Presentation Type
Poster Presentation
Abstract/Artist Statement
Introduction. North Dakota collects syndromic surveillance data, de-identified reason-for-visit data, for all visits to participating emergency departments (EDs), walk in clinics, and ambulatory clinics throughout the state. Syndromic data is submitted daily and automatically through an electronic connection between health care providers and the state. Animal bites are not required to be specifically reported to the North Dakota Department of Health (NDDoH) by providers; however, NDDoH has utilized North Dakota’s syndromic surveillance program, BioSense 2.0, for routine surveillance by using an existing animal bites syndrome. We sought to evaluate the existing BioSense 2.0 animal bites syndrome and explore how the syndrome definition could be refined to enhance surveillance capabilities.
Methods. A table summarizing the findings from “animal bite” visits from July 2014 through June 2015 was created from BioSense 2.0 data using specific clinic assigned diagnosis codes to pull the data. Health Department staff reviewed each entry and coded it as animal bite, not animal bite, or unknown. Visits that were identified as animal bite related were further categorized into bite type. Animal bite types consisted of: dog, cat, and other. Words that contributed to correct data related to animal bites or incorrect data not related to animal bites were noted and applied in a new syndrome definition. The new syndrome was applied, and changes between the original and resulting data sets were noted.
Results. The table created from BioSense 2.0 data included 1,516 unique visits, representing 953 ED visits and 563 outpatient visits. More than half (66%) of all syndrome visits were not related to animal bites; partially due to the associated diagnose code, which also covers other animal related injuries and insect bites. Although the BioSense 2.0 syndrome definition for animal bites excludes the term “insect” from chief complaint, 40% of our animal bite syndrome data related to insect bites. For 95% of insect bites, a diagnosis code or diagnosis terms for insect bites was triggering the inclusion. The remaining 5% of visits came from the chief complaint. While the word insect was not used, the name of the insect (spider, mosquito, tick, etc.) or the term arthropod was used instead. Based on these results, we changed the diagnosis codes used in the new definition to be specific to animal bites only. Other common errors identified in our analysis: “bite” pulling human bites, dental disorders, bites of food, and tongue bites; “pica” pulled pic line related visits; and “cat” brought in visits referencing cat scans. We also identified many visits referencing people falling while walking their dog, service dogs, and allergic reactions to dogs and cats. We added many insect names as well as words pertaining to falls, food, allergies and scans to the chief complaint exclusion list. Applying these changes identified in our analysis reduced the total number of visits in our animal bites syndrome to 691. The number of animal bites remained the same as the original number of animal bites identified by the BioSense data (n=513, 74%), with 28 undetermined and 150 unrelated visits remaining, an 81% reduction in misclassification. The remaining visits unrelated to actual animal bites referenced dogs and cats. However, we found these terms could not be removed from the search list because these visits often lacked diagnosis codes. Non-bite animal injury codes used were not consistent enough to exclude from the syndrome.
Discussion. Thirty percent of the animal bites identified came from outpatient visit data, demonstrating use for outpatient syndromic surveillance data in tracking animal bite trends. We found many unrelated visits for human and insect bites and animal injuries in our data. We were able to remove all insect and human bites, and some non-bite injuries using our own syndrome. Better diagnosis code data may allow “dog” and “cat” to be removed from chief complaint searches, reducing the number of instances that were misclassified due to non-bite injuries that included reports of these animals. The new syndrome definition will enhance animal bite surveillance at the NDDoH.
Mentor Name
Kari Harris
Taking a Bite out of Unwanted Data in the Animal Bite Syndrome
UC Ballroom (Center)
Introduction. North Dakota collects syndromic surveillance data, de-identified reason-for-visit data, for all visits to participating emergency departments (EDs), walk in clinics, and ambulatory clinics throughout the state. Syndromic data is submitted daily and automatically through an electronic connection between health care providers and the state. Animal bites are not required to be specifically reported to the North Dakota Department of Health (NDDoH) by providers; however, NDDoH has utilized North Dakota’s syndromic surveillance program, BioSense 2.0, for routine surveillance by using an existing animal bites syndrome. We sought to evaluate the existing BioSense 2.0 animal bites syndrome and explore how the syndrome definition could be refined to enhance surveillance capabilities.
Methods. A table summarizing the findings from “animal bite” visits from July 2014 through June 2015 was created from BioSense 2.0 data using specific clinic assigned diagnosis codes to pull the data. Health Department staff reviewed each entry and coded it as animal bite, not animal bite, or unknown. Visits that were identified as animal bite related were further categorized into bite type. Animal bite types consisted of: dog, cat, and other. Words that contributed to correct data related to animal bites or incorrect data not related to animal bites were noted and applied in a new syndrome definition. The new syndrome was applied, and changes between the original and resulting data sets were noted.
Results. The table created from BioSense 2.0 data included 1,516 unique visits, representing 953 ED visits and 563 outpatient visits. More than half (66%) of all syndrome visits were not related to animal bites; partially due to the associated diagnose code, which also covers other animal related injuries and insect bites. Although the BioSense 2.0 syndrome definition for animal bites excludes the term “insect” from chief complaint, 40% of our animal bite syndrome data related to insect bites. For 95% of insect bites, a diagnosis code or diagnosis terms for insect bites was triggering the inclusion. The remaining 5% of visits came from the chief complaint. While the word insect was not used, the name of the insect (spider, mosquito, tick, etc.) or the term arthropod was used instead. Based on these results, we changed the diagnosis codes used in the new definition to be specific to animal bites only. Other common errors identified in our analysis: “bite” pulling human bites, dental disorders, bites of food, and tongue bites; “pica” pulled pic line related visits; and “cat” brought in visits referencing cat scans. We also identified many visits referencing people falling while walking their dog, service dogs, and allergic reactions to dogs and cats. We added many insect names as well as words pertaining to falls, food, allergies and scans to the chief complaint exclusion list. Applying these changes identified in our analysis reduced the total number of visits in our animal bites syndrome to 691. The number of animal bites remained the same as the original number of animal bites identified by the BioSense data (n=513, 74%), with 28 undetermined and 150 unrelated visits remaining, an 81% reduction in misclassification. The remaining visits unrelated to actual animal bites referenced dogs and cats. However, we found these terms could not be removed from the search list because these visits often lacked diagnosis codes. Non-bite animal injury codes used were not consistent enough to exclude from the syndrome.
Discussion. Thirty percent of the animal bites identified came from outpatient visit data, demonstrating use for outpatient syndromic surveillance data in tracking animal bite trends. We found many unrelated visits for human and insect bites and animal injuries in our data. We were able to remove all insect and human bites, and some non-bite injuries using our own syndrome. Better diagnosis code data may allow “dog” and “cat” to be removed from chief complaint searches, reducing the number of instances that were misclassified due to non-bite injuries that included reports of these animals. The new syndrome definition will enhance animal bite surveillance at the NDDoH.