False-Positive Results with a Commercially Available West Nile Virus Immunoglobulin M Assay --- United States, 2008

DF Neitzel, MS, MM Kemperman, MPH, S Semple, MS, S Wong, PhD, J Hallisey, MPH, MA Feist, TK Miller, WM Chung, MD, S Hojvat, PhD, P Summers, MS, RS Lanciotti, PhD, AJ Panella, MPH, J Laven, O Kosoy, MS, JA Lehman, RS Nasci, PhD, M Fischer, MD, JE Staples, MD, E Zielinski-Gutierrez, DrPH, KB Janusz, DVM


Morbidity and Mortality Weekly Report. 2009;58(17):458-460. 

In This Article


False-Positive Results with a Commercially Available West Nile Virus Immunoglobulin M Assay --- United States, 2008

In September 2008, CDC, the Food and Drug Administration (FDA), and state health departments began a nationwide investigation into an increase in false-positive test results obtained with a commercially available West Nile virus (WNV) immunoglobulin M (IgM) capture enzyme-linked immunosorbent assay (ELISA). The investigation revealed that, in the United States, one lot of the commercially available test kits was the source of the false-positive results.[1] That lot was recalled, and a second lot distributed outside the United States also was recalled.[1] During July 1--September 30, 2008, the kit lot implicated in the United States resulted in positive tests on 568 specimens collected from 518 patients in 42 states and the District of Columbia (DC). A total of 166 (29%) specimens were retested at CDC, and 119 (72%) had false-positive results. A higher false-positive percentage were found among patients without evidence of neuroinvasive disease (77%) than patients with evidence of neuroinvasive disease (47%). Of the 518 patients, 249 (48%) had been reported to CDC as persons with WNV disease; however, only 45 (18%) had confirmatory testing that supported their inclusion in national surveillance data. Commercially available WNV test kits should be used to determine a presumptive diagnosis of WNV neuroinvasive disease. These kits should not be used to test specimens from persons without compatible illness, and any positive result should be confirmed by additional testing at a state health department or CDC.

WNV infection is a nationally notifiable disease. Cases of WNV disease are reported by state health departments to CDC through ArboNET, an Internet-based, passive surveillance system.* Cases reported to ArboNET must have clinical evidence of compatible illness and laboratory evidence of recent WNV infection.[2] Based on patients' clinical signs and symptoms, WNV cases are classified as neuroinvasive disease (i.e., encephalitis, meningitis, or acute flaccid paralysis) or nonneuroinvasive disease (i.e., other febrile illness). Four FDA-cleared WNV serologic assays are commercially available for use in the United States. These assays are labeled for use on serum to aid in a presumptive diagnosis of WNV infection in patients who have clinical symptoms consistent with neuroinvasive disease. According to product inserts,[3,4,5,6] all positive results obtained with these assays should be confirmed by plaque reduction neutralization test (PRNT) or by using current CDC guidelines for laboratory diagnosis of this disease.[7]

Initial Investigation

In summer 2008, three state health departments independently contacted CDC regarding positive WNV IgM antibody test results in patients who lacked clinical or epidemiologic evidence of WNV infection. All of these tests results originated from one large commercial laboratory that was using the PanBio WNV IgM ELISA test kit manufactured by Inverness Medical (Princeton, New Jersey). On September 5, 2008, the New York State Department of Health's Wadsworth Center laboratory reported that 13 (86%) of 15 specimens testing positive for WNV IgM antibodies at the commercial laboratory in August were negative upon retesting at the state laboratory. On September 10, CDC notified all state health departments of the potential problem and initiated an investigation into the cause of the false-positive test results.[1]

In late September, one of the affected commercial laboratories sent a convenience sample of 64 specimens that had yielded positive or negative WNV IgM antibodies results to CDC and the kit's manufacturer for retesting. This evaluation identified two lots of the kit with higher false-positive rates (20% and 56%) than the expected rate calculated from data in the package insert (2% [95% confidence interval = 0%--9%]). On October 8, these two lots were recalled voluntarily by the manufacturer. The lot with the 56% false-positive rate had been distributed to four laboratories in the United States and was used for testing specimens during July--September (Figure 1). The other lot was distributed outside the United States.[1] On October 14, a CDC health advisory was distributed,[1] and the investigation was expanded to determine the scope and impact of the problem in the United States.

Figure 1.

Number of specimens (N = 568) testing positive for West Nile virus immunoglobulin M antibodies, using one lot from a commercially available test kit that was later recalled, by week --- United States, July--September 2008

Expanded Investigation

In September, CDC, along with state and local health departments, surveyed the four laboratories that had received the recalled kit lot to determine the number of positive specimens obtained using the lot and to collect corresponding demographic information regarding these patients. State health departments provided additional information regarding WNV confirmatory testing performed in state laboratories, patient clinical syndromes (e.g., neuroinvasive or nonneuroinvasive), and case status as reported to ArboNET.

The recalled WNV ELISA kit lot had produced positive results for 568 specimens obtained from 518 patients in 42 states and DC (Figure 2). Of the 488 patients for whom clinical information was known, 83 (17%) had symptoms consistent with WNV neuroinvasive disease, 242 (50%) had symptoms consistent with nonneuroinvasive disease, and 163 (33%) had no symptoms consistent with WNV disease.

Figure 2.

Number of persons (N = 518) testing positive for West Nile virus immunoglobulin M antibodies using one lot from a commercially available test kit that was later recalled --- United States, July--September 2008

During October--December, 166 (29%) available specimens of the 568 that tested positive with the implicated kit lot were identified and sent to CDC to be retested using WNV IgM microsphere immunofluorescence assay (MIA) and IgM capture ELISA. Based on retesting, specimens were classified as false-positive, true-positive, or indeterminate. Of the 166 retested, 45 (27%) were classified as true-positive and 119 (72%) as false-positive results; two specimens had an indeterminate result. The retested specimens came from 160 patients; clinical syndrome was known for 157 of these patients. Of the 157, a higher percentage of false-positives was found among patients without evidence of neuroinvasive disease (77% [98 of 127]) than among patients with evidence of neuroinvasive disease (47% [14 of 30)]) (p<0.001 by chi-square test).

Of the 518 patients testing positive for WNV with the implicated kit lot, 249 (48%) had been reported to ArboNET as having WNV disease. However, only 45 (18%) of these 249 cases had confirmatory testing supporting their inclusion as WNV disease cases; 77 (31%) cases did not have evidence of WNV infection based on subsequent laboratory testing, and 127 (51%) cases had no further testing performed. For the remaining 269 (52%) of the 518 patients, case investigation by state health departments found no illness clinically compatible with WNV disease; therefore, these patients were not reported to ArboNET.

* Available at http://www.cdc.gov/ncidod/dvbid/westnile/index.htm.
 An additional 58 (10%) of the 568 positive specimens were retested at state public health laboratories. Various assays were used; therefore, the results are not directly comparable to those from CDC. Nonetheless, of the 58 retested, a percentage similar to that found at CDC (64%) had false-positive results. In addition, false-positive percentages similar to those found at CDC were detected in persons without evidence of neuroinvasive disease (88%) and with evidence of neuroinvasive disease (33%) (p<0.001 by chi-square).