The Burden of Mental Disorder in Sierra Leone: an Evaluation of Programmatic Data From the Roll Out of Decentralised Nurse-led Mental Health Units

Background: In sub-Saharan Africa the treatment gap for mental disorders is high. In 2009, 98·0% of people with mental illness in Sierra Leone were not receiving treatment, partly due to the absence of public psychiatric facilities outside the capital. In response, the Ministry of Health and Sanitation rolled out nurse-led mental health units (MHU) to every district. Methods: We evaluated the roll out of MHU using summary data from all units between 1 st January 2015 and 1 st January 2017, to establish the burden of diagnoses among service users, pathways to care, treatments provided, and treatment gaps. Negative binomial regressions examine bivariate relationships between diagnoses, treatments, and medication inaccessibility with demographics (age and sex), location (Freetown vs the rest and Ebola endemic regions vs the rest) and year. Results: We collected data from 15 MHU covering 13 districts in 24 months. There were 2401 referrals. The largest age category was 25-34 (23·4%). The prominent diagnoses were epilepsy (43·5%, associated with children) and psychosis (17·5%, associated with males). Reported depression (8·6%) and suicide attempts (33 patients) were low. Ebola endemic regions reported higher rates of grief, trauma, and medically unexplained symptoms. In 24·7% of cases where medication was required, it was not accessible. Conclusions: Nurse-led MHU can have a modest effect on the treatment gap in resource constrained environments such as Sierra Leone, particularly in epilepsy and psychosis. with (Freetown,


Background
Mental health and substance use disorders constitute a high burden of disease, accounting for 23% of disability-associated burden (years lived with a disability, YLD) globally and 19% in sub-Saharan Africa. 1 The treatment gap is highest in low-and middle-income countries (LMIC), ranging from 76·3% to 85·4% 2 , and research on mental health services in these countries is limited. 3 The gap is attributed to human resource constraints, stigma, weak technology and infrastructure, lack of political commitment, culture and traditional beliefs, and lack of research to guide policy formulation and implementation. 4 Task-sharing and health system restructuring have been shown to improve access to healthcare in non-communicable and infectious diseases. 5,6 Evidence around task-sharing in mental health mostly focuses on primary care. 7 Although models of task-sharing with trained nurses have been implemented, [8][9][10] there is little evidence about the effectiveness of these programs outside of primary care in Africa.
In Sierra Leone, there are no recent mental health prevalence studies; however, a 2002 Ministry of Health and Sanitation (MOHS) survey at the end of the civil war, indicated prevalence rates of 2% psychosis, 4% depression, 4% severe substance abuse, 1% mental retardation, and 1% epilepsy. 11 The prevalence of mental health problems is likely to have been adversely affected by the 1991-2002 civil war, 12 economic problems, 13 gender based violence, 14,15 poor health outcomes including high rates of maternal and child mortality, 13 and the recent Ebola epidemic. 16 In 2009, only 2058 of an estimated 102 000 people with severe mental illness received treatment, resulting in a treatment gap of 98·0%. 11 Until recently, the only MOHS treatment facility was the Sierra Leone Psychiatric Hospital (SLPH) in Freetown.
In response, the MOHS collaborated with partner organisations to train 21 nurses in a 12-18 month mental health program in 2012. 17,18 Of those trained, 19 remain in active service in addition to three nurses who completed training abroad. This created a mental health nurse workforce on par with the median for low-income countries (0·26 per 100 000 population). 19 In 2015 these staff established Mental Health Units (MHU) in all 14 districts. A centralised Child and Adolescent (CAMH) Unit was established in 2016.
Medication supply is sporadic and largely relies on donations from non-governmental organisations (NGOs). 20 The nurses assess and treat patients in accordance with the mental health Gap Action Plan (mhGAP). 21 The availability of monitoring and evaluation records over 2 years provides an opportunity to establish the burden of disease that present to district nurse-led mental health services and range of treatments provided in decentralised MHU. This analysis will contribute towards understanding the mechanisms for addressing the treatment gap in resource constrained environments.

Data
The data includes monthly reports on all persons receiving treatment at 15 MHUs in Sierra Leone between 1st January 2015 and 1st January 2017. These facilities are clinics embedded within general government hospitals, including one general referral hospital, one children's hospital, one military hospital, and 12 district hospitals.
Clinic staff collated monthly summary data. Where data sheets were missing, staff collected them from clinic ledgers resulting in 290 data sheets. Data collected include socio-demographic details, diagnoses according to World Health Organization criteria used by the mhGAP, 21 number of new patients, referral sources, contacts (number of patients seen, new and follow-up), signposting to external organisations, and the number of patients offered psychotropic The Ethics and Research Committee of the MOHS, Sierra Leone granted ethical approval to conduct this study.

Analysis
The analysis begins with a descriptive summary of total monthly records including patients' demographics, diagnoses, pathways to care, treatments, treatment gaps and differences across facilities. Additional analysis identi ed statistically signi cant differences using bivariate regressions in total monthly counts of diagnoses and treatments by patients' demographic characteristics (age groups and total number by sex), location in Freetown vs other districts, location in areas with high Ebola endemicity (Freetown, Kenema, Bombali, and Kailahun) vs low Ebola endemicity, and time (2015 vs 2016). The aggregate monthly counts and subsequent proportions were summed by facility. There was one clinic per district other than Freetown which had three facilities, whose counts were combined to examine variations across districts. We used negative binomial regressions because the outcome variables (total monthly counts per clinic of diagnoses and treatments) took non-negative integer values. Series of bivariate negative binomial regressions estimated counts of diagnoses (epilepsy/seizures, substance abuse, intellectual disability, psychosis, depression, other, and medically unexplained somatic complaints) and interventions (psychological, social, medication, medication not available, and medication unaffordable). The total monthly caseload was used as the exposure variable.

Results
Over the 2-year period, the total number of new patients was 2401.
A total of 33 patients presented to services having attempted suicide or self-harm (1·4% of 2401
Social interventions were positively associated with age group 0-14 (p = 0·001) and negatively associated with age groups 35-44 (p = 0·006) and 45-54 (p = 0·002, Table 2). Medication and psychological interventions were less strongly associated with age. Table 2 Comparison of the number of each category of intervention given with different demographic characteristics and year -relationships between the number provided with each treatment (psychological, social, medication, and those for whom medication with unavailable or unaffordable) and selected independent variables (number in each group: age, sex, residence within Freetown vs the rest, residence within districts with high Ebola endemicity vs the rest, and year 2015 vs 2016) using bivariate negative binomial regressions *p < 0.05 The number of females was negatively associated with the number of psychotropics prescribed (p = 0·02). Other interventions were less strongly associated with sex.
The total number of prescriptions required was 3719. 24·7% (919 of 3719) were for medications that were inaccessible to the patient. The medication was unavailable in 52·8% (485) of these 919 cases, and unaffordable in 47·2% (434).
However, while no clinic was exempt from these problems, the few clinics with the highest numbers of contacts accounted for the majority of medication problems. For example, one MHU represented 74·0% (359) of all 485 cases of medication unavailability, which is 70·3% of all 511 prescriptions in that unit.
Another MHU represented 43·3% (188) of the 434 cases in which patients could not afford medication, which is 40·6% of all 463 prescriptions in that clinic.
Medication inaccessibility was less signi cant in clinics that had fewer contacts.
A total of 45 referrals were made to the SLPH. Two MHU -neither based in Freetown -referred 82·2% (37)  There were a further 74 patients who had not contracted Ebola, but someone in their family had suffered from the virus.

Variation by district
There was variation in total numbers of new patients and each diagnosis across districts (Fig. 3). The diagnoses with the highest variability were epilepsy/seizures (ranging from 0·0 to 93·5%, which was 315 of 337 patients seen in one clinic over 2 years), alcohol and other substance abuse (0·0 to 55·8%, 29 of 52), and psychosis (0·0 to 18·4%, 216 of 1176).
Medication unavailability was signi cantly associated with year 2015 compared to 2016 (p = 0·02, Table 2). There was no change in medication unaffordability over time.

Treatment gap
Coverage by the MHUs (based on prevalence estimates from other studies 1,11,22 ) is 0·2% of the population experiencing schizophrenia and bipolar disorder, 0·2% with alcohol and substance use disorders, and 0·4% with epilepsy (Table 3).

Patient characteristics
Men and women accessed services equally. The number of males was associated with numbers with psychosis, depression and substance abuse. The largest age category was 25-34 years (23·4%). However, globally half of mental health disorders start by the mid-teens and three quarters by the mid-twenties, 23 and in Sierra Leone 48·3% of the population is aged under 18. 24 This emphasises the need for robust systems that are suitable for younger people.
Epilepsy/seizures accounted for the highest proportion of diagnoses seen. This is unlikely to indicate a greater prevalence amongst the general population compared to mental health conditions, but may re ect improved referral mechanisms and health literacy resulting from organisations including the Epilepsy Association of Sierra Leone. An estimated 1·43% of the population in West Africa have active epilepsy. 22 There is only one neurologist in Sierra Leone, 11 and MHUs may be playing an important role in reducing the treatment gap.
Psychosis was the most frequently diagnosed mental health disorder, surpassing more common disorders such as depression (17·5% and 8·6% of MHU patients respectively).
The under-representation of depression in newly established services replicates studies elsewhere. 25 The comparative population prevalence rates for psychosis and depression are 1·14% and 3·67% respectively in Western Africa. 1 It is possible that patients with less visible disorders such as depression may not prompt the same level of concern as psychotic disorders, that there may be differences in perceptions about treatment, and that affective disorders are not viewed in the same explanatory model as psychotic disorders. 26,27 Whilst we expect the rates of depression to be much higher, it is important to consider factors that promote resilience such as social support through community acceptance. 28,29 The 327 Ebola survivors attending MHU represent 3·3-6·9% of the total survivors (4750 laboratory-con rmed, 10 000 estimated). 30,31 Studies in Ebola sequelae show prevalence rates of 10·9-49% for depression, and up to 34% for post-traumatic stress disorder. [32][33][34][35][36] The MHU appear to be helping reduce the treatment gap for mental health disorders among Ebola survivors, particularly for people with grief, stress and trauma, and medically unexplained somatic complaints. The majority presented in 2015 (80·1%) compared to 2016 (19·9%), indicating mental health problems may present as early sequelae of Ebola. 19 However, this result may be affected by targeted health initiatives to screen Ebola survivors mental health problems. This could explain the differences in data between districts, which are not fully accounted for by endemicity.
Low numbers of reported suicide and self-harm attempts in this study (33 in 2 years) could be further explored to evaluate if this is an issue of reporting or re ects a true picture. The WHO estimates the suicide rate in Sierra Leone is 9·7 per 100 000, higher than the sub-Saharan average of 7·5. 37 This would equate to 733·0 per year in Sierra Leone, and for every completed suicide there are 50 attempts. 38 Among psychiatry patients the suicide attempt rate is 15-50%. 39 The results from this study should be interpreted in a context of wider social, cultural and stigma related issues around suicide in Sierra Leone.

Patient numbers and pathways to care
The increase in service uptake by year is likely a re ection of service establishment due to the successive opening of MHU. The last MHU (the CAMH Unit) opened in March 2016.
There is wide variation in the number of referrals into different MHU. These differences appear un-related to district population and should be investigated further. Issues around sta ng levels, geography, community access to mental health information, levels of education, traditional beliefs, stigma, and local opportunities to align resources towards mental health and epilepsy are likely contributing factors. These factors also might help explain the geographic variation in the types of disorders presenting to services.
The most common referral sources are self-referrals and referrals from other general hospital departments. A wide range of organisations are aware of, and referring to, the MHU. There are no data on how self-referred patients heard about the service. Possibilities include radio campaigns, word-of-mouth and healthcare professionals. Most patients are outpatients, indicating these services are making treatment available to people living in their homes.

Mental health treatments
Among clinics with the highest numbers of contacts there are sizeable problems with accessing medicines. These problems transcend all patient ages and sexes. Medication inaccessibility potentially undermines quality of clinical care and trust in the service. This highlights the need for the MOHS to include psychotropic medications in procurement plans, to strengthen supply chain mechanisms and prevent leaks within the supply. 40 Whilst supply is more accessible in the capital city, medication often remained unaffordable. Innovative means of mental health nancing may be required.
Few other psycho-social services are available, aligning with the low rates of signposting to other services. Treatment and rehabilitation frequently require the involvement of family and community, therefore the most common interventions applied are likely to be psychoeducation and attempts to engage wider supports. These activities are captured in the number of social and psychological interventions. Psychological interventions beyond psychoeducation would mostly have consisted of counselling and problem-solving therapy, which is appropriate to the level of training of MHU staff. Referrals from the MHU to the Psychiatric Hospital are low (45 in 2 years). It is possible that the nurses felt adequately prepared to handle most cases; patients or their families blocked the referral due to the challenges associated with travel, or were against admission for other factors including the hospital being under-resourced; or there was a preference for alternative forms of care. Since deinstitutionalisation, psychiatric inpatient facilities only provide 7% of psychiatric care globally. 41 Home visits formed a signi cant part of the workload. Home visits may be preferred by patients because of travel di culties or stigma associated with the MHU or may be the only option to see people who are very unwell. As much as this is a good indicator of patient-centred care uptake, the increasing demand on mental health nurses' time poses a risk towards the quality of services.

Treatment gap
Despite the positive trend in service uptake, coverage by the MHUs for the two most common disorders, epilepsy and psychosis, are only 0·4% and 0·2% respectively. This may slightly reduce the treatment gap for mental health disorders, which in 2009 was 98·0% (the remaining 2·0% treated by the Psychiatric Hospital and NGOs). 11 In 2016, the treatment gap for childhood disorders was estimated at over 99·8%. 42 This demonstrates that services are still not meeting the needs of the population and further exploration of how to develop services is warranted. This is especially important given that the population has risenby 2 million since the most recent epidemiological survey for mental health in 2002 -and the burden of mental health disorders is expected to greatly increase over the next decade. 1 Challenges in service delivery Efforts to increase service uptake are threatened by a multitude of challenges, ranging from problems with the wider health system to issues speci c to mental health. These include: human resource constraints, inadequate medication supply chains and the availability of only one psychiatric inpatient facility. Of the 15 MHUs, 13 are each staffed by one nurse working alone. These staff are responsible for all clinical and service management. Staff ill-health can result in no patients being seen.
Mental health nurses are not fully integrated into the health system. 20 They receive low remuneration for their work and have limited opportunities for career progression, potentially impacting future attrition rates. There is limited potential for the MOHS to conduct training or supportive supervision, both of which are provided by NGOs. The MOHS has no speci c budget for mental health outside SLPH, which has its own funding di culties.
Patient-related factors that may have contributed to low rates of presentation at MHU include stigma, low mental health literacy, preference for traditional/faith-based healers, and inability to nance health cost highlighted by the number of patients who could not afford medications. 7

Wider efforts in task sharing
Through the support of partners the roll out of mental health services to primary care has led to training in Psychological First Aid 30  Research and monitoring and evaluation are at the forefront of the policy. As yet nance options are uncertain.

Limitations
The summary reports in this study use diagnoses and treatments according to broad categories. Psychosis, for example, includes schizophrenia and bipolar disorder, which have different prognoses. It was not possible to assess the diagnoses for reliability. However, mhGAP was developed by experts in the eld as a viable method of scaling up mental health treatment using non-specialist services. 44 The de nition of some variables like psychological intervention and social intervention was not distinct. Data on individual medications are not available. However, the information collected was readily available and collected to guide policy and planning.
The results are limited by the accuracy and completeness of the data. There were differences between clinics on the source of patient data, from either new patients or all patient contacts. Missing data was often accounted for by clinics not reporting sections of the data form and reviewing the missing data from individual clinics did not demonstrate any patterns that might indicate resultant bias. Issues with routine collection of health data in Sierra Leone are certainly not limited to mental health. 31,32 Conclusions In resource-poor environments the roll-out of decentralised nurse-led MHU such as these new services in Sierra Leone can provide vital treatment, particularly to younger people, and people with epilepsy, psychosis, and psychological symptoms associated with stress, grief and trauma. The rate of depression was less than expected indicating services were not meeting need in this area. The numbers of referrals and subsequent interventions increased during the study period but there is a need to further improve coverage and medication accessibility. The variation in the provision of clinical services warrants further exploration. This study addresses a research gap for the service utilisation of specialised nurse-led MHU in sub-Saharan Africa.