A simulation study of three methods for detecting disease clusters

Enhancing Public Health Surveillance: Outbreak Detection Algorithms Deployed for Syndromic Surveillance During Arbaeenia Mass Gatherings in Iraq

BACKGROUND

Large gatherings often involve extended and intimate contact among individuals, creating environments conducive to the spread of infectious diseases. Despite this, there is limited research utilizing outbreak detection algorithms to analyze real syndrome data from such events. This study sought to address this gap by examining the implementation and efficacy of outbreak detection algorithms for syndromic surveillance during mass gatherings in Iraq.

METHODS

For the study, 10 data collectors conducted field data collection over 10 days from August 25, 2023, to September 3, 2023. Data were gathered from 10 healthcare clinics situated along Ya Hussein Road, a major route from Najaf to Karbala in Iraq. Various outbreak detection algorithms, such as moving average, cumulative sum, and exponentially weighted moving average, were applied to analyze the reported syndromes.

RESULTS

During the 10 days from August 25, 2023, to September 3, 2023, 12202 pilgrims visited 10 health clinics along a route in Iraq. Most pilgrims were between 20 and 59 years old (77.4%, n=9444), with more than half being foreigners (58.1%, n=7092). Among the pilgrims, 40.5% (n=4938) exhibited syndromes, with influenza-like illness (ILI) being the most common (48.8%, n=2411). Other prevalent syndromes included food poisoning (21.2%, n=1048), heatstroke (17.7%, n=875), febrile rash (9.0%, n=446), and gastroenteritis (3.2%, n=158). The cumulative sum (CUSUM) algorithm was more effective than exponentially weighted moving average (EWMA) and moving average (MA) algorithms for detecting small shifts.

CONCLUSION

Effective public health surveillance systems are crucial during mass gatherings to swiftly identify and address emerging health risks. Utilizing advanced algorithms and real-time data analysis can empower authorities to improve their readiness and response capacity, thereby ensuring the protection of public health during these gatherings.

Evaluation and comparison of spatial cluster detection methods for improved decision making of disease surveillance: a case study of national dengue surveillance in Thailand

BACKGROUND

Dengue is a mosquito-borne disease that causes over 300 million infections worldwide each year with no specific treatment available. Effective surveillance systems are needed for outbreak detection and resource allocation. Spatial cluster detection methods are commonly used, but no general guidance exists on the most appropriate method for dengue surveillance. Therefore, a comprehensive study is needed to assess different methods and provide guidance for dengue surveillance programs.

METHODS

To evaluate the effectiveness of different cluster detection methods for dengue surveillance, we selected and assessed commonly used methods: Getis Ord [Formula: see text], Local Moran, SaTScan, and Bayesian modeling. We conducted a simulation study to compare their performance in detecting clusters, and applied all methods to a case study of dengue surveillance in Thailand in 2019 to further evaluate their practical utility.

RESULTS

In the simulation study, Getis Ord [Formula: see text] and Local Moran had similar performance, with most misdetections occurring at cluster boundaries and isolated hotspots. SaTScan showed better precision but was less effective at detecting inner outliers, although it performed well on large outbreaks. Bayesian convolution modeling had the highest overall precision in the simulation study. In the dengue case study in Thailand, Getis Ord [Formula: see text] and Local Moran missed most disease clusters, while SaTScan was mostly able to detect a large cluster. Bayesian disease mapping seemed to be the most effective, with adaptive detection of irregularly shaped disease anomalies.

CONCLUSIONS

Bayesian modeling showed to be the most effective method, demonstrating the best accuracy in adaptively identifying irregularly shaped disease anomalies. In contrast, SaTScan excelled in detecting large outbreaks and regular forms. This study provides empirical evidence for the selection of appropriate tools for dengue surveillance in Thailand, with potential applicability to other disease control programs in similar settings.

Space-time cluster detection techniques for infectious diseases: A systematic review

BACKGROUND

Public health organizations have increasingly harnessed geospatial technologies for disease surveillance, health services allocation, and targeting place-based health promotion initiatives.

METHODS

We conducted a systematic review around the theme of space-time clustering detection techniques for infectious diseases using PubMed, Web of Science, and Scopus. Two reviewers independently determined inclusion and exclusion.

RESULTS

Of 2,887 articles identified, 354 studies met inclusion criteria, the majority of which were application papers. Studies of airborne diseases were dominant, followed by vector-borne diseases. Most research used aggregated data instead of point data, and a significant proportion of articles used a repetition of a spatial clustering method, instead of using a "true" space-time detection approach, potentially leading to the detection of false positives. Noticeably, most articles did not make their data available, limiting replicability.

CONCLUSION

This review underlines recent trends in the application of space-time clustering methods to the field of infectious disease, with a rapid increase during the COVID-19 pandemic.

Impact of healthcare capacity disparities on the COVID-19 vaccination coverage in the United States: A cross-sectional study

Background

The impact of the COVID-19 vaccination campaign in the US has been hampered by a substantial geographical heterogeneity of the vaccination coverage. Several studies have proposed vaccination hesitancy as a key driver of the vaccination uptake disparities. However, the impact of other important structural determinants such as local disparities in healthcare capacity is virtually unknown.

Methods

In this cross-sectional study, we conducted causal inference and geospatial analyses to assess the impact of healthcare capacity on the vaccination coverage disparity in the US. We evaluated the causal relationship between the healthcare system capacity of 2417 US counties and their COVID-19 vaccination rate. We also conducted geospatial analyses using spatial scan statistics to identify areas with low vaccination rates.

Findings

We found a causal effect of the constraints in the healthcare capacity of a county and its low-vaccination uptake. Counties with higher constraints in their healthcare capacity were more probable to have COVID-19 vaccination rates ≤50, with 35% higher constraints in low-vaccinated areas (vaccination rates ≤ 50) compared to high-vaccinated areas (vaccination rates > 50). We also found that COVID-19 vaccination in the US exhibits a distinct spatial structure with defined "vaccination coldspots".

Interpretation

We found that the healthcare capacity of a county is an important determinant of low vaccine uptake. Our study highlights that even in high-income nations, internal disparities in healthcare capacity play an important role in the health outcomes of the nation. Therefore, strengthening the funding and infrastructure of the healthcare system, particularly in rural underserved areas, should be intensified to help vulnerable communities.

Funding

None.

Comparative Study of Approaches for Detecting Crime Hotspots with Considering Concentration and Shape Characteristics

Hotspot detection is an important exploratory technique to identify areas with high concentrations of crime and help deploy crime-reduction resources. Although a variety of methods have been developed to detect crime hotspots, few studies have systematically evaluated the performance of various methods, especially in terms of the ability to detect complex-shaped crime hotspots. Therefore, in this study, a comparative study of hotspot detection approaches while simultaneously considering the concentration and shape characteristics was conducted. Firstly, we established a framework for quantitatively evaluating the performance of hotspot detection for cases with or without the "ground truth". Secondly, accounting for the concentration and shape characteristics of the hotspot, we additionally defined two evaluation indicators, which can be used as a supplement to existing evaluation indicators. Finally, four classical hotspot-detection methods were quantitatively compared on the synthetic and real crime data. Results show that the proposed evaluation framework and indicators can describe the size, concentration and shape characteristics of the detected hotspots, thus supporting the quantitative comparison of different methods. From the selected methods, the AMOEBA (A Multidirectional Optimal Ecotope-Based Algorithm) method was more accurate in describing the concentration and shape characteristics and was powerful in discovering complex hotspots.

Sporadic SARS-CoV-2 cases at the neighbourhood level in Toronto, Ontario, 2020: a spatial analysis of the early pandemic period

BACKGROUND

As the largest city in Canada, Toronto has played an important role in the dynamics of SARS-CoV-2 transmission in Ontario, and the burden of disease across Toronto neighbourhoods has shown considerable heterogeneity. The purpose of this study was to investigate the spatial variation of sporadic SARS-CoV-2 cases in Toronto neighbourhoods by detecting clusters of increased risk and investigating effects of neighbourhood-level risk factors on rates.

METHODS

Data on sporadic SARS-CoV-2 cases, at the neighbourhood level, for Jan. 25 to Nov. 26, 2020, were obtained from the City of Toronto COVID-19 dashboard. We used a flexibly shaped spatial scan to detect clusters of increased risk of sporadic COVID-19. We then used a generalized linear geostatistical model to investigate whether average household size, population density, dependency ratio and prevalence of low-income households were associated with sporadic SARS-CoV-2 rates.

RESULTS

We identified 3 clusters of elevated risk of SARS-CoV-2 infection, with standardized morbidity ratios ranging from 1.59 to 2.43. The generalized linear geostatistical model found that average household size (relative risk [RR] 2.17, 95% confidence interval [CI] 1.80-2.61) and percentage of low-income households (RR 1.03, 95% CI 1.02-1.04) were significant predictors of sporadic SARS-CoV-2 cases at the neighbourhood level.

INTERPRETATION

During the study period, 3 clusters of increased risk of sporadic SARS-CoV-2 infection were identified, and average household size and percentage of low-income households were found to be associated with sporadic SARS-CoV-2 rates at the neighbourhood level. The findings of this study can be used to target resources and create policy to address inequities that are shown through heterogeneity of SARS-CoV-2 cases at the neighbourhood level in Toronto, Ontario.

Spatial clustering and contextual factors associated with hospitalisation and deaths due to COVID-19 in Sweden: a geospatial nationwide ecological study

INTRODUCTION

In Sweden, thousands of hospitalisations and deaths due to COVID-19 were reported since the pandemic started. Considering the uneven spatial distribution of those severe outcomes at the municipality level, the objective of this study was, first, to identify high-risk areas for COVID-19 hospitalisations and deaths, and second, to determine the associated contextual factors with the uneven spatial distribution of both study outcomes in Sweden.

METHODS

The existences of spatial autocorrelation of the standardised incidence (hospitalisations) ratio and standardised mortality ratio were investigated using Global Moran's I test. Furthermore, we applied the retrospective Poisson spatial scan statistics to identify high-risk spatial clusters. The association between the contextual demographic and socioeconomic factors and the number of hospitalisations and deaths was estimated using a quasi-Poisson generalised additive regression model.

RESULTS

Ten high-risk spatial clusters of hospitalisations and six high-risk clusters of mortality were identified in Sweden from February 2020 to October 2020. The hospitalisations and deaths were associated with three contextual variables in a multivariate model: population density (inhabitants/km²) and the proportion of immigrants (%) showed a positive association with both outcomes, while the proportion of the population aged 65+ years (%) showed a negative association.

CONCLUSIONS

Our study identified high-risk spatial clusters for hospitalisations and deaths due to COVID-19 and the association of population density, the proportion of immigrants and the proportion of people aged 65+ years with those severe outcomes. Results indicate where public health measures must be reinforced to improve sustained and future disease control and optimise the distribution of resources.

"The devil is in the detail": geographical inequalities of femicides in Ecuador

Background

Femicide is a very important public health problem in Ecuador. Since regional and country-level femicide rates can obscure significant variations at the sub-national level, it is important to provide information at the lowest relevant level of disaggregation to be able to develop targeted preventive policies. The aim of this study was to assess the spatial distribution of the femicide rate and to examine its spatial clustering at the canton level in Ecuador in the period 2018–2019.

Methods

Data on cases were collected by a national network of non-governmental organizations. Two age-disaggregated analyses were done, one for the 15 to 24 years-olds and the other for the female population of 15 and older. Age-specific population data were obtained from the National Institute of Statistics for the study period. Standardized mortality ratios for mapping the mortality were calculated using hierarchical Bayesian models and spatial scan statistics were applied to identify local clusters. Thematic maps of age-specific femicide rates were also constructed.

Results

During the two-year period, 61 and 183 women were killed in the age ranges 15–24 and 15 years and older, respectively. The annual rate of femicides in Ecuador was 1.0 and 0.8 per 100,000 in the female population aged 15–24 and 15+, respectively, with substantial variations among cantons. The spatial analysis contributed to visualize high risk cantons, which were mainly located in a small area in the central part of the country (for those 15+) but especially in the Amazon region, for both of the studied age groups.

Conclusions

This study has shown the usefulness of applying spatial analysis to the problem of femicides in Ecuador. The study has revealed important variations among cantons but also a spatial clustering, mainly in the Amazon region of the country. The results should help policymakers to focus on current prevention programmes for violence against women into these high-risk areas. Continuous monitoring of femicides at low-level geographical areas is highly recommended.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12939-021-01454-x.

Statistical methods for spatial cluster detection in childhood cancer incidence: A simulation study

BACKGROUND AND OBJECTIVE

The potential existence of spatial clusters in childhood cancer incidence is a debated topic. Identification of such clusters may help to better understand etiology and develop preventive strategies. We evaluated widely used statistical approaches to cluster detection in this context.

METHODS

Incidence of newly diagnosed childhood cancer (140/1,000,000 children under 15 years) and nephroblastoma (7/1,000,000) was simulated. Clusters of defined size (1-50) were randomly assembled on the district level in Germany. Each cluster was simulated with different relative risk levels (1-100). For each combination 2000 iterations were done. Simulated data was then analyzed by three local clustering tests: Besag-Newell method, spatial scan statistic and Bayesian Besag-York-Mollié with Integrated Nested Laplace Approximation approach. The operating characteristics (sensitivity, specificity, predictive values, power and correct classification) of all three methods were systematically described.

RESULTS

Performance varied considerably within and between methods, depending on the simulated setting. Sensitivity of all methods was positively associated with increasing size, incidence and RR of the high-risk area. Besag-York-Mollié showed highest specificity for minimally increased RR in most scenarios. The performance of all methods was lower in the nephroblastoma scenario compared with the scenario including all cancer cases.

CONCLUSION

This study illustrates the challenge to make reliable inferences on the existence of spatial clusters based on single statistical approaches in childhood cancer. Application of multiple methods, ideally with known operating characteristics, and a critical discussion of the joint evidence seems recommendable when aiming to identify high-risk clusters.

Spatial and spatio-temporal analysis of malaria cases in Zimbabwe

Background

Although effective treatment for malaria is now available, approximately half of the global population remain at risk of the disease particularly in developing countries. To design effective malaria control strategies there is need to understand the pattern of malaria heterogeneity in an area. Therefore, the main objective of this study was to explore the spatial and spatio-temporal pattern of malaria cases in Zimbabwe based on malaria data aggregated at district level from 2011 to 2016.

Methods

Geographical information system (GIS) and spatial scan statistic were applied on passive malaria data collected from health facilities and aggregated at district level to detect existence of spatial clusters. The global Moran’s I test was used to infer the presence of spatial autocorrelation while the purely spatial retrospective analyses were performed to detect the spatial clusters of malaria cases with high rates based on the discrete Poisson model. Furthermore, space-time clusters with high rates were detected through the retrospective space-time analysis based on the discrete Poisson model.

Results

Results showed that there is significant positive spatial autocorrelation in malaria cases in the study area. In addition, malaria exhibits spatial heterogeneity as evidenced by the existence of statistically significant (P < 0.05) spatial and space-time clusters of malaria in specific geographic regions. The detected primary clusters persisted in the eastern region of the study area over the six year study period while the temporal pattern of malaria reflected the seasonality of the disease where clusters were detected within particular months of the year.

Conclusions

Geographic regions characterised by clusters of high rates were identified as malaria high risk areas. The results of this study could be useful in prioritizing resource allocation in high-risk areas for malaria control and elimination particularly in resource limited settings such as Zimbabwe. The results of this study are also useful to guide further investigation into the possible determinants of persistence of high clusters of malaria cases in particular geographic regions which is useful in reducing malaria burden in such areas.

Spatiotemporal Tools for Emerging and Endemic Disease Hotspots in Small Areas: An Analysis of Dengue and Chikungunya in Barbados, 2013-2016

Dengue fever and other febrile mosquito-borne diseases place considerable health and economic burdens on small island nations in the Caribbean. Here, we used two methods of cluster detection to find potential hotspots of transmission of dengue and chikungunya in Barbados, and to assess the impact of input surveillance data and methodology on observed patterns of risk. Using Moran's I and spatial scan statistics, we analyzed the geospatial and temporal distribution of disease cases and rates across Barbados for dengue fever in 2013-2016, and a chikungunya outbreak in 2014. During years with high numbers of dengue cases, hotspots for cases were found with Moran's I in the south and central regions in 2013 and 2016, respectively. Using smoothed disease rates, clustering was detected in all years for dengue. Hotspots suggesting higher rates were not detected via spatial scan statistics, but coldspots suggesting lower than expected rates of disease activity were found in southwestern Barbados during high case years of dengue. No significant spatiotemporal structure was found in cases during the chikungunya outbreak. Spatial analysis of surveillance data is useful in identifying outbreak hotspots, potentially complementing existing early warning systems. We caution that these methods should be used in a manner appropriate to available data and reflecting explicit public health goals-managing for overall case numbers or targeting anomalous rates for further investigation.

Bayesian Spatiotemporal Modeling of Routinely Collected Data to Assess the Effect of Health Programs in Malaria Incidence During Pregnancy in Burkina Faso

Control of malaria in pregnancy (MiP) remains a major challenge in Burkina Faso. Surveillance of the burden due to MiP based on routinely collected data at a fine-scale level, followed by an appropriate analysis and interpretation, may be crucial for evaluating and improving the effectiveness of existing control measures. We described the spatio-temporal dynamics of MiP at the community-level and assessed health program effects, mainly community-based health promotion, results-based financing, and intermittent-preventive-treatment with sulphadoxine-pyrimethamine (IPTp-SP). Community-aggregated monthly MiP cases were downloaded from Health Management Information System and combined with covariates from other sources. The MiP spatio-temporal pattern was decomposed into three components: overall spatial and temporal trends and space-time interaction. Bayesian hierarchical spatio-temporal Poisson models were used to fit the MiP incidence rate and assess health program effects. The overall annual incidence increased between 2015 and 2017. The findings reveal spatio-temporal heterogenicity throughout the year, which peaked during rainy season. From the model without covariates, 96 communities located mainly in the Cascades, South-West, Center-West, Center-East, and Eastern regions, exhibited significant relative-risk levels. The combined effect (significant reducing effect) of RBF, health promotion and IPTp-SP strategies was greatest in 17.7% (17/96) of high burden malaria communities. Despite intensification of control efforts, MiP remains high at the community-scale. The provided risk maps are useful tools for highlighting areas where interventions should be optimized, particularly in high-risk communities.

Waterfowl occurrence and residence time as indicators of H5 and H7 avian influenza in North American Poultry

Avian influenza (AI) affects wild aquatic birds and poses hazards to human health, food security, and wildlife conservation globally. Accordingly, there is a recognized need for new methods and tools to help quantify the dynamic interaction between wild bird hosts and commercial poultry. Using satellite-marked waterfowl, we applied Bayesian joint hierarchical modeling to concurrently model species distributions, residency times, migration timing, and disease occurrence probability under an integrated animal movement and disease distribution modeling framework. Our results indicate that migratory waterfowl are positively related to AI occurrence over North America such that as waterfowl occurrence probability or residence time increase at a given location, so too does the chance of a commercial poultry AI outbreak. Analyses also suggest that AI occurrence probability is greatest during our observed waterfowl northward migration, and less during the southward migration. Methodologically, we found that when modeling disparate facets of disease systems at the wildlife-agriculture interface, it is essential that multiscale spatial patterns be addressed to avoid mistakenly inferring a disease process or disease-environment relationship from a pattern evaluated at the improper spatial scale. The study offers important insights into migratory waterfowl ecology and AI disease dynamics that aid in better preparing for future outbreaks.

Malaria Hotspots: Is There Epidemiological Evidence for Fine-Scale Spatial Targeting of Interventions?

As data at progressively granular spatial scales become available, the temptation is to target interventions to areas with higher malaria transmission - so-called hotspots - with the aim of reducing transmission in the wider community. This paper reviews literature to determine if hotspots are an intrinsic feature of malaria epidemiology and whether current evidence supports hotspot-targeted interventions. Hotspots are a consistent feature of malaria transmission at all endemicities. The smallest spatial unit capable of supporting transmission is the household, where peri-domestic transmission occurs. Whilst the value of focusing interventions to high-burden areas is evident, there is currently limited evidence that local-scale hotspots fuel transmission. As boundaries are often uncertain, there is no conclusive evidence that hotspot-targeted interventions accelerate malaria elimination.

Analyzing spatial and space-time clustering of facility-based deliveries in Bangladesh

Background

A spatial and temporal study of the distribution of facility-based deliveries can identify areas of low and high facility usage and help devise more targeted interventions to improve delivery outcomes. Developing countries like Bangladesh face considerable challenges in reducing the maternal mortality ratio to the targets set by the Sustainable Development Goals. Recent studies have already identified that the progress of reducing maternal mortality has stalled. Giving birth in a health facility is one way to reduce maternal mortality.

Methods

Facility delivery data from a demographic surveillance site was analyzed at both village and Bari (comprising several households with same paternal origins) level to understand spatial and temporal heterogeneity. Global spatial autocorrelation was detected using Moran’s I index while local spatial clusters were detected using the local Getis G_i* statistics. In addition, space-time scanning using a discrete Poisson approach facilitated the identification of space-time clusters. The likelihood of delivering at a facility when located inside a cluster was calculated using log-likelihood ratios.

Results

The three cluster detection approaches detected significant spatial and temporal heterogeneity in the distribution of facility deliveries in the study area. The hot and cold spots indicated contiguous and relocation type diffusion and increased in number over the years. Space-time scanning revealed that when a parturient woman is located in a Bari inside the cluster, the likelihood of delivering at a health facility increases by twenty-seven times.

Conclusions

Spatiotemporal studies to understand delivery patterns are quite rare. However, in resource constraint countries like Bangladesh, detecting hot and cold spot areas can aid in the detection of diffusion centers, which can be targeted to expand regions with high facility deliveries. Places and periods with reduced health facility usages can be identified using various cluster detection techniques, to assess the barriers and facilitators in promoting health facility deliveries.

Electronic supplementary material

The online version of this article (10.1186/s41182-019-0170-9) contains supplementary material, which is available to authorized users.

Utility of Spatial Point-Pattern Analysis Using Residential and Workplace Geospatial Information to Localize Potential Outbreak Sources

Identifying the source of an outbreak facilitates its control. Spatial methods are not optimally used in outbreak investigation, due to a mix of the complexities involved (e.g., methods requiring additional parameter selection), imperfect performance, and lack of confidence in existing options. We simulated 30 mock outbreaks and compared 5 simple methods that do not require parameter selection but could select between mock cases’ residential and workplace addresses to localize the source. Each category of site had a unique spatial distribution; residential and workplace address were visually and statistically clustered around the residential neighborhood and city center sites respectively, suggesting that the value of workplace addresses is tied to the location where an outbreak might originate. A modification to centrographic statistics that we propose—the center of minimum geometric distance with address selection—was able to localize the mock outbreak source to within a 500 m radius in almost all instances when using workplace in combination with residential addresses. In the sensitivity analysis, when given sufficient workplace data, the method performed well in various scenarios with only 10 cases. It was also successful when applied to past outbreaks, except for a multisite outbreak from a common food supplier.

Spatial overlaps in the distribution of HIV/AIDS and malaria in Zimbabwe

BACKGROUND

In most developing economies particularly in Africa, more people are likely to die of HIV/AIDS and malaria compared to other diseases. HIV/AIDS tends to be superimposed on the long standing malaria burden particularly in sub-Saharan Africa. The detection and understanding of spatial overlaps in disease occurrence is important for integrated and targeted disease control. Integrated disease control can enhance efficiency and cost-effectiveness through the development of drugs targeting multiple infections in the same geographic space.

METHODS

Using Zimbabwe as a case study, this study tests the hypothesis that malaria clusters coincide with HIV/AIDS clusters in space. Case data for the two diseases were obtained from the Ministry of Health and Child Care in Zimbabwe at district level via the District Health Information System (DHIS). Kulldorff's spatial scan statistic was used to test for spatial overlaps in clusters of high cases of HIV/AIDS and malaria at district level. The spatial scan test was used to identify areas with higher cases of HIV/AIDS and malaria than would be expected under spatial randomness.

RESULTS

Results of this study indicate that primary clusters of HIV/AIDS and malaria were not spatially coincident in Zimbabwe. While no spatial overlaps were detected between primary clusters of the two diseases, spatial overlaps were detected among statistically significant secondary clusters of HIV/AIDS and malaria. Spatial overlaps between HIV/AIDS and malaria occurred in five districts in the northern and eastern regions of Zimbabwe. In addition, findings of this study indicate that HIV/AIDS is more widespread in Zimbabwe compared to malaria.

CONCLUSIONS

The results of this study may therefore be used as a basis for spatially-targeted control of HIV/AIDS and malaria particularly in high disease burden areas. This is important as previous interventions have targeted the two diseases separately. Thus, targeted control could assist in resource allocation through prioritising areas in greatest need hence maximising the impact of disease control.

Methods used in the spatial analysis of tuberculosis epidemiology: a systematic review

BACKGROUND

Tuberculosis (TB) transmission often occurs within a household or community, leading to heterogeneous spatial patterns. However, apparent spatial clustering of TB could reflect ongoing transmission or co-location of risk factors and can vary considerably depending on the type of data available, the analysis methods employed and the dynamics of the underlying population. Thus, we aimed to review methodological approaches used in the spatial analysis of TB burden.

METHODS

We conducted a systematic literature search of spatial studies of TB published in English using Medline, Embase, PsycInfo, Scopus and Web of Science databases with no date restriction from inception to 15 February 2017. The protocol for this systematic review was prospectively registered with PROSPERO ( CRD42016036655 ).

RESULTS

We identified 168 eligible studies with spatial methods used to describe the spatial distribution (n = 154), spatial clusters (n = 73), predictors of spatial patterns (n = 64), the role of congregate settings (n = 3) and the household (n = 2) on TB transmission. Molecular techniques combined with geospatial methods were used by 25 studies to compare the role of transmission to reactivation as a driver of TB spatial distribution, finding that geospatial hotspots are not necessarily areas of recent transmission. Almost all studies used notification data for spatial analysis (161 of 168), although none accounted for undetected cases. The most common data visualisation technique was notification rate mapping, and the use of smoothing techniques was uncommon. Spatial clusters were identified using a range of methods, with the most commonly employed being Kulldorff's spatial scan statistic followed by local Moran's I and Getis and Ord's local Gi(d) tests. In the 11 papers that compared two such methods using a single dataset, the clustering patterns identified were often inconsistent. Classical regression models that did not account for spatial dependence were commonly used to predict spatial TB risk. In all included studies, TB showed a heterogeneous spatial pattern at each geographic resolution level examined.

CONCLUSIONS

A range of spatial analysis methodologies has been employed in divergent contexts, with all studies demonstrating significant heterogeneity in spatial TB distribution. Future studies are needed to define the optimal method for each context and should account for unreported cases when using notification data where possible. Future studies combining genotypic and geospatial techniques with epidemiologically linked cases have the potential to provide further insights and improve TB control.

Spatial, temporal, and spatiotemporal analysis of mumps in Guangxi Province, China, 2005-2016

Background

The resurgence of mumps around the world occurs frequently in recent years. As the country with the largest number of cases in the world, the status of mumps epidemics in China is not yet clear. This study, taking the relatively serious epidemic province of Guangxi as the example, aimed to examine the spatiotemporal pattern and epidemiological characteristics of mumps, and provide a scientific basis for the effective control of this disease and formulation of related health policies.

Methods

Geographic information system (GIS)-based spatiotemporal analyses, including spatial autocorrelation analysis, Kulldorff’s purely spatial and space-time scan statistics, were applied to detect the location and extent of mumps high-risk areas. Spatial empirical Bayesian (SEB) was performed to smoothen the rate for eliminating the instability of small-area data.

Results

A total of 208,470 cases were reported during 2005 and 2016 in Guangxi. Despite the fluctuations in 2006 and 2011, the overall mumps epidemic continued to decline. Bimodal seasonal distribution (mainly from April to July) were found and students aged 5–9 years were high-incidence groups. Though results of the global spatial autocorrelation based on the annual incidence largely varied, the spatial distribution of the average annual incidence of mumps was nonrandom with the significant Moran’s I. Spatial cluster analysis detected high-value clusters, mainly located in the western, northern and central parts of Guangxi. Spatiotemporal scan statistics identified almost the same high-risk areas, and the aggregation time was mainly concentrated in 2009–2012.

Conclusion

The incidence of mumps in Guangxi exhibited spatial heterogeneity in 2005–2016. Several spatial and spatiotemporal clusters were identified in this study, which might assist the local government to develop targeted health strategies, allocate health resources reasonably and increase the efficiency of disease prevention.

Electronic supplementary material

The online version of this article (10.1186/s12879-018-3240-4) contains supplementary material, which is available to authorized users.

Childhood and Adolescence Cancers in the Palermo Province (Southern Italy): Ten Years (2003⁻2012) of Epidemiological Surveillance

Italy has one of the highest paediatric cancer incidence rates in Europe. We compared cancer incidence and survival rates in children (0⁻14 years) and adolescents (15⁻19 years) residing in Palermo Province (PP) with statistics derived from Italian and European surveillance systems. We included all incident cancer cases, malignant tumours and non-malignant neoplasm of central nervous system (benign and uncertain whether malignant or benign), detected in children and adolescents by the Palermo Province Cancer Registry (PPCR) between 2003 and 2012. A jointpoint regression model was applied. Annual Average Percentage Changes were calculated. The Besag⁻York-Mollie model was used to detect any cluster. The 5-year survival analysis was computed using Kaplan-Meier and actuarial methods. We identified 555 paediatric cancer incident cases (90% “malignant tumours”). No difference in incidence rates was highlighted between PPCR and Italy 26 registries and between PPCR and Southern Europe. No jointpoint or significant trend was identified and no cluster was detected. The 5-year overall survival didn’t differ between PP and the Italian AIRTUM pool. A borderline higher statistically significant survival was observed in age-group 1⁻4 when comparing PPCR to EUROCARE-5. The epidemiological surveillance documented in the PP was a paediatric cancer burden in line with Italy and southern Europe. The study supports the supplementary role of general population-based cancer registries to provide paediatric cancer surveillance of local communities.

Exploring spatial and temporal patterns of visceral leishmaniasis in endemic areas of Bangladesh

Background

Visceral leishmaniasis is a considerable public health burden on the Indian subcontinent. The disease is highly endemic in the north-central part of Bangladesh, affecting the poorest and most marginalized communities. Despite the fact that visceral leishmaniasis (VL) results in mortality, severe morbidity, and socioeconomic stress in the region, the spatiotemporal dynamics of the disease have largely remained unexplored, especially in Bangladesh.

Methods

Monthly VL cases between 2010 and 2014, obtained from subdistrict hospitals, were studied in this work. Both global and local spatial autocorrelation techniques were used to identify spatial heterogeneity of the disease. In addition, a spatial scan test was used to identify statistically significant space-time clusters in endemic locations of Bangladesh.

Results

Global and local spatial autocorrelation indicated that the distribution of VL was spatially autocorrelated, exhibiting both contiguous and relocation-type of diffusion; however, the former was the main type of VL spread in the study area. The spatial scan test revealed that the disease had ten times higher incidence rate within the clusters than in non-cluster zones. Both tests identified clusters in the same geographic areas, despite the differences in their algorithm and cluster detection approach.

Conclusion

The cluster maps, generated in this work, can be used by public health officials to prioritize areas for intervention. Additionally, initiatives to control VL can be handled more efficiently when areas of high risk of the disease are known. Because global environmental change is expected to shift the current distribution of vectors to new locations, the results of this work can help to identify potentially exposed populations so that adaptation strategies can be formulated.

Spatio-temporal analysis of malaria incidence in the Peruvian Amazon Region between 2002 and 2013

Malaria remains a major public health problem in the Peruvian Amazon where the persistence of high-risk transmission areas (hotspots) challenges the current malaria control strategies. This study aimed at identifying significant space-time clusters of malaria incidence in Loreto region 2002-2013 and to determine significant changes across years in relation to the control measures applied. Poisson regression and purely temporal, spatial, and space-time analyses were conducted. Three significantly different periods in terms of annual incidence rates (AIR) were identified, overlapping respectively with the pre-, during, and post- implementation control activities supported by PAMAFRO project. The most likely space-time clusters of malaria incidence for P. vivax and P. falciparum corresponded to the pre- and first two years of the PAMAFRO project and were situated in the northern districts of Loreto, while secondary clusters were identified in eastern and southern districts with the latest onset and the shortest duration of PAMAFRO interventions. Malaria in Loreto was highly heterogeneous at geographical level and over time. Importantly, the excellent achievements obtained during 5 years of intensified control efforts totally vanished in only 2 to 3 years after the end of the program, calling for sustained political and financial commitment for the success of malaria elimination as ultimate goal.

Spatiotemporal epidemic characteristics and risk factor analysis of malaria in Yunnan Province, China

BACKGROUND

Malaria remains an important public health concern in China and is particularly serious in Yunnan, a China's provincial region of high malaria burden with an incidence of 1.79/10⁵ in 2012. This study aims to examine the epidemiologic profile and spatiotemporal aspects of epidemics of malaria, and to examine risk factors which may influence malaria epidemics in Yunnan Province.

METHODS

The data of malaria cases in 2012 in 125 counties of Yunnan Province was used in this research. The epidemical characteristics of cases were revealed, and time and space clusters of malaria were detected by applying scan statistics method. In addition, we applied the geographically weighted regression (GWR) model in identifying underlying risk factors.

RESULTS

There was a total of 821 cases of malaria, and male patients accounted for 83.9% (689) of the total cases. The incidence in the group aged 20-30 years was the highest, at 3.00/10⁵. The majority (84.1%) of malaria cases occurred in farmers and migrant workers, according to occupation statistics. On a space-time basis, epidemics of malaria of varying severity occurred in the summer and autumn months, and the high risk regions were mainly distributed in the southwest counties. Annual average temperature, annual cumulative rainfall, rice yield per square kilometer and proportion of rural employees mainly showed a positive association with the malaria incidence rate, according to the GWR model.

CONCLUSIONS

Malaria continues to be one of serious public health issues in Yunnan Province, especially in border counties in southwestern Yunnan. Temperature, precipitation, rice cultivation and proportion of rural employees were positively associated with malaria incidence. Individuals, and disease prevention and control departments, should implement more stringent preventative strategies in locations with hot and humid environmental conditions to control malaria.

Spatial and space-time distribution of Plasmodium vivax and Plasmodium falciparum malaria in China, 2005-2014

BACKGROUND

Despite the declining burden of malaria in China, the disease remains a significant public health problem with periodic outbreaks and spatial variation across the country. A better understanding of the spatial and temporal characteristics of malaria is essential for consolidating the disease control and elimination programme. This study aims to understand the spatial and spatiotemporal distribution of Plasmodium vivax and Plasmodium falciparum malaria in China during 2005-2009.

METHODS

Global Moran's I statistics was used to detect a spatial distribution of local P. falciparum and P. vivax malaria at the county level. Spatial and space-time scan statistics were applied to detect spatial and spatiotemporal clusters, respectively.

RESULTS

Both P. vivax and P. falciparum malaria showed spatial autocorrelation. The most likely spatial cluster of P. vivax was detected in northern Anhui province between 2005 and 2009, and western Yunnan province between 2010 and 2014. For P. falciparum, the clusters included several counties of western Yunnan province from 2005 to 2011, Guangxi from 2012 to 2013, and Anhui in 2014. The most likely space-time clusters of P. vivax malaria and P. falciparum malaria were detected in northern Anhui province and western Yunnan province, respectively, during 2005-2009.

CONCLUSION

The spatial and space-time cluster analysis identified high-risk areas and periods for both P. vivax and P. falciparum malaria. Both malaria types showed significant spatial and spatiotemporal variations. Contrary to P. vivax, the high-risk areas for P. falciparum malaria shifted from the west to the east of China. Further studies are required to examine the spatial changes in risk of malaria transmission and identify the underlying causes of elevated risk in the high-risk areas.

The changing malaria landscape in Aseer region, Kingdom of Saudi Arabia: 2000-2015

Background

In 2004, a revised action plan was developed, supported by the World Health Organization, to eliminate malaria from Saudi Arabia by preventing re-introduction of malaria into regions since declared malaria free, eliminating foci of transmission in the Mecca and Medina areas and a concerted effort of foci surveillance and control, to eliminate malaria from the regions of Jazan and Aseer. This paper provides the context, activities, progress, and possible contributions toward malaria elimination in the Aseer region since 2000, with a more detailed analysis of the spatial location of locally acquired case incidence since 2012.

Methods

This is a descriptive study of all available Ministry of Health surveillance data and process reports since 2000, with higher spatial resolution analysis of data between 2012 and 2015.

Results

In 2000, there were 511 cases of Plasmodium falciparum locally acquired infection. The following 4 years witnessed a dramatic decline in cases to only 18 locally acquired infections reported in 2005. A resurgence in local infections was reported in 2006 (93) and 2007 (165), thereafter (2008–2014) local cases continued to decline to fewer than 40 per year across the region. However, in 2015, a small rise was noted (51). All locally acquired infections were P. falciparum. There has been a constant flow of imported infections into Aseer since 2000, mostly among immigrant labour from Pakistan, India, Sudan, and Yemen. Imported infections have included both Plasmodium vivax and P. falciparum. The spatial extent of malaria appears to be changing, but there remain two intractable areas Sarat Abeda and Dhran Aljanub, where risks per reporting centre have changed little since 2001, remaining above 0.5 per 10,000 population. Only seven villages contributed 55% of all locally acquired infection since 2012.

Discussion

Aseer has reached a state of very low incidence of locally acquired infections, despite a constant source of imported infections from outside the country. How many of the local infections are F2 generations from imported infections or how many are a result of residual active transmission between asymptomatic carriers of infections transmitted by pockets of existing Anopheles arabiensis populations remains unknown. A more detailed investigation of the spatial and temporal patterns of infected hosts, parasites and vectors would help define whether this region has managed to effectively prevent local transmission of new infections.

Geographical patterns of malaria transmission based on serological markers for falciparum and vivax malaria in Ratanakiri, Cambodia

Background

Malaria transmission is highly heterogeneous, especially in low endemic countries, such as Cambodia. This results in geographical clusters of residual transmission in the dry, low transmission season, which can fuel the transmission to wider areas or populations during the wet season. A better understanding of spatial clustering of malaria can lead to a more efficient, targeted strategy to reduce malaria transmission. This study aims to evaluate the potential of the use of serological markers to define spatial patterns in malaria exposure.

Methods

Blood samples collected in a community-based randomized trial performed in 98 high endemic communities in Ratanakiri province, north-eastern Cambodia, were screened with a multiplex serological assay for five serological markers (three Plasmodium falciparum and two Plasmodium vivax). The antibody half-lives range from approximately six months until more than two years. Geographical heterogeneity in malaria transmission was examined using a spatial scan statistic on serology, PCR prevalence and malaria incidence rate data. Furthermore, to identify behavioural patterns or intrinsic factors associated with malaria exposure (antibody levels), risk factor analyses were performed by using multivariable random effect logistic regression models. The serological outcomes were then compared to PCR prevalence and malaria incidence data.

Results

A total of 6502 samples from two surveys were screened in an area where the average parasite prevalence estimated by PCR among the selected villages is 3.4 %. High-risk malaria pockets were observed adjacent to the ‘Tonle San River’ and neighbouring Vietnam for all three sets of data (serology, PCR prevalence and malaria incidence rates). The main risk factors for all P. falciparum antigens and P. vivax MSP1.19 are age, ethnicity and staying overnight at the plot hut.

Conclusion

It is possible to identify similar malaria pockets of higher malaria transmission together with the potential risk factors by using serology instead of PCR prevalence or malaria incidence data. In north-eastern Cambodia, the serological markers show that malaria transmission occurs mainly in adults staying overnight in plot huts in the field. Pf.GLURP.R2 showed a shrinking pocket of malaria transmission over time, and Pf.MSP1.19, CSP, PvAMA1 were also informative for current infection to a lesser extent. Therefore, serology could contribute in future research. However, further in-depth research in selecting the best combination of antigens is required.

Electronic supplementary material

The online version of this article (doi:10.1186/s12936-016-1558-1) contains supplementary material, which is available to authorized users.

Impact of socioeconomic inequalities on geographic disparities in cancer incidence: comparison of methods for spatial disease mapping

BACKGROUND

The reliability of spatial statistics is often put into question because real spatial variations may not be found, especially in heterogeneous areas. Our objective was to compare empirically different cluster detection methods. We assessed their ability to find spatial clusters of cancer cases and evaluated the impact of the socioeconomic status (e.g., the Townsend index) on cancer incidence.

METHODS

Moran's I, the empirical Bayes index (EBI), and Potthoff-Whittinghill test were used to investigate the general clustering. The local cluster detection methods were: i) the spatial oblique decision tree (SpODT); ii) the spatial scan statistic of Kulldorff (SaTScan); and, iii) the hierarchical Bayesian spatial modeling (HBSM) in a univariate and multivariate setting. These methods were used with and without introducing the Townsend index of socioeconomic deprivation known to be related to the distribution of cancer incidence. Incidence data stemmed from the Cancer Registry of Isère and were limited to prostate, lung, colon-rectum, and bladder cancers diagnosed between 1999 and 2007 in men only.

RESULTS

The study found a spatial heterogeneity (p < 0.01) and an autocorrelation for prostate (EBI = 0.02; p = 0.001), lung (EBI = 0.01; p = 0.019) and bladder (EBI = 0.007; p = 0.05) cancers. After introduction of the Townsend index, SaTScan failed in finding cancers clusters. This introduction changed the results obtained with the other methods. SpODT identified five spatial classes (p < 0.05): four in the Western and one in the Northern parts of the study area (standardized incidence ratios: 1.68, 1.39, 1.14, 1.12, and 1.16, respectively). In the univariate setting, the Bayesian smoothing method found the same clusters as the two other methods (RR >1.2). The multivariate HBSM found a spatial correlation between lung and bladder cancers (r = 0.6).

CONCLUSIONS

In spatial analysis of cancer incidence, SpODT and HBSM may be used not only for cluster detection but also for searching for confounding or etiological factors in small areas. Moreover, the multivariate HBSM offers a flexible and meaningful modeling of spatial variations; it shows plausible previously unknown associations between various cancers.

Geographical Patterns of HIV Sero-Discordancy in High HIV Prevalence Countries in Sub-Saharan Africa

INTRODUCTION

Variation in the proportion of individuals living in a stable HIV sero-discordant partnership (SDP), and the potential drivers of such variability across sub Saharan Africa (SSA), are still not well-understood. This study aimed to examine the spatial clustering of HIV sero-discordancy, and the impact of local variation in HIV prevalence on patterns of sero-discordancy in high HIV prevalence countries in SSA.

METHODS

We described the spatial patterns of sero-discordancy among stable couples by analyzing Demographic and Health Survey data from Cameroon, Kenya, Lesotho, Tanzania, Malawi, Zambia, and Zimbabwe. We identified spatial clusters of SDPs in each country through a Kulldorff spatial scan statistics analysis. After a geographical cluster was identified, epidemiologic measures of sero-discordancy were calculated and analyzed.

RESULTS

Spatial clusters with significantly high numbers of SDPs were identified and characterized in Kenya, Malawi, and Tanzania, and they largely overlapped with the clusters with high HIV prevalence. There was a positive correlation between HIV prevalence and the proportion of SDPs among all stable couples across within and outside clusters. Conversely, there was a negative, but weak and not significant, correlation between HIV prevalence and the proportion of SDPs among all stable couples with at least one HIV-infected individual in the partnership.

DISCUSSION

There does not appear to be distinct spatial patterns for HIV sero-discordancy that are independent of HIV prevalence patterns. The variation of the sero-discordancy measures with HIV prevalence across clusters and outside clusters demonstrated similar patterns to those observed at the national level. The spatial variable does not appear to be a fundamental nor independent determinant of the observed patterns of sero-discordancy in high HIV prevalence countries in SSA.

Satellite Hyperspectral Imagery to Support Tick-Borne Infectious Diseases Surveillance

This study proposed the use of satellite hyperspectral imagery to support tick-borne infectious diseases surveillance based on monitoring the variation in amplifier hosts food sources. To verify this strategy, we used the data of the human rickettsiosis occurrences in southeastern Brazil, region in which the emergence of this disease is associated with the rising capybara population. Spatio-temporal analysis based on Monte Carlo simulations was used to identify risk areas of human rickettsiosis and hyperspectral moderate-resolution imagery was used to identify the increment and expansion of sugarcane crops, main food source of capybaras. In general, a pixel abundance associated with increment of sugarcane crops was detected in risk areas of human rickettsiosis. Thus, the hypothesis that there is a spatio-temporal relationship between the occurrence of human rickettsiosis and the sugarcane crops increment was verified. Therefore, due to the difficulty of monitoring locally the distribution of infectious agents, vectors and animal host's, satellite hyperspectral imagery can be used as a complementary tool for the surveillance of tick-borne infectious diseases and potentially of other vector-borne diseases.

Typhoid fever acquired in the United States, 1999-2010: epidemiology, microbiology, and use of a space-time scan statistic for outbreak detection

Although rare, typhoid fever cases acquired in the United States continue to be reported. Detection and investigation of outbreaks in these domestically acquired cases offer opportunities to identify chronic carriers. We searched surveillance and laboratory databases for domestically acquired typhoid fever cases, used a space-time scan statistic to identify clusters, and classified clusters as outbreaks or non-outbreaks. From 1999 to 2010, domestically acquired cases accounted for 18% of 3373 reported typhoid fever cases; their isolates were less often multidrug-resistant (2% vs. 15%) compared to isolates from travel-associated cases. We identified 28 outbreaks and two possible outbreaks within 45 space-time clusters of ⩾2 domestically acquired cases, including three outbreaks involving ⩾2 molecular subtypes. The approach detected seven of the ten outbreaks published in the literature or reported to CDC. Although this approach did not definitively identify any previously unrecognized outbreaks, it showed the potential to detect outbreaks of typhoid fever that may escape detection by routine analysis of surveillance data. Sixteen outbreaks had been linked to a carrier. Every case of typhoid fever acquired in a non-endemic country warrants thorough investigation. Space-time scan statistics, together with shoe-leather epidemiology and molecular subtyping, may improve outbreak detection.

Cluster Detection Tests in Spatial Epidemiology: A Global Indicator for Performance Assessment

In cluster detection of disease, the use of local cluster detection tests (CDTs) is current. These methods aim both at locating likely clusters and testing for their statistical significance. New or improved CDTs are regularly proposed to epidemiologists and must be subjected to performance assessment. Because location accuracy has to be considered, performance assessment goes beyond the raw estimation of type I or II errors. As no consensus exists for performance evaluations, heterogeneous methods are used, and therefore studies are rarely comparable. A global indicator of performance, which assesses both spatial accuracy and usual power, would facilitate the exploration of CDTs behaviour and help between-studies comparisons. The Tanimoto coefficient (TC) is a well-known measure of similarity that can assess location accuracy but only for one detected cluster. In a simulation study, performance is measured for many tests. From the TC, we here propose two statistics, the averaged TC and the cumulated TC, as indicators able to provide a global overview of CDTs performance for both usual power and location accuracy. We evidence the properties of these two indicators and the superiority of the cumulated TC to assess performance. We tested these indicators to conduct a systematic spatial assessment displayed through performance maps.

Evaluation of the performance of smoothing functions in generalized additive models for spatial variation in disease

Generalized additive models (GAMs) with bivariate smoothing functions have been applied to estimate spatial variation in risk for many types of cancers. Only a handful of studies have evaluated the performance of smoothing functions applied in GAMs with regard to different geographical areas of elevated risk and different risk levels. This study evaluates the ability of different smoothing functions to detect overall spatial variation of risk and elevated risk in diverse geographical areas at various risk levels using a simulation study. We created five scenarios with different true risk area shapes (circle, triangle, linear) in a square study region. We applied four different smoothing functions in the GAMs, including two types of thin plate regression splines (TPRS) and two versions of locally weighted scatterplot smoothing (loess). We tested the null hypothesis of constant risk and detected areas of elevated risk using analysis of deviance with permutation methods and assessed the performance of the smoothing methods based on the spatial detection rate, sensitivity, accuracy, precision, power, and false-positive rate. The results showed that all methods had a higher sensitivity and a consistently moderate-to-high accuracy rate when the true disease risk was higher. The models generally performed better in detecting elevated risk areas than detecting overall spatial variation. One of the loess methods had the highest precision in detecting overall spatial variation across scenarios and outperformed the other methods in detecting a linear elevated risk area. The TPRS methods outperformed loess in detecting elevated risk in two circular areas.

Spatial, temporal, and spatiotemporal analysis of malaria in Hubei Province, China from 2004-2011

Background

Malaria remains a public health concern in Hubei Province despite the significant decrease in malaria incidence over the past decades. Furthermore, history reveals that malaria transmission is unstable and prone to local outbreaks in Hubei Province. Thus, understanding spatial, temporal, and spatiotemporal distribution of malaria is needed for the effective control and elimination of this disease in Hubei Province.

Methods

Annual malaria incidence at the county level was calculated using the malaria cases reported from 2004 to 2011 in Hubei Province. Geographical information system (GIS) and spatial scan statistic method were used to identify spatial clusters of malaria cases at the county level. Pure retrospective temporal analysis scanning was performed to detect the temporal clusters of malaria cases with high rates using the discrete Poisson model. The space-time cluster was detected with high rates through the retrospective space-time analysis scanning using the discrete Poisson model.

Results

The overall malaria incidence decreased to a low level from 2004 to 2011. The purely spatial cluster of malaria cases from 2004 to 2011 showed that the disease was not randomly distributed in the study area. A total of 11 high-risk counties were determined through Local Moran’s I analysis from 2004 to 2011. The method of spatial scan statistics identified different 11 significant spatial clusters between 2004 and 2011. The space-time clustering analysis determined that the most likely cluster included 13 counties, and the time frame was from April 2004 to November 2007.

Conclusions

The GIS application and scan statistical technique can provide means to detect spatial, temporal, and spatiotemporal distribution of malaria, as well as to identify malaria high-risk areas. This study could be helpful in prioritizing resource assignment in high-risk areas for future malaria control and elimination.

Geographical variation analysis of all-cause hospital readmission cases in Winnipeg, Canada

BACKGROUND

Hospital readmission is costly and potentially avoidable. The concept of virtual wards as a new model of care is intended to reduce hospital readmissions by providing short-term transitional care to high-risk and complex patients in the community. In order to provide information regarding the development of virtual wards in the Winnipeg Health Region, Canada, this study used spatial statistics to identify geographic variations of hospital readmissions in 25 neighborhood clusters.

METHODS

The data were obtained from the Population Health Research Data Repository housed at the Manitoba Centre for Health Policy. We used a Bayesian Disease Mapping approach which applied Markov chain Monte Carlo (MCMC) for cluster detection.

RESULTS

Between 2005/06 and 2008/09, 123,842 patients were hospitalized in all Winnipeg hospitals. Of these, 41,551 (33%) were readmitted to hospital in the year following discharge. Most of these readmitted patients (89.4%) had 1-2 readmissions, while 11.6% of readmitted patients had more than 2 readmissions after initial discharge. The smoothed age- and sex- adjusted relative risk rates of hospital readmission in 25 Winnipeg neighborhood clusters ranged between 0.73 and 1.27. We found that there were spatial cluster variations of hospital readmission across the Winnipeg Health Region. Seven neighborhood clusters are more likely to be significant potential clusters for hospital readmissions (p < .05), while six neighborhood clusters are less likely to be significant potential clusters.

CONCLUSIONS

This study provides the foundation and implementation guide for the Winnipeg Regional Health Authority virtual ward program. The findings will also help to improve long-term condition management in community settings and will help program planners to assure the efficient use of healthcare resources.

Spatio-temporal analysis of the relationship between climate and hand, foot, and mouth disease in Shandong province, China, 2008-2012

Background

Hand, foot, and mouth disease (HFMD) is the most common communicable disease in China. Shandong Province is one of the most seriously affected areas. The distribution of HFMD had spatial heterogeneity and seasonal characteristic in this setting. The aim of this study was to explore the associations between climate and HFMD by a Bayesian approach from spatio-temporal interactions perspective.

Methods

The HFMD data of Shandong Province during 2008–2012 were derived from the China National Disease Surveillance Reporting and Management System. And six climatic indicators were obtained from the Meteorological Bureau of Shandong Province. The global spatial autocorrelation statistic (Moran’s I) was used to detect the spatial autocorrelation of HFMD cases in each year. The optimal one among four Bayesian models was further adopted to estimate the relative risk of the occurrence of HFMD via Markov chain Monte Carlo.

Results

The annual average incidence rate of HFMD was 104.40 per 100,000 in Shandong Province. Positive spatial autocorrelation appeared at county level (Moran’s I ≥0.30, P < 0.001). The best fitting Spatio-temporal interactive model showed that annual average temperature, annual average pressure, annual average relative humidity, annual average wind speed and annual sunshine hours were significantly positive related to the occurrence of HFMD. The estimated relative risk of 36, 87, 91, 79, 65 out of 140 counties for 2008–2012 respectively were significantly more than 1.

Conclusions

There were obvious spatio-temporal heterogeneity of HFMD in Shandong Province, and the climatic indicators were associated with the epidemic of HFMD. Bayesian approach should be recommended to capture the spatial-temporal pattern of HFMD.

Comparing methods of measuring geographic patterns in temporal trends: an application to county-level heart disease mortality in the United States, 1973 to 2010

PURPOSE

To demonstrate the implications of choosing analytical methods for quantifying spatiotemporal trends, we compare the assumptions, implementation, and outcomes of popular methods using county-level heart disease mortality in the United States between 1973 and 2010.

METHODS

We applied four regression-based approaches (joinpoint regression, both aspatial and spatial generalized linear mixed models, and Bayesian space-time model) and compared resulting inferences for geographic patterns of local estimates of annual percent change and associated uncertainty.

RESULTS

The average local percent change in heart disease mortality from each method was -4.5%, with the Bayesian model having the smallest range of values. The associated uncertainty in percent change differed markedly across the methods, with the Bayesian space-time model producing the narrowest range of variance (0.0-0.8). The geographic pattern of percent change was consistent across methods with smaller declines in the South Central United States and larger declines in the Northeast and Midwest. However, the geographic patterns of uncertainty differed markedly between methods.

CONCLUSIONS

The similarity of results, including geographic patterns, for magnitude of percent change across these methods validates the underlying spatial pattern of declines in heart disease mortality. However, marked differences in degree of uncertainty indicate that Bayesian modeling offers substantially more precise estimates.

Distance to Anopheles sundaicus larval habitats dominant among risk factors for parasitemia in meso-endemic Southwest Sumba, Indonesia

BACKGROUND

The decline in intensity of malaria transmission in many areas now emphasizes greater importance of understanding the epidemiology of low to moderate transmission settings. Marked heterogeneity in infection risk within these populations creates opportunities to understand transmission and guide resource allocation to greater impact.

METHODS

In this study, we examined spatial patterns of malaria transmission in a hypo- to meso-endemic area of eastern Indonesia using malaria prevalence data collected from a cross-sectional socio-demographic and parasitological survey conducted from August to November 2010. An entomological survey performed in parallel, identified, mapped, and monitored local anopheline larval habitats.

RESULTS

A single spatial cluster of higher malaria prevalence was detected during the study period (relative risk=2.13; log likelihood ratio=20.7; P<0.001). In hierarchical multivariate regression models, risk of parasitemia was inversely correlated with distance to five Anopheles sundaicus known larval habitats [odds ratio (OR)=0.21; 95% confidence interval (CI)=0.14-0.32; P<0.001], which were located in a geographically restricted band adjacent to the coastline. Increasing distance from these sites predicted increased hemoglobin level across age strata after adjusting for confounders (OR=1.6; 95% CI=1.30-1.98; P<0.001).

CONCLUSION

Significant clustering of malaria parasitemia in close proximity to very specific and relatively few An. sundaicus larval habitats has direct implications for local control strategy, policy, and practice. These findings suggest that larval source management could achieve profound if not complete impact in this region.

Spatial clustering and risk factors of malaria infections in Ratanakiri Province, Cambodia

Background

Malaria incidence worldwide has steadily declined over the past decades. Consequently, increasingly more countries will proceed from control to elimination. The malaria distribution in low incidence settings appears patchy, and local transmission hotspots are a continuous source of infection. In this study, species-specific clusters and associated risk factors were identified based on malaria prevalence data collected in the north-east of Cambodia. In addition, Plasmodium falciparum genetic diversity, population structure and gene flows were studied.

Method

In 2012, blood samples from 5793 randomly selected individuals living in 117 villages were collected from Ratanakiri province, Cambodia. Malariometric data of each participant were simultaneously accumulated using a standard questionnaire. A two-step PCR allowed for species-specific detection of malaria parasites, and SNP-genotyping of P. falciparum was performed. SaTScan was used to determine species-specific areas of elevated risk to infection, and univariate and multivariate risk analyses were carried out.

Result

PCR diagnosis found 368 positive individuals (6.4%) for malaria parasites, of which 22% contained mixed species infections. The occurrence of these co-infections was more frequent than expected. Specific areas with elevated risk of infection were detected for all Plasmodium species. The clusters for Falciparum, Vivax and Ovale malaria appeared in the north of the province along the main river, while the cluster for Malariae malaria was situated elsewhere. The relative risk to be a malaria parasite carrier within clusters along the river was twice that outside the area. The main risk factor associated with three out of four malaria species was overnight stay in the plot hut, a human behaviour associated with indigenous farming. Haplotypes did not show clear geographical population structure, but pairwise Fst value comparison indicated higher parasite flow along the river.

Discussion

Spatial aggregation of malaria parasite carriers, and the identification of malaria species-specific risk factors provide key insights in malaria epidemiology in low transmission settings, which can guide targeted supplementary interventions. Consequently, future malaria programmes in the province should implement additional specific policies targeting households staying overnight at their farms outside the village, in addition to migrants and forest workers.

Electronic supplementary material

The online version of this article (doi:10.1186/1475-2875-13-387) contains supplementary material, which is available to authorized users.

Spatial heterogeneity of type I error for local cluster detection tests

Background

Just as power, type I error of cluster detection tests (CDTs) should be spatially assessed. Indeed, CDTs’ type I error and power have both a spatial component as CDTs both detect and locate clusters. In the case of type I error, the spatial distribution of wrongly detected clusters (WDCs) can be particularly affected by edge effect. This simulation study aims to describe the spatial distribution of WDCs and to confirm and quantify the presence of edge effect.

Methods

A simulation of 40 000 datasets has been performed under the null hypothesis of risk homogeneity. The simulation design used realistic parameters from survey data on birth defects, and in particular, two baseline risks. The simulated datasets were analyzed using the Kulldorff’s spatial scan as a commonly used test whose behavior is otherwise well known. To describe the spatial distribution of type I error, we defined the participation rate for each spatial unit of the region. We used this indicator in a new statistical test proposed to confirm, as well as quantify, the edge effect.

Results

The predefined type I error of 5% was respected for both baseline risks. Results showed strong edge effect in participation rates, with a descending gradient from center to edge, and WDCs more often centrally situated.

Conclusions

In routine analysis of real data, clusters on the edge of the region should be carefully considered as they rarely occur when there is no cluster. Further work is needed to combine results from power studies with this work in order to optimize CDTs performance.

Bowel disorders and its spatial trend in Manitoba, Canada

BACKGROUND

Bowel disorders have destructive impacts on the patients social and mental aspects of life and can cause emotional distress. The risk of developing bowel incontinence also increases with age. The rate of incidence of inflammatory bowel disease in Manitoba, Canada, has been unusually raised. Therefore, it is important to identify trends in the incidence of bowel disorders that may suggest further epidemiological studies to identify risk factors and identify any changes in important factors.

METHODS

An important part of spatial epidemiology is cluster detection as it has the potential to identify possible risk factors associated with disease, which in turn may lead to further investigations into the nature of diseases. To test for potential disease clusters many methods have been proposed. The focused detection methods including the circular spatial scan statistic (CSS), flexible spatial scan statistic (FSS), and Bayesian disease mapping (BYM) are among the most popular disease detection procedures. A frequentist approach based on maximum likelihood estimation (MLE) has been recently used to identify potential focused clusters as well. The aforementioned approaches are studied by analyzing a dataset of bowel disorders in the province of Manitoba, Canada, from 2001 to 2010.

RESULTS

The CSS method identified less regions than the FSS method in the south part of the province as potential clusters. The same regions were identified by the BYM and MLE methods as being potential clusters of bowel disorders with a slightly different order of significance. Most of these regions were also detected by the CSS or FSS methods.

CONCLUSIONS

Overall, we recommend using the methods BYM and MLE for cluster detection with the similar population and structure of regions as in Manitoba. The potential clusters of bowel disorders are generally located in the southern part of the province including the eastern part of the city of Winnipeg. These results may represent real increases in bowel disorders or they may be an indication of other covariates that were not adjusted for in the model used here. Further investigation is needed to examine these findings, and also to explore the cause of these increases.

Hot spot or not: a comparison of spatial statistical methods to predict prospective malaria infections

Background

Within affected communities, Plasmodium falciparum infections may be skewed in distribution such that single or small clusters of households consistently harbour a disproportionate number of infected individuals throughout the year. Identifying these hotspots of malaria transmission would permit targeting of interventions and a more rapid reduction in malaria burden across the whole community. This study set out to compare different statistical methods of hotspot detection (SaTScan, kernel smoothing, weighted local prevalence) using different indicators (PCR positivity, AMA-1 and MSP-1 antibodies) for prediction of infection the following year.

Methods

Two full surveys of four villages in Mwanza, Tanzania were completed over consecutive years, 2010-2011. In both surveys, infection was assessed using nested polymerase chain reaction (nPCR). In addition in 2010, serologic markers (AMA-1 and MSP-1₁₉ antibodies) of exposure were assessed. Baseline clustering of infection and serological markers were assessed using three geospatial methods: spatial scan statistics, kernel analysis and weighted local prevalence analysis. Methods were compared in their ability to predict infection in the second year of the study using random effects logistic regression models, and comparisons of the area under the receiver operating curve (AUC) for each model. Sensitivity analysis was conducted to explore the effect of varying radius size for the kernel and weighted local prevalence methods and maximum population size for the spatial scan statistic.

Results

Guided by AUC values, the kernel method and spatial scan statistics appeared to be more predictive of infection in the following year. Hotspots of PCR-detected infection and seropositivity to AMA-1 were predictive of subsequent infection. For the kernel method, a 1 km window was optimal. Similarly, allowing hotspots to contain up to 50% of the population was a better predictor of infection in the second year using spatial scan statistics than smaller maximum population sizes.

Conclusions

Clusters of AMA-1 seroprevalence or parasite prevalence that are predictive of infection a year later can be identified using geospatial models. Kernel smoothing using a 1 km window and spatial scan statistics both provided accurate prediction of future infection.

Detecting cancer clusters in a regional population with local cluster tests and Bayesian smoothing methods: a simulation study

Background

There is a rising public and political demand for prospective cancer cluster monitoring. But there is little empirical evidence on the performance of established cluster detection tests under conditions of small and heterogeneous sample sizes and varying spatial scales, such as are the case for most existing population-based cancer registries. Therefore this simulation study aims to evaluate different cluster detection methods, implemented in the open soure environment R, in their ability to identify clusters of lung cancer using real-life data from an epidemiological cancer registry in Germany.

Methods

Risk surfaces were constructed with two different spatial cluster types, representing a relative risk of RR = 2.0 or of RR = 4.0, in relation to the overall background incidence of lung cancer, separately for men and women. Lung cancer cases were sampled from this risk surface as geocodes using an inhomogeneous Poisson process. The realisations of the cancer cases were analysed within small spatial (census tracts, N = 1983) and within aggregated large spatial scales (communities, N = 78). Subsequently, they were submitted to the cluster detection methods. The test accuracy for cluster location was determined in terms of detection rates (DR), false-positive (FP) rates and positive predictive values. The Bayesian smoothing models were evaluated using ROC curves.

Results

With moderate risk increase (RR = 2.0), local cluster tests showed better DR (for both spatial aggregation scales > 0.90) and lower FP rates (both < 0.05) than the Bayesian smoothing methods. When the cluster RR was raised four-fold, the local cluster tests showed better DR with lower FPs only for the small spatial scale. At a large spatial scale, the Bayesian smoothing methods, especially those implementing a spatial neighbourhood, showed a substantially lower FP rate than the cluster tests. However, the risk increases at this scale were mostly diluted by data aggregation.

Conclusion

High resolution spatial scales seem more appropriate as data base for cancer cluster testing and monitoring than the commonly used aggregated scales. We suggest the development of a two-stage approach that combines methods with high detection rates as a first-line screening with methods of higher predictive ability at the second stage.

How to interpret and choose a Bayesian spatial model and a Poisson regression model in the context of describing small area cancer risks variations

BACKGROUND

The statistical Bayesian approach is widely used in disease mapping and Poisson regression. Results differ depending on the underlying hypothesis. Our objective is to give a comprehensive presentation of the tools that can be used to interpret results and choose between the different hypotheses. Data from the Isere cancer registry (France) illustrate this presentation.

METHOD

We consider, first, Bayesian models for disease mapping. Classic heterogeneity (Potthoff-Whithinghill statistic) and spatial autocorrelation tests (Moran statistic) of the SIRs, the DIC criteria of the different Bayesian models and finally the comparison of the empirical variance of the unstructured and structured heterogeneity components of the BYM model are considered. The last two criteria are considered for Bayesian Poisson regression including a covariate. Mapping the components of the BYM model with a covariate is also considered.

RESULTS

Four cancer sites (prostate, lung, colon-rectum and bladder) in men diagnosed during the 1998-2007 period are used to illustrate our presentation. We show that the different criteria used to interpret and to choose a model give coherent results.

CONCLUSION

A relevant interpretation of results is a necessary step in choosing the best-adapted Bayesian model. This choice is easy to make with criteria such as the DIC. The comparison of the empirical variance of the unstructured and structured heterogeneity components of the BYM model is also informative.

Further shrinking the malaria map: how can geospatial science help to achieve malaria elimination?

Malaria is one of the biggest contributors to deaths caused by infectious disease. More than 30 countries have planned or started programmes to target malaria elimination, often with explicit support from international donors. The spatial distribution of malaria, at all levels of endemicity, is heterogeneous. Moreover, populations living in low-endemic settings where elimination efforts might be targeted are often spatially heterogeneous. Geospatial methods, therefore, can help design, target, monitor, and assess malaria elimination programmes. Rapid advances in technology and analytical methods have allowed the spatial prediction of malaria risk and the development of spatial decision support systems, which can enhance elimination programmes by enabling accurate and timely resource allocation. However, no framework exists for assessment of geospatial instruments. Research is needed to identify measurable indicators of elimination progress and to quantify the effect of geospatial methods in achievement of elimination outcomes.

Performance map of a cluster detection test using extended power

Background

Conventional power studies possess limited ability to assess the performance of cluster detection tests. In particular, they cannot evaluate the accuracy of the cluster location, which is essential in such assessments. Furthermore, they usually estimate power for one or a few particular alternative hypotheses and thus cannot assess performance over an entire region. Takahashi and Tango developed the concept of extended power that indicates both the rate of null hypothesis rejection and the accuracy of the cluster location. We propose a systematic assessment method, using here extended power, to produce a map showing the performance of cluster detection tests over an entire region.

Methods

To explore the behavior of a cluster detection test on identical cluster types at any possible location, we successively applied four different spatial and epidemiological parameters. These parameters determined four cluster collections, each covering the entire study region. We simulated 1,000 datasets for each cluster and analyzed them with Kulldorff’s spatial scan statistic. From the area under the extended power curve, we constructed a map for each parameter set showing the performance of the test across the entire region.

Results

Consistent with previous studies, the performance of the spatial scan statistic increased with the baseline incidence of disease, the size of the at-risk population and the strength of the cluster (i.e., the relative risk). Performance was heterogeneous, however, even for very similar clusters (i.e., similar with respect to the aforementioned factors), suggesting the influence of other factors.

Conclusions

The area under the extended power curve is a single measure of performance and, although needing further exploration, it is suitable to conduct a systematic spatial evaluation of performance. The performance map we propose enables epidemiologists to assess cluster detection tests across an entire study region.