Interventions to increase use of services; Mental Health Awareness in Nigeria

The Mental Health Awareness Programme (MHAP) was an initiative of a local Non-Governmental Organisation (NGO) in South East Nigeria called Amaudo (or Village of Peace in the local Igbo language). Amaudo was initially established in 1990 to support homeless people with severe mental illness, and the Community Mental Health Programme was added to this work 10 years later [18]. The services are provided as a partnership with the State and Local Governments in South-East Nigeria. Nigeria’s health system reflects its Federal structure of governance, with specialist tertiary services being largely the mandate of Federal Government, each of the 36 States running secondary hospital services, and Local Government Areas (LGAs) running primary care. Local Government Areas are a legislative level that may be considered equivalent to a large health district, under which there are usually many primary care centres of different sizes, organised under health wards. Under CMHP then, LGAs were identified as the unit under which the services would be organised. Services are run by a psychiatric nurse placed in one primary care clinic—usually the largest—in each LGA in the State. This nurse provides outpatient services, and also makes domiciliary visits and acts as a point of referral for other health workers. The awareness programme was specifically designed to strengthen this service and used the network of CMHP clinics (and psychiatric nurses) in its three States of operation in Nigeria (Abia, Imo and Ebonyi States) as well as in Anambra and Enugu States where the CMHP planned to expand. Its aims were to increase awareness about human rights of people with mental illness (and epilepsy); to change attitudes and reduce discrimination; and to increase the number of people using primary health care for mental/neurological illness, in line with the Nigerian National Policy for Mental Health [19].

The MHAP intervention was designed to increase use of a mental health service in Nigeria, by utilising Village Health Workers (VHWs), who work in communities, usually in a health promotion role such as vaccination programmes or HIV/AIDS awareness. As local volunteers linked to health services, they have a unique engagement with members of communities, including people with mental health conditions, and their carers [20]. 15 VHWs in each Local Government Area were identified using clear criteria designed to engage those who were motivated to work in mental health, and were likely to remain in the community and role for some time. A total of 315 VHW were identified and each group of 15 received a 5 day training by the psychiatric nurse in their local clinic, using a package that had been previously developed and tested by Amaudo. The content focused on gaining knowledge about mental illness, promoting human rights and dignity, and sharing practical ways of supporting social integration. The fact that the nurse provided the training reinforced relationships for ongoing work between the nurse and the VHWs. In addition, key local actors such as Primary Health Care Coordinators and other local health leaders were invited to supervise the training, and local community leaders participated in opening ceremonies. Further details are available from http://​www.​amaudouk.​org. Following the training, they were then expected to visit each village in their catchment area, share information using the provided materials, and encourage people identified as having mental health problems to attend the clinic. The awareness-raising activities in communities involved speaking to community leaders, at community gatherings (including women’s and youth meetings), and in church meetings. At the same time, a media campaign was run involving brief messages on the radio, and use of posters and leaflets. VHWs received a 1 day refresher training 6 months after the initial training, and maintained a relationship with the psychiatric nurse. This intervention has been described previously in previous published work [17] and a programme report [21].

The study used the phased roll-out of the programme to measure impact during the MHAP implementation in Imo State (intervention), and used Anambra States as a control during the year prior to implementation there. We selected Imo as the intervention state (population 3.9 million, 21 active clinics) and Anambra State (population 4.2 million, 21 clinics) as the control because they have similar demography and health systems, and the CMHP services were identical, including VHW presence.

Routinely available attendance data from all clinics in the service in the intervention and control state were collected (42 in total). Unfortunately, there was a nurse strike in our control state (Anambra) during the period where we needed to collect service use data, so we were unable to compare concurrent data using comparison of mean attendance with the t test. We adjusted the analysis instead to rely on a comparison with data from Imo prior to the intervention.

The statistical analysis had to be adjusted to take account of the lack of a control site. The negative binomial model used is an extension to the Poisson model for analyzing count data in that it allows for the over-dispersion of zeros that would be uncharacteristic of the more familiar Poisson distribution. The results however can be interpreted in the normal fashion in the form of incident rate ratios (IRR). In the context of the model above, IRRs higher than 1 indicate an increased risk of the event occurring (i.e. higher number of new patients) and IRRs lower than 1 indicates a lower rate of new patients.

The model is adjusted for an underlying baseline risk that has been tested for non-linearity but appears to be linear in nature (i.e. constant). With increasing months, the risk of increased number of new patients is 1.000 indicating that the risk does not change with time (i.e. the gradient of the slope is 0, flat). Thus there was no significant trend for the number of new patients seen from beginning of the data collection up to and including June 2011.

The incident rate for the initial period of the intervention was 5.1 times higher than the incident rate for the baseline rate (which we have shown to be a flat trend) and this increase was highly statistically significant (95% CI 3.42–7.56, p < 0.001).

This initial significant effect quickly diminished, and the intervention effect reduced before leveling off above the initial baseline (see Fig. 1).

Figure 1 shows the expected results calculated from the first model above. The fitted regression line (maroon) closely fits the observed data (black dash line) that was seen previously in Fig. 2, and closely describes the change in the number of new patients seen each month. The red line shows fitted results unadjusted by site for reference.

This shape of curve is seen frequently following an intervention whereby the initial effect is not sustained and eventually levels at a smaller effect or returns to the baseline. Unfortunately we do not have enough points of data collection after the intervention to estimate the stable intervention effect as the rate of new patients seen each month may or may not be changing by the end of data collection. Also, local fluctuations towards the end of the data collection appear slightly increased, but the figures fall within the range of fluctuations seen in the data.

Another way of describing the intervention effect is to use this regression equation to see at which points following the intervention is the rate of new patients seen higher than the baseline rate. See Table 1.

Table 1 compares the intervention effect at each time-point and compares it to the baseline rate of new patients seen per month. As described previously, the rate of new patients seen in the initial intervention was 5.09 times higher than the baseline. By the following month, the number of new patients seen per month was 4.18 times higher than the baseline (p < 0.001). Up to and including the 10th month following the intervention, the number of new patients seen per month was higher than the baseline rate. Following this, the rate of new patients seen each month remains above baseline until the end of data collection, but the results are not statistically significant for a period, before tending in this direction again.