The Mindanao project in the Philippines shows that simple mental health approaches such as psychological first aid and brief psychotherapy can be integrated into primary health care in an emergency humanitarian context. Furthermore, retrospective analysis of patient data suggest that brief psychotherapy sessions provided at primary level to patients with common mental disorders can potentially improve patients' symptoms of distress, within a few sessions.
Although there were a high number of dropouts from the program, it is important to note that patients did improve before they dropped out. This high proportion of dropouts could be linked to the volatile security context and regular displacements occurring in this population, which may prevent patients from attending consultations. We do not think that this reflects failure of care. Flexibility in the pattern of follow-up is a necessity in such an unstable environment, where regular attendance to appointments at fixed points in time cannot be expected. However, our data show that even a brief and sometimes irregular intervention can lead to substantial improvements in patients' conditions.
Whereas other case series conducted in violent contexts such as Darfur , Palestine  and Colombia  have already described characteristics of patients affected by mental disorders, our data have the advantage of having used standardized outcome measures and not only psychologist's opinion. Interestingly, our series consisted of a higher proportions of patients with common mental disorders when compared to the patients in Darfur , which showed a high proportion of severe disorders. This may be a reflection of the active case detection approach used in Mindanao, integrated into primary care, which allowed for detection of non-severe cases of mental disorders.
The creation of a strong network of community health workers was crucial to identify potential patients and to ensure good follow-up. CHWs also played an important role for adherence to psychological support and pharmacological treatment, by speaking with the patient about the importance of finishing treatment. Indeed, without the work done by the CHWs in this project, the proportion of defaulters would probably have been much higher. It was also important to find local psychologists able to speak and understand local languages and cultural issues. This gave patients the opportunity to express themselves in their own language, while receiving professional care from someone coming from the same cultural background. The good collaboration between the medical staff and the mental health team was also an important factor of success of the project. This was facilitated by previous sensitization and training of medical team on mental health issues.
It is worth noting that changes on median GAF scores reflected a progression from moderate symptoms to mild symptoms and good functionality. Although the GAF score has been used previously to measure patient outcome, the SRQ20 score was not validated as such for this purpose. However, we do find this scale useful in this situation, as it is referring to items related to distress not directly related to a specific diagnosis. Besides, it has been used in a number of different cultural contexts. Interestingly, GAF and SRQ scores showed a linear relationship in our dataset (regression coefficient -1.5; 95%CI -1.53, -1.41; p < 0.001), which strengthens our conclusions. Clinicians (doctors and nurses) also judged the SRQ20 to be a useful tool to perform screening of a suspected case before referring them for specialized assessment. Further research on the development and use of outcome measures that can be standardized, acceptable to primary health care practitioners and feasible for routine use in humanitarian settings is of the utmost importance .
One of the limitations of this work is the absence of a control group. Indeed, we cannot exclude that the positive outcomes seen in this project are not due to the intervention, but may only reflect the healing effect of time itself. The possibility of bias due to the fact that professionals providing mental health services were the same ones that measured outcome scores can not be excluded. We tried to minimize this by implementing continuous training and quality control on the use of the scales. Further, outside of a study context, inclusion criteria into the program were not strictly defined, allowing for the follow-up of some very paucisymptomatic cases. This inclusion of patients with light symptoms may have accentuated the positive impact of the intervention. This highlights the need for continued formal research in this area.