Accuracy of residential geocoding in the Agricultural Health Study.

BACKGROUND: Environmental exposure assessments often require a study participant's residential location, but the positional accuracy of geocoding varies by method and the rural status of an address. We evaluated geocoding error in the Agricultural Health Study (AHS), a cohort of pesticide applicators and their spouses in Iowa and North Carolina, U.S.A. METHODS: For 5,064 AHS addresses in Iowa, we compared rooftop coordinates as a gold standard to two alternate locations: 1) E911 locations (intersection of the private and public road), and 2) geocodes generated by matching addresses to a commercial street database (NAVTEQ) or placed manually. Positional error (distance in meters (m) from the rooftop) was assessed overall and separately for addresses inside (non-rural) or outside town boundaries (rural). We estimated the sensitivity and specificity of proximity-based exposures (crops, animal feeding operations (AFOs)) and the attenuation in odds ratios (ORs) for a hypothetical nested case-control study. We also evaluated geocoding errors within two AHS subcohorts in Iowa and North Carolina by comparing them to GPS points taken at residences. RESULTS: Nearly two-thirds of the addresses represented rural locations. Compared to the rooftop gold standard, E911 locations were more accurate overall than address-matched geocodes (median error 39 and 90 m, respectively). Rural addresses generally had greater error than non-rural addresses, although errors were smaller for E911 locations. For highly prevalent crops within 500 m (>97% of homes), sensitivity was >95% using both data sources; however, lower specificities with address-matched geocodes (more common for rural addresses) led to substantial attenuation of ORs (e.g., corn <500 m ORobs = 1.47 vs. ORtrue = 2.0). Error in the address-matched geocodes resulted in even greater ORobs attenuation for AFO exposures. Errors for North Carolina addresses were generally smaller than those in Iowa. CONCLUSIONS: Geocoding error can be minimized when known coordinates are available to test alternative data and methods. Our assessment suggests that where E911 locations are available, they offer an improvement upon address-matched geocodes for rural addresses. Exposure misclassification resulting from positional error is dependent on the geographic database, geocoding method, and the prevalence of exposure.

The impact of data suppression on local mortality rates: the case of CDC WONDER.

CDC WONDER (Centers for Disease Control and Prevention Wide-Ranging Online Data for Epidemiologic Research) is the nation's primary data repository for health statistics. Before WONDER data are released to the public, data cells with fewer than 10 case counts are suppressed. We showed that maps produced from suppressed data have predictable geographic biases that can be removed by applying population data in the system and an algorithm that uses regional rates to estimate missing data. By using CDC WONDER heart disease mortality data, we demonstrated that effects of suppression could be largely overcome.

Where do patients with cancer in Iowa receive radiation therapy?

PURPOSE: Multiple studies have shown survival benefits in patients with cancer treated with radiation therapy, but access to treatment facilities has been found to limit its use. This study was undertaken to examine access issues in Iowa and determine a methodology for conducting a similar national analysis. PATIENTS AND METHODS: All Iowa residents who received radiation therapy regardless of where they were diagnosed or treated were identified through the Iowa Cancer Registry (ICR). Radiation oncologists were identified through the Iowa Physician Information System (IPIS). Radiation facilities were identified through IPIS and classified using the Commission on Cancer accreditation standard. RESULTS: Between 2004 and 2010, 113,885 invasive cancers in 106,603 patients, 28.5% of whom received radiation treatment, were entered in ICR. Mean and median travel times were 25.8 and 20.1 minutes, respectively, to the nearest facility but 42.4 and 29.1 minutes, respectively, to the patient's chosen treatment facility. Multivariable analysis predicting travel time showed significant relationships for disease site, age, residence location, and facility category. Residents of small and isolated rural towns traveled nearly 3× longer than urban residents to receive radiation therapy, as did patients using certain categories of facilities. CONCLUSION: Half of Iowa patients could reach their nearest facility in 20 minutes, but instead, they traveled 30 minutes on average to receive treatment. The findings identified certain groups of patients with cancer who chose more distant facilities. However, other groups of patients with cancer, namely those residing in rural areas, had less choice, and some had to travel considerably farther to radiation facilities than urban patients.

Access to chemotherapy services by availability of local and visiting oncologists.

PURPOSE: Geographic disparities have raised important questions about factors related to treatment choice and travel time, which can affect access to cancer care. PATIENTS AND METHODS: Iowa residents who received chemotherapy regardless of where they were diagnosed or treated were identified through the Iowa Cancer Registry (ICR), a member of the SEER program. Oncologists and their practice locations, including visiting consulting clinics (VCCs), were tracked through the Iowa Physician Information System. Oncologists, VCCs, and patients were mapped to hospital service areas (HSAs). RESULTS: Between 2004 and 2010, 113,885 newly diagnosed invasive cancers were entered into ICR; among patients in whom these cancers were diagnosed, 31.6% received chemotherapy as a first course of treatment. During this period, 106 Iowa oncologists practiced in 14 cities, and 82 engaged in outreach to 85 VCCs in 77 rural communities. Of patients receiving chemotherapy, 63.0% resided in an HSA that had a local oncologist and traveled 21 minutes for treatment on average. In contrast, 29.3% of patients receiving chemotherapy resided in an HSA with a VCC, and 7.7% resided in an HSA with no oncology provider. These latter two groups of patients traveled 58 minutes on average to receive chemotherapy. Availability of oncologists and VCCs affected where patients received chemotherapy. The establishment of VCCs increased access to oncologists in rural communities and increased the rate that chemotherapy was administered in rural communities from 10% to 24%, a notable increase in local access. CONCLUSION: Access to cancer care is dependent on the absolute number of providers, but it is also dependent on their geographic distribution.

How many suffice? A computational framework for sizing sentinel surveillance networks.

BACKGROUND: Data from surveillance networks help epidemiologists and public health officials detect emerging diseases, conduct outbreak investigations, manage epidemics, and better understand the mechanics of a particular disease. Surveillance networks are used to determine outbreak intensity (i.e., disease burden) and outbreak timing (i.e., the start, peak, and end of the epidemic), as well as outbreak location. Networks can be tuned to preferentially perform these tasks. Given that resources are limited, careful site selection can save costs while minimizing performance loss. METHODS: We study three different site placement algorithms: two algorithms based on the maximal coverage model and one based on the K-median model. The maximal coverage model chooses sites that maximize the total number of people within a specified distance of a site. The K-median model minimizes the sum of the distances from each individual to the individual's nearest site. Using a ground truth dataset consisting of two million de-identified Medicaid billing records representing eight complete influenza seasons and an evaluation function based on the Huff spatial interaction model, we empirically compare networks against the existing Iowa Department of Public Health influenza-like illness network by simulating the spread of influenza across the state of Iowa. RESULTS: We show that it is possible to design a network that achieves outbreak intensity performance identical to the status quo network using two fewer sites. We also show that if outbreak timing detection is of primary interest, it is actually possible to create a network that matches the existing network's performance using 59% fewer sites. CONCLUSIONS: By simulating the spread of influenza across the state of Iowa, we show that our methods are capable of designing networks that perform better than the status quo in terms of both outbreak intensity and timing. Additionally, our results suggest that network size may only play a minimal role in outbreak timing detection. Finally, we show that it may be possible to reduce the size of a surveillance system without affecting the quality of surveillance information produced.

Who does not receive treatment for cancer?

PURPOSE: Little has been published on nontreatment of cancer, yet the National Cancer Data Base (NCDB) indicates that 9.2% of patients receive no first course of treatment. Because the NCDB is limited to accredited cancer programs, there is potential for the actual rate to differ. We sought to understand the rate and characteristics of patients with cancer who receive no first course of treatment in a more population-representative data source. MATERIALS AND METHODS: The Iowa Cancer Registry (ICR) strives to capture 100% of newly diagnosed cancer cases among Iowa residents, regardless of where they are diagnosed or treated. RESULTS: In the ICR from 2004 to 2010, 12.3% of newly diagnosed patients with cancer did not receive a first course of treatment, which is 48% higher than the NCDB data for the state of Iowa (8.3%) during the same time period. Logistic regression indicated that nontreatment was more common in certain cancers (ie, small-cell and non-small-cell lung/bronchial cancers and low-grade non-Hodgkin lymphoma), advanced stages, older patients, those receiving treatment recommendations at nonaccredited cancer programs, and patients who never consulted an oncologist, radiation therapist, or surgeon. Distance to treatment facilities was not related to nontreatment. CONCLUSION: The rate of nontreatment varies by cancer type and stage and is higher in patients receiving initial treatment recommendations in nonaccredited cancer programs than in accredited cancer programs. This pattern seems to be correlated with patient characteristics but also may be related to provider and facility characteristics available to people locally that influence both patient and provider decision making.

Primary stroke centers should be located using maximal coverage models for optimal access.

BACKGROUND AND PURPOSE: The current self-initiated approach by which hospitals acquire Primary Stroke Center (PSC) certification provides insufficient coverage for large areas of the United States. An alternative, directed, algorithmic approach to determine near optimal locations of PSCs would be justified if it significantly improves coverage. METHODS: Using geographic location-allocation modeling techniques, we developed a universal web-based calculator for selecting near optimal PSC locations designed to maximize the population coverage in any state. We analyzed the current PSC network population coverage in Iowa and compared it with the coverage that would exist if a maximal coverage model had instead been used to place those centers. We then estimated the expected gains in population coverage if additional PSCs follow the current self-initiated model and compared it against the more efficient coverage expected by use of a maximal coverage model to select additional locations. RESULTS: The existing 12 self-initiated PSCs in Iowa cover 37% of the population, assuming a time-distance radius of 30 minutes. The current population coverage would have been 47.5% if those 12 PSCs had been located using a maximal coverage model. With the current self-initiated approach, 54 additional PSCs on average will be needed to improve coverage to 75% of the population. Conversely, only 31 additional PSCs would be needed to achieve the same degree of population coverage if a maximal coverage model is used. CONCLUSIONS: Given the substantial gain in population access to adequate acute stroke care, it appears justified to direct the location of additional PSCs or recombinant tissue-type plasminogen activator-capable hospitals through a maximal coverage model algorithmic approach.

A probabilistic sampling method (PSM) for estimating geographic distance to health services when only the region of residence is known.

BACKGROUND: The need to estimate the distance from an individual to a service provider is common in public health research. However, estimated distances are often imprecise and, we suspect, biased due to a lack of specific residential location data. In many cases, to protect subject confidentiality, data sets contain only a ZIP Code or a county. RESULTS: This paper describes an algorithm, known as "the probabilistic sampling method" (PSM), which was used to create a distribution of estimated distances to a health facility for a person whose region of residence was known, but for which demographic details and centroids were known for smaller areas within the region. From this distribution, the median distance is the most likely distance to the facility. The algorithm, using Monte Carlo sampling methods, drew a probabilistic sample of all the smaller areas (Census blocks) within each participant's reported region (ZIP Code), weighting these areas by the number of residents in the same age group as the participant. To test the PSM, we used data from a large cross-sectional study that screened women at a clinic for intimate partner violence (IPV). We had data on each woman's age and ZIP Code, but no precise residential address. We used the PSM to select a sample of census blocks, then calculated network distances from each census block's centroid to the closest IPV facility, resulting in a distribution of distances from these locations to the geocoded locations of known IPV services. We selected the median distance as the most likely distance traveled and computed confidence intervals that describe the shortest and longest distance within which any given percent of the distance estimates lie. We compared our results to those obtained using two other geocoding approaches. We show that one method overestimated the most likely distance and the other underestimated it. Neither of the alternative methods produced confidence intervals for the distance estimates. The algorithm was implemented in R code. CONCLUSIONS: The PSM has a number of benefits over traditional geocoding approaches. This methodology improves the precision of estimates of geographic access to services when complete residential address information is unavailable and, by computing the expected distribution of possible distances for any respondent and associated distance confidence limits, sensitivity analyses on distance access measures are possible. Faulty or imprecise distance measures may compromise decisions about service location and misdirect scarce resources.

Explaining place-based colorectal cancer health disparities: evidence from a rural context.

A growing body of work examines geographical setting as a source of health disparity, hypothesizing individual as well as larger, environmental sources of risk. However, mechanisms by which this influence operates, especially in rural settings, are not well understood. This study investigates the problem of colorectal cancer in a rural US community through the lens of geographical setting. Statewide maps of colorectal cancer burdens show a place-based disparity in colorectal cancer in the region surrounding a small, diverse Iowa community. Within a research partnership framework, we use these maps to engage community residents in discussions of high colorectal cancer rates. We ask how a rural community experiencing higher than expected rates of colorectal cancer late-stage diagnosis and mortality perceives and explains their increased risk, interpreting available epidemiological evidence based on their lived experience. We use concept mapping to organize these perceptions and situate our findings in the context of previous work. Our findings reveal a complex understanding of risk that should be taken into account in crafting effective public health interventions and messages. Our work informs the growing literature on how context influences individual experiences of health problems, with specific relevance for rural populations.

Optimizing influenza sentinel surveillance at the state level.

Influenza-like illness data are collected via an Influenza Sentinel Provider Surveillance Network at the state level. Because participation is voluntary, locations of the sentinel providers may not reflect optimal geographic placement. The purpose of this study was to determine the "best" locations for sentinel providers in Iowa by using a maximal coverage model (MCM) and to compare the population coverage obtained with that of the current sentinel network. The authors used an MCM to maximize the Iowa population located within 20 miles (32.2 km) of 1-143 candidate sites and calculated the coverage provided by each additional site. The first MCM location covered 15% of the population; adding a second increased coverage to 25%. Additional locations provided more coverage but with diminishing marginal returns. In contrast, the existing 22 Iowa sentinel locations covered 56% of the population, the same coverage achieved with just 10 MCM sites. Using 22 MCM sites covered more than 75% of the population, an improvement over the current site placement, adding nearly 600,000 Iowa residents. Given scarce public health resources, MCMs can help surveillance efforts by prioritizing recruitment of sentinel locations.

Mapping cancer for community engagement.

INTRODUCTION: Two research strategies may reduce health disparities: community participation and the use of geographic information systems. When combined with community participation, geographic information systems approaches, such as the creation of disease maps that connect disease rates with community context, can catalyze action to reduce health disparities. However, current approaches to disease mapping often focus on the display of disease rates for political or administrative units. This type of map does not provide enough information on the local rates of cancer to engage community participation in addressing disparities. METHODS: We collaborated with researchers and cancer prevention and control practitioners and used adaptive spatial filtering to create maps that show continuous surface representations of the proportion of all colorectal cancer cases diagnosed in the late stage. We also created maps that show the incidence of colorectal cancer. RESULTS: Our maps show distinct patterns of cancer and its relationship to community context. The maps are available to the public on the Internet and through the activities of Iowa Consortium for Comprehensive Cancer Control partners. CONCLUSION: Community-participatory approaches to research are becoming more common, as are the availability of geocoded data and the use of geographic information systems to map disease. If researchers and practitioners are to engage communities in exploring cancer rates, maps should be made that accurately represent and contextualize cancer in such a way as to be useful to people familiar with the characteristics of their local areas.

Geocoding accuracy and the recovery of relationships between environmental exposures and health.

BACKGROUND: This research develops methods for determining the effect of geocoding quality on relationships between environmental exposures and health. The likelihood of detecting an existing relationship - statistical power - between measures of environmental exposures and health depends not only on the strength of the relationship but also on the level of positional accuracy and completeness of the geocodes from which the measures of environmental exposure are made. This paper summarizes the results of simulation studies conducted to examine the impact of inaccuracies of geocoded addresses generated by three types of geocoding processes: a) addresses located on orthophoto maps, b) addresses matched to TIGER files (U.S Census or their derivative street files); and, c) addresses from E-911 geocodes (developed by local authorities for emergency dispatch purposes). RESULTS: The simulated odds of disease using exposures modelled from the highest quality geocodes could be sufficiently recovered using other, more commonly used, geocoding processes such as TIGER and E-911; however, the strength of the odds relationship between disease exposures modelled at geocodes generally declined with decreasing geocoding accuracy. CONCLUSION: Although these specific results cannot be generalized to new situations, the methods used to determine the sensitivity of results can be used in new situations. Estimated measures of positional accuracy must be used in the interpretation of results of analyses that investigate relationships between health outcomes and exposures measured at residential locations. Analyses similar to those employed in this paper can be used to validate interpretation of results from empirical analyses that use geocoded locations with estimated measures of positional accuracy.

Community-level risk factors for depression hospitalizations.

This study measured geographic variation in depression hospitalizations and identified community-level risk factors. Depression hospitalizations were identified from the Statewide Inpatient Database. The dependent variable was specified as the indirectly standardized hospitalization rate. County-level data for 14 states were collected from federal agencies. The Bayesian spatial regression model included socio-demographic, economic, and health system characteristics as independent variables. There were 8.5 depression hospitalizations per 1,000 residents. 8.8% of counties had hospitalization rates 33% greater than the standardized rate. Significant risk factors included unemployment, poverty, physician supply, and hospital bed supply. Significant protective factors included rurality, economic dependence, and housing stress.

Modeling the probability distribution of positional errors incurred by residential address geocoding.

BACKGROUND: The assignment of a point-level geocode to subjects' residences is an important data assimilation component of many geographic public health studies. Often, these assignments are made by a method known as automated geocoding, which attempts to match each subject's address to an address-ranged street segment georeferenced within a streetline database and then interpolate the position of the address along that segment. Unfortunately, this process results in positional errors. Our study sought to model the probability distribution of positional errors associated with automated geocoding and E911 geocoding. RESULTS: Positional errors were determined for 1423 rural addresses in Carroll County, Iowa as the vector difference between each 100%-matched automated geocode and its true location as determined by orthophoto and parcel information. Errors were also determined for 1449 60%-matched geocodes and 2354 E911 geocodes. Huge (> 15 km) outliers occurred among the 60%-matched geocoding errors; outliers occurred for the other two types of geocoding errors also but were much smaller. E911 geocoding was more accurate (median error length = 44 m) than 100%-matched automated geocoding (median error length = 168 m). The empirical distributions of positional errors associated with 100%-matched automated geocoding and E911 geocoding exhibited a distinctive Greek-cross shape and had many other interesting features that were not capable of being fitted adequately by a single bivariate normal or t distribution. However, mixtures of t distributions with two or three components fit the errors very well. CONCLUSION: Mixtures of bivariate t distributions with few components appear to be flexible enough to fit many positional error datasets associated with geocoding, yet parsimonious enough to be feasible for nascent applications of measurement-error methodology to spatial epidemiology.

Geocoding in cancer research: a review.

There is now widespread agreement that geographic identifiers (geocodes) should be assigned to cancer records, but little agreement on their form and how they should be assigned, reported, and used. This paper reviews geocoding practice in relation to major purposes and discusses methods to improve the accuracy of geocoded cancer data. Differences in geocoding methods and materials introduce errors of commission and omission into geocoded data. A common source of error comes from the practice of using digital boundary files of dubious quality to place addresses into areas of interest. Geocoded data are linked to demographic, environmental, and health services data, and each data type has unique accuracy considerations. In health services applications, the accuracy of distances computed from geocodes can differ markedly. Privacy and confidentiality issues are important in the use and release of geocoded cancer data. When masking methods are used for disclosure limitation purposes, statistical methods must be adjusted for the locational uncertainty of geocoded data. We conclude that selection of one particular type of geographic area as the geocode may unnecessarily constrain future work. Therefore, the longitude and latitude of each case is the superior basic geocode; all other geocodes of interest can be constructed from this basic identifier.

Using software agents to preserve individual health data confidentiality in micro-scale geographical analyses.

Confidentiality constraints often preclude the release of disaggregate data about individuals, which limits the types and accuracy of the results of geographical health analyses that could be done. Access to individually geocoded (disaggregate) data often involves lengthy and cumbersome procedures through review boards and committees for approval (and sometimes is not possible). Moreover, current data confidentiality-preserving solutions compatible with fine-level spatial analyses either lack flexibility or yield less than optimal results (because of confidentiality-preserving changes they introduce to disaggregate data), or both. In this paper, we present a simulation case study to illustrate how some analyses cannot be (or will suffer if) done on aggregate data. We then quickly review some existing data confidentiality-preserving techniques, and move on to explore a solution based on software agents with the potential of providing flexible, controlled (software-only) access to unmodified confidential disaggregate data and returning only results that do not expose any person-identifiable details. The solution is thus appropriate for micro-scale geographical analyses where no person-identifiable details are required in the final results (i.e., only aggregate results are needed). Our proposed software agent technique also enables post-coordinated analyses to be designed and carried out on the confidential database(s), as needed, compared to a more conventional solution based on the Web Services model that would only support a rigid, pre-coordinated (pre-determined) and rather limited set of analyses. The paper also provides an exploratory discussion of mobility, security, and trust issues associated with software agents, as well as possible directions/solutions to address these issues, including the use of virtual organizations. Successful partnerships between stakeholder organizations, proper collaboration agreements, clear policies, and unambiguous interpretations of laws and regulations are also much needed to support and ensure the success of any technological solution.

Analyzing geographic patterns of disease incidence: rates of late-stage colorectal cancer in Iowa.

This study, using geocodes of the locations of residence of newly diagnosed colorectal cancer patients from the Iowa Cancer Registry, computed continuous spatial patterns of late-stage rates of colorectal cancer in Iowa. Variations in rates in intrahospital service regions were as great as interhospital service regions, shown by analysis of variance tests. Some of the spatial variations observed could be explained, using a general linear regression model on individual-level data, by spatial variations in attributes of individuals and their relationships to health resources. We show how this source of variation can be removed from the original map leaving a new map showing the remaining variation in late-stage rate not explained by these relationships. We argue that it would be more appropriate to organize prevention and control activities targeted at the areas with higher than expected late-stage rates shown on this map. The originality of this approach is in the integration of geocoded data from a cancer registry with methods of spatial analysis that provide considerable geographic detail in the cancer rate while controlling for rate stabilization and reliability due to the small number problem.

Public health, GIS, and spatial analytic tools.

We review literature that uses spatial analytic tools in contexts where Geographic Information Systems (GIS) is the organizing system for health data or where the methods discussed will likely be incorporated in GIS-based analyses in the future. We conclude the review with the point of view that this literature is moving toward the development and use of systems of analysis that integrate the information geo-coding and data base functions of GISystems with the geo-information processing functions of GIScience. The rapidity of this projected development will depend on the perceived needs of the public health community for spatial analysis methods to provide decision support. Recent advances in the analysis of disease maps have been influenced by and benefited from the adoption of new practices for georeferencing health data and new ways of linking such data geographically to potential sources of environmental exposures, the locations of health resources and the geodemographic characteristics of populations. This review focuses on these advances.