Baseline predictors of treatment outcome in Internet-based alcohol interventions: a recursive partitioning analysis alongside a randomized trial

Table 3 Performance characteristics of the predictors in the classification trees with 95% confidence intervals

Classification	Sensitivity	Specificity	Negative predictive value	Positive predictive value
Chance (random classification)	0.50 [0.40, 0.58]	0.50 [0.43, 0.58]	0.56 [0.46, 0.66]	0.45 [0.33, 0.56]
Screener conservative	0.34 [0.21, 0.48]	0.89 [0.82, 0.96]	0.63 [0.53, 0.73]	0.72 [0.54, 0.88]
Screener progressive	0.87 [0.74, 0.93]	0.30 [0.19, 0.39]	0.73 [0.54, 0.86]	0.49 [0.39, 0.58]

Bootstrapped (200 iterations) 95% confidence intervals are displayed within brackets [lower, upper]; Screener conservative interprets subgroup III as responding negative to treatment; Screener progressive interprets subgroup III as responding positive to treatment; Sensitivity is the proportion of actual positive treatment responders which are correctly identified; Specificity is the proportion of negative treatment responders which are correctly identified; Negative predictive value is the proportion of participants with negative predicted outcome who are correctly identified; Positive predictive value is the proportion of participants with positive predicted outcome who are correctly identified.

ISSN: 1471-2458