Open Access

Utilization of Google enterprise tools to georeference survey data among hard-to-reach groups: strategic application in international settings

  • Leo Beletsky1, 2Email author,
  • Jaime Arredondo2,
  • Dan Werb2,
  • Alicia Vera2, 3, 4,
  • Daniela Abramovitz2,
  • Joseph J. Amon5,
  • Kimberly C. Brouwer2,
  • Steffanie A. Strathdee2 and
  • Tommi L. Gaines2
International Journal of Health Geographics201615:24

https://doi.org/10.1186/s12942-016-0053-9

Received: 8 January 2016

Accepted: 13 July 2016

Published: 28 July 2016

Abstract

Background

As geospatial data have become increasingly integral to health and human rights research, their collection using formal address designations or paper maps has been complicated by numerous factors, including poor cartographic literacy, nomenclature imprecision, and human error. As part of a longitudinal study of people who inject drugs in Tijuana, Mexico, respondents were prompted to georeference specific experiences.

Results

At baseline, only about one third of the 737 participants were native to Tijuana, underscoring prevalence of migration/deportation experience. Areas frequented typically represented locations with no street address (e.g. informal encampments). Through web-based cartographic technology and participatory mapping, this study was able to overcome the use of vernacular names and difficulties mapping liminal spaces in generating georeferenced data points that were subsequently analyzed in other research.

Conclusion

Integrating low-threshold virtual navigation as part of data collection can enhance investigations of mobile populations, informal settlements, and other locations in research into structural production of health at low- or no cost. However, further research into user experience is warranted.

Keywords

Resource-poor settings Data collection Innovation Vulnerable groups Liminal spaces Technology Tools

Background

Dating back to John Snow’s maps of cholera cases in London, the collection of geospatial data is foundational to epidemiology and public health research. In recent decades, however, the advent of geographic information systems (GIS) technology and methodological tools has enabled remarkable advances in the scope, accuracy and power of multilevel analyses of such datasets. This has coincided with the development of novel applications of geospatial analytic methods, including those used to better target programmatic interventions to key populations and enriching our understanding of the relationship between health and our social, cultural, and physical environment, with particular focus on concentrated disadvantage, conflict, and other political and structural determinants of health [1, 2]. Examples of these applications include the use of the built environment to understand obesity, water access and quality research, modeling the diffusion of disease among displaced populations, geolocating adverse law enforcement encounters among drug users to identify service barriers, and a variety of GIS applications to inform supply chains and service delivery [211]. More broadly, historic population displacement, mounting globalization, and increased recognition of the interplay of biological, environmental, and structural factors in production of health underscore the significance of georeference data in public health and human rights research.

Despite these opportunities, there remain important challenges in collecting reliable geospatial information, especially in research on mobile and marginalized populations. In survey research, recall of physical address or cross street information may be poor, especially among itinerant, unstably housed, and lower literacy respondents. For instance, in locales of high mobility such as transit hubs, border areas, and refugee havens, survey respondents may be unfamiliar with place names or designations when asked to site particular experiences that occurred within that location. Lack of systematic nomenclature for addresses and street names in certain settings, especially in middle- and lower-income countries hinders precision in georeferenced data collection. Pervasive use of liminal spaces for residential, commercial, and other activity by under-served or criminalized groups (e.g. railroad tracks, canals, informal settlements, etc.) further complicates investigations of the environmental factors shaping their health.

We have experienced many of these challenges first-hand in the context of our research among drug users, sex workers, migrants, and other marginalized and stigmatized groups in the Global South and elsewhere [2, 1218]. For instance, our research assessing health and human rights domains among people who inject drugs (PWID) along the US–Mexico border typically samples substantial numbers of migrants and deportees who are relatively new to the locales of research [15]. To collect georeferenced data, prior studies relied on paper maps during the interview process to approximate specific locations [7, 15]. As a result, our field staff would expend considerable up-front effort orienting respondents to the map of the locale (as well as—at times—basic cartographic conventions), noting the identified locations using cross-street designation, and later identifying geo-coordinates and transferring these data to the survey database. This laborious process was hampered by limited geographical literacy, as well as open to numerous sources of human error and map imprecision.

The recent advent of free cloud-based mapping tools using satellite and street-view data has proven to be an asset in a variety of research activities, including those in resource-poor settings. In rural sub-Saharan Africa, Google Earth tools have been used to develop a spatial sampling frame to inform subsequent recruitment into a longitudinal survey [9]. In rural Haiti, a combination of Google Earth and GIS software enabled the mapping and random selection of households for water sampling and ethnographic surveys [10]. Among substance users, including PWID, cloud-based mapping tools have been adopted to examine the local areas in which individuals routinely travel and where their daily activities typically occur; to generate geographic coordinates of respondents’ activity spaces, however, these studies relied on a participant’s ability to provide a physical address or nearest street intersection [19, 20].

To our knowledge, no prior published research had employed web mapping technologies as part of the actual data collection process. In an effort to streamline and improve field-based data collection, we developed a novel methodology for the application of Google enterprise tools as part of a larger inquiry into the role of law and law enforcement in shaping infectious disease risk among PWID in Mexico [21]. Thus, the objective of this study is to describe the development and deployment of online mapping technologies in survey research targeting hard-to reach, vulnerable individuals.

Methods

Target population

The target population was PWID in Tijuana, recruited as part of a mixed-methods longitudinal study. Study rationale, recruitment, and analytical methods have been detailed elsewhere [5, 22, 23]. Structured interviews identifying the physical spaces in which individuals experienced law enforcement encounters were conducted with those who agreed to participate. The study was approved by the Institutional Review Board at UCSD School of Medicine and Collegio de la Frontera Norte, Tijuana (Project Number 141109).

Web-based data georeferencing technique

Our quantitative survey instrument assessed sociodemographics, sexual and drug use risk behaviors, migration history, knowledge of criminal laws, and police encounter history. For items related to recent police detention and abuse, we assessed the physical location of the last reported encounter. The instrument was administered in English or Spanish by trained, bilingual interviewers using computer-assisted interview software (QDS™ Systems, NOVA Research, Bethesda, USA).

Our methodological innovation was to integrate Google Enterprise tools including Google Earth and Google Street View into the structured interview protocol. Our laptop workstations running the QDS interview software on a Windows (Microsoft Corporation, Seattle, USA) platform were utilized to assist with georeferencing during data collection. Specifically, when asked for physical location information linked to a particular event (e.g. last instance of physical altercation with a police officer), respondents were able to virtually navigate and pinpoint the location using the integrated Google Street View and Google Earth cloud tools. During this process, the interviewer invited the participant to describe the general area of the encounter, then working with the participant to zoom in and identify the precise location based on narrative description of landmarks and visual anchors. Initially, once the specific location was pinned, interviewers entered the resulting geocoordinates in appropriate data field in the QDS interview database. As the study progressed, we created a software tool that directly transmitted geocoordinate data from the Google Streetview pin to the appropriate field in the QDS database, eliminating the need for human data entry. Paper maps were used as a backup in the 3–5 % of the cases when the Internet connection failed.

Applications

Our research team conducted multiple geospatial analyses using the georeferenced data collected under the aforementioned protocol. For instance, we triangulated spatial data collected through the innovative technique described above with the Mexican Census to identify concentrated areas of police activity and modeled this relationship to further understand the structural determinants of HIV, hepatitis, and other disease risk among PWID [23]. We also employed geographic weighted regression techniques to determine the spatial association between addiction treatment center locations and the spatial pattern of police interactions with PWID [5]. In addition, we triangulated PWID geospatial data with official crime statistics to identify places of high-risk drug-related activity [24].

Results and discussion

Our sample covered 737 PWID at baseline, who were 61.9 % male with a median 8 years of education and 50.6 % reporting monthly income equal or less than 2500 pesos (proximately 200 US dollars). Only 37.2 % reported being native to Tijuana, underscoring the prevalence of migrants and deportees in the sample. Our sample reported a median 16 years of injection drug use and 74 % prevalence of incarceration over their lifetime. These characteristics signal high prevalence of mobility, vulnerability and marginalization in the sample, underscoring the instrumental value of the data collection technique described here.

To our knowledge, this study is the first to use cost-free Google enterprise tools to assist participants in identifying geo-coordinates during field data collection. This method provides a low-cost alternative to the modal paper-based georeferencing that poses many challenges, especially among highly mobile and marginalized populations. Unique features of the street navigation and satellite imagery enabled an interactive experience whereby the respondents were able to base their responses on particular geographical features or landmarks without requiring the respondent to be versed in formal nomenclature or the interviewer to have personal familiarity with the full range of possible locales.

The application of these tools proved to be feasible and efficient in terms of minimizing logistical barriers to field collection of georeferenced data, eliminating the cumbersome, labor-intensive, and error-prone utilization of paper maps. Many respondents displayed a high level of engagement and interest in the technology used in this data collection technique. Utilization of the participatory virtual navigation using Google Street view proved almost universally intuitive, even for the many respondents who were not initially familiar with Google Street View or Google Earth tools. This, in contrast to the substantial efforts required to orient many respondents with cartographic conventions during prior studies using paper maps. Further research into user experiences with data collection through participatory victual navigation, including level of coaching necessary, impact of technological literacy, and other elements is warranted.

Through web-based mapping, the research team was able to create a rich dataset of georeferenced encounter points that were subsequently applied in analytical research [5, 23, 24]. Given that these and other, open-sourced mapping tools are freely available on the Internet, this could be a significant methodological innovation, particularly for research in low- and middle-income settings, among itinerant or unstably housed populations, or in liminal spaces. More recent advent of downloadable mapping and navigation applications that can be used off-line (including Google Earth) further extends the utility and promise of this technique, especially in rural and other areas with inadequate or non-existent mobile data or Internet service. At a time of major population mobility and displacement, growth in informal settlements, and increased interest in structural and rights-based determinants of global health, the integration of this technique can serve as a key innovation to streamlining and improving the data collection process.

There are important ethical considerations related to data security and the level of precision of geolocated results. We did not collect identifiable data and no data were actually stored with Google; the mapping software was utilized only to pinpoint the geocoordinates, which were then entered into a separate interview database. Therefore, we did not have privacy concerns about using a publicly-available private enterprise tool for research purposes. However, concerns have been raised that mapping locations of drug use (or sex work, or other clandestine activity by hard-to-reach populations) can provide insight into hotspots of illicit activity [25]. This information can be of interest to law enforcement, who may put pressure on researchers and study participants at risk. Meanwhile, identifiable reports of police abuse can create the risk of police retaliation. Therefore, the protection of both individual as well as community-level data must be considered both when designing and storing datasets containing sensitive georeferenced or other data.

Limitations included human error in recording geocoordinates, difficulty validating recall accuracy, and some challenges familiarizing participants with technological tools. The overall prevalence of missing data for georeferenced items (instances where the respondent reported the police encounter in question, but the geographical information was not provided) was marginally higher than that for other kinds of data in our survey. Across the set of georeferenced items, 17.3–22.8 % respondents answered “do not know” and 1.3–8.8 % refused to answer. Instances where the data was missing for unknown reasons ranged between 2.7 and 3.8 %. Given the sensitivity of the experiences where geo-coordinate information was sought (e.g. instances of physical abuse by police), deploying novel technologies to document precise locations of the encounters may have engendered concerns about possible negative repercussions among some respondents. Whether utilizing geo-referencing technology may impact recall in data collection relating to sensitive or extra-legal behavior or experiences warrants further study. To reduce human error, in subsequent study application of this tool, we have developed a software plug-in that transmits geocoordinate information directly to QDS, minimizing the risk of human data entry error. Open-source tools such as OpenStreetMap are available as alternatives to the use of proprietary Google Enterprise tools described here.

Conclusion

In this study, using Google Enterprise tools to identify and pinpoint geo-coordinates for survey responses among a sample of PWID proved feasible and efficient. Using this methodology, we were able to operationalize spatial surveillance of the structural determinants of HIV among hard-to-reach populations, making it possible to better target structural public health interventions. Future research should include validation and reliability analyses, cost-effectiveness, and qualitative research drawing on these geospatial data to assess user experiences, respondent confidentiality concerns, and to construct geonarratives.

Declarations

Authors’ contributions

LB, SS, and KB conceptualized the study. AV supervised protocol design and implementation, with input from DA, LB, SS, DW, TG and KB. DW and TG led geospatial analyses, with input from LB, SS, and KB. JA contributed to the interpretation of findings and formulation of international implications. LB drafted the manuscript. All authors read and approved the final manuscript.

Acknowledgements

This study was supported by the National Institute of Drug Abuse (NIDA), National Institutes of Health under awards R01DA039073 (L. Beletsky and S. Strathdee: MPIs) and R37 DA019829 (S. Strathdee: PI). Werb is also supported by a NIDA Avenir Award (DP2 DA040256-01) and the Canadian Institutes of Health Research (MOP–79297). Gaines is supported, in part by NIDA award K01DA034523. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health or the Canadian Institutes of Health Research. The sponsors had no role in study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.

Competing interests

The authors declare that they have no competing interests.

Availability of data and supporting materials

All data and software tools described in this manuscript are available to any scientist wishing to use them to replicate the methods described, without breaching participant confidentiality and for testing by reviewers in a way that preserves the reviewers’ anonymity.

Ethics approval and consent to participate

The study was approved by the Institutional Review Board at UCSD School of Medicine and Collegio de la Frontera Norte, Tijuana (Protocol Number 141109).

Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.

Authors’ Affiliations

(1)
Northeastern University School of Law and Bouvé College of Health Sciences
(2)
Division of Global Public Health, University of California, San Diego, School of Medicine
(3)
Universidad Xochicalco, School of Medicine
(4)
Universidad Autonoma de Baja California
(5)
Health and Human Rights Division, Human Rights Watch

References

  1. Amon JJ. The political epidemiology of HIV. JAIDS. 2014;17:19327.Google Scholar
  2. Beletsky L, Cochrane J, Sawyer A, et al. Police encounters among needle exchange clients in Baltimore: drug law enforcement as a structural determinant of health. Am J Public Health. 2015;105(9):1872–9.View ArticlePubMedGoogle Scholar
  3. Aral SO, Torrone E, Bernstein K. Geographical targeting to improve progression through the sexually transmitted infection/HIV treatment continua in different populations. Curr Opin HIV AIDS. 2015;10(6):477–82.View ArticlePubMedGoogle Scholar
  4. Lyons J. Documenting Violations of international humanitarian law from space: a critical review of geospatial analysis of satellite imagery during armed conflicts in Gaza (2009), Georgia (2008), and Sri Lanka (2009). Int Rev Red Cross. 2012;94(886):739–63.View ArticleGoogle Scholar
  5. Werb D, Strathdee S, Vera A, et al. Spatial Patterns of arrests, police brutality, and addiction treatment center locations in Tijuana, Mexico. Addiction. 2016;111(7):1246–56. doi:10.1111/add.13350.
  6. Cooper HLF, Bossak B, Tempalski B, Jarlais DCD, Friedman SR. Geographic approaches to quantifying the risk environment: drug-related law enforcement and access to syringe exchange programmes. Inte J Drug Policy. 2009;20(3):217–26.View ArticleGoogle Scholar
  7. Brouwer KC. Spatial epidemiology of HIV among injection drug users in Tijuana, Mexico. Ann Assoc Am Geogr. 2012;102(S1):1190–9.View ArticlePubMedPubMed CentralGoogle Scholar
  8. Blacksher E, Lovasi G. Place-focused physical activity research, human agency, and social justice in public health: taking agency seriously in studies of the built environment. Health Place. 2012;18(2):172–9.View ArticlePubMedGoogle Scholar
  9. Pearson AL, Rzotkiewicz A, Zwickle A. Using remote, spatial techniques to select a random household sample in a dispersed, semi-nomadic pastoral community: utility for a longitudinal health and demographic surveillance system. Int J of Health Geogr. 2015;14(1):1–10.View ArticleGoogle Scholar
  10. Wampler PJ, Rediske RR, Molla AR. Using ArcMap, Google Earth, and Global Positioning Systems to select and locate random households in rural Haiti. Int J Health Geogr. 2013;12:3. doi:10.1186/1476-072X-12-3.
  11. Thornton LE, Pearce JR, Kavanagh AM. Using Geographic Information Systems (GIS) to assess the role of the built environment in influencing obesity: a glossary. Int J Behav Nutr Phys Act. 2011;8(1):1–9.View ArticleGoogle Scholar
  12. Brouwer K, Weeks J, Lozada R, Strathdee SA. Integrating GIS into the study of contextual factors affecting injection drug use along the Mexico/U.S. border. In: Thomas Y, Richardson D, Cheung I, editors. Geography and drug addiction. New York: Springer; 2008. p. 27–42.View ArticleGoogle Scholar
  13. Beletsky L, Grau LE, White E, Bowman S, Heimer R. The roles of law, client race, and program visibility in shaping police interference with the operation of US syringe exchange programs. Addiction. 2011;106(2):357–65.View ArticlePubMedGoogle Scholar
  14. Beletsky L, Heller D, Jenness S, Neaigus A, Gelpi-Acostae C, Hagan H. Syringe access, syringe sharing, and police encounters among people who inject drugs in New York City: a community-level perspective. Int J Drug Policy. 2014;25(1):105–11.View ArticlePubMedGoogle Scholar
  15. Beletsky L, Martinez G, Gaines T, et al. Mexico’s northern border conflict: collateral damage to health and human rights of vulnerable groups. Pan Am J Public Health. 2012;31(5):403–10.View ArticleGoogle Scholar
  16. Beletsky L, Thomas R, Shumskaya N, Artamonova I, Smelyanskaya MS. Police education as a component of a national HIV response: lessons from Kyrgyzstan. Drug Alcohol Depend. 2013;132S:S48–52.View ArticleGoogle Scholar
  17. Strathdee SA, Lozada R, Martinez G, et al. Social and structural factors associated with HIV infection among female sex workers who inject drugs in the Mexico-US border region. PLoS One. 2011;6(4)e19048. doi:10.1371/journal.pone.0019048.
  18. Strathdee SA, Lozada R, Ojeda VD, et al. Differential effects of migration and deportation on HIV infection among male and female injection drug users in Tijuana, Mexico. PLoS One. 2008;3(7):e2690.View ArticlePubMedPubMed CentralGoogle Scholar
  19. Martinez AN, Lorvick J, Kral AH. Activity spaces among injection drug users in San Francisco. Int J Drug Policy. 2014;25(3):516–24.View ArticlePubMedGoogle Scholar
  20. Gibson C, Perley L, Bailey J, Barbour R, Kershaw T. Social network and census-tract level influences on substance use among emerging adult males: an activity spaces approach. Health Place. 2015;35:28–36.View ArticlePubMedGoogle Scholar
  21. Strathdee SA, Arredondo J, Rocha T, et al. Implementation of a police education program to integrate occupational safety and HIV prevention: protocol for a modified stepped-wedge study design with parallel prospective cohorts to assess behavioral outcomes. BMJ Open. 2015;5:e008958.View ArticlePubMedPubMed CentralGoogle Scholar
  22. Beletsky L, Wagner KD, Arredondo J, Palinkas L, Magis Rodríguez C, Strathdee SA. Implementing Mexico’s “Narcomenudeo” drug law reform: a mixed-methods assessment of early experiences among people who inject drugs. J Mixed Methods Res. 2015.Google Scholar
  23. Gaines T, Beletsky L, et al. Examining the spatial distribution of law enforcement encounters among people who inject drugs after implementation of Mexico’s drug policy reform. J Urban Health. 2015;92(2):338–51.View ArticlePubMedGoogle Scholar
  24. Gaines T, Werb D, Arredondo J, Alaniz V, Vilalta C, Beletsky L. The spatial-temporal pattern of policing following a drug policy reform: triangulating self-reported arrests with official crime statistics. Subst Use Misuse (in press).Google Scholar
  25. Odek WO. Estimating the size of the female sex worker population in Kenya to inform HIV prevention programming. PLoS One. 2014;9(3):e8918.View ArticleGoogle Scholar

Copyright

© The Author(s) 2016