An Original Stepwise Multilevel Logistic Regression Analysis of Discriminatory Accuracy: The Case of Neighbourhoods and Health

Abstract

Background and AimMany multilevel logistic regression analyses of “neighbourhood and health” focus on interpreting measures of associations (e.g., odds ratio, OR). In contrast, multilevel analysis of variance is rarely considered. We propose an original stepwise analytical approach that distinguishes between “specific” (measures of association) and “general” (measures of variance) contextual effects. Performing two empirical examples we illustrate the methodology, interpret the results and discuss the implications of this kind of analysis in public health.MethodsWe analyse 43,291 individuals residing in 218 neighbourhoods in the city of Malmö, Sweden in 2006. We study two individual outcomes (psychotropic drug use and choice of private vs. public general practitioner, GP) for which the relative importance of neighbourhood as a source of individual variation differs substantially. In Step 1 of the analysis, we evaluate the OR and the area under the receiver operating characteristic (AUC) curve for individual-level covariates (i.e., age, sex and individual low income). In Step 2, we assess general contextual effects using the AUC. Finally, in Step 3 the OR for a specific neighbourhood characteristic (i.e., neighbourhood income) is interpreted jointly with the proportional change in variance (i.e., PCV) and the proportion of ORs in the opposite direction (POOR) statistics.ResultsFor both outcomes, information on individual characteristics (Step 1) provide a low discriminatory accuracy (AUC = 0.616 for psychotropic drugs; = 0.600 for choosing a private GP). Accounting for neighbourhood of residence (Step 2) only improved the AUC for choosing a private GP (+0.295 units). High neighbourhood income (Step 3) was strongly associated to choosing a private GP (OR = 3.50) but the PCV was only 11% and the POOR 33%.ConclusionApplying an innovative stepwise multilevel analysis, we observed that, in Malmö, the neighbourhood context per se had a negligible influence on individual use of psychotropic drugs, but appears to strongly condition individual choice of a private GP. However, the latter was only modestly explained by the socioeconomic circumstances of the neighbourhoods. Our analyses are based on real data and provide useful information for understanding neighbourhood level influences in general and on individual use of psychotropic drugs and choice of GP in particular. However, our primary aim is to illustrate how to perform and interpret a multilevel analysis of individual heterogeneity in social epidemiology and public health. Our study shows that neighbourhood “effects” are not properly quantified by reporting differences between neighbourhood averages but rather by measuring the share of the individual heterogeneity that exists at the neighbourhood level.

Investigating societal determinants of oral health—Opportunities and challenges in multilevel studies

“…Multilevel studies offer the opportunity to separate the contextual effect of income inequality on individual health from the compositional effect of individual income by allowing to adjust for individual income within the same regression . However, ecological studies analyse ‘population risk’ according to area‐level income inequality, while the multilevel studies assess ‘individual risk’ according to area‐level income inequality . The population‐level aspect of health outcome in multilevel studies is studied through investigating the share of individual‐level variation in health outcomes that exist at the population level through decomposition of variance …”

“…However, ecological studies analyse ‘population risk’ according to area‐level income inequality, while the multilevel studies assess ‘individual risk’ according to area‐level income inequality . The population‐level aspect of health outcome in multilevel studies is studied through investigating the share of individual‐level variation in health outcomes that exist at the population level through decomposition of variance …”

“…Methodological experts argue that multilevel modelling has not been used to its potential to answer questions on population‐level variations in health status and its determinants in the field of social epidemiology . Studies have mostly focussed on average associations between individual and societal determinants, and health outcomes, ignoring a thorough analysis of heterogeneity around average associations examined through the variance estimates obtained from multilevel studies . The variance component informs to what extent individuals within a group are correlated with one another in relation to health.…”

“…Two additional measures, 80% interval odds ratio (IOR) and proportion of odds ratio in opposite direction (POOR), can be quantified by combining regression coefficients obtained from averaged associations between societal determinants and individual oral health and the variance attributed to the contextual level. These two measures estimate the heterogeneity in the associations between societal exposures and individual health outcomes among contexts/population groups . Measures of variation in individual health and its decomposition are critical for inferences on population‐level variations in health.…”

Inter-regional metric disadvantages when comparing countries’ happiness on a global scale. A Rasch based consequential validity analysis

“…The differences in the estimation of the inter-regional inequalities in happiness were compared by four techniques: 1) contrasting descriptive statistics obtained by the original HPI and the two Rasch measures (DIF-unadjusted and DIF-adjusted); 2) comparing the variance partition coefficients (VPC) estimated by random intercept linear regression variance decomposition analyses, where VPC expresses the percentage of variance attributable to inter-regional differences (Merlo, Wagner, Ghith, & Leckie, 2016); 3) graphically comparing the distribution of happiness according to HPI and DIF-adjusted Rasch measure of happiness on the world map; and 4) comparing the top 20 happiest countries according to HPI and the DIF-adjusted Rasch measure.…”