Oral Fluid Testing during 10 Years of Rubella Elimination, England and Wales

Surveillance of rubella in England and Wales has included immunoglobulin M testing of oral (crevicular) fluid from reported case-patients since 1994. The need for laboratory confirmation to monitor rubella elimination is emphasized by poor sensitivity (51%, 95% confidence interval 48.9%–54.0%) and specificity (55%, 95% confidence interval 53.7%–55.6%) of the clinical case definition. During 1999–2008, oral fluid from 11,709 (84%) of 13,952 reported case-patients was tested; 143 (1.0%) cases were confirmed and 11,566 (99%) were discarded (annual investigation and discard rate of clinically suspected rubella cases was 2,208/100,000 population). Incidence of confirmed rubella increased from 0.50 to 0.77/1 million population when oral fluid testing was included. Oral fluid tests confirmed that cases were more likely to be in older, unvaccinated men. Testing of oral fluid has improved ascertainment of confirmed rubella in children and men and provided additional information for assessing UK progress toward the World Health Organization elimination goal.


I n 1970, rubella vaccination was introduced in the United
Kingdom for prepubertal girls and nonimmune women of childbearing age to protect them from the risks for rubella during pregnancy. Although this selective vaccination policy effectively reduced the number of cases of congenital rubella syndrome (CRS) and terminations of pregnancy, rubella during pregnancy continued to occur (1). In 1988, measles, mumps, and rubella (MMR) vaccine was introduced for universal vaccination at 13-15 months of age with the goal of eliminating circulating rubella.
A considerable decrease in rubella in young children followed, but in 1993, clinically diagnosed and laboratory-confi rmed rubella increased; the increase occurred predominantly in older men who had previously not been offered a rubella-containing vaccine (2). Therefore, in November 1994, rubella vaccine was included in a school catch-up campaign to prevent a predicted measles epidemic (3). Approximately 92% of children 5-16 years of age received combined measles-rubella vaccine. In 1996, to maintain measles control, a second dose of MMR was recommended for children 5 years of age.
For any disease in the elimination phase, accurate surveillance is necessary to identify reservoirs of infection and susceptible groups (2). In 2005, the World Health Organization (WHO) European Region adopted a resolution to eliminate indigenous rubella by 2010 (elimination goal of confi rmed rubella incidence <1 per 1 million population) (4). WHO has developed a clinical case defi nition for rubella (5), but identifi cation of cases based on clinical suspicion alone becomes less reliable as disease incidence decreases. Therefore, for countries trying to eliminate rubella, laboratory confi rmation of all suspected cases is recommended (4).
Before 1994, surveillance of laboratory-confi rmed rubella in England and Wales was based mainly on detection of immunoglobulin (Ig) M against rubella in serum. However, because rubella infection is usually mild, physicians are reluctant to obtain blood samples for serum confi rmation, especially from young children. There is also some reluctance to obtain serum from men because the diagnosis is not of major clinical signifi cance. Oral or crevicular fl uid is a noninvasively obtained clinical specimen that is likely to be more acceptable, especially for children, and is safe and easy to obtain (6)(7)(8)(9). Transudates from the capillary bed situated beneath the margin between the tooth and gum are obtained by rubbing an absorptive device between the gum and the cheek. These samples, which are distinguishable from saliva samples, contain mucosal cells that enable detection of the rubella virus by PCR. Methods for obtaining, extracting, and storing oral fl uid samples are well established (7,(10)(11)(12)(13). Detection of rubella IgM in oral fl uid has been validated and shown to be ≈90% sensitive and 99% specifi c compared with detection in serum (2). Samples are also suitable for genome detection (14,15). Therefore, since late 1994, the enhanced surveillance program in England and Wales has relied on oral fl uid testing to provide laboratory confi rmation for clinically diagnosed cases of measles, mumps, and rubella (however, serum testing is still recommended for confi rmation of infection during pregnancy).
An additional increase in rubella incidence occurred during 1995-1998. Reports of rubella peaked in 1996 (a total of 9,081 clinically diagnosed cases were reported) (16). This situation offered an opportunity to evaluate the sensitivity and specifi city of the WHO clinical case defi nition for rubella. In addition, we describe the added value of oral fl uid testing during the subsequent 10 years of rubella elimination (1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008).

Methods
Since 1988, physicians in England and Wales have been required by law to report clinically suspected cases of rubella to the proper offi cer at the local health authority (usually a public health consultant in a Health Protection Unit [HPU]). Since late 1994, when a report is received, the HPU sends an oral fl uid kit to the primary-care physician or patient for confi rmatory testing. The kit is then returned by prepaid envelope to the Virus Reference Department at the Health Protection Agency Centre for Infections, for analysis. A request form contains vaccination history and, until July 2003, some brief clinical features (presence of a rash, fever, conjunctivitis, cough, and lymphadenopathy [type not specifi ed]). Oral fl uid testing was also used to test cases that were not formally reported as part of outbreaks in 3 universities associated with imported virus from Greece in 1999 (17). A similar process is used for measles (and mumps) (18,19). If there is a strong clinical or epidemiologic suspicion of rubella in samples tested for measles and for measles in samples tested for rubella, dual testing is performed.
Oral (crevicular) fl uid specimens, obtained by wiping a specially designed sponge swab (oral test kit [Oracol; Malvern Medical Developments, Worcester, UK]) around the gum margins, were tested for rubella-specifi c IgM initially by using a solid-phase IgM-antibody capture radioimmunoassay (20). After 2002, an in-house assay for rubella IgG was introduced (21), and after 2003, all samples taken within 1 week after symptom onset that were negative for rubella IgM and rubella IgG were tested by reverse transcription-PCR for rubella virus (14). In 2006, the rubella solid-phase IgM-antibody capture radioimmunoassay was replaced by a commercial enzyme immunoassay (22).
Results of testing were sent to the reporting physician, and copies were sent to the relevant HPU. Confi rmed cases were defi ned as samples positive for rubella-specifi c IgM in oral fl uid or detection of rubella virus genome by PCR in persons without a history of receipt of rubella vaccine in the previous 6 weeks. These cases were reconciled with confi rmed rubella infections (positive for rubella IgM in serum) reported to the Health Protection Agency Centre for Infections, from laboratories in England and Wales, and duplicates are removed. Since 1999, residual samples from cases reported by local laboratories are requested to be sent to the national reference laboratory for confi rmation by an alternative IgM assay (with or without avidity testing) (23). Cases with negative results for the second IgM test or with high avidity are then excluded from the confi rmed total. Further details (including travel and contact history) are requested, and vaccination status is checked for all confi rmed case-patients.
Data obtained during January 1995-July 2003 were analyzed to estimate the sensitivity and specifi city of a clinical case defi nition. The WHO defi nition of a case of rubella is a generalized maculopapular rash and fever and arthralgia/arthritis or cervical, suboccipital, or postauricular lymphadenopathy (5). Because information for arthritis was not routinely obtained, the accuracy of a modifi ed case defi nition based on rash, fever, and lymphadenopathy was calculated against the standard of presence or absence of rubella-specifi c IgM in an oral fl uid sample. The sensitivity, specifi city, and positive predictive value were compared by patient age, sex, and year of report.
Data obtained during January 1999-December 2008 from 3 sources (clinically reported cases, confi rmation in oral fl uid samples, and laboratory reports) were then compared with respect to age, sex, vaccination status, and region. Rates were calculated by using 2001 population estimates from the Offi ce for National Statistics. The incidence rate for cases confi rmed positive by oral fl uid was adjusted for the proportion of reported cases tested in each region and compared with the rate from laboratory reports alone. All analyses were conducted by using Stata/SE version 9.2 (StataCorp LP, College Station, TX, USA). Proportions were compared by using the χ 2 test.

Accuracy of Clinical Case Defi nition
During January 1995-July 2003, of 29,825 reported case-patients, oral fl uid from 17,042 (57%) was tested for IgM; complete clinical information was obtained for 12,220 (72%) patients who submitted oral fl uid samples. The overall sensitivity of the clinical case defi nition of maculopapular rash and fever and lymphadenopathy was 51% (95% confi dence interval 48.9%-54.0%) and the specifi c-ity was 55% (95% confi dence interval 53.7%-55.6%). The sensitivity and specifi city of this case defi nition did not show signifi cant variation by age, sex, and year of reporting (Table 1). However, the positive predictive value was signifi cantly higher for persons >15 years of age (71% for persons 15-24 years of age compared with 1% for persons 5-9 years of age), for men (21% compared with 5% for women), and during the 1995-1998 epidemic period (20% compared with 1% in January 1999-July 2003).

Enhanced Surveillance, 1999-2008
During 1999-2008, a total of 13,952 clinically suspected rubella cases were reported, and the number of cases per year (1,000-2,000) remained stable (Figure). Oral fl uid was tested for 11,709 (84%) case-patients; 143 (1.0%) positive results were confi rmed, and 11,566 (99%) of the results were discarded ( Table 2) This fi nding is equivalent to an annual investigation and discard rate of 2,208 clinically suspected rubella cases per 100,000 population. Over the 10-year period, the proportion of confi rmed cases for which oral fl uid was tested increased from 39% (49/127) in 1999 to 49% (16/33) in 2008. The annual number of cases confi rmed positive by oral fl uid test remained <30 (Figure). The proportion of cases with a confi rmatory test result has decreased since 1999 and remained <2%, except for a temporary increase in 2007 to 2.8%. During the 10-year period, 263 additional cases were confi rmed by serum testing, resulting in 406 confi rmed cases compared with 13,952 clinically diagnosed cases. Because laboratories in England and Wales report only cases that were confi rmed positive, the number of additional cases tested and results discarded after serum testing alone is not known.
The age and sex distribution of case-patients reported from each surveillance source differed markedly ( Table 2). The proportion of reported cases for which oral fl uid was tested was higher for children <15 years of age (10, Reports of rubella showed an approximately equal sex distribution among persons <15 years of age but showed a slight preponderance of female patients among adults ( Table 2). The proportion of reported cases for which oral fl uid was tested did not differ by sex ( Regional distribution varied by data source (Table 3). The annual incidence based on reported cases was highest in the North East region and lowest in Wales, whereas the annual incidence based on laboratory reports of serum confi rmation was highest in London. Incidence based on oral fl uid test results also differed, even after adjustment for the proportion of reported cases tested. The estimated overall incidence of confi rmed rubella increased by 54% (from 0.50 cases/1 million population to 0.77 cases/1 million population) when data for oral fl uid testing were included. Oral fl uid data also changed the ranking of regions; the Eastern region overtook London in reporting the highest overall incidence. Although the West Midlands and Yorkshire and Humberside regions reported the lowest incidence on the basis of serum testing alone, after oral fl uid testing was included, the East Midlands region reported the lowest overall incidence.

Discussion
Before vaccination was introduced, epidemics of rubella occurred regularly and caused mild rash illness, predominantly in children. In 1970, introduction of a selective vaccination policy in the United Kingdom aimed to reduce the risk for infection in early pregnancy and the risk for fetal death and CRS. Despite the success of the selective policy, MMR was adopted into the routine childhood vac-cination schedule to eliminate circulating rubella and to further reduce the risk for CRS.
Since 1988, a clinical diagnosis of rubella has been reportable by registered medical practitioners in England and Wales under the statutory Notifi cation of Infectious Diseases; there is no case defi nition. When an infection is commonly occurring in an area, the positive predictive value of a clinical diagnosis may be suffi cient for accurate surveillance (24). However, because rubella has become less common, an increasing proportion of reported cases are likely to be caused by other infections that have similar clinical manifestations. The rash of rubella may be temporary and can resemble the rash caused by other viruses. For example, infection with parvovirus B19, human herpesvirus 6 (roseola infantum), and human herpesvirus 7 all involve rash and fever and may be misdiagnosed as rubella (25,26).
We have confi rmed the low sensitivity and specifi city of the clinical case defi nition and that this defi nition is not affected by age, sex, and period of reporting. Despite some missing clinical information and the absence of information about arthralgia, this fi nding suggests that the WHO clinical case defi nition is not suffi ciently accurate for surveillance in the postvaccine era. This fi nding also emphasizes the need to have laboratory confi rmation of all clinically diagnosed cases to accurately monitor rubella elimination (27).
Over the 10-year period of elimination, only ≈1 of 100 persons reported with clinically diagnosed rubella and who underwent oral fl uid testing had confi rmed cases. In addition, reported cases differed from laboratory-confi rmed cases with respect to patient age, sex, vaccination status, and geographic distribution. We have therefore shown that surveillance based only on clinical reports would substantially overestimate the true incidence of rubella, particularly in children, and therefore give a misleading epidemiologic picture. Furthermore, we have shown that testing of oral fl uid is acceptable in the United Kingdom and can be used to augment routine serologic diagnosis. Approximately one third of confi rmed rubella cases were diagnosed by testing of oral fl uid, which improved ascertainment of confi rmed infections in children and men. In addition, by using  PCR, we obtained genotype information on 12 samples that would have otherwise not been available. Currently, there is only 1 commercial assay for testing rubella IgM in oral fl uid, and this assay does not have an In Vitro Diagnostics license, thus limiting its use in some countries. However, in many regions, WHO is evaluating this assay as a tool for surveillance of infection (28; K. Brown, pers. comm.) Although the UK system relies upon a well-organized postal service, alternative approaches for delivering and receiving specimens directly from the patient's home or family practice may be required in other countries considering the use of oral fl uid.
In addition to antibody testing, oral fl uid can be used for rubella RNA detection in samples obtained early during infection (13) and for genotyping and molecular epidemiologic studies (14,15). Tests for detecting rubella-specifi c IgM and RNA in oral fl uid samples are also suitable for confi rming a diagnosis of CRS (11). The same system is also being used to monitor measles and mumps incidence and to inform MMR vaccine policy in England and Wales (18,19,29,30).
Oral fl uid testing can also be used to evaluate the completeness of rubella surveillance. Since 1999, the number of reports of rubella and oral fl uid tests performed annually has remained constant, and the proportion of cases tested has remained high. When combined with a low rate of confi rmation, including cases diagnosed as clinical rubella, in a country with free universal access to high-quality primarycare physicians, this fi nding suggests that surveillance of confi rmed rubella is nearly complete in England and Wales. Despite this suggestion, annual incidence of confi rmed rubella remains <1 per 1 million population, which is the goal for elimination (4).
In recently published surveillance guidelines, WHO has recommended IgM detection, which can be performed with serum and oral fl uid (31). These guidelines also describe performance indicators to assess the quality of national surveillance systems in the elimination phase. These indicators include a laboratory investigation rate (proportion of clinically suspected cases with adequate specimens for IgM testing) >80% and a detection rate for the number of clinically suspected rubella cases investigated and discarded by laboratory testing >2/100,000 population/year. Data from the enhanced surveillance program show that for ≈84% of reported cases, oral fl uid was tested, and results for 2,208 clinically suspected rubella cases per 100,000 population were investigated and discarded. This high discard rate would be feasible only with noninvasive testing and contributes to the high quality of the UK enhanced surveillance program. Information about the low rate of rubella supplements surveillance that confi rms that CRS incidence in the United Kingdom was 0.14/100,000 live-born infants in 2007 (32), which was far below the WHO elimination goal of 1/100,000 live-born infants (4).
We confi rmed that a clinical case defi nition alone is not suffi ciently specifi c for surveillance of rubella in the elimination era and that laboratory confi rmation by testing serum samples is biased and incomplete. Since 1999, a substantial proportion of confi rmed cases of rubella identifi ed through the enhanced surveillance scheme have occurred in unvaccinated men. With the availability of oral fl uid testing, a high number and high proportion of suspected cases have been tested. However, numbers of confi rmed rubella cases in children remain low, which is consistent with high levels of vaccine coverage and low levels of susceptibility in this younger age group (33). The enhanced oral fl uid surveillance system has proven valuable for accurately assessing progress toward achieving the WHO goal of eliminating circulating rubella and CRS from the population of England and Wales.