Interventions for End of Life Decision Making for Patients with Limited English Proficiency

Abstract

Background:

Patients with limited English proficiency (LEP) experience disparities in end-of-life decision making and advance care planning. Our objective was to conduct a systematic review to assess the literature about interventions addressing these issues.

Methods:

Our search strategy was built around end-of-life (EOL), LEP, ACP, and goals of care. The databases included Ovid MEDLINE(R), and Epub Ahead of Print, In-Process & Other Non-Indexed Citations and Daily from 1946 to November 9, 2018, Ovid EMBASE.

Results:

Eight studies from the US and Australia were included (7 studies in Spanish and 1 study in Greek and Italian). Interventions used trained personnel, video images, web-based programs, and written materials. Interventions were associated with increased advance directive completion and decreased preferences for some life-prolonging treatments. Interventions were deemed to be feasible and acceptable.

Conclusion:

Few interventions exist to improve end-of-life care for patients with LEP. Data are limited regarding intervention effectiveness.

Introduction:

The population of people within the US with limited English proficiency (LEP) continues to rise and increase in age.[1] Those with LEP have difficulties understanding and speaking English and face unique challenges when interacting with the healthcare system as well as accessing it.[2–4] Those with LEP experience higher rates of adverse health outcomes and perceive that the quality of their care is worse than those who speak English.[4–15] These adverse events are more likely to incur serious harm such as longer hospital stays, higher rates of readmission, and poor understanding of discharge instructions. The adverse events occur in pediatric and adult populations, and within in-patient and outpatient settings.[16–18] These differences may be directly correlated with their LEP status, health literacy and access to appropriate healthcare options as well as cultural factors.[15, 19]

There is mounting evidence that decision making and care at end of life among those with LEP differs substantially from the population that speaks English. These differences include prolonged ICU stays, death in the ICU, increased and aggressive interventions at end of life, increased use of restraints, lower rates of comfort measures only, and do-not-resuscitate orders despite imminent death.[20, 21] As with other minority and vulnerable groups, uptake of advance care planning (ACP) is also lower.[22–25] ACP is a process providing people with an opportunity to understand and share their values, goals, preferences, and wishes about future medical care.[26] The cause of these disparities is multifaceted. However, it is essential to develop healthcare interventions for end of life care and ACP for patients with LEP. These interventions must be linguistically tailored and incorporate culturally sensitive communication strategies. This approach has the potential to improve decision making at end of life and adoption of ACP.

Since the 2015 Institute of Medicine’s (now the National Academy of Science) report “Dying in America,” there have been efforts to increase public engagement in end of life issues, including ACP. However, minorities may not be experiencing the potential benefits.[27, 28] These efforts extend to promoting practices and policies that can achieve higher quality end of life care.[29] The introduction of the Current Procedural Terminology (CPT) codes in 2016 gave clinicians a mechanism to claim reimbursement for ACP discussions with Medicare and Medicaid patients.[30]

Unfortunately, recently released data about the use of these codes suggests that the use of the CPT codes is low overall. Furthermore, among Hispanic and Asian populations, the use of these codes is even lower.[31–33] Although the language proficiency of these populations was not recorded, the data suggests the need for a different approach for minority populations who may have linguistic, health literacy, and cultural differences that influence their use of ACP. Strategies to improve the process and experience of complex medical decision making in LEP patients with serious illnesses are also lacking. Although the use of interpreters to navigate the healthcare system is mandated for patients with LEP, interpreters are frequently not used and family members serve as interpreters, further complicating challenging discussions [34, 35].

The objective of this systematic review is to evaluate existing interventions whose purpose is to improve end-of-life decision making and ACP for populations with LEP. We will determine the types of interventions created and used, assess the types of populations in which the interventions were used, describe the types of decision -making for which the interventions were used, and assess the success of the interventions used. This systematic review will inform future approaches to address disparities in complex medical decision making among LEP patients at end of life as well as disparities in rates of ACP.

Methods:

Registration

The systematic review was registered with the International Prospective Register of Systematic Reviews (PROSPERO, CRD42019123021 https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=123021). No Institutional Review Board approval was sought since this was secondary analysis of published work.

Data Sources and Search Strategy

Using controlled headings and text terms, we designed a search strategy built around end-of-life (EOL), LEP, ACP, and goals of care. The databases included Ovid MEDLINE(R), and Epub Ahead of Print, In-Process & Other Non-Indexed Citations and Daily from 1946 to November 9, 2018, Ovid EMBASE. The search strategy was designed and conducted by an experienced librarian (PE) with input from the study’s principal investigator (AB). We did not find additional articles with manual search strategies using reference lists of selected studies. The search provided 4429 articles after removing duplicates (Prisma Diagram Figure 1).

Figure 1.

Study Selection

The review process was structured using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist.[36] Each title, abstract and full-text article was reviewed independently by two reviewers (AB, JS, NES, BB, KC, SK, HK, BV, BK). Studies that met all of the following criteria were included: 1) study participants were aged 18 years and older, 2) the study was conducted in an English-speaking country where not speaking English is a barrier (included USA, Canada, UK, Ireland, Australia, New Zealand), 3) LEP included “English as a second language,” 4) participants required translators or interpreters, or identified as “recent immigrant.” or having ”language barriers.” Studies were excluded on the basis of one or more of the following: 1) case reports, 2) case series, commentary, editorial or review articles, 3) empirical studies, 4) non-English articles, 5) publication as an abstract only, 6) studies that described an intervention, but did not test the intervention in patients, 7) studies that tested the intervention empirically or in simulation (not real world); and 8) conference reports and/or presentations. Conflicts were adjudicated by a third reviewer in each phase. (Prisma Diagram Figure 1)

Data Extraction

Data abstraction on each of the 8 included articles was performed by two independent reviewers (BB, KC, SK, HK, BV) using a tailored data abstraction form which was piloted beforehand. Appraisal of data abstraction was performed by senior study team members (AB, JS, NES, BK) and any disagreements were addressed. Data regarding participants’ demographic characteristics, study design, study setting, and country, type of intervention and type of decision making used for, were abstracted.

Quality Assessment/Appraisal

To assess the quality of included studies we used the Cochrane risk-of-bias assessment tool for randomized clinical trials (RCT).[37] This was done during the data abstraction phase by two reviewers independently and in duplicate. Discrepancies were resolved by senior team members. This tool assesses the random sequence generation, allocation concealment, blinding of participants, blinding of outcome assessment, incomplete outcome data, and selective reporting. For non-RCT studies, we used the Newcastle-Ottawa scale for risk of bias assessment.[38]

Data Synthesis and Analysis

Due to the small number of studies and the diverse interventions and outcomes assessed in the included studies we were unable to conduct a meta-analysis. We descriptively summarized and qualitatively synthesized the data. The principal summary measures were actual rates of ACP, Advance Directives (AD), or other documentation including surrogate decision-making preferences, attitude, and comfort with ACP as well as decisions about hospice and quality of life.

Results:

Study Characteristics

The search strategy identified 4443 studies with no additional studies identified through manual searches. Prisma Diagram Figure 1.shows the study selection process and reasons for exclusion. A total of 8 studies evaluating 1846 patients were included.

Description of Eligible Studies

Seven of the eight studies were conducted in the US and 6 of those included interventions in both English and Spanish.[39–44] One study was done in Australia and included interventions for Greek and Italian speaking populations.[45] Five of the included studies were multi-center and three of the included studies were single center. Two of the studies were observational studies with either a pre-/post-test or cross-sectional design.[45, 46] The remaining six studies were randomized control trials. Four of the interventions addressed advance care planning specifically,[42–45] two primarily addressed palliative care and hospice use and ACP,[39, 40] and two focused on end of life care and decision making.[41, 46] Types of interventions included the following: one used video images[46], five used trained personnel to enhance current practice,[39–42, 45] one used a patient-facing web-based intervention,[44] and another used redeveloped paper forms for AD completion.[43] Of note, three studies were described as feasibility studies.[39, 42, 45] Table 1 provides a description of the included studies and participant characteristics.

Table 1:

Eligible Study and Participant Characteristics

Reference	Country multi or single center	Study design and population	Language	N=total or Spanish speaking /total	Age, years mean(SD)	Male, n (%)
Volandes1 2008	USA, Multi	Observational pre/post study in Spanish speaking adults >40 at a primary care clinic	Spanish	104	55(10)	11(10)
Fischer2 2015	USA, Single	Pilot feasibility RCT in Spanish speaking adults with a life-limiting illness at home	Spanish/English	64	57.4(13.5)	22(34)
Fischer3 2018	USA, Multi	RCT in English and Spanish speaking Latinos with stage III/IV cancer at home	Spanish/English	223	58.1(13.6)	99 (44)
Heyman4 2010	USA, Multi	Randomized 3 arm group design trial in English and Spanish speaking elderly Latinos	Spanish/English	84	73.5(10.6)***	5(17)***
Nedjat-Haiem5 2017	USA, Multi	Feasibility Study as part of a pilot randomized mixed methods study in Latinos ≥ 50 with serious illness	Spanish/English	68	65.8(8.7)	16(24)
Sudore6 2007	USA, Single	RCT of English and Spanish speaking adults ≥ 50 in urban, general medical clinic	Spanish/English	60/205	61.9 (9.0)***	57 (56)***
Sudore7 2018	USA, Multi	RCT of English and Spanish speaking older adults with chronic illness in primary care clinics	Spanish/English	445/986	64(7.2)**	63(27.9)**
Detering8 2015	Australia, Single	Cross-sectional feasibility and acceptability study of Greek and Italian speaking inpatients ≥ 65 (non-randomized)	Greeks/Italians/English	112(total)	82 (11)*	19 (39)

Baseline characteristics of participants are reported for full cohort unless otherwise noted

^*median (IQR)

^**Control arm (easy to read AD only) Spanish speaking participants

^***Control arm only, may include both languages

RCT=Randomized Clinical Trial

Participant characteristics

Most participants were Spanish and/or English speaking except for those in the Detering study based in Australia where the participants spoke Greek/Italian and /or English. This reflects the different demographics of the Australian immigrant population.[45] The mean age of participants in all studies was 55 or older. Four out of eight studies included participants with serious or chronic illness, including advanced cancer.[39, 40, 42, 44] Two studies included healthy participants during routine visits.[43, 46] One study recruited hospital inpatients[45] and one study recruited those needing some agency care at home[41]. Most studies recruited more women than men except for the Sudore 2007 study that included just over half male participants.

Outcome Measures

Overall, the interventions were successful in increasing ACP, ADs and other similar documentation including living wills and designation of a surrogate decision maker. However, the small number of studies that measured comfort and attitude with end of life and hospice use had mixed results. Table 2 summarizes the study outcomes.

Table 2:

Intervention Outcomes

Reference	Interventions	Study design and population	Clinical outcomes
Volandes1 2008	2 minute video in Spanish depicting a patient with advanced dementia	Observational pre/post study in Spanish speaking adults at a primary care clinic	Significant Increased preference for “comfort care:” 40% (pre) vs 75% (post), p=0.001 Decreased preference for “life-prolonging care”: 41% (pre) vs 8% (post), p=0.001 Preference for “limited care”: 11% (pre) vs 13% (post), p=0.001 Unsure of preference: 8% (pre) vs 4% (post), p=0.001
Fischer2 2015	Group 1: Home visits (up to 5) from a bilingual (English/Spanish), bicultural patient navigator. Home visits addressed barriers to ACP and hospice use. Participants also received a linguistically matched and culturally tailored packet of educational materials on palliative care, including ADs, a study-specific AD form, hospice care, and pain management. Group 2: Control group received a linguistically matched and culturally tailored packet of educational materials on palliative care only including ADs, a study-specific AD form, hospice care, and pain management.	Pilot feasibility RCT among Latinos who speak English or Spanish with a life-limiting illness at home	Significant c Increased study pecific AD “Apoyo con Cariño” form in medical records at 12 months: 22% vs 0%, p=0.01 Nonsignificant Any form of AD(MDPOA, AD, living will) in medical records at 12 months 47% vs 25%,p=0.06 Completed MDPOA in medical records at 12 months:38% vs 19%,p=0.1 Hospice enrollment (decedents) before death: 7 vs 6, p=1 Hospice length of stay before death (days): 36.4(51.6)vs 19.7(33.6),p=0.39 Hospice stay of at least 7 days before death: 71% vs 33% (p= 0.15)
Fischer3 2018	Group 1: Home visits (at least 5) from a bilingual (English/Spanish), bicultural patient navigator. Home visits addressed barriers to ACP and hospice use. Participants also received a linguistically matched and culturally tailored packet of educational materials on palliative care, including ADs, a study-specific AD form, hospice care, and pain management. Group 2: Control group received a linguistically matched and culturally tailored packet of educational materials on palliative care only including ADs, a study-specific AD form, hospice care, and pain management.	RCT among Latinos who speak English or Spanish with stage III/IV cancer at home	Significant d Increased any form of AD documentation in medical records at 6 months: 65% vs 36%,p<0.001 Increased study-specific AD “Apoyo con Cariño” form in medical records at 6 months: 46% vs 0,p=NA Increased living will documented in medical records at 6 months:40% vs 13%, p<0.001^ Increased MDPOA documented in medical records at 6 months:62% vs 25%,p<0.001 Increased reported discussion about future health care preferences with family at 3 months: 84% vs 52%,p<0.001 Increased reported discussion about future health care preferences with health care providers at 3 months: 60% vs 35%,p=0.001 Nonsignificant McGill Quality of Life Questionnaire score at 3 months: 7.7 (1.9) vs 7.2 (2.1),p=0.14 McGill Quality of Life Questionnaire score change from baseline at 3 months: +0.5(2) vs +0.04(1.7),p= 0.08) Proportion who received hospice care before death: 80% vs 84%,p=0.58 Hospice length of stay(days)before death: 54.7(103.4) vs 58.7 (78.8),p=0.63
Heyman4 2010	Group 1: One on one conversation with bilingual medical social worker in Spanish at home about ADs importance of ACP, and healthcare proxies, designed to influence attitudes towards and comfort with end-of-life planning (Conversacion A) Group 2: One on one conversation with bilingual medical social worker in Spanish at home including topics in Conversacion A plus discussion of additional topics, designed to influence attitudes towards and comfort with end-of-life planning (Conversacion B) Group 3: Control received standard information routinely provided in Spanish only	Randomized 3 arm group design trial for Spanish-speaking elderly Latinos (intervention delivered in Spanish only)	Significant Improved attitude towards end-of-life planning (groups 1 and 2 compared to 3),p<0.001 Group 1 (Conversacion A)attitude scores: 36.63* (4.48) Group 2: (Conversacion B)attitude scores: 36.72 (2.97) Group 3: (Control) 32.69 (2.98) Improved comfort with end-of-life planning (group 1 compared to group 3), p<0.01 Group 1: (Conversacion A) comfort scores: 20.57 (3.10) Group 2: (Conversacion B) comfort scores: 18.93 (2.55) Group 3: (Control) comfort scores: 18.03 (3.10) Note Conversacion A improved attitudes and comfort Conversacion B only improved attitudes, not comfort
Nedjat-Haiem5 2017	Group 1: Advance care planning-Intervention (ACP-I) received the general AD education including standard verbal and written ACP education (designed for informational and communication needs to increase ADs), as well as 1 counseling session by a social worker Group 2: Usual Care group received the general AD education including standard verbal and written ACP education	Feasibility Study as part of a pilot randomized mixed methods study in Latinos ≥ 50 years old with serious illness who spoke English and/or Spanish	72% enrolled in study e Completion rate of post-test surveys: 92% vs 92% 100% of participants reported having a positive experience with ACP-I plan ACP-I considered feasible to implement and acceptable as a program No comparisons were made with standard ACP education^^
Sudore6 2007	Group 1: Redesigned written advance directive (5th-grade reading level with graphics) Group 2: Standard written advance directive (> 12th-grade reading level)	RCT of English and/or Spanish speaking adults ≥ 50 years old in urban, general medical clinic	Significant (includes English and Spanish speaking) Increased new advance directive completion at 6 months: 19% vs 8%,p = 0.03 Increased ease of use and understanding 69<0.001Increased personal usefulness in treatment decisions and discussions 89% vs 76%,p=0.001 Increased general value in care planning 86% vs 79%,p=0.03 Increased preference for use: 73% (redesigned AD) vs 17% (standard AD) vs 10% (no preference) OR preferring redesigned vs standard 4.25(95%CI 2.93–6.34)-no differences by group assignment, literacy or language Significant (Spanish speaking only)*** Increased ease of use and understanding 70% vs 49%,p<0.001 Non-Significant (includes English and Spanish speaking) Knowledge about advance directives: 71% vs 71%,p=0.30 Non-Significant (Spanish speaking only)*** Increased usefulness in treatment decisions 90% vs 83%,p = 0.25 Value in care planning:87% vs 86%,p=0.81
Sudore7 2018	Group 1: PREPARE (an online module about ACP) plus an easy-to-read written advance directive Group 2: Easy-to-read written advance directive only (Control)	RCT of English and/or Spanish speaking older adults with chronic illness in primary care clinics	Significant**** Increased overall documentation of ACP at 15 months in medical records: 39% vs 24%, p=0.002 (adjusted for health literacy, baseline documentation of ACP, clustering by physician) Increased engagement in all ACP activities ( composite measure of behavior change and action scores)at 12 months: 97% vs 84%,p<0.001 Increased documentation of ACP at 12 months: 98% vs 81%,p<0.001 Increased discussions about ACP at 12 months: 99% vs 87%,p<0.001
Detering8 2015	Group 1: Greek and Italian speaking patients with use of interpreter and facilitator to do ACP using Respecting Choices model Group 2: Greek and Italian speaking patients without use of interpreter to do ACP (facilitator only) using the Respecting Choices model Group 3: English speaking patients using facilitator to do ACP using “Respecting Choices” model	Cross-sectional feasibility and acceptability study of Greek and Italian speaking inpatients ≥ 65 (non-randomized)	Significant: (Group 1 vs Group 2) Increased Advance care directive completion: 86% vs 52%, p<0.005 Increased Advance care directive completion - including wishes about other decision-maker and documentation of wishes: 50% vs 19%, p=0.02 Non-Significant: Groups (1 and 2 vs group 3) Accepted discussion about ACP: 96%vs 98%,p=0.58 Advance care directive completion: 60% vs 56%, p=0.85

AD = advance directive; ACP=advance care planning; MDPOA=medical durable power of attorney; RCT= randomized controlled trial; OR=odds ratio; CI=confidence interval

^*mean (standard deviation)

^{^}a living will is one type of advance directive that takes effect when the person is terminally ill

^coutcome details about those who received the intervention in Spanish versus English are not available, although the proportions of those speaking Spanish in each group was similar, about 30% spoke Spanish in total

^doutcome details about those who received the intervention in Spanish versus English are not available, although the proportions of those speaking Spanish in each group was similar, about 50% spoke Spanish in total

^eoutcome details about those who received the intervention in Spanish versus English are not available, although the proportions of those speaking Spanish overall was about 50% in total

^***Separated into those receiving intervention in Spanish only: n=33/103 received redesigned AD, n=22/102 received standard AD

^****All results reported separately for Spanish speaking only n= 445/986, PREPARE Arm=219, AD-only Arm=226

Volandes’ study showed that video images could overcome educational barriers to change end of life decision making and reduce preferences for life prolonging care and increase preferences for comfort care (p=0.001).[46] Both Fischers’ studies demonstrated improved ACP in the intervention groups who received “Apoyo con Cariño (Support With Caring)”compared to usual care (p=0.01,p<0.001). Neither study had effects on enrollment for hospice or length of stay in hospice.[39, 40] Sudore’s studies established that a literacy appropriate ACP intervention and a web-based intervention could both increase ACP completion among Latinos(p=0.03,p=0.002).[43, 44] Detering’s study demonstrated improved AD completion with the use of an interpreter in conjunction with a ACP-facilitator for Greek and Italian speaking populations when compared to no interpreter(p<0.005).[45] Nedjat-Haiem’s study directed towards improved ACP uptake showed the intervention was feasible however the study did not compare outcomes to standard care or a control.[42] Heyman’s study examining comfort with and attitude about end of life suggested that moderate levels of discussion may be better than discussion that is too intense(p<0.01).[41] In general, interventions were deemed to be feasible and acceptable. Due to the diverse outcome measures and intervention types we were unable to conduct a meta-analysis with the abstracted data.

See supplemental appendix 1 for narrative synthesis.

Risk of Bias/Quality Appraisal

Of the six RCTs, three had a low risk of bias and three had a moderate risk of bias. Of the observational studies, one had a low risk of bias and one had a moderate risk of bias. Table 3 demonstrates the risk of bias assessment scores.

Table 3:

Risk of Bias Assessment (quality scores) of included RCT studies and observational studies

Author (Year)	Did the study use a random sequence generation?	Did investigator who enrolled patients know what the next treatment allocation will be?	Were study participants and study personnel blinded to randomization?	Were the people who measured the outcome blinded to randomization?	Was the outcome of interest measured in ≥ 90% of people who received the intervention? (<10% lost to follow up)	Selective outcome reporting. Were all the pre-specified outcomes (in the study protocol) reported?	Overall risk of bias
Fischer (2015)	Yes	No	Yes	Yes	No	Yes	low
Fischer (2018)	Yes	No	No	Yes	No	Yes	low
Heyman (2010)	Unclear	Unclear	Yes	Unclear	No	Yes	moderate
Nedjat-Haiem (2017)	Unclear	Unclear	No	No	Yes	Yes	moderate
Sudore (2007)	Yes	No	No	No	Yes	Yes	moderate
Sudore (2018)	Yes	No	Yes	Yes	No	Yes	low
Author (Year)	Did the intervention group represent dying patients(or end of life) and were consecutive patients evaluated for enrollment?	Did the study have a comparison group (another cohort or before and after of same group)?	Was the primary efficacy outcome measured before the intervention?	Was the primary outcome statistically adjusted for potentially confounding variables (age, gender, severity of illness, etc.)	Was the outcome of interest measured in ≥ 90% of people who received the intervention? (<10% lost to follow up)	NA	Overall risk of bias
Volandes (2008)	No	Yes	Yes	Yes	Yes	NA	low
Detering (2015)	Yes	Yes	No	No	Yes	NA	moderate

Discussion:

This systematic review of interventions to improve ACP and end-of-life decision making among patients with LEP demonstrates several findings. First, there are few interventions to address care planning for patients with LEP. Seven interventions have been developed for patients who speak Spanish, 1 for patients who speak Greek or Italian, and none for other languages. There is limited evidence, at low to moderate risk of bias that demonstrates that such interventions do increase ACP but have less influence on end of life decision making including use of hospice.

Overall, the interventions aimed at influencing ACP were considered feasible to implement and effective when tested. However, several of the interventions required intensive human resources for implementation, including home visits, counseling, and specifically trained personnel, which may limit the ease with which they can be implemented in practice.[39–42] Sudore’s 2018 study examined a patient-facing web-based intervention to avoid the need for systemic or clinician level modifications and this may mean it is more replicable.[44] Some interventions were not successful in achieving the desired outcomes for improved end of life decision-making or increased hospice use and this suggests other approaches should be examined in future research to explore these issues further.[39, 40] Volandes’ study in a healthy population directed towards overcoming educational barriers, used videos to positively influence decisions about comfort measures and life prolonging care in advanced dementia.[46]

Despite the increased use of decision aids at end of life to promote shared decision making, we did not identify any studies that developed or tested a decision aid for those with LEP considering ACP or end of life decisions.[47] It is also noteworthy that no studies tested interventions among LEP populations in the US that spoke a language other than Spanish, although Chinese is the second most common language spoken at home by almost 3 million people in the US.[1]

Two studies that included Spanish language materials were also directed at overcoming health literacy barriers. The use of video aids and modified AD forms written at a 5^th-grade reading level was helpful for those with limited health literacy.[43, 46]Language barriers and poor health literacy may be concurrent and interventions that address both shortcomings simultaneously are vital. These, along with cultural barriers, are considered a “triple threat” and addressing any of these alone may not be sufficient to reduce disparities.[15, 19]

In recent years, there has been growth in efforts to promote discussions about end of life and ACP. The “Death Cafe” movement is an international movement that seeks to increase awareness of death.[48] ”The Conversation Project” aims to encourage initiation of discussions about end of life topics among families.[49] “Respecting choices,” promotes strategies for communities and organizations to educate people about ACP.[50] However, many of these resources concentrate their efforts on English speakers. “PREPARE for your care” is accessible to Spanish speakers, and “Five wishes” which supports pre-emptive discussions of end of life goals of care and wishes among couples is available in several languages.[51] However the costs may limit their use among some marginalized populations.[22] “ACP talk,” developed in Australia, is directed towards sensitive religious and cultural advance care planning and also has resources in several languages. Resources and videos can be accessed free of charge by health care professionals. Improving access to basic informational resources for the lay public is needed and making them free would assist with this endeavor.

Some limitations with our systematic review include the following. As with all systematic reviews, we are reliant on our search strategy to identify relevant studies. We are confident that our expert librarian captured the pertinent studies in the comprehensive search strategy but it is possible studies may have been missed. To avoid this possibility we modified the strategy over several weeks, and also did a manual search from references and appropriate known literature. The search was finalized in November 2018, therefore, it is also possible new research has been published since then that we did not include. For example Dr. Nedjat-Haiem has just published further work on motivational interviewing to enhance AD completion among Latinos with chronic illness. [52] Most of the studies that were included recruited considerably more women than men and this may also affect the generalizability of the studies’ findings. Some studies were limited by their sample size and did not reach statistical significance to demonstrate some outcomes within Spanish speaking groups, therefore overall outcomes for the ethnic group(Latinos) was reported. Difficulties with recruiting those with LEP may have contributed to this and it is possible the studies were underpowered to show effectiveness in some domains. [53]. Furthermore a few studies also used convenience samples of healthy individuals attending their providers to assess interventions. Healthy individuals may not be an adequate representation of those who need or would use the intervention and this may limit the findings of those studies. Our conclusions are tempered by our risk of bias assessments for each study also. Loss to follow up was an issue in several studies.

Strengths of the study include the following. We used robust methods while conducting this systematic review, including having all studies reviewed independently and in duplicate during all phases as well as for risk of bias assessment. We used the PRISMA guidelines for reporting and the study was registered on PROSPERO.[36] All studies targeted older individuals and several studies targeted those with illnesses where considerations about end of life care and ACP would be more salient. This suggests that the findings are applicable in LEP patients with a serious illness or chronic co-morbidities. The findings of this systematic review are somewhat generalizable outside the US for countries whose primary language is English and in which Spanish speaking minorities reside, however, cultural factors may limit the reach of some of the culturally tailored interventions.

Overall the quality of the RCT studies was high with rigorous randomization methods and blinding of study outcome assessors where practical. Of the observational studies, one had a moderate and one had a low risk of bias.

New Contribution to literature

This systematic review shows that while there are few studies that test interventions to improve ACP and AD completion among patients with Limited English proficiency, there are even fewer that are targeted at improving end of life decision making and other outcomes in this population. Interventions for ACP were generally successful although those addressing end of life decision making (such as hospice use) were not so successful. Some of the interventions, while effective, may be difficult to replicate at scale due to the intense resources required. Future research should focus on developing strategies to improve communication about end of life and ACP. These interventions must include a scalable implementation plan that is culturally and linguistically appropriate for patients with LEP as well as being mindful of health literacy issues that may be concurrent. With 350 different languages spoken at home in the US and the population of those with LEP increasing, the scope for this research agenda is vast and remains a priority.