We evaluated the performance of surveillance case definitions for Zika virus disease recommended by the CSTE, WHO, PAHO, ECDC, and the Singapore MOH by using a cohort of 359 adult patients with suspected Zika virus disease who came to the Institute of Infectious Diseases and Epidemiology, Tan Tock Seng Hospital, Singapore, the national referral center for Zika virus disease during the containment phase of the Zika virus outbreak during August 26– September 5, 2016. All adults living or working in the outbreak area who were sick and had symptoms that partially or fully met the MOH definition were screened for Zika virus disease.

At their first visit to the hospital, all patients had their signs and symptoms documented, and blood and urine samples were obtained for detection of Zika virus nucleic acids by reverse transcription PCR (RT-PCR) (15). Parallel testing in the hospital laboratory and at the National Public Health Laboratory (Singapore) was conducted to maximize sensitivity and negative predictive values to rule out Zika virus infection.

A total of 42.0% of the cohort had Zika virus infection confirmed in blood (4%), urine (36%), or both (60%) samples (Table 1). Most (80%) infected and noninfected patients were tested <5 days after illness onset (infected patients, mean 3.6 days; noninfected patients, mean 4.6 days). Infected and noninfected patients were similar in age and sex. No female patients were pregnant. Among Zika virus–infected patients, rash (93.3%) was the most common symptom, followed by fever (79.2%) and myalgia (42.3%). Headache, arthralgia, and conjunctivitis were reported in <25% patients with Zika virus disease. Pruritus (11.4%) and gastrointestinal symptoms (6.7%) were relatively uncommon. For patients not infected with Zika virus, fever (86.2%) was the most common symptom, followed by myalgia (59.1%) and rash (44.8%).

The case definition recommended by CSTE for use in the United States (US definition) had a sensitivity of 100% and a specificity of 2% in detecting Zika virus in the cohort (Table 2). The WHO case definition had the lowest sensitivity (38%). The Singapore MOH case definition had a sensitivity of 54% and a high specificity of 76%, and performed well in diagnosing Zika virus disease (positive likelihood ratio [LR+] 2.2, 95% CI 1.7–3.0). The performances of PAHO (LR+ 2.1, 95% CI 1.5–2.8) and ECDC (LR+ 2.1, 95 % CI 1.6–2.8) case definitions were similar.

Despite increasing incidence of Zika virus disease and its spread across the Americas and Asia, there is no internationally adopted common clinical criteria for the surveillance of this disease. We report a large outbreak cohort of patients with suspected Zika virus infection and comprehensive documentation of clinical symptoms and parallel RT-PCR conducted on blood and urine samples for these patients by 2 laboratories. Evaluation of the performance of surveillance case definitions in such a cohort would provide useful findings that would contribute to development of guidance for Zika virus disease surveillance.

Diagnosis of Zika virus disease remains suboptimal because of limited availability of confirmatory testing by RT-PCR during acute illness and cross-reactivity of serologic tests for Zika virus with other co-circulating flaviviruses (3,4). Thus, a good discriminatory clinical criteria for disease surveillance is crucial for prevention and control of Zika virus transmission.

The US case definition would identify all Zika virus infections and be useful for prevention of autochthonous transmission by imported cases. However, because this definition is not specific, considerable resources would be required for confirmatory testing of identified cases. The definition requires laboratory testing to report a case. Thus, sensitivity of the definition is most likely appropriate in the US setting. Conversely, the WHO case definition might miss 60% of Zika virus infections. For Zika virus disease surveillance in the absence of commercially available diagnostic laboratory tests, case definitions incorporating rash as a required clinical criteria, such as the PAHO, ECDC, and Singapore MOH case definitions, would be useful (LR+ >2), although ≈50% (range 44%–51%) of cases of Zika virus disease could be missed.

The main limitation of this study is that it included only adults. However, the small number of children infected with Zika virus during the containment phase of the outbreak in Singapore had symptoms similar to those for adults (A. Chow et al., unpub. data). Some Zika virus infections could have been misclassified as noninfections because RT-PCR could have missed infections late in the illness course or after development of antibodies against Zika virus.

In conclusion, we evaluated the performance of 5 case definitions for Zika virus disease surveillance. In the current effort to halt transmission of this virus worldwide, and with laboratory tests being largely inaccessible, use of surveillance case definitions that include rash as a required clinical criteria would provide a high yield in identifying Zika virus disease.