8
• Whenever possible, RHINO encourages you to collaborate with hospitals and clinics. They may have
additional context regarding trends and workflows, which could enhance your analysis. If you do not
have a contact for the facilities in your data, RHINO can help facilitate a connection.
• Consider alternative explanations for the trends you observe. Consult with subject matter experts and
the literature on the health issue to see if your data align with expected trends.
• Know what is normal for your data quality.
o Know the formats of diagnoses. Do they provide one diagnosis or multiple? Do they include
the decimal point in their ICD-10 codes?
o Know the formats of chief complaints. Do your facilities report a single term, standardized
terms, or free text?
o Which optional data elements do your facilities report (e.g., triage notes, procedure codes,
clinical impression)? How complete are they?
• Check that your syndrome definitions and queries are appropriately defined for the question you
would like to answer. Invite collaboration with colleagues.
• Know which of your facilities are sending production-quality data and when they started sending
data. Watch for new facilities which can change visit volumes if you are querying based on counts.
• Know which kinds of facilities you have (e.g., emergency department, inpatient, or outpatient –
urgent care, primary, and specialty care).
• Know the reporting patterns of your data. Do facilities send their visits every hour or every 24 hours?
Weekly counts may give you a more stable picture than daily counts because of reporting procedures.
Remember counts from the most recent weeks may not yet be complete.
• Use counts and percentages. After you query, check that counts are the expected magnitude and
have not changed dramatically. If counts are much higher or lower than expected, you may need to
modify your query parameters. As a result of this potential variability, consider using percentages
instead of counts as they can provide more stable trend information.
• Establish and maintain relationships with your facilities. Knowing your data providers will increase the
likelihood both that you are informed of potential changes in the data (e.g., data drop-offs,
implementation of pick lists) and of successful collaborations during an outbreak.
o Let your facilities know you use and value their data!
• View RHINO data as a tool in your public health surveillance and preparedness toolbox, rather than as
a standalone surveillance system.
o Healthcare encounter data are not cleaned or curated. Data is automatically received directly
from the electronic medical record (EMR) and, consequently, can be noisy or occasionally
lead to inaccurate conclusions.
• RHINO data is appropriate for:
o Generating hypotheses,
o Strengthening information gathered from other sources,
o Investigating rumors or interventions, and
o Conducting preliminary assessments of the health effects of an emergency.
Clinical Data Limitations
• Data drop offs can occur for brief periods (1-2 days) and occasionally for longer (weeks to months).
These are likely the result of electronic medical record system migrations and data gaps can often be
filled in with time.