Table 4.
Pooled relative risk and standardized mean difference estimates and 95% confidence intervals from weighted meta-analysis relative to healthy controls by study numbers and size with heterogeneity statistics
| Outcome | n | No. of Studies | Effect Estimate (95% CI) | I2, % (95% CI)a | τ2 | Refs. |
|---|---|---|---|---|---|---|
| Sociodemographic | Risk Ratio | |||||
| Unemployed | 733 | 8 | 1.89 (1.47 to 2.44) | 49 (0 to 76) | 0.06 | 20,23,24,27,29,33,35–37 |
| Married/with partner | 594 | 4 | 0.71 (0.53 to 0.95) | 81 (21 to 91) | 0.06 | 20,22,30,36,37 |
| Higher education | 395 | 3 | 1.05 (0.73 to 1.51) | 81 (26 to 91) | 0.1 | 22,29,36 |
| Live in family home | 418 | 2 | 1.84 (1.40 to 2.43) | 40 | 0.02 | 36,37 |
| Lifestyle | ||||||
| Alcohol abstainer | 536 | 3 | 1.96 (0.84 to 4.67) | 90 (67 to 95) | 0.5 | 30,36,38 |
| Current smoker | 487 | 2 | 0.72 (0.36 to 1.44) | 94 | 0.2 | 30,36 |
| Psychologic health | SMD | |||||
| Quality of lifeb | 678 | 10 | −0.65 (−0.88 to −0.43) | 83 (74 to 88) | 0.2 | 30,31,35,39,40,53,62,65,66,72; control data from refs. 78–81 |
| Transplantc,d | 517 | 9 | −0.42 (−0.64 to −0.20) | 77 (54 to 86) | 0.09 | 30,31,35,39,40,53,62,65,72; control data from refs. 78,79,81 |
| Dialysisd | 161 | 7 | −1.01 (−1.32 to -0.70) | 62 (0 to 81) | 0.1 | 31,35,39,53,62,65,66; control data from refs. 79–81 |
| Positive affecte | 121 | 3 | 0.40 (−0.12 to 0.91) | 84 (15 to 93) | 0.2 | 28,32,74 |
| Negative affecte | 121 | 3 | 0.18 (−0.26 to 0.61) | 79 (12 to 90) | 0.2 | 28,32,74 |
| Self-perception/self-imagef | 86 | 2 | −0.31 (−1.08 to 0.47) | 84 | 0.3 | 21,38 |
Although clustering (potential correlation from including subgroups taken from the same study) may underestimate between-study variance, the overall quality of life result showed a compensatory larger variance in effect size (τ2 value). To overcome any effect of clustering, we also undertook a sensitivity analysis by meta-analyzing the treatment by modality interaction in studies that reported data for both patients with transplants and patients on dialysis (n=6). This showed a pooled difference in SMD between dialysis and transplant groups of −0.62 (95% CI, −0.88 to −0.37), I2=0% (95% CI, 0 to 61), and τ2<0.0001. This was equivalent to the difference in effect estimates between transplant and dialysis presented in the table. 95% CI, 95% confidence interval; SMD, standardized mean difference.
95% CI for I2 incalculable with one degree of freedom.
Quality of life scales and forest plot can be seen in Figure 2. 36-Item Short Form Health Survey scores were converted to utility scores using model EQ1 by Ara and Brazier (12), and SDs were derived using a model by Wyld et al. (13). If studies did not report a normative comparator and country- and age-specific control data were readily available, we included them in the meta-analysis. When performing the meta-analysis without control data sourced externally to the studies from the systematic search, the following results were obtained: five studies, overall SMD of −0.42 (95% CI, −0.66 to −0.17), I2=78% (95% CI, 51 to 88), and τ2=0.09; five studies, transplant SMD of −0.31 (95% CI, −0.60 to −0.02), I2=84% (95% CI, 56 to 91), and τ2=0.09; and three studies, dialysis SMD of −0.67 (95% CI, −0.97 to −0.38), I2=0% (95% CI, 0 to 73), and τ2<0.0001.