PHLT 8076 Week 5 Map Layer Provenance Audit Example

Six Layers, Six Publishers, One Mismatch That Matters: A Provenance Audit of the Data Under a County Map of Late-Stage Breast Cancer and Mammography Access

Student Name

Walden University

PHLT 8076: Fundamentals of Mapping and Geographic Information Systems for Public Health

Instructor Name

Month Day, Year

Six Layers, Six Publishers, One Mismatch That Matters: A Provenance Audit of the Data Under a County Map of Late-Stage Breast Cancer and Mammography Access

Why Audit the Layers

The composite county's map of late-stage breast cancer rests on six layers, each produced by a different organization for its own purpose, at its own time and for its own unit. A map drawn from them looks like one smooth picture, but it is only as consistent as its least compatible layer. This audit records, for each layer, who published it, for what purpose, what period it describes, what unit it uses and how it was obtained. It then reports the one mismatch that changes what the map can claim, along with two smaller ones that require adjustment.

The Audit Table

Table 1

Provenance of the Six Layers (Composite Details)

Note. Composite details for a teaching example. The social vulnerability index follows the method described by Flanagan et al. (2011).

Layer by Layer

Tract Boundaries

The 2020 tract boundaries define the map's units. Several tracts in the county were split or redrawn between the 2010 and 2020 censuses, which matters for the registry layer below.

Cancer Diagnoses

The registry geocodes each patient's address at diagnosis to a point. Geocoding quality varies: rural route and post office box addresses more often land on a postal-area center point or fail to match, and those failures concentrate in rural areas (Zandbergen, 2009). The registry reports that 94% of addresses in the county matched to a street location; the remainder will be listed by tract of best match and examined for clustering in the rural south. Because the registry assigned 2018 and 2019 diagnoses to 2010 tracts and later diagnoses to 2020 tracts, points rather than tract codes will be used, so every diagnosis can be placed in a 2020 tract consistently.

Population Denominators

The American Community Survey's five-year estimates for 2018 to 2022 match the diagnosis period and use 2020 tracts. They are estimates with margins of error, which are large for small tracts; the margins will be reported for tracts with few women.

Social Vulnerability Index

The index combines census variables on poverty, housing, language and other factors into a tract-level score intended for emergency planning (Flanagan et al., 2011). It is used here as context, since late-stage diagnosis is linked to poverty and insurance more than to distance in large registry studies (Henry et al., 2011), but its inputs were chosen for disaster vulnerability, not cancer screening, and the map will say so.

Road Network

The 2023 road network is used to compute travel time from each tract's population-weighted center to the nearest mammography facility. Roads changed little in the county over the study period, so this layer's vintage causes no concern.

The Mismatch That Matters

The mammography facility list describes facilities certified on the day it was downloaded in 2024. The diagnoses describe women diagnosed from 2018 through 2022. Two facilities in the rural south closed during that period, one in 2019 and one in 2021, and a new facility opened in the county seat in 2023. A travel-time layer built from the current list would show women in the south as farther from screening than they were for much of the period and women near the county seat as closer than they were. Because the map's question concerns whether late-stage diagnosis follows distance to screening, this mismatch goes to the heart of its claim. The remedy is to rebuild the facility layer for each year of the period from archived certification records and the state health department's licensing history, then compute each woman's travel time to the facilities open in the year before her diagnosis. Wang and Luo (2005) emphasize that access measures must reflect the supply actually available to a population, and a current list does not.

Two Smaller Adjustments

The 2010 to 2020 tract change is handled by using geocoded points. The margins of error in small-tract denominators are handled by reporting them and by flagging tracts with few women alongside tracts with few diagnoses.

Recording Provenance for Future Users

The audit table will be published with the map as a data appendix, together with the date each layer was obtained, any processing steps and the version of the facility list used for each year. Future analysts updating the map will then know which layers need refreshing and which mismatches to watch for. Elliott and Wartenberg (2004) emphasize that spatial analyses depend on data quality that readers rarely see, and a provenance appendix makes that quality visible. It also protects the program if the map is challenged: anyone questioning a pattern can trace each layer to its source. The appendix will also note which layers carry known error, the geocoding gaps in the rural south and the margins of error in small-tract denominators, so future users do not mistake a fragile pattern for a firm one. Keeping that record current takes little time each year, and it prevents the next analyst from repeating the facility-list mistake this audit caught.

What the Audit Changes

Without the audit, the map would have compared diagnoses from 2018 to 2022 with facilities from 2024 and might have reported a link between distance and late stage driven partly by closures that happened after many of the diagnoses. With the facility layer rebuilt by year, the map can make the claim it was designed to make, or show that the claim does not hold.

References

Elliott, P., & Wartenberg, D. (2004). Spatial epidemiology: Current approaches and future challenges. Environmental Health Perspectives, 112(9), 998-1006. https://doi.org/10.1289/ehp.6735

Flanagan, B. E., Gregory, E. W., Hallisey, E. J., Heitgerd, J. L., & Lewis, B. (2011). A social vulnerability index for disaster management. Journal of Homeland Security and Emergency Management, 8(1), Article 3. https://doi.org/10.2202/1547-7355.1792

Henry, K. A., Boscoe, F. P., Johnson, C. J., Goldberg, D. W., Sherman, R., & Cockburn, M. (2011). Breast cancer stage at diagnosis: Is travel time important? Journal of Community Health, 36(6), 933-942. https://doi.org/10.1007/s10900-011-9392-4

Wang, F., & Luo, W. (2005). Assessing spatial and nonspatial factors for healthcare access: Towards an integrated approach to defining health professional shortage areas. Health & Place, 11(2), 131-146. https://doi.org/10.1016/j.healthplace.2004.02.003

Zandbergen, P. A. (2009). Geocoding quality and implications for spatial analysis. Geography Compass, 3(2), 647-680. https://doi.org/10.1111/j.1749-8198.2008.00205.x