Factors associated with quality of services for marginalized groups with mental health problems in 14 European countries

The QISO was developed to facilitate identification of organisational good practice in the context of providing mental health care for socially marginalized people. Its components were defined by the multidisciplinary team of experts involved in the PROMO consortium. The experts’ professional backgrounds were in mental health and social care, public health and social sciences, encompassing both clinical and research expertise. The team of experts discussed and refined each potential quality indicator and its contribution to the overall index score until a consensus was reached on a final set. Evidence generated within the scope of this and other projects in which participating experts were involved was taken into account when developing the QISO [15, 16]. This, in turn, resulted in different weightings of each component as a reflection of their importance to the provision of care to marginalised groups. An emphasis was put on self-referrals as the overall service accessibility and networking were highlighted in other PROMO data and in the findings of previous studies on the provision of care in the context of marginalisation. Clinicians working in deprived areas struggle to find adequate services to provide relevant care to the individuals from marginalised groups, with service coordination often being insufficient [13, 15]. Amongst the four components of good practice identified in 154 interviews with experts from the 14 capital cities, three directly relate to access and referrals, specifically, facilitating access to services that provide different aspects of health care (reducing the need for further referrals), strengthening the collaboration and co-ordination between different services, and disseminating information on services both to marginalised groups and to practitioners in the area [13].

Therefore, information concerning service organisation comprised indicators covering six domains, with final organisation scores ranging from 0 to a possible maximum score of 15. Quality provision domains and their contribution to the overall score were: accessibility (8), supervision (1), multidisciplinary team (1), programmes provided (2), coordination (1) and evaluation (2).

Quality indicators within each domain correspond to specific service characteristics and account for up to two points of the score as detailed in Table 2. Accessibility includes indicators on service opening hours, the existence of exclusion criteria for clients, and accepting self-referrals. Supervision refers to the provision of internal or external staff supervision of any type. Multidisciplinary team is defined as having staff with both mental health and social care professional backgrounds. Programmes provided refers to active outreach programmes and/or home visits to clients as well as case-finding. Coordination refers to services having routine meetings with other services. Finally, Evaluation includes indicators on recording data on input and attendance, as well as data on client satisfaction.

In addition to service characteristics, which correspond to the indicators of quality of service organisation, a number of other service features were recorded during the PROMO assessments. In the current analysis, the total number of staff (measured in whole time equivalents, with the number of hours per week defined by each respondent according to his/her national norm) and the number of care programmes provided were used as service-level covariates, due to their importance to the quality of health provision, as asserted in the relevant literature [17], including mental health care studies [18]. Programmes were defined as specific health care or social interventions that each service potentially provides to their clients. Each service was assessed using a specific list of programmes: active outreach, case-finding, home visits, counselling, individual psychotherapy, group psychotherapy, self-help groups, occupational therapy, medication, detoxification and acute withdrawal treatment, drug addiction treatment, alcohol addiction treatment, direct practical help in clients’ homes, befriending, leisure activities support, mental health advocacy, social welfare support, housing/accommodation advice and support, legal advice and support, job coaching/finding, mental health promotion measures and any other programmes specified by the service being assessed.

Three Eurostat country-level socioeconomic indicators were included and tested: the country Gross Domestic Product (GDP), the Material Deprivation rate and the Gini coefficient. The GDP is a commonly used measure for assessing a country’s wealth or socioeconomic status, while the Gini coefficient is a measure of income inequality which has been correlated with the prevalence of poor health outcomes and mental disorders [19]. The Material deprivation rate was also chosen because of its direct relevance to the marginalized groups studied, and is considered as an ecological measure of country’s burden of social marginalization [20‐22].

The Gross Domestic Product per capita in Purchasing Power Standards (PPS) (2008) has been defined by Eurostat as the value of all goods and services produced less the value of any goods or services used in their creation. The volume index of GDP per capita in Purchasing Power Standards is expressed in relation to the European Union (EU-27) average set to equal 100. A country index higher than 100 corresponds to GDP per capita higher than the EU average. Basic figures are expressed in PPS, a common currency that eliminates differences in price levels between countries, thus allowing meaningful volume comparisons of GDP between countries. This index is intended for cross-country rather than for temporal comparisons.

The Gini coefficient (2008) as a measure of income inequality is conceptualised as the relationship of cumulative shares of the population arranged according to the level of equalized disposable income, to the cumulative share of the equalized total disposable income received by them. The higher the Gini coefficient, the more inequality exists.

The Material Deprivation rate by poverty status (2008) is the percentage of the population with an enforced lack of at least three out of nine material deprivation items depicting material living conditions, such as housing conditions, possession of durables, and capacity to afford basic requirements [23]. The term ‘enforced lack’ refers to people wishing to possess items, but not being able to afford them and the items in question are part of a predefined ‘economic strain and durables’ dimension. ‘Economic strain’ refers to people not being able to afford to do things they would like to do, such as taking a week’s annual holiday away from home, paying a mortgage, rent, utility bills or hire purchase instalments, having a meal with meat, chicken or fish every second day, keeping their home adequately warm, or being able to face unexpected expenses. The durables dimension corresponds to enforced lack of items such as a colour TV, a telephone, a personal car or a washing machine [24].

The overall QISO was normally distributed, with a mean (SD) of 10.03 (2.13) (Figure 1).

This exploratory factor analysis revealed a 5-factor model as the solution with best fit, generically supporting the theoretical domains for the quality indicators proposed [see Additional file 1].

The mean (SD) number of staff per service across all participating countries was 33.69 (124.28), (Table 3). Services in Hungary had the highest mean (SD) number of staff, 499.20 (885.51), although with very few services contributing to that value. The country with the lowest mean (SD) number of programmes provided per service was Austria with 5.85 (3.00); Poland had the highest with 10.43 (3.84) programmes per service.

According to Eurostat for the year 2008, national GDP ranged from 134 for the Netherlands to 56 for Poland. The Gini coefficient varied between 35.8 in Portugal and 24.0 in Sweden. The Material deprivation rate was highest in Hungary (37.1) and Poland (32.3) and lowest in Sweden (4.6) and the Netherlands (5.2).

As shown in Table 4, both the number of staff per service, the number of programmes provided per service and the Gini coefficient correlated positively with the QISO across all countries while the GDP correlated negatively. There were no significant differences for mean QISO between generic and specific services: t(591) = -0.77, p = 0.44; nor between mental health care, social care and general health care services: F(2,590) = 1.32, p = 0.27, nor when considering all six types of services: F(5, 587) = 0.83, p = 0.53.

Figure 2 shows the relation of the total number of programmes with the QISO score, in the overall sample. As shown in Figure 3, the intercept and slope of the fitted regression line varies, indicating that the relationship between the QISO and the average number of programmes provided per service varies from one country to the next (Table 4).

As shown in Table 5, in the null random effects model, 14.8% of the variance was explained by country-level traits, as expressed by the Interclass Correlation Coefficient (ICC). The intercept in the empty model was equal to the overall average QISO score, which for this sample was 10.20; the variance component corresponding to the random intercept was 0.67.

When adding the service-level variables using Model 1 (total number of programmes, total number of staff), the percentage of variance in QISO attributable to country-level variables was 13.2%; in Model 2, the addition of the Gini coefficient and the Material deprivation rate did not represent a marked difference in the ICC (12.3%); only in Model 3 with the addition of the GDP, did the ICC drop to 2.7%, i.e. the country GDP played a role in explaining differences in the QISO independently of the number of staff and programmes, the country Gini coefficient and the Material deprivation rate.

Considering only statistically significant service-level variables (total number of programmes) in the linear regression model, the same proportional decrease in the variance attributable to country-level variables was observed (results not shown).

As shown in Table 6, a decrease in the total variance observed in QISO scores across countries due to country-level variables was observable for all groups of services, although this was more evident for generic services, with a decrease of nearly 8% with the addition of GDP per capita: the ICC dropped from 21.6% to 13.8%. This percentage remained above 20% in the final model for mental health care services and reached 28.0% for general health care services. Only 1.1% of the variance remained attributable to country-level features for social care services from the initial 7.4% in the null model.

In summary, country GDP is important for explaining differences in QISO scores across all countries, independently of the number of programmes each service was providing. This remains true regardless of service typology, although the trend is more evident in generic services, i.e. services not specifically focussing on any particular marginalised groups.