Place effects for areas defined by administrative boundaries: a life course analysis of mortality and cause specific morbidity in Scania, Sweden

To understand the origin of disease risk in adulthood, factors in all stages of life and on different contextual levels should be considered. Therefore, the aim of this study was to investigate the relevance of a person's area of residence over their life course with regard to four outcomes: all-cause mortality; ischemic heart disease mortality and morbidity; cancer mortality and morbidity; and respiratory diseases and related mortality. We applied a cross-classified multilevel model for three age groups on a longitudinal data set spanning a 35 year period in Scania, Sweden. According to our analyses, the proportion of the total variance at the district level for all the outcomes studied was below 2% for the 65 to 84 age group, below 4.5% for those ages 50-64 years, and below 6.5% for those 30-49 years old. Our results suggest that the parish of residence, at four different time points during the individual life course, had little influence on individual all-cause mortality, or on mortality or morbidity from IHD, cancer, and respiratory diseases; i.e., knowing when and where an individual resided during their life course gives little indication of future mortality and morbidity. Such knowledge is essential in assisting decision makers determine the relevant geographical level of intervention (in our case whether to direct interventions toward the entire region of Scania or to specific parishes) needed. Valuable information for planning public health interventions might be obtained by considering measures of variance and clustering from specific contexts before implementing strategic programs.     

Geographies of health perception in Québec: a multilevel perspective.

Self perceived health is a widely used measure and, in Quebec, it has been shown to vary significantly between geographical areas. In the present study, these geographical variations are examined in a multilevel analysis in order to disentangle compositional (individual characteristics) and contextual (place) effects. The analysis recognizes four levels of variation: individual, household, local and regional. Similar analyses carried out in Britain, have considered only two levels: individual and local. Data come from the 1992-1993 Quebec Health and Social Survey, a general household survey using a stratified two-stage sampling design. Health perception (the response variable) is considered with a set of individual predictor characteristics reflecting gender, lifestyle, socio-economic conditions, marital status and social support. Results show the existence of significant local area variations in health perception after having allowed for individual characteristics and variations at the household level. At the regional level, however, no systematic and significant variations remain although some individual regions are found to have a significant impact on health perception.     

Assessment of the magnitude of geographical variations and socioeconomic contextual effects on ischaemic heart disease mortality: a multilevel survival analysis of a large Swedish cohort

BACKGROUND: In a public health perspective, it is of interest to assess the magnitude of geographical variations in ischaemic heart disease (IHD) mortality and quantify the strength of contextual effects on IHD. OBJECTIVE: To investigate whether area effects vary according to the individual and contextual characteristics of the population, socioeconomic contextual influences were assessed in different age groups and within territories of differing population densities. DESIGN: Multilevel survival analysis of a 28-year longitudinal database. PARTICIPANTS: 341 048 residents of the Scania region in Sweden, reaching age 50-79 years in 1996, followed up over 7 years. RESULTS: After adjustment for several individual socioeconomic indicators over the adult age, Cox multilevel models indicated geographical variations in IHD mortality and socioeconomic contextual effects on the mortality risk. However, the magnitude of geographical variations and strength of contextual effects were modified by the age of individuals and the population density of their residential area: socioeconomic contextual effects were much stronger among non-elderly than among elderly adults, and much larger within urban territories than within rural ones. As a consequence, among non-elderly residents of urban territories, the socioeconomic contextual effect was almost as large as the effect of individual 20-year cumulated income. CONCLUSIONS: Non-elderly residents of deprived urban neighbourhoods constitute a major target for both contextual epidemiology of coronary disease and public health interventions aimed at reducing the detrimental effects of the social environment on IHD.     

Income distribution and mortality: implications from a comparison of individual-level analysis and multilevel analysis with Swedish data

BACKGROUND: This follow-up study analyses whether there is an association between income distribution in Swedish municipalities and risk of death from all causes in the total Swedish population aged 40-64 years and compares the results obtained with analyses performed on individual-level analysis and multilevel analysis. METHODS: Individual-level data on social and economic circumstances were obtained from various official records and were linked to the national cause-of-death register. Analyses were made with two methods, an individual-level regression and a multilevel regression. The study population comprised all people 40-64 years of age in the 1990 Swedish census, altogether 2.57 million people in 284 municipalities. RESULTS: The main results showed that in the individual-level regression the income distribution showed a positive and significant association (risk ratio = 1.29; 95% CI = 1.24-1.34) with higher mortality for those living in municipalities with higher income inequality. This association was not found in the multilevel regression analysis (RR = 1.03; 95%CI = 0.94-1.13). CONCLUSION: There seems to be no association between income distribution and mortality in Sweden when considering the possibility of clustering in municipalities. Further studies on the relationship between income inequality and health should aim at elucidate processes within area-level units.     

A brief conceptual tutorial on multilevel analysis in social epidemiology: investigating contextual phenomena in different groups of people

STUDY OBJECTIVE: (1) To provide a didactic and conceptual (rather than mathematical) link between multilevel regression analysis (MLRA) and social epidemiological concepts. (2) To develop an epidemiological vision of MLRA focused on measures of health variation and clustering of individual health status within areas, which is useful to operationalise the notion of "contextual phenomenon". The paper shows how to investigate (1) whether there is clustering within neighbourhoods, (2) to which extent neighbourhood level differences are explained by the individual composition of the neighbourhoods, (3) whether the contextual phenomenon differs in magnitude for different groups of people, and whether neighbourhood context modifies individual level associations, and (4) whether variations in health status are dependent on individual level characteristics.Design and PARTICIPANTS: Simulated data are used on systolic blood pressure (SBP), age, body mass index (BMI), and antihypertensive medication (AHM) ascribed to 25 000 subjects in 39 neighbourhoods of an imaginary city. Rather than assessing neighbourhood variables, the paper concentrated on SBP variance between individuals and neighbourhoods as a function of individual BMI. RESULTS: The variance partition coefficient (VPC) showed that clustering of SBP within neighbourhoods was greater for people with a higher BMI. The composition of the neighbourhoods with respect to age, AHM use, and BMI explained about one fourth of the neighbourhood differences in SBP. Neighbourhood context modified the individual level association between BMI and SBP. Individual level differences in SBP within neighbourhoods were larger for people with a higher BMI. CONCLUSIONS: Statistical measures of multilevel variations can effectively quantify contextual effects in different groups of people, which is a relevant issue for understanding health inequalities.     

Social capital and health-purely a question of context?

Debate still surrounds which level of analysis (individual vs. contextual) is most appropriate to investigate the effects of social capital on health. Applying multilevel ecometric analyses to British Household Panel Survey data, we estimated fixed and random effects between five individual-, household- and small area-level social capital indicators and general health. We further compared the variance in health attributable to each level using intraclass correlations. Our results demonstrate that association between social capital and health depends on indicator type and level investigated, with one quarter of total individual-level health variance found at the household level. However, individual-level social capital variables and other health determinants appear to influence contextual-level variance the most.     

No association of income inequality with adult mortality within New Zealand: a multi-level study of 1.4 million 25-64 year olds

STUDY OBJECTIVE: To determine the association of regional income inequality within New Zealand with mortality among 25-64 year olds. DESIGN: Individual census and mortality records were linked over the 1991-94 period. Income inequality (Gini coefficients) and average household income variables were calculated for 35 regions. "Individual level" variables were sex, age, ethnicity, household income, rurality, and small area socioeconomic deprivation. Logistic regression was used for the analyses. Sensitivity analyses for the level of regional aggregation were conducted. PARTICIPANTS: 1.4 million New Zealand census respondents aged 25-64 years followed up for mortality for three years. Main results: Controlling for age, ethnicity, rurality, household income, and regional mean income, there was no association of income inequality with all cause mortality for either men (OR=1.007 for a 0.01 increase in the Gini, 95% confidence intervals 0.989 to 1.024) or women (OR=1.004, 0. 983 to 1.026). By cause of death (cancer, cardiovascular disease, unintentional injury, and suicide) there was some suggestion of a positive association for female unintentional injury (OR=1.068, 0.952 to 1.198) and suicide (OR=1.087, 0.957 to 1.234) but the 95% confidence intervals all included 1.0. Failure to control for ethnicity at the individual level resulted in some association of increasing regional income inequality with increasing mortality risk. Using fewer (n=14) or more (n=73) regional divisions did not substantially change the findings. CONCLUSION: There is no convincing evidence of an association of income inequality within New Zealand with adult mortality. Previous ecological analyses within New Zealand suggesting an association of income inequality with mortality were confounded by ethnicity at the individual level. However, this study does not refute the possibility that income inequality at the national level affects health.     

Misspecification of the effect of race in fixed effects models of health inequalities

The purpose of this study is to characterize the different results obtained when analyzing health inequalities data in which individuals are nested within their neighborhoods and a single level model is used to characterize risk rather than a multilevel model. The inability of single level models to characterize between neighborhood variance in risk may affect the level of risk attributed to black race if blacks are differentially distributed in high risk neighborhoods. The research replicates in Los Angeles an approach applied by a different group of researchers in Massachusetts (Subramanian, Chen, Rehkopf, Waterman, & Krieger, 2005). Single level and multilevel models were used to analyze Los Angeles County, California, US all-cause mortality data for the years 1989–1991, modeled as 29,936 cells (deaths and population denominators cross-tabulated by age, gender, and race/ethnicity) nested within 1552 census tracts. Overall blacks had 1.27 times the risk of mortality compared to whites. However, multilevel models demonstrated considerable between census tract variance in mortality for both blacks and whites which was partially explained by neighborhood poverty. Comparing the results of equivalent single level and multilevel models, the mortality odds ratio for blacks compared to the white reference group reversed itself, indicating greater risk for blacks in the single level model and lower risk in the multilevel model. Adding an area based socioeconomic measure (ABSM) to the single level model reduced but did not remove the discrepancy. Predictions of mortality risk for the interaction of race and age group demonstrate that all single level models exaggerated the mortality risk associated with black race. We conclude that characterizing health inequalities in mortality for blacks using single level models, which do not account for the cross level interaction created by the greater likelihood of black residence in neighborhoods where the risk of mortality is greater regardless of race, can exaggerate the risk of mortality attributable to the individual level effects of black race.     

Comparison of a spatial approach with the multilevel approach for investigating place effects on health: the example of healthcare utilisation in France

STUDY OBJECTIVE: Most studies of place effects on health have followed the multilevel analytical approach that investigates geographical variations of health phenomena by fragmenting space into arbitrary areas. This study examined whether analysing geographical variations across continuous space with spatial modelling techniques and contextual indicators that capture space as a continuous dimension surrounding individual residences provided more relevant information on the spatial distribution of outcomes. Healthcare utilisation in France was taken as an illustrative example in comparing the spatial approach with the multilevel approach. DESIGN: Multilevel and spatial analyses of cross sectional data. PARTICIPANTS: 10,955 beneficiaries of the three principal national health insurance funds, surveyed in 1998 and 2000 on continental France. MAIN RESULTS: Multilevel models showed significant geographical variations in healthcare utilisation. However, the Moran's I statistic showed spatial autocorrelation unaccounted for by multilevel models. Modelling the correlation between people as a decreasing function of the spatial distance between them, spatial mixed models gave information not only on the magnitude, but also on the scale of spatial variations, and provided more accurate standard errors for risk factors effects. The socioeconomic level of the residential context and the supply of physicians were independently associated with healthcare utilisation. Place indicators better explained spatial variations in healthcare utilisation when measured across continuous space, rather than within administrative areas. CONCLUSIONS: The kind of conceptualization of space during analysis influences the understanding of place effects on health. In many contextual studies, viewing space as a continuum may yield more relevant information on the spatial distribution of outcomes.     

Average household income, crime, and smoking behaviour in a local area: the Finnish 10-Town study

Social environments, like neighbourhoods, are increasingly recognised as determinants of health. While several studies have reported an association of low neighbourhood socio-economic status with morbidity, mortality and health risk behaviour, little is known of the health effects of neighbourhood crime rates. Using the ongoing 10-Town study in Finland, we examined the relations of average household income and crime rate measured at the local area level, with smoking status and intensity by linking census data of local area characteristics from 181 postal zip codes to survey responses to smoking behaviour in a cohort of 23,008 municipal employees. Gender-stratified multilevel analyses adjusted for age and individual occupational status revealed an association between low local area income rate and current smoking. High local area crime rate was also associated with current smoking. Both local area characteristics were strongly associated with smoking intensity. Among ever-smokers, being an ex-smoker was less likely among residents in areas with low average household income and a high crime rate. In the fully adjusted model, the association between local area income and smoking behaviour among women was substantially explained by the area-level crime rate. This study extends our knowledge of potential pathways through which social environmental factors may affect health.     

The influence of neighborhood environment on the incidence of childhood asthma: a multilevel approach

Some ecological analyses suggest an influence of neighborhood environment on asthma outcomes. However, no previous study has applied a multilevel approach to assess an ecological effect of neighborhood environment on the incidence of childhood asthma accounting for individual risk factors. This study assessed the influence of neighborhood and individual-level factors on the incidence of childhood asthma among all children born in Rochester, Minnesota, between 1976 and 1979. We identified asthmatics among all children born in Rochester, between 1976 and 1983. We applied a multilevel survival model with the frailty term to assess the effects of neighborhood characteristics, such as mean family income per census tract (n = 16) from the 1980 census report and the status of whether a census tract faces intersections with major highways or railroads, on asthma incidence. The relative risks (RR) of neighborhood socioeconomic status (SES), the status of whether census tracts face intersections with highways or railroads and the variance of random effect of census tracts were calculated adjusting individual-level covariates for asthma, including gender, birth weight, mother's age at birth and parental educational level at birth. We found that the RR of developing asthma among children living in census tracts facing intersections with highways or railroads was 1.6 (95% CI: 1.1-2.2) compared to those who lived in census tracts not facing intersections, adjusting individual- and neighborhood-level covariates. The variance of the frailty term attributable to census tracts was small (0.0085) and was modified (from 0.004 to 0.0085, 112% change) by adding neighborhood covariates. The overall effects of individual-level factors on asthma incidence were independent of neighborhood environment. The influence of neighborhood environment on childhood asthma in a non-inner-city setting, like Rochester, Minnesota, was small to modest. Incorporating pertinent neighborhood-level covariates into multilevel models needs to be considered in assessing the random effect of clusters.     

Studying place effects on health by synthesising individual and area-level outcomes

It is well established that there exist substantial area-level socio-demographic variations in population health. However, area-level associations between deprivation and health cannot necessarily be interpreted as place effects on individual health. We demonstrate how recently developed statistical models for combining individual and aggregate data can help to separate the effects of place of residence and personal circumstances. We apply these to two health outcomes: risk of hospitalisation for cardiovascular disease (CVD) and risk of self-reported limiting long-term illness (LLTI). A combination of small-area data from UK hospital episode statistics and the UK census and individual data from the Health Survey for England are analysed, using a new multilevel modelling method termed hierarchical related regression (HRR). The standard multilevel model for place and health explains outcomes from individual data in terms of individual and area-level characteristics. HRR models increase precision by also explaining population aggregate outcomes, in terms of the same predictors. Aggregate outcomes are modelled by averaging the individual-level exposure-outcome relationship over the area, which can alleviate the ecological bias associated with interpreting the relationship between aggregate quantities as an individual-level relationship. We find that there are associations between area-level deprivation indicators and both area-level rates of hospital admission for CVD and area-level rates of LLTI. Multilevel models fitted to the individual data alone had insufficient power to determine whether these associations were due to compositional or contextual effects. Using HRR models which incorporate area-level outcomes in addition to individual outcomes, we found that for CVD, the area-level differences were mostly explained by individual-level effects, in particular the increased risk for individuals from non-white ethnic backgrounds. In contrast, there remained a significant association between LLTI and area-level deprivation even after adjusting for the significant increased risk associated with individual-level ethnicity and income. Our study illustrates that extending multilevel models to incorporate both individual and area-level outcomes increases power to distinguish between contextual and compositional effects.     

Modelling inequality in reported long term illness in the UK: combining individual and area characteristics.

STUDY OBJECTIVE: To assess the nature of the relation between health and social factors at both the aggregated scale of geographical areas and the individual scale. DESIGN AND SETTING: The individual data are derived from the sample of anonymised records (SAR) from the census of 1991 in Great Britain, and are combined with area data from this census. The ecological setting (context) was defined using multivariate methods to classify the 278 districts of residence identifiable in the SAR. The outcome health variable is the 1991 census long-term limiting illness question. Health variations were analysed by multilevel logistic regression to examine the compositional variation (at the level of the individual) and the contextual variation (variability operating at the level of districts) in reported illness. PARTICIPANTS: 10 per cent randomised subsample of the SAR who are aged 16+ and are resident in households. MAIN RESULTS: The multi-level modelling revealed that area factors have a significant association with individual health outcome but their effect is smaller than that of individual attributes. The results show evidence for both compositional and contextual effects in the pattern of variation in propensity to report illness. CONCLUSIONS: The results suggest generally higher levels of ill health for individuals who are older, not married, in a semi/unskilled manual social class, and socioeconomically deprived (as measured by a composite deprivation score). All individuals living in areas with high levels of illness (which tend to be more deprived areas) show greater morbidity, even after allowing for their individual characteristics. However, within affluent areas, where morbidity was generally lower, the health inequality (health gradient) between rich and poor individuals was particularly strong. We consider the implications of these findings for health and resource allocation policy.     

A brief conceptual tutorial on multilevel analysis in social epidemiology: interpreting neighbourhood differences and the effect of neighbourhood characteristics on individual health

STUDY OBJECTIVE: Using a conceptual rather than a mathematical approach, this article proposed a link between multilevel regression analysis (MLRA) and social epidemiological concepts. It has been previously explained that the concept of clustering of individual health status within neighbourhoods is useful for operationalising contextual phenomena in social epidemiology. It has been shown that MLRA permits investigating neighbourhood disparities in health without considering any particular neighbourhood characteristic but only information on the neighbourhood to which each person belongs. This article illustrates how to analyse cross level (neighbourhood-individual) interactions, how to investigate associations between neighbourhood characteristics and individual health, and how to use the concept of clustering when interpreting those associations and geographical differences in health.Design and PARTICIPANTS: A MLRA was performed using hypothetical data pertaining to systolic blood pressure (SBP) from 25 000 subjects living in the 39 neighbourhoods of an imaginary city. Associations between individual characteristics (age, body mass index (BMI), use of antihypertensive drug, income) or neighbourhood characteristic (neighbourhood income) and SBP were analysed. RESULTS: About 8% of the individual differences in SBP were located at the neighbourhood level. SBP disparities and clustering of individual SBP within neighbourhoods increased along individual BMI. Neighbourhood low income was associated with increased SBP over and above the effect of individual characteristics, and explained 22% of the neighbourhood differences in SBP among people of normal BMI. This neighbourhood income effect was more intense in overweight people. CONCLUSIONS: Measures of variance are relevant to understanding geographical and individual disparities in health, and complement the information conveyed by measures of association between neighbourhood characteristics and health.     

When does geography matter most? Age-specific geographical effects in the patterning of,and relationship between,mental wellbeing and mental illness

Mental illness and mental wellbeing are related but distinct constructs. Despite this, geographical enquiry often references the two as interchangeable indicators of mental health and assumes the relationship between the two is consistent across different geographical scales. Furthermore, the importance of geography in such research is commonly assumed to be static for all age groups, despite the large body of evidence demonstrating contextual effects in age-specific populations. We leverage simultaneous measurement of a mental illness and mental wellbeing metric from Understanding Society, a UK population-based survey, and employ bivariate, cross-classified multilevel modelling to characterise the relationship between geographical context and mental health. Results provide strong evidence for contextual effects for both responses before and after covariate adjustment, with weaker evidence for area-classification and PSU-level contextual effects for the GHQ-12 after covariate adjustment. Results support a two-continua model of mental health at the individual level, but indicates that consensual benefit may be achieved across both dimensions by intervening at household and regional levels. There is also some evidence of a greater contextual effects for mental wellbeing than for mental illness. Results highlight the potential of the household as a target for intervention design for consensual benefit across both constructs. Results also suggest the increased importance of geographical context for older respondents across both responses. This research supports an area-based approach to improving both mental illness and mental wellbeing in older populations.     

Mortality among lifelong nonsmokers exposed to secondhand smoke at home: cohort data and sensitivity analyses

Evidence is growing that secondhand smoke can cause death from several diseases. The association between household exposure to secondhand smoke and disease-specific mortality was examined in two New Zealand cohorts of lifelong nonsmokers ("never smokers") aged 45-77 years. Individual census records from 1981 and 1996 were anonymously and probabilistically linked with mortality records from the 3 years that followed each census. Age- and ethnicity-standardized mortality rates were compared for never smokers with and without home exposure to secondhand smoke (based on the reported smoking behavior of other household members). Relative risk estimates adjusted for age, ethnicity, marital status, and socioeconomic position showed a significantly greater mortality risk for never smokers living in households with smokers, with excess mortality attributed to tobacco-related diseases, particularly ischemic heart disease and cerebrovascular disease, but not lung cancer. Adjusted relative risk estimates for all cardiovascular diseases were 1.19 (95% confidence interval: 1.04, 1.38) for men and 1.01 (95% confidence interval: 0.88, 1.16) for women from the 1981-1984 cohort, and 1.25 (95% confidence interval: 1.06, 1.47) for men and 1.35 (95% confidence interval: 1.11, 1.64) for women from the 1996-1999 cohort. Passive smokers also had nonsignificantly increased mortality from respiratory disease. Sensitivity analyses indicate that these findings are not due to misclassification bias.     

No association of neighbourhood volunteerism with mortality in New Zealand: a national multilevel cohort study

BACKGROUND: The association of social capital with health and mortality is contentious, and empirical findings are inconsistent. This study tests the association of neighbourhood-level volunteerism with mortality. METHODS: Cohort study of 1996 New Zealand census respondents aged 25-74 years (4.75 million person years) using multilevel Poisson regression analyses. Neighbourhood (average population 2,034) measures included indices of social capital (volunteering activities for all census respondents) and deprivation. RESULTS: Adjusting for just age and marital status, the mortality rate ratios for people living in the quintile of neighbourhoods with the lowest compared with highest volunteerism were 1.16 (95% confidence interval 1.08-1.24) and 1.09 (1.01-1.18), for males and females, respectively. Adjusting for potential individual-level and neighbourhood-level socioeconomic confounders reduced the rate ratios to 0.94 (0.88-1.01) and 0.92 (0.85-1.01), respectively. There was no significant association with any cause of death, including suicide [rate ratios 0.89 (0.64-1.22) and 0.57 (0.31-1.05), respectively]. Restricting the analyses to only those census respondents living at their census night address for five or more years, and therefore 'exposed' to that level of volunteerism for a longer period, did not substantially alter findings. CONCLUSIONS: This study, one of the largest multilevel studies yet, found no statistically significant independent association of a structural measure of neighbourhood social capital with mortality-including suicide. Assuming social features of neighbourhoods are important determinants of health, future research should examine other features (e.g. social fragmentation) and other outcomes (e.g. behaviour).     

Individual factors explain neighbourhood variations in accidents to children under 5 years of age

Previous studies have identified possible neighbourhood-level influences on the risk of injuries to preschool children, but none have had sufficient data at both household and area level to explain these neighbourhood effects. We used data from the Avon Longitudinal Study of Parents and Children, which recruited over 14,062 children at birth in the former county of Avon, UK, and collected information about accidents, as well as extensive social, health and developmental data throughout the first 5 years of life. This information was combined with census and geographical data in order to identify neighbourhood influences on accident risks and then attempt to explain these using multilevel regression modelling. A small but statistically significant amount of between-neighbourhood variance in accident risk was found, with neighbourhood intraclass correlation coefficients of 0.82% for any accident, and 0.84% for accidents resulting in injury requiring medical attention. This was entirely accounted for by a variety of child, parental and household level variables. Independent risk factors for both outcomes were children who were developmentally more advanced or displayed greater conduct and behavioural problems, mothers who were of younger age, who were without work, who were smokers, whose partners were unemployed or drank alcohol excessively, and households in which there had recently been adverse life events, or which were under financial stress. The mother's perceptions of neighbourhood quality also explained some of the risks for any accident, but not for medically attended accidents, and this was a variable that operated at the level of individual households rather than at the level of neighbourhoods. The implications of this study are that differences in accident risk between neighbourhoods are explained by geographical clustering of similar types of children, families and households. Interventions should focus more on parental factors and household social circumstances than on the physical environment or community based risks. However, many of these factors are those most resistant to modification without broader societal change.     

Diastolic blood pressure and area of residence: multilevel versus ecological analysis of social inequity

STUDY OBJECTIVES—To study geographical differences in diastolic blood pressure and the influence of the social environment (census percentage of people with low educational achievement) on individual diastolic blood pressure level, after controlling for individual age and educational achievement. To compare the results of multilevel and ecological analyses.
DESIGN—Cross sectional analysis performed by multilevel linear regression modelling, with women at the first level and urban areas at the second level, and by single level ecological regression using areas as the unit of analysis.
SETTING—Malmö, Sweden (population 250 000).
PARTICIPANTS—15 569 women aged 45 to 73, residing in 17 urban areas, who took part in the Malmö Diet and Cancer Study (1991-1996).
MAIN RESULTS—In the "fixed effects" multilevel analysis, low educational achievement at both individual (β=1.093, SE=0.167) and area levels (β=2.966, SE=1.250) were independently associated with blood pressure, although in the "random effects" multilevel analysis almost none of the total variability in blood pressure across persons was attributable to areas (intraclass correlation=0.3%). The ecological analysis also found an association between the area educational variable and mean diastolic blood pressure (β=4.058, SE=1.345).
CONCLUSIONS—The small intraclass correlation found indicated very marginal geographical differences and almost no influence of the urban area on individual blood pressure. However, these slight differences were enough to detect an effect of the social environment on blood pressure. The ecological study overestimated the associations found in the "fixed" effects multilevel analysis, and neither distinguished individual from area levels nor provided information on the intraclass correlation. Ecological analyses are inadequate to evaluate geographical differences in health.


     

Community- and individual-level socioeconomic status and breast cancer risk: multilevel modeling on Cape Cod, Massachusetts

BACKGROUND: Previous research demonstrated increased risk of breast cancer associated with higher socioeconomic status (SES) measured at both the individual and community levels. However, little attention has been paid to simultaneously examining both measures. OBJECTIVES: We evaluated the independent influences of individual and community SES on the risk of breast cancer using case-control data. Because our previous work suggests that associations may be stronger after including a latency period, we also assessed the effect of community-level SES assuming a 10-year latency period. METHODS: We obtained individual education for cases and matched controls diagnosed between 1987 and 1993 on Cape Cod, Massachusetts (USA). We acquired community-level SES from census data for 1980 and 1990. Using SES data at diagnosis and 10 years earlier, we constructed models for breast cancer risk using individual-level SES only, community-level SES only, and a multilevel analysis including both. We adjusted models for other individual-level risk factors. RESULTS: Women with the highest education were at greater risk of developing breast cancer in both 1980 and 1990 [odds ratio (OR) = 1.17 and 1.19, respectively]. Similarly, women living in the highest-SES communities in 1990 had greater risk (OR = 1.30). Results were stronger in the analyses considering a latency period (OR = 1.69). Adjusting for intragroup correlation had little effect on the analyses. CONCLUSIONS: Models including individual- or community-level measures of SES produced associations similar to those observed in previous research. Results for models including both measures are consistent with a contextual effect of SES on risk of breast cancer independent of individual SES.