Improving Adherence and Clinical Outcomes in Self-Guided Internet Treatment for Anxiety and Depression: Randomised Controlled Trial

The researchers tested whether adding automated, non-therapeutic emails to a self-guided internet-delivered cognitive behavioural therapy (CBT) program—called the "Wellbeing Course"—improved adherence and clinical outcomes compared to the same course without emails, and compared to a waitlist control group.

**Intervention groups:**

**Treatment Group 1 (Course + Emails):** Participants completed the 8-week Wellbeing Course and received automated emails at regular intervals (frequency and content described below).

**Treatment Group 2 (Course Only):** Participants completed the same 8-week Wellbeing Course but received no automated emails.

**Waitlist Control Group:** Participants received no treatment during the 8-week study period and were offered the course after the study ended.

**The Wellbeing Course itself** was a transdiagnostic (i.e., designed to treat both anxiety and depression simultaneously) self-guided online program. It consisted of:

5 core online lessons delivered weekly (weeks 1–5)

3 consolidation weeks (weeks 6–8) with no new content

Each lesson included: psychoeducation, CBT skills training (e.g., cognitive restructuring, behavioural activation, graded exposure), and downloadable homework assignments

No therapist contact or guidance was provided

**The automated emails** were sent at specific time points:

Welcome email immediately after enrolment

Lesson reminder emails before each new lesson was released

Encouragement emails if participants had not logged in for 7 days

Mid-course check-in email at week 4

Post-treatment email at week 8

The emails contained no therapeutic content—only reminders, encouragement to continue, and links to the course website. They were fully automated and identical for all recipients.

**Primary outcome measures:**

**Depression severity:** Patient Health Questionnaire 9-Item (PHQ-9, range 0–27, higher = more severe)

**Anxiety severity:** Generalized Anxiety Disorder 7-Item (GAD-7, range 0–21, higher = more severe)

**Secondary outcome measures:**

**Psychological distress:** Kessler Psychological Distress Scale (K-10, range 10–50)

**Disability:** Sheehan Disability Scale (SDS, range 0–30)

**Course completion:** Defined as completing all 5 lessons and all post-treatment questionnaires

**Adherence:** Number of lessons completed, time spent on course, number of logins

**Treatment satisfaction:** 7-item Client Satisfaction Questionnaire (CSQ-8, range 8–32)

**Negative effects:** Open-ended question about any worsening of symptoms

**Sample size:** 257 participants randomised (86 to Course + Emails, 86 to Course Only, 85 to Waitlist)

**Age:** Mean age 42.4 years (SD = 12.5, range 18–76)

**Gender:** 73% female (187 women, 70 men)

**Location:** Australia, recruited via online advertisements and community outreach

**Inclusion criteria:**

- Aged 18 years or older

- Residing in Australia

- Elevated symptoms of anxiety and/or depression (scoring ≥5 on either PHQ-9 or GAD-7)

- Access to internet and email

- Willing to provide contact details and a general practitioner's name

**Exclusion criteria:**

- Current severe depression (PHQ-9 total >23 or score >2 on item 9 about suicidal thoughts)

- Current or recent (within 6 weeks) psychological treatment

- Current or recent (within 6 weeks) change in psychotropic medication

- Psychosis or bipolar disorder

- Substance dependence

**Baseline severity:** Mean PHQ-9 = 11.2 (moderate depression), mean GAD-7 = 10.1 (moderate anxiety). Approximately 60% met criteria for a probable anxiety or depressive disorder based on diagnostic questionnaires.

All measures were self-reported via online questionnaires at three time points:

**Pre-treatment (baseline):** Before randomisation

**Post-treatment (week 8):** Immediately after the 8-week course ended

**3-month follow-up:** Only for the two treatment groups (waitlist was offered treatment after post-treatment)

**Primary outcome measures:**

**PHQ-9:** 9 items measuring DSM-IV depression symptoms over the past 2 weeks. Each item scored 0–3. Cut-offs: 5–9 = mild, 10–14 = moderate, 15–19 = moderately severe, ≥20 = severe. Reliable change defined as a reduction of ≥5 points.

**GAD-7:** 7 items measuring GAD symptoms over the past 2 weeks. Each item scored 0–3. Cut-offs: 5–9 = mild, 10–14 = moderate, ≥15 = severe. Reliable change defined as a reduction of ≥4 points.

**Secondary outcome measures:**

**K-10:** 10 items measuring non-specific psychological distress over the past 4 weeks. Score range 10–50. Higher scores = greater distress.

**SDS:** 3 items measuring impairment in work, social life, and family life. Each item scored 0–10. Total range 0–30.

**Course completion:** Binary variable (yes/no) based on completing all 5 lessons and all post-treatment questionnaires.

**Adherence metrics:** Number of lessons completed (0–5), number of logins, total time spent on the course (minutes).

**CSQ-8:** 8 items measuring satisfaction with treatment. Score range 8–32. Higher = more satisfied.

**Negative effects:** Open-ended question: "Did you experience any worsening of your symptoms during the course? If yes, please describe."

**Study design:** This was a three-arm randomised controlled trial (RCT) with parallel groups. Participants were randomly allocated in a 1:1:1 ratio to Course + Emails, Course Only, or Waitlist Control.

**Randomisation:**

Computer-generated random number sequence using a randomisation algorithm built into the study website

Allocation was concealed from participants until after they completed baseline questionnaires

The research team was not blinded to group allocation (open-label design)

**Blinding:**

**Participants:** Not blinded—they knew whether they were receiving emails or not, and waitlist participants knew they were on a waitlist

**Researchers:** Not blinded—they knew group assignments for data analysis

**Outcome assessors:** Not applicable (all outcomes were self-reported online)

**Duration:**

Active treatment period: 8 weeks (5 weeks of lessons + 3 consolidation weeks)

Follow-up: 3 months post-treatment (12 weeks after the 8-week assessment)

Total study duration per participant: approximately 20 weeks (8 weeks treatment + 12 weeks follow-up)

**Statistical approach:**

Primary analysis: Intention-to-treat (ITT) using linear mixed models (LMM) to handle missing data

Between-group effect sizes reported as Cohen's d (0.2 = small, 0.5 = medium, 0.8 = large)

Reliable change indices calculated for PHQ-9 and GAD-7

Completer analyses also reported (participants who finished all assessments)

Subgroup analysis: Participants with "elevated symptoms" defined as PHQ-9 ≥10 or GAD-7 ≥10 at baseline

**What this design can and cannot prove:**

**Can prove:**

Causal effects of the automated emails on adherence and outcomes (because randomisation balances known and unknown confounders between groups)

The magnitude of benefit attributable specifically to the emails (by comparing Course + Emails vs. Course Only)

The overall efficacy of the Wellbeing Course vs. no treatment (by comparing both treatment groups vs. waitlist)

**Cannot prove:**

Why the emails worked—was it reminders, encouragement, or simply increased awareness of the course? The design cannot separate these mechanisms

Long-term durability beyond 3 months (follow-up was only 3 months)

Generalisability to non-internet users, people without elevated symptoms, or clinical populations with severe disorders (excluded)

Whether the emails would work with other types of self-guided treatments (different content, different platforms)

Whether the emails would work without the specific structure of the Wellbeing Course (weekly lessons, downloadable materials)

**Major methodological weaknesses:**

**No blinding:** Participants knew their group assignment, which could inflate placebo effects in the email group (they might expect more benefit from a "more supported" program)

**Self-report only:** No clinician-administered diagnostic interviews or objective adherence measures (e.g., server logs were used but only for login counts)

**High attrition:** At post-treatment, 67% of Course + Emails, 56% of Course Only, and 82% of Waitlist completed questionnaires. Differential attrition (more dropouts in Course Only) could bias results

**No active control for emails:** The Course Only group received no contact at all, so the email effect could be partly due to increased attention rather than the specific content of the emails

**Subgroup analysis was post-hoc:** The finding that emails improved outcomes only in those with elevated symptoms was not pre-specified and should be interpreted cautiously

**Primary outcomes (PHQ-9 and GAD-7 at post-treatment):**

**Depression (PHQ-9):**

- Course + Emails: Mean reduction of 5.8 points (from 11.4 to 5.6)

- Course Only: Mean reduction of 4.1 points (from 11.0 to 6.9)

- Waitlist: Mean reduction of 1.2 points (from 11.2 to 10.0)

- Between-group comparison (Course + Emails vs. Waitlist): Cohen's d = 0.74, p < 0.001

- Between-group comparison (Course Only vs. Waitlist): Cohen's d = 0.52, p < 0.01

- Between-group comparison (Course + Emails vs. Course Only): Cohen's d = 0.22, p = 0.14 (not statistically significant)

**Anxiety (GAD-7):**

- Course + Emails: Mean reduction of 4.8 points (from 10.2 to 5.4)

- Course Only: Mean reduction of 3.9 points (from 10.0 to 6.1)

- Waitlist: Mean reduction of 1.0 points (from 10.1 to 9.1)

- Between-group comparison (Course + Emails vs. Waitlist): Cohen's d = 0.72, p < 0.001

- Between-group comparison (Course Only vs. Waitlist): Cohen's d = 0.53, p < 0.01

- Between-group comparison (Course + Emails vs. Course Only): Cohen's d = 0.19, p = 0.20 (not statistically significant)

**Secondary outcomes:**

**Psychological distress (K-10):**

- Course + Emails: Mean reduction of 6.2 points

- Course Only: Mean reduction of 4.5 points

- Waitlist: Mean reduction of 1.1 points

- Course + Emails vs. Waitlist: d = 0.65, p < 0.001

- Course + Emails vs. Course Only: d = 0.24, p = 0.11

**Disability (SDS):**

- Course + Emails: Mean reduction of 3.8 points

- Course Only: Mean reduction of 2.9 points

- Waitlist: Mean reduction of 0.5 points

- Course + Emails vs. Waitlist: d = 0.55, p < 0.01

- Course + Emails vs. Course Only: d = 0.16, p = 0.30

**Adherence and completion:**

**Course completion rate:** 58% in Course + Emails vs. 35% in Course Only (χ² = 8.2, p = 0.004)

**Mean number of lessons completed:** 4.1 (out of 5) in Course + Emails vs. 3.2 in Course Only (p < 0.01)

**Mean number of logins:** 12.4 in Course + Emails vs. 8.9 in Course Only (p < 0.01)

**Mean time spent on course:** 142 minutes in Course + Emails vs. 108 minutes in Course Only (p = 0.06, not significant)

**Subgroup analysis (participants with elevated symptoms at baseline, PHQ-9 ≥10 or GAD-7 ≥10):**

Among this subgroup (n = 195), the email effect was larger:

- Course + Emails vs. Course Only for PHQ-9: d = 0.35, p = 0.04

- Course + Emails vs. Course Only for GAD-7: d = 0.31, p = 0.06 (trend)

This suggests the emails were particularly helpful for those with more severe symptoms

**Reliable change (clinically meaningful improvement):**

PHQ-9: 58% of Course + Emails achieved reliable improvement vs. 44% of Course Only vs. 18% of Waitlist

GAD-7: 55% of Course + Emails achieved reliable improvement vs. 42% of Course Only vs. 15% of Waitlist

**Treatment satisfaction (CSQ-8):**

Course + Emails: Mean 26.4 (out of 32)

Course Only: Mean 25.1

Difference not statistically significant (p = 0.12)

**Negative effects:**

4% of participants reported worsening of symptoms (no difference between groups)

No serious adverse events reported

**3-month follow-up (treatment groups only):**

Gains were maintained in both groups

No significant differences between Course + Emails and Course Only at follow-up

PHQ-9: Course + Emails = 5.2, Course Only = 6.1 (d = 0.18, p = 0.30)

GAD-7: Course + Emails = 5.0, Course Only = 5.8 (d = 0.16, p = 0.35)

**In plain English:**

**Depression:** The Course + Emails group improved by about 6 points on the PHQ-9 (from moderate depression to mild depression). This is roughly equivalent to moving from "feeling depressed nearly every day" to "feeling depressed only several days a week." The Course Only group improved by about 4 points—still meaningful, but less. The email group's improvement was about 50% larger than the no-email group.

**Anxiety:** The email group improved by about 5 points on the GAD-7 (from moderate anxiety to mild anxiety). The no-email group improved by about 4 points. Again, the email group did about 25% better.

**Adherence:** The most striking effect was on completion rates. Adding emails nearly doubled the proportion of people who finished the course (58% vs. 35%). This means that for every 4 people who received emails, 1 additional person completed the course compared to those who didn't get emails.

**Practical significance:** A 5-point drop on the PHQ-9 is considered a clinically meaningful improvement. The email group achieved this on average, while the no-email group came close