Lisa Domegan1,2, Patricia Garvey3, Paul McKeown3, Howard Johnson4, Paul Hynds5,6, Jean O'Dwyer6,7,8, Coilín ÓhAiseadha9. 1. European Programme for Intervention Epidemiology Training (EPIET), European Centre for Disease Prevention and Control (ECDC), Stockholm, Sweden. lisa.domegan@hse.ie. 2. Health Service Executive-Health Protection Surveillance Centre, Dublin, Ireland. lisa.domegan@hse.ie. 3. Health Service Executive-Health Protection Surveillance Centre, Dublin, Ireland. 4. Health Service Executive-Health Intelligence Unit, Dublin, Ireland. 5. Environmental Sustainability & Health Institute, Technological University Dublin, Dublin, Ireland. 6. Irish Centre for Research in Applied Geosciences, University College Dublin, Dublin, Ireland. 7. School of Biological, Earth and Environmental Sciences, University College Cork, Cork, Ireland. 8. Water and Environment Research Group, Environmental Research Institute, University College Cork, Cork, Ireland. 9. Health Service Executive-Department of Public Health-East, Dublin, Ireland. coilin.ohaiseadha@hse.ie.
Abstract
BACKGROUND: Geocoding (the process of converting a text address into spatial data) quality may affect geospatial epidemiological study findings. No national standards for best geocoding practice exist in Ireland. Irish postcodes (Eircodes) are not routinely recorded for infectious disease notifications and > 35% of dwellings have non-unique addresses. This may result in incomplete geocoding and introduce systematic errors into studies. AIMS: This study aimed to develop a reliable and reproducible methodology to geocode cryptosporidiosis notifications to fine-resolution spatial units (Census 2016 Small Areas), to enhance data validity and completeness, thus improving geospatial epidemiological studies. METHODS: A protocol was devised to utilise geocoding tools developed by the Health Service Executive's Health Intelligence Unit. Geocoding employed finite-string automated and manual matching, undertaken sequentially in three additive phases. The protocol was applied to a cryptosporidiosis notification dataset (2008-2017) from Ireland's Computerised Infectious Disease Reporting System. Outputs were validated against devised criteria. RESULTS: Overall, 92.1% (4266/4633) of cases were successfully geocoded to one Small Area, and 95.5% (n = 4425) to larger spatial units. The proportion of records geocoded increased by 14% using the multiphase approach, with 5% of records re-assigned to a different spatial unit. CONCLUSIONS: The developed multiphase protocol improved the completeness and validity of geocoding, thus increasing the power of subsequent studies. The authors recommend capturing Eircodes ideally using application programming interface for infectious disease or other health-related datasets, for more efficient and reliable geocoding. Where Eircodes are not recorded/available, for best geocoding practice, we recommend this (or a similar) quality driven protocol.
BACKGROUND: Geocoding (the process of converting a text address into spatial data) quality may affect geospatial epidemiological study findings. No national standards for best geocoding practice exist in Ireland. Irish postcodes (Eircodes) are not routinely recorded for infectious disease notifications and > 35% of dwellings have non-unique addresses. This may result in incomplete geocoding and introduce systematic errors into studies. AIMS: This study aimed to develop a reliable and reproducible methodology to geocode cryptosporidiosis notifications to fine-resolution spatial units (Census 2016 Small Areas), to enhance data validity and completeness, thus improving geospatial epidemiological studies. METHODS: A protocol was devised to utilise geocoding tools developed by the Health Service Executive's Health Intelligence Unit. Geocoding employed finite-string automated and manual matching, undertaken sequentially in three additive phases. The protocol was applied to a cryptosporidiosis notification dataset (2008-2017) from Ireland's Computerised Infectious Disease Reporting System. Outputs were validated against devised criteria. RESULTS: Overall, 92.1% (4266/4633) of cases were successfully geocoded to one Small Area, and 95.5% (n = 4425) to larger spatial units. The proportion of records geocoded increased by 14% using the multiphase approach, with 5% of records re-assigned to a different spatial unit. CONCLUSIONS: The developed multiphase protocol improved the completeness and validity of geocoding, thus increasing the power of subsequent studies. The authors recommend capturing Eircodes ideally using application programming interface for infectious disease or other health-related datasets, for more efficient and reliable geocoding. Where Eircodes are not recorded/available, for best geocoding practice, we recommend this (or a similar) quality driven protocol.
Authors: Kate Zinszer; Christian Jauvin; Aman Verma; Lucie Bedard; Robert Allard; Kevin Schwartzman; Luc de Montigny; Katia Charland; David L Buckeridge Journal: Spat Spatiotemporal Epidemiol Date: 2010-03-20
Authors: Julia Krolik; Allison Maier; Gerald Evans; Paul Belanger; Geoffrey Hall; Alan Joyce; Anna Majury Journal: Geospat Health Date: 2013-11 Impact factor: 1.212
Authors: K G J Pollock; H E Ternent; D J Mellor; R M Chalmers; H V Smith; C N Ramsay; G T Innocent Journal: Zoonoses Public Health Date: 2010-12 Impact factor: 2.702
Authors: M Norman Oliver; Kevin A Matthews; Mir Siadaty; Fern R Hauck; Linda W Pickle Journal: Int J Health Geogr Date: 2005-11-10 Impact factor: 3.918
Authors: Yolanda J McDonald; Michael Schwind; Daniel W Goldberg; Amanda Lampley; Cosette M Wheeler Journal: Geospat Health Date: 2017-05-11 Impact factor: 1.212