This pilot model, focusing on the diagnosis and treatment of 5 syndromic mental illnesses and completely integrated into existing primary health care services was feasible in Habilla.
Our findings seem to demonstrate that patients improve their functionality as is seen in non-war contexts.
To our knowledge, this is the first evaluation based on real life activities to demonstrate positive outcomes for patients treated for severe mental health problems in a humanitarian emergency.
As others, we also observed that after the initiation of mental health services in an area that had never had access to diagnosis and treatment of mental illnesses, the first patients to attend are patients with severe mental illness and epilepsy . We believe that it is fundamental to address immediately the needs of these patients.
We also speculate that the low numbers of patients with CMD diagnosed by our services was due to the fact that in the first 6 months, we were unable to train properly the medical staff on the recognition and diagnosis of CMD. As a result, we plan to address this problem by developing and testing algorithms where CMD can be diagnosed in patients presenting at outpatient level with multiple unexplained somatic complaints as these are known to be strongly associated with depression and anxiety in developing countries .
We believe that the low numbers of cases of alcohol and drug abuse identified in our services was due to the religions restrictions on consumptions of these substances in Darfur.
We have decided not to measure outcomes for children identified with mental health problems due to the lack of availability of simple and short instruments that can be incorporated into mental health services in remote areas and humanitarian emergencies. Most of the cases labelled as child mental health problems identified by us were cases of intellectual disability. We are also aware that we have not properly developed clear case definitions for specific children's mental health problems that can be found in conflict situations (e.g. enuresis, post traumatic stress disorder). There is a clear need to identify children's mental health problems of public health importance in humanitarian emergencies in order to help organisations to prioritise the delivery of services for key mental health conditions.
We are of course worried with levels of defaulting and future research should investigate possible determinants of defaulting in mental health services in humanitarian emergencies. We speculate that defaulting was associated with poor family support and distance from the health centre. Unfortunately, we have not included these variables in our monitoring system and therefore can not test these hypothesis here.
There is a strong need for organisations to share their strategies on how the integration of mental health into primary health care services can be better accomplished and especially in contexts like Darfur, or other emergencies.
We believe that this integration is highly context specific. Here, we struggled to treat patients that had never received any treatment before, staff needed considerable training and we are in an extremely violent area. In Beirut, in 2006, we received patients that were already on treatment but left tablets at home while escaping from aerial bombardments. In that setting, we struggled to keep them on treatment by having an essential list of drugs and relying on qualified local professionals. (MSF internal unpublished report)
This study has very important limitations. Being a small study as it is based on real life activities, caution is needed on the interpretation of the findings.
We acknowledge that having chosen only 5 syndromic diagnoses in order to integrate mental health into primary health care can be considered too simplistic. Our 5 syndromes approaches could also have missed cases of mental disorders that are expected to be seen in war contexts, such as acute stress disorder and post-traumatic stress disorder .
During the development of our intervention, we in fact included those diagnoses under CMD as we are aware about the frequent comorbidity among those conditions.  It is important to remember that there is an urgent need to develop simple packages of mental health care that can be implemented in remote settings and humanitarian emergencies. The recent Mental Health Gap project, launched by the World Health Organisation, focuses on 9 selected mental health conditions of public health importance. We believe that this is the way forward in order to scale up mental health care in under serviced areas .
As our main measurement instrument was developed through a process of simplification and adaptation of the SF-36, we should be cautious on interpreting levels of improvement on functionality. Although we acknowledge that using standardised instruments would have helped us to interpret our results and compare them with other studies, our experience has shown that most of the standardised instruments in mental health can not be used in routine mental health services in poor resource settings due to their length and need for specialised personnel. If outcome assessments are to be implemented in parallel to the process of scaling up of mental health services by using non-specialised health workers, simplified instruments need to be developed.
The lack of control group also limits a clear understanding of aspects of the intervention package that are essential. Different packages of care should be compared in different contexts in order to better understand other possible models for integration of mental health care into primary care in the context of a humanitarian emergency.
It is possible that the interviewer could have tended to mark better scores for patients at follow up. This would have biased our estimates. We believe this is unlikely as there was quality control by the supervisor psychiatrist.
Lastly, by having a sampling framework consisting of the first 6 months consecutive patients that attended mental health consultations, our results might only reflect the patients with higher chances of achieving good outcomes as they were able to reach our health facility and return to follow up consultations. Therefore, our positive results can not be generalised to the whole population with mental disorders in Habilla.
As we were unable to reach patients who were lost to follow up, the fact that they were not introduced in the analysis of outcomes remains as an important limitation of our study. In order to better understand bias due to loss to follow up, we have reanalysed the data by giving the defaulters the same disability scores as their last consultation. Although there was a reduction on the size of the mean differences, the results remained positive.
Conducting research in humanitarian emergencies poses numerous challenges. First, there are security risks to the staff and the possibility of rapid evacuation due to deterioration of the security situation can largely undermine the continuation of the medical activities. Second, the constant supply of psychotropic medication can be also interrupted in case of lack of continuous access to the implementation sites. This has to be strongly considered in advance and stocks of medication have to be kept at health centre level. Thirdly, the simplification and adaptation of diagnostic tools, treatment guidelines, training protocols and measurement instruments to be used need the support of experienced staff, which is not always available and willing to work in extremely violent contexts.