Table 2.

Detailed data extraction for IPS economic studies (quality scores above 50% only).

			Economic analysis		Outcomes and key findings
References and country	Setting and study population (age, sex, and size)	Intervention details (study design, description of intervention, comparator, and type of intervention)	Perspective Price year Currency	Type of economic analysis	Effect on mental health	Work-related outcome	Economic or financial outcome	Quality score
Christensen et al. [18] Denmark	482 people with severe mental illness: schizophrenia, schizotypal, or delusional disorders (F20–F29) 75.7 or 77.8% or bipolar disorder (F31) 13.2 or 10.5%, or recurrent depression (F33) 11.1 or 11.7% according to ICD 10 in both IPS and SAU groups. Mean age 33 in both IPS and SAU groups. 61.3 and 60.3% male in IPS and SAU groups, respectively. Data collected between 2012 and 2018	RCT comparing IPS (n = 243) compared with service as usual (SAU) (n = 239). Duration 1.5 years	Societal 2016 Euros	CEA, CUA	QALY gains were nonsignificantly higher in the intervention group than the control group (0.0329 vs. 0.0074, p = 0.2960). Mental health hospital care costs were lower in the IPS group €14,549 versus €18,279. (p = 0.0901). No significant difference in somatic hospital, primary health care or prescription costs	IPS participants earned an average of €1,792 more than the control group. Productivity gains in the intervention group −€7214 versus −€5422, p = 0.205). Labour market intervention costs were significantly lower in the IPS group. €403 versus €3,395 (p < 0.0001)	IPS was less costly, with nonsignificantly improved QALY gains compared to SAU. Overall costs, including productivity losses were significantly lower by a mean of €9,543 in the IPS group (p = 0.001). If there was a societal threshold of €35,000 for willingness-to-pay for a QALY, there is a probability of 95.6% of IPS being cost-effective compared to SAU. IPS also dominated with significantly lower costs and nonsignificantly higher hours spent in work or education	95%
Deloitte [19] UK	126 adults with unspecified mental illness using IPS services in Glasgow, Scotland. Age unspecified	Modelling study drawing on published literature, IPS data and expert opinion. The intervention was IPS and the comparator traditional vocational schemes (TVS). Duration: 1 year	Health, Public purse and service users combined 2016 UK pounds	CBA	40–60% reductions in Community Psychiatric Nurse (CPN) appointments and three less psychiatric appointments after 1 year after having secured employment. Total costs avoided of £96,710. NHS costs avoided per service user with more than 3 months competitive employment £1,981	Total service user benefit due to increased earnings £84,020	The annual cost of the IPS service was £116,000. Benefits of IPS: increased earnings £84,020, health care costs avoided £96,710, welfare benefits avoided £59,210. Net benefits £123,940. Additional net costs of TVS avoided £57,030 Total net benefit £180,970 Net benefits per user if employment sustained for less than or more than 3 months: £2,300 and £7,870	70%
Hellström et al. [20] Denmark	326 people aged 18–60 with an anxiety or affective disorder recruited from mental health centres and private psychiatrists in Copenhagen. Gender not reported. Participants should not have been in contact with mental health services for more than 3 years.	The intervention was IPS intervention modified for people with mood and anxiety disorders (IPS-MA) (162 people) compared to services as usual (SAU) (164 people). These could be social services (e.g., group therapy or psycho-social support interventions) or labour market services. Duration of study 2 years but only 12 month outcomes used in economic analysis	Societal 2016 Euros	CUA, CEA	When imputed cases data included mean QALYs gained were 0.056 and −0.17 in the IPS-MA and SAU groups (p < 0.05). The difference was not significant for complete cases only. 1 year mental health service use in the IPS-MA group was €5,489 compared to €8,161 in the SAU group (p = 0.078) Overall health care costs were not significantly different between the two groups	Mean wage earnings in the IPS-MA group were significantly lower, €5,034 versus €8,410 (p = 0.017) Labour market service costs were significantly lower in the IPS-MA group, €1,329 versus €5,591 (p = 0.009)	IPS-MA had a mean cost per person per year of €1,183. Overall, there was no significant difference in costs between the two groups, although costs were lower in the IPS-MA group €5,485 versus €7,706. p = 0.423. There was between an 83 and 95% chance at €30,000 per QALY gained of IPS-MA being cost-effective versus SAU. If cost per hour of worked gained was used instead the intervention would not be cost-effective with significantly lower levels of hours worked than SAU	78%
Heslin et al. [21] UK	219 individuals, with severe mental illness recruited between November 2004 and September 2006. Mean age 38, 66–69% male, 41 and 34% white, in IPS and control groups. 69 and 76% had a psychotic disorder. 31 and 24% had a mood disorder	RCT comparing IPS (n = 109) with treatment as usual/local vocational services (TAU) (n = 110). Duration: 2 years	Health care 2006/2007 UK pounds	CEA	There were no differences between intervention and control groups at follow-up on any clinical measures. Over 24 months, health care costs in intervention group were lower than controls (£9,571 versus £11,932), p-value not reported	Intervention had a significantly higher proportion in competitive employment than control group (22% vs. 11%, p = 0.041)	With lower costs and higher outcomes IPS was dominant. In probabilistic sensitivity analysis there was a 90% chance of IPS being the most cost-effective option	56%
Hoffmann et al. [22] Switzerland	100 individuals aged 18–64 with severe mental illness including schizophrenia spectrum, affective disorder), male (65%). Mean age 33.5 and 34.1 in intervention and support groups	RCT comparing IPS (n = 46) with traditional vocational rehabilitation (TVR) (n = 54). Duration: 5 years	Society Price year not stated Swiss Francs (CHF)	ROI	Intervention group had significantly less hospitalisation (21% vs. 46.7% p = 0.015), fewer psychiatric hospital admissions (0.4 vs. 1.1 p = 0.026) and spent fewer days in the hospital (38.6 vs. 96.8 p = 0.027)	Intervention group had higher rates of competitive work than traditional vocational rehabilitation (65% vs. 33%, p = 0.002), worked more hours per year (689 vs. 392, p = 0.023), earned more wages (CHF 11,826 vs. CHF 6,885, p = 0.004), had longer job tenures (104.8 vs. 35.5, p < 0.001)	Earning per client over 5 years CHF 66,977 versus CHF 37,093 in IPS and TVR. Mental health treatment costs per client CHF 25,484 versus 40,093. Vocational programme costs CHF 80,917 and CHF 43,701 Social ROI: 132.2%	52%
Holmås et al. [23] Norway	327 individuals (mean age = 35) with moderate (depression and anxiety disorders) to severe mental illness (psychotic or bipolar disorder with or without comorbid substance abuse/dependency), women (50%), from regional primary and secondary mental health care settings for 43 months	Original study based on RCT comparing IPS (n = 184) with treatment as usual (n = 143) 327 people before and after the IPS intervention. Duration: 3.6 years	Societal 2016 Norwegian kroner	CBA	Not reported	During 43 months, the intervention group had 8.8% higher rates of regular employment than in the control group, and 5% higher in regular employment with a half-time job or more (16.5% vs. 10.7%)	Net social benefit: NOK 217,000 (gain in productivity = 65,000 + cost-savings from traditional VR programme costs = 211,000 − programme cost(100,000) + cost-savings from excess burden of taxes 41,000, so 65,000 + 211,000000 − 100,000 + 41,000 = 217,000	65%
Howard et al. [24] UK	150 individuals (mean age = 38) with psychotic or chronic affective disorder in South London	RCT comparing IPS (n = 109) with local traditional vocational service, TAU (n = 110). Duration: 1 year	Health care	CEA	Psychiatric inpatient costs were lower in the intervention group (£719 vs. £2241), also lower costs for community mental health nurse costs than the control group (£49 versus £65)	There were no significant differences between the treatment as usual and intervention groups in obtaining competitive employment (13% in the intervention group and 7% in controls; p = 0.15), nor in secondary outcomes	Total costs were £2176 significantly higher in the control group (bootstrapped 95% CI £445–£4168). No significant differences in outcomes During the 1-year follow-up, two-thirds of the intervention group received input from employment workers at mean cost of £296. There were no substantial differences in the number of people using other services. However, control group participants who were admitted spent substantially more days in hospital than inpatients in the intervention group. This resulted in a difference in inpatient costs of £1522.	56%
Khalifa et al. [25] UK	18 individuals (mean age = 39.2) with schizophrenia, depression, personality disorder, with offending histories in community forensic settings over 12 months, male (88.9%)	RCT comparing IPS (n = 11) with Treatment as usual (n = 7). Duration: 1 year	Health care 2016 UK pounds	CEA	Brief Psychiatric Rating Scale scores were higher in the intervention group (34 vs. 25.5, p-value not reported), SF-12 scores in mental health were higher (53.1 vs. 43.8), EQ-5D (64.3 vs. 70), p-values not reported. Health care costs not reported separately from cost of IPS	Intervention group had higher rates in open employment at 12 months (9.1 and 0%) than TAU	Mean baseline costs £29,444 in IPS group versus £1,898 in TAU group. IPS less costly than TAU at 12 month follow up £1,799 vs. 1,940, significance not tested. Sample too small to draw conclusions on cost-effectiveness	70%
Knapp et al. [17] Netherlands, UK, Italy, Bulgaria, Switzerland, and Germany	312 individuals with SMI (schizophrenia and schizophrenia-like disorders, bipolar disorder, or depression with psychotic features, using ICD-10 criteria for 18 months in six European cities: Groningen (Netherlands), London (UK), Rimini (Italy), Sofia (Bulgaria), Ulm-Gunzburg (Germany), and Zurich (Switzerland)	RCT comparing IPS (n = 156) with standard vocational services (n = 156). Duration: 1.5 years	Health and social care Societal 2003 UK pounds	CBA, CEA	Readmission rates were lower in IPS than the control group (13% vs. 20%). Mean health care costs in the IPS group at first 6-month follow up were significantly lower than for control £4,688 versus £6,926 (Mean difference £2,720 95% CI −£4,624, −£813. Costs were lower in the following two 6-month follow ups but the difference was not significant	Over 18 months, IPS had higher rates of being at least 1 day in employment (55% vs. 28%) than those in vocational services	Mean costs for IPS across all sites were £18,877 versus £25,455 in controls (Mean difference £7,880 95% CI −£12,249, −£3151. Costs were significantly lower in three of the six sites: London, Ulm, and Zurich and the intervention was dominant, with lower costs and better outcomes in all areas except Groningen. In Groningen the additional cost per additional 1% of people working at least 1 day was £30 and additional cost per additional day worked was £10 Boostrapped mean net monetary benefits comparing the costs of intervention with value of employment achieved for both IPS and control were favourable at £17,005 in the IPS group	91%
Mavranezouli et al. [26] UK	Model drew on previous study of people with formal diagnosis of autism and IQ ≥70. Mean IQ score 98.8 (Wechsler Adult Intelligence Scale)	Markov modelling study comparing IPS versus standard care (day services). Duration: 8 years	Health care 2012 UK pounds	CEA, CUA	Mean QALYs gained over 8 years were 5.42 in the IPS group and 5.31 in the control group Secondary analysis including other health and social care costs, including mental health-care costs, other primary and secondary care costs and local authority costs revealed mean health care costs of £16,005 and £16,663 in the IPS and control groups	Over 8 years mean weeks in employment were 136 and 102 for the IPS and control groups	For the primary analysis, just including the costs of IPS and day care, the cost per QALY gained was £5,600 and cost per extra week of employment was £18. In probabilistic sensitivity analysis there was a 67 and 75.2% chance of being cost-effective at £20,000 or £30,000 per QALY gained. In secondary analysis including health and social care costs, the IPS intervention was dominant with lower costs and better outcomes. In probabilistic sensitivity analysis, there was an 80% chance of being cost-effective at £30,000 per QALY gained	92%
Mental Health Reform [27] Ireland	95 adults with severe and enduring mental health problems, not in paid employment who received IPS services from 2015 to 2017	A pilot IPS study, control group not reported. Duration: 2 years	Provider Price year not reported Euros	CCA	Not reported	In the project, 36% had at least one job placement. The average number of hours worked per week by successful applicants was 21 h, the average weekly wage for successful participants was €230. The combined staff and project costs totaled €276,326, with a cost per participant of €2,909	Cost per job outcome was €8,374. If start-up costs and project management excluded the cost per participant was €2,451 and the cost per job outcome was €7,057	61%
Parlettaa and Waghorn [28] Australia	175 individuals aged 15–64 (47% male) with schizophrenia or bipolar affective disorder, major depression, anxiety disorders, Posttraumatic Stress Disorder, personality disorder and substance abuse disorder	Observational cohort study comparing IPS (n = 68) with pre-IPS cohort (n = 107). Duration: 1.5 years	Provider and public purse	CBA	Not reported	Intervention group had significantly higher rates of job starts than pre-IPS services (67.6% vs. 56.1%) (Significance not reported).	Net revenues were higher in IPS than pre-IPS groups. The IPS enhanced service achieved higher gross revenue per participant ($9,062) than pre-IPS services ($7,514). The IPS enhanced programme generated more net revenue (gross revenue less direct costs) per participant compared to pre-IPS services ($6,929 versus $6,161), Cost per 26 week employment outcome achieved to government was $38,958 compared to $42,813 for pre IPS group. For higher severity group cost per 26 week employment outcome in IPS enhanced group was $48,693 vs. $167,199 in pre IPS group	52%
Perkins et al. [29] UK	Hypothetical 135,000 new IPS participants each year with unspecified mental health problems	Modelling study comparing IPS with Traditional service or no intervention. Duration: 2 years	Public Purse Price year not stated UK pounds	ROI	Not reported	Unpublished survey data for study involving employment workers across private, public and voluntary sectors was used to assume all clients would receive support for 6 months, with 35% continuing for 1 year and 25% for 2 years	Total cost of the programme £180 million per annum. 27,000 jobs would need to be created for the service to break even. This would mean a cost per job before fiscal benefits of £6,600; if 47,000 jobs were created the return on investment would be 1.72	56%
Rosenheck et al. [30] USA	404 individuals aged 15–40 with First Episode Psychosis, less than 6 months on lifetime antipsychotics in clinical treatment clinics. Demographic information not provided	RCT comparing IPS: Navigate (NAV), a comprehensive, multidisciplinary, team-based treatment approach for first episode psychosis, including IPS (n = 223) with community care (CC) (n = 181). Duration: 2 years	Health care US dollars	CBA, CUA	The NAV group had significantly greater improvement in PANSS total scores and improved significantly more on as a one standard deviation change on the Quality of Life Scale (QLS-SD) (p < 0.02). There was no significant difference in overall costs between the two groups. However, the intervention group had higher outpatient mental health ($1870 vs. 1379, p = 0.05), antipsychotic medication costs ($1739 vs. 1060. p = 0.01) Costs for all mental health and medical surgical inpatient care were lower in the intervention group but no significant difference ($3,694 vs. $3,780, p = 0.91)	Not reported	The incremental cost-effectiveness ratio was $12,081/QLS-SD, with a 94% probability that NAV was more cost-effective than CC at $40,000/QLS-SD. When converted to monetised quality-adjusted life years, NAV benefits exceeded costs, especially using future generic drug prices	87%
Saha et al. [31] Sweden	55 individuals with unemployed, depressive episodes, recurrent depression or bipolar disorder. Demographic information not reported	RCT comparing IES (individual enabling and support) (an IPS intervention) vs. TVR (traditional vocational rehabilitation) Duration: 1 year	Societal 2014 Euros	CEA, CUA	Intervention group was more effective using Manchester Short Assessment of Quality of Life (MANSA), but not EQ-5D. There were no significant differences in QALY improvement between groups. But quality of life measured by the MANSA scale significantly improved over the study period in IE. Health care costs were not included	The value of productivity gains was higher in the intervention group (€6059) than the control group (€111), p-value not reported	The cost of IES was €7247 lower per person per year, compared to TVR. The total cost for IES were €528 per person per year compared to €7775 for TVR. Intervention was dominant with no change in quality of life but lower costs	74%
Sambo [32] Canada	109 individuals aged 18–30 (mean age = 23) with Schizophrenia, psychosis, bipolar disorder, male (45%)	RCT comparing IPS: IPS + Early intervention for psychosis (TAU) (n = 56) with TAU (n = 53). Duration: 1 year	Health care Public payer 2016 Canadian dollars	CCA, CUA	Although the sample was small due to data collection issues, the EQ-5D-5L index scores were consistently higher for those in the TAU group compared with those in the IPS+ group. Scores on the QLS were consistently higher for the IPS+ TAU group compared with those in the TAU group. Overall health care costs, including primary care as well as specialist mental health care were nonsignificantly higher in the TAU group $3,884 vs. $3,656	At 12 months, proportion employed (IPS+ TAU vs. TAU):60% versus 50% but not significant. Nonsignificantly increased working days in intervention group: mean 8.38 more days, 57.24 vs. 48.86 days.	Total costs per patient in the IPS + TAU group were lower than those in TAU (mean difference $228, p = 0.823, 95% CI, $-2,261 to $1,806). Also improvements in employment outcomes in IPS group and quality of life but not significant	91%
Shi [33] Canada	149 individuals aged 18–64 (mean age = 40 or 41, female 37.3% and 39.2% in IPS or control groups 40.6) with severe mental illness: psychosis, bipolar disorder, major depression; and using outpatient psychiatric hospital services between 2001–2004	RCT comparing IPS (n = 75) with usual vocational services, including sheltered workshops, creative workshops, a consumer-run boutique, horticultural programs, job-finding-skills training and psychosocial intervention (n = 74) Duration of study: 1.5 years	Health Care, Public Purse, and Societal 2010–11 Canadian dollars	CEA	No significant differences in mental health care costs between the two groups: Inpatient costs ($1,421 versus $6,443, p = 0.2258), other mental health service ($639 versus $1,286, p = 0.9032), out-of-pocket costs of psychologist services ($11 vs. $10, p = 0.9921)	Over 12 months, significantly longer hours in competitive employment in IPS than the control group. Mean 126 h versus 72 h, p = 0.0004), higher wages in competitive employment ($935 vs. $514, p = 0.004)	Overall costs were lower for IPS compared to controls from all three perspectives: health and social care perspective: $25,709 (IPS) vs. $26,683 (UC) (p = 0.011). Public Purse: $32,984 versus $33,945, Society: $27,014 versus $27,678. The IPS programme was less expensive while improving outcomes, which means IPS dominates usual services	87%
Stant et al. [34] and van Busschbach et al. [35] Netherlands	151 with severe and long-term mental disorders who want to work. In IPS and control groups respectively: 55 and 64% psychosis, 17 and 10% mood disorders, 22 and 23% personality disorders, mean age 34.1 and 35.6, male 73 and 75%	RCT comparing IPS (n = 71) versus regular vocational rehabilitation (n = 80) Duration of study: 2.5 years	Health care Societal 2008 Euros	CEA, CUA	There were no differences between groups in quality of life at any time point measured using the MANSA – Manchester Short Assessment of Quality of Life. Mental health and general health care cost were higher but significance not reported. Overall mean costs including health, net of productivity gains were €57,285 and €43,819 in the IPS and control groups	After 2.5 years significantly more people in the IPS group were in regular paid work. Paid work during the study was significantly higher in the IPS group (44% vs. 25%) p < 0.05. More hours were also worked in the IPS group. Mean rehabilitation costs, including the cost of intervention were greater in the IPS group €1,705 versus €1,176	IPS has higher costs and better outcomes than regular vocational rehabilitation. The cost per additional 1 percent of individuals in paid work was €1,084. However, averted social welfare costs due to increased work participation are not included in the cost-effectiveness ratio. Here is an 80% probability of being cost-effective if society is willing to pay €2,000 per additional 1% in employment. The incremental cost per additional 1 point on the MANSA scale is €76,359	83%
Stroupe et al. [36] USA	541 military veterans (mean age = 41.2) with PTSD, men (81.7%)	RCT comparing IPS (n = 271) with transitional work (TW) programmes (n = 270) Duration: 1.5 years	Health care Societal 2019 US dollars	ROI, CEA	Mental health costs were insignificantly higher for IPS than TW ($1687 vs. $1498, p = 0.75) The annual mean cost per person of outpatient care were $3970 higher for IPS compared to TW ($23,245 vs. $19,276, p = 0.004) Overall mean health care costs including costs of vocational rehabilitation were significantly higher in the IPS group $29,691 vs. $23,298	The average number of hours worked in competitive employment per person per year was significantly higher in the IPS than the TW group (632 h vs. 458 h, p = 0.002). The mean annual income from competitive employment was higher in the IPS than the TW group ($9,762 vs. $7,326, p = 0.02)	IPS is more costly and more effective. 95% probability of being cost-effective if willing to pay $81 per additional hour worked. The average total costs per person per year were similar between groups ($29,828 vs. $26,772, p = 0.17). The incremental cost-effectiveness was $28 per additional hour of competitive employment. The return on investment (excluding TW income) was 32.9% for IPS ($9762 mean income/$29,691 mean total costs) and 29.6% for TW ($7326 mean income/$24,781 mean total costs)	87%
Szplit [37] UK	45 individuals with moderate to severe mental health problems in collaboration with community mental health teams from April 2010 to March 2011	1 year observational study, with longer term impacts modelled. Duration: 5 years	Employment 2010/11 UK pounds	CBA, ROI	Not reported	For 1 year, 40% (18 out of 45) of participants secured permanent employment	For every £1 invested with IPS there would be a return ranging between £5.01 and £6.77 in social added value The total present value (PV) of IPS for 2010/2011 is valued at £526,885. The total investment is £77,822 Total net present value less total investment figure (NPV) is £449,063	57%
van Stolk et al. [38] UK	People with depression, anxiety (common mental disorders) including employment, but also some people on sick leave	Modelling study comparing vocational support based on the Individual Placement and Support (IPS) model in IAPT or other suitable psychological therapy services. Duration: 1 year	Healthcare Public purse 2011 UK pounds	CBA	Assumes IPS service would support 120 clients per year. 35 would have reduced healthcare utilisation costs of £300 per year, including savings from fewer GP visits and limited use of secondary care	24.5 people would stop claiming job seekers allowance of £3,900	Positive cost benefit ratio = 1.41 from IPS; note this assumes that savings from reduced statutory sick leave so cost benefit ratio for long term unemployed alone not stated	70%
Washington State Institute for Public Policy [39] USA	Those with severe mental illness	Modelling study (Monte Carlo Simulation analysis for risk/uncertainty analysis) IPS versus traditional vocational rehabilitation. Duration: 50 years	Taxpayers, Participants	CBA	Net health care costs for psychiatric hospitalisation reduced by $8 per participant	Not reported. Net programme cost per participant: $849. Additional taxes from additional earnings to taxpayers $2,090 and to participants $4,910	Total positive benefits net of deadweight costs: $5,741. Benefit cost ratio = $7.7. Chance the program will produce benefits greater than costs 80%	74%
Whitworth [40] UK	Hypothetical 5,000 IPS programme starts over 30 months. Time limited to maximum 15 months. Assumed to have mental and physical health problems	Modelling study for IPS comparing alternative modified scenarios, the control group not reported. Duration: 10 years	Public purse Price year not reported UK pounds	CBA	Impacts on health care costs not included in analysis	Not detailed but costs averted include welfare benefits avoided, including council tax benefit and universal credit. Average annual earnings from employment assumed to be £11,800	ROI = 0.32 to 7.47, depending on models at 10 years. ROI at 5 years ranges from 0.19 to 4.53 depending on model scenario	70%