Predictive value of rapid influenza tests varies with prevalence

Article Outline

• Grijalva CG, Poehling KA, Edwards KM, Weinberg GA, Staat MA, Iwane MK, et al. Accuracy and interpretation of rapid influenza tests in children. Pediatrics 2007;119:e6-11
• Copyright

Back to Article Outline

Grijalva CG, Poehling KA, Edwards KM, Weinberg GA, Staat MA, Iwane MK, et al. Accuracy and interpretation of rapid influenza tests in children. Pediatrics 2007;119:e6-11 

Question Among young children, how does prevalence of influenza alter the predictive ability of rapid influenza tests?

Design Longitudinal cohort.

Setting Three counties in the Unites States.

Participants Children enrolled in the New Vaccine Surveillance Network, age 5 years and younger, who were hospitalized with respiratory symptoms or fever between October 2000 and September 2004. Outpatients from 2002 to 2004 also were included.

Intervention Nasal and throat swabs were obtained, and influenza virus was detected by culture and reverse-transcription polymerase chain reaction (RT-PCR). Provider-ordered rapid influenza tests were compared with the criterion standard (culture and RT-PCR) to determine their sensitivity and specificity.

Outcome Trends in weekly predictive values of the rapid tests estimated during the influenza seasons.

Main Results Rapid influenza tests had an overall sensitivity of 63% and specificity of 97%. In 2002-2003, the prevalence of influenza in symptomatic outpatient children peaked at 21% and remained above 10% for ∼4 weeks. In contrast, in 2003-2004, the prevalence peaked at 60% and remained above 20% for ∼6 weeks. The positive predictive value (PPV) of the rapid tests approached 80% when the influenza prevalence was ≥15% but decreased to <70% when the prevalence was <10%.

Conclusions The prevalence of influenza varies between and within seasons. On the basis of these estimates, rapid tests are of limited use when the prevalence is <10%. The appropriate interpretation of rapid influenza tests requires local influenza surveillance and timely communication of this information to practitioners.

Commentary Influenza is a very common childhood disease, responsible for many pediatric ambulatory visits and hospitalizations during the winter months in the northern hemisphere. This study conveys an important message about interpreting diagnostic tests and reiterates the importance of disease prevalence in test interpretation. That is, when the disease prevalence is low, the PPV of a test also will be low, and when the disease prevalence is high, the PPV will be high. Several important points should be considered when examining the results of the present study. First, although all the tests were compared using a gold standard procedure (viral culture or RT-PCR), several antigen tests with differing performance characteristics, were lumped together. This affects interpretation of the results of a single rapid test used in an institution. Second, to assess whether a test is valid, we need to ask whether the diagnostic test was evaluated in an appropriate spectrum of patients (eg, those in whom it would be used in practice). Rapid tests to either rule in or rule out disease are useful primarily in outpatient settings for such reasons as to reduce the need for further testing and/or antibiotic therapy and to reassure parents. In this study, the rapid tests were performed only on inpatients, and the results were translated to the outpatient population. This limits the generalizability of the results to the outpatient and emergency department settings, where these tests are most often used.