Included studies

56 randomised controlled trials (RCTs) were identified for inclusion in this review and the model of care described was identified.

For this review, a coding system for classifying the complexity and type of service delivery model was developed by the committee specifically for the purpose of this guideline. The service delivery model was rated on this 17-item coding system to generate an overall rating between 0-20 (see Figure 1). Service delivery models scoring at least 6 were considered a collaborative care intervention. Collaborative care interventions were further sub-divided into simple collaborative care (score of 6-12) and complex collaborative care (score ≥13). Service delivery models scoring below 6 were classified as an alternative service delivery model (e.g. care coordination) or a stand-alone psychological intervention (e.g. self-help with support).

Figure 1

Coding system for service delivery models (Collaborative Care Component Score Method).

39 RCTS were categorised as collaborative care (Aragones 2012; Araya 2003; Berghofer 2012; Bjorkelund 2018; Bosanquet 2017; Bruce 2004; Buszewicz 2016; Capoccia 2004; Chen 2015; Curth 2020; Dobscha 2006; Ell 2007; Finley 2003; Fortney 2007; Gensichen 2009; Gilbody 2017/Lewis 2017; Harter 2018; Holzel 2018; Huang 2018; Huijbregts 2013; Jarjoura 2004; Jeong 2013; Katon 1999; Katzelnick 2000; Landis 2007; Ludman 2007; Morriss 2016; Ng 2020; Oladeji 2015; Richards 2013/2016; Simon 2004 (CM); Simon 2004 (CM + psych); Simon 2006; Smit 2006; Swindle 2003; Unutzer 2002/Arean 2005; Wells 2000; Yeung 2010; Yeung 2016.

Of the 39 RCTs categorised as collaborative care, 6 were categorised as complex collaborative care (score ≥13) (Fortney 2007; Holzel 2018; Huijbregts 2013; Morris 2016; Simon 2004 CM+psych; Unutzer 2002/Arean 2005) and the remaining 33 RCTs were categorised as simple collaborative care (score of 6 to 12).

1 RCT was categorised as collaborative care for relapse prevention (Katon 2001).

5 RCTs were categorised as stepped care (Adewuya 2019; Callahan 1994; Gureje 2019; Knapstad 2020; Van Der Weele 2012).

1 RCT was categorised as stepped care for relapse prevention (Apil 2012).

5 RCTs were categorised as medication management (Akerblad 2003; Aljumah 2015; Rickles 2005; Rubio-Valera 2013a; Sirey 20105).

2 RCTs were categorised as care coordination (McMahon 2007; Salisbury 2016).

1 RCT was categorised as attached professional model (Bedoya 2014).

1 RCT was categorised as shared care (Banerjee 1996).

1 RCT was categorised as measurement-based care (Guo 2015).

The included studies are summarised in Table 2 to Table 10.

Planned subgroup analyses were outlined in the full review protocol (see appendix A) to include (where possible) for all reviews, the influence of the following subgroups: chronic depression; depression with coexisting personality disorder; psychotic depression; older adults; BME populations; men. For the collaborative care review, planned subgroup analyses included the following which were informed by the collaborative care component score method (in Figure 1): type of collaborative care; stepped care component; case manager background; psychological interventions delivered as part of the model of care; number of contacts/sessions/follow-up visits provided as part of the intervention. The committee were also interested in post-hoc subgroup analyses comparing outcomes by baseline severity. Subgroup analysis was considered for all critical outcomes with at least 2 studies in each subgroup. Subgroup analysis was only possible for the collaborative care dataset, where subgroup analyses were possible for older adults, BME groups, baseline severity, and the different collaborative care components outlined above.

See the literature search strategy in appendix B and study selection flow chart in appendix C.

Excluded studies

Studies not included in this review with reasons for their exclusions are provided in appendix K.

Comparison 1. Collaborative care (simple or complex) versus standard care/enhanced standard care

Collaborative care is defined as a multi-professional approach to care for people with depression, involving a structured management plan, scheduled follow-ups and enhanced inter-professional communication. Collaborative care may also include elements of other models, such as stepped care, psychoeducation, psychological interventions or medication management.

Summaries of the studies included for the comparison of collaborative care versus standard care or enhanced standard care are presented in Table 2.

Subgroup analysis of the collaborative care dataset was possible for:

• Older adults (mean age ≥ 60 years) versus younger adults (mean age <60 years) for the following outcomes: depression symptomatology at 6 months; depression symptomatology at 12 months; response at 6 months; response at 12 months; remission at 6 months; remission at 12 months
• BME groups, comparing studies where less than 50% of the population were from a BME group with studies where 50-100% of the population were from a BME group, for the following outcome: remission at 6 months
• Baseline severity, comparing studies where the mean depression scale score indicated less severe depression (corresponding to the traditional categories of mild and subthreshold) with more severe depression (corresponding to the traditional categories of moderate and severe depression), for the following outcomes: depression symptomatology at 6 months; depression symptomatology at 12 months; response at 6 months; response at 12 months; remission at 6 months; remission at 12 months
• Type of collaborative care, simple versus complex, for the following outcomes: depression symptomatology at 6 months; depression symptomatology at 12 months; response at 6 months; response at 12 months; remission at 6 months; remission at 12 months
• Stepped care component, comparing interventions that included a stepped care component, interventions that included only a medication algorithm, and interventions with no stepped care component or algorithm, for the following outcomes: depression symptomatology at 6 months; depression symptomatology at 12 months; response at 6 months; response at 12 months; remission at 6 months; remission at 12 months
• Case manager background, comparing studies where the case manager had a prior mental health background and studies where the case manager did not have a prior mental health background, for the following outcomes: depression symptomatology at 6 months; depression symptomatology at 12 months
• Inclusion of psychological interventions, comparing studies where psychological interventions were delivered as part of the model of care with studies where psychological interventions were not part of the service delivery model, for the following outcomes: depression symptomatology at 6 months; depression symptomatology at 12 months; response at 6 months; response at 12 months; remission at 6 months; remission at 12 months
• Number of contacts provided as part of the intervention, comparing less than 13 contacts with 13 or more contacts, for the following outcomes: depression symptomatology at 6 months; depression symptomatology at 12 months; response at 6 months; response at 12 months; remission at 6 months; remission at 12 months

Table 2

Summary of included studies for Comparison 1: Collaborative care (simple or complex) versus standard care/enhanced standard care.

There were no statistically significant subgroup differences between older and younger adults for the comparison collaborative care versus standard care or enhanced standard care on: depression symptomatology at 6 months (Test for subgroup differences: Chi² = 0.74, df = 1, p = 0.39); depression symptomatology at 12 months (Test for subgroup differences: Chi² = 1.01, df = 1, p = 0.32); response at 6 months (Test for subgroup differences: Chi² = 1.34, df = 1, p = 0.25); response at 12 months (Test for subgroup differences: Chi² = 1.34, df = 1, p = 0.25); remission at 6 months (Test for subgroup differences: Chi² = 1.20, df = 1, p = 0.27); remission at 12 months (Test for subgroup differences: Chi² = 0.52, df = 1, p = 0.47). Although there was a consistent trend for larger benefits for older adults, for example, for younger adults the effect estimate for collaborative care versus standard care/enhanced standard care on depression symptomatology at 12 months was SMD −0.25 [−0.33, −0.17] (K=7; N=2865) relative to older adults where the effect estimate was SMD −0.47 [−0.88, −0.05] (K=6; N=2543).

There was no statistically significant subgroup differences between studies with a predominantly white population and studies where the majority of participants were from BME groups for the comparison collaborative care versus standard care or enhanced standard care on: remission at 6 months (Test for subgroup differences: Chi² = 0.79, df = 1, p = 0.38).

There was a statistically significant subgroup difference between studies where the mean depression scale score indicated less severe depression and studies where participants had more severe depression, for the comparison collaborative care versus standard care or enhanced standard care, on remission at 6 months (Test for subgroup differences: Chi² = 8.54, df = 1, p = 0.003). Larger benefits were observed for more severe depression populations (RR 2.31 [1.59, 3.36]; K=6; N=1273), relative to less severe depression (RR 1.21 [0.97, 1.51]; K=4; N=1076). However, this pattern was not consistent across outcomes, and subgroup differences were not statistically significant for: depression symptomatology at 6 months (Test for subgroup differences: Chi² = 0.07, df = 1, p = 0.79); depression symptomatology at 12 months (Test for subgroup differences: Chi² = 0.47, df = 1, p = 0.49); response at 6 months (Test for subgroup differences: Chi² = 0.49, df = 1, p = 0.49); response at 12 months (Test for subgroup differences: Chi² = 1.05, df = 1, p = 0.31); remission at 12 months (Test for subgroup differences: Chi² = 0.32, df = 1, p = 0.57).

There were no statistically significant subgroup differences between simple and complex collaborative care for the comparison collaborative care versus standard care or enhanced standard care on: depression symptomatology at 12 months (Test for subgroup differences: Chi² = 0.69, df = 1, p = 0.41); response at 12 months (Test for subgroup differences: Chi² = 0.17, df = 1, p = 0.68); remission at 12 months (Test for subgroup differences: Chi² = 2.79, df = 1, p = 0.09).

There were no statistically significant subgroup differences between interventions that included a stepped care component, interventions that included only a medication algorithm, and interventions with no stepped care component or algorithm for the comparison collaborative care versus standard care or enhanced standard care on: depression symptomatology at 6 months (Test for subgroup differences: Chi² = 2.33, df = 2, p = 0.31); depression symptomatology at 12 months (Test for subgroup differences: Chi² = 5.44, df = 2, p = 0.07); response at 6 months (Test for subgroup differences: Chi² = 2.07, df = 2, p = 0.36); response at 12 months (Test for subgroup differences: Chi² = 3.96, df = 2, p = 0.14); remission at 6 months (Test for subgroup differences: Chi² = 4.02, df = 2, p = 0.13); remission at 12 months (Test for subgroup differences: Chi² = 4.30, df = 2, p = 0.12). Although there was a consistent trend for larger benefits for interventions that included a stepped care component, for example, for interventions that included a stepped care component the effect estimate for collaborative care versus standard care/enhanced standard care on depression symptomatology at 12 months was SMD −0.61 [−1.10, −0.11] (K=5; N=1717) relative to interventions that included a medication algorithm-only where the effect estimate was SMD −0.10 [−0.23, 0.03] (K=3; N=1081), or no stepped care component where the effect estimate was SMD −0.25 [−0.39, −0.12] (K=5; N=2610).

There were no statistically significant subgroup differences between interventions where the case manager had a prior mental health background and interventions where the case manager did not have a prior mental health background, for the comparison collaborative care versus standard care or enhanced standard care on: depression symptomatology at 6 months (Test for subgroup differences: Chi² = 0.18, df = 1, p = 0.67); depression symptomatology at 12 months (Test for subgroup differences: Chi² = 1.02, df = 1, p = 0.31).

There were no statistically significant subgroup differences between studies where psychological interventions were delivered as part of the model of care and studies where psychological interventions were not part of the service delivery model, for the comparison collaborative care versus standard care or enhanced standard care on: depression symptomatology at 6 months (Test for subgroup differences: Chi² = 0.00, df = 1, p = 0.98); depression symptomatology at 12 months (Test for subgroup differences: Chi² = 0.01, df = 1, p = 0.91); response at 6 months (Test for subgroup differences: Chi² = 0.01, df = 1, p = 0.94); response at 12 months (Test for subgroup differences: Chi² = 0.14, df = 1, p = 0.71); remission at 6 months (Test for subgroup differences: Chi² = 1.12, df = 1, p = 0.29); remission at 12 months (Test for subgroup differences: Chi² = 0.09, df = 1, p = 0.76).

There was a statistically significant subgroup difference between interventions with fewer than 13 contacts and interventions with 13 or more contacts, for the comparison collaborative care versus standard care or enhanced standard care, on remission at 12 months (Test for subgroup differences: Chi² = 4.23, df = 1, p = 0.04). Interventions with 13+ contacts showed larger benefits (RR 1.97 [1.33, 2.91]; K=8; N=3188) than interventions with <13 contacts (RR 1.25 [1.06, 1.48]; K=6; N=3067). Although heterogeneity remained fairly high within (as well as between) subgroups, with I² values of 79% for interventions with 13+ contacts and 56% for interventions with <13 contacts. There was a trend for larger benefits associated with more contacts across other outcomes, although subgroup differences were not statistically significant for: depression symptomatology at 6 months (Test for subgroup differences: Chi² = 0.35, df = 1, p = 0.55); depression symptomatology at 12 months (Test for subgroup differences: Chi² = 1.13, df = 1, p = 0.29); response at 6 months (Test for subgroup differences: Chi² = 0.02, df = 1, p = 0.88); response at 12 months (Test for subgroup differences: Chi² = 0.41, df = 1, p = 0.52); remission at 6 months (Test for subgroup differences: Chi² = 0.84, df = 1, p = 0.36).

Comparison 2. Collaborative care versus standard care for relapse prevention

Collaborative care can also be used for those in full or partial remission from depression, particularly those at higher risk of relapse, as a strategy to keep well.

A summary of the study included for the comparison of collaborative care versus standard care for relapse prevention is presented in Table 3.

Table 3

Summary of included studies for Comparison 2: Collaborative care versus standard care for relapse prevention.

Comparison 3. Stepped care versus standard care/enhanced standard care

Stepped care provides the most effective yet least burdensome treatment for people with depression first, but if a person does not benefit from an initial intervention they are ‘stepped up’ to a more complex intervention. Typically, stepped care starts by providing a low intensity intervention, but in patient-specific stepped care a higher intensity intervention may be commenced if, for example, a person is very ill or suicidal and a low intensity intervention would not be appropriate.

Summaries of the studies included for the comparison of stepped care versus standard care or enhanced standard care are presented in Table 4.

Table 4

Summary of included studies for Comparison 3: Stepped care versus standard care/enhanced standard care.

Comparison 4. Stepped care versus standard care for relapse prevention

Stepped care can also be used for those in full or partial remission from depression, as a strategy to keep well.

A summary of the study included for the comparison of stepped care versus standard care for relapse prevention is presented in Table 5.

Table 5

Summary of included studies for Comparison 4: Stepped care versus standard care for relapse prevention.

Comparison 5. Pure medication management versus standard care

Medication management can be a component of a broader service delivery model (for example, as part of collaborative care) or as a stand-alone intervention (pure medication management). Medication management is an intervention to ensure medication taken for depression has the greatest opportunity to be effective, by working with people to increase understanding of their medication, promote adherence, ensure adequate therapeutic levels are obtained, and allow people to discuss their medicine use and so reduce unnecessary discontinuation of medication due to lack of benefits or side effects.

Summaries of the studies included for the comparison of pure medication management versus standard care are presented in Table 6Table 4.

Table 6

Summary of included studies for Comparison 5: Pure medication management versus standard care.

Comparison 6. Care coordination versus standard care/enhanced standard care

Care coordination can be a component of a broader service delivery model (for example, as part of collaborative care) or as a stand-alone intervention. Care coordination (also known as case management) is a system where an individual healthcare professional takes responsibility for the coordination of the care of a person with depression, but is not necessarily directly involved in the provision of any intervention; it may also involve the coordination of follow-up.

Summaries of the studies included for the comparison of care coordination versus standard care or enhanced standard care are presented in Table 7Table 4.

Table 7

Summary of included studies for Comparison 6: Care coordination versus standard care/enhanced standard care.

Comparison 7. Attached professional model versus enhanced standard care

In this model a mental health professional has direct responsibility for the care of a person (usually in primary care) focusing on the primary treatment of the depression. The coordination of care remains with the GP/primary care team. Contact with the attached professional is usually limited to treatment and involves little or no follow-up beyond that determined by the specific intervention offered (for example, booster sessions in CBT).

A summary of the study included for the comparison of attached professional model versus enhanced standard care is presented in Table 8.

Table 8

Summary of included studies for Comparison 7: Attached professional model versus enhanced standard care.

Comparison 8. Shared care versus standard care

Shared care is the involvement of a multidisciplinary team who work together to plan and deliver individualised care for people with depression. The team will usually include involvement from both primary care and specialist services.

A summary of the study included for the comparison of shared care versus standard care is presented in Table 9.

Table 9

Summary of included studies for Comparison 8: Shared care versus standard care.

Comparison 9. Measurement-based care versus standard care

Measurement-based care is similar to stepped care with defined levels of treatment but progression to different steps or alternative treatments is guided by the use of a predefined algorithm that utilises objective measures of efficacy.

A summary of the study included for the comparison of measurement-based care versus standard care is presented in Table 10.

Table 10

Summary of included studies for Comparison 9: Measurement-based care versus standard care.

See the full evidence tables in appendix D and the forest plots in appendix E.

Included studies

A single economic search was undertaken for all topics included in the scope of this guideline. See the literature search strategy in appendix B and economic study selection flow chart in appendix G. Details on the hierarchy of inclusion criteria for economic studies are provided in supplement 1 (methods supplement).

The systematic search of the literature identified 12 studies (in 13 publications) on the cost effectiveness of different models for the coordination and delivery of services for adults with depression.

There were 3 UK studies that assessed simple collaborative care (Bosanquet 2017; Green 2014; Lewis 2017) and 1 UK study that assessed complex collaborative care (Morriss 2016). Following the hierarchy of inclusion criteria regarding country settings, 1 Dutch (Goorden 2015) and 1 German (Grochtdreis 2019) studies assessing the cost effectiveness of complex collaborative care were also included in the review. In addition, the search identified 1 US study assessing the cost effectiveness of simple collaborative care in relapse prevention (Simon 2002) and given that the study focused on a different population that was not covered by UK studies or other studies ranking higher on the hierarchy of inclusion criteria, this study was also included in the review.

One UK study assessed the cost effectiveness of stepped care (Mukuria 2013). Following the hierarchy of inclusion criteria regarding country settings, 2 Dutch (van der Weele 2012, Meeuwissen 2019) and 1 Canadian economic study (Health Quality Ontario 2019) were also included in the economic review of stepped care.

No UK studies on the cost effectiveness of medication management for adults with depression were identified. Following the hierarchy of inclusion criteria regarding country settings, 1 Spanish study (Rubio-Valera 2013) was included in the review.

No UK studies on the cost effectiveness of shared care for adults with depression were identified. Following the hierarchy of inclusion criteria regarding country settings, 1 US study (Wiley-Exley 2009) was included in the review.

No studies assessing the cost effectiveness of care coordination, the attached professional model, or measurement-based care for adults with depression were identified.

Economic evidence tables are provided in appendix H. Economic evidence profiles are shown in appendix I.

Excluded studies

A list of excluded economic and utility studies, with reasons for exclusion, is provided in supplement 3 - Economic evidence included & excluded studies.

Summary of studies included in the economic evidence review

Clinical evidence statements

Economic evidence statements

Interpreting the evidence

Cost effectiveness and resource use

The committee agreed that, overall, the published economic evidence indicated that simple collaborative care is potentially a cost-effective model for delivering services to adults with depression, including older adults. This is because out of the 3 UK cost-utility studies included in the review, 2 found simple collaborative care cost-effective when added to usual primary care compared with usual primary care alone at the NICE lower cost-effectiveness threshold of £20,000/QALY. The third study reported an ICER for simple collaborative care between the NICE lower and upper (£30,000/QALY) cost-effectiveness thresholds. The two studies that found simple collaborative care cost-effective were also somewhat larger than the one that found it cost-ineffective at the lower NICE cost-effectiveness threshold. The committee also noted that, among the 3 studies, there was one with minor methodological limitations (the other two were characterised by potentially serious limitations), and this found simple collaborative care to be cost-effective. In contrast, the only UK study on more resource-intensive complex collaborative care included in the review suggested this is unlikely to be cost-effective compared with usual secondary mental health care, as its ICER was well above the NICE upper cost-effectiveness threshold of £30,000/QALY. Therefore, the committee decided to recommend collaborative care with the characteristics of the less resource-intensive, simple collaborative care, as defined in this review, for organising the delivery of care and treatment of people with depression.

The committee noted, based on the evidence, that stepped care might also be cost-effective for adults with depression. They therefore agreed that the recommendations they made on treatment, which reflected the key principles of stepped (or, more accurately, matched) care, ensured efficient use of resources.

The committee noted that no UK economic evidence was available and non-UK evidence did not provide any substantial support for the cost effectiveness of medication management as an independent service delivery model for adults with depression. They also noted that non-UK economic evidence on shared care was inconclusive.

The committee acknowledged that referring people with more severe depression or chronic depressive symptoms and multiple complicating problems (such as unemployment, poor housing or financial problems) or significant coexisting conditions to specialist mental health services, if they have not benefitted from treatment or if they have impaired functioning, is likely to incur additional costs compared with no referral. However, they agreed that the number of people affected would be small and any additional costs were likely to be offset by cost-savings resulting from more appropriate care for this population following referral (compared with treatment in primary care settings), leading to improved outcomes and reduction in the need for potentially costly care further down the care pathway.

Included studies

No randomised controlled trial (RCT) evidence was identified that specifically addressed the following settings: primary care, and inpatient care, therefore, as specified in the full protocol (see Appendix A), indirect evidence was considered in the form of sub-analyses of the NMA dataset (Evidence report B: Treatment of a new episode of depression).

Comparison 1. Primary care versus secondary care

As outlined above, no RCT evidence was identified that specifically addressed this comparison. Therefore the committee considered indirect evidence in the form of sub-analyses of the NMA dataset (Evidence report B: Treatment of a new episode of depression). Primary versus secondary care differences were examined for critical outcomes that had more than 2 studies in each subgroup.

Subgroup analysis of primary care versus secondary care was possible for 5 comparisons in the NMA dataset, and all 5 comparisons were for adults with more severe depression (corresponding to the categories of moderate and severe depression):

• Comparison 1a. Cognitive and cognitive behavioural therapies individual + antidepressant versus antidepressant, with 2 RCTs included for primary care (Naeem 2011; Scott 1997) and 4 RCTs included for secondary care (Ashouri 2013; Hautzinger 1996; Hollon 1992; Zu 2014). Primary care versus secondary care subgroup analysis was possible for the depression symptoms endpoint outcome only.
• Comparison 1b. Selective serotonin reuptake inhibitors (SSRIs) versus placebo, with 5 RCTs included for primary care (Bjerkenstedt 2005; Doogan 1994; Lepola 2003; Lopez-Rodriguez 2004; Wade 2002) and 78 RCTs included for secondary care (29060 07 001; Andreoli 2002/ Dubini 1997/ Massana 1998_study 1 [1 study reported across 3 papers]; Baune 2018; Binnemann 2008; Bose 2008; Burke 2002; Byerley 1988; Claghorn 1992a; Claghorn 1992b; Clayton 2006_study 1; Clayton 2006_study 2; CL3-20098-022; CL3-20098-023; CL3-20098-024; Detke 2004; Dube 2010; Dunbar 1993; Eli Lilly HMAT-A; Emsley 2018; Fabre 1992; Fabre 1995a; Fava 1998a; Fava 2005; FDA 245 (EMD 68 843-010); FDA 246 (SB 659746-003); Forest Laboratories 2000; Forest Research Institute 2005; Golden 2002_448; Golden 2002_449; Goldstein 2002; Goldstein 2004; Gual 2003; Higuchi 2009; Hirayasu 2011a; Hirayasu 2011b; Jefferson 2000; Kasper 2012; Katz 2004; Keller 2006_Study 062; Kramer 1998; Kranzler 2006_Group A; Lam 2016; Macias-Cortes 2015; Mathews 2015; Mendels 1999; Miller 1989a; Montgomery 1992; Mundt 2012; MY-1042/BRL-029060/CPMS-251; MY-1042/BRL-029060/1 (PAR 128); Nemeroff 2007; Nierenberg 2007; NKD20006 (NCT00048204); Nyth 1992; Olie 1997; PAR 01 001 (GSK & FDA); Perahia 2006; Peselow 1989a; Peselow 1989b; Rapaport 2009; Ratti 2011_study 096; Ravindran 1995; Reimherr 1990; Rickels 1992; Rudolph 1999; SER 315 (FDA); Sheehan 2009b; Smith 1992; Stark 1985; Study 62b (FDA); Study F1J-MC-HMAQ- Study Group B; Tollefson 1993/1995 [1 study reported across 2 papers]; Valle-Cabrera 2018; VEN XR 367 (FDA); Wang 2014c; WELL AK1A4006; Wernicke 1987; Wernicke 1988). Primary care versus secondary care subgroup analyses were possible for the depression symptoms endpoint, depression symptoms change score, and response outcomes.
• Comparison 1c. SSRIs versus tricyclic antidepressants (TCAs), with 10 RCTs included for primary care (Christiansen 1996; Freed 1999; Hutchinson 1992; Kyle 1998; Moon 1994; Moon 1996; PAR 29060/281; PAR MDUK 032; Rosenberg 1994; Serrano-Blanco 2006) and 47 RCTs included for secondary care (29060 07 001; 29060/299; Akhondzadeh 2003; Arminem 1992; Beasley 1993b; Bersani 1994; Bhargava 2012; Bremner 1984; Byerley 1988; Chiu 1996; Cohn 1984b; Cohn 1990b; Danish University Antidepressant Group 1986; Danish University Antidepressant Group 1990; De Ronchi 1998; Demyttenaere 1998; Deuschle 2003; Fabre 1991; Fabre 1992; Fawcett 1989; Feighner 1993; Forlenza 2001; Geretsegger 1995; GSK_29060/103; Hashemi 2012; Keegan 1991; Laakmann 1988; Laakmann 1991; Levine 1989; Marchesi 1998; MDF/29060/III/070/88/MC; Miura 2000; Moller 1993; Moller 1998; Mulsant 1999; Navarro 2001; Ontiveros Sanchez 1998; Peselow 1989a; Peselow 1989b; Peters 1990; Preskorn 1991; Reimherr 1990; Ropert 1989; SER 315 (FDA); Staner 1995; Stark 1985; Suleman 1997). Primary care versus secondary care subgroup analyses were possible for the depression symptoms endpoint, depression symptoms change score, remission and response outcomes.
• Comparison 1d. TCAs versus placebo, with 6 RCTs included for primary care (Barge-Schaapveld 2002; Blashki 1971; Lecrubier 1997; Mynors-Wallis 1995; Philipp 1999; Schweizer 1998) and 30 RCTs included for secondary care (29060 07 001; Amsterdam 1986; Bakish 1992b; Bremner 1995; Byerley 1988; Cassano 1986; Elkin 1989/Imber 1990 [1 study reported across 2 papers]; Escobar 1980; Fabre 1992; Feiger 1996; Feighner 1982; Feighner 1989b; Fontaine 1994; Goldberg 1980; Kusalic 1993; McCallum 1975; MIR 003-020 (FDA); Peselow 1989a; Peselow 1989b; Reimherr 1990; Rickels 1982e; Rickels 1991; Rickels 1995_Study 006-1; Rickels 1995_Study 006-2; Schweizer 1994; SER 315 (FDA); Silverstone 1994; Smith 1990; Stark 1985; White 1984). Primary care versus secondary care subgroup analyses were possible for the depression symptoms endpoint, depression symptoms change score, and response outcomes.
• Comparison 1e. Serotonin–norepinephrine reuptake inhibitors (SNRIs) versus SSRIs, with 2 RCTs included for primary care (Montgomery 2004; Tylee 1997) and 29 RCTs included for secondary care (Allard 2004; Alves 1999; Bielski 2004; Clerc 1994; Costa 1998; DeNayer 2002; Detke 2004; Diaz-Martinez 1998; Dierick 1996; Eli Lilly HMAT-A; Goldstein 2002; Goldstein 2004; Hao 2014; Higuchi 2009; Hwang 2004; Jiang 2017; Khan 2007; Kornaat 2000; Mehtonen 2000; Nemeroff 2007; Nierenberg 2007; Perahia 2006; Rickels 2000; Rudolph 1999; Sheehan 2009b; Shelton 2006; Sir 2005; Study F1J-MC-HMAQ-Study Group B; Tzanakaki 2000). Primary care versus secondary care subgroup analyses were possible for the remission and response outcomes.

Comparison 2. Crisis resolution team care versus standard care (for adults with non-psychotic severe mental illness)

No RCT evidence was identified that specifically addressed this comparison for adults with depression. The committee therefore agreed to consider a wider evidence base including non-psychotic severe mental illness. A systematic review (Murphy 2015; updated version of Joy 2003 used in 2009 guideline) was identified that examined crisis intervention for people with severe mental illness. This Cochrane review was used as a source of studies with inclusion criteria into this review of over 50% of the population having a non-psychotic disorder. Of the 8 RCTs included in Murphy 2015, 1 of these studies met the >50% non-psychotic disorder inclusion criterion (Johnson 2005).

Comparison 3. Inpatient versus outpatient settings

No randomised controlled trial (RCT) evidence was identified that specifically addressed this comparison. Therefore the committee considered indirect evidence in the form of sub-analyses of the NMA dataset (Evidence report B: Treatment of a new episode of depression). Differences between inpatient and outpatient settings were examined for critical outcomes that had more than 2 studies in each subgroup.

Subgroup analysis of inpatient versus outpatient settings was possible for 6 comparisons in the NMA dataset, and all 6 comparisons were for adults with more severe depression (corresponding to the categories of moderate and severe depression):

• Comparison 3a. Selective serotonin reuptake inhibitors (SSRIs) versus placebo, with 3 RCTs included for inpatient settings (29060 07 001; Katz 2004; Sheehan 2009b) and 74 RCTs included for outpatient settings (Baune 2018; Binnemann 2008; Bjerkenstedt 2005; Blumenthal 2007/Hoffman 2011 [1 study reported across 2 papers]; Bose 2008; Burke 2002; Byerley 1988; Claghorn 1992a; Claghorn 1992b; Clayton 2006_study 1; Clayton 2006_study 2; Detke 2004; Doogan 1994; Dube 2010; Dunbar 1993; Eli Lilly HMAT-A; Emsley 2018; Fabre 1992; Fava 1998a; Fava 2005; FDA 245 (EMD 68 843-010); Forest Laboratories 2000; Forest Research Institute 2005; Golden 2002_448; Golden 2002_449; Goldstein 2002; Goldstein 2004; Gual 2003; Hirayasu 2011a; Hirayasu 2011b; Hunter 2010_study 1; Hunter 2011; Jefferson 2000; Keller 2006_Study 062; Komulainen 2018; Kramer 1998; Kranzler 2006_Group A; Lam 2016; Lepola 2003; Macias-Cortes 2015; Mathews 2015; Mendels 1999; Miller 1989a; Mundt 2012; MY-1042/BRL-029060/CPMS-251; MY-1045/BRL-029060/1 (PAR 128); Nemeroff 2007; Nierenberg 2007; NKD20006 (NCT00048204); Olie 1997; PAR 01 001 (GSK & FDA); Perahia 2006; Peselow 1989a; Peselow 1989b; Rapaport 2009; Ratti 2011_study 096; Ravindran 1995; Reimherr 1990; Rickels 1992; Roose 2004; Rudolph 1999; SER 315 (FDA); Smith 1992; Stark 1985; Study 62b (FDA); Study F1J-MC-HMAQ – Study Group B; Tollefson 1993/1995 [1 study reported across 2 papers]; Valle-Cabrera 2018; VEN XR 367 (FDA); Wade 2002; Wang 2014c; WELL AK1A4006; Wernicke 1987; Wernicke 1988). Inpatient versus outpatient subgroup analysis was possible for the depression symptoms change score and response outcomes.
• Comparison 3b. SSRIs versus tricyclic antidepressants (TCAs), with 11 RCTs included for inpatient settings (29060/299; 29060 07 001; Arminen 1992; Danish University Antidepressant Group 1986; Danish University Antidepressant Group 1990; Deushle 2003; Geretsegger 1995; Laakmann 1991; Moller 1993; Moller 1998; Staner 1995), and 40 RCTs included for outpatient settings (Akhondzadeh 2003; Beasley 1993b; Bersani 1994; Bhargava 2012; Bremner 1984; Byerley 1988; Christiansen 1996; Cohn 1984b; Cohn 1990b; De Ronchi 1998; Demyttenaere 1998; Fabre 1991; Fabre 1992; Fawcett 1989; Feighner 1993; Forlenza 2001; Freed 1999; Hashemi 2012; Hutchinson 1992; Kyle 1998; Laakmann 1988; Marchesi 1998; MDF/29060/III/070/88/MC; Moller 2000; Moon 1994; Moon 1996; Ontiveros Sanchez 1998; PAR 29060/281; PAR MDUK 032; Peselow 1989a; Peselow 1989b; Peters 1990; Preskorn 1991; Reimherr 1990; Ropert 1989; Rosenberg 1994; SER 315 (FDA); Serrano-Blanco 2006; Stark 1985; Suleman 1997). Inpatient versus outpatient subgroup analysis was possible for the depression symptoms endpoint, depression symptoms change score, remission, and response outcomes.
• Comparison 3c. Serotonin–norepinephrine reuptake inhibitors (SNRIs) versus placebo, with 2 RCTs included for inpatient settings (Guelfi 1995; Sheehan 2009b), and 26 RCTs included for outpatient settings (Brannan 2005; Cutler 2009; Detke 2002a; Detke 2002b; Detke 2004; Eli Lilly HMAT-A; Goldstein 2002; Goldstein 2004; Hewett 2009; Hewett 2010; Higuchi 2016; Khan 1998; Levin 2013; Mendels 1993; Nemeroff 2007; Nierenberg 2007; Perahia 2006; Raskin 2007; Robinson 2014; Rudolph 1999; Schweizer 1994; Study F1J-MC-HMAQ-Study Group B; Thase 1997; VEN 600A-303 (FDA); VEN 600A-313 (FDA); VEN XR 367 (FDA)). Inpatient versus outpatient subgroup analysis was possible for the depression symptoms endpoint, depression symptoms change score, and remission outcomes.
• Comparison 3d. SNRIs versus SSRIs, with 4 RCTs included for inpatient settings (Clerc 1994; Hwang 2004; Sheehan 2009b; Tzanakaki 2000), and 32 RCTs included for outpatient settings (Allard 2004; Alves 1999; Bielski 2004; Casabona 2004; Chang 2015; Costa 1998; DeNayer 2002; Detke 2004; Diaz-Martinez 1998; Dierick 1996; Eli Lilly HMAT-A; Goldstein 2002; Goldstein 2004; Hackett 1996; Heller 2009; Jiang 2017; Khan 2007; Kornaat 2000; Mehtonen 2000; Montgomery 2004; Mowla 2016; Nemeroff 2007; Nierenberg 2007; Perahia 2006; Rickels 2000; Rudolph 1999; Shelton 2006; Sir 2005; Study F1J-MC-HMAQ-Study Group B; Tylee 1997; VEN XR 367 (FDA); Wade 2007) . Inpatient versus outpatient subgroup analysis was possible for the depression symptoms endpoint, depression symptoms change score, remission, and response outcomes.
• Comparison 3e. Mirtazapine versus TCAs, with 2 RCTs included for inpatient settings (Richou 1995; Zivkov 1995), and 4 RCTs included for outpatient settings (Bremner 1995; MIR 003-020 (FDA); MIR 003-021 (FDA); Smith 1990). Inpatient versus outpatient subgroup analysis was only possible for the response outcome.
• Comparison 3f. Acupuncture + antidepressants versus antidepressants, with 2 RCTs included for inpatient settings (Wang 2014a; Zhang 2007a), and 2 RCTs included for outpatient settings (Qu 2013; Zhao 2019a). Inpatient versus outpatient subgroup analysis was only possible for the depression symptoms change score outcome.

Comparison 4. Acute psychiatric day hospital care versus inpatient care (for adults with depression and non-psychotic severe mental illness)

Acute psychiatric day hospitals are units that provide diagnostic and treatment services for acutely ill individuals who would otherwise be treated in traditional psychiatric inpatient units. 1 RCT (Dinger 2014) was identified that specifically addressed acute psychiatric day hospital care for adults with depression. The committee therefore agreed to consider a wider evidence base including non-psychotic severe mental illness. A systematic review (Marshall 2011) was identified that compared day hospital to inpatient care for people with acute psychiatric disorders. This Cochrane review was used as a source of studies with inclusion criteria into this review of over 50% of the population having a non-psychotic disorder.

Of the 10 RCTs included in Marshall 2011, 5 of these studies met the >50% non-psychotic disorder inclusion criterion (Creed 1990; Creed 1997; Dick 1985; Kallert 2007; Schene 1993).

Comparison 5. Non-acute day hospital care versus outpatient care (for adults with depression and non-psychotic severe mental illness)

No RCT evidence was identified that specifically addressed this setting for adults with depression. The committee therefore agreed to consider a wider evidence base including non-psychotic severe mental illness. A systematic review (Marshall 2001) was identified that examined the use of day hospitals as an alternative to outpatient care for people with psychiatric disorders. This Cochrane review was used as a source of studies with inclusion criteria into this review of over 50% of the population having a non-psychotic disorder.

Of the 8 studies included in Marshall 2001, 3 of these studies met the >50% non-psychotic disorder inclusion criterion (Dick 1991; Glick 1986; Tyrer 1979).

Comparison 6. Community mental health teams versus standard care (for adults with non-psychotic severe mental illness)

No RCT evidence was identified that specifically addressed this setting for adults with depression. The committee therefore agreed to consider a wider evidence base including non-psychotic severe mental illness. A systematic review (Malone 2007) was identified that examined community mental health teams (CMHTs) for people with severe mental illnesses and disordered personality. This Cochrane review was used as a source of studies with inclusion criteria into this review of over 50% of the population having a non-psychotic disorder.

Of the 3 studies included in Malone 2007, 1 of these studies met the >50% non-psychotic disorder inclusion criterion (Merson 1992).

See the literature search strategy in appendix B and study selection flow chart in appendix C.

Excluded studies

Studies not included in this review with reasons for their exclusions are provided in appendix K.

Comparison 1. Primary care versus secondary care

Summaries of the studies included in the primary care versus secondary care subgroup analysis of the cognitive and cognitive behavioural therapies individual + antidepressant versus antidepressant comparison are presented in Table 12.

There were no significant subgroup differences between primary care and secondary care for the comparison cognitive and cognitive behavioural therapies individual + antidepressant versus antidepressant on: depression symptoms endpoint (Test for subgroup differences: Chi² = 0.27, df = 1, p = 0.60).

Table 12

Summary of included studies for primary care versus secondary care subgroup analysis for comparison 1a Cognitive and cognitive behavioural therapies individual + antidepressant versus antidepressant.

Summaries of the studies included in the primary care versus secondary care subgroup analysis of the selective serotonin reuptake inhibitors (SSRIs) versus placebo comparison are presented in Table 13.

There were no significant subgroup differences between primary care and secondary care for the comparison SSRIs versus placebo on: depression symptoms endpoint (Test for subgroup differences: Chi² = 0.01, df = 1, p = 0.91); depression symptoms change score (Test for subgroup differences: Chi² = 0.26, df = 1, p = 0.61); response (Test for subgroup differences: Chi² = 1.75, df = 1, p = 0.19).

Table 13

Summary of included studies for primary care versus secondary care subgroup analysis for comparison 1b Selective serotonin reuptake inhibitors (SSRIs) versus placebo.

Summaries of the studies included in the primary care versus secondary care subgroup analysis of the SSRIs versus tricyclic antidepressants (TCAs) comparison are presented in Table 14.

There were no significant subgroup differences between primary care and secondary care for the comparison SSRIs versus TCAs on: depression symptoms endpoint (Test for subgroup differences: Chi² = 0.09, df = 1, p = 0.76); depression symptoms change score (Test for subgroup differences: Chi² = 1.46, df = 1, p = 0.23); remission (Test for subgroup differences: Chi² = 2.19, df = 1, p = 0.14); response (Test for subgroup differences: Chi² = 2.22, df = 1, p = 0.14).

Table 14

Summary of included studies for primary care versus secondary care subgroup analysis for comparison 1c SSRIs versus tricyclic antidepressants (TCAs).

Summaries of the studies included in the primary care versus secondary care subgroup analysis of the TCAs versus placebo comparison are presented in Table 15.

There were no significant subgroup differences between primary care and secondary care for the comparison TCAs versus placebo on: depression symptoms endpoint (Test for subgroup differences: Chi² = 0.49, df = 1, p = 0.49); depression symptoms change score (Test for subgroup differences: Chi² = 0.32, df = 1, p = 0.57); response (Test for subgroup differences: Chi² = 2.87, df = 1, p = 0.09).

Table 15

Summary of included studies for primary care versus secondary care subgroup analysis for comparison 1d TCAs versus placebo.

Summaries of the studies included in the primary care versus secondary care subgroup analysis of the serotonin–norepinephrine reuptake inhibitors (SNRIs) versus SSRIs comparison are presented in Table 16.

There were no significant subgroup differences between primary care and secondary care for the comparison SNRIs versus SSRIs on: remission (Test for subgroup differences: Chi² = 1.55, df = 1, p = 0.21); response (Test for subgroup differences: Chi² = 0.62, df = 1, p = 0.43).

Table 16

Summary of included studies for primary care versus secondary care subgroup analysis for comparison 1e Serotonin–norepinephrine reuptake inhibitors (SNRIs) versus SSRIs.

Comparison 2. Crisis resolution team care versus standard care (for adults with non-psychotic severe mental illness)

Summary of the study included in the crisis resolution team care and standard care comparison is presented in Table 17.

Table 17

Summary of included studies for comparison 2 Crisis resolution versus standard care.

Comparison 3. Inpatient versus outpatient settings

Summaries of the studies included in the inpatient versus outpatient subgroup analysis of the selective serotonin reuptake inhibitors (SSRIs) versus placebo comparison are presented in Table 18Table 38.

There were no significant subgroup differences between inpatient and outpatient settings for the comparison SSRIs versus placebo on: depression symptoms change score (Test for subgroup differences: Chi² = 2.47, df = 1, p = 0.12); response (Test for subgroup differences: Chi² = 0.11, df = 1, p = 0.74).

Table 18

Summary of included studies for inpatient versus outpatient subgroup analysis for comparison 3a Selective serotonin reuptake inhibitors (SSRIs) versus placebo.

Summaries of the studies included in the inpatient versus outpatient subgroup analysis of the SSRIs versus tricyclic antidepressants (TCAs) comparison are presented in Table 19.

There were no significant subgroup differences between inpatient and outpatient settings for the comparison SSRIs versus TCAs on: depression symptoms endpoint (Test for subgroup differences: Chi² = 1.08, df = 1, p = 0.30); remission (Test for subgroup differences: Chi² = 2.11, df = 1, p = 0.15); response (Test for subgroup differences: Chi² = 1.03, df = 1, p = 0.31). There was a statistically significant subgroup difference between inpatient and outpatient settings for depression change score (Test for subgroup differences: Chi² = 7.03, df = 1, p = 0.008). In inpatient settings TCAs showed a small benefit over SSRIs (SMD 0.27 [0.08, 0.47]), whereas in outpatient settings SSRIs showed a small benefit over TCAs (SMD −0.05 [−0.19, 0.09]), however, in both inpatient and outpatient settings the difference between TCAs and SSRIs was non-significant.

Table 19

Summary of included studies for inpatient versus outpatient subgroup analysis for comparison 3b SSRIs versus tricyclic antidepressants (TCAs).

Summaries of the studies included in the inpatient versus outpatient subgroup analysis of the serotonin–norepinephrine reuptake inhibitors (SNRIs) versus placebo comparison are presented in Table 20.

There were no significant subgroup differences between inpatient and outpatient settings for the comparison SNRIs versus placebo on: depression symptoms endpoint (Test for subgroup differences: Chi² = 0.03, df = 1, p = 0.87); depression symptoms change score (Test for subgroup differences: Chi² = 3.12, df = 1, p = 0.08); remission (Test for subgroup differences: Chi² = 0.25, df = 1, p = 0.62).

Table 20

Summary of included studies for inpatient versus outpatient subgroup analysis for comparison 3c Serotonin–norepinephrine reuptake inhibitors (SNRIs) versus placebo.

Summaries of the studies included in the inpatient versus outpatient subgroup analysis of the SNRIs versus SSRIs comparison are presented in Table 21.

There were no significant subgroup differences between inpatient and outpatient settings for the comparison SNRIs versus SSRIs on: depression symptoms endpoint (Test for subgroup differences: Chi² = 2.03, df = 1, p = 0.15); remission (Test for subgroup differences: Chi² = 1.08, df = 1, p = 0.30); response (Test for subgroup differences: Chi² = 0.49, df = 1, p = 0.48). There was a statistically significant subgroup difference between inpatient and outpatient settings for depression change score (Test for subgroup differences: Chi² = 8.03, df = 1, p = 0.005). SNRIs showed a benefit over SSRIs in both settings, although this effect was larger in inpatient settings (SMD −0.48 [−0.73, −0.23]) relative to outpatient settings (SMD −0.09 [−0.19, 0.01]), however, this was a difference in magnitude rather than direction and even in inpatient settings the difference was not clinically important.

Table 21

Summary of included studies for inpatient versus outpatient subgroup analysis for comparison 3d SNRIs versus SSRIs.

Summaries of the studies included in the inpatient versus outpatient subgroup analysis of the mirtazapine versus TCAs comparison are presented in Table 22Table 38.

There was not a significant subgroup difference between inpatient and outpatient settings for the comparison mirtazapine versus TCAs on response (Test for subgroup differences: Chi² = 0.19, df = 1, p = 0.66).

Table 22

Summary of included studies for inpatient versus outpatient subgroup analysis for comparison 3e Mirtazapine versus TCAs.

Summaries of the studies included in the inpatient versus outpatient subgroup analysis of the acupuncture + antidepressants versus antidepressants comparison are presented in Table 23Table 38.

There was not a significant subgroup difference between inpatient and outpatient settings for the comparison acupuncture + antidepressants versus antidepressants on depression symptoms change score (Test for subgroup differences: Chi² = 1.18, df = 1, p = 0.28).

Table 23

Summary of included studies for inpatient versus outpatient subgroup analysis for comparison 3f Acupuncture + antidepressants versus antidepressants.

Comparison 4. Acute psychiatric day hospital care versus inpatient care (for adults with depression and non-psychotic severe mental illness)

Table 24

Summary of included studies for comparison 4 acute psychiatric day hospital versus inpatient care.

Comparison 5. Non-acute day hospital care versus outpatient care (for adults with depression and non-psychotic severe mental illness)

Table 25

Summary of included studies for comparison 5 non-acute day hospital versus outpatient care.

Comparison 6. Community mental health teams versus standard care (for adults with non-psychotic severe mental illness)

Table 26

Summary of included studies for comparison 6 community mental health teams versus standard care.

See the full evidence tables in appendix D and the forest plots in appendix E.

Included studies

A single economic search was undertaken for all topics included in the scope of this guideline but no economic studies were identified which were applicable to this review question. See the literature search strategy in appendix B and economic study selection flow chart in appendix G.

Excluded studies

A list of excluded economic and utility studies, with reasons for exclusion, is provided in supplement 3 - Health economic included & excluded studies.

Clinical evidence statements

Economic evidence statements

No economic evidence was identified which was applicable to this review question.

Interpreting the evidence

Cost effectiveness and resource use

No evidence on the cost-effectiveness of different settings for the delivery of care for adults with depression was identified and no further economic analysis was undertaken. The committee considered the costs associated with crisis resolution and home treatment and estimated that these are higher than routine primary care but significantly lower than inpatient care. The committee expressed the opinion that, compared with routine primary care, crisis resolution treatment is often more appropriate for people with more severe depression who are at significant risk of suicide, harm to self or to others, self-neglect or complications in response to their treatment, leading to better outcomes and reduced need for more costly inpatient care.

The committee took into account the high costs associated with inpatient care, and decided to recommend inpatient treatment only for people with more severe depression who cannot be adequately supported by a crisis resolution and home treatment team.

Considering the benefits and costs of crisis resolution and home treatment teams (CRHT teams) relative to other care settings, the committee expressed the opinion that CRHT comprises an effective and likely cost-effective model of care for people with depression who would benefit from early discharge from hospital after a period of inpatient care.

The committee took into account the cost effectiveness of psychological treatments in the acute treatment of people with depression based on the results of the economic analysis undertaken for this guideline (Evidence report B: Treatment of a new episode of depression), and expressed the view that the full range of such treatments should also be available in inpatient settings, to allow provision of clinically and cost-effective care in populations treated in such settings. The committee acknowledged the fact that increasing the intensity and duration of psychological interventions for people with depression in inpatient settings has resource implications, but expressed the view that the benefits of more intensive treatment in this group would outweigh the additional intervention costs. Moreover, if improved outcomes result in earlier discharge, then cost-savings may outweigh the intervention costs of more intensive psychological treatment.