Introduction
Currently, social and economic needs are typically underassessed and poorly addressed by mental health services (Boardman, Killaspy, & Mezey, Reference Boardman, Killaspy and Mezey2022; Lambri, Chakraborty, Leavey, & King, Reference Lambri, Chakraborty, Leavey and King2012), despite pronounced social and economic need in people with mental ill-health (Jones et al., Reference Jones, Gicas, Seyedin, Willi, Leonova, Vila-Rodriguez, Procyshyn, Smith, Schmitt, Vertinsky, Buchanan, Rauscher, Lang, MacEwan, Lima, Montaner, Panenka, Barr, Thornton and Honer2020; Nuyen et al., Reference Nuyen, Tuithof, De Graaf, Van Dorsselaer, Kleinjan and Have2020; Pevalin, Reeves, Baker, & Bentley, Reference Pevalin, Reeves, Baker and Bentley2017; Phillips et al., Reference Phillips, Finkel, Petkus, Muñoz, Pahlen, Johnson, Reynolds and Pedersen2023; Sareen, Afifi, McMillan, & Asmundson, Reference Sareen, Afifi, McMillan and Asmundson2011; Stain et al., Reference Stain, Galletly, Clark, Wilson, Killen, Anthes, Campbell, Hanlon and Harvey2012; Topor et al., Reference Topor, Stefansson, Denhov, Bülow and Andersson2019). A range of effective interventions have been developed to address these needs (Barnett et al., Reference Barnett, Steare, Dedat, Pilling, McCrone, Knapp, Cooke, Lamirel, Dawson, Goldblatt, Hatch, Henderson, Jenkins, T, Machin, Simpson, Shah, Stevens, Webber and Lloyd-Evans2022; Killaspy et al., Reference Killaspy, Harvey, Brasier, Brophy, Ennals, Fletcher and Hamilton2022).
The social and economic adversities experienced by people with mental ill-health are further pronounced among those from marginalised groups (Giebel et al., Reference Giebel, Corcoran, Goodall, Campbell, Gabbay, Daras, Barr, Wilson and Kullu2020) who may experience multiple, intersecting disadvantages resulting from their identity. This may include minority ethnic groups (Morgan et al., Reference Morgan, Kirkbride, Hutchinson, Craig, Morgan, Dazzan, Boydell, Doody, Jones, Murray, Leff and Fearon2008, Reference Morgan, Fearon, Lappin, Heslin, Donoghue, Lomas, Reininghaus, Onyejiaka, Croudace, Jones, Murray, Doody and Dazzan2017), people living in unstable housing or facing homelessness (Queen, Lowrie, Richardson, & Williamson, Reference Queen, Lowrie, Richardson and Williamson2017; Quirouette, Reference Quirouette2016), and people experiencing economic hardship (Boardman et al., Reference Boardman, Killaspy and Mezey2022). Marginalised groups also experience reduced access to (Schlief et al., Reference Schlief, Rich, Rains, Baldwin, Rojas-Garcia, Nyikavaranda, Persaud, Dare, French, Lloyd-Evans, Crawford, Smith, Kirkbride and Johnson2023), and poorer outcomes from (Barnett et al., Reference Barnett, Oshinowo, Cooper, Taylor, Smith and Pilling2023), existing mental health interventions as a result of these unmet needs. As such, targeted intervention that addresses the specific social and economic needs of marginalised communities may work toward addressing these inequalities and achieving equity of care.
Indeed, such approaches have offered promising impacts for some minoritised groups with mental ill-health in the receipt of targeted psychological intervention (Arundell et al., Reference Arundell, Barnett, Buckman, Saunders and Pilling2021; Ellis, Draheim, & Anderson, Reference Ellis, Draheim and Anderson2022). However, there is currently no systematic evidence synthesis reviewing targeted interventions addressing social and economic needs of marginalised groups living with mental ill-health. As such, it is not clear which interventions currently exist and for which communities. This topic is even more pressing given the disproportionately harmful impacts of the recent COVID-19 pandemic and economic crises on marginalised groups (Camara et al., Reference Camara, Surkan, Van Der Waerden, Tortelli, Downes, Vuillermoz and Melchior2023; Das-Munshi et al., Reference Das-Munshi, Bakolis, Bécares, Dyer, Hotopf, Ocloo, Stewart, Stuart and Dregan2023; England et al., Reference England, Jarrom, Washington, Hasler, Batten, Edwards and Lewis2024; Siimsen et al., Reference Siimsen, Orru, Naevestad, Nero, Olson, Kaal and Meyer2023; Thomeer, Moody, & Yahirun, Reference Thomeer, Moody and Yahirun2023).
Therefore, we aimed to: (i) review existing evidence to identify interventions addressing social and/or economic needs that have either been adapted or developed bespoke for people from marginalised or minoritised sociodemographic or socioeconomic groups with mental ill-health and (ii) narratively examine the types of interventions studied and their respective outcomes.
Methods
We conducted a two-stage systematic review in line with a predefined protocol. This review was conducted as part of a broader research program which sought to identify interventions designed to address social and/or economic needs in people living with mental ill-health (Greenburgh et al., Reference Greenburgh, Baldwin, Weir, Asif, Laporte, Bertram and Morgan2025). Here, we review studies that reported targeted interventions to directly support the social and/or economic needs of marginalised groups experiencing mental ill-health. See Supplementary Materials I for the full inclusion criteria.
We first utilised bibliography searches of two recent reviews on this topic (Barnett et al., Reference Barnett, Steare, Dedat, Pilling, McCrone, Knapp, Cooke, Lamirel, Dawson, Goldblatt, Hatch, Henderson, Jenkins, T, Machin, Simpson, Shah, Stevens, Webber and Lloyd-Evans2022; Killaspy et al., Reference Killaspy, Harvey, Brasier, Brophy, Ennals, Fletcher and Hamilton2022) to avoid duplication of efforts. Together, these two reviews represent rigorous, broad, and relatively recent narratives on the subject area of social interventions for people living with mental ill-health. However, this current review represents a related but distinct topic of targeted intervention. Furthermore, the global context has shifted since the searches for these reviews were conducted, given the COVID-19 pandemic and worsening economic crises. As such, we then replicated the original search strategies from both reviews to identify recent literature (January 2020–February 2024). Searches were conducted in MEDLINE (Supplementary Materials II), PsycINFO, Web of Science (SciELO database), and the Cochrane Central Register of Controlled Trials (Supplementary Materials III). All records were double-blind-screened by two reviewers. Data extraction was conducted within a fit-for-purpose extraction form (Supplementary Materials I) by one researcher and checked by a second independent researcher. Quality appraisal was conducted using the Kmet quality assessment checklist (Kmet, Cook, & Lee, Reference Kmet, Cook and Lee2004) by one researcher, with a random sample (10% derived from a random sequence generator) conducted by two reviewers. Conflicts in decisions were discussed with the wider review team until a consensus was reached.
Data synthesis was conducted via a narrative synthesis of the identified interventions, whereby we provided a summary of the content and results for each of the included studies. We did not plan to conduct meta-analyses due to the expected heterogeneity of evidence.
Results
Seventy-eight studies were included that reported on interventions adapted or developed bespoke for a specific sociodemographic or socioeconomic group (Figure 1). These groups included: people experiencing or at risk of homelessness, people with an offending history, mothers, caregivers, minoritised ethnic groups, older adults, people experiencing economic disadvantage, women with experience of intimate partner violence, and people with intellectual disabilities. The studies were conducted across 16 countries: USA (n = 36), Canada (n = 18), UK (n = 5), France (n = 4), the Netherlands (n = 2), Spain (n = 2), Australia (n = 2), Switzerland (n = 2), Portugal (n = 1), Norway (n = 1), Vietnam (n = 1), Pakistan (n = 1), Germany (n = 1), Finland (n = 1), India (n = 1), and Bangladesh (n = 1). Kmet quality scores ranged from 81–100 (quantitative) and 40–100 (qualitative). Summaries of the evidence from randomised (Table 1) and nonrandomised studies (Table 2) are described later. Key intervention terms are summarised in a glossary (Supplementary Materials IV).
Figure 1. A PRISMA diagram demonstrating the flow of studies in the review.
*Please see Greenburgh et al. (Reference Greenburgh, Baldwin, Weir, Asif, Laporte, Bertram and Morgan2025) for details regarding the broader systematic review of social and/or economic interventions for people living with mental ill-health.
Table 1. A summary of the characteristics of the included randomised controlled trials
Abbreviations: NR = Not reported; B = Sourced from Barnett et al., Reference Barnett, Steare, Dedat, Pilling, McCrone, Knapp, Cooke, Lamirel, Dawson, Goldblatt, Hatch, Henderson, Jenkins, T, Machin, Simpson, Shah, Stevens, Webber and Lloyd-Evans2022; K = Sourced from Killaspy et al., Reference Killaspy, Harvey, Brasier, Brophy, Ennals, Fletcher and Hamilton2022; U = Sourced from updated searches. CMD = common mental disorders; SMI = severe mental illness; ACT = assertive community treatment; IACT = integrated assertive community treatment; FACT = forensic assertive community treatment; ACTO = assertive community treatment only; ICM = intensive case management; CBT = cognitive behavioural therapy; SSDI = social security disability income.
a Quality scores were conducted using the Kmet tool for both the updated searches and studies included in Killaspy et al. (Reference Killaspy, Harvey, Brasier, Brophy, Ennals, Fletcher and Hamilton2022). Studies included in Barnett et al. (Reference Barnett, Steare, Dedat, Pilling, McCrone, Knapp, Cooke, Lamirel, Dawson, Goldblatt, Hatch, Henderson, Jenkins, T, Machin, Simpson, Shah, Stevens, Webber and Lloyd-Evans2022) were appraised using the Cochrane Risk of Bias tool.
Table 2. A summary of the characteristics of the included nonrandomised studies
Abbreviations: NR = Not reported; B = Sourced from Barnett et al., Reference Barnett, Steare, Dedat, Pilling, McCrone, Knapp, Cooke, Lamirel, Dawson, Goldblatt, Hatch, Henderson, Jenkins, T, Machin, Simpson, Shah, Stevens, Webber and Lloyd-Evans2022; K = Sourced from Killaspy et al., Reference Killaspy, Harvey, Brasier, Brophy, Ennals, Fletcher and Hamilton2022; U = Sourced from updated searches. CMD = common mental disorders; SMI = severe mental illness.
People experiencing or at risk of homelessness
Targeted interventions for people experiencing homelessness or unstable housing were highly researched (n = 50 studies). Most interventions in this domain focused on housing for homeless/precariously housed populations (n = 35); the remaining literature addressed housing for people at risk of homelessness, living in sheltered/supported housing, residential care, or transitioning to community housing from sheltered accommodation.
Evidence from randomised studies
Fourteen randomised controlled trials (RCTs) evaluated housing first (HF) interventions (Aubry et al., Reference Aubry, Goering, Veldhuizen, Adair, Bourque, Distasio, Latimer, Stergiopoulos, Somers, Streiner and Tsemberis2016, Reference Aubry, Bourque, Goering, Crouse, Veldhuizen, LeBlanc, Cherner, Bourque, Pakzad and Bradshaw2019; Kerman et al., Reference Kerman, Aubry, Adair, Distasio, Latimer, Somers and Stergiopoulos2020; Kirst et al., Reference Kirst, Friesdorf, Ta, Amiri, Hwang, Stergiopoulos and O’Campo2020; Lachaud et al., Reference Lachaud, Mejia-Lancheros, Nisenbaum, Stergiopoulos, O’Campo and Hwang2021; Latimer et al., Reference Latimer, Rabouin, Cao, Ly, Powell, Aubry, Distasio, Hwang, Somers, Bayoumi, Mitton, Moodie and Goering2020; Lemoine et al., Reference Lemoine, Loubière, Boucekine, Girard, Tinland and Auquier2021; Loubière et al., Reference Loubière, Lemoine, Boucekine, Boyer, Girard, Tinland and Auquier2022; Mejia-Lancheros et al., Reference Mejia-Lancheros, Lachaud, Stergiopoulos, Matheson, Nisenbaum, O’Campo and Hwang2020; O’Campo et al., Reference O’Campo, Nisenbaum, Crocker, Nicholls, Eiboff and Adair2023; Somers et al., Reference Somers, Moniruzzaman, Patterson, Currie, Rezansoff and Palepu2017; Stergiopoulos et al., Reference Stergiopoulos, Hwang, Gozdzik, Nisenbaum, Latimer and Rabouin2015; Stergiopoulos et al., Reference Stergiopoulos, Gozdzik, Misir, Skosireva, Sarang, Connelly, Whisler and McKenzie2016; Tinland et al., Reference Tinland, Loubiere, Boucekine, Boyer, Fond and Girard2020) or supplemented housing first (Caplan et al., Reference Caplan, Nelson, Distasio, Isaak, Edel, Macnaughton, Piat, Patterson, Kirst, Aubry, Stergiopoulos and Goering2023; Tsemberis, Gulcur, & Nakae, Reference Tsemberis, Gulcur and Nakae2004). This approach draws on harm reduction principles, providing immediate access to housing through rent supplements and recovery-oriented support, without requirements such as sobriety. The literature mostly reported improved housing outcomes for those who received HF, namely stable housing and better-quality housing for homeless participants (Table 1).
Other included RCTs evaluated similar approaches to support people experiencing chronic homelessness into more stable housing, such as supported housing (Adamus, Mötteli, Jäger, & Richter, Reference Adamus, Mötteli, Jäger and Richter2022; Mötteli et al., Reference Mötteli, Adamus, Deb, Fröbel, Siemerkus, Richter and Jäger2022; Raven, Niedzwiecki, & Kushel, Reference Raven, Niedzwiecki and Kushel2020), residential treatment (Lipton, Nutt, & Sabatini, Reference Lipton, Nutt and Sabatini1988), integrated housing (McHugo et al., Reference McHugo, Bebout, Harris, Cleghorn, Herring and Xie2004), housing placements (Burnam et al., Reference Burnam, Morton, McGlynn, Petersen, Stecher, Hayes and Vaccaro1996; Goldfinger et al., Reference Goldfinger, Schutt, Tolomiczenko, Seidman, Penk, Turner and Caplan1999), and interventions involving rent subsidy (Hurlburt, Hough, & Wood, Reference Hurlburt, Hough and Wood1996; O’Connell, Tsai, & Rosenheck, Reference O’Connell, Tsai and Rosenheck2023). Types of assertive community treatment (ACT) alongside standard or integrated case management were also common in this population (Fletcher et al., Reference Fletcher, Cunningham, Calsyn, Morse and Klinkenberg2008; Korr & Joseph, Reference Korr and Joseph1995; Lehman, Reference Lehman1997; Morse et al., Reference Morse, Calsyn, Allen, Tempethoff and Smith1992; Morse et al., Reference Morse, Calsyn, Klinkenberg, Trusty, Gerber and Smith1997, Reference Morse, Calsyn, Klinkenberg, Helminiak, Wolff and Drake2006; Shern et al., Reference Shern, Tsemberis, Anthony, Lovell, Richmond and Felton2000). The remaining studies evaluated other structured programs, such as the critical time intervention involving case management (Herman et al., Reference Herman, Conover, Gorroochurn, Hinterland, Hoepner and Susser2011; Susser et al., Reference Susser, Valencia, Conover, Felix, Tsai and Wyatt1997), and the Maintaining Independence and Sobriety through Systems Integration, Outreach and Networking-Veterans Edition (MISSION-VET) intervention (Ellison et al., Reference Ellison, Schutt, Yuan, Mitchell-Miland, Glickman, McCarthy, Smelson, Schultz and Chinman2020). Broadly, all of these housing interventions were associated with improved housing stability or fewer nights spent homeless. The final intervention described a supplemented long-term psychotherapy (Laurila, Lindfors, Knekt, & Heinonen, Reference Laurila, Lindfors, Knekt and Heinonen2024) for people experiencing homelessness and reported improved social support outcomes.
Evidence from nonrandomised studies
The nonrandomised studies mostly evaluated HF interventions (Brown et al., Reference Brown, Jason, Malone, Srebnik and Sylla2016; Holmes et al., Reference Holmes, Carlisle, Vale, Hatvani, Heagney and Jones2017; Macnaughton et al., Reference Macnaughton, Nelson, Worton, Tsemberis, Stergiopoulos, Aubry, Hasford, Distasio and Goering2018; Rhenter, Moreau, & L, Reference Rhenter, Moreau and L2018; Stergiopoulos et al., Reference Stergiopoulos, Zerger, Jeyaratnam, Connelly, Kruk, O’Campo and Hwang2016; Worton et al., Reference Worton, Hasford, Macnaughton, Nelson, MacLeod, Tsemberis, Stergiopoulos, Goering, Aubry, Distasio and Richter2018), which similarly broadly reported favorable housing outcomes, experiences, and high fidelity of HF, alongside other types of supported housing (Dehn et al., Reference Dehn, Beblo, Richter, Wienberg, Kremer, Steinhart and Driessen2022; Gutman & Raphael-Greenfield, Reference Gutman and Raphael-Greenfield2017; Killaspy et al., Reference Killaspy, Priebe, Bremner, McCrone, Dowling, Harrison, Krotofil, McPherson, Sandhu, Arbuthnott, Curtis, Leavey, Shepherd, Eldridge and King2016; Killaspy et al., Reference Killaspy, Priebe, McPherson, Zenasni, Greenberg, McCrone, Dowling, Harrison, Krotofil, Dalton-Locke, McGranahan, Arbuthnott, Curtis, Leavey, Shepherd, Eldridge and King2020; Stanhope et al., Reference Stanhope, Choy-Brown, Tiderington, Henwood and Padgett2016), sheltered housing (Padmakar et al., Reference Padmakar, Wit, Mary, Regeer, Bunders-Aelen and Regeer2020; Roos et al., Reference Roos, Bjerkeset and Søndenaa2016), and specialist ACT (Doré-Gauthier et al., Reference Doré-Gauthier, Miron, Jutras-Aswad, Ouellet-Plamondon and Abdel-Baki2020), which broadly reported improved housing and social inclusion outcomes and experiences (Table 2).
People with an offending history
Nine papers reported targeted interventions for people with a current or past offending history, all of which were RCTs.
Evidence from randomised studies
An ACT model of case management with nonadversarial court proceedings in the USA was compared with treatment as usual (TAU), assessing outcomes over a 2-year period (Cosden, Ellens, Schnell, & Yamini‐Diouf, Reference Cosden, Ellens, Schnell and Yamini‐Diouf2005). Across both conditions, offenders with a high conviction rate experienced increased arrests (F 1,20 = 33.46, p < .001), convictions (F 1,20 = 17.74, p < 0.001), and jail days (F 1,20 = 43.51, p < .001) postintervention. However, for the remaining participants, an increase in arrests postintervention was observed in the ACT group (F 1,185 = 5.05, p < .05), whereas the number of convictions (treatment, pre = 1.84, post = 1.82; TAU, pre = 2.32, post = 2.04) and number of days in jail reduced across both groups (treatment: pre = 39.44, post = 24.55; TAU: pre = 47.30, post = 37.51).
An integrated dual disorders treatment (IDDT) program was compared with service as usual in recidivists with severe mental illness (SMI) and substance use disorders after leaving custody (Chandler & Spicer, Reference Chandler and Spicer2006). Both groups showed reduced arrests per year, where this difference was greater for those receiving IDDT (arrests per person/year: IDDT pre = 2.89, post = 2.21; control pre = 2.84, post = 2.61). Conviction rates reduced for those receiving IDDT only (convictions per person/year: IDDT pre = 0.69, post = 0.59; control pre = 0.61, post = 0.73). Felony convictions increased slightly for both groups (felony conviction per person/year: IDDT pre = 0.29, post = 0.31; control pre = 0.25, post = 0.28) and jail days decreased for both groups (jail days per person/year: IDDT pre = 96.74, post = 60.71; control pre = 79.43, post = 59.39).
Two RCTs evaluated interventions where adaptations to ACT were applied to create forensic assertive community treatment (FACT), including accepting referrals from criminal justice agencies and making re-arrest prevention an explicit goal. FACT led to fewer bookings (12-month follow-up mean: FACT = 0.64, TAU = 1.42; 13- to 24-month follow-up: FACT = 0.57, TAU = 0.89), an increased likelihood of staying out of jail (12-month follow-up mean: FACT = 0.75, TAU = 0.85; 13- to 24-month follow-up: FACT = 0.38, TAU = 0.55), and a shorter time in jail (12-month follow-up mean: FACT = 18.5, TAU = 35.3; 13- to 24-month follow-up: FACT = 20.5, TAU = 30.5) (Cusack et al., Reference Cusack, Morrissey, Cuddeback, Prins and Williams2010). In the second RCT, FACT led to fewer convictions (mean: 0.4 vs .0.9, p = .023), days in jail (mean: 21.56 vs 43.5, p = .025), arrests (mean: 0.8 vs 1.3, p = .165), and number of incarcerations relating to new offences (mean: 1.3 vs 1.5, p = .967) compared with TAU (Lamberti et al., Reference Lamberti, Weisman, Cerulli, Williams, Jacobowitz and Mueser2017).
A bespoke cognitive-behavioural program targeting antisocial attitudes and recidivism was compared with TAU (Kingston, Olver, McDonald, & Cameron, Reference Kingston, Olver, McDonald and Cameron2018). Recidivism data were available for 80 participants, out of 101, who were followed up with for an average of 1.5 years after release, whereby those in the treatment group had a slightly lower rate of violent recidivism (13.6% vs 16.7%), but a slightly higher rate of general recidivism compared with TAU (59.1% vs 52.8%).
A bespoke peer support group intervention encouraging social participation and sobriety and reducing criminality was tested in 114 adults who had criminal charges within two years of enrolment in the study (Rowe et al., Reference Rowe, Bellamy, Baranoski, Wieland, O’Connell and Benedict2007). Controlling for baseline levels of criminal justice charges, both the control (standard services) group and intervention group showed lower rates of criminal charges over time (mean total charges: control, pre = 1, time 1 = 0.76, time 2 = 0.32; intervention, pre = 1.40, time 1 = 1.18, time 2 = 0.75; F = 4.301,111, p < .05, η2 = .04).
Two RCTs examined a modified therapeutic community (MTC) program for men who were in prison with comorbid substance use problems. The intervention aimed to change attitudes and lifestyles associated with substance abuse, mental ill-health, and criminal thinking (Sacks et al., Reference Sacks, Chaple, Sacks, McKendrick and Cleland2012, Reference Sacks, Sacks, McKendrick, Banks and Stommel2004). The first study compared MTC with a mental health treatment program (MH) in prison settings, alongside a comparison of MTC with an aftercare option when inmates were released. Those in the MTC group had lower rates of reincarceration compared with those assigned to the MH program, and aftercare decreased reincarceration rates further (MH only = 33%, MTC-prison only = 17%, and MTC-prison + MTC aftercare = 5%). The intervention was associated with lower rates of criminal activity (MH only = 67%, MTC-prison only = 53%, and MTC-prison + MTC aftercare = 42%) and a longer time to subsequent incarceration (mean days: MH only = 108.43, MTC only = 124.80, MTC + aftercare = 169.50) or first offence (mean days: MH only = 66.19, MTC only = 84.06, MTC + aftercare = 67.11).
The second study (Sacks et al., Reference Sacks, Chaple, Sacks, McKendrick and Cleland2012) extended this work to test the effectiveness of MTC as a re-entry treatment in community correction facilities after prison release (RMTC) in comparison with parole supervision and case management. Here, reincarceration rates and self-reported criminal activity were substantially lower in the RMTC group at 12-month postrelease from prison (reincarceration: RMTC = 19%, Parole group = 38%, OR = 0.387, 95% CI: 0.155–0.97, p = 0.43; criminal activity: RMTC = 39%, Parole group = 62%, OR = 0.394, 95% CI: 0.166–0.937, p = .35).
Finally, a network coaching intervention to strengthen social networks of forensic psychiatric outpatients was compared with TAU (Swinkels et al., Reference Swinkels, Van Der Pol, Twisk, Ter Harmsel, Dekker and Popma2023). Participants in the intervention group reported fewer criminal behaviours compared with TAU at a 12-month follow-up (RR = 0.575, 95% CI: 0.225–1.47) and an 18-month follow-up (RR = 0.180, 95% CI: 0.053–0.611, p = .006).
Mothers
Six studies evaluated targeted interventions for mothers living with mental ill-health.
Evidence from randomised studies
The ‘HUGS’ intervention (Holt, Gentilleau, Gemmill, & Milgrom, Reference Holt, Gentilleau, Gemmill and Milgrom2021) aimed to improve mother–infant interactions. Seventy-seven new mothers with postnatal depression in Australia were randomised to receive either a CBT session followed by a group-based mother–infant interaction intervention (‘HUGS’) or a control playgroup. HUGS was associated with improvements in parental positive affective involvement and verbalisation (F 1,47 = 4.96, η p2 = 0.10, p = .03) and reductions in measures of impaired bonding (F 1,45 = 4.55, η p2 = .09, p = .04) compared with the control group at 6 months.
An online peer-delivered 1-day CBT–based group workshop was adapted to address content such as social support and sleep difficulties for mothers (Van Lieshout et al., Reference Van Lieshout, Layton, Savoy, Brown, Ferro, Streiner, Bieling, Feller and Hanna2021). Mothers with postpartum depression (n = 403) in Canada were assigned to either the workshop or a waitlist control group. Mothers reported improvements in bonding with their infant (B = −3.22; 95% CI, −4.72 to −1.71; p < .001; Cohen d = 0.34) and in ratings of social support (B = 3.31; 95% CI, 1.04–5.57; p < .001; Cohen d = 0.24).
The ‘Promoting First Relationship’ (PFR) intervention, initially developed to target child welfare, was adapted to support low-income new mothers with depression, anxiety, or PTSD accessing community or primary care in the USA (Oxford et al., Reference Oxford, Hash, Lohr, Bleil, Fleming, Unützer and Spieker2021). Two hundred fifty-two mothers received either PFR or were mailed a resource pack. The authors report small positive effects of PFR on parenting sensitivity (6 months: ds = .25, 12 months: ds = .26) and a small effect on maternal understanding of infant behaviour at 6 months (d = .21) and a small-to-medium effect at 12 months (d = .45).
The ‘Songs from Home’ intervention is a songwriting program designed to address loneliness in new mothers (Perkins, Spiro, & Waddell, Reference Perkins, Spiro and Waddell2023). Mothers with postnatal depression and experiences of loneliness in the UK (n = 89) were allocated to either ‘Songs from Home’ or a waitlist control. Both the intervention group and control group reported lower loneliness scores at week six (intervention drop: 38% relative and 25% absolute; control drop: 10% relative and 7% absolute). A large effect between social connectedness and treatment group was also identified (F 2,114 = 11.949, p < .001, ηp2 = 0.173), with greater improvements observed in the intervention group (14% relative increase and 7% absolute increase, respectively).
Evidence from nonrandomised studies
One open pilot trial study evaluated the effects of community family treatment for 32 postpartum couples in the USA (Battle et al., Reference Battle, Cardemil, Rossi, O’Hara and Miller2023). Improvements, with medium-to-large effects, were observed postintervention in family functioning. A feasibility study evaluated a culturally adapted integrated parenting intervention for 26 depressed mothers in a low-income setting in Pakistan compared with routine community care (Chaudhry et al., Reference Chaudhry, Sattar, Kiran, Wan, Husain, Hidayatullah, Ali, Shafique, Suhag, Saeed, Maqbool and Husain2023) and reported 100% retention and attendance.
People experiencing economic disadvantage
Three studies described interventions targeted toward people experiencing specific economic disadvantage. Two further studies tested interventions developed for populations with multiple marginalised characteristics, including economic disadvantage (Chaudhry et al., Reference Chaudhry, Sattar, Kiran, Wan, Husain, Hidayatullah, Ali, Shafique, Suhag, Saeed, Maqbool and Husain2023; Oxford et al., Reference Oxford, Hash, Lohr, Bleil, Fleming, Unützer and Spieker2021), which are discussed in the ‘Mothers’ section. Results of interventions relating to homelessness are also relevant.
Evidence from randomised studies
A follow-up RCT evaluated adapted-IPS using administrative records of 2,160 individuals with schizophrenia or affective disorder who also received Social Security Disability Insurance (SSDI) payments in the USA (Baller et al., Reference Baller, Blyler, Bronnikov, Xie, Bond, Filion and Hale2020). Adaptations to the IPS intervention for SSDI beneficiaries included payments of the beneficiary’s share of health insurance premiums; access to other evidence-based behavioural health services; and suspension of medical disability reviews for three years after study enrollment. Participants in the intervention group were 2.6 times more likely than those in the control group to receive any earnings, and on average earned more over the year than the control group.
The ‘ASHA’ project aimed to evaluate an integrated depression and economic strengthening intervention in rural Bangladesh (Karasz, Anne, Hamadani, & Tofail, Reference Karasz, Anne, Hamadani and Tofail2021). ASHA was developed via a woman-centered framework that emphasised a woman’s right to respect, dignity, and care. Low-income women with depression (n = 48) were randomised to a pilot RCT of either fortnightly depression management and a financial literacy intervention followed by a cash transfer, or no intervention. The authors report improvements from baseline to 12-month follow-up in social support, such as tangible support (ASHA mean difference: 3.4, control mean difference: 1.5, p = .153, 95% CI: −4.6 to 0.7), positive social interaction (ASHA mean difference: 4.1, control mean difference: 1.0, p = .015, 95% CI: −5.6 to −0.6) and emotional support (ASHA mean difference: 8.8, control mean difference: 6.6, p = .443, 95% CI: −8.0 to 3.5), as well as household economic decision-making (ASHA mean difference: 1.5, control mean difference: −0.1, p = .011, 95% CI: −2.8 to −0.4), and reductions in experiences of physical/mental coercion compared with controls (ASHA mean difference: −0.5, control mean difference: 0.1, p = .011, 95% CI: 0.2–1.2).
Evidence from nonrandomised studies
The second study tested the acceptability, feasibility, and impact of a community mental health support group for households living in poverty, including 68 individuals with SMI and caregivers (Nguyen, Tran, & G, Reference Nguyen, Tran and G2020). Group support sessions, facilitated by trained Women’s Union staff, covered topics such as personal hygiene, nutrition, physical and mental health care, rights and privileges of people with SMI, rehabilitation, community integration, and reducing caregiver stress. The intervention was reported to be acceptable and feasible, with increased annual household income and decreased annual expenditure reported.
Older adults
Three studies considered targeted interventions for older adults.
Evidence from randomised studies
Three RCTs evaluated targeted interventions for older adults. Two of these (Granholm et al., Reference Granholm, McQuaid, McClure, Auslander, Perivoliotis, Pedrelli, Patterson and Jeste2005; Rajji et al., Reference Rajji, Mamo, Holden, Granholm and Mulsant2022) described modifications made to a cognitive behavioural social skills training (CBSST) intervention for older adults with schizophrenia, such as developing aids to compensate for possible cognitive impairment and integrating age-relevant content (e.g. challenging ageist beliefs and role-playing age-relevant situations). Granholm et al. (Reference Granholm, McQuaid, McClure, Auslander, Perivoliotis, Pedrelli, Patterson and Jeste2005) reported that, of 76 middle- and older-adults recruited to either CBSST or usual care in the USA, those receiving CBSST performed social functioning activities more frequently than those allocated to usual care postintervention (F = 6.96, df = 1, 68, p = 0.02, η 2 = 0.08). Rajji et al. (Reference Rajji, Mamo, Holden, Granholm and Mulsant2022) reported that of the 63 participating older adults in Canada, CBSST was more efficacious in preventing decline in social function over one-year period than usual care, as the trajectories of the Independent Living Skills Survey demonstrated better function in this group at both 36 weeks (Cohen’s d = 0.75) and 52 weeks (Cohen’s d = 0.92).
The third RCT evaluated a physical activity intervention designed to alleviate loneliness in older adults with anxiety or depression (Ruiz-Comellas et al., Reference Ruiz-Comellas, Valmaña, Catalina, Baena, Mendioroz Peña, Roura Poch, Sabata Carrera, Cornet Pujol, Casaldàliga Solà, Fusté Gamisans, Saldaña Vila, Vázquez Abanades and Vidal-Alaball2022). Participants accessing primary care services in Spain (n = 90) were allocated to the physical activity program or usual care. The intervention group improved in social support outcomes (intervention change scores: −3.59 (11.68), 95% CI: −7.66 to 0.49; control change scores: 2.97 (9.81), 95% CI: −0.35 to 6.29, p = .078).
Caregivers
Two studies evaluated targeted intervention for caregivers.
Evidence from randomised studies
One multicenter RCT, conducted in Spain and Portugal, allocated 109 family primary caregivers of individuals living with schizophrenia or schizoaffective disorder to a psychoeducational intervention program (PIP) or usual care (Martin-Carrasco et al., Reference Martin-Carrasco, Fernandez-Catalina, Domínguez-Panchón, Gonçalves-Pereira, González-Fraile, Muñoz-Hermoso and Ballesteros2016). PIP aimed to alleviate caregiver burden and improve relationships and was associated with reduced caregiver burden at follow-up compared with usual care (4 months: mean difference = −4.33; 95% CI −7.96, −0.71; 8 months: mean difference = −4.46; 95% CI −7.79, −1.13), and reduced social dysfunction (p = .005).
A further RCT evaluated family-focused treatment health promoting intervention (FFT-HPI) compared with standard health education among 46 caregivers of individuals living with bipolar disorder in the USA (Perlick, Jackson, & G, Reference Perlick, Jackson and G2018). FFT-HPI was associated with greater reductions in caregiver burden postintervention and at 6-month follow-up (baseline = 0.76, 6-month follow-up = 0.26) compared with health education (baseline = 0.70, 6-month follow-up = 0.41).
Minoritised ethnic groups
Two studies considered targeted interventions for minoritised ethnic groups.
Evidence from randomised studies
An adaptation of the HF (adapted-HF) intervention was trialed for use in Canada for individuals from Black or Asian minority ethnic backgrounds (Stergiopoulos et al., Reference Stergiopoulos, Gozdzik, Misir, Skosireva, Sarang, Connelly, Whisler and McKenzie2016). Individuals with SMI who were experiencing homelessness (n = 237) were recruited to an unblinded RCT of either adapted-HF or usual care. The adapted-HF intervention employed anti-racist and anti-oppressive frameworks of practice [see (Stergiopoulos et al., Reference Stergiopoulos, O’Campo, Gozdzik, Jeyaratnam, Corneau, Sarang and Hwang2012)]. Those assigned to adapted-HF reported improved community integration over the study period (change in mean difference = 2.2, 95% CI 0.06–4.3). Assignment to adapted-HF was also associated with more housing stability compared with those assigned to usual care (adapted-HF: 75%, 95% CI 70–81, CAU: 41%, 95% CI 35–48).
Evidence from nonrandomised studies
A culturally adapted family intervention (CaFI) was co-produced to support individuals from Black African or Caribbean heritage living with schizophrenia, and their respective family members and/or key workers in the UK (Edge, Degnan, Cotterill, et al., Reference Edge, Degnan and Cotterill2018). A cultural adaptation framework was derived from a systematic review to identify and implement the essential elements required to tailor the family intervention to develop therapy and training manuals for CaFI. 92% of the family units who started CaFI completed all sessions, demonstrating feasibility. Qualitative findings also indicated acceptability of CaFI for service users, families/support members, and healthcare professionals alike.
Women experiencing intimate partner violence
Only one study reported a targeted intervention adapted for women who were accessing shelter following domestic violence.
Evidence from randomised studies
The ‘HOPE’ intervention (Helping to Overcome PTSD through Empowerment) was developed specifically for women who had been violently assaulted by a partner and were accessing shelter. Treatment modules focused on establishing safety, improving relationships, assertiveness, anger management, and postshelter concerns. HOPE was compared with an attention-matched control, ‘Present-Centered Therapy’, among 172 women in the USA (Johnson et al., Reference Johnson, Zlotnick, Hoffman, Palmieri, Johnson, Holmes and Ceroni2020). Both interventions had small-to-medium effects on mean difference severity scores for intimate partner violence between baseline and postintervention (PCT: −1.33, 95% CI: −1.63 to −1.03, HOPE: −1.32, 95% CI –1.62 to −1.02) baseline and 6-month follow-up (PCT: −1.35, 95% CI: −1.65 to −1.05, HOPE: −1.12, 95% CI: −1.42 to −0.83), and baseline and 12-month follow-up (PCT: −1.27, 95% CI: −1.57 to −0.98, HOPE: −1.02, 95% CI: −1.32 to −0.72) – and similarly for self-rated empowerment.
People with intellectual disabilities
Only one study reported targeted intervention adapted for people with an intellectual disability.
Evidence from randomised studies
In a pilot RCT conducted in the UK, participants with a comorbid intellectual disability were randomised to a befriending intervention or usual care plus access to a resource booklet of local activities (Ali et al., Reference Ali, McKenzie, Hassiotis, Priebe, Lloyd‐Evans, Jones, Panca, Omar, Finning, Moore, Roe and King2021). Befrienders were matched with participants based on shared interests and availability, aiming to provide emotional and social support and facilitate access to local activities. Befriending was found to be acceptable; however, challenges in recruiting to this study occurred, indicating a lack of feasibility for a larger RCT.
Discussion
We identified a range of targeted interventions to improve social and economic circumstances of particularly vulnerable people with mental ill-health. The interventions summarised here showed strong feasibility, acceptability and/or effectiveness across at least one social or economic outcome and highlight the potential utility for targeted interventions to improve socioeconomic inclusion for marginalised or minoritised groups. Most of these interventions were conducted in well-resourced, high-income settings, and this may limit the generalisability of findings to low- and middle-income countries or underresourced settings.
Key findings across subgroups
The evidence base was particularly strong for targeted interventions for people experiencing or at risk of homelessness. HF represented more than half of the included studies, and these studies reported replicated positive housing outcomes. The success of this bespoke intervention emphasised the benefits of interventions designed for groups with specific needs. Rather than testing generalised interventions on broader populations first, improvements in social inclusion may be most effectively achieved if interventions are designed specifically to address the needs of the most vulnerable first, in line with the framework of proportionate universalism (Carey, Crammond, & De Leeuw, Reference Carey, Crammond and De Leeuw2015).
Strikingly, with the exception of HF, there were very few replication studies resulting in a broad but heterogeneous literature base and making it difficult to draw comparisons between studies. Nevertheless, a consistent narrative emerged of the encouraging impact of targeted interventions for people with an offending history on outcomes relating to criminal behaviours, for mothers on parenting-related outcomes, and older adults on social functioning outcomes. Evidence was more disparate or sparse in relation to caregivers, people experiencing economic disadvantage, women experiencing intimate partner violence, and people with intellectual disabilities.
In particular, despite extensive research evidencing greater social adversities in people from minoritised ethnic groups, we observed a notable lack of targeted interventions for these communities – just two studies were identified (Edge et al., Reference Edge, Degnan and Cotterill2018; Stergiopoulos et al., Reference Stergiopoulos, Gozdzik, Misir, Skosireva, Sarang, Connelly, Whisler and McKenzie2016). People with mental ill-health from minoritised ethnic groups typically experience a range of social adversities, including social isolation (Morgan et al., Reference Morgan, Kirkbride, Hutchinson, Craig, Morgan, Dazzan, Boydell, Doody, Jones, Murray, Leff and Fearon2008), poorer access to vocational support (Bertram & Howard, Reference Bertram and Howard2006), and barriers to financial health (Stacey & Smith, Reference Stacey and Smith2023). Furthermore, the lack of consideration of experiences of racism, complex trauma, and migration stress in the delivery of mental health services contributes to pervasive ethnic inequalities (Bansal et al., Reference Bansal, Karlsen, Sashidharan, Cohen, Chew-Graham and Malpass2022). Together, this highlights the need for further intervention development.
Only a few of the identified interventions involved modifications for multiple marginalised characteristics, such as economic hardship, social roles, and demographics simultaneously. This is important from an intersectional perspective, as the most marginalised in society experience social exclusion across multiple domains (Filia et al., Reference Filia, Menssink, Gao, Rickwood, Hamilton, Hetrick, Parker, Herrman, Hickie, Sharmin, McGorry and Cotton2022; Kuran et al., Reference Kuran, Morsut, Kruke, Krüger, Segnestam, Orru, Nævestad, Airola, Keränen, Gabel, Hansson and Torpan2020; Villatoro, Mays, Ponce, & Aneshensel, Reference Villatoro, Mays, Ponce and Aneshensel2018), and intersectionality theory emphasises that these adversities should not be conceptualised as separable when experienced together (Crenshaw, Reference Crenshaw1989). More research is therefore warranted in this area. In a concurrent review, we identified an extensive underreporting of basic sociodemographic and intersectional features and associated stratified analyses, demonstrating key barriers to understanding what works for whom (Greenburgh et al., Reference Greenburgh, Baldwin, Weir, Asif, Laporte, Bertram and Morgan2025).
Limitations and conclusions
Several methodological limitations need to be considered in interpreting our findings. We screened for samples with diagnosed mental disorders or those who had accessed mental health services. However, many interventions exist for populations that may be vulnerable to mental ill-health but without a formal diagnosis. Thus, our approach may have missed relevant literature which is not modelled on diagnostic frameworks but rather by social circumstances. Furthermore, as we restricted our search to articles in English language and peer-reviewed journals, we likely overlooked interventions evaluated in non-English speaking countries as well as those within the grey literature. This highlights a broader problem in social intervention research, namely that key providers of support in social domains, for example third-sector organisations and local authorities, struggle to contribute to the evidence base given limited resources in tandem with day-to-day service demands.
Overall, our findings highlight that targeted social and economic interventions for people from marginalised communities who are experiencing mental ill-health may work towards addressing systemic inequalities present in mental health care. The literature base, albeit broad, is highly heterogeneous with little replication between studies. As such, these findings warrant concentrated research efforts toward existing, promising interventions to replicate findings and ultimately strengthen the evidence base to enable widespread implementation.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S0033291725101128.
Data availability statement
The data extraction spreadsheets for this review are available upon request to the corresponding author.
Acknowledgments
We would like to acknowledge Dr. Phoebe Barnett and Prof. Helen Killaspy, and their co-authors, for the huge amount of work conducted in the two reviews upon which this research was based and for their helpful advice on aspects of this review. This manuscript represents just one output from a broader program of research conducted throughout the ENRICHED project; we would like to thank Prof. Claire Henderson, Katie Chamberlain, and Madison Wempe for their valuable contributions to the ENRICHED project.
Funding statement
This work was funded by a project grant to Prof. Craig Morgan by the Maudsley Charity (The ENRICHED Project; Funding Number #2859). This work was also supported by the Economic and Social Research Council, Centre for Society and Mental Health at King’s College London [ES/S012567/1]. The views expressed are those of the author(s) and not necessarily those of the ESRC or King’s College London. JD has received funding from the Health Foundation working together with the Academy of Medical Sciences, for a Clinician Scientist Fellowship, and has received funding from the ESRC through the Centre for Society and Mental Health at King’s College London (ESRC Reference: ES/S012567/1), the National Institute for Health Research (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London and is in receipt of funding from UK Research and Innovation funding for the Population Mental Health Consortium (Grant no MR/Y030788/1) which is part of Population Health Improvement UK (PHI-UK), a national research network which works to transform health and reduce inequalities through change at the population level. The views expressed are those of the author[s] and not necessarily those of the funders, NIHR, the Department of Health and Social Care or King’s College London.
Competing interests
The authors have no conflicts of interest to disclose.
Open access
