Improving Pneumonia and Influenza Mortality Brandon Ramsey Florida Department of Health Heather Rubino Janet Hamilton David Atrubin Improving Pneumonia and Influenza Mortality Surveillance: Using Literal Cause of Death Fields to Improve Accuracy and Timeliness Mortality Data for Influenza Surveillance DOH uses pneumonia and influenza (P&I) mortality as an indicator for influenza severity and for situational awareness. International Classification of Disease (ICD- 10) codes are the most commonly used cause of death field. Underlying ICD codes process slowly (see chart below) and can change over time. DOH has access to the less commonly used literal cause of death (“literals”) fields via the state syndromic surveillance system (ESSENCE-FL), which are available 24-48 hours after death. “Literals” are causes of death that are physician-recorded at time of death (see comparison below). Research Question What are the benefits of using “literals” as opposed to underlying ICD code for mortality surveillance? *Misspelled words excluded from table Using the “literals” can increase the total number of pneumonia and influenza deaths identified. Although the “literals” query for pneumonia is likely too broad, queries can be created to limit noise while still capturing all deaths of interest. Creating specific queries makes it possible to include P&I-associated deaths even when they do not receive a P&I UCOD. The “literals” help explain what caused a death with an “unspecified UCOD.” Greater detail about deaths can improve analytic capacity to understand these events in terms of other conditions (see chart below). Real-World Epidemiologic Applications DOH uses a daily query of the literals to identify pediatric influenza deaths; often before ICD underlying cause of death is established. Early detection of pediatric mortality increases the chance of retrieving and testing samples. General P&I mortality is monitored weekly as a measure of influenza season severity. Background Methods Results Discussion CONCLUSIONS All death records from 2010 to 2014 (N=917,683) were obtained from Vital Statistics data using ESSENCE-FL. Queries for influenza and pneumonia death detection were created (see below); one for underlying ICD code and one for “literals.” The number of death records detected using the “literals” queries were compared to the number of deaths detected using the underlying cause of death ICD queries. Deaths detected using the “literals” query were examined further to determine which ICD code they most frequently obtained as an Underlying Cause of Death (UCOD). 77% of influenza deaths detected with a “literals” query had a matching underlying influenza ICD code. Top 5 ICD-10 Codes for Deaths with Influenza "Literals“ (Bold indicates “literals” and UCOD do not match) Underlying ICD-10 Code Deaths ICD-10 Code Type J110-With Pneumonia 168 Influenza J111-With Other Respiratory Manifest 90 J09-Influenza due to identified avian influenza virus 87 J100-With Pneumonia Influ Virus 64 J101-With Other Resp Manifest Influ Virus 46 Words within “Literals” of Influenza Deaths Queries Used in Analysis “Literals” Query for Influenza* UCOD Query for Influenza Terms Negations Influenza, H1N1 Haemophilus, Fluid, Flutter, Non-flu, Bacteremia, Flur, Conflu, Fluo J09, J10, J11 - 24% of pneumonia deaths detected with a “literals” query had a matching underlying pneumonia ICD code. Example Death Record: “Literals” and Underlying ICD Code "literal" 1 "literal“ 2 "literal" 3 "literal" 4 Underlying Cause Pneumonia Flu COPD None J110 (Influenza with Pneumonia) Top 5 ICD-10 Codes for Deaths with Pneumonia "Literals“ (Bold indicates “literals” and UCOD do not match) Underlying ICD-10 Code Deaths ICD-10 Code Type J189-Unspecified 10,413 Pneumonia J440-With Acute Lower Respiratory Infection 5,704 Not Pneumonia F03-Unspecified Dementia 2,833 C349-Unspecified 2,513 I251-Atherosclerotic Heart Disease 1,730 “Literals” Query for Pneumonia* UCOD Query for Pneumonia Terms Negations Pneumonia Aspir, Pneumonitis, Pneumococcal J12, J13, J14, J15, J16, J17, J18 -