Prevalence of treated diabetes: Geographical variations at the small-area level and their association with area-level characteristics. A multilevel analysis in Southeastern France. Diabetes Metab. 2011;37:39-46. [PMID: 20980182 DOI: 10.1016/j.diabet.2010.07.004]

Influence of neighborhood-level socioeconomic deprivation and individual socioeconomic position on risk of developing type 2 diabetes in older men: a longitudinal analysis in the British Regional Heart Study cohort

INTRODUCTION

Evidence from longitudinal studies on the influence of neighborhood socioeconomic deprivation in older age on the development of type 2 diabetes mellitus (T2DM) is limited. This study investigates the prospective associations of neighborhood-level deprivation and individual socioeconomic position (SEP) with T2DM incidence in older age.

RESEARCH DESIGN AND METHODS

The British Regional Heart Study studied 4252 men aged 60-79 years in 1998-2000. Neighborhood-level deprivation was based on the Index of Multiple Deprivation quintiles for participants' 1998-2000 residential postcode. Individual SEP was defined as social class based on longest-held occupation. A cumulative score of individual socioeconomic factors was derived. Incident T2DM cases were ascertained from primary care records; prevalent cases were excluded. Cox proportional hazard models were used to examine the associations.

RESULTS

Among 3706 men, 368 incident cases of T2DM were observed over 18 years. The age-adjusted T2DM risk increased from the least deprived quintile to the most deprived: HR per quintile increase 1.14 (95% CI 1.06 to 1.23) (p=0.0005). The age-adjusted T2DM HR in social class V (lowest) versus social class I (highest) was 2.45 (95% CI 1.36 to 4.42) (p=0.001). Both associations attenuated but remained significant on adjustment for other deprivation measures, becoming non-significant on adjustment for body mass index and T2DM family history. T2DM risk increased with cumulative individual adverse socioeconomic factors: HR per point increase 1.14 (95% CI 1.05 to 1.24).

CONCLUSIONS

Inequalities in T2DM risk persist in later life, both in relation to neighborhood-level and individual-level socioeconomic factors. Underlying modifiable risk factors continue to need to be addressed in deprived older age populations to reduce disease burden.

Spatiotemporal analyses of the Epidemiologic characteristics of Diabetes Mellitus

Research based on spatiotemporal analysis has been conducted to identify various factors that can affect an individual’s or community’s degree of health and disease. These spatiotemporal studies can effectively illustrate patterns in disease frequency, features, and temporal flow in different parts of a country. Furthermore, identifying these regional characteristics can aid in the development of disease prevention or intervention strategies.

Impact of deprivation on glycaemic control in youth with type 1 diabetes in the southwestern region of France

OBJECTIVES

The objective of this multicenter cross-sectional study was to determine predictors of poor glycaemic control in children with type 1 diabetes mellitus (T1DM), particularly with respect to socioeconomic status (SES).

METHODS

Our study population consisted of 1154 children who attended T1DM follow-up consultation with a pediatric diabetes specialist. Clinical and demographic data were retrieved retrospectively from patients' records. Individual deprivation was defined by an EPICES (Evaluation of the Deprivation and Inequalities of Health in Healthcare Centers) score ≥ 30. Patients were assigned to quintiles of the European Deprivation Index (EDI) based on their area deprivation scores. We used multivariable linear regression models to detect potential associations between glycaemic control and indicators of low SES.

RESULTS

In total, 33% (n = 376) of patients had an EPICES score ≥ 30 and 23% (n = 268) were in the 5th EDI quintile. Multivariable linear regression analysis showed that poor glycaemic control was associated with both individual (β 0.38; 95%CI 0.26-0.5; p < 0.001) and area deprivation (β 0.26; 95%CI 0.08-0.43; p = 0.004). Demographic factors, body mass index (BMI) and insulin regimen were also independently associated with poor glycaemic control (p < 0.001). Interestingly, access to diabetes technologies was not related to SES or either glycaemic control.

CONCLUSION

Low SES is associated with a higher risk of poor glycaemic control, independently of insulin regimen. BMI, age at the time of consultation, duration of diabetes, and insulin regimen. Also have an impact on HbA1c. These parameters need to be considered when developing novel treatment strategies for children with T1DM to better target at-risk patients.

Spatial epidemiology of diabetes: Methods and insights

Diabetes mellitus (DM) is a growing epidemic with global proportions. It is estimated that in 2019, 463 million adults aged 20-79 years were living with DM. The latest evidence shows that DM continues to be a significant global health challenge and is likely to continue to grow substantially in the next decades, which would have major implications for healthcare expenditures, particularly in developing countries. Hence, new conceptual and methodological approaches to tackle the epidemic are long overdue. Spatial epidemiology has been a successful approach to control infectious disease epidemics like malaria and human immunodeficiency virus. The implementation of this approach has been expanded to include the study of non-communicable diseases like cancer and cardiovascular diseases. In this review, we discussed the implementation and use of spatial epidemiology and Geographic Information Systems to the study of DM. We reviewed several spatial methods used to understand the spatial structure of the disease and identify the potential geographical drivers of the spatial distribution of DM. Finally, we discussed the use of spatial epidemiology on the design and implementation of geographically targeted prevention and treatment interventions against DM.

Regional variation in type 2 diabetes: evidence from 137 820 adults on the role of neighbourhood body mass index

BACKGROUND

Body mass index (BMI) is a key covariate in the study of type 2 diabetes, but can also be theorized as a contextual effect. The purpose of this study was to explore the extent to which variation in individual risk factors and neighbourhood BMI explain the variation in type 2 diabetes prevalence across neighbourhoods and municipalities.

METHODS

Cross-sectional data were collected from 137 820 adults aged ≥18 years from 3296 neighbourhoods in 296 municipalities in the Northern Netherlands. The odds of type 2 diabetes was assessed using a multilevel model. Median odds ratios were calculated and choropleth maps were created to visually assess neighbourhood variation in type 2 diabetes prevalence.

RESULTS

The overall prevalence of type 2 diabetes was 4%, ranging from 0 to ≥10 and 0-7% across neighbourhoods and municipalities, respectively. Of the regional variation, 67.0 and 71.6% is explained through variation of individual risk factors at the neighbourhood and municipality level, respectively. Analysis on the smallest spatial scale, i.e. the neighbourhood, best captured the regional variance. Statistically significant interaction between individual and neighbourhood BMI was found (OR = 1.06; 95% CI = 1.03-1.08, P for interaction < 0.001), adjusted for the individual risk profile.

CONCLUSION

The results suggest a more cautious interpretation of neighbourhood effects in type 2 diabetes is warranted, and reveals the need for further investigation into risk-prone groups to guide the design of community-level interventions to halt the rise in type 2 diabetes prevalence.

Neighbourhood property value and type 2 diabetes mellitus in the Maastricht study: A multilevel study

OBJECTIVE

Low individual socioeconomic status (SES) is known to be associated with a higher risk of type 2 diabetes mellitus (T2DM), but the extent to which the local context in which people live may influence T2DM rates remains unclear. This study examines whether living in a low property value neighbourhood is associated with higher rates of T2DM independently of individual SES.

RESEARCH DESIGN AND METHODS

Using cross-sectional data from the Maastricht Study (2010-2013) and geographical data from Statistics Netherlands, multilevel logistic regression was used to assess the association between neighbourhood property value and T2DM. Individual SES was based on education, occupation and income. Of the 2,056 participants (aged 40-75 years), 494 (24%) were diagnosed with T2DM.

RESULTS

Individual SES was strongly associated with T2DM, but a significant proportion of the variance in T2DM was found at the neighbourhood level (VPC = 9.2%; 95% CI = 5.0%-16%). Participants living in the poorest neighbourhoods had a 2.38 times higher odds ratio of T2DM compared to those living in the richest areas (95% CI = 1.58-3.58), independently of individual SES.

CONCLUSIONS

Neighbourhood property value showed a significant association with T2DM, suggesting the usefulness of area-based programmes aimed at improving neighbourhood characteristics in order to tackle inequalities in T2DM.

Geographical variation in the incidence of type 1 diabetes in the Nordic countries: A study within NordicDiabKids

BACKGROUND

The incidence of type 1 diabetes (T1D) is high in the Nordic countries with geographic differences between as well as within countries.

OBJECTIVE

To describe the geographical distribution of the incidence of T1D among children in four Nordic countries, an area where the population is considered genetically similar.

METHODS

Data on children 0 to 14 years of age and diagnosed with T1D 2006 to 2011 was collected from four Nordic national pediatric quality diabetes registries. Data included year of diagnosis (2006-2011), sex, and age at diagnosis. Figures for number of children at risk during 2006 to 2011-as well as total population, proportion with foreign background and size of populated areas of geographic regions-were collected from official statistics.

RESULTS

The total incidence during the study period for all four countries was 35.7/100 000 person years but differed between the countries (range 18.2-44.1; P < .001). The incidence difference between the countries was most obvious in the highest age group, 10 to 14 years of age, whereas there was no difference in the youngest age group 0 to 5 years of age. Iceland had similar incidence in the entire country, whereas the other countries had areas with different incidence. Densely populated areas, such as major cities, had the lowest incidence.

CONCLUSION

The incidence of T1D differed between the Nordic countries and also between the neighboring countries and generally decreased with population density. This indicates that environmental factors may contribute to the level of incidence of T1D.

Serious Mental Illness, Neighborhood Disadvantage, and Type 2 Diabetes Risk: A Systematic Review of the Literature

Aim of the Study: This review aims to systematically synthesize the body of literature examining the association between neighborhood socioeconomic disadvantage and serious mental illness (SMI)–type 2 diabetes (T2D) co-occurrence. Methods: We conducted an electronic search of four databases: PubMed, Scopus, Medline, and Web of Science. Studies were considered eligible if they were published in English, peer reviewed, quantitative, and focused on the association between neighborhood disadvantage and SMI-T2D comorbidity. Study conduct and reporting complied with PRISMA guidelines, and the protocol is made available at PROSPERO (CRD42017083483). Results: The one eligible study identified reported a higher burden of T2D in persons with SMI but provided only a tentative support for the association between neighborhood disadvantage and SMI-T2D co-occurrence. Conclusion: Research into neighborhood effects on SMI-T2D comorbidity is still in its infancy and the available evidence inconclusive. This points to an urgent need for attention to the knowledge gap in this important area of public health. Further research is needed to understand the health resource implications of the association between neighborhood deprivation and SMI-T2D comorbidity and the casual pathways linking them.

Built environmental characteristics and diabetes: a systematic review and meta-analysis

BACKGROUND

The built environment influences behaviour, like physical activity, diet and sleep, which affects the risk of type 2 diabetes mellitus (T2DM). This study systematically reviewed and meta-analysed evidence on the association between built environmental characteristics related to lifestyle behaviour and T2DM risk/prevalence, worldwide.

METHODS

We systematically searched PubMed, EMBASE.com and Web of Science from their inception to 6 June 2017. Studies were included with adult populations (>18 years), T2DM or glycaemic markers as outcomes, and physical activity and/or food environment and/or residential noise as independent variables. We excluded studies of specific subsamples of the population, that focused on built environmental characteristics that directly affect the cardiovascular system, that performed prediction analyses and that do not report original research. Data appraisal and extraction were based on published reports (PROSPERO-ID: CRD42016035663).

RESULTS

From 11,279 studies, 109 were eligible and 40 were meta-analysed. Living in an urban residence was associated with higher T2DM risk/prevalence (n = 19, odds ratio (OR) = 1.40; 95% CI, 1.2-1.6; I² = 83%) compared to living in a rural residence. Higher neighbourhood walkability was associated with lower T2DM risk/prevalence (n = 8, OR = 0.79; 95% CI, 0.7-0.9; I² = 92%) and more green space tended to be associated with lower T2DM risk/prevalence (n = 6, OR = 0.90; 95% CI, 0.8-1.0; I² = 95%). No convincing evidence was found of an association between food environment with T2DM risk/prevalence.

CONCLUSIONS

An important strength of the study was the comprehensive overview of the literature, but our study was limited by the conclusion of mainly cross-sectional studies. In addition to other positive consequences of walkability and access to green space, these environmental characteristics may also contribute to T2DM prevention. These results may be relevant for infrastructure planning.

[Trends in social inequities in diabetes care in the Provence-Alpes-Côte-d'Azur region of France between 2008 and 2011]

BACKGROUND

Social inequities in healthcare are known to exist for the management of many chronic diseases in France, including diabetes. The recession that began in 2008 has led to increased income disparities but has it also exacerbated health inequities. The aim of this study was to describe trends in inequities in diabetes-related healthcare between 2008 and 2011 in the PACA region (Provence-Alpes-Côte-d'Azur).

METHODS

This analysis used two sources of data: the regional national health insurance fund (PACA region) reimbursement database and the socio-demographic databases of the national statistics office (INSEE) for four full years (2008 to 2011). It included individuals who had been reimbursed for three purchases of oral diabetes drugs during the previous year and assessed the association between the median household income (weighted by number and age of household members) of each patient's municipality of residence and seven indicators of diabetes monitoring and care. Using adjusted mixed logistic models, including an interaction term between time (number of years) and the median household income of the municipality, we examined the performance of the indicators for each year.

RESULTS

The total number of patients with diabetes in the 941 municipalities of the PACA region varied by year from 142,055 to 164,929. Models showed that living in a town with a high or intermediate household income was significantly associated with better performance of the seven indicators and that the interaction term was significant for six of them. The effect of the municipal median income decreased significantly between 2008 and 2011 for five indicators: HbA1c, LDL-cholesterol, microalbuminuria, ophthalmoscopy and diabetes specialist visit.

CONCLUSION

Social inequities in diabetes-related healthcare persisted between 2008 and 2011 but appeared to be decreasing, despite the recession.

Membership in a diabetes-care network and adherence to clinical practice guidelines for treating type 2 diabetes among general practitioners: A four-year follow-up

OBJECTIVES

To assess whether private general practitioners (GPs) belonging to a diabetes-care network adhered more closely to clinical practice guidelines for diabetes care than GPs not in such a network, for all their patients with type 2 diabetes treated with medication (patients with diabetes), regardless of whether they received care through a network (that is, whether a halo effect occurred).

RESEARCH DESIGN AND MEASURES

The study, based on health insurance reimbursement databases in southeastern France, included 468 GPs in two networks and 468 non-network GPs in the same geographical area, matched one-to-one by propensity scores. We followed up their patients with diabetes (n=22,808) from 2008 through 2011, conducting multivariate time-to-event analyses (Cox models) that took the matching design into account to evaluate time from inclusion until performance of the given number of each of six recommended examinations/tests.

RESULTS

GPs belonging to a diabetes-care network adhered more closely to clinical practice guidelines but our result were slightly pronounced. Hazard ratios (HR) were significantly higher for patients of network GPs for the implementation of 3 HbA1C assays (HRa=1.13; [95%CI=1.10-1.16]), or 1 microalbuminuria assay (1.4 [1.35-1.45]); they were lower for LDL-cholesterol assays (1.04 [1.01-1.07]) and ophthalmological checkups (1.07 [1.04-1.10]), and not significant for creatinemia or cardiac monitoring.

CONCLUSIONS

Network GPs had better diabetes monitoring practices for all their patients with diabetes than the other GPs, especially for the most diabetes-specific tests. Further research is needed in other settings to confirm the existence of this halo effect.

Buprenorphine and methadone treatment for opioid dependence by income, ethnicity and race of neighborhoods in New York City

BACKGROUND

Geographic and demographic variation in buprenorphine and methadone treatment use in U.S. cities has not been assessed. Identifying variance in opioid maintenance is essential to improving treatment access and equity.

PURPOSE

To examine the differential uptake of buprenorphine treatment in comparison to methadone treatment between 2004 and 2013 in neighborhoods in New York City characterized by income, race and ethnicity.

METHODS

Social area (SA) analysis of residential zip codes of methadone and buprenorphine patients in NYC, which aggregated zip codes into five social areas with similar percentages of residents below poverty, identifying as Black non-Hispanic and as Hispanic, to examine whether treatment rates differed significantly among social areas over time. For each rate, mixed model analyses of variance were run with fixed effects for social area, year and the interaction of social area by year.

RESULTS

Buprenorphine treatment increased in all social areas over time with a significantly higher rate of increase in the social area with the highest income and the lowest percentage of Black, Hispanic, and low-income residents. Methadone treatment decreased slightly in all social areas until 2011 and then increased bringing rates back to 2004 levels. Treatment patterns varied by social area.

CONCLUSIONS

Buprenorphine treatment rates are increasing in all social areas, with slower uptake in moderate income mixed ethnicity areas. Methadone rates have remained stable over time. Targeted investments to promote public sector buprenorphine prescription may be necessary to reduce disparities in buprenorphine treatment and to realize its potential as a public health measure.

Rates of admission for ambulatory care sensitive conditions in France in 2009-2010: trends, geographic variation, costs, and an international comparison

BACKGROUND

Admissions for ambulatory care sensitive conditions (ACSCs) are considered preventable and indicators of poor access to primary care. We wondered whether per-capita rates of admission for ACSCs in France demonstrated geographic variation, were changing, were related to other independent variables, or were comparable to those in other countries; further, we wanted to quantify the resources such admissions consume.

METHODS

We calculated per-capita rates of admission for five categories (chronic, acute, vaccination preventable, alcohol-related, and other) of ACSCs in 94 departments in mainland France in 2009 and 2010, examined measures and causes of geographic variation in those rates, computed the costs of those admissions, and compared rates of admission for ACSCs in France to those in several other countries.

RESULTS

The highest ACSC admission rates generally occurred in the young and the old, but rates varied across French regions. Over the 2-year period, rates of most categories of ACSCs increased; higher ACSC admission rates were associated with lower incomes and a higher supply of hospital beds. We found that the local supply of general practitioners was inversely associated with rates of chronic and total ACSC admission rates, but that this relationship disappeared if we accounted for patients' use of general practitioners in neighboring departments. ACSC admissions cost 4.755 billion euros in 2009 and 5.066 billion euros in 2010; they consumed 7.86 and 8.74 million bed days of care, respectively. France had higher rates of ACSC admissions than most other countries examined.

CONCLUSIONS

Because admissions for ACSCs are generally considered a failure of outpatient care, cost French taxpayers substantial monetary and hospital resources, and appear to occur more frequently in France than in other countries, policymakers should prioritize targeted efforts to reduce them.

Characteristics of patients and physicians correlated with regular influenza vaccination in patients treated for type 2 diabetes: a follow-up study from 2008 to 2011 in southeastern France

We studied a cohort of 110 823 patients treated with oral hypoglycaemic agents for type 2 diabetes in southeastern France from 1 January 2008 to 31 December 2011, to identify influenza vaccination coverage trends and the patient and physician correlates of influenza vaccine (IFV) uptake. We used French national health insurance fund (NHIF) databases to identify these patients and collect data on their IFV reimbursement claims (IFVC) and patient and physician characteristics. We used multilevel multivariate polytomous logistic regressions to test the correlates of IFVC. Between 2008 and 2011 the annual IFVC rate varied from 33.7% to 32.3% in the 18-64 age group and from 69.5% to 61.1% in the 65 + age group, among whom we saw a clear trend towards reduced vaccination after 2008. In the younger group, the probability of regular vaccination each year from 2008 to 2011 increased with diabetes severity and duration, comorbidities, and the number of general practitioner and nurse visits; it was higher among patients seeing endocrinologists and lower among low-income patients than in other patients. In the older group, there was no association with either diabetes severity or physician specialty. These results suggest different patterns of correlates of influenza vaccination according to age. Endocrinologists might help to improve IFV uptake in the younger group of patients with type 2 diabetes. Communication strategies regarding influenza vaccination should be adapted to age, and collaboration between healthcare professionals should be reinforced to achieve vaccination objectives for these patients.

Spatial modelling of type II diabetes outcomes: a systematic review of approaches used

With the rising incidence of type II diabetes mellitus (DM II) worldwide, methods to identify high-risk geographical areas have become increasingly important. In this comprehensive review following Cochrane Collaboration guidelines, we outline spatial methods, outcomes and covariates used in all spatial studies involving outcomes of DM II. A total of 1894 potentially relevant citations were identified. Studies were included if spatial methods were used to explore outcomes of DM II or type I and 2 diabetes combined. Descriptive tables were used to summarize information from included studies. Ten spatial studies conducted in the USA, UK and Europe met selection criteria. Three studies used Bayesian generalized linear mixed modelling (GLMM), three used classic generalized linear modelling, one used classic GLMM, two used geographic information systems mapping tools and one compared case:provider ratios across regions. Spatial studies have been effective in identifying high-risk areas and spatial factors associated with DM II outcomes in the USA, UK and Europe, and would be useful in other parts of the world for allocation of additional services to detect and manage DM II early.

[Relevance of the health insurance databases to study spatial disparities in asthma prevalence: A study in southeastern France]

BACKGROUND

Data on asthma prevalence at a small-area level would be useful to set up and monitor French local public health policies. This study, based on drug reimbursement databases in southeastern France, aimed to (1) compare asthma-like disorders prevalence estimated by using three different indicators; (2) study sociodemographic characteristics associated with these indicators; (3) verify whether these indicators are equivalent to study geographical disparities of the asthma-like disorders prevalence at a small-area level.

METHODS

The study was conducted among the beneficiaries of the National Health Insurance Fund aged 18-44 years residing in southeastern France in 2010 (n=1,371,816). Using data on asthma drugs reimbursements (therapeutic class R03), we built three indicators to assess asthma-like disorders prevalence: at least 1, 2 or 3 purchase(s) in 2010. We analyzed sociodemographic characteristics associated with these indicators, and their geographical disparities at a small-area level using multilevel logistic regression models.

RESULTS

The crude asthma-like disorders prevalence varied between 2.6 % and 8.4 % depending on the indicator. It increased with age, was higher for women than for men, and among low-income people for all three indicators. We measured significant geographical disparities. Areas with high prevalence rates were the same regardless of the indicator.

CONCLUSION

The indicators built in this study can be useful to identify high prevalence areas. They could contribute to launch discussion on environmental health issues at the local level.

Area deprivation and the prevalence of type 2 diabetes and obesity: analysis at the municipality level in Germany

Background

The objective of this study was to analyse the association between area deprivation at municipality level and the prevalence of type 2 diabetes (T2D) and obesity across Germany, controlling for individual socioeconomic status (SES).

Methods

The analyses are based on a large survey conducted in 2006. Information was included from 39,908 adults aged 20 years or above. Area deprivation was assessed using the German Index of Multiple Deprivation (GIMD) at municipality level. About 4,700 municipalities could be included and assigned to a deprivation quintile. Individual SES was assessed by income and educational level. Multilevel logistic models were used to control for individual SES and other potential confounders such as age, sex and physical activity.

Results

We found a positive association of area deprivation with T2D and obesity. Controlling for all individual-level variables, the odds ratios for municipalities in the most deprived quintile were significantly increased for T2D (OR 1.35; 95% CI 1.12–1.64) as well as for obesity (OR 1.14; 95% CI 1.02–1.26). Further analyses showed that these associations were relatively similar for both men and women.

Conclusions

Based on a nationwide dataset, we were able to show that area deprivation at municipality level is significantly associated with the prevalence of T2D and obesity. It will be important to focus preventive efforts on very deprived municipalities.

Type 2 diabetes care in North Karelia Finland: do area-level socio-economic factors affect processes and outcomes?

AIMS

This research assessed the impact of area-level socio-economic factors on the prevalence and outcomes of type 2 diabetes in North Karelia, Finland.

METHODS

All type 2 diabetes patients (n=10,204) were analyzed from the regional electronic patient database during the years 2011 and 2012. The patient's individual laboratory data was used to assess whether hemoglobin A1c (HbA1c) was measured and whether the recommended level of HbA1c <7% (<53 mmol/l) was achieved. The variables describing socio-economic characteristics of postal code areas were retrieved from the database of Statistics Finland. Linear and logistic regression analyses were used to determine associations.

RESULTS

HbA1c had been measured in 83% of patients. Over 70% of those with HbA1c measured reached the recommended level of HbA1c. The worse the area-level socio-economic status, the more probably HbA1c was not measured. Achieving the recommended HbA1c level was associated with being female and having a better area-level socio-economic status. The age-adjusted prevalence of type 2 diabetes was not linearly dependent on the socio-economic circumstances of the postal code areas.

CONCLUSIONS

This study shows that socio-economic factors at the small area-level are associated with treatment outcomes. The information from the regional electronic patient database linked with area-level socio-economic information could be effectively utilized to improve diabetes care.

Missing in space: an evaluation of imputation methods for missing data in spatial analysis of risk factors for type II diabetes

Background

Spatial analysis is increasingly important for identifying modifiable geographic risk factors for disease. However, spatial health data from surveys are often incomplete, ranging from missing data for only a few variables, to missing data for many variables. For spatial analyses of health outcomes, selection of an appropriate imputation method is critical in order to produce the most accurate inferences.

Methods

We present a cross-validation approach to select between three imputation methods for health survey data with correlated lifestyle covariates, using as a case study, type II diabetes mellitus (DM II) risk across 71 Queensland Local Government Areas (LGAs). We compare the accuracy of mean imputation to imputation using multivariate normal and conditional autoregressive prior distributions.

Results

Choice of imputation method depends upon the application and is not necessarily the most complex method. Mean imputation was selected as the most accurate method in this application.

Conclusions

Selecting an appropriate imputation method for health survey data, after accounting for spatial correlation and correlation between covariates, allows more complete analysis of geographic risk factors for disease with more confidence in the results to inform public policy decision-making.

Electronic supplementary material

The online version of this article (doi:10.1186/1476-072X-13-47) contains supplementary material, which is available to authorized users.

Area level deprivation is an independent determinant of prevalent type 2 diabetes and obesity at the national level in Germany. Results from the National Telephone Health Interview Surveys 'German Health Update' GEDA 2009 and 2010

Objective

There is increasing evidence that prevention programmes for type 2 diabetes mellitus (T2DM) and obesity need to consider individual and regional risk factors. Our objective is to assess the independent association of area level deprivation with T2DM and obesity controlling for individual risk factors in a large study covering the whole of Germany.

Methods

We combined data from two consecutive waves of the national health interview survey ‘GEDA’ conducted by the Robert Koch Institute in 2009 and 2010. Data collection was based on computer-assisted telephone interviews. After exclusion of participants <30 years of age and those with missing responses, we included n = 33,690 participants in our analyses. The outcome variables were the 12-month prevalence of known T2DM and the prevalence of obesity (BMI ≥30 kg/m²). We also controlled for age, sex, BMI, smoking, sport, living with a partner and education. Area level deprivation of the districts was defined by the German Index of Multiple Deprivation. Logistic multilevel regression models were performed using the software SAS 9.2.

Results

Of all men and women living in the most deprived areas, 8.6% had T2DM and 16.9% were obese (least deprived areas: 5.8% for T2DM and 13.7% for obesity). For women, higher area level deprivation and lower educational level were both independently associated with higher T2DM and obesity prevalence [highest area level deprivation: OR 1.28 (95% CI: 1.05–1.55) for T2DM and OR 1.28 (95% CI: 1.10–1.49) for obesity]. For men, a similar association was only found for obesity [OR 1.20 (95% CI: 1.02–1.41)], but not for T2DM.

Conclusion

Area level deprivation is an independent, important determinant of T2DM and obesity prevalence in Germany. Identifying and targeting specific area-based risk factors should be considered an essential public health issue relevant to increasing the effectiveness of diabetes and obesity prevention.

Using spatial analysis to predict health care use at the local level: a case study of type 2 diabetes medication use and its association with demographic change and socioeconomic status

Local health status and health care use may be negatively influenced by low local socio-economic profile, population decline and population ageing. To support the need for targeted local health care, we explored spatial patterns of type 2 diabetes mellitus (T2DM) drug use at local level and determined its association with local demographic, socio-economic and access to care variables. We assessed spatial variability in these associations. We estimated the five-year prevalence of T2DM drug use (2005-2009) in persons aged 45 years and older at four-digit postal code level using the University of Groningen pharmacy database IADB.nl. Statistics Netherlands supplied data on potential predictor variables. We assessed spatial clustering, correlations and estimated a multiple linear regression model and a geographically weighted regression (GWR) model. Prevalence of T2DM medicine use ranged from 2.0% to 25.4%. The regression model included the extent of population ageing, proportion of social welfare/benefits, proportion of low incomes and proportion of pensioners, all significant positive predictors of local T2DM drug use. The GWR model demonstrated considerable spatial variability in the association between T2DM drug use and above predictors and was more accurate. The findings demonstrate the added value of spatial analysis in predicting health care use at local level.

The impact of regional deprivation and individual socio-economic status on the prevalence of Type 2 diabetes in Germany. A pooled analysis of five population-based studies

AIM

Our objective was to test the hypothesis that the prevalence of Type 2 diabetes increases with increasing regional deprivation even after controlling for individual socio-economic status.

METHODS

We pooled cross-sectional data from five German population-based studies. The data set contained information on n = 11,688 study participants (men 50.1%) aged 45-74 years, of whom 1008 people had prevalent Type 2 diabetes (men 56.2%). Logistic multilevel regression was performed to estimate odds ratios (OR) and 95% confidence intervals (CI) for diabetes prevalence. We controlled for sex, age and lifestyle risk factors, individual socio-economic status and regional deprivation, based on a new small-area deprivation measure, the German Index of Multiple Deprivation.

RESULTS

Adjusted for sex, age, body mass index (BMI), physical activity, smoking status and alcohol consumption, the prevalence of Type 2 diabetes showed a stepwise increase in risk with increasing area deprivation [OR 1.88 (95% CI 1.16-3.04) in quintile 4 and OR 2.14 (95% CI 1.29-3.55) in quintile 5 compared with the least deprived quintile 1], even after controlling for individual socio-economic status. Focusing on individual socio-economic status alone, the risk of having diabetes was significantly higher for low compared with medium or high educational level [OR 1.46 (95% CI 1.24-1.71)] and for the lowest compared with the highest income group [OR 1.53 (95% CI 1.18-1.99)].

CONCLUSION

Regional deprivation plays a significant part in the explanation of diabetes prevalence in Germany independently of individual socio-economic status. The results of the present study could help to target public health measures in deprived regions.

Multilevel Bayesian networks for the analysis of hierarchical health care data

OBJECTIVE

Large health care datasets normally have a hierarchical structure, in terms of levels, as the data have been obtained from different practices, hospitals, or regions. Multilevel regression is the technique commonly used to deal with such multilevel data. However, for the statistical analysis of interactions between entities from a domain, multilevel regression yields little to no insight. While Bayesian networks have proved to be useful for analysis of interactions, they do not have the capability to deal with hierarchical data. In this paper, we describe a new formalism, which we call multilevel Bayesian networks; its effectiveness for the analysis of hierarchically structured health care data is studied from the perspective of multimorbidity.

METHODS

Multilevel Bayesian networks are formally defined and applied to analyze clinical data from family practices in The Netherlands with the aim to predict interactions between heart failure and diabetes mellitus. We compare the results obtained with multilevel regression.

RESULTS

The results obtained by multilevel Bayesian networks closely resembled those obtained by multilevel regression. For both diseases, the area under the curve of the prediction model improved, and the net reclassification improvements were significantly positive. In addition, the models offered considerable more insight, through its internal structure, into the interactions between the diseases.

CONCLUSIONS

Multilevel Bayesian networks offer a suitable alternative to multilevel regression when analyzing hierarchical health care data. They provide more insight into the interactions between multiple diseases. Moreover, a multilevel Bayesian network model can be used for the prediction of the occurrence of multiple diseases, even when some of the predictors are unknown, which is typically the case in medicine.

[Potentially inappropriate prescriptions for the elderly: a study of health insurance reimbursements in Southeastern France]

BACKGROUND

This study conducted in the region of Provence-Alpes-Côte d'Azur (PACA) sought to assess the feasibility of constructing and using indicators of potentially inappropriate prescriptions for the elderly from health insurance reimbursement data. We present and discuss different indicators of inappropriate prescriptions for people aged 70 years or older (at-risk prescriptions, dangerous or at-risk coprescriptions, absence of necessary coprescriptions) and reports their prevalence in PACA.

METHODS

The indicators were constructed from the French list of inappropriate prescriptions, national agency guidelines, and the advice of experts in the field. The indicators selected were applied to the databases of the PACA Salaried Workers' Health Insurance Fund for 2008 for all recipients aged 70 years or older and compared according to age, sex, chronic disease status, and, after standardization for age and sex, according to district of residence.

RESULTS

In January 2009, 500,904 recipients aged 70 years or older were identified in the data base of the Salaried Workers' Health Insurance Fund, 60.8% of whom were women and 52.1% of whom had approved coverage for a chronic disease. The potentially inappropriate prescriptions most frequently observed here, in decreasing order, were: prescription of an NSAID without the coprescription of gastric protection (28.1%); long-term benzodiazepine treatment (21.5%); prescription of long half-life benzodiazepine (14.9%), and long-term treatment with NSAIDs (11.6%). Overall, the prevalence of each increased significantly with age and was higher among women and people with chronic diseases. Significant variations were also observed between the different districts of PACA.

CONCLUSION

Our results confirm that a substantial proportion of elderly people receive potentially inappropriate prescriptions. They also suggest that health insurance reimbursement data could be used in some prescription domains for monitoring trends in the potentially inappropriate prescriptions in the populations of various territories, provided that specific limitations are considered.