Spatial measurement errors in the field of spatial epidemiology

Comparing mapped park and greenspace boundaries in Philadelphia: implications for exposure assessment in health studies

An important consideration in studies of the relationship between greenspace exposure and health is the use of mapped data to assign geographic exposures to participants. Previous studies have used validated data from municipal park departments to describe the boundaries of public greenspaces. However, this approach assumes that these data accurately describe park boundaries, that formal parks fully capture the park and greenspace exposure of residents, and (for studies that use personal GPS traces to assign participant exposures) that time spent within these boundaries represents time spent in greenspace. These assumptions are tested using a comparison and ground-truthing of four sources of mapped park and greenspace data in Philadelphia, Pennsylvania: PAD-US-AR, Philadelphia Parks and Recreation, the Delaware Valley Regional Planning Commission, and Open Street Maps. We find several important differences and tradeoffs in these data: the incorporation of highways and building lots within park boundaries, the inclusion or exclusion of formal park spaces (federal, state, and nonprofit), the exclusion of informal parks and greenspaces, and inconsistent boundaries for a linear park. Health researchers may wish to consider these issues when conducting studies using boundary data to assign park exposure.

Capturing emergency dispatch address points as geocoding candidates to quantify delimited confidence in residential geolocation

BACKGROUND

In response to citizens' concerns about elevated cancer incidence in their locales, US CDC proposed publishing cancer incidence at sub-county scales. At these scales, confidence in patients' residential geolocation becomes a key constraint of geospatial analysis. To support monitoring cancer incidence in sub-county areas, we presented summary metrics to numerically delimit confidence in residential geolocation.

RESULTS

We defined a concept of Residential Address Discriminant Power (RADP) as theoretically perfect within all residential addresses and its practical application, i.e., using Emergency Dispatch (ED) Address Point Candidates of Equivalent Likelihood (CEL) to quantify Residential Geolocation Discriminant Power (RGDP) to approximate RADP. Leveraging different productivity of probabilistic, deterministic, and interactive geocoding record linkage, we simultaneously detected CEL for 5,807 cancer cases reported to North Carolina Central Cancer Registry (NC CCR)- in January 2022. Batch-match probabilistic and deterministic algorithms matched 86.0% cases to their unique ED address point candidates or a CEL, 4.4% to parcel site address, and 1.4% to street centerline. Interactively geocoded cases were 8.2%. To demonstrate differences in residential geolocation confidence between enumeration areas, we calculated sRGDP for cancer cases by county and assessed the existing uncertainty within the ED data, i.e., identified duplicate addresses (as CEL) for each ED address point in the 2014 version of the NC ED data and calculated ED_sRGDP by county. Both summary RGDP (sRGDP) (0.62-1.00) and ED_sRGDP (0.36-1.00) varied across counties and were lower in rural counties (p < 0.05); sRGDP correlated with ED_sRGDP (r = 0.42, p < 0.001). The discussion covered multiple conceptual and economic issues attendant to quantifying confidence in residential geolocation and presented a set of organizing principles for future work.

CONCLUSIONS

Our methodology produces simple metrics - sRGDP - to capture confidence in residential geolocation via leveraging ED address points as CEL. Two facts demonstrate the usefulness of sRGDP as area-based summary metrics: sRGDP variability between counties and the overall lower quality of residential geolocation in rural vs. urban counties. Low sRGDP for the cancer cases within the area of interest helps manage expectations for the uncertainty in cancer incidence data. By supplementing cancer incidence data with sRGDP and ED_sRGDP, CCRs can demonstrate transparency in geocoding success, which may help win citizen trust.

Human-Biting Ixodes scapularis Submissions to a Crowd-Funded Tick Testing Program Correlate with the Incidence of Rare Tick-Borne Disease: A Seven-Year Retrospective Study of Anaplasmosis and Babesiosis in Massachusetts

Tick-borne zoonoses pose a serious burden to global public health. To understand the distribution and determinants of these diseases, the many entangled environment-vector-host interactions which influence risk must be considered. Previous studies have evaluated how passive tick testing surveillance measures connect with the incidence of human Lyme disease. The present study sought to extend this to babesiosis and anaplasmosis, two rare tick-borne diseases. Human cases reported to the Massachusetts Department of Health and submissions to TickReport tick testing services between 2015 and 2021 were retrospectively analyzed. Moderate-to-strong town-level correlations using Spearman's Rho (ρ) were established between Ixodes scapularis submissions (total, infected, adult, and nymphal) and human disease. Aggregated ρ values ranged from 0.708 to 0.830 for anaplasmosis and 0.552 to 0.684 for babesiosis. Point observations maintained similar patterns but were slightly weaker, with mild year-to-year variation. The seasonality of tick submissions and demographics of bite victims also correlated well with reported disease. Future studies should assess how this information may best complement human disease reporting and entomological surveys as proxies for Lyme disease incidence in intervention studies, and how it may be used to better understand the dynamics of human-tick encounters.

Exploring agricultural land-use and childhood malaria associations in sub-Saharan Africa

Agriculture in Africa is rapidly expanding but with this comes potential disbenefits for the environment and human health. Here, we retrospectively assess whether childhood malaria in sub-Saharan Africa varies across differing agricultural land uses after controlling for socio-economic and environmental confounders. Using a multi-model inference hierarchical modelling framework, we found that rainfed cropland was associated with increased malaria in rural (OR 1.10, CI 1.03-1.18) but not urban areas, while irrigated or post flooding cropland was associated with malaria in urban (OR 1.09, CI 1.00-1.18) but not rural areas. In contrast, although malaria was associated with complete forest cover (OR 1.35, CI 1.24-1.47), the presence of natural vegetation in agricultural lands potentially reduces the odds of malaria depending on rural-urban context. In contrast, no associations with malaria were observed for natural vegetation interspersed with cropland (veg-dominant mosaic). Agricultural expansion through rainfed or irrigated cropland may increase childhood malaria in rural or urban contexts in sub-Saharan Africa but retaining some natural vegetation within croplands could help mitigate this risk and provide environmental co-benefits.

Spatial Heterogeneity in Positional Errors: A Comparison of Two Residential Geocoding Efforts in the Agricultural Health Study

Geocoding is a powerful tool for environmental exposure assessments that rely on spatial databases. Geocoding processes, locators, and reference datasets have improved over time; however, improvements have not been well-characterized. Enrollment addresses for the Agricultural Health Study, a cohort of pesticide applicators and their spouses in Iowa (IA) and North Carolina (NC), were geocoded in 2012–2016 and then again in 2019. We calculated distances between geocodes in the two periods. For a subset, we computed positional errors using “gold standard” rooftop coordinates (IA; N = 3566) or Global Positioning Systems (GPS) (IA and NC; N = 1258) and compared errors between periods. We used linear regression to model the change in positional error between time periods (improvement) by rural status and population density, and we used spatial relative risk functions to identify areas with significant improvement. Median improvement between time periods in IA was 41 m (interquartile range, IQR: −2 to 168) and 9 m (IQR: −80 to 133) based on rooftop coordinates and GPS, respectively. Median improvement in NC was 42 m (IQR: −1 to 109 m) based on GPS. Positional error was greater in rural and low-density areas compared to in towns and more densely populated areas. Areas of significant improvement in accuracy were identified and mapped across both states. Our findings underscore the importance of evaluating determinants and spatial distributions of errors in geocodes used in environmental epidemiology studies.

Association of ambient air pollution with cardiovascular disease risks in people with type 2 diabetes: a Bayesian spatial survival analysis

BACKGROUND

Evidence is limited on excess risks of cardiovascular diseases (CVDs) associated with ambient air pollution in diabetic populations. Survival analyses without considering the spatial structure and possible spatial correlations in health and environmental data may affect the precision of estimation of adverse environmental pollution effects. We assessed the association between air pollution and CVDs in type 2 diabetes through a Bayesian spatial survival approach.

METHODS

Taiwan's national-level health claims and air pollution databases were utilized. Fine individual-level latitude and longitude were used to determine pollution exposure. The exponential spatial correlation between air pollution and CVDs was analyzed in our Bayesian model compared to traditional Weibull and Cox models.

RESULTS

There were 2072 diabetic patients included in analyses. PM_2.5 and SO₂ were significant CVD risk factors in our Bayesian model, but such associations were attenuated or underestimated in traditional models; adjusted hazard ratio (HR) and 95% credible interval (CrI) or confidence interval (CI) of CVDs for a 1 μg/m³ increase in the monthly PM_2.5 concentration for our model, the Weibull and Cox models was 1.040 (1.004-1.073), 0.994 (0.984-1.004), and 0.994 (0.984-1.004), respectively. With a 1 ppb increase in the monthly SO₂ concentration, adjusted HR (95% CrI or CI) was 1.886 (1.642-2.113), 1.092 (1.022-1.168), and 1.091 (1.021-1.166) for these models, respectively.

CONCLUSIONS

Against traditional non-spatial analyses, our Bayesian spatial survival model enhances the assessment precision for environmental research with spatial survival data to reveal significant adverse cardiovascular effects of air pollution among vulnerable diabetic patients.

Geocoding Error, Spatial Uncertainty, and Implications for Exposure Assessment and Environmental Epidemiology

Although environmental epidemiology studies often rely on geocoding procedures in the process of assigning spatial exposure estimates, geocoding methods are not commonly reported, nor are consequent errors in exposure assignment explored. Geocoding methods differ in accuracy, however, and, given the increasing refinement of available exposure models for air pollution and other exposures, geocoding error may account for an increasingly larger proportion of exposure misclassification. We used residential addresses from a reasonably large, dense dataset of asthma emergency department visits from all New York City hospitals (n = 21,183; 26.9 addresses/km²), and geocoded each using three methods (Address Point, Street Segment, Parcel Centroid). We compared missingness and spatial patterning therein, quantified distance and directional errors, and quantified impacts on pollution exposure estimates and assignment to Census areas for sociodemographic characterization. Parcel Centroids had the highest overall missingness rate (38.1%, Address Point = 9.6%, Street Segment = 6.1%), and spatial clustering in missingness was significant for all methods, though its spatial patterns differed. Street Segment geocodes had the largest mean distance error (µ = 29.2 (SD = 26.2) m; vs. µ = 15.9 (SD = 17.7) m for Parcel Centroids), and the strongest spatial patterns therein. We found substantial over- and under-estimation of pollution exposures, with greater error for higher pollutant concentrations, but minimal impact on Census area assignment. Finally, we developed surfaces of spatial patterns in errors in order to identify locations in the study area where exposures may be over-/under-estimated. Our observations provide insights towards refining geocoding methods for epidemiology, and suggest methods for quantifying and interpreting geocoding error with respect to exposure misclassification, towards understanding potential impacts on health effect estimates.

Spatial predictive properties of built environment characteristics assessed by drop-and-spin virtual neighborhood auditing

Background

Virtual neighborhood audits have been used to visually assess characteristics of the built environment for health research. Few studies have investigated spatial predictive properties of audit item responses patterns, which are important for sampling efficiency and audit item selection. We investigated the spatial properties, with a focus on predictive accuracy, of 31 individual audit items related to built environment in a major Metropolitan region of the Northeast United States.

Methods

Approximately 8000 Google Street View (GSV) scenes were assessed using the CANVAS virtual audit tool. Eleven trained raters audited the 360^° view of each GSV scene for 10 sidewalk-, 10 intersection-, and 11 neighborhood physical disorder-related characteristics. Nested semivariograms and regression Kriging were used to investigate the presence and influence of both large- and small-spatial scale relationships as well as the role of rater variability on audit item spatial properties (measurement error, spatial autocorrelation, prediction accuracy). Receiver Operator Curve (ROC) Area Under the Curve (AUC) based on cross-validated spatial models summarized overall predictive accuracy. Correlations between predicted audit item responses and select demographic, economic, and housing characteristics were investigated.

Results

Prediction accuracy was better within spatial models of all items accounting for both small-scale and large- spatial scale variation (vs large-scale only), and further improved with additional adjustment for rater in a majority of modeled items. Spatial predictive accuracy was considered ‘Excellent’ (0.8 ≤ ROC AUC < 0.9) for full models of all but four items. Predictive accuracy was highest and improved the most with rater adjustment for neighborhood physical disorder-related items. The largest gains in predictive accuracy comparing large- + small-scale to large-scale only models were among intersection- and sidewalk-items. Predicted responses to neighborhood physical disorder-related items correlated strongly with one another and were also strongly correlated with racial-ethnic composition, socioeconomic indicators, and residential mobility.

Conclusions

Audits of sidewalk and intersection characteristics exhibit pronounced variability, requiring more spatially dense samples than neighborhood physical disorder audits do for equivalent accuracy. Incorporating rater effects into spatial models improves predictive accuracy especially among neighborhood physical disorder-related items.

Association of neighbourhood migrant density and risk of non-affective psychosis: a national, longitudinal cohort study

BACKGROUND

Elevated risk of psychotic disorders in migrant groups is a public mental health priority. We investigated whether living in areas of high own-region migrant density was associated with reduced risk of psychotic disorders among migrants and their children, and whether generation status, probable visible minority status, or region-of-origin affected this relationship.

METHODS

We used the Swedish registers to identify migrants and their children born between Jan 1, 1982, and Dec 31, 1996, and living in Sweden on or after their 15th birthday. We tracked all included participants from age 15 years or date of migration until emigration, death, or study end (Dec 31, 2016). The outcome was an ICD-10 diagnosis of non-affective psychosis (F20-29). We calculated own-region and generation-specific own-region density within the 9208 small areas for market statistics neighbourhoods in Sweden, and estimated the relationship between density and diagnosis of non-affective psychotic disorders using multilevel Cox proportional hazards models, adjusting for individual confounders (generation status, age, sex, calendar year, lone dwelling, and time since migration [migrants only]), family confounders (family income, family unemployment, and social welfare), and neighbourhood confounders (deprivation index, population density, and proportion of lone dwellings), and using the Akaike information criterion (AIC) to compare model fit.

FINDINGS

Of 468 223 individuals included in the final cohort, 4582 (1·0%) had non-affective psychotic disorder. Lower own-region migrant density was associated with increased risk of psychotic disorders among migrants (hazard ratio [HR] 1·05, 95% CI 1·02-1·07 per 5% decrease) and children of migrants (1·03, 1·01-1·06), after adjustment. These effects were stronger for probable visible minority migrants (1·07, 1·04-1·11), including migrants from Asia (1·42, 1·15-1·76) and sub-Saharan Africa (1·28, 1·15-1·44), but not migrants from probable non-visible minority backgrounds (0·99, 0·94-1·04). Among migrants, adding generation status to the measure of own-region density provided a better fit to the data than overall own-region migrant density (AIC 36 103 vs 36 106, respectively), with a 5% decrease in generation-specific migrant density corresponding to a HR of 1·07 (1·04-1·11).

INTERPRETATION

Migrant density was associated with non-affective psychosis risk in migrants and their children. Stronger protective effects of migrant density were found for probable visible minority migrants and migrants from Asia and sub-Saharan Africa. For migrants, this risk intersected with generation status. Together, these results suggest that this health inequality is socially constructed.

FUNDING

Wellcome Trust, Royal Society, Mental Health Research UK, University College London, National Institute for Health Research, Swedish Research Council, and FORTE.

Developing a surveillance system of sub-county data: Finding suitable population thresholds for geographic aggregations

The Centers for Disease Control and Prevention's National Environmental Public Health Tracking Program created standardized sub-county geographies that are comparable over time, place, and outcomes. Expected census tract-level counts were calculated for asthma emergency department visits and lung cancer. Census tracts were aggregated for various total population and sub-population thresholds, then suppression and stability were examined. A total of 5,000 persons was recommended for the more common outcome scheme and a total of 20,000 persons was recommended for the rare outcome scheme. Health outcomes with a median case count of 17.0 cases or higher should produce stable estimates at the census tract level. This project generated recommendations for three sub-county geographies that will be useful for surveillance purposes: census tract, a more common outcome aggregation scheme, and a rare outcome aggregation scheme. This methodology can be applied anywhere to aggregate geographic units and produce stable rates at a finer resolution.

Modelling the impact of MAUP on environmental drivers for Schistosoma japonicum prevalence

BACKGROUND

The modifiable areal unit problem (MAUP) arises when the support size of a spatial variable affects the relationship between prevalence and environmental risk factors. Its effect on schistosomiasis modelling studies could lead to unreliable parameter estimates. The present research aims to quantify MAUP effects on environmental drivers of Schistosoma japonicum infection by (i) bringing all covariates to the same spatial support, (ii) estimating individual-level regression parameters at 30 m, 90 m, 250 m, 500 m and 1 km spatial supports, and (iii) quantifying the differences between parameter estimates using five models.

METHODS

We modelled the prevalence of Schistosoma japonicum using sub-provinces health outcome data and pixel-level environmental data. We estimated and compared regression coefficients from convolution models using Bayesian statistics.

RESULTS

Increasing the spatial support to 500 m gradually increased the parameter estimates and their associated uncertainties. Abrupt changes in the parameter estimates occur at 1 km spatial support, resulting in loss of significance of almost all the covariates. No significant differences were found between the predicted values and their uncertainties from the five models. We provide suggestions to define an appropriate spatial data structure for modelling that gives more reliable parameter estimates and a clear relationship between risk factors and the disease.

CONCLUSIONS

Inclusion of quantified MAUP effects was important in this study on schistosomiasis. This will support helminth control programmes by providing reliable parameter estimates at the same spatial support and suggesting the use of an adequate spatial data structure, to generate reliable maps that could guide efficient mass drug administration campaigns.

Enhancing Models and Measurements of Traffic-Related Air Pollutants for Health Studies Using Dispersion Modeling and Bayesian Data Fusion

INTRODUCTION

The adverse health effects associated with exposure to traffic-related air pollutants (TRAPs) remain a key public health issue. Often, exposure assessments have not represented the small-scale variation and elevated concentrations found near major roads and in urban settings. This research explores approaches aimed at improving exposure estimates of TRAPs that can reduce exposure measurement error when used in health studies. We consider dispersion models designed specifically for the near-road environment, as well as spatiotemporal and data fusion models. These approaches are implemented and evaluated utilizing data collected in recent modeling, monitoring, and epidemiological studies conducted in Detroit, Michigan.

APPROACH

Dispersion models, which estimate near-road pollutant concentrations and individual exposures based on first principles - and in particular, high fidelity models - can provide great flexibility and theoretical strength. They can represent the spatial variability of TRAP concentrations at locations not measured by conventional and spatially sparse air quality monitoring networks. A number of enhancements to dispersion modeling and mobile on-road emissions inventories were considered, including the representation of link-based road networks and updated estimates of temporal allocation of traffic activity, emission factors, and meteorological inputs. The recently developed Research LINE-source model (RLINE), a Gaussian line-source dispersion model specifically designed for the near-road environment, was used in an operational evaluation that compared predicted concentrations of nitrogen oxides (NOx), carbon monoxide (CO), and PM2.5 (particulate matter ≤ 2.5 µm in aerodynamic diameter) with observed concentrations at air quality monitoring stations located near high-traffic roads. Spatiotemporal and data fusion models provided additional and complementary approaches for estimating TRAP exposures. We formulated both nonstationary universal kriging models that exploit the spatial correlation in the monitoring data, and data fusion models that leverage the information contained in both the monitoring data and the output of numerical models, specifically RLINE. These models were evaluated using observations of nitric oxide (NO), NOx, black carbon (BC), and PM2.5 monitored along transects crossing major roads in Detroit. We also examined model assumptions, including the appropriateness of the covariance functions, errors in RLINE outputs, and the effects of jointly modeling two pollutants and using an updated emission inventory.

RESULTS

For CO and NOx, dispersion model performance was best when monitoring sites were close to major roads, during downwind conditions, during weekdays, and during certain seasons. The ability to discern local and particularly the traffic-related portion of PM2.5 was limited, a result of high background levels, the sparseness of the monitoring network, and large uncertainties for certain sources (e.g., area, fugitive) and some processes (e.g., formation of secondary aerosols). Sensitivity analyses of alternative meteorological inputs and updated emission factors showed some performance gain when using local (on-site) meteorological data and updated inventories. Overall, the operational evaluation suggested RLINE's usefulness for estimating spatially and temporally resolved exposure estimates. The application of the universal kriging models confirmed that wind speed and direction are important drivers of nonstationarity in pollutant concentrations, and that these models can predict exposure estimates that have lower prediction errors than do stationary model counterparts. The application of the Bayesian data fusion models suggested that the RLINE output had a spatially varying additive bias for NOx and PM2.5 and provided little additional information for NOx, besides what is already contained in traffic and geographical information system (GIS) covariates, but had improved estimates of PM2.5 concentrations. Results of the nonstationary Bayesian data fusion model that used RLINE output across a field spanning the measurement sites were similar to a regression-based Bayesian data fusion approach that used only RLINE output at the monitoring locations, with the latter being computationally less burdensome. Using the regression-based Bayesian data fusion model, we found that RLINE with the updated emission inventory provided results that were more useful for estimating NOx concentration at unmonitored sites, but the updated emission inventory did not improve predictions of PM2.5 concentrations. Joint modeling of NOx and PM2.5 was not useful, a result of differences in RLINE's utility in predicting PM2.5 and NOx - useful for the former, but not for the latter - and differences in the spatial dependence structures of the two pollutants. Overall, information provided by RLINE was shown to have the potential to improve spatiotemporal estimates of TRAP concentrations.

CONCLUSIONS

The study results should be interpreted and generalized cautiously given the limitations of the data used. Similar analyses in other settings are recommended for confirming and extending our findings. Still, the study highlights considerations that are relevant for exposure estimates used in health studies. The ability of a dispersion model to accurately reproduce and predict a pollutant depends on the pollutant as well as on spatial and temporal factors, such as the distance and direction from the road, time-of-day, and day-of-week. The nature and source of exposure measurement errors should be taken into consideration, particularly in health studies that take advantage of time- activity information that describes where and when individuals are exposed to pollution. Efforts to refine model inputs and improve model performance can be helpful; meteorological inputs may be the most critical. For both dispersion and spatiotemporal statistical models, sufficient and high-quality monitoring data are essential for developing and evaluating these models. Our analyses using Bayesian data fusion models confirm the presence of spatially varying errors in dispersion model outputs and allow quantification of both the magnitude and the spatial nature of these errors. This valuable information can be leveraged in health studies examining air pollution exposure as well as in studies informing regulatory responses.

Evaluation of data accuracies within a comprehensive geospatial-health data surveillance platform: SOMAARTH Demographic Development and Environmental Surveillance Site, Palwal, Haryana, India

Evidence exists of an increasing prevalence of chronic conditions within developed and developing nations, notably for priority population groups. The need for the collection of geospatial data to monitor the health impact of rapid social-environmental and economic changes occurring in these countries is being increasingly recognized. Rigorous accuracy assessment of such geospatial data is required to enable error estimation, and ultimately, data utility for exploring population health. This research outlines findings from a field-based evaluation exercise of the SOMAARTH DDESS geospatial-health platform. Participatory-based mixed methods have been employed within Palwal-India to capture villager perspectives on built infrastructure across 51 villages. This study, conducted in 2013, included an assessment of data element position and attribute accuracy undertaken in six villages, documenting mapping errors and land parcel changes. Descriptive analyses of 5.1% (n = 455) of land parcels highlighted some discrepancies in position (6.4%) and attribute (4.2%) accuracy, and land parcel changes (17.4%). Furthermore, the evaluation led to a refinement of the existing geospatial health platform incorporating ground-truthed reflections from the participatory field exercise. The evaluation of geospatial data accuracies contributes to understandings on global public health surveillance systems, outlining the need to systematically consider assessment of environmental features in relation to lifestyle-related diseases.

Modeling Schistosoma japonicum Infection under Pure Specification Bias: Impact of Environmental Drivers of Infection

Uncertainties in spatial modeling studies of schistosomiasis (SCH) are relevant for the reliable identification of at-risk populations. Ecological fallacy occurs when ecological or group-level analyses, such as spatial aggregations at a specific administrative level, are carried out for an individual-level inference. This could lead to the unreliable identification of at-risk populations, and consequently to fallacies in the drugs’ allocation strategies and their cost-effectiveness. A specific form of ecological fallacy is pure specification bias. The present research aims to quantify its effect on the parameter estimates of various environmental covariates used as drivers for SCH infection. This is done by (i) using a spatial convolution model that removes pure specification bias, (ii) estimating group and individual-level covariate regression parameters, and (iii) quantifying the difference between the parameter estimates and the predicted disease outcomes from the convolution and ecological models. We modeled the prevalence of Schistosoma japonicum using group-level health outcome data, and city-level environmental data as a proxy for individual-level exposure. We included environmental data such as water and vegetation indexes, distance to water bodies, day and night land surface temperature, and elevation. We estimated and compared the convolution and ecological model parameter estimates using Bayesian statistics. Covariate parameter estimates from the convolution and ecological models differed between 0.03 for the nearest distance to water bodies (NDWB), and 0.28 for the normalized difference water index (NDWI). The convolution model presented lower uncertainties in most of the parameter estimates, except for NDWB. High differences in uncertainty were found in night land surface temperature (0.23) and elevation (0.13). No significant differences were found between the predicted values and their uncertainties from both models. The proposed convolution model is able to correct for a pure specification bias by presenting less uncertain parameter estimates. It shows a good predictive performance for the mean prevalence values and for a positive number of infected people. Further research is needed to better understand the spatial extent and support of analysis to reliably explore the role of environmental variables.

Geographic Imputation of Missing Activity Space Data from Ecological Momentary Assessment (EMA) GPS Positions

This research presents a pilot study to develop and compare methods of geographic imputation for estimating the location of missing activity space data collected using geographic ecological momentary assessment (GEMA). As a demonstration, we use data from a previously published analysis of the effect of neighborhood disadvantage, captured at the U.S. Census Bureau tract level, on momentary psychological stress among a sample of 137 urban adolescents. We investigate the impact of listwise deletion on model results and test two geographic imputation techniques adapted for activity space data from hot deck and centroid imputation approaches. Our results indicate that listwise deletion can bias estimates of place effects on health, and that these impacts are mitigated by the use of geographic imputation, particularly regarding inflation of the standard errors. These geographic imputation techniques may be extended in future research by incorporating approaches from the non-spatial imputation literature as well as from conventional geographic imputation and spatial interpolation research that focus on non-activity space data.

Modelling local areas of exposure to Schistosoma japonicum in a limited survey data environment

Background

Spatial modelling studies of schistosomiasis (SCH) are now commonplace. Covariate values are commonly extracted at survey locations, where infection does not always take place, resulting in an unknown positional exposure mismatch. The present research aims to: (i) describe the nature of the positional exposure mismatch in modelling SCH helminth infections; (ii) delineate exposure areas to correct for such positional mismatch; and (iii) validate exposure areas using human positive cases.

Methods

To delineate exposure areas to Schistosoma japonicum, a spatial Bayesian network (sBN) was constructed. It uses data on exposure risk factors such as: potential sites for snails’ accessibility, geographical distribution of snail infection rate, and cost of the community to access nearby water bodies. Prior and conditional probabilities were obtained from the literature and inserted as weights based on their relative contribution to exposure; these probabilities were then used to calculate joint probabilities of exposure within the sBN.

Results

High values of probability of S. japonicum exposure correspond to polygons where snails could potentially be present, for instance in wet soils and areas with low slopes, but also where people can easily access water bodies. Low correlation (R² = 0.3) was found between the percentage of human cases and the delineated probabilities of exposure when validation buffers are generated over the human cases.

Conclusions

The utility of a probabilistic method for the identification of exposure areas for S. japonicum, with wider application for other water-borne infections, was demonstrated. From a public health perspective, the schistosomiasis exposure sBN developed in this study could be used to guide local schistosomiasis control teams to specific potential areas of exposure, and improve efficiency of mass drug administration campaigns in places where people are likely to be exposed to the infection.

Electronic supplementary material

The online version of this article (10.1186/s13071-018-3039-6) contains supplementary material, which is available to authorized users.

Spatial clustering of childhood leukaemia with the integration of the Paediatric Environmental History

BACKGROUND

Leukaemia remains the most common type of paediatric cancer and its aetiology remains unknown, but considered to be multifactorial. It is suggested that the initiation in utero by relevant exposures and/or inherited genetic variants and, other promotional postnatal exposures are probably required to develop leukaemia. This study aimed to map the incidence and analyse possible clusters in the geographical distribution of childhood acute leukaemia during the critical periods and to evaluate the factors that may be involved in the aetiology by conducting community and individual risk assessments.

MATERIALS AND METHODS

We analysed all incident cases of acute childhood leukaemia (<15 years) diagnosed in a Spanish region during the period 1998-2013. At diagnosis, the addresses during pregnancy, early childhood and diagnosis were collected and codified to analyse the spatial distribution of acute leukaemia. Scan statistical test methodology was used for the identification of high-incidence spatial clusters. Once identified, individual and community risk assessments were conducted using the Paediatric Environmental History.

RESULTS

A total of 158 cases of acute leukaemia were analysed. The crude rate for the period was 42.7 cases per million children. Among subtypes, acute lymphoblastic leukaemia had the highest incidence (31.9 per million children). A spatial cluster of acute lymphoblastic leukaemia was detected using the pregnancy address (p<0.05). The most common environmental risk factors related with the aetiology of acute lymphoblastic leukaemia, identified by the Paediatric Environmental History were: prenatal exposure to tobacco (75%) and alcohol (50%); residential and community exposure to pesticides (62.5%); prenatal or neonatal ionizing radiation (42.8%); and parental workplace exposure (37.5%) CONCLUSIONS: Our study suggests that environmental exposures in utero may be important in the development of childhood leukaemia. Due to the presence of high-incidence clusters using pregnancy address, it is necessary to introduce this address into the childhood cancer registers. The Paediatric Environmental History which includes pregnancy address and a careful and comprehensive evaluation of the environmental exposures will allow us to build the knowledge of the causes of childhood leukaemia.

Measurement Error and Environmental Epidemiology: a Policy Perspective

PURPOSE OF REVIEW

Measurement error threatens public health by producing bias in estimates of the population impact of environmental exposures. Quantitative methods to account for measurement bias can improve public health decision making.

RECENT FINDINGS

We summarize traditional and emerging methods to improve inference under a standard perspective, in which the investigator estimates an exposure-response function, and a policy perspective, in which the investigator directly estimates population impact of a proposed intervention. Under a policy perspective, the analyst must be sensitive to errors in measurement of factors that modify the effect of exposure on outcome, must consider whether policies operate on the true or measured exposures, and may increasingly need to account for potentially dependent measurement error of two or more exposures affected by the same policy or intervention. Incorporating approaches to account for measurement error into such a policy perspective will increase the impact of environmental epidemiology.

Accuracy of two geocoding methods for geographic information system-based exposure assessment in epidemiological studies

BACKGROUND

Environmental exposure assessment based on Geographic Information Systems (GIS) and study participants' residential proximity to environmental exposure sources relies on the positional accuracy of subjects' residences to avoid misclassification bias. Our study compared the positional accuracy of two automatic geocoding methods to a manual reference method.

METHODS

We geocoded 4,247 address records representing the residential history (1990-2008) of 1,685 women from the French national E3N cohort living in the Rhône-Alpes region. We compared two automatic geocoding methods, a free-online geocoding service (method A) and an in-house geocoder (method B), to a reference layer created by manually relocating addresses from method A (method R). For each automatic geocoding method, positional accuracy levels were compared according to the urban/rural status of addresses and time-periods (1990-2000, 2001-2008), using Chi Square tests. Kappa statistics were performed to assess agreement of positional accuracy of both methods A and B with the reference method, overall, by time-periods and by urban/rural status of addresses.

RESULTS

Respectively 81.4% and 84.4% of addresses were geocoded to the exact address (65.1% and 61.4%) or to the street segment (16.3% and 23.0%) with methods A and B. In the reference layer, geocoding accuracy was higher in urban areas compared to rural areas (74.4% vs. 10.5% addresses geocoded to the address or interpolated address level, p < 0.0001); no difference was observed according to the period of residence. Compared to the reference method, median positional errors were 0.0 m (IQR = 0.0-37.2 m) and 26.5 m (8.0-134.8 m), with positional errors <100 m for 82.5% and 71.3% of addresses, for method A and method B respectively. Positional agreement of method A and method B with method R was 'substantial' for both methods, with kappa coefficients of 0.60 and 0.61 for methods A and B, respectively.

CONCLUSION

Our study demonstrates the feasibility of geocoding residential addresses in epidemiological studies not initially recorded for environmental exposure assessment, for both recent addresses and residence locations more than 20 years ago. Accuracy of the two automatic geocoding methods was comparable. The in-house method (B) allowed a better control of the geocoding process and was less time consuming.