Determining accuracy of ICD-10 influenza diagnosis codes in ED, inpatient settings

Determining accuracy of ICD-10 influenza diagnosis codes in ED, inpatient settings | Image Credit: © Rido - © Rido - stock.adobe.com.

Coded diagnosis of pediatric influenza plays a role in the study of the disease in children, according to authors of a recent study published in JAMA Network Open. However, the ability of International Classification of Diseases, Ninth Revision codes to identify flu in the emergency department (ED) and hospital settings can be "highly variable."

As a result, the investigative team sought to determine the accuracy of the newer International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes to identify influenza in children, as accuracy is unknown.

ICD-10 codes are commonly used to identify influenza in clinical studies, but a better understanding of the accuracy is needed "to evaluate potential misclassification in studies using these codes to identify influenza as either an exposure or an outcome," wrote the study authors.

The investigative team was lead by James W. Antoon, MD, PhD, MPH, of the Department of Pediatrics at Vanderbilt University Medical Center in Nashville, Tennessee.

Study design consisted of children younger than 18 years that presented to the ED or inpatient settings with fever and/or respiratory symptoms. Presentations occurred at 7 pediatric medical centers in the United States that are affiliated with the Centers for Disease Control and Prevention (CDC)-sponsored New Vaccine Surveillance Network. Data was collected from December 1, 2016 to March 31, 2020, with analysis performed in August 2023.

Diagnosis of influenza (ICD-10-positive) was defined as the presence of an ICD-10 code for patients discharged from the ED or ICD-10 code among hospitalized patients. True-positive cases were characterized as those with an ICD-10 code and an influenza-positive molecular test (influenza positive). False-negative cases were defined as influenza positive, but without an influenza ICD-10 code. False-positive cases were those with an influenza-negative molecular test (influenza negative), but an ICD-10 positive code. Measures included specificity, sensitivity, positive predictive value (PPV), and negative predictive value (NPV).

All told, there were 16,867 children in the ED (median age, 2.0 90.0-4.0] years, 55.2% boys) and 17,060 inpatients (median age, 1.0 [0.0-4.0] years, 57.4% boys). ICD-10 influenza diagnoses were highly specific in the ED (98.0%; 95% CI, 97.8%-98.3%). PPV was 88.6% (95% CI, 88.0-89.2%) and NPV was 85.9% (95% CI, 85.3%-86.6%). Sensitivity in the ED was lower, at 48.6% (95% CI, 47.6%-49.5%).

For inpatients, specificity was 98.2% (95% CI, 98.0%-98.5%), PPV was 82.8% (95% CI, 82.1%-83.5%), sensitivity was 70.7% (95% CI, 69.8%-71.5%), and NPV was 96.5% (95% CI, 96.2%-96.9%), according to the posted results from the investigators, who noted that ICD-10 accuracy varied by age of the patient, influenza season definition, time between onset and testing, and clinical setting.

In conclusion, the study authors wrote, "influenza ICD-10 discharge diagnoses were highly specific but moderately sensitive in identifying laboratory-confirmed influenza; the accuracy of influenza diagnoses varied by clinical and epidemiological factors. In the ED and inpatient settings, an ICD-10 diagnosis likely represents a true-positive influenza case."

Reference:

Antoon JW, Stopczynski T, Amarin JZ, et al. Accuracy of Influenza ICD-10 Diagnosis Codes in Identifying Influenza Illness in Children. JAMA Netw Open. 2024;7(4):e248255. doi:10.1001/jamanetworkopen.2024.8255