A point-of-care test for measles diagnosis: detection of measles-specific IgM antibodies and viral nucleic acid
Lenesha Warrener a, Rimantas Slibinskas b, Kaw Bing Chua c, Wondatir Nigatu d, Kevin E Brown a, Kestutis Sasnauskas b, Dhanraj Samuel a & David Brown a
a. Virus Reference Department, Centre for Infections, Health Protection Agency, 61 Colindale Avenue, London, NW9 5EQ, England.
b. Institute of Biotechnology of Vilnius University, Vilnius, Lithuania.
c. National Public Health Laboratory, Selangor, Malaysia.
d. Ethiopian Health and Nutrition Research Institute, Addis Ababa, Ethiopia.
Correspondence to Lenesha Warrener (e-mail: firstname.lastname@example.org).
(Submitted: 04 April 2011 – Revised version received: 27 June 2011 – Accepted: 27 June 2011 – Published online: 05 July 2011.)
Bulletin of the World Health Organization 2011;89:675-682. doi: 10.2471/BLT.11.088427
Measles is a severe, vaccine-preventable disease that causes extensive morbidity and mortality in large parts of the world. Despite the widespread use of measles vaccine, either as a single antigen vaccine or as a component of the triple vaccine against measles, mumps and rubella (MMR), 278 358 reported cases of measles and an estimated 164 000 deaths from measles occurred worldwide in 2008.1 Vaccine coverage is highly variable between World Health Organization (WHO) global regions. Measles has been eliminated in the Americas but continues to be endemic in the African and South-East Asia regions, where vaccine coverage is less than 80%. These regions account for approximately 94% of all global measles deaths.1 Outbreaks continue to occur in other global regions, primarily as a result of measles virus importation into areas where vaccine coverage has fallen to a suboptimal level and a susceptible cohort has accumulated.2–4 Most regions have elimination goals and elimination strategies based on the maintenance of high vaccination coverage, for which political commitment is required. A key component of elimination plans is surveillance to monitor impact.
Laboratory confirmation of cases of measles is a vital aspect of surveillance at all stages of control programmes because clinical diagnosis is unreliable.5 The mainstay of laboratory confirmation is the detection of measles-specific immunoglobulin M (IgM) antibodies in serum samples.6,7 More recently, alternative samples such as dried blood spots and oral fluids have been used for diagnosis by antibody detection. Oral fluids can also be used to detect viral ribonucleic acid (RNA) and their use is becoming increasingly common because samples can be obtained safely and non-invasively, without the risks associated with blood collection, and it improves patient compliance with specimen collection, as the procedure is simple and painless.7–10
Point-of-care tests (POCTs) are increasingly used for the rapid diagnosis of infections. They can be performed in a single incubation step at ambient temperature without complex electrical equipment and their results can be read visually. By increasing diagnostic capacity and facilitating rapid diagnosis in resource-poor countries, they have the potential to improve measles surveillance and the response of health authorities to possible outbreaks.
We have developed a POCT capable of detecting measles-specific IgM antibodies in both serum and oral fluid specimens. In this paper we describe the diagnostic performance of this POCT for each specimen type. We also investigated whether viral RNA could be amplified from the used test strips, as that would enhance their use in measles surveillance.
The POCT strips for the detection of measles-specific IgM were constructed essentially as described for mumps-specific IgM POCT strips,11 except that affinity purified F(ab’)2 fragment goat anti-human IgM, 1.2mg/ml (Jackson Immunoresearch Laboratories, Inc., West Grove, United States of America) was used as the test line capture antibody, and monoclonal antibody to recombinant measles virus nucleoprotein12 was conjugated to 40 nm gold colloid (BBInternational Ltd, Cardiff, England) to produce the measles-specific gold conjugate (MAb-gold conjugate). Assembled POCT strips are illustrated in Fig. 1.
Fig. 1. Components of the point-of-care test (POCT) and examples of signal intensities obtained during POCT testing of sera from four patients and of the cut-off control serum
Note: Each POCT strip measures 60 × 5 mm. The photograph shows: an accuflow G glass fibre conjugate pad, on which MAb-gold conjugate is localized before use; the nitrocellulose membrane with an immobilized anti-human IgM antibody test line (T) and an anti-mouse IgG antibody control line (C) positioned approximately 14 mm and 18 mm from the conjugate pad, respectively (not visible before testing); and a cotton linter paper wick. The five POCT strips demonstrate from left to right the results of testing the cut-off control serum (CO), one negative (strip 1, enzyme immunoassay (EIA) T/CO of 0.34) and three positive specimens of varying intensity (strips 2–4, EIA T/COs of 5.4, 1.7 and 5.5, respectively).
Recombinant measles virus nucleoprotein (rNP), expressed in yeast Pichia pastoris and purified by caesium chloride ultracentrifugation,13 was diluted to 3.5 μg/ml in oral fluid extraction diluent, prepared as described previously.14 A cut-off serum giving a weakly reactive anti-human IgM antibody test line signal in the POCT was prepared. This was done by diluting one part of a serum that tested positive for measles IgM on capture enzyme immunoassay (EIA) in 20 parts of a measles-IgM-negative human serum. The cut-off serum was tested in parallel with patient sera.
The cut-off serum and patient sera were diluted 1 part in 100 in oral fluid extraction diluent and 100 μl of the resulting dilution were mixed with 5 μl of the 3.5 μg/ml rNP in a plastic tube. POCT strips were inserted in the serum– rNP mixture to initiate the flow of reagents and were incubated at room temperature for 10 minutes. Test results were interpreted by comparing the intensity of the pink to red signal generated at the test line with patient sera, with that of the signal generated with the cut-off serum run in parallel. If the test line signal was more intense in colour than the signal obtained with the cut-off serum, the test was considered positive; if the test line signal was less intense in colour than the signal obtained with the cut-off serum or if both signals were equally intense in colour, the test was considered negative (Fig. 1).
Oral fluid protocol
The POCT protocol used for oral fluid testing was essentially the same as the one used to test sera, except that oral fluid specimens were tested undiluted, rNP was used at 35 μg/ml and the test strips were incubated for 20 minutes at room temperature. A cut-off control serum was not required for interpreting the results since background signals were not observed when testing oral fluid specimens. If the signal at the anti-human IgM test line ranged from pink to red in colour, the test was interpreted as positive for measles-specific IgM; if no signal was visible at the test line, the test was interpreted as negative for measles-specific IgM. For both serum and oral fluid testing, a signal had to be visible at the control line for the test to be considered valid.
Test performance evaluation
The performance of the POCT for the detection of measles-specific IgM in both serum and oral fluid specimens was evaluated by comparing the results with those obtained with the measles IgM capture EIA (Microimmune Ltd, Hounslow, England). Two individuals interpreted the POCT results independently at the specified incubation times. The results were then classified as positive or negative for measles-specific IgM by consensus. When only one reader observed a weak signal, the result was interpreted as negative. The EIA was performed and the results, expressed as a numerical value (quotient) calculated by dividing the optical density obtained for each test specimen (450/620 nm) by the cut-off value (T/CO), were interpreted in accordance with the manufacturer's instructions.
A total of 170 serum specimens were evaluated; 100 of them had been collected from suspected cases of measles in Malaysia in 2004 (age range: 6 months to 42 years; mean: 12.4), when outbreaks were reported. The specimens originated from four federal states within Malaysia: 83 from Selangor, 11 from Melaka, 5 from Perak and 1 from Pahang.
Sixty-two serum specimens had been collected from children (age range: 6 months to 17 years; mean: 5.8) who presented with a rash during an outbreak of rubella in Ethiopia in 2003–2004. Of these children, 42 had been vaccinated against measles. Specimens were collected a mean of 7 days (range: 1–23) after the onset of the rash.
Eight serum specimens from acute or suspected cases of rubella in the Russian Federation were kindly donated by the World Health Organization’s Regional Office for Europe, in Denmark. All sera were also tested by rubella IgM capture EIA (Microimmune Ltd).
Oral fluid specimens
All oral fluids from suspected measles and rubella cases received by the Virus Reference Department of the Health Protection Agency of the United Kingdom of Great Britain and Northern Ireland over a 10-day period in July 2008 were used to evaluate the POCT. The specimens had been collected as part of the national MMR surveillance programme using Oracol swabs (Malvern Medical Developments Ltd, Worcester, England) and extracted as described previously.14 Specimens included 232 oral fluids for measles investigation, 11 for both measles and rubella testing and 39 for rubella investigation only (age range of all oral fluids: 1 month to 59 years; mean: 5.9).The 50 oral fluids submitted for rubella investigation were tested by EIA for the detection of measles IgM as well as by Microimmune rubella IgM capture EIA.
Virus nucleic acid amplification
An additional 24 oral fluid specimens received for measles surveillance testing in September 2010 were used to evaluate the extraction and amplification of viral genome from POCT strips after IgM detection. Following routine surveillance testing, the specimens were stored for up to 6 weeks at 2–8 °C before this evaluation.
To investigate the stability of measles virus nucleic acid on POCT strips, the oral fluids were tested in duplicate by POCT as described earlier, then dried and stored at 20–25 °C. Nucleic acid was extracted and complementary DNA (cDNA) was prepared as described previously,11 from an aliquot of each oral fluid on the day of point-of-care testing, from one set of dried POCT strips after overnight storage and from a second set of dried strips after 5 weeks of storage. Measles virus cDNA amplicons were generated by real-time polymerase chain reaction (PCR) for the haemagglutinin (H) gene15 and by nested PCR targeting the nucleocapsid (N) gene.16 Amplified N-gene products were sequenced in forward and reverse directions by dideoxynucleotide sequencing in a Genetic Analyser (Applied Biosystems, Foster City, USA), analysed with Bionumerics software version 6.1 (Applied Maths NV, Sint-Martens-Latem, Belgium) and genotyped by comparison with reference strain sequences according to WHO guidelines.17–19
Concordant results were obtained for 157 (92.4%) of the 170 sera tested by measles-specific IgM capture EIA and POCT (Table 1). Of 76 sera that tested positive for measles-specific IgM in EIA, 69 also tested positive by POCT. This included 5 sera that also tested positive for rubella-specific IgM. Of the 94 sera that tested negative for measles-specific IgM by EIA, 88 were also negative by POCT, including 69 that had tested positive for rubella-specific IgM. Two of the 7 sera that tested positive for measles-specific IgM by capture EIA and negative by POCT had low T/COs (1.5 and 1.9) in the EIA. They had been collected 1 and 5 days, respectively, after the onset of fever. Of 74 sera that tested positive for rubella-specific IgM only, 69 tested negative by the POCT. Thus, with this set of samples the POCT had a specificity of 93.2%.
Table 1. Comparison of the results obtained for 170 serum specimens with point-of-care test (POCT) for the detection of measles-specific IgM and with measles- and rubella-specific IgM capture enzyme immunoassays (EIAs)
The overall performance of the POCT with all serum samples was as follows: sensitivity: 90.8% (95% confidence interval, CI: 81.94–96.22); specificity: 93.6% (95% CI: 86.62–97.62); positive predictive value: 92.0% (95% CI: 83.4−97.01); negative predictive value: 92.6% (95% CI: 85.41−96.99).
The performance of the POCT on oral fluids was evaluated using a total of 282 specimens received for measles and rubella investigations (Table 2). Of the 50 oral fluids submitted for rubella investigation, 48 tested negative for both measles- and rubella-specific IgM by EIA. One oral fluid tested positive for rubella-specific IgM and one other tested positive for measles-specific IgM by EIA.
Table 2. Comparison of the results obtained for 282 oral fluid specimens with point-of-care test (POCT) for the detection of measles-specific IgM and with Microimmune measles-specific IgM capture enzyme immunoassay (EIA)
Concordant POCT and EIA results were obtained for 263 of the 278 (94.6%) oral fluids after excluding the four specimens that gave equivocal results on the measles-specific IgM capture EIA. The overall performance of the POCT with all oral fluids was as follows: sensitivity: 90.0% (95% CI: 80.5–95.9); specificity, 96.2% (95% CI: 92.6 – 98.3%); positive predictive value: 88.7% (95% CI: 79.0 – 95.0%); negative predictive value: 96.6% (95% CI: 93.2 – 98.6%).
The POCT data were further analysed as a function of T/CO on EIA and of the timing of specimen collection following the onset of symptoms (Table 3). Approximately 20% of the specimens were collected within 7 days of the onset of symptoms; 25% were collected between 7 and 14 days after onset and 31% were collected after the 14th day. The timing of specimen collection in relation to the onset of symptoms was not known for 24% of oral fluids. Seven out of the 70 oral fluids that tested positive by EIA gave discrepant POCT results. All such specimens had T/COs on EIA just above the positive cut-off value of 1.0 (range: 1.05–1.94). Collection dates were known for six of these specimens. Five of them had been collected more than 14 days after the onset of symptoms (range: 19–38). The sixth oral fluid had been collected three days after the onset of symptoms. However, this patient had also received a single dose of measles-virus containing vaccine 32 days before the specimen was collected.
Table 3. Results obtained for oral fluid specimens with point-of-care test (POCT) for the detection of measles-specific IgM, analysed with respect to the result quotient (T/CO) obtained in Microimmune measles-specific IgM capture enzyme immunoassay (EIA) and timing of specimen collection
Table 3 also suggests that the detection of IgM by POCT may depend on the quantity of specific IgM antibody present. All oral fluids in this panel with a T/CO of 5 or higher on IgM capture EIA were identified by POCT, regardless of the timing of specimen collection.
Virus nucleic acid amplification
When measles H gene real-time PCR was performed on oral fluid aliquots and their respective POCT strips, concordant results were obtained in 23 of 24 cases (Table 4). A comparison of the threshold cycle (Ct) values obtained in the H-gene real-time PCR for oral fluids with those obtained for their matched POCT strips showed little difference between the two in the recovery of viral nucleic acid.
Table 4. Results of polymerase chain reaction (PCR) and IgM antibody testing obtained for 24 oral fluid (OF) specimens used to compare viral nucleic acid recovery from OF aliquots and from used point-of-care test (POCT) strips
The H gene was detected in 16 oral fluids and in 15 of the 16 corresponding POCT strips. The measles virus N gene was amplified by nested PCR from 14 of these oral fluids and the corresponding POCT strips. The N gene was also amplified from the oral fluid aliquot and POCT strip of specimen 20, although the H gene had not been detected in either. Nucleic acid sequence analysis of the nested N-gene PCR amplicons identified viral genotype D4 in 14 of these oral fluids and viral genotype D9 in one. The genotypes and individual strains identified from the corresponding POCT strips were identical to those identified from the oral fluids (Table 4).
The remaining nine oral fluid aliquots were negative for the N gene by nested PCR. However, when nested PCR was performed on the corresponding POCT strips, N-gene amplicons were generated on five occasions. These amplicons were not derived from sequences in the specimens, but from the rNP that had been added during IgM detection, which contained trace levels of residual N-gene sequence that amplified sporadically in the absence of specimen-derived N-gene target. The fact that the contaminating N-gene sequence was not detected after one round of N-gene PCR confirms that it was present in minute quantities. The generated amplicons were confirmed to have the same Schwarz vaccine strain sequence that was used to synthesize the rNP antigen.
The measles virus N gene was amplified from oral fluid 23 and identified as a Schwarz vaccine strain during the routine surveillance testing performed before this study, but it was not detected in the oral fluid during this evaluation. Although the same vaccine strain was identified from the corresponding POCT strip, it was not possible to determine whether this originated from the oral fluid specimen or from the added rNP.
POCT results for this small panel of oral fluids were highly concordant with the results of measles-specific IgM EIA, as were the results obtained with the larger panel of oral fluids. Measles-specific IgM was not detected by POCT in only one EIA-positive oral fluid, specimen 19 (Table 4). In this case, measles was confirmed by PCR.
Stability of measles nucleic acid on POCT strips
The measles virus H gene was amplified from 16 POCT strips stored at 20–25 °C for 5 weeks. This included all 15 specimens in which the H gene was amplified from the first POCT strip as well as specimen 20, in which the H gene was not detected initially. The measles virus N gene was amplified from 12 of these 16 POCT strips after 5 weeks of storage. The nucleotide sequences obtained from 11 of these stored strips were identical to those identified from the original POCT strips. Measles viral genotype A, consistent with the nucleic acid sequence present in the rNP antigen, was identified from the stored POCT strip of specimen 20; viral genotype D4 had been previously identified in this specimen.
Sensitive and specific tests for measles diagnosis that can be used in low-resource settings can greatly enhance surveillance. Field-based tests would facilitate the diagnosis of measles in patients with a rash illness accompanied by fever, as well as the rapid implementation of control measures. With these objectives in mind we developed a rapid POCT for the detection of measles-specific IgM based on the principle of immunochromatographic flow. In this study, the POCT showed good sensitivity, specificity and positive and negative predictive values (> 88%) and is therefore suitable for field use. However, the potential role of POCT in measles surveillance will remain unknown until wider field-based evaluations are conducted in a range of settings, including those where control activities are in the elimination phase.
An added advantage of the POCT, apart from its ability to detect measles-specific IgM, is that the test strip is an excellent matrix for capturing viral nucleic acid from oral fluid samples and preserving it for up to 5 weeks at ambient temperature. This not only makes it possible to diagnose measles cases immediately, but also permits subsequent molecular characterization of the virus at national laboratories to track transmission pathways and confirm the diagnostic accuracy of the POCT in the field.
The results obtained with the larger panel of oral fluids suggest that IgM detection by POCT improves as the concentration of IgM antibody increases, since all specimens with an EIA T/CO greater than 5 were identified (Table 3). However, one oral fluid specimen, number 19 in Table 4, was highly reactive in the measles-specific IgM capture EIA but was negative by POCT. That oral fluid was also found to have a high concentration of measles-specific IgG by EIA – a T/CO of 62 – but remained IgM-negative by POCT after IgG was absorbed from the sample using protein A sepharose (data not shown). On dilution of the sample one part in three and one part in six, a positive IgM result was obtained in the POCT. The total measles-specific antibody concentration in this sample may have been so high that it bound the entire rNP added, leaving no sites available for the MAb-gold conjugate to bind and form a complex that could be captured at the anti-human IgM antibody line in the POCT. Other oral fluids with high IgM and IgG EIA T/CO values were positive by POCT. This suggests that the failure to detect IgM in specimen 19 may not be simply due to a high concentration of antibody. Alternatively, some oral fluids may contain inhibitors that require dilution to enable complex formation with both specific antibody and the MAb-gold conjugate to occur. The antibody-negative result did not preclude the diagnosis of measles, since H and N genes were amplified from the POCT strips used for oral fluid testing.
The presence of trace levels of measles N-gene sequences from the rNP used in the POCT could have interfered with the amplification of the N-gene target in the oral fluids tested by POCT, but in most cases it did not. We surmise that the viral nucleic acid sequences in oral fluids from acute measles cases were present in greater quantity and were amplified preferentially from the POCT strips. This was confirmed by the identification of the same genotype that was found in the oral fluid aliquots. The possibility of removing the contaminating nucleic acid in the rNP by deoxyribonuclease and ribonuclease treatment is currently being investigated. However, the amplification of the Schwarz vaccine strain N-gene sequence from the rNP is not likely to seriously impair the identification of current wild-type measles strains, as genotype A viruses have not circulated widely in the past 20 years and the few sporadic cases of infection with these viruses have been frequently associated with recent vaccination.20
In its present configuration the POCT relies on some basic laboratory equipment. A robust field-based test would require further development of methods for extracting oral fluids from swabs without centrifugation, the use of disposable, volumetric pipettes or loops to deliver accurate volumes, and a plastic cassette to enclose the test strip and reduce the handling of the POCT’s analytical membrane. The estimated cost of manufacturing one POCT is approximately one United States dollar.
The authors would like to acknowledge the valuable contributions to this study of Pravesh Dhanilall of the Virus Reference Department of the Health Protection Agency, United Kingdom, and Joe Vincini of the Department of Evaluations, Standards and Training of the Health Protection Agency, United Kingdom.
Dhanraj Samuel is also a Director of Microimmune Limited, Hounslow, United Kingdom.
This project was funded by the Health Protection Agency, United Kingdom.
- Centers for Disease Control and Prevention. Global measles mortality, 2000–2008. MMWR Morb Mortal Wkly Rep 2009; 58: 1321-6 pmid: 19959985.
- Measles Mortality Reduction and Regional Elimination Strategic Plan 2001–2005. Geneva: World Health Organization; 2001.
- Choi YH, Gay N, Fraser G, Ramsay M. The potential for measles transmission in England. BMC Public Health 2008; 8: 338- doi: 10.1186/1471-2458-8-338 pmid: 18822142.
- Vyse AJ, Gay NJ, Hesketh LM, Pebody R, Morgan-Capner P, Miller E. Interpreting serological surveys using mixture models: the seroepidemiology of measles, mumps and rubella in England and Wales at the beginning of the 21st century. Epidemiol Infect 2006; 134: 1303-12 doi: 10.1017/S0950268806006340 pmid: 16650326.
- Featherstone D, Brown D, Sanders R. Development of the global measles laboratory network. J Infect Dis 2003; 187: S264-9 doi: 10.1086/368054 pmid: 12721924.
- Manual for the laboratory diagnosis of measles and rubella virus infection. 2nd ed. Geneva: World Health Organization; 2007.
- Centers for Disease Control and Prevention. Recommendations from an ad hoc Meeting of the WHO Measles and Rubella Laboratory Network (LabNet) on use of alternative diagnostic samples for measles and rubella surveillance. MMWR Morb Mortal Wkly Rep 2008; 57: 657-60 pmid: 18566565.
- World Health Organization. Measles and rubella laboratory network: 2007 meeting on use of alternative sampling techniques for surveillance. Wkly Epidemiol Rec 2008; 83: 229-32 pmid: 18572479.
- Nigatu W, Jin L, Cohen BJ, Nokes DJ, Etana M, Cutts FT, et al., et al. Measles virus strains circulating in Ethiopia in 1998–1999: molecular characterisation using oral fluid samples and identification of a new genotype. J Med Virol 2001; 65: 373-80 doi: 10.1002/jmv.2044 pmid: 11536247.
- Jin L, Vyse A, Brown DW. The role of RT-PCR assay of oral fluid for diagnosis and surveillance of measles, mumps and rubella. Bull World Health Organ 2002; 80: 76-7 pmid: 11884979.
- Warrener L, Slibinskas R, Brown D, Sasnauskas K, Samuel D. Development and evaluation of a rapid immunochromatographic test for mumps-specific IgM in oral fluid specimens and use as a matrix for preserving viral nucleic acid for RT-PCR. J Med Virol 2010; 82: 485-93 doi: 10.1002/jmv.21693 pmid: 20087926.
- Samuel D, Sasnauskas K, Jin L, Gedvilaite A, Slibinskas R, Beard S, et al., et al. Development of a measles specific IgM ELISA for use with serum and oral fluid samples using recombinant measles nucleoprotein produced in Saccharomyces cerevisiae. J Clin Virol 2003; 28: 121-9 doi: 10.1016/S1386-6532(02)00273-1 pmid: 12957182.
- Slibinskas R, Samuel D, Gedvilaite A, Staniulis J, Sasnauskas K. Synthesis of the measles virus nucleoprotein in yeast Pichia pastoris and Saccharomyces cerevisiae. J Biotechnol 2004; 107: 115-24 doi: 10.1016/j.jbiotec.2003.10.018 pmid: 14711495.
- Vyse AJ, Knowles WA, Cohen BJ, Brown DW. Detection of IgG antibody to Epstein-Barr virus viral capsid antigen in saliva by antibody capture radioimmunoassay. J Virol Methods 1997; 63: 93-101 doi: 10.1016/S0166-0934(96)02118-0 pmid: 9015279.
- Thomas B, Beard S, Jin L, Brown KE, Brown DWG. Development and evaluation of a real-time PCR assay for rapid identification and semi-quantitation of measles virus. J Med Virol 2007; 79: 1587-92 doi: 10.1002/jmv.20997 pmid: 17705181.
- Kühne M, Brown DW, Jin L. Genetic variability of measles virus in acute and persistent infections. Infect Genet Evol 2006; 6: 269-76 doi: 10.1016/j.meegid.2005.08.003 pmid: 16172023.
- World Health Organization. Nomenclature for describing the genetic characteristics of wild-type measles viruses (update). Part I. Wkly Epidemiol Rec 2001; 76: 242-7 pmid: 11515240.
- World Health Organization. New genotype of measles virus and update on global distribution of measles genotypes. Wkly Epidemiol Rec 2005; 80: 347-51 pmid: 16240986.
- World Health Organization. Expanded Programme on Immunization (EPI). Standardization of the nomenclature for describing the genetic characteristics of wild-type measles viruses. Wkly Epidemiol Rec 1998; 73: 265-9 pmid: 9745371.
- Sanders R. Risk analysis for measles reintroduction post global certification of eradication. Geneva: World Health Organization; 2010. Available from: http://www.who.int/immunization/sage/7._Measles_post_eradication_risk_analysis.pdf [accessed 25 Feb 2011].