policy. Examples of serious medical events in this study include 10-fold medication errors, missed or delayed diagnoses, failure to monitor patient, diagnostic procedures performed on the wrong patient, wrong diagnostic procedure performed, and administration of breast milk to the wrong patient. We did not list the specific observed serious medical events to retain the confidentiality of the patients.
We identified case patients through an electronic database of all serious medical events compiled by the quality improvement staff at the study hospital. The staff is made up of physicians, nurses, and pharmacists who have expertise in hospital quality evaluation and medical error root-cause analysis. When a potentially serious event occurs, it is reported to the quality improvement staff by supervisors or directly by clinical staff involved. Serious medical events are also compiled from review of incident reports and complaints. The associate medical director for quality improvement, the nurse executive, or the medical director review all potentially serious events. When an event is deemed serious by the quality improvement staff as defined above, a multidisciplinary team is charged with analyzing the event and developing an action plan aimed at instituting system change to reduce the likelihood of recurrence. All serious events are systematically followed up to assess implementation of the action plan. A hospital quality improvement committee that consists of doctors, nurses, other health professionals, administrators, and board members periodically reviews serious medical events.
Control Patients
For each case, we chose 5 matched control hospitalizations without a reported serious medical event. We chose to match 5 control patients to each case to maximize the power of our study given the number of available cases. We matched control patients with case patients on 4 criteria: age of patient, admitting service, admission to intensive care, and date of admission. We matched on these criteria to make case patients and control patients crudely comparable on diagnosis, indication for admission, severity of underlying condition, and hospital staffing and location. We chose matched control patients who were admitted during the period 3 months to 2 weeks before the date of admission of the case to control for time of year of admission. We did not choose matched control patients from the period during or after the admission date of the case because we believed that medical care might have changed temporarily at the study hospital in the period after a serious medical event had occurred. To avoid introducing bias, the same hospitalization could be randomly chosen as a matched control for >1 case.15 If a hospitalization for a patient who had experienced a serious medical event was chosen as a control for another case, then those hospitalizations that happened after the serious medical event were dropped from the study.
Exposure Variable
The main exposure variable was a language barrier as defined by self- or provider-reported request for an interpreter. In-person and telephone interpreters are available throughout the hospital 24 hours a day every day of the week. On admission to the hospital, a patient representative asks all patients and families whether they need an interpreter. If the patient or the patient's family requests an interpreter, then the language of interpreter requested is recorded. When patients do not request an interpreter on admission but a medical provider, the patient, or the patient's family decides that an interpreter is necessary during the hospitalization, then the hospital's administrative database is updated to reflect accurately the request for an interpreter.
Data Analysis
Conditional logistic regression with control patients matched to case patients on age of patient, admitting service, admission to intensive care, and date of admission was used to estimate the odds ratios (ORs) for whether a language barrier was associated with serious medical events. We tested for effect modification among subgroups of patients on the basis of age, gender, race and ethnicity, admitting service, admission to intensive care, insurance, and admit year.
RESULTS
There were 97 cases of hospitalizations with serious medical events during the study period. We matched 5 control patients to all but 3 case patients. Because of our matching criteria, 1 case had only 1 control patient and 2 case patients had only 3 control patients. Two control hospitalizations were dropped from the study because the patients had previously experienced serious medical events. This left 475 matched control patients in the study.
The case patients were similar to control patients on gender and all of the matching criteria (age, admitting service, admission to intensive care, and date of admission; Table 1). There were similar percentages of black patients in the case and control groups; however, there were more white, Latino, and American Indian patients and fewer Asian American patients and patients who classified their race as other in the case group. There were more patients with Medicaid insurance in the case group.
Fourteen (14.4%) of the case patients and 53 (11.2%) of the control patients were assigned an interpreter during their hospitalization (Table 2). Overall, we did not detect an increased risk for serious medical events in patients and families who requested an interpreter compared with patients and families who did not request an interpreter (OR: 1.36; 95% confidence interval [CI]: 0.73–2.55). The largest subgroup of patients who requested an interpreter in our study, Spanish-speaking patients, comprised 11 (11.3%) of the case patients and 26 (5.5%) of the control patients. This subgroup had a twofold increased risk for serious medical events compared with patients who had no language barrier (OR: 2.26; 95% CI: 1.06–4.81). Not all Latino patients had a language barrier and were assigned a Spanish interpreter. All Latino patients who had a serious medical event (11 of 11) and 60% of Latino control patients (26 of 43) were assigned an interpreter during their hospitalization.
All other language groups (Amharic, Bulgarian, Cambodian, Cantonese, French, Haitian, Ilocano, Italian, Japanese, Korean, Lao, Mandarin, Russian, Somali, Tagalog, Ukrainian, and Vietnamese) each composed <1% of the total patient population. This collective subgroup of patients who spoke neither English nor Spanish fluently composed 3% of the case patients and 6% of the control patients and had no detectable increased risk for serious medical errors (OR: 0.55; 95% CI: 0.16–1.89). We found no increased risk for a language barrier among subgroups of patients on the basis of age, gender, race and ethnicity, admitting service, admission to intensive care, insurance, and admit year.
DISCUSSION
Serious medical events in hospitalized pediatric patients are attributable to many different causes. Adverse events in hospitalized pediatric patients are more common in older children16 and adolescents.3 Medical errors in hospitalized pediatric patients are more common in children with special medical needs or who are dependent on technology.16 Events in which patient safety for hospitalized pediatric patients is compromised have a high likelihood of representing medical errors and are associated with greater severity of illness.8 Medical errors that lead to adverse drug events in pediatric inpatients are more common in neonates and patients in intensive care.7 The serious medical events in our study encompassed a wide range of ages, hospital settings, and disease processes. A communication barrier between patient and provider could have contributed to all of the serious medical events that we studied and is 1 of the few features that unites them.
Compared with patients and families who did not request an interpreter, Spanish-speaking patients and families who requested an interpreter seemed to have a significantly increased risk for serious medical events during pediatric hospitalization. We did not find an increased risk for the patient group that was fluent in neither English nor Spanish. Language barriers, cultural barriers, and inadequate interpreter use may underlie this finding. One hypothesis for this finding is that some medical providers may erroneously believe that they have an adequate command of Spanish. We speculate that although interpreters were requested more frequently for Spanish-speaking families who incurred a serious medical event, providers may not rely on interpreters for Spanish-speaking patients as much as for patients who speak neither English nor Spanish. Many patients with language barriers may not be given interpreters as often as they should, and poor interpretation can lead to medical errors17 and adverse outcomes.18
We were able only to infer language barriers in our study, because we did not have access to information on when and how often interpreters were used, adequacy of interpretation, availability of interpreters, when interpreters were requested, or language fluency of patient and provider. Errors occur frequently in medical interpretation,19 and adequacy of interpretation may be different for different language groups. A reporting or ascertainment bias may exist for patients who do not speak English, and errors in one language, racial, or ethnic group may be more likely to be reported than in another group. Interpreters may be used more frequently in patients who experience a serious medical event, which may explain why all Latino patients who had a serious medical event were assigned an interpreter compared with 60% of Latino control patients. This trend of increased interpreter use in patients who experience a serious medical event was not seen among patients who were fluent in neither English nor Spanish.
Race and ethnicity are highly associated with language barriers.20 We could not adjust for race or ethnicity in our analysis because language, race, and ethnicity were highly correlated in our study population. The subgroup of patients who are fluent in neither English nor Spanish encompasses many languages, races, and ethnicities. Because of the small numbers within each non-English and non-Spanish language group, we were unable to evaluate the risk for serious medical events for language groups other than Spanish.
In Washington State, Hispanic ethnicity is associated with lower median household income than white, non-Hispanic race,21 and families with language barriers may have even lower socioeconomic status (SES). Because Hispanic ethnicity is associated with speaking Spanish, lower SES may in theory contribute to worse outcomes and increased serious medical events among Spanish-speaking patients compared with other language groups. Because very few families with a language barrier had employer-based insurance in our study population, we could not adjust for insurance as a proxy for SES. In addition, a language barrier may affect insurance status directly and may lie along the causal pathway between language barriers and serious medical events; therefore, statistical adjustment for this variable may be inappropriate. We did not have data to adjust for parental education level, another marker for SES. Similarly, acculturation in immigrants is associated with both language barriers and health status,22 and less acculturated individuals may be at increased risk for serious medical errors. Acculturation is complex to measure, and we did not have data available other than request for an interpreter to evaluate acculturation in our study.
The number of serious medical events increased with each year as a result of improved identification of events at the study hospital over the course of the study. Because we relied on serious medical events identified by the hospital quality improvement staff and not on chart review, we did not have an exhaustive list of all serious events at the study hospital during the study period. In addition, our list of serious medical events underestimates the number of medical errors of all types that occurred at the study hospital during the study period. The number of adverse events from medical errors in our study is less than the 1% reported nationally,3 which reflects the study hospital's identification of only serious medical events as defined by the quality improvement staff. We specifically evaluated the risk for serious medical events instead of the risk for all medical errors because serious medical events by definition have led to a serious adverse outcome, and it is these events that are particularly important to explore to improve the safety of the patients. We included cases from all 6 years of the study to maximize our power, and we controlled for year of admission in our analysis to reduce any bias that this might have introduced.
In our analysis, we controlled for factors that are associated with adverse events in pediatric inpatients, such as age and severity of illness as measured by admission to intensive care. Children with high medical complexity, chronic medical conditions, or special health care needs may be more likely to be assigned an interpreter. We were unable to adjust specifically for these risk factors for serious medical events as potential confounders in our analysis.
These results are important because they highlight a group of hospitalized children who are at increased risk for serious medical errors. Our results reflect all reported serious medical events over a 6-year period at 1 large, academic, regional children's hospital with a relatively high percentage of non–English-speaking patients. Because this hospital serves a racially and ethnically diverse pediatric population over a 5-state region, we believe that these results are generalizable to other regional pediatric hospitals. These results have large implications for improving quality of pediatric care and suggest that minimizing language barriers may be an important key to limiting medical errors in hospitalized pediatric patients.
ACKNOWLEDGMENTS
Dr Cohen was supported by a National Research Service Award during the course of this study.
We thank Decision Support at Children's Hospital and Regional Medical Center for data collection.
FOOTNOTES
Accepted May 19, 2005.
No conflict of interest declared.
Dr. Cohen’s current affiliation is the Epidemic Intelligence Service, Division of Healthcare Quality Promotion, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia.
REFERENCES
Brennan TA, Leape LL, Laird NM, et al. Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med. 1991;324 :370 –376
Thomas EJ, Studdert DM, Burstin HR, et al. Incidence and types of adverse events and negligent care in Utah and Colorado. Med Care. 2000;38 :261 –271
Woods D, Thomas E, Holl J, Altman S, Brennan T. Adverse events and preventable adverse events in children. Pediatrics. 2005;115 :155 –160
Institute of Medicine (U.S.), Committee on Quality of Health Care in America. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press; 2001
Corrigan J, Donaldson MS, Kohn LT. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press; 2000
Fortescue EB, Kaushal R, Landrigan CP, et al. Prioritizing strategies for preventing medication errors and adverse drug events in pediatric inpatients. Pediatrics. 2003;111 :722 –729
Kaushal R, Bates DW, Landrigan C, et al. Medication errors and adverse drug events in pediatric inpatients. JAMA. 2001;285 :2114 –2120
Miller MR, Elixhauser A, Zhan C. Patient safety events during pediatric hospitalizations. Pediatrics. 2003;111 :1358 –1366
Jenista JA. The immigrant, refugee, or internationally adopted child. Pediatr Rev. 2001;22 :419 –429
Nielsen-Bohlman L, Panzer AM, David A. Kindig E, Committee on Health Literacy. Health Literacy: A Prescription to End Confusion. Washington, DC: National Academy Press; 2004
Hampers LC, Cha S, Gutglass DJ, Binns HJ, Krug SE. Language barriers and resource utilization in a pediatric emergency department. Pediatrics. 1999;103 :1253 –1256
Hampers LC, McNulty JE. Professional interpreters and bilingual physicians in a pediatric emergency department: effect on resource utilization. Arch Pediatr Adolesc Med. 2002;156 :1108 –1113
Gandhi TK, Burstin HR, Cook EF, et al. Drug complications in outpatients. J Gen Intern Med. 2000;15 :149 –154
Wacholder S, Silverman DT, McLaughlin JK, Mandel JS. Selection of controls in case-control studies. III. Design options. Am J Epidemiol. 1992;135 :1042 –1050
Slonim AD, LaFleur BJ, Ahmed W, Joseph JG. Hospital-reported medical errors in children. Pediatrics. 2003;111 :617 –621
Flores G, Abreu M, Schwartz I, Hill M. The importance of language and culture in pediatric care: case studies from the Latino community. J Pediatr. 2000;137 :842 –848
Harsham P. A misinterpreted word worth $71 million. Med Econ. 1984:289 –292
Flores G, Laws MB, Mayo SJ, et al. Errors in medical interpretation and their potential clinical consequences in pediatric encounters. Pediatrics. 2003;111 :6 –14
Smedley BD, Stith AY, Nelson AR, Institute of Medicine (U. S.), Committee on Understanding and Eliminating Racial and Ethnic Disparities in Health Care. Unequal Treatment Confronting Racial and Ethnic Disparities in Healthcare. Washington, DC: National Academy Press; 2002
Washington Kids Count. 2003 State of Washington's Children Report. Seattle, WA: Human Services Policy Center, University of Washington; 2003
Franzini L, Fernandez-Esquer ME. Socioeconomic, cultural, and personal influences on health outcomes in low income Mexican-origin individuals in Texas. Soc Sci Med. 2004;59 :1629 –1646
Ku L, Flores G. Pay now or pay later: providing interpreter services in health care. Health Aff (Millwood). 2005;24 :435 –444
Baker DW, Parker RM, Williams MV, Coates WC, Pitkin K. Use and effectiveness of interpreters in an emergency department. JAMA. 1996;275 :783 –788
US Census. U.S. Interim projections by age, sex, race, and Hispanic origin, Table 1a. Projected population change in the United States, by race and Hispanic origin: 2000 to 2050. Available at: www.census.gov/ipc/www/usinterimproj. Accessed September 5, 2004(Adam L. Cohen, MD, MPH, F)
Office of Genomics, Centers for Disease Control and Prevention, Atlanta, Georgia
ABSTRACT
Objective. Language barriers may lead to medical errors by impeding patient-provider communication. The objective of this study was to determine whether hospitalized pediatric patients whose families have language barriers are more likely to incur serious medical errors than patients whose families do not have language barriers.
Methods. A case-control study was conducted in a large, academic, regional children's hospital in the Pacific Northwest. Case patients (n = 97) included all hospitalizations of patients who were younger than 21 years and had a reported serious medical event from January 1, 1998, to December 31, 2003. Control patients (n = 475) were chosen from hospitalizations without a reported serious medical event and were matched with case patients on age, admitting service, admission to intensive care, and date of admission. The main exposure was a language barrier defined by self- or provider-reported need for an interpreter. Serious medical events were defined as events that led to unintended or potentially adverse outcomes identified by the hospital's quality improvement staff.
Results. Fourteen (14.4%) of the case patients and 53 (11.2%) of the control patients were assigned an interpreter during their hospitalization. Overall, we found no increased risk for serious medical events in patients and families who requested an interpreter compared with patients and families who did not request an interpreter (odds ratio: 1.36; 95% confidence interval: 0.73–2.55). Spanish-speaking patients who requested an interpreter comprised 11 (11.3%) of the case patients and 26 (5.5%) of the control patients. This subgroup had a twofold increased risk for serious medical events compared with patients who did not request an interpreter (odds ratio: 2.26; 95% confidence interval: 1.06–4.81).
Conclusions. Spanish-speaking patients whose families have a language barrier seem to have a significantly increased risk for serious medical events during pediatric hospitalization compared with patients whose families do not have a language barrier.
Key Words: language barriers medical errors
Abbreviations: OR, odds ratio CI, confidence interval SES, socioeconomic status
Medical errors are a serious and often preventable problem in hospitals in the United States. Adverse events as a result of medical error occur in 3% to 4% of adult hospitalizations1,2 and in 1% of pediatric hospitalizations.3 The Institute of Medicine has called for broad improvements in the quality of health care in the United States,4 in part by decreasing preventable medical errors.5 Medication errors6,7 and events in which patient safety is compromised8 are common in hospitalized pediatric populations and occur at similar or increased rates to adults. Hospitalized children are at high risk for medical errors because of patient-provider communication barriers and dependence on parents or guardians who cannot continuously oversee their care.
Language barriers may contribute to medical errors by impeding patient-provider communication. One particularly vulnerable group is immigrant children and the children of immigrant parents, the fastest growing segment of the US child population.9 Several barriers may hamper the delivery of optimal care to children of immigrant parents, and foremost among these barriers may be language.10 More than 25% of Hispanic, Asian, and Pacific Islander families are linguistically isolated, meaning that no member of the family who is older than 14 years speaks English well.11 Limited English proficiency in pediatric patients and their families is associated with increased use of diagnostic tests in the emergency department12 and increased rates of hospitalization.12,13 In outpatient adult clinics, patients with a primary language other than English or Spanish were at increased risk for reporting prescription drug complications.14
Very few if any previous studies in children or adults have examined the effect of language barriers on inpatient medical errors. The objective of this study was to determine whether hospitalized pediatric patients whose families have language barriers are more likely to incur serious medical events than patients whose families do not have language barriers.
METHODS
Study Design and Setting
We conducted a case-control study at a large, academic, regional children's hospital in the Pacific Northwest from January 1, 1998, to December 31, 2003. Data on case patients and control patients were obtained primarily from hospital admission and billing databases. Electronic and paper patient charts were reviewed to gather additional data, when data were missing. For both case patients and control patients, we gathered information on the patient's age (<1, 1–3, 4–8, 9–13, or >13 years), gender, race and ethnicity (white, Hispanic, black, Asian American or Pacific Islander, American Indian or Alaskan Native, other, or not identified), need for an interpreter, language of interpreter requested, admitting service (surgical or medical), admission to intensive care (admission to ICU or admission to general medical/surgical ward), insurance (employer-based, Medicaid, or self-pay), and date of admission.
Race and ethnicity were assessed because of the potential correlation among language barriers, race, and ethnicity. On admission to the hospital, a patient representative asked the patient or patient's caregiver the race and ethnicity of the patient with options defined by the hospital admitting system as listed above. At the time of this study, the study hospital offered patients a limited number of choices for race and ethnicity that did not allow for Latino individuals to identify with both Hispanic ethnicity and another race. For the purposes of this study, patients who were identified as Hispanic in the hospital database are referred to as Latino. References to subjects in other studies that identified subjects as Hispanic are referred to as Hispanic.
This study was approved by the institutional review board of Children's Hospital and Regional Medical Center (Seattle, WA). Informed consent was not needed because we analyzed only de-identified, retrospective data.
Case Patients
Case patients included all hospitalizations of patients who were younger than 21 years and had a reported serious medical event during the study period. We included patients up to 21 years of age because the study hospital admits adolescents and young adults up to that age. Serious medical events were defined as events that led to unintended or potentially adverse outcomes in a patient and were attributable to deviation from the hospital's usual practice or policy. Examples of serious medical events in this study include 10-fold medication errors, missed or delayed diagnoses, failure to monitor patient, diagnostic procedures performed on the wrong patient, wrong diagnostic procedure performed, and administration of breast milk to the wrong patient. We did not list the specific observed serious medical events to retain the confidentiality of the patients.
We identified case patients through an electronic database of all serious medical events compiled by the quality improvement staff at the study hospital. The staff is made up of physicians, nurses, and pharmacists who have expertise in hospital quality evaluation and medical error root-cause analysis. When a potentially serious event occurs, it is reported to the quality improvement staff by supervisors or directly by clinical staff involved. Serious medical events are also compiled from review of incident reports and complaints. The associate medical director for quality improvement, the nurse executive, or the medical director review all potentially serious events. When an event is deemed serious by the quality improvement staff as defined above, a multidisciplinary team is charged with analyzing the event and developing an action plan aimed at instituting system change to reduce the likelihood of recurrence. All serious events are systematically followed up to assess implementation of the action plan. A hospital quality improvement committee that consists of doctors, nurses, other health professionals, administrators, and board members periodically reviews serious medical events.
Control Patients
For each case, we chose 5 matched control hospitalizations without a reported serious medical event. We chose to match 5 control patients to each case to maximize the power of our study given the number of available cases. We matched control patients with case patients on 4 criteria: age of patient, admitting service, admission to intensive care, and date of admission. We matched on these criteria to make case patients and control patients crudely comparable on diagnosis, indication for admission, severity of underlying condition, and hospital staffing and location. We chose matched control patients who were admitted during the period 3 months to 2 weeks before the date of admission of the case to control for time of year of admission. We did not choose matched control patients from the period during or after the admission date of the case because we believed that medical care might have changed temporarily at the study hospital in the period after a serious medical event had occurred. To avoid introducing bias, the same hospitalization could be randomly chosen as a matched control for >1 case.15 If a hospitalization for a patient who had experienced a serious medical event was chosen as a control for another case, then those hospitalizations that happened after the serious medical event were dropped from the study.
Exposure Variable
The main exposure variable was a language barrier as defined by self- or provider-reported request for an interpreter. In-person and telephone interpreters are available throughout the hospital 24 hours a day every day of the week. On admission to the hospital, a patient representative asks all patients and families whether they need an interpreter. If the patient or the patient's family requests an interpreter, then the language of interpreter requested is recorded. When patients do not request an interpreter on admission but a medical provider, the patient, or the patient's family decides that an interpreter is necessary during the hospitalization, then the hospital's administrative database is updated to reflect accurately the request for an interpreter.
Data Analysis
Conditional logistic regression with control patients matched to case patients on age of patient, admitting service, admission to intensive care, and date of admission was used to estimate the odds ratios (ORs) for whether a language barrier was associated with serious medical events. We tested for effect modification among subgroups of patients on the basis of age, gender, race and ethnicity, admitting service, admission to intensive care, insurance, and admit year.
RESULTS
There were 97 cases of hospitalizations with serious medical events during the study period. We matched 5 control patients to all but 3 case patients. Because of our matching criteria, 1 case had only 1 control patient and 2 case patients had only 3 control patients. Two control hospitalizations were dropped from the study because the patients had previously experienced serious medical events. This left 475 matched control patients in the study.
The case patients were similar to control patients on gender and all of the matching criteria (age, admitting service, admission to intensive care, and date of admission; Table 1). There were similar percentages of black patients in the case and control groups; however, there were more white, Latino, and American Indian patients and fewer Asian American patients and patients who classified their race as other in the case group. There were more patients with Medicaid insurance in the case group.
Fourteen (14.4%) of the case patients and 53 (11.2%) of the control patients were assigned an interpreter during their hospitalization (Table 2). Overall, we did not detect an increased risk for serious medical events in patients and families who requested an interpreter compared with patients and families who did not request an interpreter (OR: 1.36; 95% confidence interval [CI]: 0.73–2.55). The largest subgroup of patients who requested an interpreter in our study, Spanish-speaking patients, comprised 11 (11.3%) of the case patients and 26 (5.5%) of the control patients. This subgroup had a twofold increased risk for serious medical events compared with patients who had no language barrier (OR: 2.26; 95% CI: 1.06–4.81). Not all Latino patients had a language barrier and were assigned a Spanish interpreter. All Latino patients who had a serious medical event (11 of 11) and 60% of Latino control patients (26 of 43) were assigned an interpreter during their hospitalization.
All other language groups (Amharic, Bulgarian, Cambodian, Cantonese, French, Haitian, Ilocano, Italian, Japanese, Korean, Lao, Mandarin, Russian, Somali, Tagalog, Ukrainian, and Vietnamese) each composed <1% of the total patient population. This collective subgroup of patients who spoke neither English nor Spanish fluently composed 3% of the case patients and 6% of the control patients and had no detectable increased risk for serious medical errors (OR: 0.55; 95% CI: 0.16–1.89). We found no increased risk for a language barrier among subgroups of patients on the basis of age, gender, race and ethnicity, admitting service, admission to intensive care, insurance, and admit year.
DISCUSSION
Serious medical events in hospitalized pediatric patients are attributable to many different causes. Adverse events in hospitalized pediatric patients are more common in older children16 and adolescents.3 Medical errors in hospitalized pediatric patients are more common in children with special medical needs or who are dependent on technology.16 Events in which patient safety for hospitalized pediatric patients is compromised have a high likelihood of representing medical errors and are associated with greater severity of illness.8 Medical errors that lead to adverse drug events in pediatric inpatients are more common in neonates and patients in intensive care.7 The serious medical events in our study encompassed a wide range of ages, hospital settings, and disease processes. A communication barrier between patient and provider could have contributed to all of the serious medical events that we studied and is 1 of the few features that unites them.
Compared with patients and families who did not request an interpreter, Spanish-speaking patients and families who requested an interpreter seemed to have a significantly increased risk for serious medical events during pediatric hospitalization. We did not find an increased risk for the patient group that was fluent in neither English nor Spanish. Language barriers, cultural barriers, and inadequate interpreter use may underlie this finding. One hypothesis for this finding is that some medical providers may erroneously believe that they have an adequate command of Spanish. We speculate that although interpreters were requested more frequently for Spanish-speaking families who incurred a serious medical event, providers may not rely on interpreters for Spanish-speaking patients as much as for patients who speak neither English nor Spanish. Many patients with language barriers may not be given interpreters as often as they should, and poor interpretation can lead to medical errors17 and adverse outcomes.18
We were able only to infer language barriers in our study, because we did not have access to information on when and how often interpreters were used, adequacy of interpretation, availability of interpreters, when interpreters were requested, or language fluency of patient and provider. Errors occur frequently in medical interpretation,19 and adequacy of interpretation may be different for different language groups. A reporting or ascertainment bias may exist for patients who do not speak English, and errors in one language, racial, or ethnic group may be more likely to be reported than in another group. Interpreters may be used more frequently in patients who experience a serious medical event, which may explain why all Latino patients who had a serious medical event were assigned an interpreter compared with 60% of Latino control patients. This trend of increased interpreter use in patients who experience a serious medical event was not seen among patients who were fluent in neither English nor Spanish.
Race and ethnicity are highly associated with language barriers.20 We could not adjust for race or ethnicity in our analysis because language, race, and ethnicity were highly correlated in our study population. The subgroup of patients who are fluent in neither English nor Spanish encompasses many languages, races, and ethnicities. Because of the small numbers within each non-English and non-Spanish language group, we were unable to evaluate the risk for serious medical events for language groups other than Spanish.
In Washington State, Hispanic ethnicity is associated with lower median household income than white, non-Hispanic race,21 and families with language barriers may have even lower socioeconomic status (SES). Because Hispanic ethnicity is associated with speaking Spanish, lower SES may in theory contribute to worse outcomes and increased serious medical events among Spanish-speaking patients compared with other language groups. Because very few families with a language barrier had employer-based insurance in our study population, we could not adjust for insurance as a proxy for SES. In addition, a language barrier may affect insurance status directly and may lie along the causal pathway between language barriers and serious medical events; therefore, statistical adjustment for this variable may be inappropriate. We did not have data to adjust for parental education level, another marker for SES. Similarly, acculturation in immigrants is associated with both language barriers and health status,22 and less acculturated individuals may be at increased risk for serious medical errors. Acculturation is complex to measure, and we did not have data available other than request for an interpreter to evaluate acculturation in our study.
The number of serious medical events increased with each year as a result of improved identification of events at the study hospital over the course of the study. Because we relied on serious medical events identified by the hospital quality improvement staff and not on chart review, we did not have an exhaustive list of all serious events at the study hospital during the study period. In addition, our list of serious medical events underestimates the number of medical errors of all types that occurred at the study hospital during the study period. The number of adverse events from medical errors in our study is less than the 1% reported nationally,3 which reflects the study hospital's identification of only serious medical events as defined by the quality improvement staff. We specifically evaluated the risk for serious medical events instead of the risk for all medical errors because serious medical events by definition have led to a serious adverse outcome, and it is these events that are particularly important to explore to improve the safety of the patients. We included cases from all 6 years of the study to maximize our power, and we controlled for year of admission in our analysis to reduce any bias that this might have introduced.
In our analysis, we controlled for factors that are associated with adverse events in pediatric inpatients, such as age and severity of illness as measured by admission to intensive care. Children with high medical complexity, chronic medical conditions, or special health care needs may be more likely to be assigned an interpreter. We were unable to adjust specifically for these risk factors for serious medical events as potential confounders in our analysis.
These results are important because they highlight a group of hospitalized children who are at increased risk for serious medical errors. Our results reflect all reported serious medical events over a 6-year period at 1 large, academic, regional children's hospital with a relatively high percentage of non–English-speaking patients. Because this hospital serves a racially and ethnically diverse pediatric population over a 5-state region, we believe that these results are generalizable to other regional pediatric hospitals. These results have large implications for improving quality of pediatric care and suggest that minimizing language barriers may be an important key to limiting medical errors in hospitalized pediatric patients.
ACKNOWLEDGMENTS
Dr Cohen was supported by a National Research Service Award during the course of this study.
We thank Decision Support at Children's Hospital and Regional Medical Center for data collection.
FOOTNOTES
Accepted May 19, 2005.
No conflict of interest declared.
Dr. Cohen’s current affiliation is the Epidemic Intelligence Service, Division of Healthcare Quality Promotion, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia.
REFERENCES
Brennan TA, Leape LL, Laird NM, et al. Incidence of adverse events and negligence in hospitalized patients. Results of the Harvard Medical Practice Study I. N Engl J Med. 1991;324 :370 –376
Thomas EJ, Studdert DM, Burstin HR, et al. Incidence and types of adverse events and negligent care in Utah and Colorado. Med Care. 2000;38 :261 –271
Woods D, Thomas E, Holl J, Altman S, Brennan T. Adverse events and preventable adverse events in children. Pediatrics. 2005;115 :155 –160
Institute of Medicine (U.S.), Committee on Quality of Health Care in America. Crossing the Quality Chasm: A New Health System for the 21st Century. Washington, DC: National Academy Press; 2001
Corrigan J, Donaldson MS, Kohn LT. To Err Is Human: Building a Safer Health System. Washington, DC: National Academy Press; 2000
Fortescue EB, Kaushal R, Landrigan CP, et al. Prioritizing strategies for preventing medication errors and adverse drug events in pediatric inpatients. Pediatrics. 2003;111 :722 –729
Kaushal R, Bates DW, Landrigan C, et al. Medication errors and adverse drug events in pediatric inpatients. JAMA. 2001;285 :2114 –2120
Miller MR, Elixhauser A, Zhan C. Patient safety events during pediatric hospitalizations. Pediatrics. 2003;111 :1358 –1366
Jenista JA. The immigrant, refugee, or internationally adopted child. Pediatr Rev. 2001;22 :419 –429
Nielsen-Bohlman L, Panzer AM, David A. Kindig E, Committee on Health Literacy. Health Literacy: A Prescription to End Confusion. Washington, DC: National Academy Press; 2004
Hampers LC, Cha S, Gutglass DJ, Binns HJ, Krug SE. Language barriers and resource utilization in a pediatric emergency department. Pediatrics. 1999;103 :1253 –1256
Hampers LC, McNulty JE. Professional interpreters and bilingual physicians in a pediatric emergency department: effect on resource utilization. Arch Pediatr Adolesc Med. 2002;156 :1108 –1113
Gandhi TK, Burstin HR, Cook EF, et al. Drug complications in outpatients. J Gen Intern Med. 2000;15 :149 –154
Wacholder S, Silverman DT, McLaughlin JK, Mandel JS. Selection of controls in case-control studies. III. Design options. Am J Epidemiol. 1992;135 :1042 –1050
Slonim AD, LaFleur BJ, Ahmed W, Joseph JG. Hospital-reported medical errors in children. Pediatrics. 2003;111 :617 –621
Flores G, Abreu M, Schwartz I, Hill M. The importance of language and culture in pediatric care: case studies from the Latino community. J Pediatr. 2000;137 :842 –848
Harsham P. A misinterpreted word worth $71 million. Med Econ. 1984:289 –292
Flores G, Laws MB, Mayo SJ, et al. Errors in medical interpretation and their potential clinical consequences in pediatric encounters. Pediatrics. 2003;111 :6 –14
Smedley BD, Stith AY, Nelson AR, Institute of Medicine (U. S.), Committee on Understanding and Eliminating Racial and Ethnic Disparities in Health Care. Unequal Treatment Confronting Racial and Ethnic Disparities in Healthcare. Washington, DC: National Academy Press; 2002
Washington Kids Count. 2003 State of Washington's Children Report. Seattle, WA: Human Services Policy Center, University of Washington; 2003
Franzini L, Fernandez-Esquer ME. Socioeconomic, cultural, and personal influences on health outcomes in low income Mexican-origin individuals in Texas. Soc Sci Med. 2004;59 :1629 –1646
Ku L, Flores G. Pay now or pay later: providing interpreter services in health care. Health Aff (Millwood). 2005;24 :435 –444
Baker DW, Parker RM, Williams MV, Coates WC, Pitkin K. Use and effectiveness of interpreters in an emergency department. JAMA. 1996;275 :783 –788
US Census. U.S. Interim projections by age, sex, race, and Hispanic origin, Table 1a. Projected population change in the United States, by race and Hispanic origin: 2000 to 2050. Available at: www.census.gov/ipc/www/usinterimproj. Accessed September 5, 2004(Adam L. Cohen, MD, MPH, F)