Defining the urban area for cross national comparison of health indicators: the EURO-URHIS 2 boundary study.

Introduction: Despite much research focusing on the impact of the city condition upon health, there still remains a lack of consensus over what constitutes an urban area (UA). This study was conducted to establish comparable boundaries for the UAs participating in EURO-URHIS 2, and to test whether the sample reflected the heterogeneity of urban living. Key UA contacts ( n = 28) completed a cross-sectional questionnaire, which included where available comparison between Urban Audit city and larger urban zone (LUZ) boundaries and public health administration areas (PHAAs). Additionally, broad health and demographic indicators were sought to test for heterogeneity of the EURO-URHIS 2 sample. Urban Audit city boundaries were found to be suitable for data collection in 100% ( n = 21) of UAs where Urban Audit data were available. The remainder ( n = 7) identified PHAA boundaries akin to the 'city' level. Heterogeneity was observed in the sample for population size and infant mortality rate. Heterogeneity could not be established for male and female life expectancy. This study was able to establish comparable boundaries for EURO-URHIS 2 data collection, with the 'city' area being selected for data collection. The homogeneity of life expectancy indicators was reflective of sub-regional similarities in life expectancy, whilst population estimates and rates of infant mortality indicated the presence of heterogeneity within the sample. Future work would trial these methods with a larger number of indicators and for a larger number of UAs.

Collecting urban health indicators from routinely available sources: development and piloting of a tool for the collection of existing data for EURO-URHIS 2.

Introduction: EURO-URHIS 2 aimed to collect comparable health indicators across a large number of urban areas (UAs) across Europe and Vietnam using four data collection tools. This paper outlines the process for the selection of indicators to be collected from routinely available sources, and the piloting of the data collection tool. A long-list of indicators potentially collectable from routinely available sources was generated by the EURO-URHIS 2 consortium. Key contacts from each UA completed an e-mail survey reporting for each indicator whether it could be collected using the given definition, an alternative definition or not at all. Additionally participants listed the 20 leading causes of death for their UAs from the Eurostat 65. Results were compiled to inform indicator selection for the main data collection phase. Responses were received for 25 of 28 eligible UAs. Of the 29 proposed indicators, 55.1% ( n = 16) were accepted without change, 24.1% ( n = 7) were re-allocated to other data collection tools and 17.2% ( n = 5) were accepted after a modification of the EURO-URHIS 2 definition. This scoping exercise and piloting phase for the 'existing data tool' for the project was useful and informative. It provided detailed information on what could be collected, and an opportunity to modify indicator definitions to maximize response rates. These results are only applicable to those UAs returning results and cannot be generalized. Detailed interrogation of definitions is essential to this sort of data collection, and the process described was designed with cross-national comparability in mind.

Elderly from lower socioeconomic groups are more vulnerable to mental health problems, but area deprivation does not contribute: a comparison between Slovak and Dutch cities.

Background: Little is known about factors associated with mental health problems (MHP) of the elderly in socioeconomically deprived neighbourhoods, and comparisons between Central European and Western European countries on this topic are lacking. We examined whether MHP occurred more frequently in deprived neighbourhoods and among deprived people. Next, we examined whether the association of MHP with area deprivation differed by country and whether this could be explained by the socioeconomic (SE) characteristics of the residents. We obtained data on non-institutionalized residents aged 65 years and above from the EU-FP7: EURO-URHIS 2 project from Slovak ( N = 665, response rate 44.0%) and Dutch cities ( N = 795, response rate 50.2%). An elevated score on General Health Questionnaire-12 (≥2) indicated MHP. Education and household income with financial strain were used as measures of individual SE status. We employed multilevel logistic regression. Overall rates of MHP were significantly higher in Slovakia (40.6%) than in the Netherlands (30.6%). The neighbourhood unemployment rate was not associated with the mental health of elderly in either country. Rates of MHP were significantly higher among elderly with low and medium income [odds ratio (OR) = 1.75, 95% confidence interval (CI) = 1.16-2.62; OR = 1.64, 95% CI = 1.12-2.41, respectively] and financial strain (OR = 2.26, 95% CI = 1.56-3.28) when compared with those with high income and no strain, respectively. Individual-level SE characteristics explained differences between the two countries. The risk of MHP among the elderly is associated with their individual-level SE position but not with neighbourhood deprivation in both Slovakia and the Netherlands.

Are area-level and individual-level socioeconomic factors associated with self-rated health in adult urban citizens? Evidence from Slovak and Dutch cities.

Background: Evidence shows that living in disadvantaged areas is associated with poor health. This may be due to the socioeconomic (SE) characteristics of both these residents and the areas where they live. Evidence regarding this on Central European (CE) countries is scarce. Our aim was to assess whether the prevalence of poor self-rated health (SRH) was higher in deprived urban areas, whether this can be explained by individual SE status (SES) and whether this differed between Slovakia and the Netherlands per age group. We examined the association of urban-level data and individual-level SE factors from different urban areas in different countries (Slovakia, the Netherlands) using comparable urban health indicators and area indicators. We also obtained unique data from the EU-FP7 EURO-URHIS 2 project. Multilevel logistic regression showed that poor SRH was associated with area deprivation in both countries. Regarding age by country, poor SRH occurred more frequently in the more deprived areas for the younger age group (≤64) in the Netherlands but for the older age group (≥65 years) in Slovakia. Moreover, Slovak citizens reported poor SRH significantly more often than Dutch residents. Individual SES was significantly associated with poor SRH in both age groups and both countries for most area-level SE measures. Individual SES is associated with SRH more strongly than area deprivation. Therefore, it is important to account for relative deprivation at an individual level when considering health-enhancing activities. Moreover, the effect of urban-area deprivation seems to differ between CE and WE countries.

Collecting standardised urban health indicator data at an individual level for adults living in urban areas: methodology from EURO-URHIS 2.

Background: An aim of the EURO-URHIS 2 project was to collect standardised data on urban health indicators (UHIs) relevant to the health of adults resident in European urban areas. This article details development of the survey instruments and methodologies to meet this aim. 32 urban areas from 11 countries conducted the adult surveys. Using a participatory approach, a standardised adult UHI survey questionnaire was developed mainly comprised of previously validated questions, followed by translation and back-translation. An evidence-based survey methodology with extensive training was employed to ensure standardised data collection. Comprehensive UK piloting ensured face validity and investigated the potential for response bias in the surveys. Each urban area distributed 800 questionnaires to age-sex stratified random samples of adults following the survey protocols. Piloting revealed lower response rates in younger males from more deprived areas. Almost 19500 adult UHI questionnaires were returned and entered from participating urban areas. Response rates were generally low but varied across Europe. The participatory approach in development of survey questionnaires and methods using an evidence-based approach and extensive training of partners has ensured comparable UHI data across heterogeneous European contexts. The data provide unique information on health and determinants of health in adults living in European urban areas that could be used to inform urban health policymaking. However, piloting has revealed a concern that non-response bias could lead to under-representation of younger males from more deprived areas. This could affect the generalisability of findings from the adult surveys given the low response rates.

Differences in adults' health and health behaviour between 16 European urban areas and the associations with socio-economic status and physical and social environment.

Background: With a growing proportion of the European population living in urban areas (UAs), exploring health in urban areas becomes increasingly important. The objective of this study is to assess the magnitude of differences in health and health behaviour between adults living in urban areas (UAs) across Europe. We also explored whether and to what extent such differences can be explained by socio-economic status (SES) and physical or social environment. Data were obtained from a cross-sectional questionnaire survey, performed between as part of the European Urban Health Indicator System Part 2 (EURO-URHIS 2) project. Using multi-level logistic regression analysis, UA differences in psychological distress, self-assessed health, overweight and obesity, daily smoking, binge drinking and physical exercise were assessed. Median Odds Ratios (MORs) were calculated to estimate the extent to which the observed variance is attributable to UA, individual-level SES (measured by perceived financial strains, education level and employment status) and/or characteristics of physical and social environment. The dataset included 14 022 respondents in 16 UAs within 9 countries. After correction for age and gender, all MORs, except that for daily smoking, indicated statistically significant UA health differences. SES indicators (partly) explained UA differences in psychological distress, decreasing the MOR from 1.43 [95% credible interval (Cr.I.) 1.27-1.67, baseline model], to 1.25 (95% Cr.I. 1.14-1.40, SES model): a reduction of 42%. Accounting for the quality of green areas reduced the MOR for psychological distress by an additional 40%, to 1.15 (95% Cr.I. 1.05-1.28). Our study showed large differences in health and health behaviour between European UAs. Reducing socio-economic disadvantage and improving the quality of the neighbourhood's green spaces may reduce UA differences in psychological distress.

Developing a European urban health indicator system: results of EURO-URHIS 1.

Introduction: More than half of the world's population now live in cities, including over 70% in Europe. Cities bring opportunities but can be unhealthy places to live. The poorest urban dwellers live in the worst environments and are at the greatest risk of poor health outcomes. EURO-URHIS 1 set out to compile a cross-EU inventory of member states use of measures of urban health in order to support policymakers and improve public health policy. Following a literature review to define terms and find an appropriate model to guide urban health research, EURO-URHIS Urban Areas in all EU member states except Luxembourg, as well as Croatia, Turkey, Macedonia, Iceland and Norway, were defined and selected in collaboration with project partners. Following piloting of the survey tool, a the EURO-URHIS 45 data collection tool was sent out to contacts in all countries with identified EUA's, asking for data on 45 Urban Health Indicators (UHI) and 10 other indicators. 60 questionnaires were received from 30 countries, giving information on local health indicator availability, definitions and sources. Telephone interviews were also conducted with 14 respondents about their knowledge of sources of urban health data and barriers or problems experienced when collecting the data. Most participants had little problem identifying the sources of data, though some found that data was not always routinely recorded and was held by diverse sources or not at local level. Some participants found the data collection instrument to not be user-friendly and with UHI definitions that were sometimes unclear. However, the work has demonstrated that urban health and its measurement is of major relevance and importance for Public Health across Europe. The current study has constructed an initial system of European UHIs to meet the objectives of the project, but has also clearly demonstrated that further development work is required. The importance and value of examining UHIs has been confirmed, and the scene has been set for further studies on this topic.

The effect of neighbourhood unemployment on health-risk behaviours in elderly differs between Slovak and Dutch cities.

BACKGROUND: Health-risk behaviours (HRB) increase the risk of disability and chronic diseases at an older age. This study aimed to compare Slovakia and the Netherlands regarding differences in the prevalence of HRB by neighbourhood and individual deprivation and to determine whether area differences could be explained by the socio-economic position (SEP) of the residents. METHODS: We obtained data on non-institutionalized residents aged ≥ 65 years from the EU-FP7: EURO-URHIS 2 project from Slovak (N = 665, response rate 44.0%) and Dutch cities (N = 795, response rate 50.2%). HRB concerned daily smoking, binge drinking, physical activity, consumption of fruits and vegetables and body mass index. Area deprivation was measured by the neighbourhood unemployment rate. Individual SEP was measured by education and household income with financial strain. We used multilevel logistic regression. RESULTS: In Slovakia, no HRB was associated with either neighbourhood unemployment or individual SEP. The elderly in the Netherlands from the least favourable neighbourhoods were more likely to be daily smokers [odds ratio (OR) 2.32; 95% confidence interval (CI) 1.25, 4.30] and overweight (OR 1.84; 95% CI 1.24, 2.75) than residents from the most favourable ones. For the Dutch elderly the gradients varied per HRB and per individual-level SEP indicator. Individual SEP explained country differences in the association of area unemployment with smoking and lack of physical activity but not that with overweight. CONCLUSION: Countries differed in the associations with HRB of both neighbourhood unemployment and individual SEP among the elderly urban residents. The local importance of socio-economic factors on both levels should be considered when developing health-promotion activities for the elderly.

Neighborhood ethnic density and suicide risk among different migrant groups in the four big cities in the Netherlands.

PURPOSE: Recent studies suggested a favorable association between the ethnic density of the neighborhood and the risk of psychotic disorders among ethnic minorities. In this study, it was investigated whether this so-called 'ethnic density hypothesis' is also relevant to suicide risk, which is not sensitive to bias associated with ethnic differences in access to health care and reflects a broad range of mental health problems. METHODS: Suicides in the four big cities in the Netherlands during 2000-2011 were ascertained using the cause of death register of Statistics Netherlands and analyzed in a multilevel Poisson model in relation to individual- and neighborhood-level characteristics. RESULTS: With increasing non-Western minority density, the adjusted rate ratio (RR) of suicide in non-Western immigrants compared to native Dutch persons decreased from 0.69 to 0.39 (P < 0.001). This was explained by higher suicide rates among Dutch persons (RR = 1.28, P = 0.048) and lower rates among non-Western persons (RR = 0.72, P = 0.004) in neighborhoods with high (>55.9 %) compared to neighborhoods with low non-Western minority density (<36.5 %). Similar results were found for Turkish, Moroccan, Surinamese/Antillean and other non-Western subgroups separately. Compared to personally matched controls, non-Western cases (i.e., those who committed suicide) more often moved house to own-group high-dense areas and less often to own-group low-dense areas in the 5 years prior to suicide. CONCLUSIONS: Our findings support the beneficial influence of the presence of the own ethnic group in the neighborhood on suicide risk among non-Western minorities. As moving to minority more dense areas prior to suicide was observed, this influence of ethnic density as measured on population level may have been underestimated.

Ethnic density is not associated with psychological distress in Turkish-Dutch, Moroccan-Dutch and Surinamese-Dutch ethnic minorities in the Netherlands.

PURPOSE: Ethnic density, the proportion of people of the same ethnic group in the neighbourhood, has been identified as a protective factor with regard to mental health in ethnic minorities. Research on the putative intermediating factors, exposure to discrimination and improved social support, has not yielded conclusive evidence. We investigated the association between ethnic density and psychological well-being in three ethnic minority groups in the Netherlands. We also assessed whether a protective ethnic density effect is related to the degree to which each group experiences discrimination and social support at group level. METHODS: Using multi-level linear regression modelling, we studied the influence of ethnic density at neighbourhood level on psychological distress, measured with the Kessler Psychological Distress scale (K10), in 13,864 native Dutch, 1,206 Surinamese-Dutch, 978 Turkish-Dutch and 784 Moroccan-Dutch citizens of the four major cities in the Netherlands. Based on a nationwide survey among ethnic minorities on social integration, ethnic groups were ordered with respect to the intermediating factors. RESULTS: Ethnic density was not associated with psychological distress in any of the three ethnic minority groups. As a consequence, we found no support for either experiences of discrimination or for own-group social interactions at group level as intermediating factors. In all three ethnic minority groups, as well as in the native Dutch group, individual demographic and socio-economic factors emerged as the main explanations for individuals' mental well-being. CONCLUSIONS: These results suggest that individual demographic and socio-economic risk characteristics outweigh the influence of neighbourhood attributes on mental health.

Stressors and resources mediate the association of socioeconomic position with health behaviours.

BACKGROUND: Variability in health behaviours is an important cause of socioeconomic health disparities. Socioeconomic differences in health behaviours are poorly understood. Previous studies have examined whether (single) stressors or psychosocial resources mediate the relationship between socioeconomic position and health or mortality. This study examined: 1) whether the presence of stressors and the absence of resources can be represented by a single underlying factor, and co-occur among those with lower education, 2) whether stressors and resources mediated the relation between education and health behaviours, and 3) addressed the question whether an aggregate measure of stressors and resources has an added effect over the use of individual measures. METHODS: Questionnaire data on sociodemographic variables, stressors, resources, and health behaviours were collected cross-sectionally among inhabitants (n = 3050) of a medium-sized Dutch city (Utrecht). Descriptive statistics and bootstrap analyses for multiple-mediator effects were used to examine the role of stressors and resources in mediating educational associations with health behaviours. RESULTS: Higher levels of stressors and lower levels of resources could be represented by a single underlying factor, and co-occurred among those with lower educational levels. Stressors and resources partially mediated the relationship between education and four health- behaviours (exercise, breakfast frequency, vegetable consumption and smoking). Financial stress and poor perceived health status were mediating stressors, and social support a strong mediating resource. An aggregate measure of the stressors and resources showed similar associations with health behaviours compared to the summed individual measures. CONCLUSIONS: Lower educated groups are simultaneously affected by the presence of various stressors and absence of multiple resources, which partially explain socioeconomic differences in health behaviours. Compared to the direct associations of stressors and resources with health behaviours, the association with socioeconomic status was modest. Therefore, besides addressing structural inequalities, interventions promoting financial management, coping with chronic disease, and social skills training have the potential to benefit large parts of the population, most notably the lower educated. Further research is needed to clarify how stressors and resources impact health behaviours, why this differs between behaviours and how these disparities could be alleviated.

Effect of nutritional counselling and nutritional plus exercise counselling in overweight adults: a randomized trial in multidisciplinary primary care practice.

OBJECTIVE: To compare the effects of nutritional counselling with nutritional plus exercise counselling on body weight and waist circumference in overweight adults in a multidisciplinary primary care setting. METHODS: One hundred and thirty-four overweight adults (body mass index 28-35) were randomly assigned to individual counselling sessions by a dietician (D) or counselling sessions by a dietician plus physiotherapist (D + E) during 6 months with one follow-up session at 12 months. Outcomes were assessed at baseline, 6 and 12 months. Difference in changes of outcome measures between groups were analysed using generalized estimating equations. RESULTS: Weight reduced from baseline to 6 months in D [-2.2 (-3.1 to -1.4) kg] and D + E [-3.0 (-4.0 to -2.0) kg] and was sustained at 12 months [-2.0 (-3.1 to -1.4) kg and -3.1 (-4.5 to -1.6) kg, respectively]. The reduction in weight did not significantly differ between D and D + E (P = 0.48). In both groups, waist circumference decreased from baseline to 6 months [-2.1 (-3.3 to -0.8) cm for D; -3.7 (-5.1 to -2.3) cm for D + E] and was sustained at 12 months [-2.1 (-3.5 to -0.7) cm and -4.2 (-6.0 to -2.5) cm, respectively]. Participants in D + E tended to decrease their waist circumference more than those in D (P = 0.14). DISCUSSION: Nutritional counselling by a dietician resulted in modest reductions in weight and waist circumference in overweight adults, which were sustained up to 12 months. Adding exercise counselling by a physiotherapist did not significantly enhance the effect on body weight. Exercise counselling may, however, further improve waist circumference.

Transitions in drug use in a new generation of problem drug users in Amsterdam: a 6-year follow-up study.

AIMS: This study describes the transitions in drug use in Amsterdam among young drug users (YDUs) who are inhaling or injecting cocaine or using illicitly obtained opiates. METHODS: From 2000 until 2003, 187 YDUs (< or = 30 years) were recruited of whom 126 were followed into 2007. RESULTS: During the 6 months prior to inclusion, 95% used cocaine, 60% used illicit opiates, and 9% injected more than once a week. During follow-up, the incidence of new-onset heroin use was 4.8/100 person-years. The incidence of injecting was much lower for never-injectors (2.1/100 person-years) than for ever-injectors who relapsed (13.2/100 person-years). Transition to abstinence or nonfrequent drug use is common (23/100 person-years), mostly followed by a relapse (73/100 person-years). Polydrug users were less likely to discontinue their frequent drug use than monodrug users. CONCLUSIONS: Treatment services should target the crack cocaine users to prevent further marginalization and, during abstinence, to prevent relapse. Although the prevalence and incidence of injecting are relatively low, it is still an option for opiate users, especially those with a history of injecting.

Prediction of relapse to frequent heroin use and the role of methadone prescription: an analysis of the Amsterdam Cohort Study among drug users.

The risk of relapse into frequent heroin use was studied among 732 participants of the Amsterdam Cohort Study (ACS) on HIV/AIDS among drug users, who experienced an episode of abstinence from or occasional use of heroin. Participants of the ACS were recruited primarily from easy access ("low-threshold") methadone programs. The duration of abstinence/occasional use and relative risks (RR) of relapse were estimated by analyzing 1577 episodes by means of survival analysis using characteristics of patients and methadone treatment as covariates. The majority of episodes (85.8%) were followed by relapse within 5 years. Less education, intense use of heroin prior to the episode of abstinence or well-controlled use, occasional use of heroin and intense use of cocaine during the episode, and having a drug-using partner or having no partner were significantly associated with a higher risk of relapse. Among frequent attendees of a "low-threshold" methadone program, relapse was associated with the daily dose of methadone: RR for dosages <40 and 40-60 mg, compared with doses of >100mg, were 1.45 (P<0.01) and 1.59 (P<0.01), respectively. No beneficial influence was revealed of methadone dosage or program attendance in itself on the risk of relapse into cocaine. High doses of methadone in a harm-reduction setting extend the duration of an episode of no or occasional use of heroin. Other factors, such as no occasional use of heroin during the episode, no use of cocaine, and having a non-using partner, seem to be equally important.

HIV prevalence and risk behaviour in young drug users in Amsterdam.

OBJECTIVES: To assess the HIV prevalence and risk behaviour among regular young drug users (YDU) in Amsterdam. DESIGN: Subjects were 282 YDU (using heroin, cocaine, methadone and/or amphetamines at least 3 days/week) aged 30 years or less who participated in a cross-sectional survey in Amsterdam in 1998. Participants were recruited both directly (41%) through healthcare institutions, and indirectly (59%) by referrals from YDU already enrolled. A comparison sample of YDU from the Amsterdam Cohort Study recruited between 1985 and 1989 was used to describe trends in HIV prevalence and risk behaviour. RESULTS: Mean age was 25 years, 22% were women, 39% had at least once injected drugs, and 22% currently injected in 1998. Cocaine was more often used regularly (70%) than heroin (42%). Of the 282 YDU, 21 (7.4%) were infected with HIV. Among those who had injected drugs, HIV prevalence was 16.2%. Among those who had never injected, the prevalence of HIV infection was 1.8%. A history of injecting declined from 83% in 1985-1989 to 56% in 1998. Compared with 1985-1989, the lack of a recent steady partner appeared to be an important risk factor for a positive HIV status among injectors in 1998, whereas borrowing of used needles/syringes was no longer a risk factor. CONCLUSION: The use of cocaine among YDU is very popular in Amsterdam. A strong decrease in injecting behaviour has taken place, but among those who inject, risk behaviour is still considerable, and HIV infection is still widespread. Preventative activities should be expanded, specifically among YDU.

Amount of self-reported illicit drug use compared to quantitative hair test results in community-recruited young drug users in Amsterdam.

AIMS: To assess the dose-effect relationship between self-reported drug intake and the concentration of drugs and/or their metabolites in hair and to examine factors that may mediate this relationship. DESIGN AND SETTING: A cohort study among young drug users (YDU) in Amsterdam, the Netherlands, which began in July 2000. At intake, YDU were asked to report their average drug intake over a 2-month period. A hair sample was taken and then analysed for cocaine, benzoylecgonine (BE), morphine, 6-monoacetylmorphine and methadone. Weighted least-squares regression analysis was used to model hair-test results as a function of reported drug use. PARTICIPANTS: Subjects were 95 YDU (using cocaine, heroin, methadone and/or amphetamines at least 3 days/week) aged 18-30 residing in Amsterdam in 2000-2001. FINDINGS: Of the 95 YDU, one-third were women; mean age was almost 26; 30% had black hair, 33% blond hair and 37% brown hair. Cocaine use was reported by 92%, heroin by 75% and methadone by 64% of participants. All hair samples contained one or more drugs. Crude correlation coefficients between reported drug doses and drug concentrations in hair ranged between 0.45 and 0.59. The multivariate regression analysis showed that, for one or more types of drug, black-haired people, women and non-western European people had relatively high drug concentrations in hair (significant slope effects). The corresponding multivariate correlation coefficients ranged between 0.63 and 0.87. CONCLUSIONS: Hair testing can be used to quantify drug use in epidemiological studies, given that factors such as hair colour and sex are taken into account.

Mortality and psychiatric disorders among public mental health care clients in Utrecht: a register-based cohort study.

BACKGROUND: Different studies have shown similar or even lower mortality among homeless persons with compared to homeless persons without a severe mental disorder. AIMS: To clarify the association between presence of a psychiatric diagnosis and mortality among the socially marginalized. METHODS: The Public Mental health care (PMHc) is a legal task of the municipal authority aiming at prevention and intervention in case of (imminent) homelessness among persons with a serious shortage of self-sufficiency. The data of PMHc clients (N=6,724) and personally matched controls (N=66,247) were linked to the registries of Statistics Netherlands and analysed in a Cox model. RESULTS: The increased mortality among PMHc clients, compared to the general population (HR=2.99, 95%-CI: 2.63-3.41), was associated with a broad range of death causes. Clients with a record linkage to the Psychiatric Case Registry Middle Netherlands ('PMHc+') had an increased risk of suicide (HR=2.63, 0.99-7.02, P=0.052), but a lower risk of natural death causes (HR=0.71, 0.54-0.92, P=0.011), compared to clients without this record linkage ('PMHc-'). Compared to controls, however, 'PMHc-' clients experienced substantially increased risks of suicide (HR=3.63, 1.42-9.26, P=0.007) and death associated with mental and behavioural disorders (ICD-10 Ch.V) (HR=7.85, 3.54-17.43, P<0.001). CONCLUSION: Psychiatric services may deliver an important contribution to the prevention of premature natural death among the socially marginalized. KEYPHRASES: The earlier observed lower mortality among vulnerably housed and homeless persons with a psychiatric diagnosis compared to vulnerably housed and homeless persons without a psychiatric diagnosis appears to be due to a significantly lower risk of natural causes of death. Compared to controls from the general population, vulnerably housed and homeless persons without registered diagnosis at a local psychiatric service have a significantly increased mortality associated both with natural death causes and with suicide and death due to mental and behavioural disorders. Services for mental health care may deliver an important contribution to the prevention of premature death due to somatic disorders among the socially marginalized.

Measuring social exclusion in routine public health surveys: construction of a multidimensional instrument.

INTRODUCTION: Social exclusion is considered a major factor in the causation and maintenance of health inequalities, but its measurement in health research is still in its infancy. In the Netherlands the Institute for Social Research (SCP) developed an instrument to measure the multidimensional concept of social exclusion in social and economic policy research. Here, we present a method to construct a similar measure of social exclusion using available data from public health surveys. METHODS: Analyses were performed on data from the health questionnaires that were completed by 20,877 adults in the four largest cities in the Netherlands. From each of the four questionnaires we selected the items that corresponded to those of the SCP-instrument. These were entered into a nonlinear canonical correlation analysis. The measurement properties of the resulting indices and dimension scales were assessed and compared to the SCP-instrument. RESULTS: The internal consistency of the indices and most of the dimension scales were adequate and the internal structure of the indices was as expected. Both generalisabiliy and construct validity were good: in all datasets strong associations were found between the index and a number of known risk factors of social exclusion. A limitation of content validity was that the dimension "lack of normative integration" could not be measured, because no relevant items were available. CONCLUSIONS: Our findings indicate that a measure for social exclusion can be constructed with available health questionnaires. This provides opportunities for application in public health surveillance systems in the Netherlands and elsewhere in the world.

Health-risk behaviours in deprived urban neighbourhoods: a comparison between Slovak and Dutch cities.

OBJECTIVES: International comparisons of the associations of area-level socioeconomic position (SEP) and health-risk behaviours (HRBs) are for the most part lacking. The aims of this study were to compare Slovakia and the Netherlands regarding differences in the prevalence of HRBs by neighbourhood and individual deprivation, and to determine whether area differences could be explained by the SEP of residents. METHODS: We obtained data on residents aged 19-64 from Slovak and Dutch cities from the FP7 EURO-URHIS2 project and employed multilevel logistic regression. RESULTS: The association between neighbourhood-level unemployment and HRBs differed between countries. In the Netherlands, the prevalence of almost all HRBs was higher in deprived areas, except for the consumption of fruits and vegetables. These area effects diminished after controlling for individual-level SEP. In Slovakia, no area effects were observed, although Slovak residents showed a higher risk for most HRBs. At the individual level, an inverse SE gradient was found for almost all HRBs in both countries. CONCLUSIONS: Local analyses of small area health differences and health determinants are critical for efficient implementation of neighbourhood-based interventions.