In a large-scale evaluation of 422 evidence-based psychological interventions for severe mental disorders, studied in 260 randomised trials, sourced from six large NMAs, we could retrieve a detailed intervention description, operationalised as a protocol or manual, for about 86% of all active psychological interventions. Considering publicly available protocols or manuals only, this rate was reduced by more than three times, at 25%. Around 45% the protocols or manuals were only available commercially (as pay-walled publications or books that could be purchased). The retrieval rate is based on a resource-intensive, multipronged approach, which involved searching multiple sources (papers, trials registries, Google queries for commercially available manuals), as well as extensive author contact. Only 8% of the trials were associated with a protocol (published separately or as supplement), describing the intervention in detail. Under 30% of the trials were registered, and in only 5% of the trials did the registration contain detailed intervention descriptions. Author contact led to the retrieval of an additional 43 protocols, amounting to 18% of the trials with no protocol previously identified from reports and trial registries and 16% of the total psychological intervention arms. Of the 450 authors contacted, we received no reply, despite reminders, for 42% of the trials. For another 30% of trials, authors declined to provide protocols or manuals, frequently suggesting we purchase them, though we had indicated our use was exclusively for research. For 5% of the trials, we could not identify contact data. Therefore, author contact led to positive, helpful replies for less than a quarter of the trials.

We selected interventions studied in RCTs included in NMAs. These allow for the combination of large collections of trials, even for interventions studied infrequently, and are considered the highest level of evidence in treatment guidelines.12 Consequently, we can be reasonably certain that our analysis covered most evidence-based interventions for these disorders. Moreover, we considered severe mental disorders, for which there are often multiple bespoke interventions, based on distinct theoretical models and often tested in few trials. Some of these interventions have specific components, sometimes not found in other treatments packages.

Like drugs, psychological interventions are used to treat mental disorders. Protocols and manuals could be viewed as analogous to the package inserts or labels in drug regulation, where the ingredients, indications, usage, dose, administration and other details are specified. Fourteen percent of the active psychological intervention arms were either non-manualised or the manual was inaccessible (unpublished or otherwise unretrievable). For these, identifying active ingredients or otherwise reproducing the intervention is not possible. Our retrieval rate of 86% of all protocols and manuals was made possible through a time- and resource-intensive activity, entailing an almost year-long effort by a team of seven researchers. A quarter of all psychological intervention arms had publicly available protocols or manuals. Contacting authors of trials testing the interventions, the most accessible retrieval method after public availability, produced modest results. Only 43 additional protocols were sent by authors, amounting to only 16% of all active intervention arms. Most authors were either not available or not willing to share, though it is possible they would have been more forthcoming in sharing protocols for clinical purposes. Therefore, reliance on open-access resources and sharing from authors would enable access to about 40% of trial protocols or manuals.

These findings are in stark contrast with recent calls for making manuals for psychological interventions freely available, particularly as many interventions were developed with public funds.21 Public access to treatment manuals would greatly aid dissemination, particularly in low-resourced settings, where access to and uptake of psychological treatments are woefully insufficient. For example, for schizophrenia, there has been limited implementation of psychosocial interventions, with a median treatment gap of 69%, reaching 89% in low-income countries.22 Conversely, commercialisation of psychotherapy manuals through publishers could be seen as ensuring better distribution, thus aiding implementation and use. Estimating the cost of psychotherapy protocols and manuals, which would involve retrieving prices from various publishers and aggregate distributors like Amazon, and comparing these to those of other treatments, like medication, was beyond the scope of the current work. Such estimates would also not consider the cost of specialised training associated with many of the treatment manuals. However, regardless of the exact amount, the average cost will likely still be too high for low-income and middle-income countries, where most people with mental disorders are located.21 It was recently argued that treatment manuals for effective psychological treatments should be free or affordable, similarly to the drugs included in the WHO Essential Medicines List (WHO-EML). It is unclear what proportion of all available drugs are on the WHO-EML, with one over 30-year-old analysis estimating 16%.23 Therefore, it is difficult to benchmark estimates of freely accessible psychological interventions against those for drugs. However, for some disorders very few protocols or manuals were publicly available, as for example, just one for bulimia, five for anorexia and eight manuals for borderline personality disorder.

Without access to protocols or manuals, at least for explicit research purposes such as identifying active ingredients in the DECOMPOSE project, the content of psychological interventions is limited to descriptions in trial reports. If these descriptions are not complete or at similar levels of detail, some components might be inadvertently merged (eg, thought monitoring and cognitive restructuring) or missed entirely. This would severely constrain the potential of meta-analyses examining the differential efficacy of treatment components (‘component network meta-analyses’).24 Completeness of reporting for psychological interventions, for example, as assessed with the Template for Intervention Description and Replication (TIDieR),25 was rarely examined. However, the few existent analyses showed inadequate reporting for interventions (including psychological) for alcohol use disorders26 and for psychological interventions for pain after knee replacement.27 We could not identify any analysis of completeness of intervention reporting for other mental disorders. Similarly, examinations of the availability of treatment manuals for psychological interventions have been quasi non-existent, except for a small analysis of 27 trials set in low-income and middle-income countries.10 Of the 19 trials that reported using a manual, only 8 were referenced in the bibliography and only 2 were publicly available. However, the authors did not attempt to also retrieve manuals, as we have done here. One planned scoping review28 aims to characterise book-based psychotherapy manuals published up to 2022, but no results have been reported yet.

Our study has several limitations. First, we only focused on severe mental disorders, for which there are often many distinct psychological interventions, tested in few trials. Therefore, it is more likely that developers would be protective of the manuals and that there are fewer similar interventions, with overlapping components. Similar evaluations could be conducted for more common mental disorders, like depression or anxiety, where protocols are tested in more trials and where distinct protocols often share multiple components. Second, we did not contact all authors involved in the trial. However, given our extensive procedure, it is very likely we reached the lead authors or treatment developers. Third, we did not organise results by unique psychological intervention arms, as many trials described changes and adaptations of existing manuals, often making it challenging to ascertain if the treatment components remained unchanged. However, when the same intervention was used in more trials and the same manual was referenced, we considered it accessible if we had already retrieved it for another trial. Fourth, we did not check whether other entities, such as funders, could have retained and publicly shared a copy of the intervention protocol.

Overall, our findings underscore the challenges of accessing detailed descriptions of psychological interventions, beyond what is reported in the paper, which in turn hampers examinations of treatment components. Developing psychological treatments and evaluating these in RCTs need to be complemented by a more streamlined and less taxing access to treatment components and other characteristics (eg, delivery modes, training required), similarly to the labels of approved drugs. Furthermore, improving dissemination of effective psychological treatments, acknowledged as a global mental health priority,29 requires public availability of a significant proportion of treatment manuals. For research purposes such as the DECOMPOSE project, funders could consider mandating sharing of treatment protocols. Some medical journals such as The Lancet and JAMA family journals, the New England Journal of Medicine, Plos Medicine or Nature Medicine currently mandate the inclusion of trial protocols for published randomised trials, a policy that should be generalised across journals. Trial registries could also require the inclusion of the trial protocol. However, it would be important to ensure that intervention descriptions in trial protocols are exhaustive. More systematic investigations into the completeness of reporting of psychological interventions for mental disorders, for example by applying TIDieR, could provide an estimate of the feasibility of identifying active ingredients based on what is reported in the paper or in trial protocols shared with publications. Finally, systematic investigations of the accessibility of treatment manuals and protocols for other mental disorders are necessary to establish the proportion of evidence-based psychological interventions described with sufficient detail to enable identification of active ingredients.