Progress Toward Eliminating Type b Disease Among Infants and Children—United States, 1987-1997False-Positive Laboratory Tests for Involving an Enzyme-Linked Immunosorbent Assay—United States, November 1997-March 1998Progress Toward Eliminating False-Positive Laboratory Tests for FREE

MMWR. 1998;47:993-998

2 tables, 1 figure omitted

Haemophilus influenzae type b (Hib) causes serious invasive diseases among previously healthy children aged less than 5 years. Before the availability of conjugate vaccines in 1988, Hib was the most common cause of bacterial meningitis among preschool-aged children.^1- 2 Since 1993, the incidence of Hib invasive disease (defined as illness clinically compatible with invasive disease such as meningitis or sepsis, with isolation of the bacterium from a normally sterile site) among children aged <5 years has declined greater than 95% in the United States.³ This report describes the continued decline of reported Hib invasive disease cases and underscores the need for investigation of Haemophilus influenzae (Hi) invasive disease cases.

State health agencies and the District of Columbia provide weekly reports of provisional cases of Hi invasive disease to CDC through the National Electronic Telecommunications System for Surveillance (NETSS).⁴ Case reports include basic demographic data about persons with Hi invasive disease, and supplemental information (e.g., the serotype that caused illness, clinical illness, outcome, and Hib vaccination status). For 1996 and 1997, all states were contacted approximately every 2 months to obtain supplemental information about cases of Hi invasive disease in children aged <5 years. Hi cases identified by the active laboratory-based surveillance system also are reported to CDC through NETSS or the National Bacterial Meningitis and Bacteremia Reporting System. Reported Hib vaccination doses were considered valid if administration dates were available and if they were given ≥14 days before illness onset. Rates were calculated using 1996 census data.

Among children aged <5 years, 280 cases of Hi invasive disease were reported in 1996 (incidence: 1.5 per 100,000 children), and 258 cases were reported in 1997 (incidence: 1.3 per 100,000 children). Incidence in 1996 and 1997 represented a decline of 97% from 1987 (41 per 100,000). From 1987 through 1997, the incidence of Hi disease varied slightly among persons aged ≥5 years (range: 0.3-0.6 per 100,000) (Figure 1).

For children aged <5 years, serotype data were available for 200 (71%) of 280 cases in 1996 and for 200 (78%) of 258 cases in 1997. Of the cases for which serotype was known, in 1996, Hib was the cause of illness in 63 (32%) cases, and in 1997, in 81 (41%) cases. By state, excluding Alaska, the average annual incidence of Hib invasive disease during 1996-1997 ranged from 0 to 2.9 per 100,000 children aged less than 5 years; in Alaska, the incidence was 15.1 per 100,000 children (Table 1). The incidence of nontype b Hi disease ranged from 0 to 3.7 (national rate: 0.7 per 100,000).

During 1996-1997, the average annual incidence of Hib invasive disease per 100,000 children aged <5 years varied by race/ethnicity: 0.5 among non-Hispanic whites, 0.7 among non-Hispanic blacks, 12.4 among American Indians/Alaskan Natives, 0.6 among Asians/Pacific Islanders, and 0.7 among Hispanics. Race/ethnicity data were missing for 12 (8%) children.

Population-based surveillance for Hi invasive disease is part of a multistate active surveillance project coordinated by CDC. From 1989 through 1997, CDC collaborated with investigators in state and local health departments and universities in several geographically dispersed areas of the United States, with a median population of 1,060,505 children aged <5 years (range: 750,534 in 1989 to 1,605,777 in 1997). During 1989-1991, surveillance was conducted in eight Atlanta area counties, three San Francisco Bay area counties, four urban counties in Tennessee, and the entire state of Oklahoma. In 1992, Maryland was added. Missouri participated during 1992-1993. In 1995, a county in Tennessee was added, and Oklahoma discontinued participation. In 1996, Connecticut and Oregon and seven counties in Minnesota were added. In 1997, active surveillance in Georgia expanded to 20 counties, and surveillance in Minnesota expanded to the entire state. Information routinely obtained for cases of Hi invasive disease was similar to that collected by the national surveillance systems. Rates were calculated using census projections from 1989 through 1996 and were race-adjusted to the U.S. population.³

From 1989 to 1997, the race-adjusted incidence of Hib invasive disease among children aged <5 years declined 99%, from 34 to 0.4 per 100,000. During 1996-1997, 79 cases of Hi invasive disease were reported among children aged <5 years. Of these, 14 (18%) were caused by Hib; 48 (61%), by nontype b Hi; and 17 (22%), by unknown serotypes. From 1989 to 1997, the median race-adjusted incidence of nontype b Hi invasive disease was 1.6 per 100,000 children (range: 1.1 to 3.8 per 100,000); the median incidence was higher among blacks (3.2) than among all others (1.4).

Of the 144 children with confirmed Hib invasive disease who were reported to CDC through national surveillance, 69 (48%) were aged <6 months and therefore were too young to have completed a three-dose primary Hib vaccination series (Table 2), and 75 (52%) children were eligible to have completed a primary series (aged ≥6 months). Of the 75 children, 48 (64%) were incompletely vaccinated or vaccination status was unknown, and 27 children had completed a primary series; 14 children also had received a booster dose. Five (4%) of 115 children with known outcome and Hib invasive disease died; the deceased children were aged <6 months and had received one or no Hib vaccine doses.

G Rothrock, MPH, Emerging Infections Program, San Francisco; D Vugia, MD, S Waterman, MD, State Epidemiologist, California State Dept of Health Svcs. N Barrett, MS, JL Hadler, MD, State Epidemiologist, Connecticut Dept of Public Health. W Baughman, MSPH, M Farley, MD, D Stephens, MD, Veterans Administration Medical Svcs and Emory Univ School of Medicine, Atlanta; K Toomey, MD, State Epidemiologist, Georgia State Dept of Health. L Billmann, MPH, L Harrison, MD, Johns Hopkins Univ, Baltimore; DM Dwyer, MD, State Epidemiologist, Maryland State Dept of Health and Mental Hygiene. J Rainbow, MPH, M Osterholm, PhD, State Epidemiologist, Minnesota Dept of Health. M Skala, Missouri Dept of Health. LM Smithee, MS, Oklahoma State Dept of Health. K Stefonek, MPH, D Fleming, MD, State Epidemiologist, State Health Div, Oregon Dept of Human Resources. B Barnes, MS, L Lefkowitz, MD, Dept of Preventive Medicine, Vanderbilt Medical Center, Nashville, Tennessee. Meningitis and Special Pathogens Br and Respiratory Diseases Br, Div of Bacterial and Mycotic Diseases, and Active Bacterial Core Surveillance/Emerging Infections Program Network, National Center for Infectious Diseases; Child Vaccine Preventable Disease Br, Epidemiology and Surveillance Div, National Immunization Program, CDC.

Since 1988, when Hib conjugate vaccines were first licensed for children aged 18-59 months in the United States, with subsequent licensure in 1990 and widespread use in infants, the number of reported Hib invasive disease cases among children aged <5 years has declined 99%. However, surveillance data indicate that circulation of Hib continued and that some children remained susceptible to disease; susceptible children include those who do not respond or are too young to complete the primary series of Hib vaccination and those who are unvaccinated or undervaccinated. In the 1997 National Immunization Survey of children aged 19-35 months, the coverage level for receipt of three Hib vaccine doses by age 7 months was 61% (CDC, unpublished data, 1997); by age 24 months, coverage for three doses reached 93%.⁵ High coverage levels will help protect susceptible children in the community by herd immunity (i.e., by less frequent exposure to pharyngeal carriers of the organism).⁶

The small number of reported Hib cases among children who had completed a primary Hib vaccine series suggests that vaccine failure occurs infrequently. However, vaccination history was known for only 54 (72%) of the 75 Hib case-patients aged ≥6 months. Vaccination history is needed to determine whether Hib invasive disease results from vaccine failure or failure to vaccinate. Protection induced by vaccination is not absolute, and cases will continue to occur as long as the Hib organism circulates in populations.

Serotype information for Hi invasive disease cases is essential to monitor progress toward elimination. This information also is needed to monitor nontype b Hi invasive disease to determine whether there is an increase in invasive disease with another serotype or with nontypeable strains, and to measure the sensitivity of the surveillance system. In 1997, information about serotype had been reported for 78% of 258 cases, compared with 41% of 340 cases in 1994.³ State health departments are encouraged to promote laboratory reporting of Hi cases and to identify laboratories that can perform serotyping on Hi isolates from children aged <15 years with invasive disease; if serotyping is not available, state health departments can contact CDC.

To strengthen national surveillance, the incidence of nontype b Hi invasive disease among children aged <5 years can be used to monitor the sensitivity of reporting; Hi invasive disease caused by any serotype and nontypeable strains, in addition to type b strains, is nationally notifiable.⁷ Although Hi invasive disease rates may vary by racial/ethnic groups, as was the case in the prevaccine era,^1- 3,8 the incidence of nontype b Hi invasive disease will occur within an expected range. For example, in California, the two regions of the state with active, laboratory-based surveillance had an incidence rate of nontype b Hi invasive disease of 1.5 per 100,000 children aged <5 years.⁸ In 1996 and 1997, 24 states reported annual rates of ≥0.5 nontype b Hi invasive disease cases per 100,000 children aged <5 years.

Age-appropriate vaccination starting at age 2 months continues to be the most important method to protect children from Hib invasive disease. Health-care providers should emphasize to parents the importance of vaccinating children against Hib invasive disease.⁹

MMWR. 1999;48:4-8

(1 table omitted)

From November 1997 through March 1998, the number of positive tests for Cryptosporidium increased in several locations in the United States. Several laboratories (e.g., the New York state laboratory and the Medical Science Laboratories in Wisconsin) retested original stool specimens and could not confirm the original positive test result. Following reports to the manufacturer by the Massachusetts, New York, and Wisconsin state health departments about possibly inaccurate test results, Alexon-Trend* (Ramsey, Minnesota) notified its laboratory customers in a March 25, 1998, letter that three lots of its enzyme-linked immunosorbent assay (ELISA) 24 well (catalog number 540-24) ProSpecT® Cryptosporidium Microplate Assay (lot numbers 970717, 975011, and 980401) and seven lots of its ELISA 96 well (catalog number 540-96) ProSpecT® Cryptosporidium Microplate Assay (lot numbers 970696, 970775, 970883, 975006, 980402, 980808, and 980809) were subject to a "non-specific reaction between some stool specimens and the microplate assay" (i.e., a false-positive test result) (K. Hood, Alexon-Trend, personal communication, March 25, 1998). Alexon-Trend directed laboratories to discontinue using kits with implicated lot numbers. This report summarizes an analysis of reports of false-positive tests and describes identification of apparent clusters in three states.

On April 2, 1998, CDC requested state epidemiologists and state laboratory directors to report suspected cases and clusters of false-positive tests. Six states (California, Idaho, Maine, Massachusetts, New York, and Wisconsin) reported apparent clusters and/or an increase in the overall number of positive test results. A working group of state and local public health laboratorians and epidemiologists from these six states participated in a conference call on May 18, 1998, to review their experiences. The findings from five states were reviewed; an apparent false-positive cluster in Idaho was omitted because it involved an ELISA kit not referenced in the manufacturer's letter.

The working group established three case definitions. A confirmed false-positive (CFP) case was one in which a stool specimen that originally tested positive by an implicated lot of the Alexon-Trend kit before March 25, 1998, was available for retesting, subsequently tested negative by an alternate ELISA kit, and if additional testing was performed (e.g., acid-fast and/or fluorescent antibody staining), tested negative by the additional method(s). A possible false-positive (PFP) case was one in which a stool specimen that originally tested positive by an implicated lot of the Alexon-Trend kit before March 25, 1998, was not available for retesting by an alternate ELISA kit but tested negative by an additional method(s) (e.g., acid-fast and/or fluorescent antibody staining). An indeterminate case was one in which a stool specimen tested positive by an implicated lot of the Alexon-Trend kit before March 25, 1998, but for which no original stool specimen was available for retesting, and the original stool specimen was not tested by any other method. Participating laboratories were given a letter designation (e.g., New York has reports from five laboratories, which are designated NY-A, NY-B, NY-C, NY-D, and NY-E).

A total of 62 CFP, eight PFP, and 155 indeterminate cases, including four clusters, were reported in the five states. Five laboratories provided information regarding their rate of positivity (i.e., the number of positive tests for Cryptosporidium expressed as a percentage of the total number of tests for Cryptosporidium) for January 1997-April 1998. For each laboratory, CFP, PFP, and indeterminate cases occurred at the same time as the highest rates of positivity. Information was not available regarding how false-positive test results may have affected patients (e.g., additional diagnostic testing or experimental therapy). Maine, Massachusetts, and Wisconsin provided details regarding their investigations to determine the cause of their suspected disease cluster.

During November-December 1997, laboratory MA-A reported four stool specimens positive by ProSpecT® Cryptosporidium Microplate Assay from residents of one town in Massachusetts. The local health department found no link between cases, and testing of the town's water supply was negative for Cryptosporidium. During January-March 1998, 27 additional positive test results were reported from this laboratory, compared with one to two positive tests per month during the same 3-month period in 1997. No stool specimens were available for retesting. The physicians who ordered the stool tests were notified that positive test results should be considered indeterminate.

During November-December 1997, laboratory WI-A noted that 10 stool specimens that were positive by ProSpecT® Cryptosporidium Microplate Assay were all negative when retested by direct fluorescent antibody (DFA); four also were negative when retested by repeat ELISA. This increase could not be explained by an increase in effluent turbidity at the water treatment plant or by an increase in morbidity measured by other surveillance systems in place in Milwaukee County since the 1993 Cryptosporidium outbreak.^1- 2 WI-A had noted a gradual increase in the rate of positive ELISAs for Cryptosporidium from a background of ≤2% in the fall of 1997 to 5% in March 1998, with peaks of ≥25% positive on March 6 and 19. The other 11 laboratories involved in statewide Cryptosporidium surveillance, all of which use DFA routinely, reported no increases in absolute number of tests or increases in the rate of positive tests. The physicians who ordered the stool tests were notified of CFP or indeterminate results.

From late January to early February 1998, 41 of 50 elderly male residents on one ward and one of 50 residents on a second ward at a 100-bed extended-care facility experienced gastrointestinal illness. The first cases of illness began approximately 10 days after a severe ice storm caused a power failure lasting several days at the facility and in surrounding communities. Stool samples were negative for bacterial pathogens. Additional persons with diarrhea were reported in mid-February; two of four initial stool specimens from these persons tested positive by ProSpecT® Cryptosporidium Microplate Assay. Stool specimens from 35 of 79 facility patients in both wards and from one outpatient tested positive for Cryptosporidium by this method. A public health investigation and water testing were performed at the facility. Because clinical and epidemiologic characteristics of this outbreak were inconsistent with cryptosporidiosis, the 36 antigen-positive specimens were re-evaluated at ME-A, a reference laboratory, and all were negative by ELISA. Water tests were negative for coliform bacteria.

JR Miller, MD, B Mojica, MD, City Epidemiologist, New York City Dept of Health. J Nadle, MPH, California Emerging Infections Program; DJ Vugia, MD, SH Waterman, MD, State Epidemiologist, California Dept of Health Svcs. B Mamer, PhD, C Hahn, MD, State Epidemiologist, Idaho Dept of Health and Welfare. KM Doing, PhD, Affiliated Laboratories, Inc, Bangor; JL Hamm, N Buker, Togus Veterans Administration Hospital, Togus; GA Beckett, MPH, KF Gensheimer, MD, State Epidemiologist, Maine Dept of Human Svcs. P Kludt, MPH, A DeMaria, MD, State Epidemiologist, Massachusetts Dept of Public Health. J Ennis, MS, J Keithly, PhD, S Kondracki, D Ackman, MD, P Smith, MD, State Epidemiologist, New York State Dept of Health. D Warshauer, PhD, Medical Science Laboratories, Milwaukee; M Proctor, PhD, J Davis, MD, State Epidemiologist, Wisconsin Dept of Health and Social Svcs. Div of Parasitic Diseases, National Center for Infectious Diseases, CDC.

ELISA and other immunoassays offer advantages over diagnostic tests based on microscopic methods, especially for laboratories that perform large numbers of tests. ELISA can be used to test multiple stool specimens simultaneously, and ELISA does not require the same high level of technical skill needed to identify parasites based on the morphologic and staining characteristics observed during microscopic examination. However, when a laboratory depends solely on ELISA for detection of Cryptosporidium, false-positive test results may go unrecognized for long periods of time because of problems associated with the kit reagents or technician error.

Retaining stool specimens, or preparing a permanent microscopic slide whenever an ELISA result is positive has implications for cost, staffing, and storage. In laboratories that rely solely on antigen tests of stool specimens for parasites and that do not routinely retain stool specimens or make permanent slides, management should consider monitoring the rate of positive test results and, when this rate noticeably increases above a certain level (e.g., two or more times the laboratory's mean positivity rate for an organism), implement confirmatory testing by microscopic methods and/or begin archiving stool specimens. Alternatively, all stool specimens could be split before testing so that an aliquot of a specimen positive by ELISA could be sent to a reference diagnostic laboratory for confirmation. This method is analogous to using ELISA as a screening test for human immunodeficiency virus, with Western blot testing used to confirm specimens positive by ELISA.³ Another advantage of retaining stool specimens is its availability for polymerase chain reaction-based genotyping, as might be warranted in an outbreak. In New York, laboratories using ELISA must either prepare a permanent microscopic slide or retain a portion of the original stool specimen, and laboratories are required to hold slides or stool specimens for 1 year. As a result of the investigation described in this report, New York state has reminded laboratories of this existing requirement and has used this incident in a statewide educational workshop for laboratorians.

In many communities, a cluster of laboratory-reported cases of cryptosporidiosis elicits a multidisciplinary investigation to find the cause. Every community should develop a plan for responding quickly and efficiently to increases in the number of reported cases of cryptosporidiosis.⁴ Essential components of an effective response plan include confirming the diagnosis, comparing current disease data with baseline data, and developing a strategy for critically and systematically determining whether there is a community outbreak. Having access to good laboratory records and stored specimens facilitates confirmation of the diagnosis and reduces the likelihood that limited health department resources will be redirected to an unnecessary community-wide epidemiologic investigation on the basis of false-positive laboratory results.

When evidence suggests a commercial laboratory diagnostic kit is yielding inaccurate test results, this information should be forwarded to the kit manufacturer and the appropriate local and state health department. These departments will inform the state certifying authority for laboratory practice, the Food and Drug Administration, and CDC.

*Use of trade names and commercial sources is for identification only and does not imply endorsement by CDC or the U.S. Department of Health and Human Services.