Seventy-eight articles reporting 82 patient cohorts were included (the number of patients was unclear).
Impedance plethysmography (42 cohorts, data reported separately for distal and proximal DVT in 28 cohorts).
The pooled sensitivity was 75% (95% confidence interval, CI: 73, 77) for all DVT, 88% (95% CI: 86, 90) for proximal DVT and 28% (95% CI: 24, 33) for distal DVT. The pooled specificity was 90% (95% CI: 89, 91). There was evidence of statistical heterogeneity in all accuracy measures (p<0.001). Setting of recruitment (p=0.098) and blind reporting of the reference standard (p=0.056) were associated with variations in sensitivity. Proportion of males in the cohort (p=0.01), DVT prevalence (p=0.043), setting for recruitment (p=0.09), consecutive recruitment (p=0.017) and prospective study (p=0.046) were associated with variations in specificity.
Strain gauge plethysmography (20 cohorts, data reported separately for distal and proximal DVT in 10 cohorts).
The pooled sensitivity was 83% (95% CI: 81, 85) for all DVT, 90% (95% CI: 88, 92) for proximal DVT and 56% (95% CI: 50, 63) for distal DVT. The pooled specificity was 81% (95% CI: 79, 82). There was evidence of statistical heterogeneity in all accuracy measures (p<0.001), except for sensitivity for distal DVT (p=0.033). Setting of recruitment (p<0.001) and proportion of males (p=0.0005) were associated with variations in sensitivity; no variables were associated with variation in specificity.
Air plethysmography (4 cohorts, data reported separately for distal and proximal DVT in 2 cohorts).
The pooled sensitivity was 85% (95% CI: 79, 90) for all DVT, 98% (95% CI: 93, 100) for proximal DVT and 39% (95% CI: 22, 58) for distal DVT. The pooled specificity was 91% (95% CI: 81, 95). There was evidence of statistical heterogeneity in the overall sensitivity and specificity (p<0.02), but not in separate estimates of sensitivity for proximal and distal DVT (p>0.18).
Light reflex rheography (9 cohorts, data reported separately for distal and proximal DVT in 4 cohorts).
The pooled sensitivity was 91% (95% CI: 87, 94) for all DVT, 94% (95% CI: 88, 98) for proximal DVT and 92% (95% CI: 74, 99) for distal DVT. The pooled specificity was 71% (95% CI: 66, 75). There was evidence of statistical heterogeneity in the overall sensitivity and specificity (p<0.001), but not in separate estimates of sensitivity for proximal and distal DVT (p>0.17).
Phleborheography (7 cohorts, data reported separately for distal and proximal DVT in 4 cohorts).
The pooled sensitivity was 86% (95% CI: 83, 89) for all DVT, 92% (95% CI: 88, 94) for proximal DVT and 58% (95% CI: 48, 68) for distal DVT. The pooled specificity was 93% (95% CI: 91, 95). There was evidence of statistical heterogeneity (p<0.001).