Table 3.
Overall meta-analysis of outcomes
| Employment rate | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Comparison | Follow-up (FU) subgroup | N studiesa | % employed at FUb | Effect size of outcomec | Heterogeneity | |||||||
| IPS | Control | ORd | 95% CI | p | I2 | 95% CI | p | |||||
| n (%) | N | n (%) | N | |||||||||
| IPS vs. active control condition | All studies | 20 | 949 (50.1%) | 1893 | 496 (26.7%) | 1859 | 3.15 [S] | 2.74–3.62 | <0.01 | 79% | 71–84% | <0.01 |
| ⩽ 12 month FU | 10 | 353 (38.3%) | 921 | 182 (20.8%) | 876 | 2.50 [S] | 2.07–3.02 | <0.01 | 68% | 49–80% | <0.01 | |
| > 12 month FU | 15 | 895 (54.2%) | 1652 | 518 (31.9%) | 1625 | 2.88 [S] | 2.48–3.36 | <0.01 | 81% | 72–87% | <0.01 | |
| European studies (including UK) | 7 | 345 (42.2%) | 817 | 199 (25.5%) | 781 | 2.27 [S] | 1.84–2.79 | <0.01 | 62% | 32–79% | <0.01 | |
| Non-European studies | 13 | 604 (56.1%) | 1076 | 297 (27.6%) | 1077 | 4.09 [M] | 3.40–4.92 | <0.01 | 79% | 68–86% | <0.01 | |
| IPS vs. passive control condition | All studies | 12 | 846 (48.3%) | 1753 | 525 (29.4%) | 1788 | 2.26 [S] | 1.97–2.68 | <0.01 | 61% | 43–74% | <0.01 |
| ⩽ 12 month FU | 7 | 144 (42.5%) | 339 | 80 (23.6%) | 338 | 2.12 [S] | 1.52–2.96 | <0.01 | 65% | 37–81% | <0.01 | |
| > 12 month FU | 6 | 719 (50.8%) | 1414 | 454 (31.3%) | 1450 | 2.30 [S] | 1.97–2.68 | <0.01 | 55% | 17–76% | 0.05 | |
| European studies (including UK) | 3 | 112 (30.3%) | 370 | 78 (21.3%) | 367 | 1.64 [S] | 1.18–2.27 | <0.01 | 34% | 0–74% | 0.22 | |
| Non-European studies | 9 | 734 (53.1%) | 1383 | 447 (31.5%) | 1421 | 2.44 [S] | 2.09–2.85 | <0.01 | 62% | 38–76% | <0.01 | |
| Overall outcomes | 31 | 1745 (48.8%) | 3578 | 1013 (28.3%) | 3578 | 2.62 [S] | 2.37–2.89 | <0.01 | 74% | 67–80% | <0.01 | |
| Test for subgroup differences | Active vs passive control condition | χ2 = 10.77; df = 1; p < 0.01 | ||||||||||
| ⩽ 12 month vs. > 12 month follow-up | χ2 = 0.29; df = 1; p = 0.59 | |||||||||||
| European vs. non-European studies | χ2 = 10.54; df = 1; p < 0.01 | |||||||||||
| Job duration | ||||||||||||
| Comparison | Follow-up subgroup | M (s.d.) at FUb | Effect size of outcomec | Heterogeneity | ||||||||
| IPS | Control | |||||||||||
| N studiesa | M (s.d.) | N | M (s.d.) | N | dd | 95% CI | p | I2 | 95% CI | p | ||
| IPS vs. active control condition | All studies | 17 | 24.5 (11.4) | 1571 | 10.9 (10.0) | 1565 | 0.47 [S] | 0.33–0.61 | <0.01 | 81% | 73–87% | <0.01 |
| ⩽ 12 month follow-up | 7 | 27.1 (13.6) | 674 | 15.2 (13.9) | 666 | 0.42 [S] | 0.16–0.68 | <0.01 | 76% | 51–88% | <0.01 | |
| > 12 month follow-up | 10 | 22.7 (9.9) | 897 | 7.9 (5.1) | 899 | 0.47 [S] | 0.30–0.63 | <0.01 | 82% | 73–89% | <0.01 | |
| European studies | 7 | 21.0 (12.2) | 688 | 10.2 (8.0) | 694 | 0.40 [S] | 0.20–0.61 | <0.01 | 83% | 68–91% | <0.01 | |
| Non-European studies | 10 | 27.8 (10.8) | 832 | 12.4 (11.5) | 828 | 0.52 [M] | 0.33–0.71 | <0.01 | 77% | 62–86% | <0.01 | |
| IPS vs. passive control condition | All studies | 7 | 27.1 (12.7) | 1354 | 17.1 (13.7) | 1393 | 0.31 [S] | 0.12–0.49 | <0.01 | 69% | 42–83% | <0.01 |
| ⩽ 12 month follow-up | 4 | 26.4 (17.9) | 155 | 18.0 (16.9) | 152 | 0.23 [S] | 0.07–0.40 | <0.01 | 0% | 0–78% | 0.63 | |
| > 12 month follow-up | 3 | 28.1 (1.9) | 1199 | 16.0 (11.3) | 1241 | 0.36 [S] | 0.02–0.71 | <0.05 | 87% | 56–96% | <0.01 | |
| European studies | 1 | 29.6 (19) | 127 | 27.7 (19.5) | 121 | −0.02 [N] | −0.27 to 0.23 | 0.88 | NA | NA | NA | |
| Non-European studies | 6 | 26.7 (13.9) | 1227 | 15.4 (14.1) | 1272 | 0.37 [S] | 0.20–0.54 | <0.01 | 54% | 15–75% | 0.06 | |
| Overall outcomes | 23 | 25.2 (11.8) | 2857 | 13.1 (11.4) | 2889 | 0.41 [S] | 0.30–0.52 | <0.01 | 77% | 69–83% | <0.01 | |
| Test for subgroup differences | Active vs passive control condition | χ2 = 1.98; df = 1; p = 0.16 | ||||||||||
| ⩽ 12 month vs. > 12 month follow-up | χ2 = 0.27; df = 1; p = 0.60 | |||||||||||
| European vs. non-European studies | χ2 = 0.65; df = 1; p = 0.42 | |||||||||||
| Wages | ||||||||||||
| Comparison | Follow-up subgroup | M (s.d.) at FUb | Effect size of outcomec | Heterogeneity | ||||||||
| IPS | Control | |||||||||||
| N studiesa | M (s.d.) | N | M (s.d.) | N | dd | 95% CI | p | I2 | 95% CI | p | ||
| IPS vs. active control condition | All studies | 10 | 286.6 (314.7) | 994 | 148.3 (196.6) | 979 | 0.39 [S] | 0.20–0.58 | <0.01 | 76% | 61–85% | <0.01 |
| ⩽ 12 month follow-up | 2 | 310.1 (328.5) | 60 | 77.9 (94.6) | 62 | 0.63 [M] | 0.26–0.99 | <0.01 | 0% | NA | 0.92 | |
| > 12 month follow-up | 8 | 277.8 (332.2) | 934 | 174.7 (223.0) | 917 | 0.35 [S] | 0.15–0.56 | <0.01 | 80% | 64–89% | <0.01 | |
| European studies (including UK) | 2 | 558.9 (491.9) | 289 | 353.1 (247.2) | 285 | 0.17 [N] | −0.07 to 0.41 | 0.17 | 52% | NA | 0.15 | |
| Non-European studies | 8 | 226.1 (266.3) | 705 | 102.8 (166.9) | 694 | 0.46 [S] | 0.22–0.69 | <0.01 | 77% | 59–87% | <0.01 | |
| IPS vs. passive control condition | All studies | 6 | 497.7 (413.9) | 1222 | 414.8 (401.1) | 1259 | 0.28 [S] | 0.14–0.42 | <0.01 | 23% | 0–43% | 0.26 |
| ⩽ 12 month follow-up | 4 | 640.4 (451.7) | 150 | 550.2 (427.3) | 141 | 0.15 [N] | –0.07 to 0.36 | 0.19 | 0% | 0–79% | 0.61 | |
| > 12 month follow-up | 2 | 252.1 (252.4) | 1072 | 144.1 (191.8) | 1120 | 0.39 [S] | 0.11–0.67 | <0.01 | 64% | NA | 0.09 | |
| European studies (including UK) | 0 | X | X | X | X | X | X | X | X | X | X | |
| Non-European studies | 6 | 497.7 (413.9) | 1222 | 414.8 (401.1) | 1261 | 0.28 [S] | 0.14–0.42 | <0.01 | 23% | 0–43% | 0.26 | |
| Overall outcomes | 15 | 379.1 (358.9) | 2148 | 257.0 (306.4) | 2172 | 0.31 [S] | 0.19–0.44 | <0.01 | 65% | 51–76% | 0.01 | |
| Test for subgroup differences | Active vs passive control condition | χ2 = 0.84; df = 1; p = 0.36 | ||||||||||
| ⩽ 12 month vs. > 12 month follow-up | χ2 = 0.14; df = 1; p = 0.70 | |||||||||||
| European vs. non-European studies | χ2 = 1.58; df = 1; p = 0.21 | |||||||||||
Some studies have used multiple follow-up assessments or have multiple treatment arms. Therefore, some studies are included in the analysis of both follow-up subgroups and one study compared IPS with both an active and passive control group. Therefore, the total amount of studies and sample sizes analysed in each comparison is sometimes lower than the sum of studies analysed in both follow-up subgroups.
Summary statistics for each of the three employment outcomes are assessed as follows: Employment rate: number and percentage of people in competitive employment at the follow-up assessment; Job duration: percentage of time within the study period that participants are employed; Wages: monthly salary in euros during the study period.
d > 0 and OR > 1 indicates outcomes are beneficial for IPS compared to the control group; d < 0 and OR < 1 indicates outcomes are beneficial for the control group compared to IPS.
Magnitude of effect (Chinn, 2000): Not clinically relevant [N]: d > −0.2 – <0.2; OR > 0.67 – <1.5; Small effect [S]: d ⩽ −0.20 and >−0.50 – ⩾0.20 and <0.50; OR ⩽ 0.67 and >0.29 – ⩾1.5 and <3.5; Medium effect [M]: d ⩽ −0.50 and >−0.80 – ⩾0.50 and <0.80; OR ⩽ 0.29 and >0.20 – ⩾3.5 and <5; Large effect [L]: d < −0.80 – >0.80; OR < 0.20 – >5.