In this MSF program, we observed a decrease of SRQ20 individual scores and an increase of individual GAF scores. This corresponded to an improvement in the functionality of our patients, indicating the success of treatments.
Treating mental disorders as early as possible, holistically and close to person’s home and community leads to the best health outcomes 
. Primary care offers unparalled opportunities for prevention of mental disorders and mental health promotion, for family and community education, and for collaboration with other sectors 
We observed a progressive increase of inclusion over time, proportionate to the acceptance and information about the centre, thanks to “word of mouth” as well as the services’ promotion made by our CHWs. Collaboration with other organizations (including UNRWA) becomes relatively effective in referring patients to the center and the camp’s clinics. The work performed by the CHWs and SWs inside the patients’ homes also brought patients to our services. Referrals from external doctors remained comparatively low.
We observed a progressive increase in the proportion of males coming to our services, as contacts with the religious authorities were reinforced.
Most of our patients were Lebanese, which is not surprising given the centre’s location and the density of population in Burj-el-Barajneh. We estimate that, even if a prevalence study were not performed there, the needs and gaps regarding mental health services would be almost identical (deprived area, with a lot of displaced population) to those found inside the camp.
The proportion of Palestinians coming to the services has risen while those services were progressively integrated in the camp health structures.
Evolutions of both scales can be considered as satisfying. As the cut-off score for SRQ 20 was arbitrarily set at 7 (based on WHO document 
and other field observations), we could consider our intervention brought most of the patients under this line. Stabilization around 6 is, in this sense, a very good result.
After several visits we no longer observed an effect. This could be the stabilization phase, corresponding to the maintenance treatment of chronic patients, but this could also questions the necessity to perform more than 6 to 8 consultations.
Being unable to test for all possible predictors, we used other studies as a base to examine which predictors to explore 
. Our intervention appears better adapted to patients suffering from anxiety than to those with severe disorders like psychosis. The baseline score is also a predictor of evolution over time tending to show we do better with less severe cases. A big and progressive effect was observed for the educational level as a predictor of a favourable evolution. This leads us to think we should increase our attention and adapt our interventions to severe mental disorders (SMD) and less educated patients.
Patient file which included both scales was tested beforehand on healthy volunteers. Those scales have been chosen in agreement with the clinicians of the project. We are considering keeping only one of the instruments for monitoring of future projects. We would prefer to opt for SRQ even if it is not adapted to severe (psychotic) patients, nor always supported to follow patients’ evolution (as a monitoring tool). The narrow distribution of GAF (between intervals 50 to 80) and the inter-rater problems of reliability create more limitations in the use of GAF only. Another problem faced here with the GAF is that we do not observe any further changes after 3 visits.
Some colleagues from other MSF sections use 1–10 Likert scales, complaint and functionality rating (patient and clinician point of view), which are very practical, but with the disadvantage of being not yet standardized tools 
For common mental disorders (CMD) there would be other possibilities such as the depression and anxiety stress scale (DASS) while for psychosis we could consider using an adaption of the positive and negative syndrome scale (PANSS) or the brief psychiatric rating scale (BPRS). Actually we may have to use different instruments for CMD and SMD.
Our study presents several limitations. First, a moderate proportion of patients have comparative measures (38.5% for SRQ20 and 46% for GAF). As in each cohort, we probably have a selection bias. The patients remaining in the programme (and for whom we have comparative measures) are the ones doing better.
We have tried to collect a lot of information through the patient file on top of SRQ 20 and GAF (a questionnaire on health behavior, the Self Functioning questionnaire with 12 items, the Harvard Trauma Scale, and the Trauma Scale Questionnaire). It is certainly one of the reasons why the questionnaires were not always filled. From this experience we believe that the project team’s efforts in collecting information can be sustained for a limited period of time (pilot or research project) but that this information should be limited to the minimum needed for effective monitoring of a regular project and should also be timely analyzed at field level.
The aim of this study was to describe individual outcomes of patients and to identify potential predictors of these outcomes in a refugee camp based mental health intervention. We believe that the efficacy of our treatment model has been demonstrated by the positive individual outcomes we got in terms of functionality. Analysis of predictors of this positive evolution shows that we need to adapt our model for the more severe and less educated patients. It also makes us reflect on the length of the individual follow-up. Those results must be taken into consideration in our other interventions, with limitations due to the context. This study provides evidence that SRQ20 can be used as a baseline and monitoring tool but that the use of GAF for the same purpose is questionable.
A further and deeper analysis of the data collected would be needed, especially regarding the link between socio-economic determinants and patients’ outcomes. We do indeed consider social determinants of mental health as predictors of mental health conditions (in protracted refugee situations in general). Advocating for better socio-economic conditions for the refugee populations is also part of our work.
Further research could also include a qualitative evaluation of the intervention.