Table 1

Performance criteria in ROC analysisa

Criteria

Anti-dsDNA-NcX

Anti-dsDNA

Antinucleosome

Farr


Area under the curve

0.9133b

0.8455

0.9118

0.8823

95% confidence interval

0.89 to 0.94

0.81 to 0.88

0.89 to 0.94

0.85 to 0.91

Sensitivity at specificity of 95%

72.5%b

55.6%

67.2%

65.7%

(cutoff)

(>49.7)

(>58.8)

(>6.8)

(>5.1)

Sensitivity at specificity of 98%

66.7%b

43.0%

58.0%

55.6%

(cutoff)

(>71.5)

(>99.7)

(>11.1)

(>6.0)

Sensitivity at specificity of 98.15%c

66.7%b

41.5%

55.6%

55.6

(cutoff)

(>72.03)

(>105.8)

(>12.1)

(>6.0)

Sensitivity at specificity of 99%

59.9%b

35.8%

53.6%

54.1%

(cutoff)

(>103.4)

(>151.6)

(>14.8)

(>6.5)

Maximum sum of sensitivity + specificity

170.9b

159.1

168.7

162.1


adsDNA, double-stranded DNA. Test criteria for the Anti-dsDNA-NcX, anti-dsDNA and antinucleosome enzyme-linked immunosorbent assays (ELISAs), as well as for the anti-dsDNA radioimmunoassay (also known as a Farr assay), were calculated using receiver operating characteristic (ROC) curve analysis based on test readings of 964 samples from 207 systemic lupus erythematosus (SLE) patients, 357 disease controls and 400 healthy donors. Outcome parameters of ROC curve analysis were diagnosis versus no diagnosis of SLE. bHighest value over all four assays. cThis specificity was calculated for the Crithidia luciliae immunofluorescence (CLIF) assay.

Biesen et al. Arthritis Research & Therapy 2011 13:R26   doi:10.1186/ar3250

Open Data