An epidemiological and gis-based analysis of mortality in selected rare diseases

Abstract

Introduction: Rare diseases (RD) are defined as those affecting no more than 5 in 10,000 people within the European Union (EU). It is estimated that between 27 and 36 million people might be affected by RD in the EU (which amounts to between 6 and 8% of its total population), with them consequently having become a Public Health issue. The epidemiological research of the over 6,000 accounted for RD proves harder than that of more frequent diseases, due to their specific characteristics (e.g. heterogeneity, low number of cases, case dispersion or difficulties in codification, to name a few). When it comes to mortality, the availability of national population-based data and standardized registries are very useful when researching RD, since the reduction of mortality is a public health policy goal. Geographic Information Systems allowing for the integration, analysis and representation of geospatial data have proved to be a valuable tool with which to study the mortality distribution and its spatio-temporal evolution. Knowing the geographical patterns of mortality and their relationships with other spatial variables helps more accurately describe its epidemiology, as well as providing important baseline elements when searching for causal relationships between environmental factors and the diseases object of analysis. Objective: The general aim of this doctoral thesis is to deepen the knowledge on mortality attributed to RDs in Spain from a geographical and epidemiological perspective. The specific objectives are the following: Objective 1: To identify difficulties related to the optimal geographic units of representation when working with RDs and to provide sound advice for the election of the most appropriate unit, aiming for a better epidemiological and cartographic result. Objective 2: To evaluate the Spanish population’s time trends in mortality due to RDs over three decades. This target will be achieved by selecting Huntington’s disease (HD) and Granulomatosis with polyangiitis (GPA). Objective 3: To identify geographic patterns in mortality attributed to RDs, in order to trace variations in mortality among spatial units throughout Spain. This objective will be tackled through the analysis of GPA, HD and Motor neuron diseases (MND) as the RDs of reference. Objective 4: To describe the characteristics of heavy metal emissions to river basins, and to explore the possible correlation that exists between their location and the mortality attributed to MND, in order to help understand the etiology of this group of diseases. Methodology: This doctoral thesis has the Spanish territory as its study framework. It covers the years 1984 to 2016, at maximum, and uses the annual death registry provided by the National Statistics Institute as the basis for analysis. The decease-related data relevant to each of the three analyzed RDs were selected taking into account the underlying cause of death, whilst data relative to the heavy metal emissions to river basins were collected from the European Pollutant Release and Transfer Register (E-PRTR). Shapefiles containing information on geographic units and hydrography were downloaded from the National Mapping Agency. As a starting point, the suitability of the available aggregation units (municipalities, districts and provinces) was analyzed, by means of the epidemiological and geospatial comparison of different indicators, such as the Standardized Mortality Ratio (SMR), smoothed SMR and local indicators of spatial association (LISA). Secondly, regarding epidemiological mortality indicators, time trends in HD and GPA were analyzed by calculating age-adjusted mortality rates, and subsequently smoothed. Time trends were assessed by a joinpoint regression. As to the geographical distribution of mortality for the three selected diseases, the SMRs and their 95% confidence intervals were calculated. SMR is defined as the existing ratio between observed and expected deaths within a geographical unit. Smoothed SMR and its posterior probability were also calculated. The smoothing process takes into account the observed and expected values of the neighboring geographical units and helps identify spatial patterns of death risk. Lastly, the sites that release heavy metal emissions into the rivers were geocoded. In order to study their influence on MND deaths, we took exposed municipalities to be those downstream from the point of emission, within 20 km of it. The mortality of the exposed municipalities was compared to that of the non-exposed ones through a Poisson regression model. The obtained epidemiological results of a spatial nature were represented cartographically. Results: The modifiable areal unit problem (MAUP) proved relevant when analyzing different aggregation units for the study of RDs, since significant differences were found when comparing epidemiological indicators among the three existing units. The number of cases occurring due to disease was a determining factor when choosing the most appropriate level of aggregation. The district level (comarca) has proved the most appropriate to display the variability of mortality in Spain, while still displaying relevant spatial detail. Regarding the evolution of mortality rates for the analyzed RDs, significant time trends were identified in both HD and GPA within the periods of study. The mortality rate from HD increased annually by 3.44% (p <0.001) throughout, while that for GPA displayed an annual increase of 20.6% between 1984 and 1992, with a 1.91% annual increase thereafter. (p <0.05). The average ages of death from GPA and EH rose by 0.78% and 0.59% (p <0.05), respectively, within the studied timeframe. As for the spatial analysis of RD-related mortality, the three reference diseases showed geographical variability within the Spanish territory. Regarding HD, higher SMRs and smoothed SMRs were found in some districts to the Southwest of the peninsula. GPA posed an increased death risk for the inhabitants of two specific districts in the North of Spain, and for a further two in the South, while in the case of MND, a comparison between municipalities showed those located in the North to be at a higher risk of death from the disease than those to the South. The Poisson regression analysis conducted on the variation of mortality attributed to MND in relation to the emission of heavy metals to river basins showed that MND mortality rates were 18.4% higher in the exposed municipalities than it was in non-exposed ones. The figure remained relevant both globally and for the analyses carried out individually for each type of metal. Conclusions: RDs are of increasing interest due to the high impact they have on affected patients, their families and society in general. Therefore, it is necessary to improve the information available on them through research, carried out over multidisciplinary perspectives. This doctoral thesis has enriched the body of knowledge on spatial variability in mortality from RDs, by adding a geographical approach to its analysis. Some of the difficulties presented by the epidemiological study of these low prevalence diseases have been analyzed. The low number of cases and the need for spatial aggregation have the MAUP as a consequence. Relevant information regarding the temporal and geographic variability in mortality for selected RDs has been provided, as well as having explored the possible associations with geographic factors. The results of this study offer very valuable information for health planning, since they support and justify the need for preventive and assistive action in those areas where there have been more inequalities in the risk of death due to these diseases. In addition, by establishing a correlation between mortality from RDs and suspect environmental factors, we offer clues for future studies that may delve into the discovery of RD causes.