Detecting Critical Decision Points during Cognitive-Behavioral Therapy and Light Therapy for Winter Depression
Conference Year
January 2019
Abstract
Introduction
Efficacious first-line treatments for winter seasonal affective disorder (SAD) include light therapy (LT) and cognitive-behavioral therapy (CBT-SAD). Our recently completed clinical trial found very similar post-treatment remissions rates for these treatments (47.6% in CBT-SAD and 47.2% in LT, Rohan et al., 2015). Therefore, nearly half of SAD patients do not remit in these treatments. The current study applied the approach of Steidtmann et al. (2013) to derive a critical decision point (treatment week) regarding whether a patient is likely to non-remit with CBT-SAD or LT. In practice, patients who are unlikely to remit can be identified at these decision points and their treatment can be modified, potentially improving outcomes.
Methods
Adults in a current episode of Major Depression, Recurrent with Seasonal Pattern (N= 177) were randomized to 6-weeks of CBT-SAD (n= 88; two 1½-hr group sessions/week; Rohan, 2008) or LT (n= 89; 10,000-lux cool-white florescent light, initiated at 30 min./day each morning and adjusted per algorithm). Depressive symptoms were measured weekly using the 29-item Structured Interview Guide for Hamilton Rating Scale for Depression–Seasonal Affective Disorder Version (SIGH-SAD; Williams et al., 1992), which includes the 21-item Hamilton Rating Scale for Depression (HAM-D) and an 8-item atypical subscale. Remission status at post-treatment was classified as either: >50% improvementin SIGH-SAD score from pre-treatment + HAM-D score <7 >+ atypical score<2 >+ atypical score
The majority of participants was female (84%) and White (93%) with a mean age of 45.6 years (SD= 12.8). At baseline, 27% had a comorbid Axis I diagnosis and 25% were taking antidepressant medications. Attrition was minimal. Of 177 randomized, 172 (97%) provided data at post-treatment.
Results
Consistent with Steidtmann et al. (2013), we conducted receiver operating curve (ROC) analyses to explore how well SIGH-SAD score at each treatment week (1, 2, 3, 4, or 5) predicted nonremission status at Week 6 (post-treatment) within each treatment. C-statistics, sensitivity, and specificity by treatment condition are presented in Table 1.
Within CBT-SAD, the SIGH-SAD score at Weeks 2, 3, 4, and 5 had adequate ability to potentially predict post-treatment remission status (C-statistics >0.70). However, Week 4 (and only in CBT-SAD) was the only timepoint at which the SIGH-SAD score had sufficient predictive ability to inform clinical decision-making (C-statistic >0.80). (See Figure 1). With a cutpoint of 13, sensitivity was .91 and specificity was .68. (See Table 1). No weekly SIGH-SAD score (at Weeks 2-5) was sufficiently predictive of nonremission in LT to inform clinical decision-making.
Discussion
This work is a first-step in identifying critical decision points to determine when a change in SAD treatment with CBT-SAD or LT is indicated, and these preliminary results require validation in an independent dataset. Week 4 of CBT-SAD may be a critical timepoint to identify likely nonremitters who need tailoring of intervention, based on SIGH-SAD score >13. This could be evaluated in future sequential multiple assignment randomized trials (SMARTs). In Week 4 of CBT-SAD, likely nonremitters could potentially be augmented with or crossed to LT. However, Week 4 represents 2/3 of the way through the 6-week CBT-SAD protocol. In tailoring CBT-SAD at Week 4, a SMART design (and a treating clinician) must consider the number of weeks remaining until spontaneous springtime remission occurs. If it is late in the winter season, tailoring of CBT-SAD may need to occur the following fall.
In the parent trial’s prospective follow-up of participants one and two winters later, CBT-SAD was associated with fewer recurrences and less severe depressive symptoms at the second winter follow-up than LT (Rohan et al., 2016). Offsetting SAD recurrence is arguably a higher priority public health challenge than achieving post-treatment remission. Therefore, we also plan to explore clinical decision points for identifying depression recurrence following treatment with CBT-SAD or LT.
Primary Faculty Mentor Name
Kelly Rohan
Faculty/Staff Collaborators
Julia Camuso, Jonah Meyerhoff, Pamela Vacek, Michael DeSarno
Status
Graduate
Student College
College of Arts and Sciences
Program/Major
Psychological Science
Primary Research Category
Social Sciences
Detecting Critical Decision Points during Cognitive-Behavioral Therapy and Light Therapy for Winter Depression
Introduction
Efficacious first-line treatments for winter seasonal affective disorder (SAD) include light therapy (LT) and cognitive-behavioral therapy (CBT-SAD). Our recently completed clinical trial found very similar post-treatment remissions rates for these treatments (47.6% in CBT-SAD and 47.2% in LT, Rohan et al., 2015). Therefore, nearly half of SAD patients do not remit in these treatments. The current study applied the approach of Steidtmann et al. (2013) to derive a critical decision point (treatment week) regarding whether a patient is likely to non-remit with CBT-SAD or LT. In practice, patients who are unlikely to remit can be identified at these decision points and their treatment can be modified, potentially improving outcomes.
Methods
Adults in a current episode of Major Depression, Recurrent with Seasonal Pattern (N= 177) were randomized to 6-weeks of CBT-SAD (n= 88; two 1½-hr group sessions/week; Rohan, 2008) or LT (n= 89; 10,000-lux cool-white florescent light, initiated at 30 min./day each morning and adjusted per algorithm). Depressive symptoms were measured weekly using the 29-item Structured Interview Guide for Hamilton Rating Scale for Depression–Seasonal Affective Disorder Version (SIGH-SAD; Williams et al., 1992), which includes the 21-item Hamilton Rating Scale for Depression (HAM-D) and an 8-item atypical subscale. Remission status at post-treatment was classified as either: >50% improvementin SIGH-SAD score from pre-treatment + HAM-D score <7>+ atypical score<2>+ atypical score
The majority of participants was female (84%) and White (93%) with a mean age of 45.6 years (SD= 12.8). At baseline, 27% had a comorbid Axis I diagnosis and 25% were taking antidepressant medications. Attrition was minimal. Of 177 randomized, 172 (97%) provided data at post-treatment.
Results
Consistent with Steidtmann et al. (2013), we conducted receiver operating curve (ROC) analyses to explore how well SIGH-SAD score at each treatment week (1, 2, 3, 4, or 5) predicted nonremission status at Week 6 (post-treatment) within each treatment. C-statistics, sensitivity, and specificity by treatment condition are presented in Table 1.
Within CBT-SAD, the SIGH-SAD score at Weeks 2, 3, 4, and 5 had adequate ability to potentially predict post-treatment remission status (C-statistics >0.70). However, Week 4 (and only in CBT-SAD) was the only timepoint at which the SIGH-SAD score had sufficient predictive ability to inform clinical decision-making (C-statistic >0.80). (See Figure 1). With a cutpoint of 13, sensitivity was .91 and specificity was .68. (See Table 1). No weekly SIGH-SAD score (at Weeks 2-5) was sufficiently predictive of nonremission in LT to inform clinical decision-making.
Discussion
This work is a first-step in identifying critical decision points to determine when a change in SAD treatment with CBT-SAD or LT is indicated, and these preliminary results require validation in an independent dataset. Week 4 of CBT-SAD may be a critical timepoint to identify likely nonremitters who need tailoring of intervention, based on SIGH-SAD score >13. This could be evaluated in future sequential multiple assignment randomized trials (SMARTs). In Week 4 of CBT-SAD, likely nonremitters could potentially be augmented with or crossed to LT. However, Week 4 represents 2/3 of the way through the 6-week CBT-SAD protocol. In tailoring CBT-SAD at Week 4, a SMART design (and a treating clinician) must consider the number of weeks remaining until spontaneous springtime remission occurs. If it is late in the winter season, tailoring of CBT-SAD may need to occur the following fall.
In the parent trial’s prospective follow-up of participants one and two winters later, CBT-SAD was associated with fewer recurrences and less severe depressive symptoms at the second winter follow-up than LT (Rohan et al., 2016). Offsetting SAD recurrence is arguably a higher priority public health challenge than achieving post-treatment remission. Therefore, we also plan to explore clinical decision points for identifying depression recurrence following treatment with CBT-SAD or LT.