Table 3 Effect size estimates of the outcomes (total and per region), test of homogeneity (I2) and GRADE quality of the evidence (below the outcome is the total number of studies mentioning the outcome)
Outcome | Effect size estimates | I2 | GRADE quality of the evidence | |||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
Included in meta-analysis | For subgroup analysis | |||||||||||
East Asia and Pacific | Europe and Central Asia | Latin America and the Caribbean | Middle East and North Africa | North America | South Asia | Sub-Saharan Africa | Multi-country studies | |||||
Anxietya (N = 518) | n | 350 | 106 | 83 | 23 | 44 | 38 | 32 | 13 | 11 | 1.00 | Very low—346 cross-sectional studies, 3 cohorts and 1 case–control, 57% medium quality/risk of bias, 43% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 39% | 31% | 40% | 44% | 48% | 37% | 45% | 49% | 32% | |||
95CI | [37; 41] | [27; 34] | [36; 45] | [36; 52] | [41; 54] | [31; 43] | [38; 52] | [35; 62] | [17; 47] | |||
Depression (N = 503) | Before 2020 | |||||||||||
n | 3 | – | – | – | – | – | – | – | – | 0.98 | Very low—3 cross-sectional studies, 1 medium quality/risk of bias, 2 high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect | |
P | 21% | – | – | – | – | – | – | – | – | |||
95CI | [5; 37] | – | – | – | – | – | – | – | – | |||
From 2020 onwards | ||||||||||||
n | 370 | 108 | 90 | 22 | 48 | 45 | 34 | 13 | 10 | 1.00 | Very low—367 cross-sectional studies, 2 cohorts and 1 case–control, 55% medium quality/risk of bias, 45% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect | |
P | 35% | 32% | 35% | 34% | 44% | 31% | 39% | 41% | 20% | |||
95CI | [33; 37] | [29; 35] | [31; 39] | [26; 42] | [39; 50] | [26; 36] | [32; 45] | [31; 51] | [11; 29] | |||
Stressb (N = 486) | Before 2020 | |||||||||||
n | 2 | – | – | – | – | – | – | – | – | 1.00 | Very low—2 cross-sectional studies, 1 medium quality/risk of bias, 1 high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect | |
P | 40% | – | – | – | – | – | – | – | – | |||
95CI | [0; 79] | – | – | – | – | – | – | – | – | |||
From 2020 onwards | ||||||||||||
n | 158 | 41 | 38 | 9 | 25 | 11 | 23 | 7 | 4 | 1.00 | Very low—157 cross-sectional studies and 1 case–control, 63% medium quality/risk of bias, 37% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect | |
P | 44% | 39% | 46% | 34% | 50% | 38% | 48% | 51% | 35% | |||
95CI | [40; 48] | [32; 46] | [38; 54] | [22; 47] | [41; 60] | [25; 50] | [37; 59] | [39; 63] | [3; 67] | |||
Burnouta (N = 235) | n | 94 | 24 | 23 | 8 | 12 | 20 | 1 | 4 | 2 | 1.00 | Very low—90 cross-sectional studies and 4 cohort, 52% medium quality/risk of bias, 48% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 46% | 52% | 46% | 34% | 55% | 42% | – | 41% | 46% | |||
95CI | [42; 51] | [43; 61] | [38; 55] | [18; 49] | [43; 66] | [32; 51] | – | [16; 67] | [18; 75] | |||
PTSDa (N = 84) | n | 53 | 12 | 23 | 4 | 2 | 8 | 1 | 2 | 1 | 1.00 | Very low—52 cross-sectional, 1 cohort, 89% medium quality/risk of bias, 43% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 26% | 26% | 25% | 22% | 44% | 24% | – | 56% | – | |||
95CI | [22; 31] | [16; 36] | [18; 31] | [6; 38] | [16; 72] | [15; 32] | – | [52; 59] | – | |||
Suicidal ideationa (N = 18) | n | 15 | 4 | 7 | 1 | – | 3 | – | – | – | 0.98 | Very low—12 cross-sectional, 60% medium quality/risk of bias, 40% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 7% | 9% | 5% | – | – | 8% | – | – | – | |||
95CI | [5; 8] | [6; 12] | [3; 7] | – | – | [8; 9] | – | – | – | |||
Headachesa,c (N = 12) | n | 12 | 2 | 2 | 1 | – | 1 | 5 | – | 1 | 0.99 | Very low—12 cross-sectional studies, 92% medium quality/risk of bias, 8% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 53% | 43% | 36% | – | – | – | 53% | – | – | |||
95CI | [38; 67] | NA | [23; 49] | – | – | – | [39; 77] | – | – | |||
Sleep disorders (N = 90) | n | 54 | 13 | 9 | 3 | 12 | 6 | 4 | 5 | 2 | 0.99 | Very low—54 cross-sectional studies, 52% medium quality/risk of bias, 48% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 36% | 40% | 31% | 33% | 38% | 50% | 34% | 24% | 32% | |||
95CI | [31; 41] | [30; 50] | [21; 41] | NA | [29; 46] | [33; 67] | [17; 51] | [12; 37] | [12; 52] | |||
Skin-related morbidity (N = 19) | n | 14 | 4 | 6 | 1 | 1 | 1 | 1 | – | – | 1.00 | Very low—14 cross-sectional studies, 86% medium quality/risk of bias, 14% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 51% | 48% | 64% | – | – | – | – | – | – | |||
95CI | [39; 64] | [43; 53] | [43; 84] | – | – | – | – | – | – | |||
Violence (N = 32) | n | 11 | 1 | 2 | 3 | 2 | – | 2 | – | 1 | 1.00 | Very low—11 cross-sectional studies, 64% medium quality/risk of bias, 36% high quality/low risk of bias, very serious inconsistency, very serious imprecision, unlikely publication bias, no large effect, no evidence of a dose–response gradient, all plausible confounding would suggest a spurious effect |
P | 48% | – | 30% | 58% | 81% | – | 19 | – | – | |||
95CI | [32; 64] | – | [28; 32] | [31; 84] | NA | – | [17; 21] | – | – | |||