1 Tables

1.1 Table 1

Characteristic

Active TBS

Sham TBS

P

N = 54

N = 54

Demographics

Gender (% Female)

36 (67)

41 (76)

0.4

Age, y

66.9 ± 4.8

66.6 ± 6.9

0.81

White - no. (%)

48 (89)

40 (74)

0.08

Years of education - mean (SD)

12.9 ± 4.5; n=53

13 ± 5.2; n=53

0.9

At least 3 minimum wages - no. (%)

29 (54)

28 (53); n=53

>0.99

Retired - no. (%)

35 (65)

32 (59)

0.69

Relationship Status (% Partnered)

24 (44)

18 (33)

0.32

Physical Activity (% Active)

11 (21); n=53

9 (17); n=53

0.8

Clinical characteristics

Hypertension - no. (%)

22 (41)

25 (47); n=53

0.64

Hypothyroidism - no. (%)

11 (20)

10 (19); n=53

>0.99

Duration of current episode, months - mean (SD)

37.9 ± 40.7; n=37

27 ± 30.7; n=36

0.2

Refractory depression - no. (%)

21 (47); n=45

22 (49); n=45

>0.99

Currently in psychotherapy - no. (%)

8 (16); n=50

13 (27); n=49

0.3

Current benzodiazepine use - no. (%)

6 (18); n=33

11 (33); n=33

0.26

No. failed AD trials in lifetime - median (IQR)

2 (1-4); n=29

2 (1-3); n=34

0.92

Onset age of MDD, years - mean (SD)

42.4 ± 15.5; n=37

43.1 ± 18.4; n=36

0.85

No. previous depressive episodes - median (IQR)

3 (1-4); n=36

3 (2-4); n=36

0.72

Scales at baseline

Hamilton Depression Rating Scale (HDRS-17)

20.9 ± 3.4

22.1 ± 3.8

0.1

Montgomery-Asberg Depression Rating Scale (MADRS)

27.1 ± 6.3

28 ± 6.4

0.48

Geriatric Depression Scale (GDS)

8.6 ± 1.7

8.8 ± 1.6

0.38

Mania (YMRS)

1.7 ± 1.6

1.6 ± 1.4

0.8

Positive affect (PANAS)

25.7 ± 5.6

24.9 ± 4.5

0.45

Negative affect (PANAS)

24.1 ± 5.8

23.3 ± 5.5

0.44

Clinical Global Impression (CGI)

4.2 ± 0.9

4.4 ± 0.9

0.47

Cumulative Illness (CIRS-G)

4.2 ± 3.1

3.6 ± 2.8

0.3

Note: (*) indicates that data from the HERAH system is missing (33 of 108 patients). There were no cases of hyperthyroidism.

1.2 Scale scores across time

1.3 Response & Remission

Group

Type

Week 1

Week 2

Week 4

Week 6

Week 8

Week 12

Active TBS

Response - no. (%)

6 (11)

11 (20)

14 (28)

19 (39)

24 (50)

25 (52)

Active TBS

Remission - no. (%)

1 (2)

3 (6)

6 (12)

8 (16)

15 (31)

16 (33)

Sham TBS

Response - no. (%)

4 (8)

6 (12)

7 (15)

12 (24)

16 (36)

13 (29)

Sham TBS

Remission - no. (%)

4 (8)

2 (4)

1 (2)

4 (8)

8 (18)

9 (20)

1.4 Hamilton Depression Rating Scale (HDRS) score reductions

## # A tibble: 2 × 5
##   condition_rev `mean(diff_hdrs_6, na.rm = TRUE)` sd(diff_hdrs_6, na.rm = TRUE…¹
##   <fct>                                     <dbl>                          <dbl>
## 1 Active TBS                                 8.65                           6.24
## 2 Sham TBS                                   7                              5.12
## # ℹ abbreviated name: ¹​`sd(diff_hdrs_6, na.rm = TRUE)`
## # ℹ 2 more variables: `mean(diff_hdrs_12, na.rm = TRUE)` <dbl>,
## #   `sd(diff_hdrs_12, na.rm = TRUE)` <dbl>

2 Figures

All error bars represent 95% confidence intervals.

2.1 Primary and secondary outcomes

2.2 Alternative figures for main outcome

2.3 Response and Remission

Condition

Type

Baseline

Week 1

Week 2

Week 4

Week 6

Week 8

Week 12

Active TBS

HDRS remission - no. (%)

0 (0)

1 (2)

3 (6)

6 (12)

8 (16)

15 (31)

16 (33)

Sham TBS

HDRS remission - no. (%)

0 (0)

4 (8)

2 (4)

1 (2)

4 (8)

8 (18)

9 (20)

Active TBS

HDRS response - no. (%)

0 (0)

6 (11)

11 (20)

14 (28)

19 (39)

24 (50)

25 (52)

Sham TBS

HDRS response - no. (%)

0 (0)

4 (8)

6 (12)

7 (15)

12 (24)

16 (36)

13 (29)

Active TBS

MADRS remission - no. (%)

1 (2)

12 (22)

7 (13)

18 (36)

14 (29)

16 (33)

22 (46)

Sham TBS

MADRS remission - no. (%)

0 (0)

8 (15)

4 (8)

8 (17)

10 (20)

11 (25)

12 (27)

Active TBS

MADRS response - no. (%)

0 (0)

21 (39)

14 (26)

22 (44)

23 (47)

23 (48)

27 (56)

Sham TBS

MADRS response - no. (%)

0 (0)

20 (38)

8 (15)

12 (25)

18 (37)

18 (41)

13 (29)

MADRS response and remission rates reflect the temporary drop in scores at week 1.

3 Antidepressive effects of treatment

3.1 Linear mixed models

## Linear mixed model fit by REML. t-tests use Satterthwaite's method [
## lmerModLmerTest]
## Formula: hdrs_total ~ condition * time_log + (1 | subject)
##    Data: dss
## 
## REML criterion at convergence: 4137.5
## 
## Scaled residuals: 
##     Min      1Q  Median      3Q     Max 
## -2.9720 -0.5970 -0.0486  0.5606  3.8498 
## 
## Random effects:
##  Groups   Name        Variance Std.Dev.
##  subject  (Intercept) 13.85    3.721   
##  Residual             16.16    4.020   
## Number of obs: 701, groups:  subject, 108
## 
## Fixed effects:
##                              Estimate Std. Error       df t value Pr(>|t|)    
## (Intercept)                   20.8514     0.6582 207.6871  31.677   <2e-16 ***
## conditionActive TBS           -1.1019     0.9288 206.2149  -1.186   0.2368    
## time_log                      -2.8662     0.2633 606.2825 -10.887   <2e-16 ***
## conditionActive TBS:time_log  -0.7424     0.3687 604.6063  -2.014   0.0445 *  
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Correlation of Fixed Effects:
##             (Intr) cnATBS tim_lg
## cndtnActTBS -0.709              
## time_log    -0.539  0.382       
## cndtnATBS:_  0.385 -0.540 -0.714
##   effect group                         term   estimate std.error  statistic
## 1  fixed  <NA>                  (Intercept) 20.8514328 0.6582475  31.677194
## 2  fixed  <NA>          conditionActive TBS -1.1019010 0.9287881  -1.186386
## 3  fixed  <NA>                     time_log -2.8661628 0.2632713 -10.886729
## 4  fixed  <NA> conditionActive TBS:time_log -0.7423533 0.3686863  -2.013509
##         df      p.value          d   CI     CI_low     CI_high
## 1 207.6871 1.826499e-81  4.3961438 0.95  3.8919460  4.89687156
## 2 206.2149 2.368351e-01 -0.1652326 0.95 -0.4384699  0.10840369
## 3 606.2825 2.471250e-25 -0.8842802 0.95 -1.0507414 -0.71715137
## 4 604.6063 4.450369e-02 -0.1637749 0.95 -0.3233939 -0.00402086

A linear mixed-effects model was used to investigate if treatment was effective. Time was assumed to be linear on a natural logarithmic scale. HDRS-17 was the outcome, group and time were included as interacting fixed effects, and subject as random-intercept effect. Results showed a reduction in HDRS-17 scores across time, independent of group (t(606) = -10.89, P < .001). There was a a significant interaction of group and time (t(604) = -2.01, P = 0.045). This indicates that active TBS treatment was effective in reducing depressive symptoms.

The following tables show the interaction tests for all outcomes at different endpoints (week 6 or 12).

Results until week 6

Scale

Estimate

SE

t

DF

P

Cohen's d

HDRS

-0.90

0.47

-1.89

412.30

0.059

-0.36 [-0.74, 0.01]

MADRS

-1.97

0.83

-2.37

412.87

0.018

-0.45 [-0.83, -0.08]

GDS

-0.20

0.18

-1.08

413.27

0.280

-0.21 [-0.58, 0.17]

PANAS negative

-0.62

0.50

-1.24

412.11

0.216

-0.24 [-0.61, 0.14]

PANAS positive

-1.07

0.49

-2.20

412.41

0.029

-0.42 [-0.8, -0.04]

YMRS

-0.17

0.19

-0.90

404.08

0.367

-0.17 [-0.55, 0.2]

CGI1

-0.32

0.12

-2.73

415.52

0.007

-0.52 [-0.9, -0.15]

Note: TMA results are based 84 patients and on 2 evaluations only, baseline and week 4.

Results until week 12

Scale

Estimate

SE

t

DF

P

Cohen's d

HDRS

-0.74

0.37

-2.01

604.61

0.045

-0.42 [-0.83, -0.01]

MADRS

-1.52

0.60

-2.53

605.49

0.012

-0.53 [-0.94, -0.12]

GDS

-0.12

0.13

-0.94

606.78

0.348

-0.2 [-0.6, 0.21]

PANAS negative

-0.32

0.34

-0.92

602.59

0.358

-0.19 [-0.6, 0.22]

PANAS positive

-0.63

0.34

-1.88

603.47

0.060

-0.39 [-0.8, 0.02]

YMRS

-0.05

0.12

-0.37

594.46

0.708

-0.08 [-0.49, 0.33]

CGI1

-0.21

0.09

-2.46

609.84

0.014

-0.51 [-0.92, -0.1]

3.2 Linear mixed models on per-protocol (PP) sample

## # A tibble: 7 × 2
##    time `sum(!is.na(hdrs_total))`
##   <int>                     <int>
## 1     0                       108
## 2     1                       106
## 3     2                       106
## 4     4                        98
## 5     6                        98
## 6     8                        92
## 7    12                        93
## # A tibble: 2 × 2
##   condition      N
##   <fct>      <int>
## 1 Sham TBS      46
## 2 Active TBS    50

Results until week 6

Scale

Estimate

SE

t

DF

P

Cohen's d

HDRS

-1.01

0.49

-2.05

381.10

0.041

-0.41 [-0.8, -0.02]

MADRS

-2.21

0.86

-2.57

381.25

0.010

-0.51 [-0.9, -0.12]

GDS

-0.19

0.19

-0.99

381.15

0.322

-0.2 [-0.59, 0.19]

PANAS negative

-0.61

0.52

-1.17

381.11

0.241

-0.23 [-0.62, 0.16]

PANAS positive

-1.14

0.50

-2.27

381.20

0.024

-0.45 [-0.84, -0.06]

YMRS

-0.19

0.18

-1.09

381.38

0.276

-0.22 [-0.61, 0.17]

CGI1

-0.37

0.12

-3.04

381.28

0.002

-0.61 [-1, -0.21]

Note: TMA results are based 84 patients and on 2 evaluations only, baseline and week 4.

Results until week 12

Scale

Estimate

SE

t

DF

P

Cohen's d

HDRS

-0.88

0.38

-2.30

561.53

0.022

-0.5 [-0.92, -0.07]

MADRS

-1.77

0.61

-2.88

562.31

0.004

-0.62 [-1.05, -0.2]

GDS

-0.13

0.13

-1.01

563.12

0.314

-0.22 [-0.64, 0.21]

PANAS negative

-0.25

0.35

-0.70

561.60

0.487

-0.15 [-0.57, 0.27]

PANAS positive

-0.69

0.35

-2.01

561.91

0.045

-0.43 [-0.86, -0.01]

YMRS

-0.08

0.12

-0.71

563.97

0.477

-0.15 [-0.58, 0.27]

CGI1

-0.26

0.09

-2.81

563.37

0.005

-0.61 [-1.03, -0.18]

3.3 LMM with Autoregressive covariance structure (HDRS only)

An autoregressive covariance structure is a type of within-subject correlation structure sometimes used in longitudinal data analysis. It assumes that the correlation between two observations within the same subject decreases as the time between the observations increases.

Until week 6:

## Linear mixed-effects model fit by REML
##   Data: filter(dss, time <= 6) 
##        AIC      BIC    logLik
##   3030.334 3060.002 -1508.167
## 
## Random effects:
##  Formula: ~1 | subject
##         (Intercept) Residual
## StdDev:    3.226422 4.073749
## 
## Correlation Structure: Continuous AR(1)
##  Formula: ~time | subject 
##  Parameter estimate(s):
##       Phi 
## 0.3074477 
## Fixed effects:  hdrs_total ~ condition * time 
##                              Value Std.Error  DF  t-value p-value
## (Intercept)              20.491977 0.6364831 406 32.19564  0.0000
## conditionActive TBS      -1.277550 0.8990065 106 -1.42107  0.1582
## time                     -1.055508 0.1282522 406 -8.22994  0.0000
## conditionActive TBS:time -0.266200 0.1810900 406 -1.46999  0.1423
##  Correlation: 
##                          (Intr) cnATBS time  
## conditionActive TBS      -0.708              
## time                     -0.550  0.389       
## conditionActive TBS:time  0.390 -0.551 -0.708
## 
## Standardized Within-Group Residuals:
##        Min         Q1        Med         Q3        Max 
## -2.9567925 -0.5533191 -0.0632191  0.5697592  2.9494388 
## 
## Number of Observations: 516
## Number of Groups: 108

Until week 12:

## Linear mixed-effects model fit by REML
##   Data: dss 
##        AIC      BIC    logLik
##   4202.237 4234.064 -2094.118
## 
## Random effects:
##  Formula: ~1 | subject
##         (Intercept) Residual
## StdDev:     3.44291 4.703078
## 
## Correlation Structure: Continuous AR(1)
##  Formula: ~time | subject 
##  Parameter estimate(s):
##       Phi 
## 0.5278129 
## Fixed effects:  hdrs_total ~ condition * time 
##                              Value Std.Error  DF   t-value p-value
## (Intercept)              19.571058 0.6965046 591 28.098963  0.0000
## conditionActive TBS      -1.552329 0.9831819 106 -1.578882  0.1173
## time                     -0.550677 0.0751718 591 -7.325574  0.0000
## conditionActive TBS:time -0.119470 0.1051168 591 -1.136548  0.2562
##  Correlation: 
##                          (Intr) cnATBS time  
## conditionActive TBS      -0.708              
## time                     -0.560  0.397       
## conditionActive TBS:time  0.401 -0.563 -0.715
## 
## Standardized Within-Group Residuals:
##         Min          Q1         Med          Q3         Max 
## -2.94323100 -0.62787867 -0.09888731  0.49768289  3.15279543 
## 
## Number of Observations: 701
## Number of Groups: 108

These models show no significant treatment effect (“conditionActive TBS:time”).

3.4 Delta analyses for HDRS

These are t-tests that compare the differences between baseline and endpoint for the 2 groups. They are a more traditional approach, but only don’t take all measures into account and have less power.

Endpoint

t

DF

P

Week 2

-0.71

103.40

0.477

Week 6

-1.43

92.50

0.155

Week 12

-1.34

90.97

0.184

3.5 Respone and Remission

3.5.1 Mixed logistic regressions

This approach considers all time points. Odds ratios indicate the change in chances for response or remission for one week of treatment.

Results until week 6

Scale

Odds ratio

z

P

HDRS response

1.38

0.57

0.569

HDRS remission

30.30

2.41

0.016

MADRS response

1.60

1.20

0.229

MADRS remission

1.20

0.39

0.699

Results until week 12

Scale

Odds ratio

z

P

HDRS response

1.55

1.08

0.278

HDRS remission

3.54

2.06

0.039

MADRS response

1.73

2.00

0.045

MADRS remission

1.22

0.60

0.548

3.5.2 Logistic regression

Logistic regression only considers data from the endpoint, it’s the more common approach for this type of analysis though.

Results for week 6

Scale

Odds ratio

z

P

HDRS response

1.95

1.51

0.131

HDRS remission

2.20

1.21

0.226

MADRS response

1.52

1.02

0.307

MADRS remission

1.56

0.94

0.349

Results for week 12

Scale

Odds ratio

z

P

HDRS response

2.68

2.25

0.025

HDRS remission

2.00

1.44

0.151

MADRS response

3.16

2.62

0.009

MADRS remission

2.33

1.90

0.057

4 Adverse events

t-Test for group difference in number of adverse events during the entire study”

## # A tibble: 1 × 7
##   statistic  t_df p_value alternative estimate lower_ci upper_ci
##       <dbl> <dbl>   <dbl> <chr>          <dbl>    <dbl>    <dbl>
## 1     -1.31  104.   0.192 two.sided    -0.0793   -0.199   0.0404

Logistic regression to test for group difference in number of adverse events for the final week

## 
## Call:
## glm(formula = adverse_effects_w12 ~ condition, family = binomial, 
##     data = dss_wide)
## 
## Coefficients:
##                     Estimate Std. Error z value Pr(>|z|)    
## (Intercept)          -1.8718     0.4385  -4.268 1.97e-05 ***
## conditionActive TBS   0.5635     0.5651   0.997    0.319    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## (Dispersion parameter for binomial family taken to be 1)
## 
##     Null deviance: 85.015  on 91  degrees of freedom
## Residual deviance: 83.995  on 90  degrees of freedom
##   (16 observations deleted due to missingness)
## AIC: 87.995
## 
## Number of Fisher Scoring iterations: 4

5 Blinding

Active TBS

Sham TBS

Chi-Square

P

Patients

Guess active

34

22

Guess sham

14

22

3.35

0.067

Raters

Guess active

25

21

Guess sham

23

23

0.04

0.835

6 Treatment response (HDRS) prediction

Predictors were evaluated for HDRS changes (deltas) from baseline to endpoint using linear regression and for HDRS response using logistic regression (generalized linear model). For each predictor, separate models were set up with the outcome referring to Week 6 or 12 as possible endpoint. Condition and predictor were included as interacting factors. A significant main effect of the predictor indicates that the predictor had an effect on HDRS changes/responses on the whole sample, independent of condition. A significant interaction of the predictor and condition indicates that the effectiveness of TBS depended on the predictor.

Variables not included because of no or only very few cases: - Hyperthyroidism (no case) - Stroke (“demarre”, only 3 cases)

6.1 Predicting HDRS change

Predictor

Main effect

Interaction

Estimate

SE

t

P

Estimate

SE

t

P

Change to week 6

Gender

0.00

1.91

0.00

1.000

-1.11

2.64

-0.42

0.674

Age

0.05

0.12

0.47

0.641

-0.42

0.21

-2.02

0.046

Ethnicity (white/non-white)

-1.10

1.82

-0.60

0.547

2.27

3.27

0.70

0.488

Years of education

-0.13

0.16

-0.84

0.402

0.37

0.25

1.47

0.145

Unemployed (yes/no)

-1.83

2.35

-0.78

0.437

1.12

3.81

0.29

0.769

Wage (low/high)

1.16

1.66

0.70

0.487

-2.31

2.34

-0.99

0.326

Relationship (yes/no)

-1.53

1.72

-0.89

0.376

2.39

2.39

1.00

0.320

Phystical activity (low/high)

-0.64

2.38

-0.27

0.787

0.86

3.16

0.27

0.787

Hypertension

1.05

1.68

0.63

0.531

-0.53

2.37

-0.22

0.824

Duration of current episode

-0.02

0.03

-0.61

0.545

0.03

0.04

0.74

0.464

Refractory depression (yes/no)

-1.57

1.67

-0.94

0.350

0.14

2.38

0.06

0.952

Current benzodiazepine use (yes/no)

0.95

2.33

0.41

0.685

1.59

3.61

0.44

0.661

Currently in psychotherapy (yes/no)

-0.73

2.01

-0.36

0.718

2.33

3.10

0.75

0.455

No. failed AD trials in lifetime

0.25

0.36

0.69

0.495

-0.54

0.66

-0.82

0.415

Onset age of MDD *

0.03

0.05

0.62

0.534

-0.09

0.08

-1.05

0.299

No. previous depressive episodes *

-0.07

0.28

-0.25

0.803

-0.15

0.42

-0.37

0.714

Change to week 12

Gender

1.81

2.30

0.79

0.434

-3.76

3.16

-1.19

0.237

Age

-0.10

0.14

-0.73

0.468

0.11

0.26

0.42

0.675

Ethnicity (white/non-white)

-1.92

2.29

-0.84

0.403

6.29

3.94

1.60

0.114

Years of education

-0.04

0.20

-0.21

0.835

0.40

0.31

1.27

0.206

Unemployed (yes/no)

-2.36

2.80

-0.84

0.402

-1.53

4.53

-0.34

0.737

Wage (low/high)

-0.39

2.05

-0.19

0.851

-2.37

2.84

-0.83

0.407

Relationship (yes/no)

-1.25

2.11

-0.59

0.556

2.55

2.90

0.88

0.381

Phystical activity (low/high)

-0.76

2.86

-0.27

0.790

-0.56

3.78

-0.15

0.882

Hypertension

-0.18

2.09

-0.09

0.930

-0.01

2.91

0.00

0.997

Duration of current episode

-0.03

0.03

-1.05

0.299

0.03

0.04

0.70

0.488

Refractory depression (yes/no)

-3.56

2.01

-1.77

0.081

4.37

2.79

1.56

0.122

Current benzodiazepine use (yes/no)

-0.42

2.36

-0.18

0.859

0.83

3.58

0.23

0.818

Currently in psychotherapy (yes/no)

-5.00

2.49

-2.00

0.048

8.86

3.83

2.31

0.023

No. failed AD trials in lifetime

-0.27

0.41

-0.67

0.505

1.25

0.74

1.69

0.096

Onset age of MDD *

0.03

0.06

0.60

0.554

-0.05

0.09

-0.52

0.605

No. previous depressive episodes *

0.25

0.29

0.88

0.385

-0.40

0.43

-0.93

0.356

Predictor

Estimate

SE

t

P

Change to week 6

Gender

-1.11

2.64

-0.42

0.674

Age

-0.42

0.21

-2.02

0.046

Duration of current episode

0.03

0.04

0.74

0.464

Refractory depression (yes/no)

0.14

2.38

0.06

0.952

Current benzodiazepine use (yes/no)

1.59

3.61

0.44

0.661

Currently in psychotherapy (yes/no)

2.33

3.10

0.75

0.455

Change to week 12

Gender

-3.76

3.16

-1.19

0.237

Age

0.11

0.26

0.42

0.675

Duration of current episode

0.03

0.04

0.70

0.488

Refractory depression (yes/no)

4.37

2.79

1.56

0.122

Current benzodiazepine use (yes/no)

0.83

3.58

0.23

0.818

Currently in psychotherapy (yes/no)

8.86

3.83

2.31

0.023

Note: (*) indicates that data from the HERAH system is missing (33 of 108 patients).

Age at week 6

Psychotherapy at week 12

6.2 Predicting HDRS response

Predictor

Main effect

Interaction

Odds ratio

z

P

Odds ratio

z

P

Week 6

Gender

1.04

0.05

0.962

0.80

-0.22

0.829

Age

1.00

-0.09

0.932

0.98

-0.28

0.778

Ethnicity (white/non-white)

0.28

-1.83

0.067

3.42

1.03

0.302

Years of education

1.01

0.08

0.937

1.13

1.19

0.233

Unemployed (yes/no)

0.47

-0.67

0.506

1.06

0.04

0.970

Wage (low/high)

0.88

-0.19

0.852

0.40

-0.99

0.323

Relationship (yes/no)

0.92

-0.11

0.909

2.08

0.79

0.427

Phystical activity (low/high)

0.81

-0.24

0.814

2.10

0.63

0.532

Hypertension

5.31

2.22

0.026

0.25

-1.43

0.152

Duration of current episode

1.00

0.02

0.987

1.00

0.07

0.947

Refractory depression (yes/no)

0.64

-0.56

0.578

1.13

0.12

0.905

Current benzodiazepine use (yes/no)

0.26

-1.16

0.245

9.12

1.47

0.142

Currently in psychotherapy (yes/no)

0.51

-0.77

0.440

10.60

1.88

0.060

No. failed AD trials in lifetime

1.03

0.21

0.837

0.84

-0.70

0.487

Onset age of MDD *

1.01

0.45

0.653

0.96

-1.09

0.274

No. previous depressive episodes *

0.96

-0.27

0.788

1.01

0.04

0.967

Week 12

Gender

1.33

0.40

0.692

0.29

-1.26

0.209

Age

1.01

0.28

0.782

0.98

-0.28

0.778

Ethnicity (white/non-white)

0.45

-1.13

0.260

11.28

1.78

0.075

Years of education

0.96

-0.71

0.476

1.11

1.05

0.295

Unemployed (yes/no)

0.36

-0.90

0.370

0.77

-0.16

0.873

Wage (low/high)

0.71

-0.49

0.623

0.86

-0.17

0.862

Relationship (yes/no)

0.65

-0.62

0.538

2.21

0.87

0.385

Phystical activity (low/high)

0.49

-0.83

0.406

2.14

0.69

0.493

Hypertension

2.53

1.37

0.171

0.48

-0.81

0.417

Duration of current episode

0.94

-1.46

0.145

1.06

1.24

0.215

Refractory depression (yes/no)

0.25

-1.57

0.115

5.29

1.52

0.129

Current benzodiazepine use (yes/no)

0.29

-1.06

0.291

4.41

0.98

0.328

Currently in psychotherapy (yes/no)

0.24

-1.27

0.202

8.38

1.46

0.144

No. failed AD trials in lifetime

0.87

-0.83

0.407

1.35

1.15

0.251

Onset age of MDD *

1.02

1.09

0.276

0.97

-0.82

0.414

No. previous depressive episodes *

0.89

-0.71

0.478

1.06

0.31

0.753

Note: (*) indicates that data from the HERAH system is missing (33 of 108 patients).