Impact Statement
Despite the high prevalence of common mental disorders and the availability of evidence-based interventions, access to treatments remains limited and adherence rates are often suboptimal, posing significant challenges to improving mental health outcomes worldwide.
Our findings underscore the potential of adherence-focused interventions not only to enhance treatment engagement but also to yield positive clinical outcomes, contributing to overall improvements in mental health and well-being. By integrating these strategies into psychological treatment services, we hope that therapists can better engage with and support their patients, potentially leading to improved mental health outcomes on a global scale. Additionally, the review discusses various gaps in evidence like the lack of focus on non-patient-related dimensions of adherence (e.g., social, economic, health systems related), the absence of studies conducted in low- and middle-income countries and the scope of scalability or wider implementation of the effective interventions.
Background
Common mental disorders (CMDs), including depression, anxiety and somatoform disorders, are among the leading causes of disability worldwide, accounting for 23% of all years lived with disability (Whiteford et al., Reference Whiteford, Ferrari and Degenhardt2016). The Global Burden of Disease study estimates that depressive and anxiety disorders accounted for the largest proportions of disability-adjusted life-years (DALYs) related to mental disorders in 2019, contributing 37% and 23% of the total DALYs, respectively (Murray, Reference Murray2022). Evidence-based psychological treatments – spanning cognitive, behavioural and interpersonal treatment elements – are effective in treating these disorders across a wide range of contexts (Patel et al., Reference Patel, Chisholm, Dua, Laxminarayan and Medina-Mora2016). While psychological treatments generally have significant therapeutic effects (Singla et al., Reference Singla, Kohrt, Murray, Anand, Chorpita and Patel2017; Munder et al., Reference Munder, Flückiger, Leichsenring, Abbass, Hilsenroth, Luyten, Rabung, Steinert and Wampold2019), the duration and frequency of contact required for clinically meaningful improvement can vary depending on the treatment (Hansen et al., Reference Hansen, Lambert and Forman2002; Robinson et al., Reference Robinson, Delgadillo and Kellett2020).
According to the World Health Organisation (WHO), patient adherence is defined as “the extent to which a person’s behaviour – taking medication, following a diet and/or executing lifestyle changes – corresponds with agreed recommendations from a healthcare provider” (Sabaté, Reference Sabaté2003). The WHO’s Adherence to Long-Term Therapies Project outlines five dimensions that impact patient adherence: (1) social- and economic-related factors (e.g., poverty, illiteracy); (2) health system and healthcare team-related factors (e.g., short consultations, overworked healthcare workers); (3) therapy-related factors (e.g., previous treatment failures, immediacy of beneficial effects); (4) condition-related factors (e.g., severity of symptoms, level of disability) and (5) patient-related factors (e.g., anxiety about possible adverse effects, low motivation) (Sabaté, Reference Sabaté2003). These dimensions challenge the traditional notion that patient adherence is solely a patient-driven problem and provide a framework to guide the design and implementation of strategies for promotion of adherence to long-term therapies.
Patient adherence to a minimally effective dose of therapy remains challenging; 20–47% patients drop out from therapy before recovering from the mental health problems that first led them to seek care (Wierzbicki and Pekarik, Reference Wierzbicki and Pekarik1993; Swift and Greenberg, Reference Swift and Greenberg2012; Fernandez et al., Reference Fernandez, Salem, Swift and Ramtahal2015). Apart from reduced treatment efficacy, poor patient adherence to psychological treatment also leads to empty appointment slots and inefficient resource allocation, ultimately resulting in the reduced cost-effectiveness of providing mental health services (Barrett et al., Reference Barrett, Chua, Crits-Christoph, Gibbons and Thompson2008; Bosworth, Reference Bosworth2010). Furthermore, receiving a suboptimal ‘dose’ of psychological treatment may exacerbate symptoms, leading to either increased or decreased future use of mental health services than would have otherwise occurred (Walitzer et al., Reference Walitzer, Dermen and Connors1999). Finally, poor patient adherence can result in therapist burnout and workforce turnover by increasing workload and wastage of time (Pekarik, Reference Pekarik1985; Piper et al., Reference Piper, Ogrodniczuk, Joyce, McCallum, Rosie, O’Kelly and Steinberg1999). These inefficiencies tend to have the highest impact in community settings where human and financial resources are already limited (Walitzer et al., Reference Walitzer, Dermen and Connors1999; Oldham et al., Reference Oldham, Kellett, Miles and Sheeran2012).
Previous reviews of interventions to improve patient adherence to psychological treatments were not systematic and examined observational or case studies (Walitzer et al., Reference Walitzer, Dermen and Connors1999; Barrett et al., Reference Barrett, Chua, Crits-Christoph, Gibbons and Thompson2008; Bosworth, Reference Bosworth2010); and those that systematically synthesised RCTs of such interventions are now more than 10 years old (Oldham et al., Reference Oldham, Kellett, Miles and Sheeran2012). Additionally, most reviews examining patient adherence have disproportionately focused on interventions to enhance psychotropic medication adherence for serious mental illness (Barkhof et al., Reference Barkhof, Meijer, De Sonneville, Linszen and De Haan2012; Steinkamp et al., Reference Steinkamp, Goldblatt, Borodovsky, LaVertu, Kronish, Marsch and Schuman-Olivier2019). In contrast, the primary aim of our review is to systematically examine the existing literature on interventions targeting patient adherence to psychological treatments for CMDs. Specifically, our objectives are to (a) synthesise evidence on the effectiveness of these interventions in enhancing patient adherence to psychological treatment; (b) describe the implementation (content, delivery methods, acceptability, feasibility) of these interventions and (c) summarise the secondary impact of these interventions on clinical outcomes related to mental health symptoms, well-being, disability and quality of life.
Methods
Our systematic review is reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement (Page et al., Reference Page, McKenzie, Bossuyt, Boutron, Hoffmann, Mulrow, Shamseer, Tetzlaff, Akl, Brennan, Chou, Glanville, Grimshaw, Hróbjartsson, Lalu, Li, Loder, Mayo-Wilson, McDonald, McGuinness, Stewart, Thomas, Tricco, Welch, Whiting and Moher2021). The protocol was developed a priori and registered on PROSPERO (CRD42021266680).
Eligibility criteria
Peer-reviewed primary research publications in English were included. There were no restrictions on geographical location or year of publication. Eligible study populations included adults (>18 years) diagnosed with any type of CMD (depression, anxiety and somatoform disorders) (Lund et al., Reference Lund, Breen, Flisher, Kakuma, Corrigall, Joska, Swartz and Patel2010). Only randomised controlled trials (RCTs) and non-RCTs (nRCTs) were included.
An intervention was eligible if it was designed and tested specifically for improving patient adherence to a psychological treatment for CMDs. Interventions designed solely to enhance treatment initiation were excluded as patient non-adherence by failing to continue or complete psychological treatment is a distinct construct from that of rejecting psychological treatment, seen when an individual does not attend their initial treatment session (Swift and Greenberg, Reference Swift and Greenberg2012). Interventions not designed or evaluated to directly target patient adherence to a psychological treatment, but which led to improved adherence as an indirect/secondary outcome or psychological treatments with adherence components within them, were excluded. Interventions designed or evaluated to target adherence to pharmacological treatment or digital psychological treatment, were excluded as they are fundamentally different in nature (Ludden et al., Reference Ludden, Van Rompay, Kelders and Van Gemert-Pijnen2015; Gast and Mathes, Reference Gast and Mathes2019; Lippke et al., Reference Lippke, Gao, Keller, Becker and Dahmen2021).
The primary outcome of interest was adherence to any kind of psychological treatment, that is, individual or group, delivered independently or in combination with pharmacological treatments, and delivered in-person or by humans via telecommunication technologies (e.g., tele-counselling). Given our definition of patient adherence, studies that measured adherence outcomes after the completion of an initial session of psychological treatment up until the completion of a final session, defined by either the fixed endpoint of a treatment protocol or by mutually agreed discharge between patient and provider, were considered. The primary outcome of interest was any objective measure of patient adherence to psychological treatment, including appointment attendance, homework compliance and indicators of treatment adherence (such as uptake, engagement, motivation, utilisation, participation, completion or retention) or nonadherence (such as treatment discontinuation, dropout, withdrawal, attrition, interruption or premature termination). Secondary outcomes of interest were clinical outcomes related to mental health symptoms, well-being, disability, and quality of life and implementation outcomes.
Search strategy
Six electronic databases were searched: MEDLINE, PsycINFO, Embase, Global Health, the Cumulative Index to Nursing and Allied Health Literature and the Cochrane Central Register of Controlled Trials. The search was first conducted in July 2021 and subsequently updated in February 2023, using search terms under the following concepts: CMDs (e.g., “major depressive disorder”), patient adherence (e.g., “compliance”), psychological treatments (e.g., “cognitive behavioural therapy” [CBT]) and controlled trial study design (e.g., “RCT”). The detailed search strategy for MEDLINE can be found in Supplementary Material, Appendix 1.
Study selection and data extraction
Search results from all electronic databases were merged and imported into EndNote for removal of duplicates. After automatic and manual de-duplication, the remaining papers were imported to Covidence, an online software for managing systematic reviews. Papers were also manually screened for duplicates on the Covidence platform. Two pairs of reviewers (SR and RB, BB and LF) independently screened all titles and abstracts as well as full texts for eligibility; and conflicts were resolved by a third reviewer (AG or AN). For the title and abstract screening, consensus was reached for 93% and 95% of publications for the two pairs of reviewers, respectively. For the full-text screening, consensus was reached for 89% and 91% of studies for the two pairs of reviewers respectively.
Forward and backward citation chaining of included studies was conducted using the Web of Science and Google Scholar to identify any additional eligible studies not identified by our primary search. A data extraction form was developed a priori on MS Excel to collect data relevant to the objectives of this review. Data were extracted by two pairs of researchers (SR and RB, BB and LF). Inter-rater reliability among raters for data extraction, as measured by Cohen’s Kappa was κ=0.81 and κ=0.84 for the two pairs, respectively.
Data analysis
Due to the heterogeneity in intervention content and outcomes evaluated in the included studies, a narrative synthesis was conducted (Popay et al., Reference Popay, Roberts, Sowden, Petticrew, Arai and Rodgers2006). This involved a descriptive analysis of the studies included in the systematic review, using a textual approach to summarise and explain the results of the synthesis (Popay et al., Reference Popay, Roberts, Sowden, Petticrew, Arai and Rodgers2006). Intervention components were categorised under common themes and the content and effectiveness outcomes were also described. These are presented in a tabular format in Supplementary Material, Appendix 2. Delivery methods, clinical outcomes and implementation outcomes were summarised under separate categories (Proctor et al., Reference Proctor, Powell and McMillen2013).
Quality assessment
Risk of bias was assessed independently using the Cochrane Risk of Bias Version 2 (RoB 2) tool for RCTs and the Cochrane ROBINS-I tool for nRCTs (Sterne et al., Reference Sterne, Hernán, Reeves, Savović, Berkman, Viswanathan, Henry, Altman, Ansari, Boutron, Carpenter, Chan, Churchill, Deeks, Hróbjartsson, Kirkham, Jüni, Loke, Pigott, Ramsay, Regidor, Rothstein, Sandhu, Santaguida, Schünemann, Shea, Shrier, Tugwell, Turner, Valentine, Waddington, Waters, Wells, Whiting and Higgins2016; Sterne et al., Reference Sterne, Savović, Page, Elbers, Blencowe, Boutron, Cates, Cheng, Corbett, Eldridge, Emberson, Hernán, Hopewell, Hróbjartsson, Junqueira, Jüni, Kirkham, Lasserson, Li, McAleenan, Reeves, Shepperd, Shrier, Stewart, Tilling, White, Whiting and Higgins2019). The domains included in RoB 2 are bias arising from the randomisation process, deviations from intended interventions, missing outcome data, measurements of the outcome and selection of the reported result. The domains included in ROBINS-I are bias due to confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement of the outcome and selection of the reported result. To mitigate subjectivity in assessment, two reviewers (BZ, BB) independently answered signalling questions embedded in the ROBINS-I Microsoft Word template and the RoB 2 Excel tool, which uses an algorithm to generate an overall risk of bias based on ratings in each domain. Both reviewers arrived at highly similar ratings in the rating of each domain and overall risk of bias; disagreements were discussed until consensus was reached.
Results
Figure 1 summarises the results of our search. Of the 33,612 reports identified, 12,526 were duplicates. From the remaining, 20,812 papers were excluded at the title and abstract screening stage. In total, 274 full texts were assessed for eligibility and based on our criteria, 20 studies were eligible for inclusion. The forward and backward citation chaining process identified three additional eligible studies, leading to 23 studies being included in our review.
Figure 1. Identification of studies via databases and registers.
Study characteristics
The 23 included studies comprised a total sample size of 2,779 participants, and the sample sizes in the individual studies ranged from 7 to 325 participants. All studies were from high-income or upper-middle-income countries: United States (n = 9), United Kingdom (n = 4), Australia (n = 3) and Canada (n = 3), Austria (n = 1), Sweden (n = 1), China (n = 1) and Chile (n = 1). Of the total, 18 were RCTs and 5 were nRCTs. 16 of the RCTs were individually randomised trials and two were cluster RCTs (Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019; Wang et al., Reference Wang, Pan, Xu and Ding2022). The five nRCTs utilised a variety of participant allocation methods, including self-selection (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020), propensity score matching (Delgadillo and Groom, Reference Delgadillo and Groom2017), scheduling convenience (Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017) and timing of trial enrolment (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998). Comparators or controls in the studies were usual care, enhanced usual care, another adherence intervention, delayed intervention or wait-list control. The characteristics of these studies are summarised in Table 1.
Table 1. Summary characteristics of included studies
Participant characteristics
The mean age of participants across the included studies ranged from 19.6 to 69 years. Depressive disorders (n = 9) were identified based on one of the following: Patient Health Questionnaire (PHQ-9), Hamilton Depression Rating Scale, Beck Depression Inventory and clinician diagnosis. Anxiety disorders (n = 6) included social anxiety disorder, panic disorder and generalised anxiety disorder, and were identified using one of the following: Anxiety Sensitivity Index (ASI), Generalised Anxiety Disorder-7, Social Interaction Anxiety Scale, State-Trait Anxiety Inventory and clinician diagnosis. Seven studies enrolled participants who had an adjustment/mood disorder, or a CMD comorbid with another condition (substance use, cancer). The oldest study, conducted in 1964, described its participants as having a ‘psychoneurotic disorder’ (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964).
The following section describes the various interventions that were tested to improve adherence to psychological treatments for CMDs.
Interventions
Seven studies involved reminders (Clough, Reference Clough and Casey2014; Delgadillo et al., Reference Delgadillo, Moreea, Murphy, Ali and Swift2015), content reinforcement between sessions (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Alfonsson et al., Reference Alfonsson, Englund and Parling2019; Pérez et al., Reference Pérez, Fernández, Cáceres, Carrasco, Moessner, Bauer, Espinosa-Duque, Gloger and Krause2021) or a combination of both (Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020). Five studies (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998; Westra and Dozois, Reference Westra and Dozois2006; Westra et al., Reference Westra, Arkowitz and Dozois2009; Barrera et al., Reference Barrera, Smith and Norton2016; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019) tested interventions involving principles of motivational communication between therapist and patient. Six studies (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Latour and Cappeliez, Reference Latour and Cappeliez1994; Reis and Brown, Reference Reis and Brown2006; Delgadillo and Groom, Reference Delgadillo and Groom2017; Stein et al., Reference Stein, Shumake, Beevers and Smits2020; Wang et al., Reference Wang, Pan, Xu and Ding2022) tested interventions consisting of pre-therapy orientation to set shared expectations and familiarise patients to the therapy. Two studies (Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003; Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019) implemented clinical case management services to address structural barriers to engaging in therapy for depression. One study (Mohr et al., Reference Mohr, Ho, Duffecy, Reifler, Sokol, Burns, Jin and Siddique2012) aimed to reduce practical challenges to therapy adherence by employing telephone-based delivery.
Two studies (Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018; Jurinec and Schienle, Reference Jurinec and Schienle2020) incorporated unique psychological principles in self-delivered interventions targeting adherence to therapy and its components. One of these studies administered a placebo to the intervention arm, alongside a psychoeducation course to enhance motivation for homework completion (Jurinec and Schienle, Reference Jurinec and Schienle2020). The other study involved implementation intentions to sustain attendance at a group psychoeducation programme (Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018).
All except five interventions (Latour and Cappeliez, Reference Latour and Cappeliez1994; Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Delgadillo et al., Reference Delgadillo, Moreea, Murphy, Ali and Swift2015; Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019; Pérez et al., Reference Pérez, Fernández, Cáceres, Carrasco, Moessner, Bauer, Espinosa-Duque, Gloger and Krause2021) were effective in improving at least one adherence outcome. Sixteen studies also reported a clinical outcome or a measure of well-being. Six studies were effective in improving at least one clinical outcome (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003; Westra and Dozois, Reference Westra and Dozois2006; Delgadillo and Groom, Reference Delgadillo and Groom2017; Jurinec and Schienle, Reference Jurinec and Schienle2020; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020) and three studies were effective in improving at least one measure of well-being or disability (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998; Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003).
All the interventions and their effect on adherence outcomes are discussed in detail under relevant categories below and presented in Supplementary Material, Appendix 2.
Reminders and between-session engagement
Seven studies utilised an automated text-messaging software to deliver reminders (Clough, Reference Clough and Casey2014; Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Delgadillo et al., Reference Delgadillo, Moreea, Murphy, Ali and Swift2015; Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017; Alfonsson et al., Reference Alfonsson, Englund and Parling2019; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020; Pérez et al., Reference Pérez, Fernández, Cáceres, Carrasco, Moessner, Bauer, Espinosa-Duque, Gloger and Krause2021) to patients’ mobile phones at times and dates relevant to their therapy sessions. This included a single text message reminder prior to the first therapy appointment (Delgadillo and Groom, Reference Delgadillo and Groom2017) or multiple texts throughout the week, reminding participants to attend therapy as well as regularly practice cognitive exercises such as meditation taught during therapy (Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020). In addition to reminders, these interventions incorporated text messages that reinforced content taught in therapy. These messages were tailored in collaboration with patients (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017; Alfonsson et al., Reference Alfonsson, Englund and Parling2019) or were interactive (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014). One study evaluated the effectiveness of a blended treatment approach as compared to treatment as usual (TAU) alone (Pérez et al., Reference Pérez, Fernández, Cáceres, Carrasco, Moessner, Bauer, Espinosa-Duque, Gloger and Krause2021). The blended approach comprised of the TAU alongside an internet intervention that ensured between-session engagement by providing psycho-educational information, supportive monitoring and regular feedback (Pérez et al., Reference Pérez, Fernández, Cáceres, Carrasco, Moessner, Bauer, Espinosa-Duque, Gloger and Krause2021).
Out of the seven studies, four reported the intervention being significantly superior to the control in at least one adherence measure: homework completion (Alfonsson et al., Reference Alfonsson, Englund and Parling2019), duration in treatment (Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017), therapy completion rate (Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020) or reduction in dropouts (Clough, Reference Clough and Casey2014). Out of the four studies that reported symptomatic/functional outcomes (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020; Pérez et al., Reference Pérez, Fernández, Cáceres, Carrasco, Moessner, Bauer, Espinosa-Duque, Gloger and Krause2021) only one study reported a statistically significant difference of a clinical outcome (depression symptoms) between trial arms (Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020).
Motivational interviewing
Five studies (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998;Westra and Dozois, Reference Westra and Dozois2006; Westra et al., Reference Westra, Arkowitz and Dozois2009; Barrera et al., Reference Barrera, Smith and Norton2016; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019) tested interventions involving principles of motivational interviewing (MI), a guided counselling style that elicits behavioural change in patients by exploring and resolving ambivalence. Four studies (Westra and Dozois, Reference Westra and Dozois2006;Westra et al., Reference Westra, Arkowitz and Dozois2009; Barrera et al., Reference Barrera, Smith and Norton2016; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019) employed MI methods that were adapted for therapy ‘pre-treatment’ by Westra and Dozois (2003). The pre-treatment was delivered individually by graduate-level clinical psychologists prior to the onset of therapy and ranged from a single pre-treatment session (Reis and Brown, Reference Reis and Brown2006) to four weekly pre-treatment sessions (Stein et al., Reference Stein, Shumake, Beevers and Smits2020). One study utilised a different approach to motivational counselling as described by Hester and Miller (Reference Hester and Miller2003), for substance use disorders and incorporated these principles into the therapy sessions (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964).
Compared to the control, MI pre-treatments were found to significantly improve the number of sessions attended (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998; Barrera et al., Reference Barrera, Smith and Norton2016), client-rated and therapist-rated homework compliance (Westra and Dozois, Reference Westra and Dozois2006; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019) and rate of treatment completion (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998; Westra and Dozois, Reference Westra and Dozois2006). Three of the studies reported no significant differences in improvement of anxiety symptoms between arms (Westra et al., Reference Westra, Arkowitz and Dozois2009; Barrera et al., Reference Barrera, Smith and Norton2016; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019). One study reported significant improvement in anxiety symptoms in the intervention arm compared with the control (Westra and Dozois, Reference Westra and Dozois2006), while another reported reduced rehospitalisations and time spent in in-patient treatment (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998).
Therapy orientation
Six studies (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Latour and Cappeliez, Reference Latour and Cappeliez1994; Reis and Brown, Reference Reis and Brown2006; Delgadillo and Groom, Reference Delgadillo and Groom2017;Stein et al., Reference Stein, Shumake, Beevers and Smits2020; Wang et al., Reference Wang, Pan, Xu and Ding2022) tested interventions to set shared expectations, orient patients to the components of therapy, and practice desirable therapy roles and behaviours in a safe and less personal environment. These interventions were delivered to individuals (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Reis and Brown, Reference Reis and Brown2006; Delgadillo and Groom, Reference Delgadillo and Groom2017; Stein et al., Reference Stein, Shumake, Beevers and Smits2020; Wang et al., Reference Wang, Pan, Xu and Ding2022) or groups (Latour and Cappeliez, Reference Latour and Cappeliez1994; Delgadillo and Groom, Reference Delgadillo and Groom2017) through a single session (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Stein et al., Reference Stein, Shumake, Beevers and Smits2020) or multiple sessions (Latour and Cappeliez, Reference Latour and Cappeliez1994; Reis and Brown, Reference Reis and Brown2006; Delgadillo and Groom, Reference Delgadillo and Groom2017; Wang et al., Reference Wang, Pan, Xu and Ding2022). Each of these interventions combined various components of pre-therapy education, expectation-setting and role-playing. Delivery formats were multimodal: videos showcasing scenarios that mimic desired therapy behaviours (Latour and Cappeliez, Reference Latour and Cappeliez1994; Reis and Brown, Reference Reis and Brown2006), and in-person sessions to learn and practice exercises in completing homework (Stein et al., Reference Stein, Shumake, Beevers and Smits2020). Expectation-setting had to do with the role of patient and therapist in the therapeutic relationship (Ogrodniczuk et al., Reference Ogrodniczuk, Joyce and Piper2005; Wang et al., Reference Wang, Pan, Xu and Ding2022), anticipated difficulties to adherence (Stein et al., Reference Stein, Shumake, Beevers and Smits2020; Wang et al., Reference Wang, Pan, Xu and Ding2022) and time taken by the therapy to take effect (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Reis and Brown, Reference Reis and Brown2006; Wang et al., Reference Wang, Pan, Xu and Ding2022).
Five out of six interventions demonstrated significant improvements in adherence outcomes compared to controls, such as higher homework completion (Stein et al., Reference Stein, Shumake, Beevers and Smits2020), longer treatment duration (Delgadillo and Groom, Reference Delgadillo and Groom2017), fewer dropouts (Reis and Brown, Reference Reis and Brown2006) and increased attendance at sessions (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Delgadillo and Groom, Reference Delgadillo and Groom2017; Wang et al., Reference Wang, Pan, Xu and Ding2022). Out of the three studies that reported clinical and well-being outcomes (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Delgadillo and Groom, Reference Delgadillo and Groom2017; Stein et al., Reference Stein, Shumake, Beevers and Smits2020), one study reported significant differences in improvement of anxiety symptoms between arms (Delgadillo and Groom, Reference Delgadillo and Groom2017). Another study reported significant differences in patient-reported symptom improvement and therapist-rated global improvement between arms (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964).
Case management services
Two interventions (Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003; Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019) attempted to increase adherence to therapy through case management with support in overcoming structural barriers. Patient-centred case management services that assisted older adults (engaging in treatment decisions, scheduling appointments and facilitating transportation) were superior to controls in getting patients to start therapy, but they did not retain them in therapy any better than controls (Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019). Another study which involved clinical case management that supported patients in resolving self-reported problems related to housing, employment and interpersonal relationships, helped improve session attendance and reduce dropout rates in a sub-sample of individuals from low socioeconomic backgrounds (Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003). It also reported significant differences in improvement of depression symptoms and social functioning between arms (Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003).
Telephone-based therapy
One intervention sought to address practical barriers to engaging in therapy for depression in a clinical setting (Mohr et al., Reference Mohr, Ho, Duffecy, Reifler, Sokol, Burns, Jin and Siddique2012). It tested telephone-based therapy against face-to-face CBT and found significant improvements in number of sessions attended and reduced dropout rates. However, significantly greater improvement in clinical outcomes was seen in face-to-face CBT as opposed to telephone-based therapy.
Placebo intervention
One study administered a placebo alongside a CBT-based outpatient psychoeducation course to enhance motivation for homework completion (Jurinec and Schienle, Reference Jurinec and Schienle2020). It was a blue glass bottle and a dropper with 30 ml sunflower oil, labelled ‘golden root oil’ (rhodiola rosea). This placebo label was chosen because rhodiola rosea was widely used as a traditional medicine for several symptoms and disorders (including depression). Intervention group participants were instructed to consume three drops (0.15 ml) of the placebo oil 10 min before the daily relaxation exercise advised in the outpatient course. The intervention was superior to control in increasing homework completion and led to a significantly greater improvement in depression symptoms (Jurinec and Schienle, Reference Jurinec and Schienle2020).
Implementation intention
One study involved implementation intentions which are brief, simple, low-cost, self-regulatory strategies in the form of an “if-then plan” (Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018). These involved anticipating barriers and creating mitigation strategies, to sustain attendance at a group psychoeducation programme. The intervention intended to help patients in downregulating feelings of concern associated with attending large group sessions. It was superior to the control in significantly increasing session attendance and treatment duration.
Delivery of interventions
Mode and setting
Eleven of the interventions were delivered in-person; mainly in university psychiatric clinics (n =5), primary healthcare centres (n = 4) or community health centres (n = 2). Seven of the interventions employed automated text messages. Others involved telephonic delivery (n = 2), a pre-programmed website (n = 1) and delivery via mail (n = 1). One intervention consisting of clinical case management, provided services both in-person at an outpatient mental healthcare centre and over telephone.
Delivery agent
Seven interventions, involving MI and therapy orientation, were delivered by clinical psychologists and doctoral graduate students of psychology. Seven interventions did not involve a delivery agent and relied on computer programmes sending automated text messages to participants or pre-programmed websites doing the same. Other interventions were delivered by a range of practitioners such as neurologists (n = 1), psychiatrists (n = 1), registered nurses (n = 1) and researchers with a post-graduate degree (n = 1). Two interventions were self-delivered (Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018; Jurinec and Schienle, Reference Jurinec and Schienle2020) and two studies did not specify the characteristics of the delivery agent (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998; Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003). One intervention was delivered by both licensed social workers with master’s degrees and doctoral graduate students of psychology (Reis and Brown, Reference Reis and Brown2006).
Implementation outcomes
Five studies reported implementation outcomes. All five studies reported on acceptability of the intervention (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Delgadillo and Groom, Reference Delgadillo and Groom2017; Alfonsson et al., Reference Alfonsson, Englund and Parling2019; Jurinec and Schienle, Reference Jurinec and Schienle2020; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020), defined as the perception among stakeholders that a given treatment or service was agreeable or satisfactory (Proctor et al., Reference Proctor, Landsverk, Aarons, Chambers, Glisson and Mittman2009). Four of the studies based their findings on perspectives of the participants and one explored perspectives of therapists on patient acceptability of the intervention (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014). Three of them explored patient perspectives using Likert scales (Delgadillo and Groom, Reference Delgadillo and Groom2017; Alfonsson et al., Reference Alfonsson, Englund and Parling2019; Jurinec and Schienle, Reference Jurinec and Schienle2020) and two conducted qualitative interviews with therapists (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014) and participants (Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020).
In one study (Alfonsson et al., Reference Alfonsson, Englund and Parling2019), participants rated their perception of the text messages in terms of helpfulness, annoyance, redundancy and value on a scale ranging from 0 (not at all) to 4 (absolutely). The threshold for being “satisfied with the intervention” was considered a score of 50%. Four participants rated the text messages at 50% or above, indicating satisfaction, while three participants rated them as neutral or irrelevant (scoring below 50%). In the open feedback section, participants expressed that the text messages served as helpful reminders for exercises and helped keep the treatment on track between sessions. However, they also felt pressured when unable to complete all assignments and found the text messages lacking personalisation. Another study employed a similar approach, where participants evaluated the effectiveness of a placebo on a 10-point Likert scale, with 10 representing extreme effectiveness. The mean score was 8.19 (SD = 1.15) (Jurinec & Schienle, Reference Jurinec and Schienle2020).
In another study (Delgadillo and Groom, Reference Delgadillo and Groom2017), a brief acceptability questionnaire was designed for the intervention, consisting of 3-point Likert-scale items rated from 0 to 10 (high). Patients rated each session based on relevance, the quality of the presentation and quality of the materials. The average ratings were combined to form a single measure of acceptability, which was then compared across different diagnostic groups. The results showed no significant differences across the diagnostic groups.
In one of the studies (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014), therapists reported that patients initially found the messages useful in therapy but perceived them as less useful over time. In another study (Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020) qualitative interviews were conducted with patients who either declined or accepted smart messaging (n = 15). Among those who declined text messaging in the control group (n = 2), the main reason given was a lack of confidence in using mobile phones. On the other hand, participants who accepted messaging in the intervention group viewed it as a helpful reminder, motivating them to complete exercises and maintain mindfulness practices in their daily lives (Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020). Interviews also revealed a secondary theme of feeling a personal connection, despite being aware that the texts were automated.
Risk of bias
The majority of RCTs assessed using the RoB 2 tool were of high quality, but a third of the studies (n = 6) demonstrated moderate or high (Westra et al., Reference Westra, Arkowitz and Dozois2009; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019; Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019; Stein et al., Reference Stein, Shumake, Beevers and Smits2020; Wang et al., Reference Wang, Pan, Xu and Ding2022) risk of bias. The most common domain of concern was in the selection of the reported result (Clough, Reference Clough and Casey2014; Barrera et al., Reference Barrera, Smith and Norton2016; Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018; Stein et al., Reference Stein, Shumake, Beevers and Smits2020; Pérez et al., Reference Pérez, Fernández, Cáceres, Carrasco, Moessner, Bauer, Espinosa-Duque, Gloger and Krause2021; Wang et al., Reference Wang, Pan, Xu and Ding2022). Some RCTs also demonstrated a moderate or high level of bias in the following domains: allocation concealment or randomisation (Reis and Brown, Reference Reis and Brown2006; Wang et al., Reference Wang, Pan, Xu and Ding2022), deviations from intended interventions (Westra et al., Reference Westra, Arkowitz and Dozois2009; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019), outcome measurement (Reis and Brown, Reference Reis and Brown2006; Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019) and missing outcome data (Stein et al., Reference Stein, Shumake, Beevers and Smits2020).
While the nRCTs included in the review were generally of high quality, some studies (Furber et al., Reference Furber, Jones, Healey and Bidargaddi2014; Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017; Wells et al., Reference Wells, Malins, Clarke, Skorodzien, Biswas, Sweeney, Moghaddam and Levene2020) had moderate risk of bias due to confounding. One study (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998) lacked enough information to be assessed for bias due to confounding, deviations from intended interventions and selection of reported results. The risk of bias assessments of all studies can be found in Supplementary Material, Appendix 3.
Discussion
We conducted this systematic review to examine interventions designed to improve patient adherence to psychological treatment for CMDs. We identified eight groups of strategies that researchers have employed across various settings to improve adherence to psychological treatment. The heterogeneity of how adherence outcomes are defined and measured, as well as the diversity in patient population, selection methods, follow-up periods, content and delivery of interventions tested, prevent us from drawing comparative conclusions about the effectiveness of specific types of adherence interventions. However, while there are not sufficient similarities among these findings to warrant statistical pooling of the results, it is important to note the trends of the different interventions. For instance, 9 out of 10 studies (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998; Reis and Brown, Reference Reis and Brown2006; Westra and Dozois, Reference Westra and Dozois2006; Westra et al., Reference Westra, Arkowitz and Dozois2009; Barrera et al., Reference Barrera, Smith and Norton2016; Delgadillo and Groom, Reference Delgadillo and Groom2017; Peters et al., Reference Peters, Romano, Byrow, Gregory, McLellan, Brockveld, Baillie, Gaston and Rapee2019; Stein et al., Reference Stein, Shumake, Beevers and Smits2020) that provided some form of preparation to patients on the practical and theoretical aspects of therapy participation were found to be effective in improving at least one adherence measure. These preparation strategies included MI, role and relationship induction and practical expectations surrounding therapy.
Therapy orientation process has been previously described as an opportunity for patients to acquire information, observe and participate in experiences and then develop skills that link the information and experience in the mind of the patient (Piper et al., Reference Piper, Ogrodniczuk, Joyce, McCallum, Rosie, O’Kelly and Steinberg1999). Research shows that this information allows patients to set realistic expectations and evokes feelings of satisfaction and hope when those expectations are met in the process of therapy (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964). Additionally, it reduces the apprehension and anxiety around therapy, which may otherwise lead to counterproductive behavioural patterns like avoidance or resistance (France and Dugo, Reference France and Dugo1985).
Similarly, after employing MI style communication between therapist and patient, Barrera and colleagues saw improvement in motivation to change (Barrera et al., Reference Barrera, Smith and Norton2016), which is necessary for action-oriented treatments such as CBT. By linking provided information with an experience, a patient might be able to change the thinking that interrupts the connection of knowledge (“I should attend therapy”, “therapy will help me”) with action (attending therapy).
Reminder-based interventions may play a different role in improving adherence. Older studies have suggested that appointment reminders (in the form of letters and phone calls) can reduce therapy dropout rates (Hochstadt and Trybula, Reference Hochstadt and Trybula1980; Sherman and Anderson, Reference Sherman and Anderson1987; Swenson and Pekarik, Reference Swenson and Pekarik1988) but studies on text messaging are still relatively new. This review suggested that, in combination with supportive monitoring and content reinforcement messages, these interventions provide a personal connection that makes patients feel supported and motivated, strengthen the therapeutic alliance and provide longitudinal engagement that increases the likelihood that a person will return to therapy even after a period of absence.
Case Study 1: Automated text messages
In the US-based study by Aguilera et al. (Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017), five types of automated text messages were sent in Spanish: (a) daily mood rating prompt (e.g., “What is your mood right now on a scale of 1 to 9?”); (b) Feedback based on mood ratings (e.g., “Treat yourself with kindness, the same way you would treat a loved one”); (c) daily message reinforcing live therapy content (e.g., “Do at least one new pleasant activity this week”); (d) weekly reminder to attend psychotherapy sent the night before session (e.g., “Are you coming to group this week?”); (e) a monthly opt-out message to terminate message delivery, if desired (e.g., “To stop receiving these messages, respond with the word STOP”). Text messages were delivered multiple times a day, and the patients were allowed to customise the time of day to receive reminders. At the start of each therapy session, patient’ mood data were displayed on a whiteboard to evaluate their mood states over the past week and apply therapeutic tools to specific events. This intervention was administered for the full duration of the 16-week manualised group CBT at a behavioural health outpatient clinic in an urban public hospital, targeting a sample of low-income Spanish-speaking Latino patients.
Given that attendance is necessary for in-person therapy, any intervention that improves session attendance should be critical for therapy outcome (Hansen et al., Reference Hansen, Lambert and Forman2002; Robinson et al., Reference Robinson, Delgadillo and Kellett2020). While Robinson and colleagues established a curvilinear relationship between number of sessions and response to treatment from naturalistic studies, our synthesised findings from clinical trials do not necessarily corroborate this idea. Out of the eight studies that found a significant improvement in session attendance (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998; Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003; Mohr et al., Reference Mohr, Ho, Duffecy, Reifler, Sokol, Burns, Jin and Siddique2012; Barrera et al., Reference Barrera, Smith and Norton2016; Delgadillo and Groom, Reference Delgadillo and Groom2017; Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018; Wang et al., Reference Wang, Pan, Xu and Ding2022), only three found a correlating and significant improvement in clinical outcomes (Hoehn-Saric et al., Reference Hoehn-Saric, Frank, Imber, Nash, Stone and Battle1964; Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003; Delgadillo and Groom, Reference Delgadillo and Groom2017). This may reflect the possibility of a more nuanced link between treatment dosage and treatment outcomes.
Case Study 2: Motivational interviewing
One study conducted in the United States (Daley et al., Reference Daley, Salloum, Zuckoff, Kirisci and Thase1998) integrated dual disorder recovery counselling and motivational therapy. This integrated treatment strategy targeted both substance use and common mental health disorders, aiming to equip patients with the skills to foresee and manage challenges that hinder adherence by leveraging their own internal resources for change. The intervention employed the FRAMES acronym, which stands for: providing Feedback on substance use and mood problems, emphasising the patient’s Responsibility for change, offering Advice on how to change, presenting a Menu of change options, showing Empathy towards the patient and fostering Self-efficacy or confidence in their ability to change. The study applied these principles to individuals who were receiving outpatient care for depression and cocaine dependence after being discharged from a university psychiatric hospital’s dual diagnosis in-patient unit, during individual and group CBT sessions over 4 weeks.
However, it is difficult to make any such assertion without a consistent definition of adherence among researchers and clinicians. There is also a lack of a robust research base with large prospective studies capturing the effect of adherence interventions on both adherence outcomes and clinical outcomes. In addition, frequently assessing patient perceptions (their expectations, motivations and progress) throughout treatment may further illuminate why patients continue to attend therapy- and perceive-related benefits.
It is important to note that all the studies found in this systematic review tested interventions dealing with patient-related factors of adherence. These interventions are critical for working within a patient’s ability to access internal resources of knowledge, skills and beliefs that engage or deter a patient from therapy. However, it is misleading to believe that patients are solely responsible for adherence to therapy. An evidence base that neglects the other factors affecting a person’s capacity to adhere to psychological treatment will only partially resolve the issue of poor session attendance and completion rates (Sabaté, Reference Sabaté2003). Relatively less research has been conducted globally on the other dimensions of adherence as defined by WHO’s framework on interacting dimensions affecting adherence (Sabaté, Reference Sabaté2003), including factors related to the social and economic status of the patient, therapist characteristics, the severity and progression of the patient’s condition, presence of comorbidities and the complexity and duration of therapy itself (Peh et al., Reference Peh, Kwan, Goh, Ramchandani, Phang, Lim, Loh, Østbye, Blalock, Yoon, Bosworth, Low and Thumboo2021).
The benefit of intervening in other dimensions can be seen in a limited number of studies included in this review (all conducted in United States). For example, Mohr et al. demonstrated the early improvement in session attendance and dropout rate by switching from in-person to telephone-based therapy (Mohr et al., Reference Mohr, Ho, Duffecy, Reifler, Sokol, Burns, Jin and Siddique2012). This intervention reduces the logistical and motivational barriers to physical attendance at a clinic. Miranda et al.’s case management approach improved session attendance and dropout rates (Miranda et al., Reference Miranda, Azocar, Organista, Dwyer and Areane2003). It worked on social- and economic-related dimensions by addressing other areas of difficulty (housing, employment, relationships) that could interfere with or disrupt consistent treatment. Several such strategies working in parallel and targeting multiple dimensions of adherence may work to improve adherence to psychological treatment (Raue et al., Reference Raue, Schulberg, Bruce, Banerjee, Artis, Espejo, Catalan and Romero2019).
Case Study 3: Therapy orientation
One study from the United Kingdom used pre-treatment lecture style seminars (Delgadillo and Groom, Reference Delgadillo and Groom2017) that were delivered in a group setting by CBT practitioners, to patients diagnosed with anxiety or depressive disorders. The seminar content included: (a) clarification on the roles of the patient and therapist; (b) psychoeducation on how transdiagnostic processes (thinking, behaviour, attention and memory) contribute to psychological distress and (c) an overview of generic aspects of CBT and common misconceptions related to it. Each seminar revolved around one central theme; such as problematic thinking processes, behavioural change and emotion regulation. The seminars used accessible language and materials (slides, booklets and videos) along with exercises that emphasised personal experiences and self-reflection. Participants had the opportunity to ask questions and engage in brief discussions about the content. The three weekly seminars of 90 min were followed by 20 sessions of individual CBT.
It is noteworthy to mention that all studies included in this review were conducted in high-income or upper-middle-income countries. The lack of studies testing the included adherence strategies in LMIC contexts limits the generalisability of our findings, especially when differences in health systems and social and economic factors in LMICs may influence how similar interventions will exert their effects. Additionally, the inclusion of only two studies with minority populations (Aguilera et al., Reference Aguilera, Bruehlman-Senecal, Demasi and Avila2017; Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018) limits drawing conclusions about the applicability of our findings to diverse contexts and populations. Recent advances in non-specialist-delivered psychosocial treatments and task sharing in LMICs (Singla et al., Reference Singla, Kohrt, Murray, Anand, Chorpita and Patel2017) could mean “adherence to treatment” would look significantly different from the strategies reported in this review since all the included studies had specialists as delivery agents. Adherence strategies applied for other chronic illness treatments (e.g., tuberculosis, cardiovascular diseases) in LMICs (Pradipta et al., Reference Pradipta, Houtsma and van Boven2020; Loots et al., Reference Loots, Goossens, Vanwesemael, Morrens, Van Rompaey and Dilles2021; Ogungbe et al., Reference Ogungbe, Byiringiro, Adedokun-Afolayan, Seal, Dennison Himmelfarb, Davidson and Commodore-Mensah2021) emphasise home-based, group-based or peer-led approaches, such as home visits by community agents and peer-support groups to enhance adherence. Other reviews focusing on adherence in LMIC contexts (Pradipta et al., Reference Pradipta, Houtsma and van Boven2020; Loots et al., Reference Loots, Goossens, Vanwesemael, Morrens, Van Rompaey and Dilles2021; Boima et al., Reference Boima, Doku, Agyekum, Tuglo and Agyemang2024) also emphasise multicomponent interventions that integrate treatment initiation with adherence strategies (e.g., patient education and home visits by healthcare workers), social with psychological interventions (e.g., monetary incentives with MI) or digital with in-person interventions (e.g., SMS reminders and nurse follow-up).”
Evidence also suggests that multi-level interventions, such as combining individual therapies with community-based or facility-based programmes, can be effective in LMICs (Reif et al., Reference Reif, Abrams, Arpadi, Elul, McNairy, Fitzgerald and Kuhn2020; Pugh et al., Reference Pugh, Roberts, Viswasam, Hahn, Ryan, Turpin, Lyons, Baral and Hansoti2022). Therapy orientation and content reminders, which have proven effective in HIC settings, also show promise in LMIC settings when combined with additional support such as case management, which targets non-patient-related barriers (structural challenges of clinic-based care, and poor provider relationships). Innovative group-based approaches like “adherence clubs” offer valuable opportunities to provide comprehensive services, including education and information on additional support resources like addressing food insecurity. However, further research is needed to tailor these interventions to address the complex, non-patient-centric aspects of treatment adherence, ensuring they cater to the specific needs of the target population and context.
Further research is needed to address the methodological limitations of currently published findings, as well as the conceptual limitations surrounding the definition and measure of adherence. We hope this review highlights the gap in intervention studies targeting social, economic, health system, healthcare team and therapeutic factors as the causes of problems with adherence to psychological treatment. A uniform yet inclusive understanding and measure of adherence to psychological treatment can serve practical purposes for research and policymaking and will maximise the benefit of therapy for patients in clinical settings.
This review further adds to the existing literature base on adherence to psychological treatments, a subject about which there is considerable discussion but little consensus or progress. To our knowledge, only one prior systematic review (Oldham et al., Reference Oldham, Kellett, Miles and Sheeran2012) has assessed the literature for interventions that improve adherence to psychological treatment. We believe that our review has relative strengths compared to the previous review in terms of choosing a broader definition of adherence, searching a higher number of databases for studies, including a wider range of studies, and acknowledging the different approaches of adherence needed for CMDs.
There are a number of limitations to our findings and review process. A broad conceptualisation of adherence inevitably means procuring a heterogeneity of outcome measures, definitions of those measures, follow-up periods, delivery methods, intervention content, intervention duration and psychological treatments itself. While we believe this proved useful in illustrating the landscape of research currently in existence on the topic, we recognise this does not provide sufficient consistency for meta-analysis or definitive conclusions on the “most effective” strategy. We did not search grey literature, publications in languages other than English, and regional databases, which may bias our results and could have excluded studies conducted in LMICs. The quality assessment of the studies did not impact the weight given to them in the narrative synthesis. Finally, the sample sizes across studies varied significantly, with some included studies having as few as seven and eight individuals, which decreases the precision and generalisability of the results (Avishai et al., Reference Avishai, Oldham, Kellett and Sheeran2018; Alfonsson et al., Reference Alfonsson, Englund and Parling2019).
We hope that our review will orient researchers, policymakers, clinic administrators and care providers to the array of interventions that have been tested to improve psychological treatment adherence for CMDs. Integrating adherence strategies in mental health services is crucial for enhancing treatment effectiveness and patient outcomes for individuals with CMDs. By incorporating these strategies into treatment services, therapists can better engage and support their patients, leading to potentially better mental health outcomes and overall well-being.
Open peer review
To view the open peer review materials for this article, please visit http://doi.org/10.1017/gmh.2024.94.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/gmh.2024.94.
Acknowledgements
The authors thank Simone Renault (SR) and Ratika Bhandari (RB) for their help in conducting abstract screening, full-text screening and contributions to manuscript development.
Author contribution
Conceptualisation: BB, YG, DRS, RV, VP, MS and AN. Data curation: BB. Formal analysis: BB. Funding acquisition: DRS, RV, VP and AN. Investigation: BB, LF, AG and AN. Methodology: MP and AN. Project administration: AN. Supervision: YG, AG, UB and AN. Visualisation: BB. Writing – original draft: BB BZ. and Writing – review and editing: BB, YG, DRS, RV, BZ, LF, VP, MP, MS, AG, UB and AN.
Financial support
This study was conducted as a part of the formative research for the IMPRESS trial, which aims to assess the effectiveness and cost-effectiveness of a community intervention in enhancing access to care and improving clinical outcomes for depression. The trial has been funded through a grant from the National Institute of Mental Health (NIMH), USA (Grant number R01MH115504).
Competing interest
The authors declare that they have no conflict of interest.
Open access
