Simon Read

Measuring the spatial accuracy of the spatial scan statistic

The spatial scan statistic is well established in spatial epidemiology. However, studies of its spatial accuracy are infrequent and vary in approach, often using multiple measures which complicate the objective ranking of different implementations of the statistic. We address this with three novel contributions. Firstly, a modular framework into which different definitions of spatial accuracy can be compared and hybridised. Secondly, we derive a new single measure, Ω, which takes account of all true and detected clusters, without the need for arbitrary weightings and irrespective of any chosen significance threshold. Thirdly, we demonstrate the new measure, alongside existing ones, in a study of the six output filter options provided by SaTScan™. The study suggests filtering overlapping detected clusters tends to reduce spatial accuracy, and visualising overlapping clusters may be better than filtering them out. Although we only address spatial accuracy, the framework and Ω may be extendible to spatio-temporal accuracy.

A graph-theory method for pattern identification in geographical epidemiology - a preliminary application to deprivation and mortality.

BackgroundGraph theoretical methods are extensively used in the field of computational chemistry to search datasets of compounds to see if they contain particular molecular sub-structures or patterns. We describe a preliminary application of a graph theoretical method, developed in computational chemistry, to geographical epidemiology in relation to testing a prior hypothesis. We tested the methodology on the hypothesis that if a socioeconomically deprived neighbourhood is situated in a wider deprived area, then that neighbourhood would experience greater adverse effects on mortality compared with a similarly deprived neighbourhood which is situated in a wider area with generally less deprivation.MethodsWe used the Trent Region Health Authority area for this study, which contained 10,665 census enumeration districts (CED). Graphs are mathematical representations of objects and their relationships and within the context of this study, nodes represented CEDs and edges were determined by whether or not CEDs were neighbours (shared a common boundary). The overall area in this study was represented by one large graph comprising all CEDs in the region, along with their adjacency information. We used mortality data from 1988–1998, CED level population estimates and the Townsend Material Deprivation Index as an indicator of neighbourhood level deprivation. We defined deprived CEDs as those in the top 20% most deprived in the Region. We then set out to classify these deprived CEDs into seven groups defined by increasing deprivation levels in the neighbouring CEDs. 506 (24.2%) of the deprived CEDs had five adjacent CEDs and we limited pattern development and searching to these CEDs. We developed seven query patterns and used the RASCAL (Rapid Similarity Calculator) program to carry out the search for each of the query patterns. This program used a maximum common subgraph isomorphism method which was modified to handle geographical data.ResultsOf the 506 deprived CEDs, 10 were not identified as belonging to any of the seven groups because they were adjacent to a CED with a missing deprivation category quintile, and none fell within query Group 1 (a deprived CED for which all five adjacent CEDs were affluent). Only four CEDs fell within Group 2, which was defined as having four affluent adjacent CEDs and one non-affluent adjacent CED. The numbers of CEDs in Groups 3–7 were 17, 214, 95, 81 and 85 respectively. Age and sex adjusted mortality rate ratios showed a non-significant trend towards increasing mortality risk across Groups (Chi-square = 3.26, df = 1, p = 0.07).ConclusionGraph theoretical methods developed in computational chemistry may be a useful addition to the current GIS based methods available for geographical epidemiology but further developmental work is required. An important requirement will be the development of methods for specifying multiple complex search patterns. Further work is also required to examine the utility of using distance, as opposed to adjacency, to describe edges in graphs, and to examine methods for pattern specification when the nodes have multiple attributes attached to them.

Bayesian hierarchical modelling of noisy spatial rates on a modestly large and discontinuous irregular lattice.

We present Bayesian hierarchical spatial model development motivated from a recent analysis of noisy small area response rate data, named the Booster data. The Booster data are postcode-level aggregates from a recent mail-out recruitment for a physical exercise intervention in deprived urban neighbourhoods in Sheffield, UK. Bayesian hierarchical Bernoulli-binomial spatial mixture zero-inflated Binomial models were developed for modelling overdispersion and for separation of systematic and random variations in the noisy and mostly low crude response rates. We present methods that enabled us to explore the underlying spatial rate variation, clustering of low or high response rate areas and neighbourhood characteristics that were associated with variations and patterns of invitation mail-outs, zero-response and response rates. Three spatial prior formulations, the intrinsic conditional autoregressive or (iCAR), the Besag-York-Mollié (BYM) and the modified BYM models, were explored for their performance on modelling sparse data on a modestly large and discontinuous irregular lattice. An in-depth Bayesian analysis of the Booster data is presented, with the resulting posterior estimation and inference implemented via Markov chain Monte Carlo simulation in WinBUGS. With increasing availability of spatial data referenced at fine spatial scales such as the postcode, the sparse-data situation and the Bayesian models and methods discussed herein should have considerable relevance to small area disease and health mapping and to spatial regression.

New developments in the spatial scan statistic

The quantity and variety of spatial data have increased over recent years, and the variety and sophistication of tools for analysing this type of data have also increased. One such tool is the spatial scan statistic, which is freely available (www.satscan.org) and has been the subject of much scholarly research since its introduction in 1995 owing to its numerous applications in epidemiology, criminology and other fields. This paper provides readers with a non-technical introduction to the spatial scan statistic, together with an overview of associated research, which focuses particularly on work conducted at the University of Sheffield’s Information School, in collaboration with the School of Health and Related Research. This work falls into three main areas. First, we provide an examination of the probability of obtaining false alerts when using the statistic, and ways in which this can be managed. Second, we describe the development of a definitive way of measuring the spatial accuracy of the statistic. Third, and potentially the most important in terms of impact, we discuss a means of substantially increasing the detection capability of the statistic by placing a realistic constraint on the strength of any cluster that is likely to be present in the data. The paper also provides a discussion of potential future research directions.

Key design issues with visualising images using Google earth

Using map visualisation tools and earth browsers to display images in a spatial context is integral to many photo-sharing sites and commercial image archives, yet little academic research has been conducted into the utility and functionality of such systems. In developing a prototype system to explore the use of Google Earth in the visualisation of news photos, we have elicited key design issues based on user evaluations of Panoramio and two custom-built spatio-temporal image browsing prototypes. We discuss the implications of these design issues, with particular emphasis on visualising news photos.

A study on the use of Gumbel approximation with the Bernoulli spatial scan statistic

The Bernoulli version of the spatial scan statistic is a well established method of detecting localised spatial clusters in binary labelled point data, a typical application being the epidemiological case-control study. A recent study suggests the inferential accuracy of several versions of the spatial scan statistic (principally the Poisson version) can be improved, at little computational cost, by using the Gumbel distribution, a method now available in SaTScan(TM) (www.satscan.org). We study in detail the effect of this technique when applied to the Bernoulli version and demonstrate that it is highly effective, albeit with some increase in false alarm rates at certain significance thresholds. We explain how this increase is due to the discrete nature of the Bernoulli spatial scan statistic and demonstrate that it can affect even small p-values. Despite this, we argue that the Gumbel method is actually preferable for very small p-values. Furthermore, we extend previous research by running benchmark trials on 12 000 synthetic datasets, thus demonstrating that the overall detection capability of the Bernoulli version (i.e. ratio of power to false alarm rate) is not noticeably affected by the use of the Gumbel method. We also provide an example application of the Gumbel method using data on hospital admissions for chronic obstructive pulmonary disease.

Associations between neighbourhood environmental factors and the uptake and effectiveness of a brief intervention to increase physical activity: findings from deprived urban communities in an English city.

Background Evidence suggests behavioural interventions may exacerbate health inequalities, potentially due to differences in uptake or effectiveness. We used a physical activity intervention targeting deprived communities to identify neighbourhood-level factors that might explain differences in programme impact. Methods Individuals aged 40-65 were sent a postal invitation offering a brief intervention to increase physical activity. We used postcodes linkage to determine whether neighbourhood indicators of deprivation, housing, crime and proximity to green spaces and leisure facilities predicted uptake of the initial invitation or an increase in physical activity level in those receiving the brief intervention. Results A total of 4134 (6.8%) individuals responded to the initial invitation and of those receiving the intervention and contactable after 3 months, 486 (51.6%) reported an increase in physical activity. Area deprivation scores linked to postcodes predicted intervention uptake, but not intervention effectiveness. Neighbourhood indicators did not predict either uptake or intervention effectiveness. Conclusions The main barrier to using brief intervention invitations to increase physical activity in deprived, middle-aged populations was the low uptake of an intervention requiring significant time and motivation from participants. Once individuals have taken up the intervention offer, neighbourhood characteristics did not appear to be significant barriers to successful lifestyle change.

A power-enhanced algorithm for spatial anomaly detection in binary labelled point data using the spatial scan statistic

This paper presents a novel modification to an existing algorithm for spatial anomaly detection in binary labeled point data sets, using the Bernoulli version of the Spatial Scan Statistic. We identify a potential ambiguity in p-values produced by Monte Carlo testing, which (by the selection of the most conservative p-value) can lead to sub-optimal power. When such ambiguity occurs, the modification uses a very inexpensive secondary test to suggest a less conservative p-value. Using benchmark tests, we show that this appears to restore power to the expected level, whilst having similarly retest variance to the original. The modification also appears to produce a small but significant improvement in overall detection performance when multiple anomalies are present.