Exercise and antiretroviral adherence in adults living with HIV: A systematic review

Abstract

This systematic review assessed the relationship between exercise and ART adherence in adults living with HIV. A comprehensive search through June 2020 for relevant studies was conducted, and PRISMA guidelines were followed. To be included, studies had to meet the following criteria: (a) published in a peer-reviewed journal; and (b) examined the relationship between exercise and ART adherence. A total of 4310 studies were identified, and nine were included. The majority (five out of nine) of studies found a significant and positive relationship between exercise and ART adherence. Strengths, limitations, and future directions are discussed.

Introduction

For people living with HIV (PLWH) antiretroviral therapy (ART) results in a significantly longer life expectancy and essentially eliminates the risk of AIDS-related complications (Kyser et al., 2011). Yet, maintaining optimal ART adherence (i.e. ⩾95%) to achieve undetectable viral loads to untransmissible levels (i.e. U=U) remains a constant challenge (Salmoirago-Blotcher and Carey, 2018). For example, rates of ART nonadherence in the United States is estimated to be as high as 30% (Bradley et al., 2014), and in sub-Saharan African, is estimated to range from 17.1% to up to 46% (Becker et al., 2020). Meta-analytic data also estimate that only 55% of PLWH in the North America achieve over 80% adherence to ART, and that upwards of 77% of patients in sub-Saharan Africa achieve 80% adherence to ART (Nachega et al., 2014). ART nonadherence is associated with significant morbidity and mortality, including an increased risk of HIV transmission and progression (Nachega et al., 2011), ART-resistant HIV strains (Sethi et al., 2003), and systemic inflammation (Castillo-Mancilla et al., 2016).

ART adherence reflects an individual’s level of compliance to medication dosing schedules and is a complex process that is partially explained by both structural and personal factors (Salmoirago-Blotcher and Carey, 2018). For example, structural factors negatively affecting ART adherence include cost and access to medication, transportation to pharmacies, and discrimination and stigma (particularly among men who have sex with men) within health care systems (Muessig et al., 2017). Personal factors, such as poor mental health (e.g. depression, fatigue, stress, body dissatisfaction, low quality of life and motivation), substance use (e.g. drug and alcohol), poor sleep quality, low self-efficacy, and low social support and education (Kalichman et al., 2017; Reisner et al., 2009; Salmoirago-Blotcher and Carey, 2018) have also been reported to have negative effects on ART adherence.

Given the deleterious effects of ART nonadherence for PLWH, many interventions have been designed and implemented to address barriers at both the structural and patient level to improve adherence. Examples include lowering the out-of-pocket cost for medication, group and individual educational and counseling programs, psychotherapy, drug therapy, memory aids, meditation training, and different mobile health (mHealth) strategies such as text messaging, smartphone applications, and social media (Duncan et al., 2012; Kalichman et al., 2017; Muessig et al., 2017; Risher et al., 2017; Salmoirago-Blotcher and Carey, 2018). These strategies have yielded mixed results, however, and somewhat paradoxically, some interventions present similar barriers to taking ART (e.g. cost) and are limited in their widespread scalability (Katz et al., 2013; Mugavero et al., 2006; Rintamaki et al., 2019; Simoni et al., 2010; Stringer et al., 2019). Therefore, alternative methods that help PLWH establish and maintain ART adherence in daily routines are needed.

Exercise may be such an intervention. Exercise is an established, evidence-based intervention to improve health and well-being for individuals in the general population as well as those with many different chronic illnesses, including diabetes (Colberg et al., 2016), cardiovascular disease (Braith and Stewart, 2006; Myers, 2003), depression (Schuch et al., 2016), and certain types of cancer (Schwartz et al., 2017). In addition to the well-established exercise guidelines from the American College of Sports Medicine to reduce to the risk of cardiovascular disease and diabetes (Nelson et al., 2007), regular exercise is also recommended as a method to prevent and treat issues with physical functioning (among others) that may occur during or after receiving cancer treatment, and also to improve overall quality of life (Cormie et al., 2017; Schwartz et al., 2017). While some physicians may be hesitant to prescribe exercise to their cancer survivor patients because of a lack of knowledge on exercise programming or the lack of clarity of how their patients will respond (Watson and Leonard, 2020), the benefits upon both quality of life and physical health are well-established (Juvet et al., 2017; Schwartz et al., 2017). Moreover, an abundance of research also highlights the beneficial effects of exercise on depression as both a stand-alone intervention and adjunct to other forms of treatment (Rethorst and Trivedi, 2013).

Exercise is particularly important for PLWH, as they are more likely than the general population to develop these co-occurring chronic illnesses (Do et al., 2014). Exercise also warrants examination as a method to improve ART adherence among PLWH due to its ability to address many of the personal and structural and personal barriers to other interventions that support ART adherence. Exercise is consistently shown to improve multiple dimensions of mental health in PLWH including depression, fatigue, stress, anxiety, and health-related quality of life (Heissel et al., 2019; Nosrat et al., 2017), which are in turn related to ART adherence. Similarly, exercise can also have positive impacts on substance use reduction, although most of this work has been in the general population (Abrantes and Blevins, 2019; Wang et al., 2014) rather than specifically focused on PLWH. Of note, exercise seems to improve these outcomes without any changes to immunological indicators of disease progression such as CD4+ cell count and viral load (Jaggers and Hand, 2016). Thus, one potential pathway between exercise and improved ART adherence is through improvements in mental health and reductions in substance use. In fact, two studies found that depression mediated the relationship between exercise and ART adherence (Bhochhibhoya et al., 2019; Blashill et al., 2013), with one of these studies also finding that depression mediated the relationship between exercise and viral load (Blashill et al., 2013). Exercise is also affordable, easily accessible, scalable, and can be performed anywhere and at any time with little to no equipment and can be supervised by non-medical professionals, ultimately avoiding many of the structural barriers associated with ART adherence interventions.

Another potential pathway between exercise and ART adherence could be explained through improvements in social cognitive theory constructs, such as self-efficacy. Self-efficacy reflects one’s perceived ability to manage their behaviors to achieve a specific goal and has been identified as an important predictor of ART adherence (Ammassari et al., 2002). For example, one study found a statistically significant and positive association between ART adherence and adherence-specific self-efficacy in HIV+ youth (Naar-King et al., 2006). Another study of predominantly African American (84%) women (68%) also found a statistically significant and positive relationship between the same two variables (Tatum and Houston, 2017). Fortunately, exercise has been shown to increase general self-efficacy (no research on adherence-specific self-efficacy) in PLWH (Fillipas et al., 2006), but whether self-efficacy (general or adherence-specific) mediates the relationship between exercise and ART adherence in PLWH remains unknown.

Despite the potential of exercise as an effective stand-alone or adjunct intervention to improve ART adherence, limited HIV research has focused on this type of intervention. In fact, in a 2018 systematic review of interventions designed to improve engagement along the HIV Care Continuum, 117 studies were identified that aimed to improve adherence, but none included exercise as a potential strategy (Risher et al., 2017). In addition, despite numerous systematic reviews on the effects of exercise on multiple dimensions of health in PLWH (Dudgeon et al., 2004; Grace et al., 2015; Jaggers and Hand, 2016; Nosrat et al., 2017; O’Brien et al., 2010), this is the first review to specifically assess the relationship between exercise and ART adherence. To address this gap, we systematically reviewed the literature exploring the relationship between exercise and ART adherence in adult PLWH with the aim of collecting evidence that could provide direction for future research to improve health outcomes in this population. Possible directions and suggestions for future interventions to improve ART adherence in PLWH are also discussed.

Methods

Preferred reporting items for systematic reviews and meta-analysis (PRISMA) guidelines were followed to conduct this systematic review (Moher et al., 2009). An electronic search of PubMed, PsycINFO, Embase, and Google Scholar was performed to locate any existing eligible studies between 1996 (the year ART was introduced) through June 2020. The following search terms were used to locate eligible studies: (“HIV” or “AIDS”), (“exercise” or “physical activity”), (“HIV medication adherence” or “antiretroviral adherence” or “ART adherence” or “HIV treatment adherence”) and the syntax was modified to accommodate differences between search databases. To be included, studies had to meet the following criteria: (a) published in a peer-reviewed journal; and (b) examined the relationship between exercise and ART adherence measures (measured by self-report or objective criteria). Studies that were not in English and did not have a translation service on the journal’s home page were entered into Google translate to identify additional studies for possible inclusion. Titles and abstracts were initially reviewed to remove duplicates and those with titles unrelated to the focus of this review. Full texts were obtained for studies that met inclusion criteria and for studies where a determination could not be made based on the abstract alone, and reference lists were scanned thoroughly for additional publications that might warrant inclusion. Since there are currently no randomized controlled trials to our knowledge that have examined the effects of exercise on ART adherence the Physiotherapy Evidence Database (PEDro) scale was not used (Verhagen et al., 1998).

Results

Study characteristics

The electronic search identified 4310 studies for screening. Of these, 4156 were excluded after reviewing titles and removing duplicates. Abstracts were then reviewed of the remaining 154, and 46 full-text articles were retrieved for further review. Of these, nine studies met full inclusion criteria. A more detailed description of the search strategy is presented in Figure 1.

Figure 1.

Flow diagram of search results.

The nine studies meeting full inclusion criteria were published from 2008 to 2020. Six studies were conducted in the United States (Blashill et al., 2013; Kyser et al., 2011; Littlewood et al., 2008; Pellowski and Kalichman, 2016; Webel et al., 2016; Wimberly et al., 2020), one in Iran (Morowatisharifabad et al., 2019), one in Vietnam (Dang et al., 2018), and one in China (Bhochhibhoya et al., 2019). Seven of the studies were cross-sectional and observational (Bhochhibhoya et al., 2019; Dang et al., 2018; Kyser et al., 2011; Morowatisharifabad et al., 2019; Pellowski and Kalichman, 2016; Webel et al., 2016), one was longitudinal and observational (Blashill et al., 2013), and one was a secondary analysis of a longitudinal randomized controlled trial (Wimberly et al., 2020).

Participant characteristics

There was a total of 4134 participants (males = 3202, 77%, females = 930, 23%, transgender = 2, <1%) who had an average age of 41.4 (SD = 8.3) years. Seven studies had a mixed-sample of both male and female participants (Dang et al., 2018; Kyser et al., 2011; Littlewood et al., 2008; Morowatisharifabad et al., 2019; Pellowski and Kalichman, 2016; Webel et al., 2016; Wimberly et al., 2020), and two studies had only male participants (Bhochhibhoya et al., 2019; Blashill et al., 2013). Seven studies reported on race/ethnicity (n = 3122; 37% White, 28% Black, 22% Asian, 8% Hispanic/Latino, 5% Other, <1% Native American) and on employment rate, which ranged from 14% to 81.4% (Bhochhibhoya et al., 2019; Dang et al., 2018; Kyser et al., 2011; Littlewood et al., 2008; Pellowski and Kalichman, 2016; Webel et al., 2016; Wimberly et al., 2020). Income was reported in five studies and the majority of PLWH in these studies made less than $10,000 (Littlewood et al., 2008; Morowatisharifabad et al., 2019; Pellowski and Kalichman, 2016; Webel et al., 2016, Wimberly et al., 2020). Six studies also reported on education and the majority of PLWH had a high school degree of less (Bhochhibhoya et al., 2019; Dang et al., 2018; Kyser et al., 2011; Morowatisharifabad et al., 2019; Webel et al., 2016, Wimberly et al., 2020). Participants who identified as men who have sex with men was reported in three studies: 860 (100% of sample) (Blashill et al., 2013); 295 (56% of sample) (Kyser et al., 2011); and 29 (54% of sample) (Webel et al., 2016). Only one study reported that 68 participants (13%) were injection drug users (Kyser et al., 2011), but one other study recruited 72 PLWH who had a substance use problem (Wimberly et al., 2020). Seven out of nine studies reported on participants viral loads and/or CD4+ counts. More specifically, one study reported that 88% of their sample had a viral load of <400 copies and had a median CD4+count of 486 cells/mm³ (Kyser et al., 2011). In another study, 37% of participants had undetectable viral loads and a mean CD4+ count of 500 (Littlewood et al., 2008). One study found that 57% of participants had undetectable viral loads and that 71.4% had CD4+ counts above 200 (Pellowski and Kalichman, 2016). The majority of participants in two other studies had undetectable viral loads (Blashill et al., 2013; Wimberly et al., 2020). Lastly, two studies reported only on CD4+ counts, with one study reported that 59% had CD4+ counts over 350 (Morowatisharifabad et al., 2019), and the other showing that participants had a mean CD4+ count of 294.7 (SD = 215.2) (Dang et al., 2018).

Lastly, six studies required participants to be on ART to be considered for inclusion in their respective studies (Blashill et al., 2013; Dang et al., 2018; Kyser et al., 2011; Morowatisharifabad et al., 2019; Pellowski and Kalichman, 2016; Webel et al., 2016), and all but one study recruited participants who were currently in an HIV-clinic receiving care (Wimberly et al., 2020).

Exercise measures

There was only one study that objectively measured exercise through attendance to study intervention sessions (Wimberly et al., 2020) while the remaining eight studies used a variety of self-report measures. In two studies, for example, a single-item question was used to collect information about participants’ exercise behaviors in the last 30 days (Kyser et al., 2011; Pellowski and Kalichman, 2016). Another study asked two questions that assessed the frequency with which participants engaged in moderate or vigorous exercise with responses ranging from “never” to “nearly every day” (Littlewood et al., 2008; Pellowski and Kalichman, 2016), while another study used the two-item Lipid Research Clinics Physical Activity Questionnaire which assesses current levels of exercise and how an individual rates their level of exercise compared to others of the same age and sex (Blashill et al., 2013). Using a pen and paper diary, another study asked participants four questions that began with whether or not they did any exercise in the past week and if participants answered yes, three additional questions asked about the type, duration, and perceived intensity (Webel et al., 2016). Another study used the Baecke questionnaire for the measurement of habitual physical activity which is a 16-item measure with three distinct physical activity related subscales (i.e. labor, exercise, leisure time) (Morowatisharifabad et al., 2019). Two other studies used the long form (Dang et al., 2018) and the short form (Bhochhibhoya et al., 2019) of the International Physical Activity Questionnaire.

ART measures

Two of the nine studies objectively measured ART adherence, one through unannounced calls to participants to assess pill counts (Pellowski and Kalichman, 2016) and the other by calculation of the percentage of days in the previous 90 days that participants refilled their prescription at their pharmacy (Wimberly et al., 2020). The remaining seven studies used varying self-report measures. For example, two studies asked participants to recall how many doses they missed over the past 3 days (Kyser et al., 2011; Webel et al., 2016), while another study assessed adherence over the past 7 days (Littlewood et al., 2008). Three other studies assessed ART adherence in two different ways. One study used both a past 30 days and a past week measure (Morowatisharifabad et al., 2019), another study used a past month and a past 3-day measure (Bhochhibhoya et al., 2019), and the last study assessed adherence in the past month and in the past 4 days (Dang et al., 2018). Another study asked participants about their ability to follow their doctor’s instructions on how to take their medications, and the number of missed doses over the last week (Blashill et al., 2013).

Main findings

The following section is divided by study design: cross-sectional and observational, longitudinal and observational, and secondary analysis of a randomized controlled trial.

Cross-sectional and observational

Seven studies were cross-sectional and observational and included 3201 participants (male = 2293, female = 908), and of the studies that reported age, had a mean age of 41.53 (SD = 7.72) years. Of the studies that reported on race and ethnicity, 732 participants were Black/African American; 668 were Asian/Pacific; 447 where White; 18 reported other; 10 were Native American; and 71 were Hispanic. Participants in all seven studies were currently receiving HIV care.

The majority of the cross-sectional and observational evidence suggests that exercise is positively and significantly associated with ART adherence, as four of the seven studies found a positive relationship between the two. The two studies that used a single-item question about exercise in the past 30 days found that those who answered yes were more adherent to their ART regimens over the last 3 days in one study (aOR = 2.13, 95% CI: 1.25–3.57) (Kyser et al., 2011) and during an unannounced phone call to assess pill count in another (p < 0.05) (Pellowski and Kalichman, 2016). The two studies that assessed exercise using the IPAQ; one used the long form (Dang et al., 2018) and one used the short form (Bhochhibhoya et al., 2019), found that those who exercised more had significantly better ART adherence over the past month (OR = .26, 95% CI: .09–.80, p < 0.05) (Dang et al., 2018), (aOR = 1.36, 95% CI:1.02–1.80, p < 0.05) (Bhochhibhoya et al., 2019). Three of the seven studies, however, failed to find a relationship between exercise and ART adherence (Littlewood et al., 2008; Morowatisharifabad et al., 2019; Webel et al., 2016). One of these studies found that the frequency of either moderate or vigorous intensity exercise performed over the previous month was not associated with the frequency of participants’ frequency of missed doses, late doses, or ignored instructions on prescription medication over the previous week (p > 0.05, r=−.08) (Littlewood et al., 2008). Another study found that the three factors within the Baecke (i.e. labor, exercise, and leisure-time) were independently and collectively unrelated to ART adherence over the last seven and 30 days (ps=.32–.56) (Morowatisharifabad et al., 2019). Finally, hours of exercise in the past week was unrelated to ART adherence as assessed with 3-day recall (p = 0.79, rho = .03) (Webel et al., 2016).

Longitudinal and observational

The single study examined the association between exercise and ART adherence over a 3-month period using data collected from a multicenter cohort of 860 HIV positive men who have sex with men who were engaged in clinical care and currently receiving ART (Blashill et al., 2013). On average, participants were aged 43.8 (SD = 9.9) years, and were mostly White (79.7%). This study found a statistically significant and modest positive association between physical activity and ART adherence at baseline (p < 0.05, r = .09) and that these two variables did not significantly change over 3 months (p = 0.25 and p = 0.64, respectively). Over time, however, physical activity was not associated with ART adherence when depression was controlled for (95% CI: −.04 to .12, p = 0.30), highlighting the potential mediating role of depression.

Secondary analysis

Lastly, one study performed a secondary analysis of a randomized controlled trial (with a different primary outcome) to explore the effects of exercise on ART adherence (Wimberly et al., 2020). The randomized controlled trial tested the effects of hatha yoga on stress and substance use in 72 PLWH (50 males, 21 females, two transgender) who were recently released from prison or jail and thought to have a substance use problem. Participants had a mean age of 44.46 (SD = 10.46) years and were mostly Black (78%). The most problematic drug for participants was crack (39.7%) and the majority of participants had no yoga experience (68%). Participants were randomized to either a weekly, 90-minute hatha yoga class for 12 weeks or treatment as usual. After controlling for ART adherence at baseline, results showed no statistically significant treatment effect of yoga on ART adherence F(1, 69) = .89, p < 0.39. Despite the lack of statistical significance, adherence increased in the yoga group from 65% to 75% compared with a decrease in adherence from 77% to 66% in the treatment as usual group. It is important to note that participants only attended an average of 4 of the 12 yoga sessions with the most prevalent reasons for missing sessions being incarceration and illness. A full description of all included studies can be found in Table 1.

Table 1.

Study characteristics.

Authors	County	Study design	Sample size	Race/ethnicity	Mean age (SD)	Exercise measure	ART adherence measure	Main findings
Blashill et al. (2013)	USA	Longitudinal	860, 100% Male	79.7% White, 10.5% Black, 2.3% Asian, 7.5% Other, 28.4% Hispanic	43.8 (9.9)	2-item, lipid research physical activities questionnaire	2-item Art- nonadherence questionnaire	Positive association between exercise and ART adherence
Kyser et al. (2011)	USA	Cross-sectional- observational	528, 78% Male, 22% Female	60% White; 28% Black; 12% Hispanic or Other	Median = 41	Single item: Aerobic exercise in the last 30 days, yes/no	3-day recall	Having engaged in aerobic exercise in the last 30 days was associated with better ART adherence
Littlewood et al. (2008)	USA	Cross-sectional- observational	221, 56% Male, 44% Female	42% AA; 46% White; 4% Native American; 4% Asian/Pacific; 4% Other	40	Two items on past week frequency of moderate and vigorous exercise	4-item, 7-day recall, assessed frequency of missed and late does and frequency of ignoring instructions	Null findings
Morowatisharifabad et al. (2019)	Iran	Cross-sectional- observational	122, 53% Male, 47% Female	N/A	41.9 (9.5)	Baecke and AIDS Clinical Trials Group	30 and 7-day recall	Null findings
Pellowski and Kalichman (2016)	USA	Cross-sectional- observational	437, 69% Male, 31% Female	92.7% Black, 4.1% White, 1.4 % Hispanic/Latino and Other, .2 % Asian/Pacific Islander	46 (7.8)	Single item on exercise in the past 30 days for health reasons	Unannounced pill counts via a phone call once per month for 3 months	Those who exercised more in the past 30 days to stay healthy were more adherent to ART
Bhochhibhoya et al. (2019)	China	Cross-sectional- observational	658, 100% Male	100% Asian	37.8 (5.7)	IPAQ short form	30 and 3-day recall	Total weekly exercise and vigorous intensity exercise was associated with better ART adherence
Wimberly et al. (2020)	USA	Secondary analysis of a longitudinal RCT, Hatha yoga, 1 90-minute yoga session a week for 12 weeks	73, 69% Male, 29% Female, 2% Transgender	78% Black, 9% White, 8% Multiple, 5% Latino	44.5 (10.5)	Weekly Yoga Classes	Number of 30-day prescription refills in the last 90 days	Null findings
Webel et al. (2016)	USA	Cross-sectional- observational	102, 53% Male, 47% Female	84% Black, 10% White, 2% Hispanic/Latina, 1 % Native American/Indian, 3% Other	48 (8.7)	Pen and Paper diary, four questions about exercise in the past week that alluded to whether or not the participant performed any, the duration of the session, the intensity, and the type	3-day recall	Null findings
Dang et al. (2018)	Vietnam	Cross-sectional- observational	1133, 60% Male, 40% Female?	N/A	35.5 (6.9)	IPAQ	two steps: visual analog scale in the last month, and ART recall over the last 4 days	Those who exercised less had worse ART adherence

Discussion

The purpose of this review was to systematically identify and review the peer-reviewed literature that has examined the relationship between exercise and ART adherence in adult PLWH. Nine studies met full inclusion criteria and were included in this review. The majority of studies were conducted in the United States and included mostly middleaged (mean = 41.4 years, SD = 8.3) males (77%) who were currently receiving HIV care, including currently being on ART. The majority of included participants had less than a $10,000 income and a high school education. The majority (five out of nine) of studies found a statistically significant and positive association between exercise and ART adherence, such that individuals who exercised more were more adherent to their medication regimen. Potential mental health and substance use mediators were highlighted but given the observational design of the majority of studies, warrant further investigation in future randomized clinical trials. Additional limitations include a heavy focus on middleaged males and the use of self-report measures to assess exercise and ART adherence. While only a small number of studies were included, these studies provide promising findings to suggest that exercise may be an intervention that merits examination in future research studies. The implications for practice and future work in this area as well as the limitations of the current review are discussed in further detail below.

Implications for practice

In addition to providing initial evidence that exercise is associated with ART adherence, the studies included in this review have provided some insight into the potential mechanisms that support this relationship. For example, results from two studies (Blashill et al., 2013; Bhochhibhoya et al., 2019) suggest that depression mediates the relationship between exercise and ART adherence, with one of these studies also showing that depression mediated the relationship between exercise and viral load (Blashill et al., 2013). Thus, higher levels of exercise can predict an improvement in ART adherence and a decrease in viral load through a reduction in depressive symptoms. ART adherence in another study was also associated with depression, but also with scores on the Drug Abuse Screening Test (DAST-10) which assesses drug-abuse related problems in the last year (Littlewood et al., 2008). In one study, ART adherence was associated with both overall functioning quality of life and life satisfaction quality of life (Webel et al., 2016), which could be improve by exercise. Two other studies found that alcohol use was negatively associated with ART adherence (Kyser et al., 2011; Pellowski and Kalichman, 2016), with one of these studies also showing that stress was negatively associated with ART adherence (Pellowski and Kalichman, 2016). Although the majority of this evidence is cross-sectional and prevents directional inferences, these findings align with the literature showing that mental health, substance use, and quality of life are consistently and strongly associated with ART nonadherence (Do et al., 2014; Reisner et al., 2009) and supports future exercise trials that target these potential mechanisms to improve ART adherence.

The studies included in this review do not provide any evidence for any specific type of exercise protocol that might be optimal to improve ART adherence. However, other reviews (Dudgeon et al., 2004; Grace et al., 2015; Jaggers and Hand, 2016; Nosrat et al., 2017; O'Brien et al., 2010) on the effects of exercise interventions in adult PLWH have provided some guidance for future trials. For example, after the individual is medically cleared, exercise (either aerobic or resistance exercise) that takes place on 2 to 4 days each week separated by one or two rest days would be ideal, as this would allow any muscle soreness from one session to resolve prior to the start of the subsequent session. Additionally, moderate to high intensity exercise seems to be more effective than low-intensity, and sessions ranging from as short as 20 minutes to as long as 90 minutes also seem to be effective at improving HIV-related outcomes. These reviews also recommended that as session intensity increases, session duration should decrease. Other studies have showed that supervision could be an important factor as this would allow the exercise professional to help correct technique, fix any problems with equipment, and motivate the participant if they are feeling down (Heissel et al., 2019). Following these guidelines may be an efficacious first step towards refining an optimal exercise protocol to examine its effects on ART adherence.

Limitations of existing studies

There are several limitations worth noting. First, the inability to develop an optimal exercise protocol for ART adherence is precipitated by the pervasiveness of self-report exercise questionnaires. For example, several of the included studies asked only two questions, or even one question in some cases, about whether or not the individual has exercised in the past month or week. In addition to the known limitations of self-reporting exercise behaviors (Sallis and Saelens, 2000), recalling exercise behaviors over the past month and week may be difficult for PLWH particularly because of the existing responsibilities of taking numerous medications, and attending to doctors’ visits, a job, and a family. Moreover, PLWH are three times more likely to be depressed compared to the general population which can be associated with cognitive and memory deficits (Do et al., 2014). Objective measures such as accelerometry are increasingly feasible with smartphones and watches, but reliable and valid self-report questionnaires should also be considered such as the IPAQ (Hagströmer et al., 2006), Modifiable Activity Questionnaire (Pettee Gabriel et al., 2011), and the Godin Leisure-Time Exercise Questionnaire (Amireault and Godin, 2015) to provide other detailed information on the most effective exercise type and intensity.

Along similar lines, the majority of the studies included in this review used self-report measures to recall ART adherence over the last month, week, or even shorter time period. Similar to the limitations mentioned in the previous paragraph about recalling exercise behaviors, recalling how many (if any) doses missed over an extended period of time may be difficult and could present issues with reliability (Blashill et al., 2013; Littlewood et al., 2008; Webel et al., 2016). Additionally, it’s possible that weekly (or less frequent) recall is not a true representation of an individual’s ART adherence. For example, an individual may have taken 100% of their doses in the last week but may have been on a “medication vacation” the previous week. To more accurately measure ART adherence, even cross-sectionally, the percentage of pills taken from an existing prescription could be used. Alternatively, other studies have assessed participant’s most recent viral load or CD4+ count as an objective measure of adherence (Reisner et al., 2009). Mobile medication monitoring devices (i.e. Wisepill) are also an option (Craker et al., 2019; Pellowski et al., 2014) but may present barriers to use (e.g. cost, scalability, acceptability, being constantly monitored) for both researchers and PLWH.

Additional limitations include the genderrelated homogeneity (77% male) of the aggregated sample, the average age of participants being 41.4 (SD = 8.3) years despite data showing that young adults and adolescents have worse ART adherence compared to older PLWH (Closson et al., 2019), and the lack of theoryguided interventions. Moreover, seven of the nine studies were conducted in the United States which may have affected the conflicting results in this review. For instance, the health care system, personal resources, and heightened levels of stigma toward HIV in countries outside the United States (Gari et al., 2013) may prevent some PLWH from seeking or adhering to an HIV care program in general, or one that includes exercise program more specifically. Lastly, it is worth noting that the five studies to have significant findings included sample sizes that were more than twice as large as the studies to have null findings. These discrepant results could be a potential artifact of the larger sample sizes but remains speculative, as no effect sizes were reported on the direct effects of exercise on ART adherence. Thus, we suggest caution in drawing conclusions from studies with null findings as small sample sizes and low session attendance relative to studies with larger samples and statistically significant associations makes it difficult to illuminate potential associations.

Future directions and conclusions

Although exercise may be an effective strategy to improve factors linked to ART adherence, it is not a stand-alone intervention that will circumvent all barriers to optimal ART adherence (e.g. cost and transportation to and from the pharmacy or doctor’s office). Thus, a multidisciplinary team-based multi-modal approach that incorporates several strategies (including exercise) using various delivery platforms may be more effective compared to a single-approach to better mitigate challenges and setbacks to adherence (Simoni et al., 2010). For example, an mHealth (i.e. mobile health) based exercise intervention that fosters online social support while providing an open- and secure based platform for interaction with physicians (regarding medication or any other medical-based concerns) would circumvent the obtrusive barriers (e.g. transportation, finances, and fatigue) to attending exercise intervention sessions and doctor’s appointments. Within these mHealth applications, workouts and progress could be tracked, intensity could be continually assessed through heart rate monitors, and leaderboards could be updated daily to cultivate a sense of friendly competition. Tips on sexual health and reminders to take medication could also be included. These types of interventions would also be scalable and would allow participants to exercise when and where it is most convenient, which may reduce the high levels of attrition seen in most exercise studies with PLWH (Nosrat et al., 2017; Wimberly et al., 2020). Although attrition was not a concern in the studies included in this review because eight of the nine studies were observational, one study found that participants only attended an average of four of 12 yoga sessions over the course of 3 months (Wimberly et al., 2020). Taking steps to reduce attrition will further our understanding of how (and how much) exercise is optimal to increase ART adherence, and may reduce bias in the current findings, as many of the positive findings may be only be valid for a unique subset of PLWH who are able to adhere to interventions and fully complete surveys (Nosrat et al., 2017). Still, PLWH may encounter some barriers to mHealth interventions such as accessibility to a cell phone and a cell phone plan, wireless internet, and physical ability to perform the intervention without on-site supervision. As such, these potential barriers warrant consideration during the design and development of future interventions.

Conclusion

The results of this systematic review suggest that exercise may be an effective strategy to support improved ART adherence in adult PLWH through factors linked to nonadherence such as mental health and substance use problems, as well as more generalized healthy lifestyles and self-efficacy for healthy behaviors. However, there are several limitations within the current body of work that should be addressed to further our understanding of this relationship. Recommendations include developing more theoretically driven, tightly controlled, randomized controlled trials that include all genders of all ages and racial/ethnic backgrounds. Further work should also elucidate the optimal exercise protocol (i.e. frequency, intensity, time, type) and whether exercise could be an effective adjunct to a multidisciplinary team-based approach aimed at improving ART adherence in adult PLWH.