This paper allows reflection on the process of recruiting, training and supervising non-specialized health workers to deliver technology-based task-shifting interventions which have been recognized as an important research gap [18]. The knowledge acquired with these experiences can be a great asset to implementation planning and future research.
Recruitment
In Brazil, primary care was previously composed of personnel with no formal training, but in a journey of healthcare improvement, mandatory qualifications were introduced increasing the number of NAs available [19]. In Sao Paulo, this opened up the possibility of task-shifting using existing NAs who play a key role in primary care services. In Lima, nurses usually deliver health programs targeting specific populations, and so they were chosen under the assumption that if CONEMO was effective and not too onerous, it could be included as an additional program, making the intervention sustainable [17]. However, as existing nurses reported being overburdened, additional nurses were hired for the purpose of the trial. Both approaches were successful in terms of tasks completion and supervision attendance, which highlighted that local service context and available resources should steer recruitment.
In terms of experience, a significant proportion of NAs in Sao Paulo had worked in the FHU region for a long time. This meant they were closer to the participants than other health professionals and were aware of their medical and personal histories. It was hypothesized that this could facilitate the intervention. However, as NAs with less experience in the region and Lima nurses who were not based in a particular health center prior to the intervention also performed tasks adequately, this experience was not essential.
A previous lack of knowledge or experience dealing with mental health issues did not appear to be a barrier to implementation. In Sao Paulo, some inaccurate beliefs about depression were identified, such as understanding it as deep sadness or lack of willpower, but could be corrected through the training and supervision provided. In Lima, even though nurses knew their tasks were not aimed at providing psychological support, they requested a brief talk about depression. Technical challenges were however flagged in Sao Paulo as an issue throughout, which highlighted the need to ensure technological literacy and support.
In Sao Paulo, a number of NAs dropped out due to logistical or personal reasons. This could have been minimized by service planning and addressing motivation to participate. Conversely, Lima did not experience nurse attrition, likely due to nurses being recruited and hired specifically for the role. Motivation and setting realistic expectations would be important to consider when planning recruitment to ensure sustainability.
Training
The duration of training differed between sites. In Sao Paulo, NAs could not be absent from the FHU for more than 1 day, since it would imply delays in their work schedule, so the training content was compressed into a shorter period of time. In Lima, the training content was delivered over a longer time frame, since they were hired full-time for this work. This enabled nurses to have more time to practice, understand procedures and gain confidence. This was evidenced by all nurses passing the training evaluation. A small number of NAs in Sao Paulo did not pass and required additional training, highlighting the importance of evaluating learning and ensuring training packages are comprehensive and allow time for assimilation in order to guarantee the delivery of the intervention with adequate fidelity. It may, of course, be that in Lima a shorter training package could deliver the same results.
Supervision
Supervision was also shown to be essential for task-shifting. Through supervision, important questions were raised, nurses/NAs improved knowledge and skills, and maintained fidelity to the intervention. Supervision was well received by nurse/NAs. In Sao Paulo, there was a significant number of missed meetings, but the majority (89%) were for reasons beyond the NAs control such as the FHU busy routine or unplanned medical leave. In Lima, nurses did not have tasks outside of CONEMO and so they were able to prioritize supervision meetings. However, as appointments were scheduled around participant availability, supervision had to be rescheduled at times. Boundaried time to attend supervision and flexibility in the supervisors’ schedule or location would be useful.
Individual supervision meetings were more frequent than group supervision in Sao Paulo, although naturally shorter in duration. However, when we consider the number of NAs and the variant interval between meetings, they required 3.3 times more supervisor time. In some instances, supervisors in Sao Paulo had to attend the FHU more than once a week (e.g., if NAs missed too many meetings in a row); therefore, they needed to be available for more time than predicted. In Lima, group supervision meetings were longer due to the number of participants each nurse was responsible for. Even though individual meetings were more frequent, they generally took place in addition to the group meetings and were performed in specific cases, especially at the beginning of the intervention to address questions regarding the intervention. Unsurprisingly, NAs with less experience with technology required more supervision support, highlighting the need for competence in this area. Given the issues discussed and caseload of each nurse/NA, weekly meetings appeared to be a reasonable frequency. When planning future implementation, the format of supervision should be based on which is more cost-effective for the health system and the best fit for the service context. Remote supervision could be a way to reduce the time required by the supervisor.
The issues brought to supervision were similar across sites over time. According to the intervention manual, it was mandatory to discuss all initial appointments. Nurses/NAs shared their experiences and challenges when training participants to use the app. Group supervision meetings were helpful in this matter, as the staff contributed with possible solutions to the issues raised. Difficulties contacting participants were a frequent subject in both sites, often due to the participant working, traveling, or providing inaccurate contact details. In Lima, lack of connectivity was frequently discussed, mainly where participants accidentally closed the app or ran out of battery. Some of these practical issues could be overcome by writing solutions into the manual and avoiding the need for supervision on these matters.
Nurses/NAs could contact the supervisor through the CONEMO system outside of supervision. This was predominately used for technological issues and never for a mental health emergency. Whilst helpful to have this safety net, it is unlikely to require heavy supervisor resources going forward. Communication between nurse/NAs and supervisors took place more frequently through an instant messaging app than on the CONEMO platform as this proved an easier and faster way to communicate. This flagged the need to utilize existing and easily accessed systems when possible, but caution is required to ensure the confidentiality of participants and security of data.
Supervision meetings in Sao Paulo were often used for the completion of tasks, similar to the experience in the pilot studies [17, 20]. This was due to NAs struggling to find time in the week to dedicate to CONEMO and so they used this time to get on top of tasks, which explains the average supervision time spent for each participant being 2.4 times higher in Sao Paulo than in Lima. This learning demonstrated the need for boundaried and realistic time frames to accomplish tasks when they are part of an existing job role, and the usefulness of a mechanism to monitor and track task completion.
In terms of capacity, supervision appeared to demand more time at the beginning of the implementation, as nurses/NAs were familiarizing themselves with the CONEMO platform and the specific issues participants presented with. Over time, they learned how to manage some difficulties and required less help from the supervisor. In that sense, the human resource management requirement would be likely to reduce over time if the staff is retained, which would increase both supervisors and nurses/NA’s work capacity.
CONEMO was able to reach significantly more people than three psychologists, such as the supervisors, could have reached on their own within the same time period (657 participants in both sites in approximately 1.5 years). Considering there are 12.4 and 9.5 psychologists per 100,000 inhabitants in Brazil and Peru [21, 22], respectively, interventions like CONEMO can help reduce the burden. Even though implementing task-shifting will not cover the high needs for mental health coverage [23], this study shows that nurses/NAs can support low-level psychological interventions. Through taking on a supervisory role in task-shifting, psychologists could facilitate access to mental health interventions for a greater proportion of the population and free up specialist resources to treat more complex cases.