Determination of clinical and demographic predictors of laboratory-confirmed influenza with subtype analysis.

BACKGROUND

Rapid influenza diagnosis is important for early identification of outbreaks, effective management of high-risk contacts, appropriate antiviral use, decreased inappropriate antibiotic use and avoidance of unnecessary laboratory testing. Given the inconsistent performance of many rapid influenza tests, clinical diagnosis remains integral for optimizing influenza management. However, reliable clinical diagnostic methods are not well-established. This study assesses predictors of influenza, and its various subtypes, in a broad population at the point of care, across age groups, then evaluates the performance of clinical case definitions composed of identified predictors.

METHODS

Respiratory specimens and demographic and clinical data were obtained from 3- to 80-year-old US military family members presenting for care with influenza-like illness (ILI) from November 2007 to April 2008. Molecular and virus isolation techniques were used to detect and subtype influenza viruses. Associations between influenza diagnosis and demographic/clinical parameters were assessed by logistic regression, including influenza type and subtype analyses. The predictive values of multiple combinations of identified clinical predictors (case definitions), and the Centers for Disease Control and Prevention (CDC) ILI case definition, were estimated.

RESULTS

Of 789 subjects, 220 (28%) had laboratory-confirmed influenza (51 A(H1), 46 A(H3), 19 A(unsubtypeable), 67 B, 1 AB coinfection), with the proportion of influenza A to B cases highest among 6- to 17-year-olds (p = 0.019). Independent predictors of influenza included fever, cough, acute onset, body aches, and vaccination status among 6- to 49-year-olds, only vaccination among 3- to 5-year-olds, and only fever among 50- to 80-year-olds. Among 6- to 49-year-olds, some clinical case definitions were highly sensitive (100.0%) or specific (98.6%), but none had both parameters over 60%, though many performed better than the CDC ILI case definition (sensitivity 37.7%, 95% confidence interval 33.6-41.9% in total study population).

CONCLUSIONS

Patterns of influenza predictors differed across age groups, with most predictors identified among 6- to 49-year-olds. No combination of clinical and demographic predictors served as a reliable diagnostic case definition in the population and influenza season studied. A standardized clinical case definition combined with a point-of-care laboratory test may be the optimal rapid diagnostic strategy available.