Research Associate Global Public Health, Johnson & Johnson Inc., Canada
Background: Healthcare claims databases are widely used in epidemiological research. However, their suitability to answer research questions for rare and non-endemic diseases requires assessment. Dengue is a hemorrhagic febrile illness with low incidence within the United States (US), but it is increasing in frequency as a cause of febrile illness in travelers returning from dengue-endemic areas.
Objectives: To evaluate whether a large healthcare insurance claims database was fit-for-purpose in identifying patient characteristics and healthcare utilization among those diagnosed with dengue infection in the US over a 10-year period (2010-2019).
Methods: The Merative MarketScan Research Database was used to identify an incident cohort of individuals with dengue infection. Individuals were required to have data 6 months prior to diagnosis and at least 30 days following diagnosis (index date), location within either of the 50 states, and no diagnosis code for other flavivirus infections within +/- 30 days of the index date. Multiple data elements were needed to ascertain healthcare utilization including population identification, dengue virus (DENV) exposure status (location of exposure and serotype), diagnostic testing (dengue laboratory testing), outcomes of infection and treatment.
Results: There were 3,185 individuals with incident DENV diagnoses over the 10-year period. About half were females (51.4%) and the majority were adults ≥20 years of age (80.0%). There was no information of dengue exposure history and little evidence of DENV infection diagnostic testing recorded ( < 1%). On the index date, healthcare utilization included 81.5% outpatient visit, 14.8% inpatient visit, 7.4% inpatient and emergency department (ED) visit, and 7.0% ED visit. 30 days post-index date, 26.4% of individuals had an expanded medical examination and the top four prescribed medications included doxycycline (4.2%), acetaminophen (4.1%), hydrocodone (2.8%), and oxycodone (2.2%).
Conclusions: Administrative claims data allowed identification of individuals with recorded diagnosis of dengue infection, but healthcare and drug utilization information lacked detail for valid inference. Diagnostic testing results to confirm DENV infection and serotypes were infrequent. Alternative sources to administrative claims data with more complete capture of symptoms, diagnostic testing results, and treatments should be considered for characterizing dengue infection in non-endemic regions.
