Initial exploration of newly implemented public health policy using geographic information systems: the case of a U.S. silver alert program.
Public health policies generally target specific subsets of the population to address unique but prevalent health problems. Once implemented, policies should be carried out such that they effectively address their missions (e.g., serving for the specific target population with a unique health problem). Previous studies have shown that utilization patterns are critical to assessing the progress of newly implemented public health programs (Glasgow, Vogt, & Boles, 1999; Handler, Issel, & Turnock, 2001). Indeed, if a program is utilized in limited geographic areas more than others, the mission of the policy is unlikely being met. Yet evaluation of utilization patterns is rarely employed as an approach to assess the efficacy of programs. A primary reason that evaluation of policies does not often begin with analysis of utilization patterns is related to the challenge of effectively and efficiently measuring utilization at a system level. We propose that implementation of Geographic Information System (GIS) approach may provide the tools to remedy this problem, providing an accessible way to assess utilization patterns, and by extension, the impact of health policies on the populations for whom they are designed. We present a case study to demonstrate a suggested utilization pattern assessment strategy and provide a discussion for possible applications in current and future public health policies.
One newly implemented US public health policy the Silver Alert (SA) policy--called for development of public media notification programs for quickly locating missing adults. These SA programs are implemented at the state level, and activate alerts that provide information about missing adults are disseminated through a variety of media outlets such as radio, TV and the internet to receive public assistance in the search. The SA policy is designed as a public health strategy for ensuring the safety of missing older adults and adults with cognitive impairment. Although SA programs have rapidly emerged throughout the United States, little is known about their utilization patterns, particularly with regard to when, where and how often SA programs are used (Carr et al., 2010). Assessment of utilization patterns in the initial stage of health policy implementation is crucial. Without an appropriate initial assessment, the outcome evaluation of a new public health program may be biased because the program may be servicing the wrong target populations or only particular areas, not necessarily those for whom it was designed (Glasgow et al., 1999; Handler et al., 2001). To date, no systematic evaluations of the utilization patterns for the North Carolina SA have been completed, and in coordination with being in the early stages of implementation, it serves as an ideal case study for examining whether the utilization of the policy adheres to its mission. Our case study could be a basis for future applications in entire or specific components of major state- and nation-wide public health policies such as the Affordable Care Act (ACA).
SILVER ALERTS, BRIEF HISTORY AND ISSUES
Implemented in 29 states in the U.S. since 2006 (Carr et al., 2010), SA are public alert programs for older adults and adults with cognitive impairments at risk of wandering and going missing. SA programs are built on the existing AMBER (America's Missing: Broadcast Emergency Response) alert program infrastructure, a program designed for locating missing children (Muschert, Young-Spillers, & Carr, 2006). A National SA Act was passed in 2008 to provide assistance to states interested in implementing their own programs; however, several states established their own SA programs prior to federal legislation. Like other state SA policies, the mission of the SA policy in North Carolina (see North Carolina General Statutes [section] 143B-499.8) is explicitly stated in its legislation: "to provide a statewide system for the rapid dissemination of information regarding a missing person who is believed to be suffering from dementia or other cognitive impairment" with parameters including adults (age 18 or older), unless they are autistic for which there is a special provision for individuals younger than 18 to be included (North Carolina Department of Crime Control & Public Safety, 2011).
The SA policies were designed, in part, to address the problem of a growing number of individuals with dementia in community-based settings who are likely to, and often do, wander outside of their homes. This problem, which is related to population aging (3) coupled with associated increases in cognitive impairment (4), is proposed to cause increased utilization of long-term care. SA policies are proposed as a way to help delay utilization of expensive institutional long-term care services by helping caregivers of adults with dementia/cognitive impairment continue caring for their loved ones so they can stay at home as long as possible. As a result, prevention of life-threatening events (e.g., accident) and causes of institutionalization (e.g., severe injuries) related to wandering have become an urgent public health policy task (Alzheimer's Association, 2007). SA programs require certain infrastructure such as coordinated systems between state/local law enforcement agency, a variety of media outlets (e.g., TV, highway message signs) and the general public. Indeed, enactment of legislation (e.g., national SA legislation in 2008) and program implementation has been rapid in the U.S., with more than two dozen states developing SA programs since 2006 (Carr et al., 2010). However, to date, virtually no evaluations of any kind have been done on SA programs. As such, the utilization and effectiveness of SA programs are not known. Therefore, assessment of utilization patterns of SA programs is needed.
FRAMEWORKS FOR GUIDING EVALUATION OF UTILIZATION PATTERNS
Although there are several approaches that have been used in the evaluation of public health policies, we look to two common frameworks to guide our analysis: the RE-AIM model (Glasgow et al., 1999) and public health system (PHS) performance model (Handler et al., 2001). The RE-AIM model provides a framework for systematically assessing the effectiveness of public health interventions, with a focus on five specific domains (reach, efficacy, adoption, implementation and maintenance) in two levels. Reach and efficacy concern individual-level impacts of public health intervention on the coverage of target population and individual outcomes. Adoption, implementation and maintenance concern organization-level impacts such as process of dissemination, utilization patterns and sustainability of interventions over time. The PHS performance model provides a framework for monitoring the operation of a public health system/policy with consideration of practical applications in the real world. The PHS performance model highlights the important evaluation domains including mission/purpose, structural capacity, process and outcomes. In principal, this model focuses on whether a public health policy is implemented based on its mission, recognizing that the implementation process often needs to be adjusted based on available resources (e.g., infrastructure, personnel). Policy outcomes are not simply end products but resources used to inform the next cycle of policy implementation. The PHS performance model emphasizes the importance of continuous improvement/adjustment of structural capacity and implementation process to fulfill the mission of a health policy. Unlike the RE-AIM model, all domains in the PHS performance model concern organization-/system-levels of public health policy. These two models inform our guiding approach to evaluating the efficacy of the SA policy, emphasizing assessment of how a specific mission addresses the needs of a specific population.
THE CASE FOR GEOGRAPHIC INFORMATION SYSTEM (GIS) IN HEALTH POLICY EVALUATIONS
Geographic Information System (GIS) provides a useful way to effectively analyze and clearly present public health policy utilization patterns to a broad audience (Higgs, 2004). GIS is computer software that manages, analyzes and visualizes geographically-referenced data (Cromley & McLafferty, 2002). One of the main advantages of the GIS approach is data visualization. In more traditional statistical analysis frameworks, intuitively identifying geographic summary measures (e.g., "average" geographical patterns, areas where a public health policy is used more often than their surrounding areas) and effectively communicating the results are demanding tasks (Sips, Schneidewind, & Keim, 2007).
Two primary challenges create barriers for carrying out an evaluation of utilization patterns for newly designed policies. First, lack of data about utilization is often a problem, particularly for new policies. In our case, despite adoption of the SA policy by the majority of US states, North Carolina is the only state that provides publicly available SA data, which is why our study focuses exclusively on the North Carolina SA policy (http://www.nccrimecontrol.org/). Even though data exists, however, limited information is provided about the number of cases, the date of utilization, location of utilization of the policy (i.e., city and county) and recovery status of individuals as a result of the policy. Second, developing a comprehensive understanding of utilization patterns requires multiple forms of data to be simultaneously examined. The most common data about program/policy utilization takes the form of frequency/count, time and location (Courtney, 2005). For this reason, despite data limitations, in the case of North Carolina's SA policy we focus on identifying the relationship between the number of alerts activated and when/where they are used. Such inquiry and dissemination of findings are cognitively intense tasks particularly when a study area is large (e.g., state, country) (Heer, Viegas, & Wattenberg, 2009; Rosling, 2007). Given that most policy-related data are collected according to political boundaries (often by state and county in the U.S.) and calendar years, the capability of GIS maximizes the information from routinely collected data (Higgs, 2004).
GIS-based data visualization (a.k.a., geovisualziation) also facilitates more effective dissemination of research findings because it provides easily interpretable visual aids despite complex combinations of numeric data (e.g., frequency, location and time) (Koua & Kraak, 2004). In part, this is because data visualization reduces the cognitive tasks necessary to understand data (Few, 2004; Tufte, 2001). That is, individuals can quickly see the relationships among the characteristics (e.g., locations, adjacent counties) of study areas, policy utilization patterns by time (e.g., increase/decrease compared to previous/next year), and patterns by locations (i.e., more/less usage in one location compared to other locations). Easily interpretable data graphics can promote active engagement in discussion among policy makers, researchers, practitioners and the general public, and in turn, produce more creative and practical policy-decisions (Cummins, Curtis, Diez-Roux, & Macintyre, 2007; Dummer, 2008; Rosling, 2007). In short, GIS-based data visualization is an effective strategy to manage, understand and present complex data in the context of health policy evaluations.
This paper presents an analysis of the utilization patterns for North Carolina to show how GIS can be effectively utilized with evaluation of new public health policy programs. Our GIS approach of the North Carolina SA policy is guided by the RE-AIM and PHS performance public health policy evaluation models. We systematically assess utilization patterns reflecting organization-level domains (i.e., adoption, implementation and maintenance). Specifically, in this study, we focus on the level of adherence with the SA policy mission targeting adult populations with the ultimate goal of showing how a GIS-based utilization pattern assessment method could be beneficial for other public health policy evaluations.
MATERIALS AND METHODS
The data for the first three years in which the North Carolina Silver Alert policy was fully implemented, 2008, 2009 and 2010, were obtained from the North Carolina Department of Crime Control & Public Safety (http://www.nccrimecontrol.org/). Altogether, 587 SA were activated in 2008 (n = 128), 2009 (n = 239) and 2010 (n = 220). The SA data available include the name of missing persons, county of residence, city where each alert was initiated, date the SA was canceled and recovery status. Additionally, map data and demographic data (e.g., number and percentage
of older population age 65 years and older) for counties in North Carolina were obtained from the U.S. Census Bureau (2010). All data are aggregated and merged in the ArcGIS 10 geodatabase (ESRI, 2011).
Descriptive summary of SA utilization and county population are computed by county. North Carolina has 100 counties with mean county population of 93,809 (SD = 141,085; maximum = 913,639; minimum = 4,078) (U.S. Census Bureau, 2010). The analysis consists of three parts including visual examination of thematic maps, overlay analysis and spatial statistical analysis. First, thematic county maps showing the count of SA program activated were developed for the year of 2008, 2009 and 2010. The color-coding (i.e., gray-scale) of maps is done using the quintile-based classification at 0, 1-2, 3-5, 6-33 and 34-85. After examining the distribution of data, Wake county (n = 85) is separated into another category due to a significantly larger number of SA than the rest of counties. Thematic map development often employs commonly used classification methods (e.g., quartile, quintile) with adjustments according to preliminary data analysis and/or expert opinion (C. A. Brewer, 2006). To better understand the utilization patterns of SA policy, color-coded bar chart (corresponding to the map color-coding) was added to the thematic maps. Second, two thematic maps, SA count between 2008 and 2010, and county adult population, are overlaid in one map. Overlay maps enable simultaneous examination of multiple measures (Mitchell, 1999). The number of adult (18+) population is represented by the height of the bar symbol in the overlay analysis. Based on visual examination, symbols and brief comments were added.
Finally, spatial statistical analysis were used to quantify the patterns of SA policy utilization in two ways: global and local measures. In the global measure, the SA utilization pattern is evaluated given all 100 counties in North Carolina. Moran's I statistic, which computes spatial autocorrelation or the relationship between values of a feature (i.e., count of SA policy activated in counties) and their spatial relationship (Waller & Gotway, 2004). Moran's I statistic ranges from -1 to 1. Near -1 indicates dispersion (i.e., the nearer the counties, the more dissimilar their SA counts are) and near 1 indicates clustering (i.e., the nearer the counties, the more similar their SA counts are). When there is no spatial autocorrelation, Moran's I is 0 (i.e., random). In the local measure, each county and its neighboring counties are examined to quantify concentration of high/low values (i.e., count of SA activated) or hot/cold spots. That is, groups of neighboring counties are classified into hot spot, cold spot or no pattern according to the frequency of SA policy activated. Anselin's Local Moran's I statistic is computed to identify hot/cold spots (Mitchell, 2005). Positive Local Moran's I value indicates neighboring counties have similar frequencies of SA policy activated and negative Local Moran's I value indicate dissimilar frequencies.
One important decision to make in spatial analysis is definition of "neighbors" or search area (i.e., which counties should be compared?) (Anselin, 2002). In this study, the five nearest neighboring counties were identified as "neighbors" for two reasons. First, because the unit of analysis is determined by county boundaries (i.e., polygon), a distance-based definition is problematic as each county has a different size and shape. Also, counties adjacent to the state boundary and ocean have fewer neighboring counties, and therefore, distance measurements are not consistent with others (i.e., edge effect) (Waller & Gotway, 2004). Second, multiple neighbor definitions (i.e., k-nearest neighbors; k = 3, 4, 5, 6, 7, 8, 9, 10) are examined in the preliminary analyses, and the five nearest neighbors approach was chosen because of the highest value of Moran's I (i.e., the greatest spatial autocorrelation) and consideration of edge effects. Results are visualized according to basic mapping guidelines (C. A. Brewer, 2006). All analyses were performed using the ESRI ArcGIS 10 software.
Table 1 presents the descriptive summary of SA activated during 2008, 2009 and 2010 in North Carolina. There was a large increase in the total number of alerts activated between 2008 and 2009. Also, 26 out of 100 counties in North Carolina never used SA between 2008 and 2010. At the same time, 85 alerts were activated in one county (Wake County--see Figure 1). As the color-coded bar chart in Figure 1 demonstrates, Wake County had exceptionally frequent use of alerts compared to other counties in North Carolina. These descriptive summary statistics are further detailed in Figure 1. The cluster of counties (in dark gray color) with relatively frequent use of the SA policy can be visually observed (Wake County in black color). In addition, many counties with zero use of SA were observed in the mountain areas (west extreme) and coast areas (east extreme).
Figure 2 shows the results of a map overlay analysis, combining the maps of alerts activated and adult (age 18 and older) populations by county. Results suggest that the frequency of alerts activated appear to follow the size of adult populations. On one hand, the counties on the western side of North Carolina had the smallest adult populations and fewest (mostly zero) SA. On the other hand, the majority of SA activations were in central North Carolina where populous counties are located. However, it must be noted that one of the most populous counties (Mecklenburg County) had only 9 SA activated. Even when adjusted for the adult population, Mecklenburg County had about 1.4 SA activations per 100,000 adults whereas Wake County (equivalently populous county to Mecklenburg County) had 13.9 cases. On a related note, the findings can be applied for older populations as the number of total adult population and older population are highly correlated (r = .93). In other words, the county with a larger population has a larger older population. Also, we verified such relationship with visual examination of the color-coded map for older population (results available from the authors upon request) across counties in North Carolina.
Finally, the global Moran's I (I = 0.24; Z-score = 5.19, p < 0.001) describe the clustering patterns of SA activated across counties in North Carolina. In other words, given all counties in North Carolina, the pattern of SA utilization is statistically significantly clustered. Additionally, the Local Moran's I (i.e., hot spot analysis) identified two specific clusters of counties where SA were frequently activated (see Figure 3). Among these counties in hot spots, the Local Moran's I index ranged from 4.33 to 0.78 (corresponding Z-scores ranged from 11.94 to 2.17, p < 0.05). On a related note, the Local Moran's I did not detect statistically significant cold spots. However, visual examination of Figure 1 suggests potential areas (e.g., extreme west) for further investigation.
Understanding utilization patterns is critical to determining whether a policy effectively addresses its mission. Guided by the RE-AIM and PHS performance models, in this paper we propose that GIS provides a valuable tool for examining mission adherence. However, GIS has been underutilized in newly implemented public health policy evaluations, particularly in initial utilization pattern assessments. Similarly, even though GIS has been frequently used in epidemiological studies (e.g., cancer prevalence, health resource availability), few previous studies have employed GIS to explicitly evaluate utilization patterns of public health policies (McLafferty, 2003; Moore, Diez Roux, Nettleton, & Jacobs, 2008; Nykiforuk & Flaman, 2011).
GIS analysis is especially useful for evaluation of newly implemented health policies for which utilization patterns consider time and space (Higgs, 2009). As shown in the RE-AIM and PHS performance models, ensuring proper execution of multiple domains such as reaching out an appropriate target population and addressing its mission in the initial stages of policy is critical. In our example, North Carolina's Silver Alert health policy, this is especially true. This recently introduced policy relies on a coordinated effort between law enforcement and the media, and is designed primarily to help caregivers of individuals with a cognitive impairment locate their missing loved ones. GIS proves especially beneficial for understanding new programs in the early stages of implementation like the SA policy because it provides a graphical representation of the spatial relationship between utilization frequency and broader demographic patterns (Ghetian, Parrott, Volkman, & Lengerich, 2008; Horev, Pesis-Katz, & Mukamel, 2004). Our GIS-based visualized data (i.e., maps) of the utilization patterns for North Carolina's Silver Alert policy unequivocally show disproportionate utilization across counties, especially when considering the demographic group at highest risk of wandering (i.e., older adults 65+, adults with cognitive impairment/mental illness). Alerts are not related to the size and distribution of the adult population across the state. In other words, our results show that the utilization of the policy is arguably not in concordance with its mission in view of the RE-AIM and PHS models (Glasgow et al., 1999; Handler et al., 2001).
Beyond assessing whether the mission is being achieved, GIS-based visualized data (i.e., maps) also provide a platform for proposing next steps for understanding why policies may not be effectively addressing their missions. GIS can inform next steps involved in improving a policy and developing hypotheses for further research on the efficacy and value of a policy (Aigner, Bertone, Miksch, Tominski, & Schumann, 2008; Heer et al., 2009). Based on our case analysis of the SA policy in North Carolina, we propose two issues that should guide next steps for evaluation and implementation of the North Carolina SA policy. First, the possible causes for the kind of utilization patterns observed could include inadequate infrastructure (e.g., radio station, electric billboard, internet) in certain parts of the state, lack of public awareness about the program, or barriers facing the law enforcement officials who are key to the implementation of the policy. Identifying how these factors contribute to utilization is a key next step.
Second, we suggest the need for a better understanding of where individuals with cognitive impairments reside so that it is possible to better assess utilization rates. Unfortunately, these data are not currently publicly available in North Carolina, but by working with the state health agencies, such data may be attainable. Careful analysis like this is critical because more alerts are not necessarily better. In fact, with a system like the SA, too many alerts could lead the public to become oversaturated and no longer respond to alerts (i.e., public fatigue) (Carr et al., 2010). Over-utilizing and underutilization are equally problematic. We propose that the unique role of Wake County as a political and media center is a sensible starting point for future investigation of the relationship between ideal utilization rates for the state and the other factors that lead to disproportionately low utilization in some areas.
Drawing from the case of North Carolina's newly implemented SA policy, we propose two reasons why GIS-based assessment of newly implemented public health policy is important. First, as demonstrated in Figure 1, 2 and 3, GIS-based data visualization and spatial inquiry produces easily interpretable maps of utilization patterns. These visualized data enhance engagement in data analysis because the data are more accessible to a variety of individuals (Rosling, Rosling-Ronnlund, & Rosling, 2004; Sieber, 2006). Often, comparing multiple measures such as frequency, time and location, and identifying patterns are challenging cognitive tasks for most individuals, and GIS mapping makes the complex information needed to assess a health policy more accessible (Sips et al., 2007). Second, maps facilitate interdisciplinary efforts to guide public health policy evaluation and facilitate engagement in decision-making among researchers, practitioners, policy makers and the public sector (Aigner et al., 2008; I. Brewer, MacEachren, Abdo, Gundrum, & Otto, 2000). Bringing a range of stake holders involved in development and implementation of a health policy together to examine patterns in visualized data is not just beneficial for improving a policy once it is in place, but it also can bring closer together the relationship between research, practice, and politics (Koua & Kraak, 2004).
GIS analysis is powerful and flexible both in the planning and evaluation stages of health policy development. As demonstrated by the case of North Carolina's Silver Alert policy, visual examination of multivariate data (e.g., count of alerts activated, adult population and location) is a sensible way to explore the data because it helps generate hypotheses and clarify tasks for the next phase of policy implementation. Our study shows that examination of utilization patterns for newly implemented health policies is needed before any conclusive assessment of the value of a policy can occur because outcomes of a health policy that are not implemented according to their mission may produce misleading information. As the ACA becomes fully implemented in 2014, GIS may provide a critical tool for assessment of whether and in what ways the policy is addressing the needs of US citizens (e.g., insurance coverage, health/preventive care service utilization). Analysis of hot spots like those found in our study, provides a way to assess the geographically differential impact a policy. For example, although the SA policy is a state-based policy, more than one-fourth of North Carolina counties never used the SA policy between 2008 and 2010, and about one-fourth of the total alerts occurred in a four-county cluster at the center of the state. As the ACA is based on providing affordable health care to all US citizens who need it, such an analysis may be an ideal tool for observing utilization/diffusion patterns (e.g., insurance coverage, service utilization rates by the poor) over the first few years of implementation, and guiding system level changes to assure adherence of the policy mission.
Results from this study should be treated with caution as our analysis is not confirmatory. Findings are limited to the county-level, and therefore, extensive discussion at the individual-level is beyond the scope of this study. Moreover, a different definition of county "neighbors" could alter the results to some degree; little theory is available to guide this definition and therefore, as practiced in this study, a combination of an iterative approach and an empirical approach is recommended (Anselin, 2002; Chi & Zhu, 2008). Finally, results from this study need to be verified through additional data collection and with different policies.
Suggested by widely regarded public health policy/system evaluation modelsthe RE-AIM and PHS performance modelsGIS-based assessment of newly implemented public health policies like SA is a suitable approach as it enables collaborative exploratory analysis of mission adherence through visualized data and inspires meaningful discussion for next phases of policy implementation. Public health evaluations underutilize the capability of GIS for analyzing utilization patterns in the initial implementation stage. Our findings show that the utilization patterns of the North Carolina SA program were geographically disproportionate, which suggests that the SA policy requires strategic modifications to meet its mission (e.g., serving for the public) and objectives in this early stage and next phase of implementation. Without modifying the policy to address these issues first, a complete outcome evaluation of this policy may produce misleading and inaccurate findings. The analytic approach used in this study could be adopted for other public health policies to ensure fair evaluations and in turn, proper maintenance as well as further improvement. Major health policies like the ACA would benefit from utilization of GIS as a tool for mission adherence and communication with the public about the way the policy is being implemented.
List of Abbreviations
GIS = Geographic Information Systems
SA = Silver Alert
AMBER = America's Missing: Broadcast
PHS = Public Health System
ACA = Affordable Care Act
There are no conflicts of interests.
This project was partially supported by the NIH grant (5T32 AG000272-09). The funding agency had no role in the process of study design, analysis, interpretation or writing of the paper.
Aigner, W., Bertone, A., Miksch, S., Tominski, C., & Schumann, H. (2008). Towards a conceptual framework for visual analytics of time and time-oriented data. Simulation Conference, 2007 (pp. 721-729). IEEE.
Alzheimer's Association. (2007). Wandering behavior: preparing for and preventing it. http://www.alz.org/national/documents/topicsheet wandering.pdf
Anselin, L. (2002). Under the hood issues in the specification and interpretation of spatial regression models. Agricultural Economics, 27(3), 247-267.
Brewer, C. A. (2006). Basic Mapping Principles for Visualizing Cancer Data Using Geographic Information Systems (GIS). American Journal of Preventive Medicine, 30(2, Supplement), S25-S36. doi: 10.1016/j.amepre.2005.09.007
Brewer, I., MacEachren, A. M., Abdo, H., Gundrum, J., & Otto, G. (2000). Collaborative geographic visualization: enabling shared understanding of environmental processes. Paper presented at the Information Visualization, 2000. InfoVis 2000. IEEE Symposium on.
Carr, D. C., Muschert, G. W., Kinney, J., Robbins, E., Petonito, G., Manning, L., & Brown, J. S. (2010). Silver Alerts and the Problem of Missing Adults with Dementia. The Gerontologist, 50(2), 149-157.
Chi, G., & Zhu, J. (2008). Spatial regresion models for demographic analysis. Population Research and Policy Review, 27(1), 17-42.
Courtney, K. L. (2005). Visualizing nursing workforce distribution: Policy evaluation using geographic information systems. International Journal of Medical Informatics, 74(11-12), 980-988. doi: 10.1016/j.ijmedinf.2005.07.012
Cromley, E. K., & McLafferty, S. (2002). GIS and Public Health: The Guilford Press.
Cummins, S., Curtis, S., Diez-Roux, A. V., & Macintyre, S. (2007). Understanding and representing 'place' in health research: A relational approach. Social Science & Medicine, 65(9), 1825-1838. doi: 10.1016/j.socscimed.2007.05.036
Dummer, T. J. B. (2008). Health geography: supporting public health policy and planning. Canadian Medical Association Journal, 178(9), 1177-1180. doi: 10.1503/cmaj.071783
ESRI. (2011). ArcMap (Version 10). Redland, CA: ESRI. Few, S. (2004). Show me the numbers: designing tables and graphcs to enlighten. Oakland, CA: Analytics Press.
Ghetian, C. B., Parrott, R., Volkman, J. E., & Lengerich, E. J. (2008). Cancer registry policies in the United States and geographic information systems applications in comprehensive cancer control. Health Policy, 87(2), 185-193. doi: 10.1016/j.healthpol.2007.12.007
Glasgow, R. E., Vogt, T. M., & Boles, S. M. (1999). Evaluating the public health impact of health promotion interventions: the RE-AIM framework. American Journal of Public Health, 89(9), 13221327. doi: 10.2105/ajph.89.9.1322
Handler, A., Issel, M., & Turnock, B. (2001). A conceptual framework to measure performance of the public health system. American Journal of Public Health, 91(8), 1235-1239.
Hebert, L. E., Scherr, P. A., Bienias, J. L., Bennett, D. A., & Evans, D. A. (2003). Alzheimer Disease in the US Population: Prevalence Estimates Using the 2000 Census. Arch Neurol, 60(8), 1119-1122. doi: 10.1001/archneur.60.8.1119
Heer, J., Viegas, F. B., & Wattenberg, M. (2009). Voyagers and voyeurs: Supporting asynchronous collaborative visualization. Communications of the ACM, 52(1), 87-97. doi: 10.1145/1435417.1435439
Higgs, G. (2004). A Literature Review of the Use of GIS-Based Measures of Access to Health Care Services. Health Services and Outcomes Research Methodology, 5(2), 119-139. doi: 10.1007/s10742005-4304-7
Higgs, G. (2009). The role of GIS for health utilization studies: literature review. Health Services and Outcomes Research Methodology, 9(2), 84-99. doi: 10.1007/s 10742-009-0046-2
Horev, T., Pesis-Katz, I., & Mukamel, D. B. (2004). Trends in geographic disparities in allocation of health care resources in the US. Health Policy, 68(2), 223-232. doi: 10.1016/j.healthpol.2003.09.011
Koua, E. L., & Kraak, M. J. (2004). Geovisualization to support the exploration of large health and demographic survey data. International Journal of Health Geographics, 5(1), 1-13.
McLafferty, S. L. (2003). GIS AND HEALTH CARE. Annual Review of Public Health, 24(1), 25-42. doi: doi :10.1146/annurev.publhealth.24.012902.141012
Mitchell, A. (1999). The ESRI Guide to GIS Analysis: Volume 1: geographic patterns & relationships. Redlands, CA: ESRI Press.
Mitchell, A. (2005). The ESRI Guide to GIS Analysis: Volume 2: Spatial Measurements and Statistics. Redlands, CA: ESRI Press.
Moore, L. V., Diez Roux, A. V., Nettleton, J. A., & Jacobs, D. R. (2008). Associations of the Local Food Environment with Diet Quality--A Comparison of Assessments based on Surveys and Geographic Information Systems. American Journal of Epidemiology, 167(8), 917-924. doi: 10.1093/aje/kwm394
Muschert, G. W., Young-Spillers, M., & Carr, D. C. (2006). 'Smart' policy decisions to combat a social problem: the case of child abductions 2002-2003. Justice Policy Journal, 3(2), 1-32.
North Carolina Department of Crime Control & Public Safety. (2011). The Silver Alert Program. Retrieved May 1, 2013, from http://www.nccrimecontrol.org/index2.cfm?a=0000 03,000014,000081,001670
Nykiforuk, C. I. J., & Flaman, L. M. (2011). Geographic Information Systems (GIS) for Health Promotion and Public Health: A Review. Health Promotion Practice, 12(1), 63-73. doi: 10.1177/1524839909334624
Rosling, H. (2007). Visual technology unveils the beauty of statistics and swaps policy from dissemination to access. Statistical Journal of the IAOS: Journal of the International Association for Official Statistics, 24(1), 103-104.
Rosling, H., Rosling-Ronnlund, A., & Rosling, O. (2004). New software brings statistics beyond the eye. Paper presented at the OECD World Forum on Key Indicators, Palermo, Italy.
Sieber, R. (2006). Public Participation Geographic Information Systems: A Literature Review and Framework. Annals of the Association of American Geographers, 96(3), 491-507. doi: 10.1111/j.14678306.2006.00702.x
Sips, M., Schneidewind, J., & Keim, D. A. (2007). Highlighting space-time patterns: Effective visual encodings for interactive decision-making. International Journal of Geographical Information Science, 21(8), 879-894.
Tufte, E. R. (2001). The visual display of quantitative information (2 ed.). Cheshire, CT: Graphics Press.
U.S. Census Bureau. (2010). North Carolina. State & County QuickFacts.
Retrieved May 1, 2011, from http://quickfacts.census.gov/qfd/states/37000.html
U.S. Census Bureau. (2011). Older Americans Month: May 2011, Facts for Features.
Waller, L. A., & Gotway, C. A. (2004). Applied spatial statistics for public health data. Hoboken, NJ: Wiley-Interscience.
(3) By 2050, the number of Americans ages 65 and older is expected to be approximately 40 million, which is nearly 20 percent of the total population (U.S. Census Bureau, 2011)
(4) The number of adults with dementia and related diseases is projected to be approximately 13.2 million by 2050 an increase from the estimated 4.5 million in 2000 (Hebert, Scherr, Bienias, Bennett, & Evans, 2003)
University of Nevada
DAWN C. CARR
J. SCOTT BROWN
Miami University, Oxford, Ohio
Table 1 Descriptive Summary of Silver Alert (SA) Activated during 2008, 2009 and 2010 in North Carolina, USA Year 2008 2009 2010 Total Mean number of SA 1.3 2.4 2.2 5.8 activated Standard deviation 2.5 4.3 3.9 10.5 Minimum 0 0 0 0 Maximum 18 32 35 85 Count 128 239 220 587 Counties with no 52 43 44 26 Silver Alerts Note: North Carolina has 100 counties. Data source: North Carolina Department of Crime Control and Public Safety
|Printer friendly Cite/link Email Feedback|
|Author:||Yamashita, Takashi; Carr, Dawn C.; Brown, J. Scott|
|Publication:||Journal of Health and Human Services Administration|
|Date:||Dec 22, 2014|
|Previous Article:||An uncertain federalism: the states and the Affordable Care Act.|
|Next Article:||Serious games for health: An introduction to the symposium.|