Table 2.
Summary of the Two Cutoff Methods: A Priori Defined Threshold of .5% and Cutoffs Based on Quintiles
| Stratum number | Stratification cutoff, % | No. per stratum | No. (%) of patients infected ≤1 year | Comparison of strata a | Sensitivity, % b | Specificity, % c | Correctly Classified, % |
| Predefined | |||||||
| 1 | ≤.5 | 1,117 | 184 (16.5) | 1 vs 2 | 86.8 | 69.9 | 70.9 |
| 2 | >.5 | 2,190 | 28 (1.3) | – | – | – | – |
| By quintiles | |||||||
| 1 | 0–.15 | 679 | 146 (21.5) | 1 vs 2, 3, 4, 5 | 68.8 | 82.8 | 81.9 |
| 2 | .16–.67 | 649 | 44 (6.8) | 1, 2 vs 3, 4, 5 | 89.6 | 63.2 | 64.9 |
| 3 | .68–1.35 | 691 | 14 (2.0) | 1, 2, 3 vs 4, 5 | 96.2 | 41.4 | 44.9 |
| 4 | 1.36–2.07 | 633 | 7 (1.1) | 1, 2, 3, 4 vs 5 | 99.5 | 21.1 | 26.2 |
| 5 | 2.08–6.65 | 655 | 1 (.2) | – | – | – | – |
a
Indicates how strata are collapsed for comparison; for example, strata 1 versus all remaining strata.
b
Proportion of patients infected ≤1 year whose human immunodeficiency virus (HIV) sequence had ≤.5% ambiguous positions.
c
Proportion of patients infected >1 year whose HIV sequence had >.5% ambiguous positions.