Our analysis of almost 15000 clients gives a comprehensive overview of who accessed MSF counselling services in 18 different humanitarian contexts. Our results show a clear gender bias, in that men were underrepresented. Two of the projects exclusively targeted women. However, all projects, irrespective of target group, had a lower percentage of men presenting than women, with a mean of 27.5% male patients. This could be a result of gender-related cultural norms, but could also represent a lack of attention to this group in community education or unrecognised barriers to attracting men .
The most common reason for seeking care was 'anxiety-related symptoms’, a significant proportion of which was linked to arousal symptoms such as hypervigilance that are associated with PTSD. Few clients presented with a serious mental health condition, despite associations between psychiatric disorders and violence and war [21, 22]. Furthermore, there was no link between more violent project contexts and presenting complaints of a serious mental health condition. Our programmes however did not target major psychiatric disease and did no active case finding.
MSF mental health programmes are designed for those affected by conflict and violence. But our results show that counsellors focused on trauma-related symptoms in fewer than 1 in 5 cases. Only half those enrolled gave a clear history of violence as the precipitating event for their complaint. A large proportion of clients (25%) presented with a precipitating event of domestic discord or violence; few (7%) could directly link this to war or conflict. It may be that underlying disruption to the family caused by the conflict was indirectly linked to the main complaint or that the client was not able to recognise this connection. There was no link between active conflict settings and a precipitating violent event. However the high relative rate of sexual violence found in conflict settings is consistent with the many reports of sexual violence as a weapon of war [23, 24].
Outcomes for the clients who returned for a second session were positive, regardless of whether the outcome measure was a client-rated scale or counsellor assessment. Equally important, the percentage of individuals whose scores worsened during treatment was below 2%. This is reassuring given the concern that some forms of psychological intervention after a traumatic event can cause psychological harm . Children were included in the analysis as they are included in our routine programmes and registered and treated using similar principles as in adults. It is unlikely that their inclusion influenced the results; when those under 18 years were excluded from the outcomes model there were no substantial changes.
As others have noted, the number of sessions was strongly linked to successful outcome of treatment [26–28]. We saw a high rate of attrition, though it was highly variable between programmes. In most cases there was no reason given for not returning, but there was an association with increased severity of complaint on presentation, indicating this needs further exploration. In some projects, there were physical, social or even security barriers that prevented clients from returning. People may have been displaced during treatment, or travel may have been insecure. In project sites in Pakistan, women needed to be accompanied by a male family member to leave the home, which may have had an impact on their retention in care. In Colombia, services were often delivered by mobile clinics in very remote, insecure areas, where access was possible only by boat or foot. We are not aware of any other reports of attrition rates in mental health programmes in humanitarian settings. However, in western settings there are reports of high drop-out rates and high rates of patients not returning after the first encounter . Baekeland reported 20-57% of single session visits, compared with our rate of 44% . This suggests that our drop-out rate is not unusual especially given the potential difficulties in our settings with returning for follow-up care.
We were unable to assess outcomes in the 44% of our clients who attended only one session. The impact of a single session of counselling is controversial given the strong association between completed treatment and outcomes . In Colombia almost 95% of individuals reported that a single session was beneficial . In Australia (North Yarra Community) 90% regarded single-session counselling as useful . In Melbourne, Australia, 78-81% were satisfied with single-session counselling as an alternative to being waitlisted for family therapy . However, the MSF model was designed to be a multi-session intervention and was not adapted for single session therapy.
In many resource-limited settings, academically-trained counsellors are not available. Although we were not able to directly compare the quality of care provided by specialised versus non-specialist providers, our findings suggest that lay counsellors can achieve similar outcomes to those academically trained, within the limitations of the outcome measures we used. This is consistent with other positive results with trained lay counsellors [8, 32].
Clients with the best outcomes attended more sessions, lived in a conflict setting, were treated in a smaller project more than a year old and did not have a serious mental health condition. Stated differently, risk factors for poor outcomes were attending few sessions, living in a stable setting or one with high levels of societal violence, having a serious mental health condition, and attending a large, recently opened project.
The better outcomes in conflict and unstable settings may be because our mental health programmes are designed for these settings. In contexts where clients had mainly social problems or in post-conflict settings, the reasons for presenting may have been less acute and therefore more difficult to address. Counsellors did not provide psychotherapy. Their aim was to reduce symptoms and improve functionality. Context was also strongly associated with severity of complaint rating at first visit, with societal violence having the strongest association with severity of complaint. Further study is needed to evaluate the best intervention approach in post-conflict and societal violence settings.
Few patients with a serious mental health condition accessed counselling, and those that did derived less benefit than other clients. It is important to note however that our counsellors were not trained to diagnose psychiatric illnesses, nor was the programme designed to target those with severe mental health disorders.
Smaller projects might have had better outcomes because counsellors received more clinical supervision from their mental health officer. The poorer outcomes in newer projects may be due to lack of experience of the counselling staff. MSF invests in training counsellors in a specific approach, and whether lay counsellor or academically-trained, it is likely that the counsellors are more effective with greater experience.
There are several limitations to our analysis which uses routinely collected programmatic data. The data come from a range of programmes, and although a standardised database was used, data quality can vary. Additionally, although intervention approaches were standardised, the individual counsellors or mental health officers could have modified the standard approach. Another potential source of bias is that data were extracted at different times. This is however unlikely to have had a major effect, as the number of sessions analysis included only 2.6% of open files, and only 19 of 7793 open files were included in the outcomes analysis for complaint rating difference. Re-running the outcomes analysis without the open files had a trivial effect on model parameters.
A limitation for the outcomes analysis is that the monitoring tools have not been externally validated. Further, the client rated tools were adapted to the specific cultural context and counsellors were taught to check for understanding, but this might not have been done consistently across and within programmes. Outcomes were consistent between client and counsellor and functional and complaint ratings, suggesting that the results are robust. Nevertheless, the outcome measures are subject to related biases and are interdependent. Finally, as no control group was used, it is possible that the outcomes are due to evolution over time rather than the counselling intervention. This natural improvement over time may also differentially select for acute problems over more chronic issues and may account for the differential improvement in areas of acute conflict as compared to post conflict areas . However this would not account for the poorer outcomes in the context of societal violence where the frequency of acute violent events is also high. Finally, and importantly, we are unable to assess the outcomes in the 44% of clients who did not return for a second visit.
The analysis of clients and their outcomes resulted in a number of lessons learned for MSF, many of which may be applicable to other programmes offering individual counselling services in similar settings. The fact that men are significantly under-represented in our programmes, suggest that pro-active targeting of men is needed. This should be supplemented by sensitising medical staff to the importance of identifying men who present in need of mental health counselling. In projects with high numbers of single sessions, especially those due to contextual reasons that are difficult to influence, tailored single-session counselling should be offered. Tailored single sessions would mean less in-depth assessment (of root causes of complaints) and more emphasis on a problem-focused approach where the objective is to provide direct support to the client . This approach has now been adopted by MSF in programmes with high attrition rates after the first session. Results are difficult to measure but feedback from counsellors is thus far positive. In programmes where the attrition rates cannot be attributed to contextual factors, increased attention to adherence is important and should include simple strategies such as reminder phone calls prior to appointments. Finally, perhaps the most important lesson learned is counselling interventions need adaptation to the context. Post conflict settings may demand different intervention models than those in acute conflict or where there is a high incidence of domestic violence. For example in Papua New Guinea where clients present with a history of extreme societal violence, often involving intimate partner violence, MSF has adapted its model of care to include couple’s counselling and anger management training.