Table 1.
Summary of included randomized controlled trials of Internet interventions for anxiety and depression
| Sample size N = total I = Intervention C = Control |
Dropout N = total I = Intervention C = Control |
Adherence to treatment | Predictors of dropout/adherence | Self-reported reason for dropout? | Type of statistical analysis: ITT, NMAR, MAR, LOCF |
|
| Depression | ||||||
| Andersson et al 2002 [23] (6 modules) |
N = 117 I = 53 C = 64 |
Post-treatment: N = 66 (56.4%) I = 23 (43.4%) C = 43 (67.2%) 1 year: N = 96 (82.1%) I = 46 (86.8%) C = 50 (78.1%) |
Not reported. Mean posting on discussion board = 8.7 (SD = 21.5) | Response rate higher in control group at post-treatment. | No formal measure described. Reported reasons: lack of time, programme too fast, lack of ideal environment to complete programme, programme is impersonal and too extensive. | Completers. Multiple regression |
| Andersson et al 2005 [24] (5 modules + discussion group) |
N = 117 I = 57 C = 60a |
Post-treatment: N = 85 (72.6%) I = 36 (63.1%) C = 49 (81.6%) 6 months: N = 71 (60.7%) I = 36 (63.1%) C = 35 (58.3%)b |
Mean number of modules completed = 3.7 out of 5 (SD = 1.9) Total postings on the discussion board: I = 233 C = 842 (C > I, P < .05) |
Lower withdrawal for control than treatment group participants at 3 months. (100% intervention completed; 71% Control group completed). No significant differences in depressive symptoms (BDI) or age, gender, educational level, place of living, or quality of life between dropouts and completers at 3 months. |
No formal measure described. Main reported reason: treatment was too demanding. | ITT LOCF, ANOVA |
| Christensen et al 2004 [6] Griffiths et al 2004 [25] Mackinnon et al 2008 [26] (12 month follow-up) (5 modules) |
N = 525 I(i) = 165 I(ii) = 182 C = 178 I(i) = BluePages depression informationI(ii) = MoodGYMc CBT |
Post-treatment: N = 435 (82.8%) I(i) = 140 (84.8%) I(ii) = 136 (74.7%) C = 159 (89.3%) 6 months: N = 352 (67%) I(i) = 115 (69.6%) I(ii) = 106 (58.2%) C = 131 (73.6%) 12 months: N = 325 (61.9%) I(i) = 107 (64.8%) I(ii) = 94 (51.6%) C = 124 (69.6%) |
Mean BluePages visits = 4.49 (SD = 1.4) Mean MoodGYM exercises completed = 14.8 (SD = 9.7) (51%) |
Greater dropout for MoodGYM (CBT) than BluePages (depression information) (P = .0001) Baseline depressive symptoms (CES-D) and knowledge of psychological treatments lower among dropouts (P < .01) Males more likely to be lost to follow-up at 12 months. |
No | ITT LOCF |
| Clarke et al 2002 [27] (7 content chapters) |
N = 299 I = 144 C = 155 |
4 weeks: N = 158 (52.8%) 8 weeks: N = 195 (65.2%) 16 weeks: N = 196 (65.6%) 32 weeks: N = 177 (59.2%) |
Not reported. Mean logons: I = 2.6 (SD = 2.5; range 1-20) | Baseline depressive symptoms (CES-D) lower in those who completed at least one follow-up questionnaire (P < .05). Age, gender, recruitment group did not predict dropout. | No | ITT random effect regression analyses |
| Clarke et al 2005 [28] (7 content chapters) |
N = 255 I(i) = 75 I(ii) = 80 C = 100 I(i) = website + postcard remindersI(ii) = website + telephone reminders |
5 weeks: N = 164 (63.1%) I(i) = 36 (48%) I(ii) = 48 (60%) C = 77 (77%) 10 weeks: N = 173 (67.8%) I(i) = 43 (57.3%) I(ii) = 50 (62.5%) C = 80 (80%) 16 weeks: N = 169 (66%) I(i) = 46 (61.3%) I(ii) = 43 (53.8%) C = 80 (80%) |
Not reported. Mean logons: I(i) = 5.9 (SD = 6.2; range 1-33) I(ii) = 5.6 (SD = 5.8; range = 1-27) |
Baseline depressive symptoms (CES-D) and age lower in those who completed at least one follow-up questionnaire (P < .05). Gender not a predictor. Control participants more likely to complete a follow-up assessment. Mean logons did not differ between postcard & telephone reminder conditions (p > .05). |
No | ITT random effect regression analyses – REML |
| Patten 2003 [29] (4 content modules) |
N = 786 I = 420 C = 366 |
1 month: I = 418 (99.5%) C = 363 (99.2%) 2 months: I = 412 (98.1%) C = 361 (98.6%) 3 months: I = 406 (96.7%) C = 358 (97.8%) |
Not reported. Mean duration signed-on = 50 min | None reported. | No | Completers |
| Spek et al 2007 [30] Spek et al 2008 [31] (12 month follow-up) I(i) = 8 modules I(ii) = 10 sessions |
N = 301 I(i) = 102 I(ii) = 99 C = 100 I(i) = Internet CBTI(ii) = Group CBT |
Post-treatment: N = 181 (60.1%) I(i) = 67 (65.7%) I(ii) = 56 (56.6%) C = 58 (58%) 12 months: N = 190 (63.1%) I(i) = 58 (56.8%) I(ii) = 66 (66.6%) C = 66 (66%) |
Mean modules/sessions completed: I(i) = 5.5 out of 8 (78.1%) I(ii) = 9.1 out of 10 (98.3%) Completed whole course: I(i) = 48.3% I(ii) = 94.5% |
Less treatment completion in Internet intervention group. | No formal measure described. Main reason reported: lack of time. |
ITT MI |
| Warmerdam et al 2008 [32] (I(i) = 9 lessons, I(ii) = 5 lessons) |
N = 263 I(i) = 88 I(ii) = 88 C = 87 I(i) = Cognitive Behavioral Therapy (CBT)I(ii)=Problem Solving Therapy (PST) |
5 weeks: N = 184 (69.9%) I(i) = 61 (69.3%) I(ii) = 52 (59.1%) C = 71 (81.6%) 8 weeks: N = 173 (65.8%) I(i) = 51 (57.9%) I(ii) = 51 (57.9%) C = 71 (81.6%) 12 weeks: N = 151 (57.4%) I(i) = 46 (52.2%) I(ii) = 42 (47.2%) C = 63 (72.4%) |
Completed at least 1 module: I(i) = 80 (90.9%) I(ii) = 74 (84.1%) Completed at least 3-4 lessons: I(i)=63 (71.6%) I(ii) = 49 (55.7%) Completed whole course: I(i) = 34 (38.6%) I(ii) = 33 (37.5%) |
Lower withdrawal in control group compared with both intervention groups. Participants who completed post-treatment measures more likely to be born in the Netherlands and older. |
No formal measure described. Reported reasons: other treatment; feeling better; lack of time; and problems understanding the program. | ITT LLM using REML |
| Depression, anxiety and stress | ||||||
| van Straten et al 2008 [33] (4 modules) |
N = 213 I = 107 C = 106 |
Post-treatment: N = 177 (83.1%) I = 81 (76%) C = 96 (91%) |
Completed 1 module = 97 (90.6%) Completed 2 modules = 79 (73.8%) Completed 3 modules = 70 (65.4%) Completed whole course = 59 (55.1%) |
Post-treatment measure response rate higher among more educated participants and those without alcohol problems. Married participants more likely to complete the intervention. |
No | ITT MI |
| Generalised anxiety disorder | ||||||
| Kenardy et al 2003 [34] (6 modules) Kenardy et al 2006 [35] (6 month follow-up) |
N = 83 I = 43 C = 40 |
Post-treatment N = 75 (90.4%) I = 37 (86%) C = 38 (95%) 6 months N = 42 (50.6%) I = 19 (44.2%) C = 23 (57.5%) |
Average modules completed = 3.33 out of 7 (SD = 2.10). Mean logons = 7.76 (SD = 7.31). Mean access time = 90.37 minutes (SD = 111.29). |
Baseline depressive symptoms (CES-D), anxiety sensitivity (ASI) lower among completers than dropouts. At 6 months: No differences between those who dropped out in this period and those who did not. |
No formal measure described. Main reason reported: time constraints. At 6 months: No reasons for additional dropout between post-test and 6 months reported. |
Completers. Excluded outlier (high post test results in the intervention group; n = 1). |
| Panic disorder | ||||||
| Carlbring et al 2001 [36] (6 modules) |
N = 41 I = not reported C = not reported |
Post-treatment: N = 36 (87.9%) I = 4 dropouts C = 1 dropout |
Completed all modules: 100% (excluding participants who dropped out) |
None reported. | No formal measure described. Reported reasons: I = lack of time (n = 3); serious physical illness (n = 1). C = no reason given. |
ITT LOCF |
| Carlbring et al 2006 [37] (10 modules) |
N = 60 I = 30 C = 30 |
Post-treatment: N = 57 (95%) I = 28d (93.3%) C = 29 (96.6%) 9 months: I = 26 (86.6%) C = not collected |
Completed all modules = 24 (80%); Mean number of modules completed = 8.9 (SD = 2.6). One participant completed 0 modules. |
None reported. | No formal measure described. Reported reason: shortage of time (n = 1). |
ITT LOCF |
| Klein and Richards, 2001 [38] | N = 23 I = 11 C = 12 |
Post-treatment: N = 22 (95.7%) C = not reported I = not reported |
Not reported. | None reported. | No | Completers |
| Klein et al 2006 [39] (6 modules) |
N = 55 I(i) = 19 I(ii) = 18 C = 18 I(i) = Online CBTI(ii) = Manualized CBT |
Post-treatment: N = 46 (83.6%) I(i) = 18 (94.7%) I(ii) = 15 (83.3%) C = 13 (72.2%) |
Those lost to follow-up did not complete the intervention. | Condition did not affect attrition. | No formal measure described. Reported reasons: I(i) = bipolar disorder episode (n = 1). I(ii) = depressive episode ( n = 1); treatment perceived to be ineffective (n = 1); lack of motivation (n = 1). C = monitoring led to recurrence of ‘bad’ memories (n = 1); no reason given (n = 4). |
ITT LOCF |
| Richards et al 2006 [40] | N = 32 I(i) = 12 I(ii) = 11 C = 9 I(i) = Online CBTI(ii) = Online CBT + stress management |
Post-treatment: N = 27 (84.4%) I(i) = 10 (83.3%) I(ii) = 10 (90.9%) C = 7 (77.8%) |
Not reported. | Completers frequency of emails I(i) = 15.3 (SD = 12.8) I(ii) = 11.6 (SD = 13.3) |
No formal measure described. Reported reasons: I(i) = lack of motivation, episode of depression I(ii) = wish to commence SSRI C = no reason given. |
ITT LOCF |
| Social phobia | ||||||
| Andersson et al 2006 [41] | N = 64 I = 32 C = 32 |
Post-treatment: N = 62 (96.9%) I = 30 (93.8%) C = 32 (100%) 12 months: N = 49 (76.6%) I = 29 (90.6%) C = 20 (62.5%) |
Completed all modules = 20 (62.5%) Mean modules completed = 7.5 (SD = 2.4) |
None reported. | No formal measure described. Reported reason: lack of time. | ITT LOCF |
| Carlbring et al 2007 [42] (9 modules) |
N = 57 C = 30 I = 30 |
Post-treatment: N = 55 (96.5%) C = 28 (93.3%) I = 28 (93.3%) 12 months: I = 27 (90%) C = Not collected |
Completed whole course = 27 (93.1%) Completed 4 modules = 1 (3.4%) Completed 1 module = 1 (3.4%) |
None reported. | No formal measure described. Reported reasons for dropout: I = began other therapy (n = 1); No computer access (n = 1) C = began other therapy (n = 1) Reasons for not completing treatment: lack of time |
Analysis excluded two participants after randomization but included two partially treatment compliant participants and one participant who did not return post-survey using LOCF. |
| Titov et al 2008 [43] (6 modules) |
N = 105 I = 50 C = 55 |
Post-treatment: N = 93 (88.6%) I = 44 (88%) C = 49 (89.1%) |
39 (78%) completed whole course | None reported | No formal measure described. Reported reasons: lack of time and motivation (n = 2); exposure too anxiety provoking (n = 1); programme not helpful (n = 1); overseas holiday (n = 1); change in work or study commitments (n = 3); medical complications (n = 1); no reason (n = 2) | ITT LOCF |
| Titov et al 2008 [44] | N = 88 I = 43 C = 45 |
Post-treatment: N = 78 (88.6%) I = 38 (88.4%) C = 40 (88.8%) |
33 (73.3%) completed whole course Mean modules completed: 5.5 out of 6 |
None reported. | No formal measure described. Reported reasons: programme not helpful ( n =1); symptoms improved significantly (n = 1) | ITT LOCF |
| Post traumatic stress disorder | ||||||
| Hirai and Clum 2005 [45] | N = 36 I = 18 C = 18 |
Post-treatment: N = 27 (75%) I = 13 (72.2%) C = 14 (77.8%) |
Not reported. | No demographic differences were found between completers and those who dropped out. | No | Completers |
| Knaevelsrud et al 2007 [46] (10 sessions) |
N = 96 I = 49 C = 47 |
Post-treatment: N = 87 (90.6%) I = 41 (83.7%) C = 46 (97.9%) 3 months: I = 41 (83.7%) C = not assessed |
Not reported | None reported | No formal measure described. Reported reasons include: technical problems (with network and computer) and emotional distress. | ITT LOCF |
| Lange et al 2001 [47] | N = 30 I = 15 C = 15 |
Post-treatment: N = 25 (83.3%) C = 12 (80%) I = 13 (86.7%) |
Not reported. | Participants who dropped out showed lower baseline intrusion scores (Impact of Events scale). | No formal measure described. Reported reasons: No quiet place for writing; could not focus on one trauma; ceased studies; marked improvement so saw no value in continuing. | Completers |
| Lange et al 2003 [48] | N = 184 I = 122 C = 62 |
Post-treatment: N = 101 (54.9%) I = 69 (56.6%) C = 32 (51.6%) 6 weeks: I = 57 (46.7%) C = not collected |
Completed treatment = 78 (63.9%) | Compliance with treatment higher for women, for older people, for those who lived with a partner, those less experienced with a computer. Education, time since trauma, amount disclosed about trauma, and psychological functioning did not predict adherence. Compliance with protocol was not predicted by any of the variables investigated. |
Formal questionnaire administered. Reasons for dropout: Technical problems with computer (n = 18, 41%) Preference for face-to face contact (n = 13, 29.5%) Burden of writing about stressful events (n = 13, 29.5%) 6 weeks Reported reasons: failure to respond; sought ‘other treatment’; did not wish to wait. |
Completers |
Note: ITT = Intention to treat; NMAR = Not Missing at Random; MAR = Missing at Random; LOCF = Last Observation Carried Forward; REML = Restricted Maximum Likelihood Estimation; LLM = Linear Mixed Modelling; MI = Multiple Imputation using NORM procedure in statistical package R; CBT = cognitive behavioral therapy.
aControl involved an online discussion group.
bhad received intervention at 3 months.
cThe same website can be offered both as open access site directly to the community or as a Web-based intervention offered in a randomized controlled trial.
dIn contrast to the authors of some papers, the dropout rate is calculated strictly using the number randomized as the denominator. Hence figures may differ from those reported by authors in some cases (e.g., Carlbringet al 2007 [42]).