Overview of Reviewed Studies

All studies were randomized controlled trials (n=20) or concurrent controlled trials (n=7). Most articles were in the area of prevention (n=19) and most targeted physicians only (n=17). The primary provider intervention was a tracking/reminder system in 10 studies, multifaceted interventions in 9, provider education in 2, bypassing the physician using nurse/nurse practitioners in 2, use of a structured patient questionnaire in 1, use of remote simultaneous translation in 1, use of subspecialty consultation in 1, and use of defibrillators on emergency medical vehicles in 1. Approximately half (n=14) of the studies had a patient intervention component, although these studies varied in whether the patient intervention was provided in addition to the provider intervention or compared with the provider intervention. The intervention was targeted to improve the quality of care specifically for racial/ethnic minorities in only two studies. The most common outcomes were related to healthcare process: appropriateness of care (n=18), quality of providers (n=9), and use of services (n=7).

Quality of Reviewed Studies

Most studies (20 of 27) clearly described healthcare providers and setting, and most (24 of 27) described the intervention sufficiently to ensure replication. Although there were 20 randomized controlled trials, the randomization was considered adequate (in that investigators could not predict assignment) in only 11 studies. Although there were seven concurrent controlled trials, there was one study in which the comparison group was considered inadequate (dissimilar).

Finally, although all studies used objective methods to evaluate outcomes, only nine of 27 had masked outcome assessment, and 13 of 27 performed a pre‐ and a post intervention evaluation. Approximately half (15 of 27) reported the numbers for and reasons for non‐inclusion in the study analysis, and almost all (21 of 27) performed a complete statistical analysis (including the magnitude of difference between groups, an index of variability, and a test statistic).

Results of Reviewed Studies

Twenty‐seven articles qualified for review, each of which used a unique combination of intervention methods in a variety of settings and patient populations. For the purpose of synthesis, we have identified the main intervention method. The categorization of the main intervention method is a simplification of what was often a complex intervention strategy.

Tracking/reminder systems. Ten studies used tracking and/or reminder systems to improve quality of care; of these studies, two were in adult general prevention,^{4, 5} six in adult cancer screening,⁵‐¹⁰ one in tobacco cessation,¹¹ and one in end‐of‐life care (completion of advance directives).¹² All ten studies demonstrated positive outcomes, primarily in the appropriateness of care (such as provision of preventive care, tobacco cessation counseling, or advance directive counseling) category. Overall, there is excellent evidence supporting the use of tracking/reminder systems aimed at providers of racial/ethnic minority patients (Evidence Grade A).

Multifaceted interventions. Nine studies used an intervention characterized as multifaceted, meaning that there were two or more main intervention methods.¹³‐²¹ Two of these interventions were in adult cancer screening,^{13, 14} one in tobacco cessation,¹⁵ one in cholesterol reduction,¹⁶ three in mental health,¹⁷‐¹⁹ one in acute upper respiratory tract infections,²⁰ and one in asthma.²¹ Outcomes of these studies are mixed, with most studies showing improvements in one or two (but not all) outcomes measured. Overall, there is fair evidence supporting the use of multifaceted interventions aimed at providers of racial/ethnic minority patients (Evidence Grade C).

Bypass the physician. Two studies (both in adult cancer screening) bypassed the physician and had either a nurse or a nurse practitioner offer screening directly to patients,^{22, 23} and both studies demonstrated improvements in the provision of preventive services to patients. Overall, there is fair evidence supporting the use of bypassing the providers of racial/ethnic minority patients to offer standardized services directly to patients (Evidence Grade C).

Provider education. Two studies used provider education as the main intervention strategy, one in the area of adult general prevention²⁴ and one in prevention of injuries in children.²⁵ Both studies demonstrated improvements in provider counseling behaviors,^{24, 25} but one measured and did not find any effect of the intervention on parent knowledge of injury prevention (the only outcome categorized as efficacy of treatment) or parent adherence to provider advice.²⁵ Overall, there is fair evidence supporting the use of provider education aimed at providers of racial/ethnic minority patients (Evidence Grade C).

Use of Safe Times Questionnaire (STQ). One study (in the area of prevention in children) used a structured questionnaire to assess adolescent health behaviors and demonstrated a positive impact on providers' counseling behaviors.²⁶ Overall, there is poor evidence supporting the use of structured questionnaires for racial/ethnic minority patients (Evidence Grade D).

Use of Remote Simultaneous Translation (RST). One study compared the accuracy of translation and quality of patient‐physician communication by using remote simultaneous and proximate consecutive interpretation and found fewer translation errors and greater patient and physician satisfaction.²⁷ Overall, there is poor evidence supporting the use of RST for racial/ethnic minority patients (Evidence Grade D).

Use of specialty consult. One study evaluated the use of nephrology consults for patients with chronic kidney disease and found no effect on health care process or patient outcomes.²⁸ Overall, there is poor evidence supporting the use of specialty consults aimed at providers of racial/ethnic minority patients (Evidence Grade D).

Use of defibrillators on emergency medical services. One study evaluated the use of defibrillators on emergency medical services and found no effect on patient outcomes.²⁹ Overall, there is poor evidence supporting the use of defibrillators on emergency medical services (Evidence Grade D).

Results for Question 1a: Strategies to Reduce Disparities

Only one study specifically addressed the question of whether an intervention could reduce disparities in healthcare quality between minorities and white persons.¹⁸ The study, in which two different culturally tailored interventions to improve the quality of depression care were evaluated and compared to a control group that received no intervention, had mixed results.

There was no differential effect of the interventions on healthcare process for white versus minority patients; all patients (African American, Latino, and white) in the intervention groups were more likely than patients in the control group to receive appropriate therapy. However, there was a mixed effect on health outcomes: there were improvements for African‐American and Latino patients in the rate of depression compared with controls (with no improvement for white patients), whereas there were no improvements for African‐American and Latino patients in the intervention groups in employment rates compared with controls (with improvement for white patients). Overall, there is poor evidence to determine which interventions might reduce disparities between racial/ethnic minority patients and majority patients (Evidence Grade D).

Results for Question 1b: Costs of Quality Improvement for Racial/Ethnic Minorities

Only one study reported on the costs of an intervention aimed at improving the quality of healthcare for racial/ethnic minority persons.²⁸ This study, which provided case management and nephrology consultation for patients with chronic renal insufficiency, estimated a minimum yearly cost of $89,355 in 1998 (or $484 per intervention patient) and it was unable to demonstrate any health benefits in its participants. Overall, there is poor evidence to determine the cost of strategies to improve the quality of care for racial/ethnic minorities (Evidence Grade D).

Overview of Reviewed Studies

Of the 64 articles that qualified for our review, only two described randomized controlled trials, eight studies were concurrent controlled trials, and four had an external (non‐concurrent) control group. Most studies were designed without a comparison group; these had either a postintervention evaluation only (n=25), a pre‐ and a postintervention evaluation (n=20), or a qualitative evaluation (n=5). Most of the interventions targeted nurses (n=32) or physicians (n=19).

The content of the curricular interventions varied. Using a previously developed framework to categorize cultural competence curricular content,³⁰ we found that most interventions focused on specific cultural content (n=45), general concepts of culture (n=43), language (n=15), and patient‐provider interaction (n=13). In terms of the specific ethnic minority groups that were the focus of the interventions, 20 studies mentioned Hispanic persons; 19, African Americans; 16, Asians/Pacific Islanders; and 5, American Indians.

Most interventions used more than one training method, and no two studies used exactly the same methods. The most common training methods were group discussion (n=29) and lectures (n=29). Most studies used more than one method for evaluation; the most common method was provider self‐assessment forms (used in 33 studies). Only four articles attempted to measure patient outcomes. Most included some measure of provider outcome; attitude (n=44), knowledge (n=30), or skills/behaviors (n=22).

Quality of Reviewed Studies

Notably, less than half (n=27) of the studies had an objective outcome assessment. Only one third (n=21) included enough detail about the intervention to ensure replication. Only 17 of the interventions were developed with a theoretical model. Only 21 studies clearly described the targeted healthcare providers, setting, and dates of study. Only 15 had a complete statistical analysis. Only 14 included the numbers and reasons for non‐inclusion in the study analysis. Only eight had an adequate comparison group (concurrent and similar). Only two had masking of outcome assessors, and only one had adequate randomization.

Results of Reviewed Studies

In our results below, we focus on the 34 studies with the strongest study design (studies that either had a comparison group and/or did a pre‐ and postintervention evaluation). We do not focus on articles that described interventions evaluated qualitatively or with only a post‐test.

Knowledge. Of the 19 studies that evaluated the effect of cultural competence training on the knowledge of healthcare providers, 17 demonstrated a positive effect, one study showed no effect, and one study demonstrated a partial/mixed effect. Eleven of these studies tested the provider's knowledge about general cultural concepts, seven evaluated culture‐specific knowledge, and one did not provide details to allow determination of content. There was no obvious pattern regarding which type of knowledge was enhanced by cultural competence training. Overall, there is excellent evidence to suggest that cultural competence training increases the knowledge of healthcare providers (Evidence Grade A).

Attitudes. Of the 25 studies that evaluated the effect of cultural competence training on the attitudes of healthcare providers, 21 demonstrated a positive effect, one showed no effect, and three showed a partial/mixed effect. The most common attitude outcome measured was cultural self‐efficacy (measured in three studies), but other types of attitudes were greater understanding of the impact of sociocultural issues on the patient‐physician relationship, more positive attitudes toward community health issues, and an increased interest in learning about patient and family backgrounds. Overall, there is good evidence to suggest that cultural competence training favorably affects the attitudes of healthcare providers (Evidence Grade B).

Skills. Of the 14 studies that evaluated the effect of cultural competence training on the skills of healthcare providers, all demonstrated a positive effect. For example, in one study, participants were given 16 one‐hour sessions in which they practiced communication skills with the community volunteers. They were subsequently shown to be significantly more competent in interviewing a non‐English‐speaking person as rated by a masked psychologist who viewed videotapes of interviews. Other types of skills/behaviors improvements were an increase in nurses' involvement in community‐based cancer education programs, an increase in self‐reported social interactions with peers of different races/ethnicities, and an improved ability of participants to conduct a behavioral analysis and treatment plan. Overall, there is good evidence to suggest that cultural competence training favorably affects the skills/behaviors of healthcare providers (Evidence Grade B).

Patient outcomes. Only three articles evaluated patient outcomes: one targeted physicians,³¹ one targeted mental health counselors,³² and one targeted a mixed group of providers.³³ All three reported favorable patient satisfaction measures,³¹‐³³ and one demonstrated improved adherence to followup among patients assigned to the intervention group providers.³²

In terms of the methods used to bring about such improvements in patient satisfaction and (in one case) adherence, one study trained four mental health counselors about the attitudes that low‐income, African‐American women bring to counseling (4 hours total),³² another trained nine physicians to speak Spanish (20 hours total),³¹ and a third implemented a state‐mandated, 3‐day training program focused on team training, recipient recovery principles, clinical issues, and cultural competence for all staff who have contact with recipients of inpatient mental healthcare.³³ Overall, there is good evidence that cultural competence training improves patient satisfaction (Evidence Grade B) and poor evidence that it affects patient adherence or health outcomes (Evidence Grade D).

Results for Question 2a: Costs of Cultural Competence Training

Of the 55 articles eligible for review, only five addressed the costs of cultural competence training.^{31 , 34}‐³⁷ Four of the five³⁴‐³⁷ described the costs of interventions that involved international travel. In all cases students paid for some portion of the trip, while the school or program paid $0 to $2,000. There are limited data on the costs of classroom or other types of instruction. One study estimated the cost of 20 total hours of Spanish‐language instruction for nine physicians to be $2,000 in 2000, not including the opportunity costs for physician time (approximately 20 hours total for each physician).³¹ In another program, 60 hours of classroom instruction (20 hours of Spanish‐language instruction and 40 hours of cultural competence training focused on Hispanic populations) provided for 19 students had an estimated local cost of $3,000 in 1994, of which each student contributed $80.³⁶ Finally, one program matched involved matching 26 students to 26 local ethnically diverse families, asked the students to visit the family six times, and paid each family $400 in 1996‐2000.³⁵ Overall, there is poor evidence to determine the costs of cultural competence training (Evidence Grade D).

General Limitations

This review was limited to reports published in English (after 1980), as our resources did not permit extensive searching of the non‐English‐language and "gray" literature. Consequently, publication bias is possible. However, recent work has suggested that results of reviews with these limits do not differ substantially from reviews with no such limits.³⁸ Only studies that specifically presented data on racial/ethnic minorities were included.

Limitations of Report and Literature for Question 1

There were limited numbers of studies in each clinical category (except prevention), and few studies focused on priority conditions for which there are documented healthcare disparities (such as HIV/AIDS, cardiovascular disease, diabetes mellitus, and infant mortality). The majority of interventions (all but two) were generic improvement interventions targeted at providers of racial/ethnic minority patients; they did not necessarily target those aspects of care for which there are demonstrated disparities between minority and nonminority populations.

Some of the targeted processes of care were not evidence‐based practices for any patient population (such as oral cavity exams or breast self‐examinations for cancer screening) and thus would be unlikely to improve the quality of care or reduce disparities for racial/ethnic minority patients. Most studies measured health processes, rather than patient outcomes. This characteristic poses a significant limitation for studies that targeted processes of care not already linked to patient outcomes (that is, not evidence‐based).

Evaluating the effectiveness of specific interventions was challenging for several reasons. Each study used slightly different intervention methods, thereby making generalizations across studies difficult. The studies used multicomponent interventions and did not examine separate components.

Very few studies involved Hispanic populations, and none included American Indians/Alaska Natives or Asians/Pacific Islanders. Most studies had no data on costs.

Only interventions targeting providers/organizations were included in this review. Although targeting patients directly may be a promising strategy to improve quality of care and reduce racial/ethnic disparities, such interventions are not reflected here. Only randomized controlled trials and concurrent controlled trials were included; there may be other worthwhile interventions that have been evaluated with other study designs.

Eligibility for our review was limited to studies in the United States, even though there may have been other promising studies conducted in other countries. Finally, we made no assessment of the generalizability of the population of providers targeted in these studies to the broader population of providers caring for racial/ethnic minorities.

Limitations of Report and Literature for Question 2

There are no standardized instruments for measuring cultural competence, and very few outcome assessments were objectively measured. Often there were no data concerning the psychometric properties of the instruments used for evaluation, and most studies were designed without a comparison group for evaluation.

Many articles did not describe the curricular interventions well enough to ensure replication. Furthermore, each curricular intervention was different, making generalizability across studies difficult.

Few studies measured patient outcomes, and none measured healthcare process quality indicators. Some studies used curriculum evaluation as the only outcome. Finally, most studies did not include data on costs.

We made no attempt to assess the psychometric properties of the instruments used to measure cultural competence. Our review focused on interventions aimed at the education of healthcare providers, rather than on an evaluation of all possible organizational strategies to provide culturally and linguistically appropriate services.