How sensitive are multilevel regression findings to defined area of context?: a case study of mammography use in California

Neighbourhood Income Inequality and General Psychopathology at 3-Years of Age

Background

Several studies have linked neighbourhood environment to preschool-aged children's behavioural problems. Income inequality is an identified risk factor for mental health among adolescents, however, little is known as to whether this relationship extends to younger children.

Objective

To explore the association between neighbourhood-level income inequality and general psychopathology problems among preschool-aged children.

Methods

We analyzed data from the All Our Families (AOF) longitudinal cohort located in Calgary, Canada at 3-years postpartum. The analytical sample consisted of 1615 mother-preschooler dyads nested within 184 neighbourhoods. Mothers completed the National Longitudinal Survey of Children and Youth Child Behaviour Checklist (NLSCY-CBCL), which assessed internalizing and externalizing symptoms. Income inequality was assessed via the Gini coefficient, which quantifies the unequal distribution of income in society. Mixed effects linear regression assessed the relationship between neighbourhood income inequality and preschooler's general psychopathology.

Results

The mean Gini coefficient across the 184 neighbourhoods was 0.33 (SD = 0.05; min, max: 0.20-0.56). In the fully adjusted model income inequality was not associated with general psychopathology in children β = 0.07 (95%CI: -0.29, 0.45). Neighbourhood environment accounted for 0.5% of the variance in psychopathology in children.

Conclusion

The lack of significant findings may be due to a lack of statistical power in the study. Future studies should investigate this relationship with appropriately powered studies, and over time, to assess if income inequality is a determinant of preschooler psychopathology in Canada.

Associations between breast cancer screening participation and residential area sociodemographic features, geographic accessibility, and features of screening venue location in Greater Sydney, Australia

Breast cancer screening (BCS) participation rates are often suboptimal and vary geographically. Environmental features may influence BCS participation, but few studies have assessed this relationship. This study assessed the associations between BCS participation, residential area sociodemographic characteristics, distance to BCS venue, and venue location attributes. Data for 384,433 women residing in Greater Sydney, Australia, invited to BCS during 2011-2014 were spatially joined to their state suburb (SSC) (n = 800). SSC sociodemographic measures included women's median age, proportion women speaking English at home, full-time employed, and university educated; and proportion dwellings with motor-vehicles. Road network distance was calculated to each BCS venue. BCS venues were coded as co-located with bus-stop, train-station, hospital, general practitioner (GP), and shop. Hot spots were calculated to quantify spatial clustering of BCS participation. Multilevel logistic models were used to estimate the associations between environmental predictors and BCS participation, accounting for SSC-level clustering. BCS participation was 53.9% and spatially clustered. BCS was positively associated with SSC-level median age for women, proportions women speaking English and university educated, and dwellings with motor-vehicles. Distance to venue was inversely associated with BCS. Venue co-location with GP was positively associated and co-location with bus-stop, train-station, and shop, hospital were negatively associated with BCS. Residential sociodemographic features, geographic access, and venue location attributes are associated with BCS participation. These findings implicate the relevance of social and built environmental factors to programmatic aims to raise BCS participation. Additional research on venue location features is required to understand where best to site BCS venues.

A Spatial Regression Analysis on the Effect of Neighborhood-Level Trust on Cooperative Behaviors: Comparison With a Multilevel Regression Analysis

There is no reason to suppose that neighborhood effects based on residents' trust vary according to administrative boundaries. We examined the relationship between neighborhood trust and cooperative behaviors using the spatial Durbin model which assumed that people are influenced by closer neighbors regardless of administrative boundaries, comparing the results with those of the multilevel model. We used data from 476 residents in Arakawa Ward, Tokyo, Japan. For each respondent, we assigned a unique 'neighborhood trust' value weighted by the inverse distance between the respondent and all other respondents as an independent variable. The dependent variables were perceived neighbors' cooperative behaviors and respondents' own cooperative behaviors. The spatial Durbin model showed that spatially weighted neighborhood trust was positively associated with cooperative behaviors. Meanwhile, the multilevel models did not show the statistically significant effect of neighborhood trust. We concluded that the spatial model might model the neighborhood effects in society more precisely.

Assessing and Explaining Geographic Variations in Mammography Screening Participation and Breast Cancer Incidence

Investigating geographic variations in mammography screening participation and breast cancer incidence help improve prevention strategies to reduce the burden of breast cancer. This study examined the suitability of health insurance claims data for assessing and explaining geographic variations in mammography screening participation and breast cancer incidence at the district level. Based on screening unit data (1,181,212 mammography screening events), cancer registry data (13,241 incident breast cancer cases) and claims data (147,325 mammography screening events; 1,778 incident breast cancer cases), screening unit and claims-based standardized participation ratios (SPR) of mammography screening as well as cancer registry and claims-based standardized incidence ratios (SIR) of breast cancer between 2011 and 2014 were estimated for the 46 districts of the German federal state of Lower Saxony. Bland-Altman analyses were performed to benchmark claims-based SPR and SIR against screening unit and cancer registry data. Determinants of district-level variations were investigated at the individual and contextual level using claims-based multilevel logistic regression analysis. In claims and benchmark data, SPR showed considerable variations and SIR hardly any. Claims-based estimates were between 0.13 below and 0.14 above (SPR), and between 0.36 below and 0.36 above (SIR) the benchmark. Given the limited suitability of health insurance claims data for assessing geographic variations in breast cancer incidence, only mammography screening participation was investigated in the multilevel analysis. At the individual level, 10 of 31 Elixhauser comorbidities were negatively and 11 positively associated with mammography screening participation. Age and comorbidities did not contribute to the explanation of geographic variations. At the contextual level, unemployment rate was negatively and the proportion of employees with an academic degree positively associated with mammography screening participation. Unemployment, income, education, foreign population and type of district explained 58.5% of geographic variations. Future studies should combine health insurance claims data with individual data on socioeconomic characteristics, lifestyle factors, psychological factors, quality of life and health literacy as well as contextual data on socioeconomic characteristics and accessibility of mammography screening. This would allow a comprehensive investigation of geographic variations in mammography screening participation and help to further improve prevention strategies for reducing the burden of breast cancer.

Neighborhood characteristics as determinants of healthcare utilization - a theoretical model

BACKGROUND

We propose using neighborhood characteristics as demand-related morbidity adjusters to improve prediction models such as the risk equalization model.

RESULTS

Since the neighborhood has no explicit 'place' in healthcare demand models, we have developed the "Neighborhood and healthcare utilization model" to show how neighborhoods matter in healthcare utilization. Neighborhood may affect healthcare utilization via (1) the supply-side, (2) need, and (3) demand for healthcare - irrespective of need. Three pathways are examined in detail to explain how neighborhood characteristics influence healthcare utilization via need: the physiological, psychological and behavioral pathways. We underpin this theoretical model with literature on all relevant neighborhood characteristics relating to health and healthcare utilization.

CONCLUSION

Potential neighborhood characteristics for the risk equalization model include the degree of urbanization, public and open space, resources and facilities, green and blue space, environmental noise, air pollution, social capital, crime and violence, socioeconomic status, stability, and ethnic composition. Air pollution has already been successfully tested as an important predictive variable in a healthcare risk equalization model, and it might be opportune to add more neighborhood characteristics.

Community social capital and inequality in depressive symptoms among older Japanese adults: A multilevel study

Although studies have suggested that community social capital contributes to narrow income-based inequality in depression, the impacts may depend on its components. Our multilevel cross-sectional analysis of data from 42,208 men and 45,448 women aged 65 years or older living in 565 school districts in Japan found that higher community-level civic participation (i.e., average levels of group participation in the community) was positively associated with the prevalence of depressive symptoms among the low-income groups, independent of individual levels of group participation. Two other social capital components (cohesion and reciprocity) did not significantly alter the association between income and depressive symptoms.

Resident Perceptions of Neighborhood Conditions, Food Access, Transportation Usage, and Obesity in a Rapidly Changing Central City

There is a lack of research on obesity that uses primary data and fine-grained information on neighborhoods. I use primary data for 367 participants in Detroit to examine neighborhood predictors of obesity. These data were supplemented with public data. I considered multilevel and spatial modeling, but the data lent itself best to ordinary least squares (OLS) regressions. I find that socioeconomic factors, the built environment, transportation usage, and perceptions of neighborhoods are important predictors of obesity. Importantly, litter is associated with higher levels of obesity. Planners can take measures to reduce litter and collaborate with other policy-makers to encourage less driving, though drawing direct lines of causality is complicated.

Contextual effects and cancer outcomes in the United States: a systematic review of characteristics in multilevel analyses

PURPOSE

There is increasing call for the utilization of multilevel modeling to explore the relationship between place-based contextual effects and cancer outcomes in the United States. To gain a better understanding of how contextual factors are being considered, we performed a systematic review.

METHODS

We reviewed studies published between January 1, 2002 and December 31, 2016 and assessed the following attributes: (1) contextual considerations such as geographic scale and contextual factors used; (2) methods used to quantify contextual factors; and (3) cancer type and outcomes. We searched PubMed, Scopus, and Web of Science and initially identified 1060 studies. One hundred twenty-two studies remained after exclusions.

RESULTS

Most studies utilized a two-level structure; census tracts were the most commonly used geographic scale. Socioeconomic factors, health care access, racial/ethnic factors, and rural-urban status were the most common contextual factors addressed in multilevel models. Breast and colorectal cancers were the most common cancer types, and screening and staging were the most common outcomes assessed in these studies.

CONCLUSIONS

Opportunities for future research include deriving contextual factors using more rigorous approaches, considering cross-classified structures and cross-level interactions, and using multilevel modeling to explore understudied cancers and outcomes.

Modeling Geospatial Patterns of Late-Stage Diagnosis of Breast Cancer in the US

In the US, about one-third of new breast cancers (BCs) are diagnosed at a late stage, where morbidity and mortality burdens are higher. Health outcomes research has focused on the contribution of measures of social support, particularly the residential isolation or segregation index, on propensity to utilize mammography and rates of late-stage diagnoses. Although inconsistent, studies have used various approaches and shown that residential segregation may play an important role in cancer morbidities and mortality. Some have focused on any individuals living in residentially segregated places (place-centered), while others have focused on persons of specific races or ethnicities living in places with high segregation of their own race or ethnicity (person-centered). This paper compares and contrasts these two approaches in the study of predictors of late-stage BC diagnoses in a cross-national study. We use 100% of U.S. Cancer Statistics (USCS) Registry data pooled together from 40 states to identify late-stage diagnoses among ~1 million new BC cases diagnosed during 2004–2009. We estimate a multilevel model with person-, county-, and state-level predictors and a random intercept specification to help ensure robust effect estimates. Person-level variables in both models suggest that non-White races or ethnicities have higher odds of late-stage diagnosis, and the odds of late-stage diagnosis decline with age, being highest among the <age 50 group. After controlling statistically for all other factors, we examine place-centered isolation and find for anyone living in an isolated Asian community there is a large beneficial association (suggesting lower odds of late-stage diagnosis) while for anyone living in an isolated White community there is a large detrimental association (suggesting greater odds of late-stage diagnosis). By contrast, living in neighborhoods among others of one’s own race or ethnicity (person-centered isolation) is associated with greater odds of late-stage diagnosis, as this measure is dominated by Whites (the majority). At the state level, living in a state that allows unfettered access to a specialist is associated with a somewhat lower likelihood of being diagnosed at a late stage of BC. Geographic factors help explain the likelihood of late-stage BC diagnosis, which varies considerably across the U.S. as heterogeneous compositional and contextual factors portray very different places and potential for improving information and outcomes. The USCS database is expanding to cover more states and is expected to be a valuable resource for ongoing and future place-based cancer outcomes research.

Demographic Disparities in Late-Stage Diagnosis of Breast and Colorectal Cancers Across the USA

BACKGROUND

We determined whether there were disparities in the likelihood of being diagnosed at a late stage for breast cancer (BC) or colorectal cancer (CRC) in each of 40 states, using the recently available US Cancer Statistics (USCS) database.

METHODS

We extracted 981,457 BC cases and 558,568 CRC cases diagnosed in 2004-2009. Separate multilevel regressions were run for each state and each cancer type. Models included person and area-level covariates and were identically specified across states. The disparities foci were race or ethnicity (white, African-American, Hispanic, Asian, all other), gender, and age (<40, 40-49, 50-64, 65-74, and 75+). Using whites, males, and the oldest age group as reference groups, we noted the statistically significant disparities coefficients (p value ≤0.05) and translated the findings via a set of maps of states in the USA.

RESULTS

National disparity estimates were not consistent with disparities identified in the states. Some states had estimates consistent with the national average, while others did not. Patterns of disparities across states were different for each covariate and mapped separately.

CONCLUSION

National disparity estimates may mask what is true at the more local, state level because national estimates can confound the effects of race with place. Cancer control efforts are local and require locally relevant information to assess needs. Findings from the period 2004-2009 establish valuable benchmarks against which to assess changes following national health reform implemented in 2010. The USCS database is a valuable new resource that will facilitate future disparities research.

Using residential segregation to predict colorectal cancer stage at diagnosis: two different approaches

PURPOSE

Studies have found a variety of evidence regarding the association between residential segregation measures and health outcomes in the United States. Some have focused on any individuals living in residentially segregated places, whereas others have examined whether persons of specific races or ethnicities living in places with high segregation of their own race or ethnicity have differential outcomes. This article compares and contrasts these two approaches in the study of predictors of late-stage colorectal cancer (CRC) diagnoses in a cross-national study. We argue that it is very important when interpreting results from studies like this to carefully consider the geographic scope of the analysis, which can significantly change the context and meaning of the results.

METHODS

We use US Cancer Statistics Registry data from 40 states to identify late-stage diagnoses among over 500,000 CRC cases diagnosed during 2004-2009. We pool data over the states and estimate a multilevel model with person, county, and state levels and a random intercepts specification to ensure robust effect estimates. The isolation index of residential segregation is defined for racial and ethnic groups at the county level using Census 2000 data. The association between isolation indices and late-stage CRC diagnosis was measured by (1) anyone living in minority-segregated areas (place-centered approach) and by (2) individuals living in areas segregated by one's own racial or ethnic peers (person-centered approach).

RESULTS

Findings from the place-centered approach suggest that living in a highly segregated African American community is associated with lower likelihood of late-stage CRC diagnosis, whereas the opposite is true for people living in highly segregated Asian communities, and living in highly segregated Hispanic communities has no significant association. Using the person-centered approach, we find that living in places segregated by one's racial or ethnic peers is associated with lower likelihood of late-stage CRC diagnosis.

CONCLUSIONS

In a model that covers a large geographic area across the nation, the place-centered approach is most likely picking up geographic disparities that may be deepened by targeted interventions in minority communities. By contrast, the person-centered approach provides a national average estimate suggesting that residential isolation may confer community cohesion or support that is associated with better CRC prevention.

Breast Cancer Screening Among Women with Medicaid, 2006-2008: a Multilevel Analysis

INTRODUCTION

Nationally, about one third of women with breast cancer (BC) are diagnosed at late stage, which might be reduced with greater utilization of BC screening. The purpose of this paper is to examine the predictors of BC mammography use among women with Medicaid, and differences among Medicaid beneficiaries in their propensity to use mammography.

METHODS

The sample included 2,450,527 women drawn from both fee-for-service and managed care Medicaid claims from 25 states, during 2006-2008. The authors used multilevel modeling of predictors at person, county, and state levels of influence and examined traditional factors affecting access and the expanded scope of practice allowed for the nurse practitioner (NP) in some states to provide primary care independent of physician oversight.

RESULTS

Black [OR = 0.87; 95 % CI (0.87-0.88)] and American Indian women [OR = 0.74; 95 % CI (0.71-0.76)] had lower odds ratio of mammography use than white women, while Hispanic [OR = 1.06; 95 % CI (1.05-1.07)] had higher odds ratio of mammography use than white women. Living in counties with higher Hispanic residential segregation [OR = 1.16; 95 % CI (1.10-1.23)] was associated with a higher odds ratio of mammography use compared to areas with low Hispanic residential segregation, whereas living among more segregated black [OR = 0.78; 95 % CI (0.75-0.81)] or Asian [OR = 0.19; 95 % CI (0.17-0.21)] communities had lower odds ratio compared to areas with low segregation. Holding constant statistically the perceived shortage of MDs, which was associated with significantly lower mammography use, the NP regulatory variable [OR = 1.03; 95 % CI (1.01-1.07)] enhanced the odds ratio of mammography use among women in the six states with expanded scope of practice, compared with women residing in 19 more restrictive states.

CONCLUSIONS

Racial and ethnic disparities exist in the use of mammography among Medicaid-insured women. More expansive NP practice privileges in states are associated with higher utilization, and may help reduce rural disparities.

Development of a cross-cultural deprivation index in five European countries

BACKGROUND

Despite a concerted policy effort in Europe, social inequalities in health are a persistent problem. Developing a standardised measure of socioeconomic level across Europe will improve the understanding of the underlying mechanisms and causes of inequalities. This will facilitate developing, implementing and assessing new and more effective policies, and will improve the comparability and reproducibility of health inequality studies among countries. This paper presents the extension of the European Deprivation Index (EDI), a standardised measure first developed in France, to four other European countries-Italy, Portugal, Spain and England, using available 2001 and 1999 national census data.

METHODS AND RESULTS

The method previously tested and validated to construct the French EDI was used: first, an individual indicator for relative deprivation was constructed, defined by the minimal number of unmet fundamental needs associated with both objective (income) poverty and subjective poverty. Second, variables available at both individual (European survey) and aggregate (census) levels were identified. Third, an ecological deprivation index was constructed by selecting the set of weighted variables from the second step that best correlated with the individual deprivation indicator.

CONCLUSIONS

For each country, the EDI is a weighted combination of aggregated variables from the national census that are most highly correlated with a country-specific individual deprivation indicator. This tool will improve both the historical and international comparability of studies, our understanding of the mechanisms underlying social inequalities in health and implementation of intervention to tackle social inequalities in health.

Individual and neighbourhood social capital and all-cause mortality in Brazilian adults: a prospective multilevel study

OBJECTIVE

The relationship between social capital and mortality is not clear-cut. There have been few longitudinal studies investigating this association so far. The objective was to assess the effect of different dimensions of social capital on mortality among adults of a Brazilian city.

STUDY DESIGN

This is a prospective multilevel study. Baseline data were obtained from a population-based random sample of 846 adults (aged 18 years or more) residing in 38 neighbourhoods (census blocks).

METHODS

Participants were interviewed in 2006-7 and their vital status investigated in 2013. Social capital was assessed by five scales (social cohesion, informal social control, neighbours' support, social action and political efficacy). The outcome was all-cause mortality. Data analysis used multilevel logistic regression models.

RESULTS

At the individual level social cohesion was positively related to mortality in the unadjusted model but this association lost significance after adjustment for other variables in multivariable models. At the neighbourhood level, high mortality rates were associated with low social action independently of demographic, socio-economic, behavioural and health-related variables.

CONCLUSION

We found more evidence for a contextual than individual level effect of social capital on mortality.

Geographic and Demographic Disparities in Late-stage Breast and Colorectal Cancer Diagnoses Across the US

PROBLEM

In 2009, breast cancer was the most common cancer in women, and colorectal cancer was the third most common cancer in both men and women. Currently, the majority of colorectal and almost 1/3 of breast cancers are diagnosed at an advanced stage in the US, which results in higher morbidity and mortality than would obtain with earlier detection. The incidence of late-stage cancer diagnoses varies considerably across the US, and few analyses have examined the entire US.

PURPOSE

Using the newly available US Cancer Statistics database representing 98% of the US population, we perform multilevel analysis of the incidence of late-stage cancer diagnoses and translate the findings via bivariate mapping, answering questions related to both Why and Where demographic and geographic disparities in these diagnoses are observed.

METHODS

To answer questions related to Why disparities are observed, we utilize a three-level, random-intercepts model including person-, local area-, and region- specific levels of influence. To answer questions related to Where disparities are observed, we generate county level robust predictions of late-stage cancer diagnosis rates and map them, contrasting counties ranked in the upper and lower quantiles of all county predicted rates. Bivariate maps are used to spatially translate the geographic variation among US counties in the distribution of both BC and CRC late-stage diagnoses.

CONCLUSIONS

Empirical modeling results show demographic disparities, while the spatial translation of empirical results shows geographic disparities that may be quite useful for state cancer control planning. Late stage BC and CRC diagnosis rates are not spatially random, manifesting as place-specific patterns that compare counties in individual states to counties across all states. Providing a relative comparison that enables assessment of how results in one state compare with others, this paper is to be disseminated to all state cancer control and central cancer registry program officials.

Multilevel built environment features and individual odds of overweight and obesity in Utah

Based on the data from the Behavioral Risk Factor Surveillance System (BRFSS) in 2007, 2009 and 2011 in Utah, this research uses multilevel modeling (MLM) to examine the associations between neighborhood built environments and individual odds of overweight and obesity after controlling for individual risk factors. The BRFSS data include information on 21,961 individuals geocoded to zip code areas. Individual variables include BMI (body mass index) and socio-demographic attributes such as age, gender, race, marital status, education attainment, employment status, and whether an individual smokes. Neighborhood built environment factors measured at both zip code and county levels include street connectivity, walk score, distance to parks, and food environment. Two additional neighborhood variables, namely the poverty rate and urbanicity, are also included as control variables. MLM results show that at the zip code level, poverty rate and distance to parks are significant and negative covariates of the odds of overweight and obesity; and at the county level, food environment is the sole significant factor with stronger fast food presence linked to higher odds of overweight and obesity. These findings suggest that obesity risk factors lie in multiple neighborhood levels and built environment features need to be defined at a neighborhood size relevant to residents' activity space.

Does patient rurality predict quality colon cancer care?: A population-based study

BACKGROUND

More than 50 million people reside in rural America. However, the impact of patient rurality on colon cancer care has been incompletely characterized, despite its known impact on screening.

OBJECTIVE

Our study sought to examine the impact of patient rurality on quality and comprehensive colon cancer care.

DESIGN

We constructed a retrospective cohort of 123,129 patients with stage 0 to IV colon cancer. Rural residence was established based on the patient medical service study area designated by the registry.

SETTINGS

The study was conducted using the 1996-2008 California Cancer Registry.

PATIENTS

All of the patients diagnosed between 1996 and 2008 with tumors located in the colon were eligible for inclusion in this study.

MAIN OUTCOME MEASURES

Baseline characteristics were compared by rurality status. Multivariate regression models then were used to examine the impact of rurality on stage in the entire cohort, adequate lymphadenectomy in stage I to III disease, and receipt of chemotherapy for stage III disease. Proportional-hazards regression was used to examine the impact of rurality on cancer-specific survival.

RESULTS

Of all of the patients diagnosed with colon cancer, 18,735 (15%) resided in rural areas. Our multivariate models demonstrate that rurality was associated with later stage of diagnosis, inadequate lymphadenectomy in stage I to III disease, and lower likelihood of receiving chemotherapy for stage III disease. In addition, rurality was associated with worse cancer-specific survival.

LIMITATIONS

We could not account for socioeconomic status directly, although we used insurance status as a surrogate. Furthermore, we did not have access to treatment location or distance traveled. We also could not account for provider or hospital case volume, patient comorbidities, or complications.

CONCLUSIONS

A significant portion of patients treated for colon cancer live in rural areas. Yet, rural residence is associated with modest differences in stage, adherence to quality measures, and survival. Future endeavors should help improve care to this vulnerable population (see video, Supplemental Digital Content 1, http://links.lww.com/DCR/A143).

Examining the complexity and variation of health care system distrust across neighborhoods: Implications for preventive health care

PURPOSE

Recently, the institutional performance model has been used to explain the increased distrust of health care system by arguing that distrust is a function of individuals' perceptions on the quality of life in neighborhood and social institutions. We examined (1) whether individuals assess two dimensions of distrust consistently, (2) if the multilevel institutional performance model explains the variation of distrust, and (3) how distrust patterns affect preventive health care behaviors.

METHODOLOGY

Using data from 9,497 respondents in 914 census tracts (neighborhoods) in Philadelphia, we examined the patterns of how individuals evaluate the competence and values distrust using the Multilevel Latent Class Analysis (MLCA), and then investigated how neighborhood environment factors are associated with distrust patterns. Finally, we used regression to examine the relationships between distrust patterns and preventive health care.

FINDINGS

The MLCA identified four distrust patterns: Believers, Doubters, Competence Skeptics, and Values Skeptics. We found that 55 % of the individuals evaluated competence and values distrust coherently, with Believers reporting low levels and Doubters having high levels of distrust. Competence and Values Skeptics assessed distrust inconsistently. Believers were the least likely to reside in socioeconomically disadvantaged and racially segregated neighborhoods than other patterns. In contrast to Doubters, Believers were more likely to use preventive health care, even after controlling for other socioeconomic factors including insurance coverage.

PRACTICAL IMPLICATIONS

Our findings suggest that distrust patterns are function of neighborhood conditions and distrust patterns are associated with preventive health care. This study provides important policy implications for health care and future interventions.

Self-rated health and residential segregation: how does race/ethnicity matter?

Despite recent declines, racial segregation remains a detriment to minority neighborhoods. However, existing research is inconclusive as to the effects racial segregation has on health. Some argue that racial segregation is related to poor health outcomes, whereas others suspect that racial segregation may actually lead to improved health for some minority communities. Even less is known about whether minority access to white neighborhoods improves health. We address these gaps with individual data from the 2010 Public Health Management Corporation's Southeastern Pennsylvania Household Health Survey and census tract data from the 2010 Decennial Census and the 2006-2010 American Community Survey. We implement logistic multilevel models to determine whether and how a resident's self-rated health is affected by the racial/ethnic segregation of their neighborhoods. Our key finding suggests that the effects of segregation on self-rated health depend on an individual's race/ethnicity, with blacks and Latino residents most likely to experience adverse effects. Particularly, minorities living in predominantly white communities have a significantly higher likelihood to report poor/fair health than they would in segregated minority neighborhoods. These findings make clear that access to white neighborhoods is not sufficient to improve minority health; fuller neighborhood integration is necessary to ensure all have health equity.

Assessing patterns of spatial behavior in health studies: their socio-demographic determinants and associations with transportation modes (the RECORD Cohort Study)

Prior epidemiological studies have mainly focused on local residential neighborhoods to assess environmental exposures. However, individual spatial behavior may modify residential neighborhood influences, with weaker health effects expected for mobile populations. By examining individual patterns of daily mobility and associated socio-demographic profiles and transportation modes, this article seeks to develop innovative methods to account for daily mobility in health studies. We used data from the RECORD Cohort Study collected in 2011-2012 in the Paris metropolitan area, France. A sample of 2062 individuals was investigated. Participants' perceived residential neighborhood boundaries and regular activity locations were geocoded using the VERITAS application. Twenty-four indicators were created to qualify individual space-time patterns, using spatial analysis methods and a geographic information system. Three domains of indicators were considered: lifestyle indicators, indicators related to the geometry of the activity space, and indicators related to the importance of the residential neighborhood in the overall activity space. Principal component analysis was used to identify main dimensions of spatial behavior. Multilevel linear regression was used to determine which individual characteristics were associated with each spatial behavior dimension. The factor analysis generated five dimensions of spatial behavior: importance of the residential neighborhood in the activity space, volume of activities, and size, eccentricity, and specialization of the activity space. Age, socioeconomic status, and location of the household in the region were the main predictors of daily mobility patterns. Activity spaces of small sizes centered on the residential neighborhood and implying a large volume of activities were associated with walking and/or biking as a transportation mode. Examination of patterns of spatial behavior by individual socio-demographic characteristics and in relation to transportation modes is useful to identify populations with specific mobility/accessibility needs and has implications for investigating transportation-related physical activity and assessing environmental exposures and their effects on health.

The small area predictors of ambulatory care sensitive hospitalizations: a comparison of changes over time

The hospital admission for ambulatory care sensitive conditions (ACSCs) is a validated indicator of impeded access to good primary and preventive care services. The authors examine the predictors of ACSC admissions in small geographic areas in two cross-sections spanning an 11-year time interval (1995-2005). Using hospital discharge data from the Healthcare Cost and Utilization Project of the Agency for Healthcare Research and Quality for Arizona, California, Massachusetts, Maryland, New Jersey, and New York for the years 1995 and 2005, the study includes a multivariate cross-sectional design, using compositional factors describing the hospitalized populations and the contextual factors, all aggregated at the primary care service area level. The study uses ordinary least squares regressions with and without state fixed effects, adjusting for heteroscedasticity. Data is pooled over 2 years to assess the statistically significant changes in associations over time. ACSC admission rates were inversely related to the availability of local primary care physicians, and managed care was associated with declines in ACSC admissions for the elderly. Minorities, aged elderly, and percent under federal poverty level were found to be associated with higher ACSC rates. The comparative analysis for 2 years highlights significant declines in the association with ACSC rates of several factors including percent minorities and rurality. The two policy-driven factors, primary care physician capacity and Medicare-managed care penetration, were not found significantly more effective over time. Using small area analysis, the study indicates that improvements in socioeconomic conditions and geographic access may have helped improve the quality of primary care received by the elderly over the last decade, particularly among some minority groups.

Reducing racial and ethnic disparities in colorectal cancer screening is likely to require more than access to care

Colorectal endoscopy, an effective screening intervention for colorectal cancer, is recommended for people age fifty or older, or earlier for those at higher risk. Rates of colorectal endoscopy are still far below those recommended by the US Preventive Services Task Force. This study examined whether factors such as the supply of gastroenterologists and the proportion of the local population without health insurance coverage were related to the likelihood of having the procedure, and whether these factors explained racial and ethnic differences in colorectal endoscopy. We found evidence that improving access to health care at the county and individual levels through expanded health insurance coverage could improve colorectal endoscopy use but might not be sufficient to reduce racial and ethnic disparities in colorectal cancer screening. Policy action to address these disparities will need to consider other structural and cultural factors that may be inhibiting colorectal cancer screening.

Revisiting the role of the urban environment in substance use: the case of analgesic overdose fatalities

OBJECTIVES

We examined whether neighborhood social characteristics (income distribution and family fragmentation) and physical characteristics (clean sidewalks and dilapidated housing) were associated with the risk of fatalities caused by analgesic overdose.

METHODS

In a case-control study, we compared 447 unintentional analgesic opioid overdose fatalities (cases) with 3436 unintentional nonoverdose fatalities and 2530 heroin overdose fatalities (controls) occurring in 59 New York City neighborhoods between 2000 and 2006.

RESULTS

Analgesic overdose fatalities were less likely than nonoverdose unintentional fatalities to have occurred in higher-income neighborhoods (odds ratio [OR] = 0.82; 95% confidence interval [CI] = 0.70, 0.96) and more likely to have occurred in fragmented neighborhoods (OR = 1.35; 95% CI = 1.05, 1.72). They were more likely than heroin overdose fatalities to have occurred in higher-income (OR = 1.31; 95% CI = 1.12, 1.54) and less fragmented (OR = 0.71; 95% CI = 0.55, 0.92) neighborhoods.

CONCLUSIONS

Analgesic overdose fatalities exhibit spatial patterns that are distinct from those of heroin and nonoverdose unintentional fatalities. Whereas analgesic fatalities typically occur in lower-income, more fragmented neighborhoods than nonoverdose fatalities, they tend to occur in higher-income, less unequal, and less fragmented neighborhoods than heroin fatalities.

Are HPV vaccination services accessible to high-risk communities? A spatial analysis of HPV-associated cancer and Chlamydia rates and safety-net clinics

PURPOSE

While HPV vaccines can greatly benefit adolescents and young women from high-risk areas, little is known about whether safety-net immunization services are geographically accessible to communities at greatest risk for HPV-associated diseases. We explore the spatial relationship between areas with high HPV risk and proximity to safety-net clinics from an ecologic perspective.

METHODS

We used cancer registry data and Chlamydia surveillance data to identify neighborhoods within Los Angeles County with high risk for HPV-associated cancers. We examined proximity to safety-net clinics among neighborhoods with the highest risk. Proximity was measured as the shortest distance between each neighborhood center and the nearest clinic and having a clinic within 3 miles of each neighborhood center.

RESULTS

The average 5-year non-age-adjusted rates were 1,940 cases per 100,000 for Chlamydia and 60 per 100,000 for HPV-associated cancers. A large majority, 349 of 386 neighborhoods with high HPV-associated cancer rates and 532 of 537 neighborhoods with high Chlamydia rates, had a clinic within 3 miles of the neighborhood center. Clinics were more likely to be located within close proximity to high-risk neighborhoods in the inner city. High-risk neighborhoods outside of this urban core area were less likely to be near accessible clinics.

CONCLUSIONS

The majority of high-risk neighborhoods were geographically near safety-net clinics with HPV vaccination services. Due to low rates of vaccination, these findings suggest that while services are geographically accessible, additional efforts are needed to improve uptake. Programs aimed to increase awareness about the vaccine and to link underserved groups to vaccination services are warranted.

Spatial Polygamy and Contextual Exposures (SPACEs): Promoting Activity Space Approaches in Research on Place and Health

Exposure science has developed rapidly and there is an increasing call for greater precision in the measurement of individual exposures across space and time. Social science interest in an individual's environmental exposure, broadly conceived, has arguably been quite limited conceptually and methodologically. Indeed, we appear to lag behind our exposure science colleagues in our theories, data, and methods. In this paper we discuss a framework based on the concept of spatial polygamy to demonstrate the need to collect new forms of data on human spatial behavior and contextual exposures across time and space. Adopting new data and methods will be essential if we want to better understand social inequality in terms of exposure to health risks and access to health resources. We discuss the opportunities and challenges focusing on the potential seemingly offered by focusing on human mobility, and specifically the utilization of activity space concepts and data. A goal of the paper is to spatialize social and health science concepts and research practice vis-a-vis the complexity of exposure. The paper concludes with some recommendations for future research focusing on theoretical and conceptual development, promoting research on new types of places and human movement, the dynamic nature of contexts, and on training. "When we elect wittingly or unwittingly, to work within a level … we tend to discern or construct - whichever emphasis you prefer - only those kinds of systems whose elements are confined to that level."Otis Dudley Duncan (1961, p. 141)."…despite the new ranges created by improved transportation, local government units have tended to remain medieval in size."Torsten Hägerstrand (1970, p.18)"A detective investigating a crime needs both tools and understanding. If he has no fingerprint powder, he will fail to find fingerprints on most surfaces. If he does not understand where the criminal is likely to have put his fingers, he will not look in the right places. Equally, the analyst of data needs both tools and understanding."John Tukey (1977, p.1)"When we observe the environment, we necessarily do so on only a limited number of scales."Simon Levin (1992, p. 1945)There is a desperate need to develop methods with the same precision for an individual's environmental exposure as we have for an individual's genome … even a partial, targeted understanding of exposure can provide substantial advantages."Christopher Wild (2005, p.1848).

Spatial Sufficiency of 5% Medicare Standard Analytic Files

The 5% Medicare Standard Analytic Files (SAF) are random samples used to analyze national trends in medical treatments, expenditures, and outcomes. Their utility in small-area or multilevel analyses is unknown. To demonstrate possible limitations of the 5% SAF for analysis of health behaviors in small areas. We use descriptive Chi-square goodness-of-fit tests and mapping to explore consistency in the 5% representation of the 100% population in states and counties. We conduct multilevel modeling of individual utilization of mammography or endoscopy services for cancer screening and contrast findings across the 5% and 100% files. Subjects are enrolled in both parts A and B Medicare coverage and ages 65-104, alive and residing in the same state, with no gaps in coverage during the study period. Identically defined groups are drawn from the 5% SAF and 100% population claims and denominator files. The Chi-square tests of homogeneous population subgroups in 5% and 100% files exhibit significant differences in 7 of 8 states. Maps confirm this among states' counties and find that one state is generally under-represented by the 5% SAF, while others show areas with variable representation. Multilevel modeling results are largely consistent across the partitions of the data, but 5% sample models have much lower statistical power. Area-level covariate effect estimates show some differences across the two datasets. Multilevel modeling with contextual variables may be misleading in small area analyses conducted using 5% Medicare SAFs. Provider supply and market characteristics show inconsistent results. Disparities research may benefit from 100% files to provide statistical power needed to detect meaningful differences. This is significant because the Centers for Medicare and Medicaid Services have recently curtailed permissions to use the 100% files. These 100% files are one of few sources of population data available in the U.S. that are representative of small areas in the U.S.. In times of constrained budgets, using population data files is essential so that resources can be targeted to areas robustly identified as having greatest need or gaps in outcomes.

Progress in Spatial Demography

BACKGROUND

Demography is an inherently spatial science, yet the application of spatial data and methods to demographic research has tended to lag that of other disciplines. In recent years, there has been a surge in interest in adding a spatial perspective to demography. This sharp rise in interest has been driven in part by rapid advances in geospatial data, new technologies, and methods of analysis.

OBJECTIVES

We offer a brief introduction to four of the advanced spatial analytic methods: spatial econometrics, geographically weighted regression, multilevel modeling, and spatial pattern analysis. We look at both the methods used and the insights that can be gained by applying a spatial perspective to demographic processes and outcomes. To help illustrate these substantive insights, we introduce six papers that are included in a Special Collection on Spatial Demography. We close with some predictions for the future, as we anticipate that spatial thinking and the use of geospatial data, technology, and analytical methods will change how many demographers address important demographic research questions.

CONCLUSION

Many important demographic questions can be studied and framed using spatial approaches. This will become even more evident as changes in the volume, source, and form of available demographic data-much of it geocoded-further alter the data landscape, and ultimately the conceptual models and analytical methods used by demographers. This overview provides a brief introduction to a rapidly changing field.

Do more hospital beds lead to higher hospitalization rates? a spatial examination of Roemer's Law

BACKGROUND

Roemer's Law, a widely cited principle in health care policy, states that hospital beds that are built tend to be used. This simple but powerful expression has been invoked to justify Certificate of Need regulation of hospital beds in an effort to contain health care costs. Despite its influence, a surprisingly small body of empirical evidence supports its content. Furthermore, known geographic factors influencing health services use and the spatial structure of the relationship between hospital bed availability and hospitalization rates have not been sufficiently explored in past examinations of Roemer's Law. We pose the question, "Accounting for space in health care access and use, is there an observable association between the availability of hospital beds and hospital utilization?"

METHODS

We employ an ecological research design based upon the Anderson behavioral model of health care utilization. This conceptual model is implemented in an explicitly spatial context. The effect of hospital bed availability on the utilization of hospital services is evaluated, accounting for spatial structure and controlling for other known determinants of hospital utilization. The stability of this relationship is explored by testing across numerous geographic scales of analysis. The case study comprises an entire state system of hospitals and population, evaluating over one million inpatient admissions.

RESULTS

We find compelling evidence that a positive, statistically significant relationship exists between hospital bed availability and inpatient hospitalization rates. Additionally, the observed relationship is invariant with changes in the geographic scale of analysis.

CONCLUSIONS

This study provides evidence for the effects of Roemer's Law, thus suggesting that variations in hospitalization rates have origins in the availability of hospital beds. This relationship is found to be robust across geographic scales of analysis. These findings suggest continued regulation of hospital bed supply to assist in controlling hospital utilization is justified.

Neighborhood social capital and crime victimization: comparison of spatial regression analysis and hierarchical regression analysis

Crime is an important determinant of public health outcomes, including quality of life, mental well-being, and health behavior. A body of research has documented the association between community social capital and crime victimization. The association between social capital and crime victimization has been examined at multiple levels of spatial aggregation, ranging from entire countries, to states, metropolitan areas, counties, and neighborhoods. In multilevel analysis, the spatial boundaries at level 2 are most often drawn from administrative boundaries (e.g., Census tracts in the U.S.). One problem with adopting administrative definitions of neighborhoods is that it ignores spatial spillover. We conducted a study of social capital and crime victimization in one ward of Tokyo city, using a spatial Durbin model with an inverse-distance weighting matrix that assigned each respondent a unique level of "exposure" to social capital based on all other residents' perceptions. The study is based on a postal questionnaire sent to 20-69 years old residents of Arakawa Ward, Tokyo. The response rate was 43.7%. We examined the contextual influence of generalized trust, perceptions of reciprocity, two types of social network variables, as well as two principal components of social capital (constructed from the above four variables). Our outcome measure was self-reported crime victimization in the last five years. In the spatial Durbin model, we found that neighborhood generalized trust, reciprocity, supportive networks and two principal components of social capital were each inversely associated with crime victimization. By contrast, a multilevel regression performed with the same data (using administrative neighborhood boundaries) found generally null associations between neighborhood social capital and crime. Spatial regression methods may be more appropriate for investigating the contextual influence of social capital in homogeneous cultural settings such as Japan.

Geographic disparities in late-stage cancer diagnosis: multilevel factors and spatial interactions

In 2009 in the United States, breast cancer was the most common cancer in women, and colorectal cancer was the third most common cancer in both men and women. Currently, over 40% of these cancers are diagnosed at an advanced stage, which results in higher morbidity and mortality than would obtain with optimal cancer screening utilization. To provide information that might improve these cancer outcomes we use spatial analysis to answer questions related to both Why and Where disparities in late-stage cancer diagnoses are observed. In examining Why, we include state level characteristics reflecting characteristics of states' cancer control planning, insurance markets and managed care environments to help model the spatial heterogeneity from place to place. To answer questions related to Where disparities are observed, we generate county level predictions of late-stage cancer rates from a random-intercept multilevel model estimated on the population data from 11 pooled SEER Registries. The findings allow for comparisons across states that reveal logical starting points for a national effort to control cancer.

Medicare costs and surgeon supply in hospital service areas

OBJECTIVE

To quantify the correlates of variations of Medicare per beneficiary costs at the hospital service area level and determine whether physician supply and the specialty of physicians has a significant relationship with cost variation.

BACKGROUND

The American Medical Association Masterfile data on physician and surgeon location, characteristics and specialty; Census derived sociodemographic data from 2006 ZIP code level Claritas PopFacts database; and Medicare per beneficiary costs from the Dartmouth Atlas of Health Care project.

METHODS

A correlational analysis using bivariate plots and fixed effects linear regression models controlling for hospital service area sociodemographics and the number and characteristics of the physician supply. Data were aggregated to the Dartmouth hospital service area level from ZIP code level files.

RESULTS

We found that costs are strongly related to the sociodemographic character of the hospital service areas and the overall supply of physicians but a mixed correlation to the specialist supply depending on the interaction of the proportion of the physician supply who are international medical graduates. The ratio of general surgeons and surgical subspecialists to population are associated with lower costs in the models, again with difference depending on the influence of international medical graduates. There is a strong association between higher costs and the local proportion of physician supply made up of graduates of non-US or Canadian medical schools and female graduates.

CONCLUSIONS

These results suggest that strategies to reduce overall costs by changing physician supply must consider more than just overall numbers.

Spatial Heterogeneity in Cancer Control Planning and Cancer Screening Behavior

Each state is autonomous in its comprehensive cancer control (CCC) program, and considerable heterogeneity exists in the program plans. However, researchers often focus on the concept of nationally representative data and pool observations across states using regression analysis to come up with average effects when interpreting results. Due to considerable state autonomy and heterogeneity in various dimensions-including culture, politics, historical precedent, regulatory environment, and CCC efforts-it is important to examine states separately and to use geographic analysis to translate findings in place and time. We used 100 percent population data for Medicare-insured persons aged 65 or older and examined predictors of breast cancer (BC) and colorectal cancer (CRC) screening from 2001-2005. Examining BC and CRC screening behavior separately in each state, we performed 100 multilevel regressions. We summarize the state-specific findings of racial disparities in screening for either cancer in a single bivariate map of the 50 states, producing a separate map for African American and for Hispanic disparities in each state relative to whites. The maps serve to spatially translate the voluminous regression findings regarding statistically significant disparities between whites and minorities in cancer screening within states. Qualitative comparisons can be made of the states' disparity environments or for a state against a national benchmark using the bivariate maps. We find that African Americans in Michigan and Hispanics in New Jersey are significantly more likely than whites to utilize CRC screening and that Hispanics in 6 states are significantly and persistently more likely to utilize mammography than whites. We stress the importance of spatial translation research for informing and evaluating CCC activities within states and over time.

Incorporating geospatial capacity within clinical data systems to address social determinants of health

Linking electronic health record (EHR) systems with community information systems (CIS) holds great promise for addressing inequities in social determinants of health (SDH). While EHRs are rich in location-specific data that allow us to uncover geographic inequities in health outcomes, CIS are rich in data that allow us to describe community-level characteristics relating to health. When meaningfully integrated, these data systems enable clinicians, researchers, and public health professionals to actively address the social etiologies of health disparities.This article describes a process for exploring SDH by geocoding and integrating EHR data with a comprehensive CIS covering a large metropolitan area. Because the systems were initially designed for different purposes and had different teams of experts involved in their development, integrating them presents challenges that require multidisciplinary expertise in informatics, geography, public health, and medicine. We identify these challenges and the means of addressing them and discuss the significance of the project as a model for similar projects.

Does context matter for the relationship between deprivation and all-cause mortality? The West vs. the rest of Scotland

Background

A growing body of research emphasizes the importance of contextual factors on health outcomes. Using postcode sector data for Scotland (UK), this study tests the hypothesis of spatial heterogeneity in the relationship between area-level deprivation and mortality to determine if contextual differences in the West vs. the rest of Scotland influence this relationship. Research into health inequalities frequently fails to recognise spatial heterogeneity in the deprivation-health relationship, assuming that global relationships apply uniformly across geographical areas. In this study, exploratory spatial data analysis methods are used to assess local patterns in deprivation and mortality. Spatial regression models are then implemented to examine the relationship between deprivation and mortality more formally.

Results

The initial exploratory spatial data analysis reveals concentrations of high standardized mortality ratios (SMR) and deprivation (hotspots) in the West of Scotland and concentrations of low values (coldspots) for both variables in the rest of the country. The main spatial regression result is that deprivation is the only variable that is highly significantly correlated with all-cause mortality in all models. However, in contrast to the expected spatial heterogeneity in the deprivation-mortality relationship, this relation does not vary between regions in any of the models. This result is robust to a number of specifications, including weighting for population size, controlling for spatial autocorrelation and heteroskedasticity, assuming a non-linear relationship between mortality and socio-economic deprivation, separating the dependent variable into male and female SMRs, and distinguishing between West, North and Southeast regions. The rejection of the hypothesis of spatial heterogeneity in the relationship between socio-economic deprivation and mortality complements prior research on the stability of the deprivation-mortality relationship over time.

Conclusions

The homogeneity we found in the deprivation-mortality relationship across the regions of Scotland and the absence of a contextualized effect of region highlights the importance of taking a broader strategic policy that can combat the toxic impacts of socio-economic deprivation on health. Focusing on a few specific places (e.g. 15% of the poorest areas) to concentrate resources might be a good start but the impact of socio-economic deprivation on mortality is not restricted to a few places. A comprehensive strategy that can be sustained over time might be needed to interrupt the linkages between poverty and mortality.

Erosion in the Healthcare Safety Net: Impacts on Different Population Groups

Safety net hospitals (SNHs) have played a critical role in the U.S. health system providing access to health care for vulnerable populations, in particular the Medicaid and uninsured populations. However, little research has examined how access for these populations changes when contraction of the safety net occurs. Institutional policies, such as hospital closure or ownership conversion, could affect the supply of minority health care providers, thus exacerbating disparities in outcomes. We use multilevel logistic modeling of person-level hospital discharge data to examine the effects of contractions in the California safety net over the period of 1990-2000 on access to care as measured by changes in ambulatory care sensitive condition (ACSC) admissions, using geographic methods to characterize proximity to a contraction event. We found that presence of a contraction event was associated with a statistically significant increase in the predicted probability of impeded access, with an increase of about 1% for Medicaid-insured populations and about 4-5% for the uninsured. The Medicaid-insured group also maintained the highest rates of ACSC admissions over time, suggesting persistent access problems for this vulnerable group. This research is timely given continued budget problems in many states, where rising unemployment has increased the number of Medicaid enrollees by 6 million and uninsured individuals by 1.5 million, increasing pressure on remaining SNHs.

Geographic disparities in late-stage breast cancer diagnosis in California

Using cancer registry data for the population of California women aged 67+ with breast cancers, we estimated random intercept logistic models to examine how two socio-ecological predictors (residential isolation and poverty) were associated with probability of late-stage diagnosis for breast cancer. Using the multilevel modeling results, we calculated fully adjusted predicted probabilities associated with women in each Medical Service Study Area (MSSA) in California and classified the areas into two distinct groups: MSSAs with predicted rates below the 25th percentile (presumably the better outcome areas) and MSSAs with predicted rates above the 75th percentile (presumably the worse outcome areas) for two minority groups. Some areas had better outcomes for one group but worse outcomes for the other, suggesting that interventions to improve outcomes need different strategies for different groups in the same areas. Using information from geographic risk factors and multilevel modeling, this study informs interventions designed to reduce disparities in breast cancer outcomes.

Community contextual predictors of endoscopic colorectal cancer screening in the USA: spatial multilevel regression analysis

Background

Colorectal cancer (CRC) is the second leading cause of cancer death in the United States, and endoscopic screening can both detect and prevent cancer, but utilization is suboptimal and varies across geographic regions. We use multilevel regression to examine the various predictors of individuals' decisions to utilize endoscopic CRC screening. Study subjects are a 100% population cohort of Medicare beneficiaries identified in 2001 and followed through 2005. The outcome variable is a binary indicator of any sigmoidoscopy or colonoscopy use over this period. We analyze each state separately and map the findings for all states together to reveal patterns in the observed heterogeneity across states.

Results

We estimate a fully adjusted model for each state, based on a comprehensive socio-ecological model. We focus the discussion on the independent contributions of each of three community contextual variables that are amenable to policy intervention. Prevalence of Medicare managed care in one's neighborhood was associated with lower probability of screening in 12 states and higher probability in 19 states. Prevalence of poor English language ability among elders in one's neighborhood was associated with lower probability of screening in 15 states and higher probability in 6 states. Prevalence of poverty in one's neighborhood was associated with lower probability of screening in 36 states and higher probability in 5 states.

Conclusions

There are considerable differences across states in the socio-ecological context of CRC screening by endoscopy, suggesting that the current decentralized configuration of state-specific comprehensive cancer control programs is well suited to respond to the observed heterogeneity. We find that interventions to mediate language barriers are more critically needed in some states than in others. Medicare managed care penetration, hypothesized to affect information about and diffusion of new endoscopic technologies, has a positive association in only a minority of states. This suggests that managed care plans' promotion of this cost-increasing technology has been rather limited. Area poverty has a negative impact in the vast majority of states, but is positive in five states, suggesting there are some effective cancer control policies in place targeting the poor with supplemental resources promoting CRC screening.

Predictors of endoscopic colorectal cancer screening over time in 11 states

Objectives

We study a cohort of Medicare-insured men and women aged 65+ in the year 2000, who lived in 11 states covered by Surveillance, Epidemiology, and End Results (SEER) cancer registries, to better understand various predictors of endoscopic colorectal cancer (CRC) screening.

Methods

We use multilevel probit regression on two cross-sectional periods (2000–2002, 2003–2005) and include people diagnosed with breast cancer, CRC, or inflammatory bowel disease (IBD) and a reference sample without cancer.

Results

Men are not universally more likely to be screened than women, and African Americans, Native Americans, and Hispanics are not universally less likely to be screened than whites. Disparities decrease over time, suggesting that whites were first to take advantage of an expansion in Medicare benefits to cover endoscopic screening for CRC. Higher-risk persons had much higher utilization, while older persons and beneficiaries receiving financial assistance for Part B coverage had lower utilization and the gap widened over time.

Conclusions

Screening for CRC in our Medicare-insured sample was less than optimal, and reasons varied considerably across states. Negative managed care spillovers were observed, demonstrating that policy interventions to improve screening rates should reflect local market conditions as well as population diversity.

Mammography facilities are accessible, so why is utilization so low?

OBJECTIVE

This study examines new socio-ecological variables reflecting community context as predictors of mammography use.

METHODS

The conceptual model is a hybrid of traditional health-behavioral and socio-ecological constructs with an emphasis on spatial interaction among women and their environments, differentiating between several levels of influence for community context. Multilevel probability models of mammography use are estimated. The study sample includes 70,129 women with traditional Medicare fee-for-service coverage for inpatient and outpatient services, drawn from the SEER-Medicare linked data. The study population lives in heterogeneous California, where mammography facilities are dense but utilization rates are low.

RESULTS

Several contextual effects have large significant impacts on the probability of mammography use. Women living in areas with higher proportions of elderly in poverty are 33% less likely to use mammography. However, dually eligible women living in these poor areas are 2% more likely to use mammography than those without extra assistance living in these areas. Living in areas with higher commuter intensity, higher violent crime rates, greater land use mix (urbanicity), or more segregated Hispanic communities exhibit -14%, -1%, -6%, and -3% (lower) probability of use, respectively. Women living in segregated American Indian communities or in communities where more elderly women live alone exhibit 16% and 12% (higher) probability of use, respectively. Minority women living in more segregated communities by their minority are more likely to use mammography, suggesting social support, but this is significant for Native Americans only. Women with disability as their original reason for entitlement are found 40% more likely to use mammography when they reside in communities with high commuter intensity, suggesting greater ease of transportation for them in these environments.

CONCLUSIONS

Socio-ecological variables reflecting community context are important predictors of mammography use in insured elderly populations, often with larger magnitudes of effect than personal characteristics such as race or ethnicity (-3% to -7%), age (-2%), recent address change (-7%), disability (-5%) or dual eligibility status (-1%). Better understanding of community factors can enhance cancer control efforts.

Heterogeneity in mammography use across the nation: separating evidence of disparities from the disproportionate effects of geography

Background

Mammography is essential for early detection of breast cancer and both reduced morbidity and increased survival among breast cancer victims. Utilization is lower than national guidelines, and evidence of a recent decline in mammography use has sparked concern. We demonstrate that regression models estimated over pooled samples of heterogeneous states may provide misleading information regarding predictors of health care utilization and that comprehensive cancer control efforts should focus on understanding these differences and underlying causal factors. Our study population includes all women over age 64 with breast cancer in the Surveillance Epidemiology and End Results (SEER) cancer registries, linked to a nationally representative 5% reference sample of Medicare-eligible women located in 11 states that span all census regions and are heterogeneous in racial and ethnic mix. Combining women with and without cancer in the sample allows assessment of previous cancer diagnosis on propensity to use mammography. Our conceptual model recognizes the interplay between individual, social, cultural, and physical environments along the pathways to health care utilization, while delineating local and more distant levels of influence among contextual variables. In regression modeling, we assess individual-level effects, direct effects of contextual factors, and interaction effects between individual and contextual factors.

Results

Pooling all women across states leads to quite different conclusions than state-specific models. Commuter intensity, community acculturation, and community elderly impoverishment have significant direct impacts on mammography use which vary across states. Minorities living in isolated enclaves with others of the same race/ethnicity may be either advantaged or disadvantaged, depending upon the place studied.

Conclusion

Careful analysis of place-specific context is essential for understanding differences across communities stemming from different causal factors. Optimal policy interventions to change behavior (improve screening rates) will be as heterogeneous as local community characteristics, so no "one size fits all" policy can improve population health. Probability modeling with correction for clustering of individuals within multilevel contexts can reveal important differences from place to place and identify key factors to inform targeting of specific communities for further study.