Three hundred and fifty apps were screened for eligibility – representing the top 40 ranked apps in each search, except where fewer i OS apps were returned for schizophrenia, self-harm and substance use.Inter-rater reliability for the binary choice to include or exclude each app was measured using Cohen’s kappa at 0.78, suggesting moderate agreement.The types of functionality described by the apps are summarised in Table 2.
Twenty-one of these apps claimed both effectiveness and acceptability.
The most common form of effectiveness claim was related to improvements in knowledge or skills to support self-management (§3.a.iii, 26/73, 36%), closely followed by improvements in symptoms or mood (§3.a.ii, 22/73, 30%), with fewer apps claiming the ability to diagnose or detect a mental health condition (§3.a.i, 7/73, 10%).
A post-hoc analysis identified that five apps (5/73, 6.8%) mentioned research or clinical trials underway.
The second most common type of support was the description of technical expertise (§4.b, 23/73, 32%).
The format of this material is standardised for commercial app stores, consisting of a written app description and, optionally, screenshots or videos of app functions.
Within this restricted context, the extent to which scientific evidence is presented as a potential marker of quality for health apps is unclear.
The descriptions of the top-ranking, consumer-focused apps were coded to identify claims of acceptability and effectiveness, and forms of supporting statement.
For apps which invoked ostensibly scientific principles, a literature search was conducted to assess their credibility.
Following screening for eligibility and removal of duplicates across search terms and platforms, 76 platform-independent apps were retained for coding.
During the coding process, an additional three apps were identified as being targeted at clinicians or health professionals; excluding these apps resulted in 73 apps being retained for full coding.