Introduction

While health literacy (HL) is a multifaceted concept [1, 2] almost all definitions relate HL to “the literacy and numeracy skills that enable individuals to obtain, understand, appraise, and use information to make decisions and take actions that will have an impact on health status” [3]. Low HL has been linked to poorer health outcomes, including increased mortality [4, 5]. HL has increasingly been recognized as a potentially important factor mediating health disparities, especially those related to race and ethnicity [5], and has been suggested as an important mediator of the relationship between socioeconomic status and health [6]. This may be due to communication barriers with physicians and difficulty understanding and making use of medical resources [5].

As a concept, HL has sometimes been poorly defined. A recent systematic review which sought to clarify the concept found that scholars commonly characterized HL along three main domains: knowledge of health/healthcare systems, processing and using information related to health and healthcare, and the ability to maintain health through collaboration with health providers [7]. Other theoretical frameworks developed for HL understand the concept through its effects. For example Nutbeam established a useful framework for understanding the benefits of health literacy through a “health outcomes model” in which HL is comprised of functional HL, the basic skills necessary for everyday health functioning, communicative/interactive HL, the more advanced skills needed to act independently with “motivation and self-confidence,” and critical HL, the ability to analyze and use information to “exert greater control over life events and situations” allowing people to respond adversity and to advocate for themselves [8, 9]. HL is sometimes understood as not only a skill, but an important social determinant of health, with community level and public health implications [10].

While many U.S. residents struggle with limited health literacy, there may be a particular barrier among those who speak Spanish preferentially or exclusively, including Hispanic, immigrant or migrant populations. In the United States, minority groups, immigrants, migrants, and nonnative English speakers have lower health literacy scores than White adults and are at higher risk of having poor HL, making them more susceptible to the adverse outcomes associated with low HL [11]. Hispanics are the largest group of nonnative English speakers and preferential Spanish speakers in the U.S. and have low rates of HL compared to other populations [5]. Limited English proficiency may be a factor that contributes to poorer health outcomes and reduced quality of care, especially in a predominantly English language-based health care system with a shortage of bilingual and culturally competent providers [12]. For example, one recent study found higher rates of obesity among Spanish speakers in the United States [13]. These factors, in combination with a lack of healthcare access and insurance coverage, may contribute to higher morbidity and mortality rates among Hispanics due to chronic diseases such as diabetes and obesity [14].

Methods to accommodate the HL needs of patients with a Spanish-language preference (SLP) may therefore be important in improving health equity [15]. While strides have been made in community-based educational efforts and the translation or cultural adaptation of health communication tools and processes [16], there are limited data on effective interventions to improve HL for patients with SLP in the United States [17]. The literature on interventions targeting HL in the United States has frequently grouped together populations of immigrants who do not share a common language [18] or, conversely, focused only on individuals from a single nationality [19, 20]. Given the gap in the literature synthesizing research on HL interventions for patients with SLP in the United States and the important association between HL and health outcomes, we conducted a systematic review of the literature that summarizes and evaluates the effectiveness of HL intervention strategies for patients with SLP in the United States. The aim of this systematic review was to describe existing HL interventions for patients with SLP and present their reported effectiveness.

Methods

Protocol and registration

The protocol for this review was registered with PROSPERO (CRD42021257655). The use of the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) enabled authors to follow best practices in conducting the review [21].

Search strategy and screening

Searches were conducted in the PubMed, MEDLINE, Web of Science, and Embase databases and data was extracted from these databases between January 20, 2020 and April 27, 2023 (Fig. 1.). The keywords for each database included: “health literacy” and “intervention” or “Spanish”, “Hispanic” or “LEP,” or “limited English proficiency.” Databases were queried to include only articles published between January 1, 2011 and April 27, 2023. In 2010, the U.S. The Department of Health and Human Services unveiled the National Action Plan to Improve Health Literacy, bringing more attention to this matter and inspiring more research on HL. Our review also avoids redundancy with a 2011 comprehensive review [5], which found no interventions focused on HL in Spanish-speaking populations, with only three mentioned measures of HL in this population.

Fig. 1
figure 1

PRISMA flow diagram for studies considered for the systematic review

Full size image

After removing duplicates, two reviewers, P.P. and L.D., independently reviewed titles and abstracts to select potentially eligible articles based on the inclusion/exclusion criteria described below. Any disagreements regarding the inclusion of a study were resolved by a third reviewer, J.H. Bibliographies of included studies were subsequently hand searched.

Inclusion & exclusion criteria

Inclusion criteria for this literature review included articles that a) featured participants with SLP, b) described interventions that occurred in the United States, c) described interventions that were designed to mitigate the effects of low HL in participants with SLP and improve the use of health services or the health outcomes in these populations, d) were shared in an online format in indexed scientific journals, e) were written in English or Spanish, f) were published in 2011–2023, g) were randomized control trials (RCTs), pre/post (PP) studies, prospective cohort (PC) studies, cross-sectional (CC) studies, or mixed methods studies and h) measured effectiveness of intervention using HL assessment tools or health outcomes.

Exclusion criteria included studies of outcomes related to numeracy or literacy alone without reference to HL because such interventions were found to differ from those that dealt with these issues in the context of HL. We also excluded studies that did not report HL interventions targeting Spanish-speakers in the United States.

Assessment of methodology quality

We assessed the methodological quality of each included study using the Revised Tool to Assess Risk of Bias in Randomized Trials (RoB 2) [22] and the Risk Of Bias In Non-randomized Studies—of Interventions (ROBINS-I) tool for assessing risk of bias in the different interventions analyzed (RCTs, PP studies, PC studies, CC studies, and mixed methods studies) [23]. Two review authors (P.P. and L.D.) independently performed quality assessments. Disagreements regarding the overall assessments were resolved through discussion, with a third reviewer as the final arbitrator (J.H.). Bibliographies of included studies were subsequently hand searched.

Data synthesis

After piloting, four reviewers (J.H., L.A., P.P., L.D.) conducted data extraction using a standardized data extraction template (Appendix 1). Due to the heterogeneity of interventions, outcomes assessed, and varying durations of interventions, we did not pool the data and instead conducted a narrative analysis. We conducted a thematic analysis of identified studies and grouped studies for synthesis on the basis of identified categories. This process consisted of iterative discussions of the studies by all members of the study team and was based on published guidelines for Synthesis without Metanalysis (SWiM) [24]. Our data synthesis specifically grouped studies based on the categories of study characteristics, measures of effectiveness, reported effectiveness by intervention type, and quality assessment. We stratified the results by intervention type. While we did not focus on migrant status specifically, this could be estimated by one of our data extraction items, country of origin.

Results

Study characteristics

After removal of duplicates, 2,823 titles and abstracts were screened for inclusion using the criteria described above. A manual search of bibliographies yielded eight additional articles for screening. A total of 121 potentially relevant articles were selected using the inclusion criteria described above. After a detailed full-text analysis of each study, 62 studies were included, and 59 were excluded, as indicated in Fig. 1. This included 17 RCTs, 35 PP studies, 3 PC studies, 3 CC studies, and 4 mixed methods studies. A summary of the study characteristics can be found in Table 1, 2, and 3. The studies encompassed mainly female, middle-aged adults (range: 30 to 50); only two studies included participants under the age of 18 [25, 26] and no studies were focused solely on pediatric populations. Only a minority of participants had graduated from college. Sample sizes varied from 10 to 943. Interventions included in-person education (n = 28), multimedia education (n = 27) and other types of multimodal strategies (n = 7). Eighteen studies made use of lay health advisors and promotores.

Table 1 Summary of the study characteristics-in person education
Full size table
Table 2 Summary of the study characteristics-multimedia education
Full size table
Table 3 Summary of the study characteristics multimodal
Full size table

Topics included prenatal care and parent education; breast, cervical, colorectal, and ovarian cancer; diet and healthy lifestyle choices; mental health literacy; diabetes; cardiovascular disease; end-stage renal disease; asthma; upper respiratory infections; inflammatory bowel disease; HIV/AIDS; skin care; hearing loss prevention; medication understanding; palliative care; family health history; chronic pain; healthcare navigation; and anesthesia education. Thirty-four studies employed a theoretical framework when designing and conducting research, and there was little heterogeneity in terms of frameworks employed. No framework was shared by more than four studies.

Studies were performed in a variety of settings, including clinics (n = 13), hospitals and health centers (n = 13), Federally Qualified Health Centers (FQHCs) or safety net clinics (n = 9) and community spaces (n = 18). Common community settings, which include community health centers and safety net clinics, frequently used curricular interventions embedded in educational curricula and educational workshops (n = 18). Larger hospital networks implemented organizational interventions, often updating their practices or replacing standard-of-care materials with language and culturally concordant materials (n = 8).

Measures of effectiveness

The measures of successful enhancement of HL used by the studies in our review were heterogenous, and were often unvalidated measures of knowledge or beliefs. Twenty-two studies had a questionnaire about beliefs, knowledge or practice that was developed by the researchers, limiting the validity of their results. Fifty-eight studies measured effectiveness quantitatively, and four were mixed methods. The two most common approaches to primary outcomes were either HL assessment tools [16, 17, 25, 25, 26, 28, 30,31,32, 34,35,36,37,38, 41, 47, 49, 50, 53,54,55,56,57,58,59,60,61, 63,64,65,66,67,68,69,70,71,72, 74,75,76, 78,79,80,81,82,83] (n = 45) or health outcomes [27, 29, 39, 42, 48, 51, 52] (n = 7), with some studies using both [15, 43, 46, 84,85,86,87,88,89,90] (n = 10). HL tools most commonly took the form of pretest/posttest questionnaires specifically developed by the researchers to assess knowledge gained over the course of a given intervention. A few studies (n = 10) utilized previously validated disease-specific assessments of HL, such as the High Blood Pressure-Health Literacy Scale for high blood pressure [32], or more standardized Test of Functional Health Literacy in Adults (TOFHLA) [31, 37, 38, 49, 50, 52, 55, 60, 68, 73] (n = 10) and/or Newest Vital Sign (NVS) [28, 43, 51, 52, 68, 76] (n = 6), to assess overall changes based on the participant’s ability to read and understand generic health-related materials.

Other outcome measures included patient satisfaction and patient attitude surveys, which were intended to predict not only knowledge of health conditions but also attitudes toward receiving treatment [84]. Higher satisfaction and improved attitude scores were thought to lead to a more positive and confident approach in obtaining healthcare. Some studies measured improvements in confidence and self-advocacy [31, 34, 70]. Medical health measurements and outcomes, such as blood pressure readings, were also commonly used as primary outcomes [27, 46, 48, 74]. Secondary measures were also varied and included measures of patient confidence, perceived support, perceived barriers to care, level of comfort, and adherence to the intervention.

Studies also varied in how they measured the long-term changes associated with their interventions. Thirty-one studies had a follow up of at least a month, ranging from 1 to 24 months, with most studies doing a 1 month follow up (n = 7) or a 3 month follow up (n = 8).

Overview of health literacy interventions

In-Person Education

In-person education health literacy programs varied in presentation of material but shared commonalities of repeated meetings in a class setting that encouraged practice and facilitated opportunities for enhanced participant engagement compared to other modalities (Table 1). A study by Cruz [30], found the use of 90 min training session conducted by promotores focusing on general knowledge for diabetes, risk factors, and prevention and control of diabetes provided significant improvement on diabetes knowledge for diabetic participants comparing pre- and posttest scores (13.7 vs. 18.6, p < 0.001; Cohen’s d = 1.2), and for nondiabetic participants (12.9 vs. 18.2, p < 0.001; Cohen’s d = 1.2).

Similarly, Buckley [27] assessed the implementation of social clubs hosted by navegantes (patient navigators) for 2 h every week over 5 weeks. The findings suggested 88.9% of 126 participants increased health literacy and over 60% decreased at least one risk factor associated with metabolic syndrome. Change for those that improved, [mean (SD)]: Weight [− 6.0 lbs (5.2)]; BMI [− 1.1 (1.0)]; Waist Circumference [− 2.2 inches (1.5)]; Blood Glucose [− 26.3 mg/dl (27.5)]; LDL Cholesterol [− 19.1 mg/dl (16.8)]; Systolic BP [− 11.1 mmHg (9.5)]; Health Literacy Test (n = 117) [+ 22.2% (19.7%)]. Castaneda [28] studied the implementation of 6-week, culturally tailored, promotora-based group for health prevention knowledge and found participants improved their self-reported cancer screening, breast cancer knowledge (Mpre = 2.64, Mpost = 3.02), daily fruit and vegetable intake, and ability to read a nutrition label (p < 0.05).

Across all the different in-person education there were common findings that repeated exposure to health education information in an engaging classroom setting provided meaningful improvements to health literacy in SLP populations that correlated with improvements in physical health and greater utilization of health screening services.

Multimedia education

Multimedia approaches to health literacy education varied from narrative films and fotonovelas to animated culturally sensitive videos and virtual workshops to assess applied knowledge (Table 2). The commonality shared with these interventions were that they could largely be independently navigated without need for transportation or cost to the participant as long as they had access to a computer and the internet.

A study related to health literacy in women’s health, Borrayo [53] found that through a 8-min narrative film to reinforce desired self-efficacy and behavioral intentions as precursors to engaging in mammography screening there was a significant increase in breast cancer knowledge ( Wilks’s Λ = 0.75, F(1, 39) = 13.15, p < 0.001, η2 = 0.25) and mammography self-efficacy ( Wilks’s Λ = 0.76, F(1, 37) = 11.64, p < 0.01, η2 = 0.24) compared to baseline and control group. Furthermore, Cabassa [54] assessed the use of a fotonovela centered around entertainment-education intervention toward mental health stigma finding a significant increase in depression treatment knowledge scores at posttest ( B = 1.22, p < 0.001, Cohen’s d = 0.91) and 1-month follow-up ( B = 0.81, p < 0.01, Cohen’s d = 0.53). Calderon [55] looked at the implementation of an animated, culturally sensitive, Spanish video to improve diabetes health literacy (DHL). The findings reported DHL survey scores improved significantly more in the experimental group than the control group (adjusted mean = 55% vs 53%, F = 4.7, df = 1, p = 0.03). Additionally, Cheney [56] studied the application of tailor MyPlate recipes to local food sources and culture, virtual cooking demonstrations, and Spanish cookbook, on diabetes education finding there was an increased confidence in adherence to two of four components of the Mediterranean diet (badded sugar = 0.24; 95%CI: 0.02, 0.46; bredmeat = 0.5; 95% CI: 0.02, 0.98).

Other types of multimodal strategies

Multimodal strategies provided a crossover between in-person and multi-media focused health literacy approaches (Table 3). A study by Auger [15], found the use of fotonovelas as an educational tool along with health education facilitation by the teacher and lay health educator provided an increased knowledge of pregnancy, childbirth, and breastfeeding (p < 0.001) and confidence in navigating pregnancy, caring for oneself and the baby, and interacting with health professionals (p ≤ 0.05).

Additionally, Calderon [78] took a multimodal approach to mental health education via workshops including a short video on possible psychotic and depressive symptoms, La CLAve mnemonic device to describe the main symptoms of psychosis, and a narrative film to discuss its portrayal of symptoms. That study demonstrated a significant increase in psychotic symptoms reported as definition of serious mental illness (pre, M = 0.69, SD = 0.61; post, M = 1.23, SD = 0.90, t(80) = − 5.64; p < 0.001; Cohen's d = 0.70) and ability to detect a serious mental illness in others (pretraining: M = 2.83, SD = 1.31; posttraining: M = 3.24, SD = 1.27, t(74) = − 2.76, p < 0.05; Cohen's d = 0.32), and decrease in participants' recommendations for nonprofessional help-seeking (pre: 49.4%, post: 25.9%, N = 81, p = 0.001). There was no significant change in recommendations for professional help (pre: 64.2%, post: 72.8%, N = 81, p = 0.25).

Reported effectiveness by intervention type

Of the interventions, 89% of in-person educational interventions (n = 25) and 89% of multimedia educational interventions (n = 24) found improvements to HL. All multimodal interventions (n = 7) provided improvements in HL. The use of lay health advisors and promotores was correlated with increased effectiveness; all 18 studies that used this technique reported that their interventions had caused statistically significant changes in HL [27, 28, 34, 46, 70]. Similarly, all nine of the studies implementing fotonovela strategies reported statistically significant improvements in HL [16, 25, 53, 54, 61, 63, 74].

Quality assessment

The risk of bias assessment for RCTs evaluated risks due to randomization, outcomes, and result reporting (Table 4). Among RCTs (n = 17), one was assessed as having a high risk of bias, and eight were assessed as having some concerns. Non-RCTs were likewise evaluated for risk of bias due to problems with recruitment, confounding factors, missing data, and selective measurement of outcome or result reporting (Table 5). Among non-RCTs (n = 36), 14 studies had a serious risk of bias, while the remaining 22 studies had a moderate risk of bias.

Table 4 Risk of bias summary for randomized studies
Full size table
Table 5 Risk of bias summary for nonrandomized studies
Full size table

Discussion

To the best of our knowledge, this review is the first to systematically describe and evaluate the effectiveness of HL interventions among patients with SLP in the United States. Recent reviews have studied the impact of different intervention strategies for increasing the HL of the general population [85] and for immigrant communities [18] but have not focused on Spanish speakers – a community largely at risk for low HL and poor health outcomes [5, 11, 17].

Our review found that, as with other populations with a non-English language preference, including migrant populations [86], there is a lack of evidence-based specific interventions to raise HL tailored to U.S. patients with SLP. Further, our review found that the few existing studies may be at risk of bias. The high risk of bias we found especially in non-RCTs on this topic likely represents both the lack of attention to research addressing this need in SLP populations, as well as difficulties inherent in testing and measuring interventions aimed at improving HL more broadly. Our review of quality was in line with other reviews on this topic [18, 87] which found that a risk of bias was introduced primarily due to difficulty blinding participants and moderators due to the nature of study designs. This made RCTs more difficult to conduct, and as a result, studies primarily used pretest/posttest and cross-sectional methodologies. This finding points not only to a need for high-quality studies of HL in this population, but also for the potential to critically rethink how to conduct research on HL in a high-quality, low risk of bias way. Additionally, studies reported sample sizes ranging from 10 to 943 participants, which made it difficult to compare effect sizes directly. This variation likely reflects the dissimilarity of study designs, sample populations and setting types, thus making it difficult to compare across studies, a challenge that has been previously acknowledged for reviews of HL.

There are also significant differences in patient populations across reviews, and many studies had a low number of participants. This small sample size was in some cases due to strict study inclusion criteria, and other cases were due to high rates of attrition. This could be partly due to primarily targeting participants already facing cultural, socioeconomic, and educational barriers, making them more difficult to recruit and retain in research. Many studies have indicated that their sample may not be representative due to sampling methodologies or that there may be limited generalizability of results. This was due in part to convenience sampling or small sample sizes, which made it difficult to determine whether findings represented a true effect due to limitations in statistical power.

Some studies were focused on only one research site and/or a highly specific Hispanic immigrant community with a SLP (i.e., Mexican immigrants [52]), limiting generalizability. At the same time, while we attempted to capture the difference between the broader category of Spanish speaking populations in the United States and specific migrant populations, most studies did not include this sort of information, indicating a potential need for studies that focus on specific SLP migrant communities. No studies addressed pediatric populations. Another factor limiting the generalizability of the reviewed studies was that the majority of study participants were women; this may be tied to a wider lack of healthcare utilization among Hispanic men, including those with a SLP [88,89,90]. The relative paucity of males in the sample population of the studies may indicate a need for research that focuses on men with a SLP. To date, only a few strategies have been developed to include males with a SLP in research, including the use of male community health workers and health outreach in workplaces and providing public transportation [41]. Finally, the studies reviewed included a predominantly adult to middle-aged population (aged 30–50) rather than older adults who are more at risk for serious medical problems. This suggests that several important populations (men, children, older adults) may be missed by most previous HL interventions in populations with SLP.

The varied, poorly standardized, heterogenous measures used to assess HL in reviewed studies demonstrate that HL as a concept is poorly defined by researchers, and the concept likely encompasses more than can be quantified by numeric scores on standardized assessments of knowledge. For example, in assessing HL, there may also be a need to address the ability of patients to advocate for themselves, ask questions, and feel empowered to change their health behaviors [8, 9, 16, 74, 85]. Existing measures of HL may not fully capture HL concepts, and thus may be a poor proxy of effectiveness. Studies in our review often used measures that were not validated and tested knowledge on a specific health topic or reported beliefs about health as proxies for HL, and relatively few measured direct behavioral changes, attempts at communication self-efficacy or advocacy, or effects on health outcomes. A key takeaway of this review is the need to critically reexamine definitions and measures of HL, and to develop and validate improved qualitative and quantitative measures of HL outcomes. Only about half of the studies used a theoretical framework to inform their intervention or research, and studies rarely employed the same frameworks, perhaps partially accounting for the variety of measures and the limitations in the ways that HL was framed by researchers.

We found that studies of HL among people with SLP in the United States therefore followed trends within the literature, in which HL is measured through knowledge of health/healthcare systems; to a lesser extent, studies included in our review also attempted to measure participants’ use of information related to health and healthcare, and their ability to maintain health through collaboration with health providers [7]. When framed in terms of Nutbeam’s health outcomes approach, the studies mostly attempted to measure functional HL, occasionally addressed communicative/interactive HL, and rarely attempted to address critical HL [8, 9]. This focus on HL as knowledge rather than personal health advocacy has important ramifications in terms of the skills that HL interventions focus on building, and may help to explain the success and failure of HL interventions.

In addition to the importance of improving individual’s health literacy there is support in the literature to improve “organizational health literacy.” Organizational health literacy refers to the responsibility for health care systems to address populations with low health literacy [91]. Methods for organizations to address populations with low health literacy include “reducing the complexity of health care; increasing patient understanding of health information and enhancing supports for patients of all levels of health literacy” [91]. Because limited health literacy has been associated with increased cost of healthcare organizations have an incentive to address health literacy. However, few if any of the studies attempted to address organizational health literacy, and placed the onus for building HL on the individual patient and their family.

Specific recommendations

Successful interventions focused on HL interventions that targeted SLP populations through linguistically and culturally concordant techniques that utilized community member liaisons and culturally relevant storytelling. Successful interventions were also often well integrated within communities and organizations.

Our review found that interventions utilizing cultural and linguistic concordance (ie. Spanish-language, culturally salient concepts/terms), liaison roles (promotores), and narrative media were effective in achieving notable improvements in HL among patients with SLP. These interventions focused more on what Nutbeam frames as communicative/interactive HL [8, 9]. The relative success of these interventions may be due to more effective communication with patients through a shared cultural background and deeper levels of trust. One particularly effective strategy is the use of narrative in media, as seen with fotonovela strategies [16, 25, 53, 54, 61, 63, 74]. Such strategies may involve a video or booklet presenting important health information in a story format. Narrative media appeared to activate study participants and lead to improvements in health knowledge and behavior change. Another important element of effective multimedia health interventions is cultural adaptation to address previously identified cultural concepts such as respeto, familismo, marianismo, and personalismo [16, 25, 55, 62, 65, 68, 73, 76]. Realistic stories with Spanish-speaking characters and culturally tailored information were key components of these interventions [25, 54, 55, 74]. Prior research has shown that identification with storytellers is an important prerequisite for patient engagement and is particularly useful in combating cultural stigma and eliciting behavioral health changes [92].

Liaison roles that employ educators and health promoters from similar cultural backgrounds as patients were also an important strategy used by reviewed studies. The lived experience and cultural understanding from these workers (promotores, navegantes, community health workers) may help boost patient comprehension and overcome distrust of the healthcare system [93]. Linguistic and culturally concordant care, including cultural competency training for providers, has also repeatedly been identified as a successful strategy for increasing HL among immigrant populations generally [18]. Furthermore, successful interventions often consider the opinions of the target population when designing content to ensure that the experience is culturally relevant [16, 28, 29, 34, 35, 40, 44, 45, 55, 62, 65, 73, 76].

Our review also included a number of multimedia intervention strategies (n = 22) that might be utilized more often in the future following the increased acceptance of online options since the COVID-19 pandemic. Interestingly, of the 14 studies published since 2020, seven were multimedia interventions. Our search also revealed the importance of including nonmedical settings such as community gathering spaces, which may serve as a hub for creating a wider network of health promotion. The integration of health promotion interventions into communities may be complimentary to the long-term reinforcement of health education, serving as a means of achieving sustained outcomes.

Other elements of successful HL interventions may include finding a fit between factors such as intervention type, size and type of setting, duration of time available, and level of community integration (Fig. 2). As described above, HL may be framed as organizational as well as individual, and successful interventions better integrate organizational setting into the structure of the HL intervention. We refer to community integration as the level of incorporation of community resources, stakeholders such as promotores, and settings into interventions aimed at improving health literacy, concepts drawn from the literature [94, 95]. These categories of community integration were inferred from the setting type since we expect large hospitals to be less involved in community initiatives than community clinics or community settings (i.e. local churches) themselves. Smaller, community-based settings and nonmedical centers such as churches and college campuses seemed to be more successful with implementing multiweek curricula interventions. This may be because these settings have the infrastructure in place for recruitment and retention of community members with a SLP. Larger hospital systems and clinics with less time and resources available may be better able to focus on culturally and linguistically concordant patient materials and replace standard of care materials written in English with multimedia health information. These recommendations are illustrated in Fig. 2, which displays fit between intervention type and setting.

Fig. 2
figure 2

Recommendations for intervention models by common combinations of setting type, duration available, and community engagement

Full size image

The findings of our review are also relevant to studies of HL in other populations with a non-English preference, including minority, migrant or immigrant populations. Previous reviews of HL interventions did not include studies measuring HL indirectly through variables such as health outcomes or behavioral change but only included those using standardized tools [85]. However, as these standardized assessment tools are available predominantly in English, this approach may limit the generalizability of past reviews to non-English speaking populations. A growing body of evidence suggests that a reframing of our understanding of HL, especially among marginalized communities, is necessary to improve health equity [2].

Finally, our review highlights a need for additional attention to the development and adaptation of HL interventions for patients with SLP in the United States. Policies promoting HL interventions may need to better address the needs of specific populations through research and the widespread promotion of effective strategies.

Limitations

A limitation of our review is that all studies were conducted in the United States, which limits the generalizability of our findings to healthcare systems in other countries. It should be noted that we did not explore gray literature. We also chose to limit our review to studies that took place after 2010, preventing a fuller historical examination of HL literature. Finally, we could not conduct a meta-analysis due to variability in design and measurement.

Conclusion

There is a small but growing body of literature that addresses the need for HL interventions among individuals in the United States with SLP. However, there is no consensus around strategies to improve or tools to assess HL, and studies vary greatly in quality and risk of bias. Important target populations, such as children, older adults and men, may be excluded from this research. Strategies that incorporate linguistic and cultural factors particular to this population, such as fotonovelas and health promoters from similar cultural backgrounds, may be of use in promoting HL. There is a need for improved research and policy on HL interventions specifically targeting this population.