Application of geospatial analysis in health research: A systematic review of methodological aspects of studies on violence against children and young people.

BACKGROUND: Geographical variation exists in violence experienced by children and young people; however, there is limited research applying geospatial techniques to study this variation, and the methodological quality of this body of work is unclear. OBJECTIVE: This study aimed to review the application of geospatial analysis in research on violence against children (VAC) and evaluate how essential methodological aspects are reported. METHODS: Twelve databases were searched for studies on VAC using geospatial techniques. Two independent reviewers screened the papers for eligibility. Findings were narratively synthesised. RESULTS: Sixty studies were included. Six studies estimated the prevalence of VAC and 54 investigated the associations between VAC and covariates. Most studies were conducted in the US (68 %), and the broad definition of 'child maltreatment' (53 %) was the most common form of violence explored. Most studies (83 %) used administrative data, whereas 23 % used an ecological study design to estimate the associations between risk factors and official reports of VAC. Frequentist modelling approaches were used in 54 % of the studies, and 47 % investigated VAC at census tract level. Model fit metrics were reported in 69 % of studies. CONCLUSIONS: Current knowledge of the geographical distribution of VAC is severely limited because of the reliance on administrative data, which vastly underestimates the prevalence of VAC compared with self-reports and poor reporting of quality characteristics. There is a huge opportunity for applying geospatial methods in VAC research in diverse geographic contexts. Future research must adopt rigorous and standardised approaches to model fitting and validation and make better use of self-reported data.

The mortality rates and the space-time patterns of John Snow's cholera epidemic map.

BACKGROUND: Snow's work on the Broad Street map is widely known as a pioneering example of spatial epidemiology. It lacks, however, two significant attributes required in contemporary analyses of disease incidence: population at risk and the progression of the epidemic over time. Despite this has been repeatedly suggested in the literature, no systematic investigation of these two aspects was previously carried out. Using a series of historical documents, this study constructs own data to revisit Snow's study to examine the mortality rate at each street location and the space-time pattern of the cholera outbreak. METHODS: This study brings together records from a series of historical documents, and prepares own data on the estimated number of residents at each house location as well as the space-time data of the victims, and these are processed in GIS to facilitate the spatial-temporal analysis. Mortality rates and the space-time pattern in the victims' records are explored using Kernel Density Estimation and network-based Scan Statistic, a recently developed method that detects significant concentrations of records such as the date and place of victims with respect to their distance from others along the street network. The results are visualised in a map form using a GIS platform. RESULTS: Data on mortality rates and space-time distribution of the victims were collected from various sources and were successfully merged and digitised, thus allowing the production of new map outputs and new interpretation of the 1854 cholera outbreak in London, covering more cases than Snow's original report and also adding new insights into their space-time distribution. They confirmed that areas in the immediate vicinity of the Broad Street pump indeed suffered from excessively high mortality rates, which has been suspected for the past 160 years but remained unconfirmed. No distinctive pattern was found in the space-time distribution of victims' locations. CONCLUSIONS: The high mortality rates identified around the Broad Street pump are consistent with Snow's theory about cholera being transmitted through contaminated water. The absence of a clear space-time pattern also indicates the water-bourne, rather than the then popular belief of air bourne, nature of cholera. The GIS data constructed in this study has an academic value and would cater for further research on Snow's map.