Table 2.
Risk factor | Effect size | Heterogeneity | Excluded studies | ||||
---|---|---|---|---|---|---|---|
nk | OR | 95% CI | I2 (%) | 95% CI | 95% PI | ||
Asthma | |||||||
Random effects model, unadjusted | 11 | 1.30 | [1.19–1.42]a | 89.22 | [83–93.00]a | 0.06–0.46 | – |
Basic outlier removal | 11 | 1.30 | [1.20–1.40]a | 89.22 | [83–93.00]a | 0.06–0.46 | – |
Influence analysis („leave-one-out”) | 10 | 1.27 | [1.20–1.40]a | 65.59 | [33–82.00]a | 1.08–1.48 | [45] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 10 | 1.27 | [1.20–1.40]a | 65.59 | [33–82.00]a | 1.08–1.48 | [45] |
Gaussian mixture model (GMM) clustering | 10 | 1.28 | [1.20–1.40]a | 89.66 | [83–94.00]a | 1.05–1.56 | [68] |
K-means clustering | 10 | 1.30 | [1.20–1.40]a | 90.11 | [84–94.00]a | 1.06–1.60 | [52] |
Autoimmune disorders | |||||||
Random effects model, unadjusted | 16 | 1.33 | [0.98–1.80] | 98.28 | [98–99.00]a | − 0.95–1.52 | – |
Basic outlier removal | 13 | 1.40 | [1.20–1.60]a | 91.11 | [87–94.00]a | − 0.14–0.81 | [40, 45, 77] |
Influence analysis („leave-one-out”) | 14 | 1.46 | [1.30–1.70]a | 92.55 | [89–95.00]a | 0.84–2.54 | [45, 77] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 9 | 1.49 | [1.20–1.90]a | 96.22 | [94–97.00]a | 0.71–3.09 | [49, 51, 52, 77, 89, 92] |
Gaussian mixture model (GMM) clustering | 9 | 1.49 | [1.20–1.90]a | 96.22 | [94–97.00]a | 0.71–3.09 | [49, 51, 52, 77, 89, 92] |
K-means clustering | 15 | 1.53 | [1.30–1.80]a | 95.35 | [94–97.00]a | 0.82–2.84 | [89] |
Cancer | |||||||
Random effects model, unadjusted | 20 | 2.42 | [1.91–3.07]a | 99.07 | [99–99.00]a | − 0.18–1.95 | – |
Basic outlier removal | 13 | 2.21 | [1.90–2.60]a | 95.93 | [94–97.00]a | 0.43–1.16 | [49, 86, 93, 116, 117] |
Influence analysis („leave-one-out”) | 18 | 2.24 | [1.80–2.80]a | 97.58 | [97–98.00]a | 0.90–5.53 | [49, 89] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 13 | 2.45 | [2.00–3.10]a | 97.26 | [96–98.00]a | 1.09–5.55 | [49, 52, 86, 89, 93, 103] |
Gaussian mixture model (GMM) clustering | 13 | 2.45 | [2.00–3.10]a | 97.26 | [96–98.00]a | 1.09–5.55 | [49, 52, 86, 89, 93, 103] |
K-means clustering | 19 | 2.25 | [1.80–2.80]a | 99.05 | [99–99.00]a | 0.94–5.37 | [103] |
Cardiovascular disorders | |||||||
Random effects model, unadjusted | 11 | 1.39 | [1.12–1.73]a | 99.33 | [99–99.00]a | − 0.42–1.08 | – |
Basic outlier removal | 8 | 1.30 | [1.00–1.60]a | 92.62 | [88–96.00]a | − 0.44–0.96 | [13, 45] |
Influence analysis („leave-one-out”) | 9 | 1.37 | [1.10–1.80]a | 98.62 | [98–99.00]a | 0.64–2.94 | [13, 45] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 7 | 1.26 | [0.92–1.70] | 99.39 | [99–100.00]a | 0.50–3.17 | [45, 103, 111] |
Gaussian mixture model (GMM) clustering | 7 | 1.26 | [0.92–1.70] | 99.39 | [99–100.00]a | 0.50–3.17 | [45, 103, 111] |
K-means clustering | 10 | 1.40 | [1.10–1.80]a | 99.40 | [99–100.00]a | 0.64–3.07 | [45] |
CHF | |||||||
Random effects model, unadjusted | 7 | 1.35 | [0.99–1.84] | 98.48 | [98–99.00]a | − 0.55–1.15 | – |
Basic outlier removal | 6 | 1.24 | [0.93–1.70] | 91.74 | [85–96.00]a | − 0.46–0.90 | [45] |
Influence analysis („leave-one-out”) | 6 | 1.24 | [0.93–1.70] | 91.74 | [85–96.00]a | 0.63–2.45 | [45] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 5 | 1.32 | [0.93–1.90] | 76.13 | [42–90.00]a | 0.64–2.73 | [83, 111] |
Gaussian mixture model (GMM) clustering | 5 | 1.32 | [0.93–1.90] | 76.13 | [42–90.00]a | 0.64–2.73 | [83, 111] |
K-means clustering | 5 | 1.26 | [0.85–1.90] | 93.35 | [87–96.00]a | 0.49–3.24 | [45, 111] |
COPD | |||||||
Random effects model, unadjusted | 10 | 1.55 | [1.04–2.31]a | 99.91 | [100–100.00]a | − 0.86–1.74 | – |
Basic outlier removal | 9 | 1.38 | [1.10–1.70]a | 99.08 | [99–99.00]a | − 0.28–0.92 | [75] |
Influence analysis („leave-one-out”) | 9 | 1.38 | [1.10–1.70]a | 99.08 | [99–99.00]a | 0.76–2.51 | [75] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 8 | 1.34 | [1.10–1.70]a | 99.19 | [99–99.00]a | 0.70–2.54 | [45, 85] |
Gaussian mixture model (GMM) clustering | 8 | 1.34 | [1.10–1.70]a | 99.19 | [99–99.00]a | 0.70–2.54 | [45, 85] |
K-means clustering | 9 | 1.38 | [1.10–1.70]a | 99.08 | [99–99.00]a | 0.76–2.51 | [85] |
Depression | |||||||
Random effects model, unadjusted | 9 | 1.27 | [1.08–1.49]a | 95.98 | [94–97.00]a | − 0.10–0.57 | – |
Basic outlier removal | 8 | 1.23 | [1.10–1.40]a | 96.18 | [94–97.00]a | − 0.08–0.50 | [106] |
Influence analysis („leave-one-out”) | 8 | 1.21 | [1.10–1.40]a | 84.52 | [71–92.00]a | 0.98–1.48 | [99] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 6 | 1.20 | [1.10–1.30]a | 82.43 | [63–92.00]a | 0.96–1.50 | [13, 51, 52] |
Gaussian mixture model (GMM) clustering | 6 | 1.20 | [1.10–1.30]a | 82.43 | [63–92.00]a | 0.96–1.50 | [13, 51, 52] |
K-means clustering | 8 | 1.23 | [1.10–1.40]a | 96.18 | [94–97.00]a | 0.93–1.64 | [13] |
Diabetes | |||||||
Random effects model, unadjusted | 17 | 1.26 | [1.03–1.54]a | 99.70 | [100–100.00]a | − 0.62–1.08 | – |
Basic outlier removal | 13 | 1.21 | [1.10–1.40]a | 96.33 | [95–97.00]a | − 0.22–0.60 | [45, 49, 73, 82] |
Influence analysis („leave-one-out”) | 14 | 1.26 | [1.10–1.40]a | 99.38 | [99–99.00]a | 0.76–2.09 | [49, 73, 82] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 15 | 1.24 | [1.00–1.50] | 99.74 | [100–100.00]a | 0.52–2.99 | [13, 45] |
Gaussian mixture model (GMM) clustering | 15 | 1.24 | [1.00–1.50] | 99.74 | [100–100.00]a | 0.52–2.99 | [13, 45] |
K-means clustering | 15 | 1.29 | [1.10–1.50]a | 99.69 | [100–100.00]a | 0.76–2.18 | [41, 100] |
Digestive disorders | |||||||
Random effects model, unadjusted | 11 | 1.26 | [0.98–1.61] | 97.18 | [96–98.00]a | − 0.60–1.06 | – |
Basic outlier removal | 6 | 1.36 | [1.20–1.60]a | 73.86 | [40–89.00]a | 0.04–0.57 | [55, 61, 72, 90] |
Influence analysis („leave-one-out”) | 9 | 1.20 | [0.92–1.60] | 96.68 | [95–98.00]a | 0.53–2.75 | [58, 72] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 7 | 1.14 | [0.78–1.70] | 98.24 | [98–99.00]a | 0.37–3.48 | [38, 53, 65, 72] |
Gaussian mixture model (GMM) clustering | 7 | 1.14 | [0.78–1.70] | 98.24 | [98–99.00]a | 0.37–3.48 | [38, 53, 65, 72] |
K-means clustering | 9 | 1.16 | [0.91–1.50] | 96.69 | [95–98.00]a | 0.53–2.54 | [38, 58] |
Endocrine and metabolic disorders | |||||||
Random effects model, unadjusted | 8 | 1.26 | [1.04–1.54]a | 90.31 | [83–94.00]a | − 0.35–0.82 | – |
Basic outlier removal | 7 | 1.33 | [1.10–1.60]a | 88.96 | [80–94.00]a | − 0.18–0.76 | [86] |
Influence analysis („leave-one-out”) | 6 | 1.29 | [1.10–1.50]a | 86.50 | [73–93.00]a | 0.79–2.09 | [51, 86] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 7 | 1.33 | [1.10–1.60]a | 88.96 | [80–94.00]a | 0.83–2.13 | [52] |
Gaussian mixture model (GMM) clustering | 7 | 1.33 | [1.10–1.60]a | 88.96 | [80–94.00]a | 0.83–2.13 | [52] |
K-means clustering | 7 | 1.29 | [1.00–1.60]a | 91.67 | [85–95.00]a | 0.67–2.49 | [50] |
Hematological disorders | |||||||
Random effects model, unadjusted | 7 | 2.16 | [1.36–3.44]a | 94.20 | [90–96.00]a | − 0.55–2.09 | – |
Basic outlier removal | 7 | 2.16 | [1.40–3.40]a | 94.20 | [90–96.00]a | − 0.55–2.09 | – |
Influence analysis („leave-one-out”) | 5 | 2.19 | [1.50–3.30]a | 81.02 | [56–92.00]a | 0.76–6.32 | [60, 118] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 5 | 2.19 | [1.50–3.30]a | 81.02 | [56–92.00]a | 0.76–6.32 | [92, 98] |
Gaussian mixture model (GMM) clustering | 5 | 2.19 | [1.50–3.30]a | 81.02 | [56–92.00]a | 0.76–6.32 | [92, 98] |
K-means clustering | 5 | 2.43 | [1.50–4.00]a | 94.94 | [91–97.00]a | 0.62–9.50 | [92] |
HIV | |||||||
Random effects model, unadjusted | 12 | 1.81 | [1.21–2.69]a | 93.67 | [91–96.00]a | − 0.77–1.95 | – |
Basic outlier removal | 10 | 1.43 | [1.00–2.00]a | 87.19 | [78–92.00]a | − 0.61–1.32 | [89, 118] |
Influence analysis („leave-one-out”) | 10 | 1.69 | [1.10–2.50]a | 87.43 | [79–93.00]a | 0.52–5.52 | [86, 92] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 8 | 1.64 | [0.90–3.00] | 92.07 | [87–95.00]a | 0.28–9.60 | [82, 86, 89, 118] |
Gaussian mixture model (GMM) clustering | 8 | 1.64 | [0.90–3.00] | 92.07 | [87–95.00]a | 0.28–9.60 | [82, 86, 89, 118] |
K-means clustering | 9 | 1.55 | [1.10–2.10]a | 87.12 | [78–93.00]a | 0.61–3.92 | [63, 71, 86] |
IBD | |||||||
Random effects model, unadjusted | 8 | 1.68 | [1.02–2.75]a | 99.73 | [100–100.00]a | − 1.02–2.05 | – |
Basic outlier removal | 6 | 1.50 | [1.30–1.80]a | 97.38 | [96–98.00]a | − 0.05–0.86 | [55, 64] |
Influence analysis („leave-one-out”) | 7 | 1.86 | [1.10–3.10]a | 99.75 | [100–100.00]a | 0.41–8.57 | [64] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 7 | 1.38 | [1.10–1.80]a | 98.02 | [97–99.00]a | 0.66–2.87 | [52] |
Gaussian mixture model (GMM) clustering | 7 | 1.38 | [1.10–1.80]a | 98.02 | [97–99.00]a | 0.66–2.87 | [52] |
K-means clustering | 7 | 1.38 | [1.10–1.80]a | 98.02 | [97–99.00]a | 0.66–2.87 | [52] |
Mental health condition | |||||||
Random effects model, unadjusted | 15 | 1.43 | [0.98–2.11] | 99.65 | [100–100.00]a | − 1.18–1.90 | – |
Basic outlier removal | 13 | 1.16 | [1.00–1.30]a | 87.85 | [81–92.00]a | − 0.19–0.48 | [106, 114] |
Influence analysis („leave-one-out”) | 14 | 1.23 | [1.00–1.50]a | 91.08 | [87–94.00]a | 0.66–2.28 | [114] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 9 | 1.64 | [0.83–3.20] | 99.80 | [100–100.00]a | 0.18–14.50 | [44, 74, 109, 110, 114] |
Gaussian mixture model (GMM) clustering | 9 | 1.64 | [0.83–3.20] | 99.80 | [100–100.00]a | 0.18–14.50 | [44, 74, 109, 110, 114] |
K-means clustering | 14 | 1.23 | [1.00–1.50]a | 91.08 | [87–94.00]a | 0.66–2.28 | [110] |
Musculoskeletal disorders | |||||||
Random effects model, unadjusted | 14 | 1.43 | [1.22–1.67]a | 96.45 | [95–97.00]a | − 0.19–0.90 | – |
Basic outlier removal | 11 | 1.3 | [1.20–1.40]a | 88.21 | [81–93.00]a | 0.00–0.52 | [51, 52, 86] |
Influence analysis („leave-one-out”) | 12 | 1.37 | [1.20–1.60]a | 91.37 | [87–94.00]a | 0.78–2.39 | [52, 86] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 9 | 1.45 | [1.10–1.90]a | 93.61 | [90–96.00]a | 0.66–3.23 | [52, 81, 99, 103] |
Gaussian mixture model (GMM) clustering | 9 | 1.45 | [1.10–1.90]a | 93.61 | [90–96.00]a | 0.66–3.23 | [52, 81, 99, 103] |
K-means clustering | 12 | 1.34 | [1.20–1.50]a | 96.48 | [95–97.00]a | 0.88–2.04 | [86, 107] |
Neurological disorders | |||||||
Random effects model, unadjusted | 6 | 1.52 | [1.08–2.14]a | 96.47 | [94–98.00]a | − 0.53–1.37 | – |
Basic outlier removal | 5 | 1.32 | [1.10–1.60]a | 88.15 | [75–94.00]a | − 0.26–0.82 | [79] |
Influence analysis („leave-one-out”) | 5 | 1.32 | [1.10–1.60]a | 88.15 | [75–94.00]a | 0.77–2.27 | [79] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 5 | 1.32 | [1.10–1.60]a | 88.15 | [75–94.00]a | 0.77–2.27 | [51] |
Gaussian mixture model (GMM) clustering | 5 | 1.32 | [1.10–1.60]a | 88.15 | [75–94.00]a | 0.77–2.27 | [51] |
K-means clustering | 4 | 1.35 | [1.00–1.80]a | 91.00 | [80–96.00]a | 0.61–2.97 | [51, 54] |
Psoriasis | |||||||
Random effects model, unadjusted | 6 | 1.25 | [1.08–1.44]a | 98.09 | [97–99.00]a | − 0.18–0.63 | – |
Basic outlier removal | 5 | 1.16 | [1.10–1.30]a | 59.33 | [0 -85.00]a | − 0.10–0.40 | [89] |
Influence analysis („leave-one-out”) | 4 | 1.21 | [1.10–1.30]a | 0.00 | [0–85.00] | 1.04–1.42 | [86, 89] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 6 | 1.25 | [1.10–1.40]a | 98.09 | [97–99.00]a | 0.83–1.87 | – |
Gaussian mixture model (GMM) clustering | 6 | 1.25 | [1.10–1.40]a | 98.09 | [97–99.00]a | 0.83–1.87 | – |
K-means clustering | 4 | 1.21 | [1.10–1.30]a | 0.00 | [0–85.00] | 1.04–1.42 | [51, 86] |
Renal disorders | |||||||
Random effects model, unadjusted | 10 | 1.17 | [0.93–1.48] | 94.82 | [92–97.00]a | − 0.59–0.91 | – |
Basic outlier removal | 8 | 1.15 | [1.00–1.30] | 84.85 | [72–92.00]a | − 0.22–0.51 | [49, 60] |
Influence analysis („leave-one-out”) | 8 | 1.28 | [1.00–1.60] | 90.00 | [83–94.00]a | 0.65–2.51 | [62, 118] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 9 | 1.18 | [0.91–1.50] | 95.39 | [93–97.00]a | 0.53–2.64 | [52] |
Gaussian mixture model (GMM) clustering | 9 | 1.26 | [1.00–1.50] | 89.47 | [82–94.00]a | 0.69–2.28 | [52] |
K-means clustering | 8 | 1.27 | [1.00–1.60] | 90.79 | [84–95.00]a | 0.66–2.43 | [52, 60] |
RA | |||||||
Random effects model, unadjusted | 13 | 1.62 | [1.29–2.02]a | 99.03 | [99–99.00]a | − 0.34–1.30 | – |
Basic outlier removal | 10 | 1.62 | [1.40–1.90]a | 93.26 | [90–96.00]a | − 0.02–0.99 | [69, 88, 89] |
Influence analysis („leave-one-out”) | 10 | 1.74 | [1.40–2.10]a | 97.16 | [96–98.00]a | 0.98–3.10 | [69, 88, 89] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 5 | 2.08 | [1.70–2.50]a | 95.61 | [92–98.00]a | 1.25–3.47 | [13, 49, 51, 86, 88, 89, 92] |
Gaussian mixture model (GMM) clustering | 5 | 2.08 | [1.70–2.50]a | 95.61 | [92–98.00]a | 1.25–3.47 | [13, 49, 51, 86, 88, 89, 92] |
K-means clustering | 12 | 1.75 | [1.50–2.10]a | 98.56 | [98–99.00]a | 0.96–3.17 | [51] |
SLE | |||||||
Random effects model, unadjusted | 10 | 2.87 | [1.99–4.13]a | 97.59 | [97–98.00]a | − 0.17–2.28 | – |
Basic outlier removal | 6 | 2.91 | [2.10–4.00]a | 91.97 | [85–96.00]a | 0.14–2.00 | [39, 66, 69, 92] |
Influence analysis („leave-one-out”) | 8 | 2.93 | [2.10–4.10]a | 96.95 | [96–98.00]a | 1.05–8.23 | [39, 66] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 6 | 2.63 | [1.40–4.90]a | 94.15 | [90–97.00]a | 0.44–15.90 | [49, 51, 52, 92] |
Gaussian mixture model (GMM) clustering | 6 | 2.63 | [1.40–4.90]a | 94.15 | [90–97.00]a | 0.44–15.90 | [49, 51, 52, 92] |
K-means clustering | 9 | 3.18 | [2.30–4.40]a | 96.76 | [95–98.00]a | 1.09–9.29 | [89] |
Transplantation | |||||||
Random effects model, unadjusted | 10 | 4.51 | [1.90–10.70]a | 98.43 | [98–99.00]a | − 1.20–4.21 | – |
Basic outlier removal | 7 | 3.68 | [2.00–6.80]a | 98.27 | [98–99.00]a | − 0.26–2.87 | [51, 52, 82] |
Influence analysis („leave-one-out”) | 8 | 3.55 | [1.30–9.80]a | 97.27 | [96–98.00]a | 0.19–64.80 | [89, 117] |
GOSH-Diagnostics Connectivity (DBSCAN) clustering | 9 | 5.59 | [2.70–12.00]a | 98.58 | [98–99.00]a | 0.56–56.10 | [52] |
Gaussian mixture model (GMM) clustering | 9 | 5.59 | [2.70–12.00]a | 98.58 | [98–99.00]a | 0.56–56.10 | [52] |
K-means clustering | 8 | 5.15 | [2.30–12.00]a | 98.75 | [98–99.00]a | 0.45–59.30 | [48, 52] |
nk, number of studies; OR, odds ratio; CI, confidence interval; PI, prediction interval
aIndicates significant values p < 0.005