Harokopio University
Development of spatio-temporal statistical models for investigating the socioeconomic and environmental determinants of health-related quality of life and nutritional habits, in the presence of missing data: a spatio-temporal epidemiological approach
Abstract
dc:descriptionAim: The aims of the present thesis are (a) to investigate the geographic determinants of population’s health in general, and in particular, the health-related quality of life, as well as nutritional habits, in terms of their level of adherence to the Mediterranean diet; and (b) to propose a robust approach for imputing missing data at an aggregate geospatial level, since missing data are widely acknowledged as one of the greatest obstacles in geospatial epidemiological analysis. Material and Methods: For the first aim, data from the ATTICA epidemiological study (2002-2012) were used, while socioeconomic and geographical data for the Athens metropolitan area were retrieved from various Official Statistics sources (i.e., ELSTAT, Urban Atlas, Directorate-General for Enterprise and Industry). The ATTICA study, being consisted of a large representative sample of the Greek population residing in the Athens metropolitan area, is a prospective, observational cohort investigation, that was initiated in 2001-2002 and included two follow- up examinations in 2006 and in 2011-2012, respectively. The sampling procedure of the ATTICA epidemiological study was random, multistage (by city) and it was based on the age – sex distribution of the province of Attica provided by the National Statistical Service, according to the census of 2001. Only one participant per household was enrolled, while according to the study's protocol all people living in institutions were excluded from the sampling. From May 2001 to December 2002, 4056 inhabitants from the above area were randomly selected and asked to participate into the study. Of them, a representative sample of 3042 apparently healthy volunteers residing in 40 municipalities of the Athens metropolitan area agreed to participate at baseline (2001–2002). For the scope of the present work and due to the incomplete sample data for geographical identification, the analyses were restricted to 40 municipalities of the Athens metropolitan Area with a population of 3,338,131. Based on their socio- economic characteristics, the remaining 40 municipalities were reformed in five sectors to be characterized by common socioeconomic status. For the second aim, a methodological Framework For Imputing missiNg spatial data at an aggregate level and guaranteeing personal daTa privacy, named AFFINITY hereinafter, is introduced. By this framework, imputing missing spatial data can be utilized to gain the most possible value of the spatially referenced information. In particular, AFFINITY focuses on finding the accuracy in estimating missing values at an aggregate- municipality level under various missing data scenarios, using the following steps: (i) for each municipality with missing spatially referenced data, the geodetic distances from other areas with full data are being calculated, based on the Haversine formula, so as to find the three nearest neighbourhood municipalities for each one, and (ii) after finding the three nearest neighbourhood municipalities, each variable with missing data is imputed with the weighted mean of the neighbours’ values in the same variable. Furthermore, to show the robustness of the proposed AFFINITY method, we compare the methodology with other imputation methodologies which are being commonly used in practise, namely the mean and median imputation, as well as the stochastic and the deterministic regression imputation. Results:Level of adherence to the Mediterranean diet: A high level of adherence to the Mediterranean diet was presented within the study area, principally in case of regions with higher socioeconomic status. Moreover, higher adherence to the Mediterranean diet was detected in areas with a greater proportion of females and older population, as well as, in areas with higher number of employed residents and lower percent of immigrant population. Besides, the availability of supermarkets or street markets contributed differently to residents’ dietary habits according to the socioeconomic status. Finally, regions covered by green urban areas at a greater extent, seemed to have a higher level of adherence to the Mediterranean diet. Health- Related Quality of Life (HRQoL): Better HRQoL was detected in areas with higher socioeconomic status, with younger residents following healthier lifestyle and eating habits, higher percent of married individuals, lower percent of illiterate people and rate of immigration, as well as in municipalities with higher land coverage by green urban areas. A significant interaction between sex and studied area on HRQoL was observed, suggesting that men and women follow different patterns regarding their quality of life. Metabolic Syndrome: Approximately 20% of the people residing in the study area were diagnosed with MetS, with its prevalence being almost two times higher in men compared to women. Areas more extensively covered by green urban spaces and sports facilities were shown to have a lower prevalence of MetS, while greater density and availability of street markets were inversely related to MetS prevalence in both sexes. In addition, the present analysis revealed that the beneficial role of the built environment’s characteristics on MetS prevalence was significantly stronger in the male population, while the preventive effect of the food environment’s characteristics was almost 1.5 times stronger in the female population. Cardiovascular Disease (CVD) and Cardiometabolic risk factors: Hypercholesterolaemia was found to be the most prevalent clinical risk factor in the study area, while higher prevalence for all the risk factors were presented in areas of lower socioeconomic status (SES), where a significantly higher number of new CVD cases in the decade were detected as well. 10-year CVD incidence was also found to be significantly elevated in areas characterized by worse health status in terms of the risk factors’ prevalence, while the sex and age- specific CVD incidence rates were found to be differently affected by these factors. Interestingly, despite the fact that CVDs are acknowledged as a man’s disease, areas of higher (South & North) and moderate (East) SES were found to be characterized by higher CVD incidence rates in females, while in areas of lower SES (West), higher CVD incidence rates were found in males. Imputation of missing spatial data at an aggregate level (AFFINITY method): Under all missing data scenarios, and under the assumptions of MAR and MCAR, as well as under the ‘pragmatic scenario’, the AFFINITY method succeeded lower absolute and relative error in the estimation of the mean values, when compared to the rest imputation methodologies. Specifically, the relative error for the AFFINITY method ranged on average between 0.5% and 2.51% in case of MCAR spatially referenced data, between 0.37% and 0.69% in case of MAR spatially referenced data and between 0.53% and 9.7% on case of the ‘pragmatic’ scenario. In contrast, both the deterministic, as well as the stochastic regression imputation methodologies, were found to be the less accurate techniques under all the examined scenarios. Conclusions: Although individualized prevention and treatment approaches are necessary to decrease the burden of diseases, environmental modifications that promote healthy behaviors represent an essential health approach. In particular, in line with the current European policy developments, our findings support the transformation of public health actions to local authorities and provide input for policy makers to better understand how population’s health is affected by the characteristics of the area in which they live. In addition, the identification of areas where the intensity of the studied associations is stronger, can be considered a crucial methodological aspect for the design and implementation of more effective CVD prevention programs at the community level. Finally, a robust, in terms of accuracy, methodological framework was proposed here, which is expected to help the decision- makers to impute accurately the missing values at aggregate level spatial datasets, improving in this way both the power, as well as the precision in further statistical analyses, while at the same time the risk of identification especially in small areas can be overcome.
Degree
thesis:*- Grantor dc:publisher
- Harokopio University
- Year dc:date
- 2023
Author and committee
dc:creator, dc:contributor.*- Authors dc:creator
-
- Tsiampalis, Thomas
- Τσιαμπαλής, Θωμάς
Subjects
dc:subject × 16- Βιοστατιστική
- Επιδημιολογία
- Χωρική ανάλυση
- Ελλιπή δεδομένα
- Καρδιαγγειακά νοσήματα
- Biostatistics
- Epidemiology
- Spatial analysis
- Missing data
- Cardiovascular disease
- Επιστήμες Υγείας
- Ιατρική και Επιστήμες Υγείας
- Δημόσια/περιβαλλοντική υγεία και Υγεία της εργασίας
- Health Sciences
- Medical and Health Sciences
- Public Health, Environmental and Occupational Health
Rights
- Language dc:language
- eng
Identifiers
dc:identifier.*- Identifier
- 10.12681/eadd/54272