For many people living with depression, mental health apps promise accessible support at moments when face to face care feels out of reach. Yet new research suggests that those struggling the most may also be the most likely to stop engaging with these digital tools early on, raising questions about how effective such apps really are in everyday use. The findings were published in JMIR Human Factors.
The study analysed data from more than 2,000 adults who took part in a large remote trial of smartphone based mental health apps designed to reduce depressive symptoms. Participants were asked to regularly complete short questionnaires about mood, sleep, functioning, and perceived improvement over a 12 week period, allowing researchers to track engagement over time .
Only around half of participants completed at least one questionnaire beyond the initial assessment. Among those who did engage, completion rates varied widely, with many users responding to fewer than half of the questionnaires they received. This pattern reflects a long recognised problem in digital mental health research, where high dropout rates can distort findings and limit real world impact.
One of the clearest findings was that people who felt their mental health was improving were more likely to keep using the app and continue completing assessments. Participants who reported meaningful reductions in depressive symptoms tended to stay engaged for longer and respond more consistently over time. In contrast, those who felt little improvement, or worsening symptoms, were more likely to disengage early.
Interestingly, people who entered the study with more severe depressive symptoms were not less likely to stick with the app if they began to feel better. In fact, those showing improvement often started with higher levels of depression than those who did not improve. What mattered most for long term engagement was not how unwell someone was at the beginning, but whether they experienced positive change during the study.
Demographic factors also played a role. Engagement and adherence differed by age, income, education, and background. Older participants and those with higher education levels generally completed more questionnaires, while lower income and lower income satisfaction were linked to poorer retention. These patterns suggest that digital mental health apps may struggle to serve groups who already face barriers to care.
The type of app also influenced engagement. Participants using simpler apps with lower time demands were more likely to keep completing assessments than those using apps requiring frequent or intensive interaction. This points to a trade off between app complexity and sustained use, especially for people dealing with low motivation and fatigue.
Researchers warn that these engagement patterns may lead to overly optimistic conclusions about how well mental health apps work. If people who do not improve are more likely to drop out, the remaining data may exaggerate treatment benefits. This kind of attrition bias is difficult to detect but has important implications for how digital mental health tools are evaluated and promoted.
The findings highlight the need for app designs that adapt to users who are struggling, rather than losing them. Strategies such as early identification of disengagement, personalised prompts, and reduced burden during periods of worsening symptoms could help keep users involved. Without this, mental health apps risk benefiting only those who are already on a path to recovery.