Skip to main content

Table 5 Area under the receiver operating characteristic curve (AUC) for depression measures for detecting any improvement

From: Responsiveness of PROMIS and Patient Health Questionnaire (PHQ) Depression Scales in three clinical trials

Depression measure* Average accuracy across trials Accuracy for detecting any improvement
Retrospective Global Rating of Change (GRC)
Accuracy for detecting any improvement
Prospective Global Rating of Change (GRC)
Retro-spective GRC Pro-spective GRC CAMEO SPACE SSM CAMEO SPACE SSM
AUC (95% CI) AUC (95% CI) AUC (95% CI) AUC (95% CI) AUC (95% CI) AUC (95% CI)
PROMIS 4-item .565 .658 .603 (.510–.697) .586 (.512–.660) .507 (.436–.579) .646 (.553–.739) .699 (.628–.770) .630 (.544–.716)
PROMIS 6-item .578 .660 .634 (.542–.727 .570 (.493–.646) .530 (.457–.603) .638 (.542–.734) .712 (.642–.782) .631 (.546–.715)
PROMIS 8-item .574 .664 .623 (.530–.716 .575 (.499–.650) .523 (.450–.596) .646 (.550–.741) .712 (.646–.785) .633 (.550–.717)
PROMIS Short-form .583 .671 .644 (.551–.736) .577 (.501–.652) .529 (.456–.602) .638 (.543–.732) .741 (.673–.808) .634 (.551–.717)
PHQ-9 .617 .649 .697 (.608–.786) .587 (.512–.662) .566 (.493–.639) .664 (.572–.755) .644 (.570–.717) .640 (.557–.723)
PHQ-2 .570 .627 .598 (.597–.690) .576 (.505–.646) .535 (.464–.606) .576 (.483–.670) .686 (.619–.752) .620 (.537–.703)
SF-36 Mental Health .627 (.534–.720) .745 (.661–.829)     
  1. *AUC is probability of correctly discriminating between patients who have improved and those who have not. Any improvement ≥ “a little better”; moderate improvement ≥ “moderately better”
  2. 6 month follow-up for CAMEO; 3 months for SPACE and SSM. The proportion of patients reporting any improvement by retrospective GRC was 57%, 40%, and 57% in CAMEO, SPACE, and SSM, respectively. The proportion reporting any improvement by prospective GRC was 40%, 39%, and 29% in CAMEO, SPACE, and SSM, respectively
  3. There were no significant differences at P < .01 (using Bonferroni’s correction for multiple comparisons) between any of the retrospective AUC’s. The prospective AUC was significantly lower for the PHQ-2 (P = .003) compared to the PROMIS Short-form in the CAMEO trial