In a recent randomized controlled trial published in The Lancet Digital Health, researchers investigated the efficacy of two versions of automated feedback following internet-based depression screening on the severity of depression.
They found that automated feedback did not significantly lower the severity of depression or lead to adequate depression care in individuals who were not previously diagnosed with depression but experienced it.
Study: The efficacy of automated feedback after internet-based depression screening (DISCOVER): an observer-masked, three-armed, randomised controlled trial in Germany. Image Credit: PeopleImages.com - Yuri A/Shutterstock.com
Background
Depressive disorders are highly disabling and prevalent but often remain undetected and untreated, leading to chronic conditions, treatment resistance, higher healthcare costs, and increased disease burden. Standardized depression screening, though debated, could potentially aid early detection.
Feedback on screening results might prompt individuals to recognize symptoms and seek help. Previous trials showed mixed results on depression severity but improved patient-physician communication and access to therapy.
In a pioneering study, termed the “DISCOVER” trial, researchers aimed to evaluate the efficacy of two forms of automated feedback following internet-based screening for moderate to severe depression, examining its effect on initiating evidence-based care, depression-related behaviors, and potential negative effects.
About the study
The present study was an observer-masked, randomized controlled trial with three arms conducted in Germany between 2021 and 2022. A total of 1,178 participants aged ≥18 years, with Patient Health Questionnaire-9 (PHQ-9) scores ≥10 (moderate depression severity), and no recent depression diagnosis or treatment were randomized in a 1:1:1 ratio.
Researchers compared the impact of automated tailored feedback (n = 394), automated non-tailored feedback (n = 393), and no feedback (n = 391) on depression severity six months after internet-based screening.
The no-feedback group received no further information post-screening. In contrast, participants in the two feedback groups had the option to access feedback immediately via a clickable link on the website.
The feedback content was developed collaboratively with individuals affected by depressive disorders. It consisted of four sections: 1) presenting the screening results, 2) encouraging consultation with a healthcare professional, 3) providing general information on depression, and 4) detailing treatment options based on German clinical guidelines.
The tailored feedback adapted content based on participants' symptom profiles, preferred specialist type, health insurance provider, symptom attributions, and local residency.
Across the three groups, the mean age was 37.1 years, 70% were women, 29% were men, 1% reported other genders, and 10% had a migrant background. The majority were well-educated (49%), single (41%), employed (72%), and lived in large cities (51%).
At the six-month follow-up, 965 participants provided PHQ-9 data. The primary outcome was the change in depression severity using the PHQ-9 scale six months post-randomization, assessing nine depressive symptoms on a scale from 0 to 3, with scores ranging from 0 to 27.
Secondary outcomes included receipt of evidence-based depression care, diagnosis of depressive disorder by healthcare professionals, engagement in depression-related health behaviors, health-related quality of life, anxiety severity, somatic symptom severity, and safety monitoring for participants with suicidal ideation.
Statistical analysis involved covariance, intention-to-treat analysis, per-protocol analysis, subgroup analysis, multiple imputations for missing data, Cohen's d calculation, and closed testing principle.
Results and discussion
Six months after random assignment, depression severity reduced similarly across groups: by 3.4 points in the no-feedback group, 3.5 points in the non-tailored feedback group, and by 3.7 points in the tailored feedback group, with no significant differences within the groups (p=0.72).
Secondary outcome analyses showed no significant intervention effects across groups. Negative effects were minimal (<1%), with isolated reports of emotional burden and distress related to trial participation.
The rates of major depressive disorder diagnosis based on SCID (short for structured clinical interview for DSM disorders) criteria and treatment initiation were found to be comparable among the groups. Sensitivity analyses were not found to alter the findings.
Overall, the trial revealed that while digital depression screening may identify undetected depression, it does not ensure evidence-based treatment, indicating the need for more effective strategies to facilitate access to care post-screening.
A large sample size strengthens the trial with a good follow-up rate and the ability to isolate the effects of screening and feedback, including untreated individuals with depression, representative recruitment, and diagnostic interviews for efficacy analysis.
However, the trial is limited by the absence of a no-screening control group, recruitment not explicitly targeting those seeking depression information, reliance on self-reported help-seeking data, potential self-selection bias, and possible influence of repeated assessments on depression outcomes.
Conclusion
In conclusion, the DISCOVER study shows that automated feedback after internet-based depression screening may not lower depression severity or trigger evidence-based care.
These findings should be considered by healthcare providers and inform guidelines for early depression detection, highlighting the need for further research to understand the path of patients from early detection to effective treatment.
Maternal stress and depression alter infant DNA, with potential lifelong impact