BACKGROUND: The NDDoH receives mandatory reports of laboratory confirmed cases of influenza every year, but does not receive information on clinically diagnosed cases. The purpose of this evaluation was to better understand how well influenza reporting reflects the true burden of influenza in North Dakota.
METHODS: ICD-9 and ICD-10 codes for influenza were provided to a select healthcare facility (HFA) to aid them in the data request. Data from HFA for the same time period was extracted from MAVEN, North Dakota’s reportable disease database. Additionally, the Biosense 2.0 influenza-like illness (ILI) syndrome data for this facility was also gathered.
RESULTS: The total number of cases of influenza submitted by HFA were 440 (EMR requested line list). Seven hundred and 16 laboratory identified cases were submitted to MAVEN from this facility and 519 ILI visits were recorded in Biosense. Across all three data sets, females were more likely to visit a healthcare facility than males. Age distribution was similar across all three data sets and total cases reported each week showed the same trend. Emergency department visits made up 41.6% of all influenza encounters followed by office visits (40.7%). Additionally, emergency departments were most often used among those younger than 5 years and older than 55 years. Two percent of influenza cases reported by Healthcare Facility A included chief complaints or secondary diagnosises containing symptoms characteristic of a GI illness, not influenza. Additionally, 25% of ICD coded influenza cases in Biosense had chief complaints indicating a GI illnesses.
CONCLUSIONS: The original hypothesis assumed more cases would be identified via clinical diagnosis; however, this was not seen. Instead, more cases were found using the MAVEN data set. MAVEN reporting showed that 700 people were diagnosed with flu, but, the line list provided identified only 400 people. This indicates that not even those with a positive laboratory result were all being ICD coded as having influenza. Additionally, GI illness was important to include because patients often report “stomach flu” or clinicians refer to GI illness as a “flu-like”. Both HFA and Biosense data were affected by this type of terminology, indicating that non-laboratory data is likely to contain some GI interference. The analysis of this project suggests that using data pulled from electronic medical records based on ICD coding alone may not be as accurate as originally anticipated. It also illustrates the importance of collecting chief complaints in addition to diagnostic codes.