Effectiveness trials on mental health and psychosocial support interventions: the current paradigm

For the management of mental health conditions, a number of MHPSS interventions have been developed and formally tested for effectiveness over the last decades. A Cochrane systematic review of psychological therapies in humanitarian settings in LMICs identified 33 RCTs with 3523 participants. For adults, meta-analyses showed that psychological therapies substantially reduced symptoms of PTSD, depression and anxiety at study endpoint (0–4 weeks after the intervention ended), consistently with a review focused on MHPSS interventions in emergency settings in LMICs (Bangpan et al., Reference Bangpan, Felix and Dickson2019), but the intervention effect and the number of participants assessed gradually decreased over time. In children and adolescents between 5 and 18 years of age, four studies of very low quality found no significant beneficial effects of interventions over waiting list in reducing PTSD symptoms. No data were available in these studies for depression and anxiety (Purgato et al., Reference Purgato, Gastaldon, Papola, van Ommeren, Barbui and Tol2018a, Reference Purgato, Gross, Betancourt, Bolton, Bonetto, Gastaldon, Gordon, O'Callaghan, Papola, Peltonen, Punamaki, Richards, Staples, Unterhitzenberger, van Ommeren, de Jong, Jordans, Tol and Barbui2018b). Similarly, a Cochrane systematic review of psychological and social prevention interventions in humanitarian settings in LMICs highlighted the paucity of randomised evidence properly evaluating prevention interventions (i.e., incidence) and confirmed no beneficial effects of interventions over waiting list in reducing symptoms of PTSD, depression and anxiety in children. For adults, a positive effect of interventions in reducing depressive and anxiety symptoms at immediate post-intervention was detected, with no data on PTSD (Papola et al., Reference Papola, Purgato, Gastaldon, Bovo, van Ommeren, Barbui and Tol2020). Recently, two large RCTs evaluated a new psychological intervention called Self-Help Plus (SH+) and developed by the WHO, with the aim of preventing the onset of mental disorders in asylum seekers and refugees. These trials identified a beneficial effect of SH+ in preventing the frequency of mental disorders at post-intervention (Purgato et al., Reference Purgato, Carswell, Tedeschi, Acarturk, Anttila, Au, Bajbouj, Baumgartner, Biondi, Churchill, Cuijpers, Koesters, Gastaldon, Ilkkursun, Lantta, Nosè, Ostuzzi, Papola, Popa, Roselli, Sijbrandij, Tarsitani, Turrini, Valimaki, Walker, Wancata, Zanini, White, van Ommeren and Barbui2021) and at 6-month follow-up (Acarturk et al., Reference Acarturk, Uygun, Ilkkursun, Carswell, Tedeschi, Batu, Eskici, Kurt, Anttila, Au, Baumgartner, Churchill, Cuijpers, Becker, Koesters, Lantta, Nosè, Ostuzzi, Popa, Purgato, Sijbrandij, Turrini, Valimaki, Walker, Wancata, Zanini, White, van Ommeren and Barbuiin press).

In an attempt to move a step beyond effectiveness evaluation, a systematic review on focused psychosocial interventions used individual participant data (IPD) meta-analytic techniques to delineate socio-demographic profiles of children who might respond better to the allocated interventions. The IPD meta-analysis, conducted on 11 RCTs with 3143 children in humanitarian settings in LMICs (mainly in war contexts), found reduced PTSD and distress symptoms in specific subgroups of children only, e.g., older children of 15–18 years, or those not displaced, or those living in smaller households with less than six people (Purgato et al., Reference Purgato, Gastaldon, Papola, van Ommeren, Barbui and Tol2018b, Reference Purgato, Tedeschi, Betancourt, Bolton, Bonetto, Gastaldon, Gordon, O'Callaghan, Papola, Peltonen, Punamaki, Richards, Staples, Unterhitzenberger, de Jong, Jordans, Gross, Tol and Barbui2020a, Reference Purgato, Tedeschi, Bonetto, de Jong, Jordans, Tol and Barbui2020b). The IPD meta-analysis identified socio-demographic variables that might impact interventions' effect, however it was restricted to focused psychosocial support interventions without collecting information for psychotherapy or social interventions and included mixed samples of children (i.e., children with increased level of distress together with those with a potential psychiatric diagnosis). Moreover, the moderators identified by the IPD meta-analysis were not subsequently tested in formal RCTs.

According to this premise, we are presently unable to predict which subgroup of individuals might benefit most from which type of MHPSS intervention, forcing a ‘trial and error’ approach that: (1) might generate harm; (2) is almost certainly implicated in high rates of intervention dropouts; (3) causes delays in recovery and/or benefits in the short-term only; (4) and generates ethical concerns. These issues are problematic in low-resource contexts, where providing the right intervention to the right individual at right time may be a matter of survival, as many people are exposed to multiple traumatic events, and there is intense scarcity of human resources in mental health.

Although it should be recognised that the current paradigm focused on effectiveness trials providing mean values at group level has been crucially important over the last decade for the development and implementation of large RCTs on MHPSS interventions' effect, progress should now be made towards the identification and testing of predictors of interventions' response, to generate evidence for matching interventions with subgroups of individuals that might respond optimally.

This could, in turn, help minimise the inequity gap between HICs and LMICs in terms of provision of safe and effective mental health interventions.

Precision medicine

Taking into account individual clinical data, biological susceptibility and social and environmental factors, precision medicine has a potential to tackle the challenges reported above, by generating a match between subgroups of individuals and the most appropriate available intervention (Cuijpers et al., Reference Cuijpers, ReynoldS, Donker, Li, Andersson and Beekman2012, Reference Cuijpers, Ebert, Acarturk, Andersson and Cristea2016, Reference Cuijpers, Reijnders and Huibers2019; Cuijpers and Christensen, Reference Cuijpers and Christensen2017; Cuijpers and Beekman, Reference Cuijpers and Beekman2018). Precision medicine aims at predicting which treatment option may work better for a particular disease in a specific group of people. Although used often synonymously with ‘personalised medicine’, the label ‘precision medicine’ has been recently adopted as some may misinterpret the word ‘personalised’ to imply treatments developed uniquely for each individual, while the term ‘precision’ implies the identification of which approach is best for a specific subgroup of individuals sharing specific characteristics. The concept of precision medicine dates back from Hippocrates in 400 BC to Claude Bernard, a physiologist in the 19th century. Archibald Garrod, who talked about the importance of individual ‘chemical differences’ in an article of 1902, is considered the father of precision medicine (Perlman and Govindaraju, Reference Perlman and Govindaraju2016). He postulated that ‘no two individuals of a species are absolutely identical in bodily structure neither are their chemical processes carried out on exactly the same lines’ (Garrod, Reference Garrod1902). Being a physician, he emphasised the importance of clinical medicine and that is ‘in the ward rather than in the laboratory that the importance of inborn factors is to be appreciated […] and tricks of gesture and action’ (Garrod, Reference Garrod1902, Reference Garrod1909, Reference Garrod1931).

Up to now, precision medicine has been applied to different medical fields in Western HICs, offering new avenues for the amelioration of diseases. In 2015, a Precision Medicine Initiative was promoted by Obama in the United States to accelerate progress and research towards curing diseases and providing access to personalised information (Collins and Varmus, Reference Collins and Varmus2015). The goal of this initiative was to improve health by tailoring the prevention and treatment interventions of health conditions to genetic, environmental and lifestyle differences among individuals. Because of the progress in sequencing the genome of cancer cells and in identifying mutations that are ‘drivers’ of malignancy, oncology was mentioned amongst the main areas of interest, as in the last 20 years different trial designs adopting a precision medicine approach were developed and tested in oncology (Manrai et al., Reference Manrai, Patel, Gehlenborg, Tatonetti, Ioannidis and Kohane2016, Reference Manrai, Patel and Ioannidis2018).

Of note, in psychiatry a precision paradigm has already been conceptualised, with emphasis on the need of accounting for the difficulties of translating genetic information into diagnosis and clinical interventions, as diagnoses are mainly based on clinical evaluations, depending on the clinicians' judgement and experience. Similarly, the subsequent prescription phase typically follows a ‘trial and error’ approach (Vieta, Reference Vieta2015). Nevertheless, research efforts have been directed to designing a new roadmap for precision psychiatry, aiming to overcome these barriers by stratifying disease processes, identifying predictors of treatment outcome and using neuroimaging and other biomarkers to evaluate the role and mechanisms of action of mental health interventions (Krystal et al., Reference Krystal, Pizzagalli, Smoski, Mathew, Nurnberger, Lisanby, Iosifescu, Murrough, Yang, Weiner, Calabrese, Sanacora, Hermes, Keefe, Song, Goodman, Szabo, Whitton, Gao and Potter2020; Williams et al., Reference Williams, Rush, Koslow, Wisniewski, Cooper, Nemeroff, Schatzberg and Gordon2011). The International Study to Predict Optimized Treatment for Depression (iSPOT-D) is an example of a large RCT (n = 2016) aimed at identifying a number of predictors and moderators of treatment outcome in adult participants with major depressive disorder and to develop and test through a replication study a model to incorporate the effects of multiple baseline predictors or moderators on antidepressant response (escitalopram, sertraline or venlafaxine-extended release) (Williams et al., Reference Williams, Rush, Koslow, Wisniewski, Cooper, Nemeroff, Schatzberg and Gordon2011). iSPOT-D represents an important step towards precision psychiatry, as it is hypothesised that baseline clinical data (e.g., symptom severity, psychological characteristics, cognitive functioning and physical parameters) may act as potential markers for treatment response.

Precision psychology in global mental health: identifying and testing predictors of intervention response through a coordinated approach

Accurate prediction of which persons will benefit the most from which intervention would be possible even in the field of clinical psychology, where a theoretical framework is being developed for mental health conditions as depression (Huibers et al., Reference Huibers, Lorenzo-Luaces, Cuijpers and Kazantzis2021).

In a systematic review and meta-analysis aimed at assessing personalised treatments for adult depression, Cuijpers et al. collected 41 RCTs focused on specific population groups with 2741 participants, which were allocated to six different intervention strategies (head-to-head comparisons) based on their specific characteristics. Target groups were defined according to the availability of clinical trials focused on them and considering socio-demographic characteristics (e.g., older adults or minority groups), clinical characteristics (e.g., type of depression) and comorbid clinical conditions (e.g., physical illness). In total, authors identified 27 clinical and socio-demographic characteristics of participants as potential moderators of intervention's effect. However, only for a few set of characteristics the statistical power was sufficient to show a clinically relevant effect size. Moreover, there was high risk of bias in the included studies, and only a selected number of potentially relevant moderators were examined (Cuijpers et al., Reference Cuijpers, Ebert, Acarturk, Andersson and Cristea2016). For example, CBT was found to be more effective than other therapies in older adults, in patients with comorbid addictive disorders and in university students, while for the other therapies there was insufficient statistical power to show that one therapy was more effective than another therapy.

Even though the results highlighted the need of using complementary strategies for developing a proper precision approach in psychology, the review approximated its application, offering an example that incorporated individual level characteristics in the choice of the intervention strategy.

In the area of global mental health, a precision psychology approach has never been conceptualised, developed and validated. Currently, the majority of clinical trials in LMICs are designed to assess the effectiveness of a target MHPSS intervention against a control condition – often no treatment or waiting list − with a focus on short-term outcomes (mainly immediate post-intervention) (Purgato et al., Reference Purgato, Jayaram, Surkan, Bass and Bolton2019). Little evidence exists on the mechanisms of action of interventions, for example considering specific interventions' ingredients or the mediation role of target variables to predict how a MHPSS intervention works. Secondly, trials are focused on mean-group differences, informing that a given characteristic was ‘on average’ different between groups, but without assessing variability across individual participants, making it impossible to predict who might benefit most from an intervention. In this way, even when an intervention is proven to be effective in reducing psychological symptoms, the evaluation refers to mean values of clinically heterogeneous groups of individuals. Thirdly, evidence does not always incorporate environmental and social variables to predict under what set of circumstances a specific intervention is optimal in terms of efficacy and safety. The social determinants of mental disorders are the social and economic conditions that have a direct influence on the prevalence and severity of mental disorders in individuals across the life course (Patel et al., Reference Patel, Saxena, Lund, Thornicroft, Baingana, Bolton, Chisholm, Collins, Cooper, Eaton, Herrman, Herzallah, Huang, Jordans, Kleinman, Medina-Mora, Morgan, Niaz, Omigbodun, Prince, Rahman, Saraceno, Sarkar, De Silva, Singh, Stein, Sunkel and UnUtzer2018). There is growing global evidence that mental disorders in populations are strongly socially determined, especially in sensitive periods as childhood and adolescence, with the mediation of biological factors (Lund et al., Reference Lund, Brooke-Sumner, Baingana, Baron, Breuer, Chandra, Haushofer, Herrman, Jordans, Kieling, Medina-Mora, Morgan, Omigbodun, Tol, Patel and Saxena2018). In terms of theoretical frameworks on the relation between social determinants and mental health, Lund et al. recently developed a new conceptual framework that summarised the major social determinants of mental health and linked them with the Sustainable Development Goals. This framework – adopted by the Lancet Commission of Global Mental Health and Sustainable Development (Patel et al., Reference Patel, Saxena, Lund, Thornicroft, Baingana, Bolton, Chisholm, Collins, Cooper, Eaton, Herrman, Herzallah, Huang, Jordans, Kleinman, Medina-Mora, Morgan, Niaz, Omigbodun, Prince, Rahman, Saraceno, Sarkar, De Silva, Singh, Stein, Sunkel and UnUtzer2018) – identified the following domains of the social determinants of mental disorders: demographic, economic, neighbourhood, environmental events and social and culture domains. In each domain, distal and proximal determinants were identified to impact on mental disorders, mediated by family-level and biological variables. This framework highlighted the importance of an ecological approach and the complex multidimensional way in which social determinants interact with key genetic determinants to affect mental disorders, and may inform the development and implementation of psychological interventions (Lund et al., Reference Lund, Brooke-Sumner, Baingana, Baron, Breuer, Chandra, Haushofer, Herrman, Jordans, Kieling, Medina-Mora, Morgan, Omigbodun, Tol, Patel and Saxena2018).

For example, adverse social and economic circumstances that are frequently experienced in humanitarian settings in LMICs, like poverty, income inequality, trauma exposure, interpersonal and collective violence and forced migration, are key determinants of mental disorders, and may impact individual symptoms and functioning over time (Miller et al., Reference Miller, Jordans, Tol and Galappatti2021).

A recent study aimed at delineating trajectories of psychological symptoms and resilience in 597 children exposed to multiple traumatic events in Burundi, Indonesia, Nepal and Sri Lanka, identified a positive association between the number of traumatic event types and psychological symptoms of PTSD, depression and anxiety (Purgato et al., Reference Purgato, Tedeschi, Betancourt, Bolton, Bonetto, Gastaldon, Gordon, O'Callaghan, Papola, Peltonen, Punamaki, Richards, Staples, Unterhitzenberger, de Jong, Jordans, Gross, Tol and Barbui2020a, Reference Purgato, Tedeschi, Bonetto, de Jong, Jordans, Tol and Barbui2020b). In particular, time and trauma were the factors exerting most influence on trends over time, and the healing effect of time was strongly contextually determined. In a precision psychology paradigm for global mental health, the integration of the social determinants of mental health in RCT designs should be prioritised and considered at an individual level, including contextual factors, e.g., the nature of trauma, the time since trauma exposure and the amount and types of traumatic events. In turn, these factors may be integrated with functional and structural neuroimaging information, that is not always easily accessible in LMICs, and might be less critical in guiding the intervention choice according to recent claims of shifting the attention from the consideration of mental health problems as mostly brain disorders (Ioannidis, Reference Ioannidis2019).

Fourth, RCTs should include head-to-head comparisons between different MHPSS interventions (Cuijpers, Reference Cuijpers2016), for example including low resource psychosocial interventions versus more complex and resource intensive psychotherapeutic interventions in homogeneous groups (Turrini et al., Reference Turrini, Tedeschi, Cuijpers, Del Giovane, Kip, Morina, Nosè, Ostuzzi, Purgato, Ricciardi, Sijbrandij, Tol and Barbui2021). Additionally, longitudinal naturalistic designs might track the psychological outcomes and resource pathways, accounting for the contextual factors from proximal to distal determinants, regardless of the administration of the intervention.

We are aware of the theoretical and practical complexities implied in the development of a precision psychology paradigm in global mental health, but new research methodologies and technologies are available. IPD and network meta-analytic techniques might be applied to all MHPSS interventions for developing large clinical datasets with up-to-date individual-level sociodemographic, clinical, biological and context-related information. The efficacy and safety of different MHPSS interventions, compared each other (head-to-head) and against inactive controls (e.g., waiting list, no treatment, care as usual) might this way be studied in relation to available individual-level information, following a precision psychology approach. Machine learning algorithms may represent an additional tool for generating visual representations of selected MHPSS intervention options and to create evidence-based matches between subgroups of individuals with a psychological condition (e.g., PTSD, depression, anxiety) and the most appropriate MHPSS intervention.

However, IPD meta-analyses typically collect information on very few individual-level variables, as the primary studies were designed to respond to specific research questions. Therefore, in order to move beyond a ‘one-size-fits-all’ rationale, we need a new generation of RCTs that use a common set of individual-level measures to be used as moderators of response. This implies identifying the moderators that might potentially impact the intervention's effect first and verifying whether the inclusion of these in RCTs indeed results in better outcomes for a selected group of individuals. Qualitative research involving end-users, key informants and stakeholders might also be useful as a preparatory phase of RCTs, to properly understand the peculiarity of specific implementation contexts (e.g., humanitarian settings), barriers, facilitators and challenges.

This process will hopefully result in a comprehensive global evidence base for MHPSS interventions open to all researchers in global mental health, with trials being easily compared and combined and no longer considered as separate entities of data collection.