Table 3.
Meta-analysis of diagnostic accuracy for patients with Parkinson’s disease vs healthy controls summary statistics for the bivariate and hierarchal summary receiver operating characteristic (HSROC) models
| Model | Variable | Coefficient estimate ± SE (95% CI) |
|---|---|---|
| Summary statistic | Sensitivity | 0.725 ± 0.038 (0.644–0.793) |
| Specificity | 0.759 ± 0.031 (0.692–0.814) | |
| DOR | 8.29 ± 2.15 (4.98–13.79) | |
| posLR | 3.00 ± 0.423 (2.28–3.96) | |
| negLR | 0.362 ± 0.053 (0.272–0.482) | |
| 1/negLR | 2.76 ± 0.403 (2.07–3.67) | |
| Bivariate | Logit-transformed sensitivity | 0.970 ± 0.191 (0.595–1.34) |
| Logit-transformed sensitivity variance | 1.14 ± 0.170 (0.812–1.48) | |
| Logit-transformed specificity | 0.813 ± 0.272 (0.421–1.57) | |
| Logit-transformed specificity variance | 0.618 ± 0.207 (0.320–1.19) | |
| Correlation between sensitivity and specificity | 0.035 ± 0.222 (-0.381–0.439) | |
| AUC (partial AUC) | 0.800 (0.692) | |
| HSROC | Lambda (Λ) | 2.13 ± 0.261 (1.62–2.64) |
| Theta (Θ) | -0.160 ± 0.172 (-0.498–0.178) | |
| Beta (β) | -0.137 ± 0.237 (-0.601–0.327) | |
| Variance Λ | 1.47 ± 0.468 (0.785–2.74) | |
| Variance Θ | 0.342 ± 0.113 (0.179–0.655) |