Abstract
Introduction
Advances in medical science afford individuals who are living with a chronic disease the opportunity for enhanced health outcomes and a productive life by taking medications daily as prescribed. 1 However, for some people who are living with a chronic condition, taking medications on a daily basis for a lifetime can be a daunting task. 2 Nonadherence to medications as prescribed, sometimes termed noncompliance, is a prevalent problem and multifaceted phenomenon that can vary in nature between individuals, even within the same disease state. 3 However, regardless of the specific disease, suboptimal medication adherence frequently results in negative health outcomes and increased morbidity and mortality. 4 In selected chronic conditions, such as HIV disease, suboptimal adherence can become a threat to public health as nonadherence to HIV medications can result in an increased risk for transmitting HIV infection to other individuals. 5 People with chronic diseases who are nonadherent to their medications often require more health care services, frequent hospital admissions, and complex treatment regimens. 6 Addressing nonadherence in chronic disease is a significant challenge facing health care providers, policy makers, researchers, and the health care system in general.
Traditionally, professionals have been called upon to facilitate interventions that address individual and population health needs and problems. However, there is growing awareness that including patients, stakeholders, and community members in health care interventions can enhance effectiveness, particularly when addressing complex health conditions. 7 There are published intervention models that have successfully used peers (ie, lay individuals) as facilitators to enhance health. Most studies using lay individuals to deliver intervention programs have targeted health promotion (eg, violence prevention) or an increase in screening uptake (eg, mammography).8,9 Lay individuals have been successful as change agents in such endeavors. One successful example is the promotora model, which has been repeatedly used within the Latino population. 10 A systemic review of obesity interventions for Latinos found strong evidence that promotora-led interventions are effective in reducing obesity. 11 The promotora model, also called the community health worker (CHW) model, is an effective strategy to enhance health outcomes. 12 The CHW model has primarily focused on educational interventions and some studies have included medication adherence as an outcome measure. 13 However, promotoras and CHWs are not typically individuals who are living with the same disease as the target population. 14
With regard to mediation adherence interventions, primarily it has been health care professionals, and not peers (ie, lay individuals living with the same disease), who have delivered such interventions. 15 Despite concerted efforts and large numbers of medication adherence intervention trials, nonadherence to medications continues to be a vexing problem particularly among individuals with chronic disease conditions that require lifelong adherence to medications. 16 The use of peers as facilitators of medication adherence interventions is a strategy that has promise and could enhance the cultural relevance of medication adherence interventions. 17 In other words, lay individuals who are living with the chronic disease serve as facilitators of the intervention in an effort to enhance adherence among others who are also living with the same chronic disease. 18
The current review was undertaken to explore the literature with regard to peer-facilitated medication adherence interventions and to examine what is known about the effectiveness of such interventions in chronic disease. In this article, published interventions that used solely peers (ie, individuals living with the same chronic disease as the target population) as facilitators and reported at least 6 months of follow-up on medication adherence are summarized. The review focuses on the characteristics of the interventions, interventionists, and the impact that peer-facilitated interventions had on adherence. Information is also presented about other health outcomes of interest to the studies included.
Methods
Data Sources
The original search of multiple databases was conducted January 2014 and updated January 2015. Selected electronic databases were searched from their start date through December 31, 2014 based on key terms, inclusion criteria, and exclusion criteria. The search engine Google was used to search for abstracts presented at conferences that might be of interest to this review. Ancestry searches were also conducted of the reference lists of publications that met the inclusion/exclusion criteria. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement guidelines were followed. 19
Search Strategy and Study Selection
For this review, “intervention” was defined as a behavioral treatment that aimed to enhance the health behavior of medication adherence. To be included in this literature review, a peer-facilitated behavioral intervention had to have targeted a group of people living with a chronic disease who had been prescribed medications for said chronic disease by a physician, nurse practitioner, or physician’s assistant. Interventions that focused on enhancing medication adherence in diseases that require taking medications for a finite time frame were excluded. Reports of medication adherence interventions in chronic diseases that were facilitated by health care professionals or paraprofessionals were excluded, as were interventions that focused strictly on enhancing healthy behaviors in chronic disease that did not involve taking medications (eg, exercise, healthy eating) and hence did not include medication adherence as an outcome measure. To be included in this review, solely peers (ie, meaning lay individuals living with the same chronic disease) delivered the intervention. For example, interventions delivered by a lay individual in partnership with a professional or paraprofessional (ie, the lay individual and professional delivered the intervention together at the same time) were excluded. Also excluded were interventions where the peer was a proxy facilitator (ie, an educational video was shown to the participant of a peer speaking or a peer showed the participant an educational video of a nurse or physician speaking).
The following databases were utilized in this search: PubMed, CINAHL, Google Scholar, Google, and PsychInfo. The search strategy for PubMed was as follows: limits were set to include only articles published in English with no start date. The search end date was December 31, 2014. The key terms “medication,” “adherence,” and “intervention” were entered into the database search box together with the term “chronic disease.” Then these 3 key terms were entered again with each of the terms “diabetes,” “HIV,” “hypertension,” and “hyperlipidemia,” in place of “chronic disease.” The same searches were also conducted using the terms “lay” and “peer.” The initial searches yielded a total of 1163 articles. There were a number of articles that could be excluded simply by reading the title. The abstracts of 95 potentially relevant publications were then read and assessed for eligibility. If eligibility could not be determined by reading the abstract, the entire article was accessed and reviewed.
Data Extraction and Synthesis
One author independently abstracted each included study. The second author contributed to the interpretation of the extracted data. The following data were extracted: chronic disease targeted, author and year of publication, study design and period, country where study was implemented, type of setting where the study took place, intervention delivery format, theoretical framework, sample size, measure of adherence, other measures of interest to the study, characteristics of peer interventionists, training of peer interventionists, length of follow-up, attrition, study findings. Table 1 shows an overview of the studies included in the review by chronic disease targeted.
Overview of Studies by Chronic Disease Targeted.
Abbreviations: AIDS, acquired immunodeficiency syndrome; BP, blood pressure; CI, confidence interval; DM, diabetes mellitus; HbA1C, hemoglobin A1C (glycated hemoglobin); HIV, human immunodeficiency virus; HTN, hypertension; IMB model, informational, motivation, and behavioral skills model; MA, medication adherence; MEMS, medication event monitoring system (ie, electronic drug monitors); NA, not applicable; NR, not reported; TB, tuberculosis; VA, Veterans Affairs.
Both studies tested same intervention.
Results
Eleven articles met the inclusion criteria for this review (see Figure 1). These articles reported findings about 10 different peer-facilitated interventions, which included medication adherence as an outcome measure. All interventions targeted adults and no published peer-facilitated interventions were located that targeted children. One of the interventions was tested twice and published as 2 separate studies.20,21 There were 6 different chronic diseases targeted by the interventions, with HIV being the most common. Most of the studies employed a randomized controlled design, were conducted in the United States, and used medical settings as their recruitment sites. The number of participants in each study ranged from 77 to 1336. The majority of studies used some type of theoretical framework and a variety of formats were used to deliver the peer-facilitated interventions. Length of follow-up ranged from 24 to 96 weeks and overall attrition was low.
Literature review flow diagram.
Theoretical Frameworks
Most of the studies reported the use of some type of theoretical framework to guide the intervention. Two studies did not report on theory utilization but overall there was emphasis on communication and social support, although a specific communication theory or social support theory was not cited. Theories and frameworks used to guide the peer-facilitated interventions included motivational interviewing, 22 social support,20,21 IMB (informational, motivation, and behavioral skills) model, 23 Orem’s self-care theory, 24 Rodgers’ person-centered approach, 25 and Lorig’s chronic disease self-management program. 26 Except for the 2 studies that examined the same intervention grounded in social support theory, no other studies utilized the same theoretical underpinnings.
Intervention Format and Length
The format and length of the peer-facilitated interventions differed. Six interventions used an individual level approach: 4 were delivered in a face-to-face format23,24,27,28 and 2 were delivered via telephone.22,29 While most interventions were delivered at the individual level, there were 3 studies that used a group format25,26,30 and 1 that used a mix of a support group together with individual telephone contact.20,21 One of the group interventions used a peer mentoring strategy 26 and the other two used a group format to provide education. Some interventions could be delivered as often as daily,22-24 others weekly,20,21,26,27,29 and still others were delivered monthly. 25 The duration of the interventions ranged from 6 weeks to 6 months.
The frequency and duration of the interventions varied greatly. Some interventions tailored the number of intervention sessions and contacts with the peer based on the needs of the participant.22,24,26,28,29 For example, the location of contact with the peer facilitator (eg, home visit vs work site vs phone support) was based on participant preference, as was the frequency of contact. Other interventions had a set number of sessions, which ranged from 3 to 12, but varied in the length of delivery. For example, some interventions sessions were held weekly, while others were held biweekly or monthly.
Populations and Diseases Targeted
All studies targeted adults living with a chronic condition. The most common condition was HIV with five studies, albeit only 4 different interventions.20,21,23,24,27 Three of the HIV studies were conducted in African countries in community-based settings and clinics.23,24,27 The other 2 HIV studies occurred in large East and West coast HIV clinics located in the United States and tested the same intervention.20,21
There were 2 studies focused on adults with diabetes: one study targeted individuals who were taking oral medications 22 while the other included individuals who were taking oral and/or injectable medications or both. 29 The 2 diabetes studies were both conducted in medical settings. One diabetes study was conducted in medical practice settings across the United Kingdom, 22 while the other was conducted in 2 US Veterans Administration settings in the same state. 29
The remaining 4 studies targeted adults with hypertension, 28 ischemic heart disease, 25 migraine headache, 30 and serious mental illness. 26 The ischemic heart disease study was conducted in Scotland and the others in the United States. The study targeting individuals with hypertension was conducted in a community setting and the other 3 studies were set in medical centers.
Characteristics and Training of Interventionists
All interventionists were lay individuals (ie, not health care professionals) who were living with the same chronic condition as the target population. The title given to the peer interventionists varied among the studies. Titles given to interventionists included such names as lay facilitators, buddies, partners, lay health mentors, and supporters, while some studies simply called them peers.
All interventions provided some form of structured training to the peer interventionists. Training ranged from as little as one 3-hour group session prior to beginning the intervention, to 30 hours of training followed by ongoing in-services throughout the course of the study. One study reported that some peers had to be dismissed from the study during training due to an inability to master the intervention technique (ie, motivational interviewing). 22 Another study reported a requirement that peer facilitators show competency by passing a written test at the end of the training. 30
Medication Adherence Outcomes
For the most part, adherence was measured using self-report via oral interview or written questionnaires. Several studies used biomarkers or vital signs as a proxy measures of adherence (ie, HbA1C, blood pressure, HIV viral load, CD4 cell count).20-23,27-29 One study used pill counts to measure adherence 27 and 2 studies used medication event monitoring systems20,21 (ie, electronic monitoring caps on pill bottles): these were all HIV studies. Furthermore, there were studies that used a combination of measures to examine adherence (eg, self-reported medication adherence and change in HbA1C).
Other Outcomes
In addition to adherence, a variety of other outcome variables were also examined by several studies.20,22-26,29,30 Like adherence, the majority of the other outcome variables were also measured using self-report. Other variables of interest included stress, 29 social support,20,23 self-efficacy,22,23 depression,20,23 exercise/physical activity,25,26 and quality of life.26,30
Attrition and Attendance at Intervention Sessions
Overall attrition (ie, dropout) rates were low and attendance at intervention sessions was high across the studies reviewed, but particularly so for the peer arms. Because several of the studies targeted populations with serious life-threatening conditions (ie, AIDS, ischemic heart disease), death was treated as a study end point, rather than attrition. There were 3 studies with 0 attrition,24,28,30 with the majority of the studies experiencing attrition rates between 2% and 10% in the peer-delivered intervention arms.22,23,25-27,29 Just 2 studies had attrition rates more than 10% in the peer groups: one study had 12% attrition 21 and the other had 17%. 20 Attrition rates in the comparison/control arms were somewhat higher and ranged from 4% to 28%. The studies with the lowest attrition also had the highest intervention attendance rates. In other words, participants attended more sessions and therefore had more contact time with peers over the course of the intervention. The highest rates of attrition, and also the lowest attendance at intervention sessions, were seen in US studies that targeted individuals with HIV/AIDS.
Intervention Impact
Ten of the 11 studies in this review reported positive findings among participants in the peer arm. Eight of the 10 studies with positive findings had statistically significant increases on medication adherence in the peer-facilitated intervention arm. The 2 interventions that did not report significant improvement with regard to medication adherence did however report other significant findings. One of these interventions was focused on enhancing patient activation, which did increase significantly (P = .03), while medication adherence was considered a secondary outcome. 26 The other study focused on reducing migraine disability as its main outcome, which decreased significantly. 30 Of note, the migraine intervention study reported that participants in the peer intervention arm experienced a 38% higher adherence rate to their prophylactic medication to prevent headache. The increased adherence in the use of prophylactic medication was considered to be clinically significant for this particular disease condition. 30
The interventions that enhanced adherence, also positively affected other variables of interest to the studies reviewed. Two studies reported significant increases on social support among participants who received the peer-facilitated interventions.20,29 Two other studies reported significant increases in exercise/physical activity in the peer arms.25,26 Finally, several studies reported significantly better emotional health/quality of life outcomes in the peer arms.20,24,26,30
All of the interventions with very low attrition (0 to 3%) reported significantly increased medication adherence in the peer arm. The lowest attrition interventions were all tailored in some fashion. Strategies to tailor the interventions involved adapting the dose, place of delivery, and/or content of the intervention based on the needs of the participants.23,24,27,30 Most of the interventions with very low attrition rates were delivered at the individual level; however, one was a group-level intervention. 30 Interventions with low attrition rates also had more frequent peer contact over longer periods of time.
The highest impact on medication adherence measures was reported for interventions that were delivered face-to-face. There was a mixed format intervention that consisted of face-to-face, telephone, and/or text messaging, which reported that the greatest impact on the main clinical outcome measure of adherence occurred among the participants who had the highest amount of face-to-face contact time with peers. 21 The 2 peer interventions delivered solely by telephone were not found to be as effective as the face-to-face peer-facilitated interventions.22,29 Both telephone interventions addressed diabetes mellitus and used HbA1C as a proxy measure of adherence. One telephone intervention study had null findings. 22 The other telephone intervention had no impact on self-reported medication adherence but participants in the peer arm did have significant improvement on HbA1C. 29
Discussion
This review explored what is known about peer-facilitated mediation adherence interventions in chronic disease. For this review, peer-facilitated meant that the intervention was led by a lay individual living with the same chronic disease as the target population. Despite setting no limit on search start dates, only 11 studies were located that solely used peers to deliver behavioral medication adherence interventions to individuals living with chronic diseases. These 11 studies reported on 10 different interventions; one of the interventions was published twice as it was tested in 2 separate randomized clinical trials.
Medication adherence and other important clinical outcomes changed significantly in the desired direction across the peer-facilitated interventions reviewed here. Overall, attrition was low, particularly in the peer-delivered intervention arms of the studies. By comparison, a recent meta-analysis of health care provider led medication adherence interventions found a modest rate of attrition (22%). 31 Thus, having a peer as the intervention facilitator may have helped decrease attrition. The strategy of using peers as intervention team members may warrant further investigation with regard to retention.
The majority of interventions had some type of theoretical underpinning. The most effective interventions were those delivered in a face-to-face format where participants had more contact with peers for a longer period of time. These interventions also had the highest engagement with regard to intervention attendance and the lowest attrition rates. Face-to-face interventions delivered over long time periods have also been shown to be effective when delivered by health care providers. 31 However, peers may be a more cost-effective strategy for delivering long-term and time-consuming medication adherence interventions.
Despite the encouraging findings of this review, some methodological issues must be considered when interpreting the results. The measure of adherence used may have been a limitation for some studies. Self-report, which can be overestimated, as the single measure of adherence was common among the studies included in this review. Historically the use of self-report has been a predominant measure of adherence in medication adherence intervention studies for a variety of reasons.31,32 In this review, many authors cited limited resources as their rationale for using self-report as the measure of adherence, instead of other more resource intense measurements such as medication event monitoring system.
Long-term follow-up of adherence postintervention was limited. Although this review required that participants have at least 6-months of follow-up, there was only one study located that was excluded from the review because the follow-up was too short. In fact, most studies that met the inclusion criteria for this review had only 6-month follow-up, with just 3 studies followed participants for a longer period of time. Hence, longer follow-up may need to be considered in future studies.
This review had other limitations, which must be considered when interpreting the results. Despite its systematic format, it is possible that articles reporting peer-facilitated medication adherence interventions could have been missed because peer-facilitated interventions are not well indexed in computerized databases. Because this review was limited to English language, peer-facilitated interventions conducted in other countries and published in other languages were excluded. In addition, some intervention characteristics may not have been reported and therefore could not be included in this review. Limited reporting of information about the interventions themselves precludes implementation in practice or efficient further research. For example, there were a handful of community health worker-led interventions in the published literature that examined medication adherence but did not describe the interventionists, except to state they were lay individuals. In those cases, it was not possible to determine if the community health workers were, or were not, living with the same chronic disease as the target population. Hence, those articles were excluded and may indeed have had promise but could not be assessed for the ability to enhance and sustain adherence.
Finally, there are lifestyle-related behavior modifications that must also be considered when interpreting the results of this review. For the studies that used biomarkers or vital signs as proxy measures of adherence, one must consider that in some chronic conditions lifestyle behaviors can significantly affect outcomes. For example, the biomarker HbA1C, which was used as a proxy measure of adherence to diabetes medications in some of the studies included in this review, is impacted by lifestyle modification. Hence, one must acknowledge that lifestyle modification behaviors, and not just medication adherence, may have affected this laboratory test which measures average plasma glucose concentrations. For example, behaviors like participating in exercise and eating a reduced calorie diet can play a significant role in the glucose levels of a person who is living with diabetes. 33 Therefore because the outcome measure of HbA1C is dependent on multiple factors, medication adherence alone may not have been the only factor that impacted the direction of this biomarker.
Albeit some limitations, the peer-facilitated interventions presented in this review had positive findings overall. Participants appeared to be highly engaged in the majority of the interventions. Peer-facilitated interventions had positive effects on medication adherence and other desired behaviors such as exercise, which suggest that this strategy has promise. Hence, interacting with a peer who is experiencing the same health problem may help the person with a chronic disease to engage in healthful behaviors.
A limited number of peer-facilitated medication adherence intervention studies were located by this review. The eleven studies that were identified were all published fairly recently, between 2004 and 2010. Given the vast amount of published literature addressing medication adherence, the small number of 11 suggests that the utilization of peers as facilitators of medication adherence interventions is in the beginning stages. Another plausible explanation may be that barriers exist to the use of peers as medication adherence facilitators; hence relatively few such studies have reached fruition. More research to examine the impact of peers as facilitators of medication adherence in clinical and community settings, together with an exploration of possible barriers to the utilization of peers as intervention facilitators, seems warranted.
Footnotes
Authors’ Note
The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by the National Institute of Nursing Research grants R01NR011990 (Conn-PI) and 5 K01 NR014409-03 (Enriquez-PI).