Background
Advance care planning (ACP) is vital for ensuring individuals receive end-of-life care that is consistent with their care preferences (Kermel-Schiffman and Werner, Reference Kermel-Schiffman and Werner2017; Gazarian et al., Reference Gazarian, Cronin and Dalto2019). A multidisciplinary Delphi panel defined ACP as a “process that supports adults at any age or stage of health in understanding and sharing their personal values, life goals, and preferences regarding future medical care” (Sudore et al., Reference Sudore, Lum and You2017b, p. 826). ACP has been shown to improve patient quality of life, promote patient-centered end-of-life care, and help identify potentially avoidable hospitalizations (Kermel-Schiffman and Werner, Reference Kermel-Schiffman and Werner2017; Jimenez et al., Reference Jimenez, Tan and Virk2018; Hickman et al., Reference Hickman, Unroe and Ersek2019).
Despite the benefits of ACP, only 11% of Americans have actually discussed ACP with a healthcare provider (Hamel et al., Reference Hamel, Wu and Brodie2017). Limited health literacy (LHL) has been identified as a significant barrier to ACP (Sudore et al., Reference Sudore, Cuervo and Tieu2018a; Nouri et al., Reference Nouri, Barnes and Volow2019). LHL is defined as the limited ability to attain, comprehend, and communicate basic health information (Nielsen-Bohlman et al., Reference Nielsen-Bohlman, Panzer and Kindig2004). Individuals with LHL often lack knowledge about potentially life-saving but invasive treatment options, including cardiopulmonary resuscitation and ventilators (Eneanya et al., Reference Eneanya, Olaniran and Xu2018), which is necessary to make decisions about their future care preferences. To ensure that individuals with LHL can participate in ACP, highly effective interventions tailored to mitigate the effects of LHL, such as redesigning health information materials using simple language, are needed.
While ACP interventions developed for individuals with LHL have been tested in randomized controlled trials (RCTs), to our knowledge there are no published reviews that systematically evaluate the quality of these RCTs. Therefore, this review will rigorously evaluate the quality of RCTs to test ACP interventions adapted for individuals with LHL using the Joanna Briggs Institute (JBI) Critical Appraisal Checklist (Tufanaru et al., Reference Tufanaru, Munn, Aromataris, Aromataris and Munn2020). Accordingly, the specific aims of this integrative review are to (1) identify RCTs designed to address the effects of LHL on ACP, (2) use the JBI Critical Appraisal Tool to evaluate the quality of each study, and (3) summarize the evaluation data to inform future research in this area.
Methods
Search strategy and inclusion criteria
A review protocol was developed based on the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA; Moher et al., Reference Moher, Liberati and Tetzlaff2009). A literature search was conducted using the PubMed, CINAHL, PsycINFO, and Scopus databases using selected keyword search terms (see Table 1). Inclusion criteria for the studies in this review were (1) peer-reviewed, (2) written in English, (3) sampled patients ages 18 years and older, (4) RCTs, (e) assessed ACP outcome measures and baseline health literacy measures, and (f) published between January 1997 and July 2020.
Table 1. Keyword search query table
Screening process and data extraction
Two independent reviewers (MH and MM) identified relevant studies (see Figure 1). After removing duplicates from the database search, one reviewer (MH) screened records for eligibility criteria. In a second step, both reviewers assessed full-text articles for eligibility. Discrepancies were discussed until consensus was reached. Reviewers used a standardized form to extract information from each article, including year of publication, study purpose, sample characteristics, research methods, measurement tools, and study results.
Fig. 1. PRISMA flow diagram. Note: RCT, randomized controlled trial.
Evaluation of study quality
RCTs were independently evaluated by the same two reviewers using the JBI Critical Appraisal Checklist. The JBI Critical Appraisal Checklist is a peer-reviewed tool that helps researchers rigorously evaluate whether individual studies should be incorporated into a review (Tufanaru et al., Reference Tufanaru, Munn, Aromataris, Aromataris and Munn2020). It assesses the methodological quality of a study and determines whether that study has addressed the prospect of bias in its methodology (Tufanaru et al., Reference Tufanaru, Munn, Aromataris, Aromataris and Munn2020). The checklist consists of 13 questions that assess randomization methods (n = 1); concealment of treatment and control group allocation (n = 1); similarities between baseline treatment and control groups (n = 1); blinding strategies (n = 3); care received for treatment and control groups (n = 1); follow-up description and analysis (n = 1); data analysis (n = 2); outcome measurement (n = 2); and trial design (n = 1) (Tufanaru et al., Reference Tufanaru, Munn, Aromataris, Aromataris and Munn2020). The reviewers assessed whether each study addressed each of the 13 questions in the checklist by coding the question as either “Yes,” “No,” “Unclear,” or “Not Applicable.” A consensus was reached, and the total number of questions answered with a “Yes” were then tabulated and percentages were then calculated to score the quality of the studies.
Results
Description of studies
The database search yielded 253 articles after elimination of duplicates. 203 irrelevant papers were excluded based on titles or abstracts. The 50 remaining papers then went on to the full-text review. Of these, five met eligibility criteria and were included in the final review (see Figure 1).
As shown in Table 2, all studies were conducted in outpatient clinics in the United States. Four studies were conducted in urban cities in the coastal United States (Sudore et al., Reference Sudore, Landefeld and Barnes2007, Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b; Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009), and one was conducted in a rural clinic in the Southern United States (Volandes et al., Reference Volandes, Ferguson and Davis2011). The five studies included in this review were conducted by two research groups (three studies were from the Sudore group and two were from the Volandes group). The two groups used two different measures of health literacy and both tested interventions they developed. Three studies used the Short Form Test of Functional Literacy in Adults (s-TOFHLA; Baker et al., Reference Baker, Williams and Parker1999) to categorize participants into limited or adequate health literacy groups (Sudore et al., Reference Sudore, Landefeld and Barnes2007, Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b), and two studies used the Rapid Assessment of Adult Literacy in Medicine (REALM) tool (Davis et al., Reference Davis, Long and Jackson1993) to categorize participants into low, marginal, or adequate health literacy groups (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009, Reference Volandes, Ferguson and Davis2011). The percentage of participants with LHL, defined as a score of ≤22 on the s-TOFHLA and defined as low or marginal literacy with a score ≤60 on the REALM, ranged from 20% to 53% across the five studies.
Table 2. Summary of the literature review (N = 5)
Notes: yr(s)., years; LHL, Limited Health Literacy; I, Intervention; C, Control; AD, Advance Directive; ACP, Advance Care Planning; mo(s)., month(s), wk(s)., week(s); NR, not reported.
a Short Test of Functional Health Literacy in Adults (S-TOFHLA; Baker et al., Reference Baker, Williams and Parker1999).
b Chi-Square Test.
c Mixed Effects Logistic or Linear Regression.
d Rapid Estimate of Adult Literacy in Medicine (REALM; Davis et al., Reference Davis, Long and Jackson1993).
e Independent Sample t-Test.
Sample sizes varied greatly among studies. Intervention group sizes ranged from 33 to 481 participants and control group sizes ranged from 43 to 505 participants. A majority of the participants in all five studies identified as White (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009, Reference Volandes, Ferguson and Davis2011; Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b). In three studies, the mean age for participants was greater than 65 years (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009, Reference Volandes, Ferguson and Davis2011; Sudore et al., Reference Sudore, Boscardin and Feuz2017a). All but one study, which was conducted in a population of veterans (Sudore et al., Reference Sudore, Boscardin and Feuz2017a), had greater than 50% female participants. Two studies included Spanish-speaking participants (Sudore et al., Reference Sudore, Landefeld and Barnes2007, Reference Sudore, Schillinger and Katen2018b).
All studies required participants to have a primary care provider at the study clinic and excluded individuals who were cognitively impaired. Two studies required participants to have two or more comorbidities (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b), and three studies required participants to own a telephone (Sudore et al., Reference Sudore, Landefeld and Barnes2007, Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b).
Several intervention types designed to mitigate the effects of LHL on ACP were tested (see Table 2). Approaches to addressing LHL included writing text at lower reading levels, adding images to improve communication, and designing materials with an appropriate layout to enhance readability. One study tested an easy-to-read advance directive intervention, which provided individuals with an advance directive written at a fifth-grade reading level and included pictures that helped explain the text (Sudore et al., Reference Sudore, Landefeld and Barnes2007). Two studies with the same first author tested an interactive website intervention that included video stories to enhance an easy-to-read advance directive (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b). Two studies with the same first author tested a short video intervention for improving participants’ comprehension of their disease and care options (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009, Reference Volandes, Ferguson and Davis2011).
Of the five studies, four included longitudinal data collection that ranged from one week to 15 months after interventions (Sudore et al., Reference Sudore, Landefeld and Barnes2007, Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b; Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009). Outcome measures included patient knowledge (n = 2), care preferences (n = 2), certainty of care preferences (n = 1), ACP engagement (n = 2), and ACP documentation (n = 3).
Reported findings
Of the two studies that measured knowledge, one specifically measured participants’ knowledge of advance directives (Sudore et al., Reference Sudore, Landefeld and Barnes2007) and the other measured participants’ knowledge of dementia (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009). Although advance directive knowledge significantly improved for both the control and intervention groups, there was no significant post-intervention difference between the groups (Sudore et al., Reference Sudore, Landefeld and Barnes2007). In contrast, post-intervention dementia knowledge scores were significantly higher for the intervention group compared with the control group (p < 0.001) (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009).
The two studies that measured patient care preferences found that participants had a higher preference for comfort care than aggressive end-of-life care after the intervention (p < 0.05) (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009, Reference Volandes, Ferguson and Davis2011). The one study that measured certainty of care preference found that participants in the intervention group were more certain about their care preferences immediately after the intervention (p < 0.0001) (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009), and less likely to change their care preferences over a six-week period compared with the control group (p < 0.001) (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009).
ACP engagement scores were significantly higher after the intervention compared with the control in both studies that measured ACP engagement (p < 0.001) (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b). In addition, all three studies that reported ACP documentation outcomes found a significant increase in advance directive completion for the intervention group compared with the control group (p < 0.05), but not a significant difference in ACP care provider discussions between groups (Sudore et al., Reference Sudore, Landefeld and Barnes2007, Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b).
JBI critical appraisal of RCTs
All studies were coded with a “Yes” for at least half of the checklist questions and two were coded with a “Yes” for more than two-thirds of the checklist questions (see Table 3). The most common methodological issues identified included baseline differences in treatment and control groups, issues with blinding, suboptimal reporting of outcome measure psychometric data, and inappropriate statistical analyses.
Table 3. JBI critical appraisal of RCTs
Notes: RCT, Randomized Controlled Trial; Y, Yes; N, No; N/A, Not Applicable; UC, Unclear. Adapted from the Joanna Briggs Institute (JBI). JBI Manual for Evidence Synthesis: 2020.
a N: Calculated using 4 articles instead of 5.
b N: Calculated using 12 criteria instead of 13.
Baseline differences between the treatment and control groups
Intervention and control groups in three of the five RCTs were not similar at baseline (see Table 3). There were significant differences in age (Sudore et al., Reference Sudore, Landefeld and Barnes2007), previous ACP documentation (Sudore et al., Reference Sudore, Cuervo and Tieu2018a), and dementia diagnosis and relationship to someone with dementia (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009). One study corrected for their dissimilarity in the analyses (Sudore et al., Reference Sudore, Schillinger and Katen2018b); however, because groups were not similar prior to correction, selection bias cannot be ruled out even in that study.
Blinding to treatment assignment
Participants and those delivering treatments were not blinded to treatment assignment for any of the five RCTS. To minimize potential biases resulting from nonblinding, two studies stated that they blinded participants to the study hypothesis (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b).
Psychometric properties of outcome measures
The psychometric properties of the outcome measures were not consistently reported among the studies. Validity and reliability of outcome measures were not reported in two studies (Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009, Reference Volandes, Ferguson and Davis2011). One study evaluated internal consistency reliability by calculating Kuder–Richardson reliability coefficients for subscales of an advance directive knowledge survey, but did not report those reliabilities (Sudore et al., Reference Sudore, Landefeld and Barnes2007). This same study also used factor analysis to explain variance in scales, but did not report results of the factor analysis (Sudore et al., Reference Sudore, Landefeld and Barnes2007). Two studies provided appropriate references for validity and reliability measures for their main outcome measure, the ACP Engagement Survey; however, the authors did not report the validity and reliability data in the paper (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b), rather, these data were reported in the study protocols (Sudore et al., Reference Sudore, Le and McMahon2015, Reference Sudore, Barnes and Le2016).
Appropriateness of statistical analyses
Appropriate statistical analyses and reporting of findings were not performed in two of the five studies. One study did not report unadjusted results for logistic regression analysis of ACP documentation completion rates between groups (Sudore et al., Reference Sudore, Boscardin and Feuz2017a). This study also reported a significant p-value (p = 0.047), but a nonsignificant confidence interval [CI = 1.0–15.1], for the ACP documentation rates between control and experimental groups (Sudore et al., Reference Sudore, Boscardin and Feuz2017a). One study also failed to report standard deviations or factor analysis results for the advance directive knowledge survey used (Sudore et al., Reference Sudore, Landefeld and Barnes2007).
Discussion
This literature review identified RCTs designed to mitigate the effects of LHL on ACP and evaluated the quality of these studies. We found only five studies that met our eligibility criteria, which were conducted mainly by two research groups. Intervention approaches included rewriting text at lower reading levels, adding images to materials, creating videos to enhance communication, and designing materials to enhance readability. Interventions identified aimed at increasing participant knowledge, preference for comfort care, engagement, and care documentation. Determining which interventions are most effective for mitigating the effects of LHL on ACP is difficult due to heterogeneity in ACP intervention approaches and outcome measures; however, we conclude that employing multiple methods to address LHL, such as introducing both an easy-to-read advance directive and an interactive website (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b), might be an advantageous approach for improving ACP outcomes in individuals with LHL.
Although we identified several interventions that significantly improved ACP outcomes, we found a lack of diversity in study settings and participants’ race, ethnicity, and age among all five studies. For instance, all studies focused on middle-aged and older adults, despite literature that suggests that ACP is a process that supports adults at any age with discussing and planning future healthcare decisions (Weathers et al., Reference Weathers, O'Caoimh and Cornally2016; Sudore et al., Reference Sudore, Lum and You2017b; de Vries et al., Reference de Vries, Banister and Dening2019). All studies were conducted during outpatient visits, which is an ideal time for providers to discuss end-of-life preferences because individuals are generally stable and not acutely ill during these visits (Houben et al., Reference Houben, Spruit and Groenen2014); however, implementing ACP interventions in outpatient clinics might be problematic since many historically underrepresented populations, including Black and Latinx Americans, lack access to health care (Manuel, Reference Manuel2018; Rhee et al., Reference Rhee, Marottoli and Van Ness2019). Additionally, studies were only conducted in settings on the East Coast, West Coast, and Southern United States, thus limiting generalizability to persons living in other regions of the United States or other countries.
We identified several methodological issues among the reviewed RCTs, including issues with baseline differences between the treatment and control groups, blinding, outcome measure reliability and validity, and statistical analyses. Three studies reported differences between the treatment and control groups at baseline (Sudore et al., Reference Sudore, Landefeld and Barnes2007, Reference Sudore, Cuervo and Tieu2018a; Volandes et al., Reference Volandes, Paasche-Orlow and Barry2009), which might be due to issues with randomization methods or adequacy of sample size. Blinding issues were also identified in all studies; however, these issues are inherent to many behavioral interventions due to ethical concerns with using a placebo or nonequivalent comparator (Edmond et al., Reference Edmond, Turk and Williams2019). To minimize bias, two studies blinded participants to the study hypothesis (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b). Blinding the study hypothesis allows researchers to provide participants with an explanation of the intervention being offered, while concealing the intent of the study (Edmond et al., Reference Edmond, Turk and Williams2019).
There were also concerns with the lack of reporting of reliability and validity data for the measures used in the studies. Complete reliability and validity data were not reported for outcome measures in all RCTs, which is a concern since suboptimal psychometric rigor is a threat to internal validity (Heale and Twycross, Reference Heale and Twycross2015); however, since the concept of ACP has expanded over the past decade and there is currently a lack of consensus on outcomes that define successful ACP (Sudore et al., Reference Sudore, Lum and You2017b), there may be a dearth of psychometrically evaluated tools available to measure ACP outcomes, such as knowledge or patient care preference. Additionally, journal restrictions or previous protocol publication might have prevented authors from publishing reliability and validity data in their papers.
Our findings have implications for clinicians. Since individuals with LHL often have difficulty comprehending important concepts discussed during ACP (Institute of Medicine, 2015; Eneanya et al., Reference Eneanya, Olaniran and Xu2018; Ladin et al., Reference Ladin, Buttafarro and Hahn2018), clinicians might need to employ different methods and invest more time into ACP interventions for this patient population. For instance, a clinician might promote the use of an interactive website or easy-to-ready advance directive prior to a clinic visit so that an individual may ask questions during their visit, or the clinician may reinforce concepts introduced. Additionally, there is evidence that engagement in ACP increases over time (Sudore et al., Reference Sudore, Boscardin and Feuz2017a, Reference Sudore, Schillinger and Katen2018b), suggesting that ACP is not a one-time occurrence, but rather a process that involves behavior change. Therefore, clinicians should evaluate the ACP needs of their patients frequently, such as during routine care visits and other medical appointments.
There are also several research implications of this review. First, since all studies in this review were performed in outpatient clinics, there is a need to determine the effectiveness of these interventions in different settings, such as nursing homes, assisted living facilities, and long-term care facilities. Determining the efficacy of these interventions in nonclinical settings is important because clinician time constraints and barriers to healthcare access might prevent ACP interventions during a clinic visit. Second, researchers should use reliable and validated instruments for ACP outcomes to encourage comparison of findings across studies and ensure that constructs are appropriately measured (Sudore et al., Reference Sudore, Lum and You2017b). Finally, it is important to note that a majority of the interventions identified in this review were created and implemented by physicians, which might have limited the representativeness of approaches and their associated frameworks for addressing the effects of LHL on ACP.
There were several limitations of this review. First, because we only reviewed papers in English, we might have missed studies that were written in other languages. Second, including only RCTs prevented inclusion of findings from quasi-experimental studies. Finally, the eligibility criteria of this review prevented the inclusion of ACP interventions that did not directly measure participant's health literacy.
This review highlights the need for more valid evaluations of interventions that address the effects of LHL on ACP. Although the studies in this review implemented approaches tailored for individuals with LHL, interventions were not implemented exclusively in populations with LHL. Our findings suggest that interventions included in this review improved ACP outcomes, regardless of health literacy levels.
Since study settings were limited to outpatient clinics, future studies should be performed in alternative settings and locations, such as community settings and in the Midwestern U.S., since environmental context might affect the efficacy of interventions. Additionally, future intervention studies should continue to include historically underrepresented populations, including Latinx or Black populations, since LHL rates remain especially high (RTI International, 2011) and ACP participation remains low, among these groups (Hong et al., Reference Hong, Yi and Johnson2018). When implementing ACP interventions that address LHL in these populations, it is necessary to consider the contributing social determinants of health that many of these underrepresented populations experience, including inequities in educational opportunities, lack of culturally appropriate health information, and racism (Stormacq et al., Reference Stormacq, den Broucke and Wosinski2019; Muvuka et al., Reference Muvuka, Combs and Ayangeakaa2020). Since social determinants of health affect health literacy, we should also consider how they affect ACP interventions designed to mitigate the effects of LHL. Future research should assess the relationships among social determinants of health, health literacy, and ACP outcomes to improve ACP interventions for individuals with LHL.
Funding
The authors did not receive any financial support for the research, authorship, and/or publication of this review article.
