Emergency Department Syndromic Surveillance Diagnostic Code Selection for Assessing Severity of Seasonal Influenza in New South Wales, Australia

TL;DR

This study identifies the best diagnostic codes for tracking seasonal influenza severity using emergency department data in New South Wales.

Contribution

The paper introduces a method to optimize diagnostic code selection for syndromic surveillance to better assess influenza severity.

Findings

Unspecified Viral and influenza-based ED syndromes showed the best similarity and timeliness for tracking influenza.

Pneumonia and lower respiratory tract infection syndromes were linked to higher hospital admission and confirmed influenza rates.

The method can improve severity assessment of seasonal influenza using ED surveillance data.

Abstract

Emergency department (ED) syndromic surveillance (EDSyS) often relies on preliminary or ED discharge diagnosis codes as indicators of influenza, but few studies provide a justification for their selection. This retrospective analytical study aimed to optimise the selection of diagnostic codes in EDSyS for monitoring influenza activity and severity.

Diagnostic codes potentially relating to a respiratory infection and assigned to people presenting to over 180 EDs in New South Wales (NSW), Australia, were grouped into 16 mutually exclusive ‘ED Syndromes’. Time series of the proportion of ED presentations for each ED syndrome by epidemiological week between 2010 and 2019 were compared to a reference series of the percentage influenza positive results from sentinel laboratories, using two similarity and three timeliness statistics. Hospital inpatient admission and laboratory notification data linked to each ED presentation allowed assessment of patient infection status and outcomes.

‘Unspecified Viral’ (any non‐specific viral illness, without reference to the respiratory system) and ED syndromes based on influenza like Illness (ILI) and influenza had the best combination of similarity and timeliness measures. Linked data identified relatively high rates of hospital admission, laboratory‐confirmed influenza and inpatient influenza diagnosis for ED syndromes based on pneumonia and lower respiratory tract infection.

In addition to ILI and influenza, ED syndromes based on unspecified viral illnesses can be used for EDSyS to assess influenza timing and transmissibility in NSW, Australia. The approach outlined in our paper can identify diagnostic codes to improve severity assessment of seasonal influenza using EDSyS.

Figures