* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Download Lack of Evidence of Measles Virus Shedding in People with
Schistosomiasis wikipedia , lookup
Oesophagostomum wikipedia , lookup
Herpes simplex wikipedia , lookup
2015–16 Zika virus epidemic wikipedia , lookup
Influenza A virus wikipedia , lookup
Neonatal infection wikipedia , lookup
Ebola virus disease wikipedia , lookup
Hepatitis C wikipedia , lookup
Orthohantavirus wikipedia , lookup
Hospital-acquired infection wikipedia , lookup
Human cytomegalovirus wikipedia , lookup
Antiviral drug wikipedia , lookup
Middle East respiratory syndrome wikipedia , lookup
Marburg virus disease wikipedia , lookup
West Nile fever wikipedia , lookup
Herpes simplex virus wikipedia , lookup
Hepatitis B wikipedia , lookup
Lymphocytic choriomeningitis wikipedia , lookup
SUPPLEMENT ARTICLE Lack of Evidence of Measles Virus Shedding in People with Inapparent Measles Virus Infections Fabio A. Lievano,1 Mark J. Papania,1 Rita F. Helfand,2 Rafael Harpaz,1 Laura Walls,2 Russell S. Katz,2 Irene Williams,2 Yvonne S. Villamarzo,2 Paul A. Rota,2 and William J. Bellini2 1 Measles Elimination Activity, Child Vaccine Preventable Diseases Branch, Epidemiology and Surveillance Division, National Immunization Program, and 2Measles Virus Section, Respiratory and Enteric Viruses Branch, Division of Viral and Rickettsial Diseases, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia Serological evidence of measles virus infection has been detected among people exposed to measles who do not exhibit classical clinical symptoms. Throat swabs, lymphocytes, and serum and urine samples were collected from contacts of individuals with confirmed measles 12–16 days after exposure, during measles outbreaks occurring in 1998. Follow-up serum samples were drawn 2 weeks later. Samples were tested for measles IgM antibody by enzyme immunoassays and plaque reduction neutralization testing. Virus isolation and reverse transcriptase–polymerase chain reaction testing was attempted for all samples. None of the 133 contacts developed classical measles disease; 11 (8%) had serological evidence of infection. Duration of exposure of ⭓3 h was the only significant risk factor for developing serological response (24% vs. 4% among contacts exposed for 1–2 h; relative risk, 6.0; 95% confidence interval, 1.9–19.2). None of the 133 contacts had virological evidence of infection by culture or polymerase chain reaction. We found no evidence that persons with inapparent measles virus infections shed measles virus. Measles control strategies are based on the assumption that measles virus transmission occurs in chains of transmission of clinically recognizable measles cases. Surveillance systems rely on the identification of persons with the clinically recognizable symptoms of measles for detecting and responding to outbreaks, vaccinating susceptible contacts, and assessing the efficacy of vaccines and the impact of vaccination programs. However, it has been postulated that the occurrence of measles virus infections in persons without classical measles symptoms may play an important role in the transmission of measles. Serological evidence of acute measles has been documented among people who are exposed to measles virus but do not develop classical symptoms [1–10]. This phenomenon has been observed most frequently in highly vaccinated populations with rates of infection of 4%–42% [2, 4, 5, 8, 9]. Inapparent measles virus infections are clinically un- Reprints or correspondence: Dr. Fabio A. Lievano, Centers for Disease Control and Prevention, 1600 Clifton Rd., Mailstop E-05, Atlanta, GA 30333 (flievano@ cdc.gov) The Journal of Infectious Diseases 2004; 189(Suppl 1):S165–70 This article is in the public domain, and no copyright is claimed. 0022-1899/2004/18909S1-0025 important, because there are no symptoms or only mild symptoms. However, inapparent infections do provide a boost in measles antibody titers, which some investigators have suggested provide increased protection against future exposures to measles [9]. However, this increased protection is probably not necessary. People with inapparent infections, by definition, have been exposed to measles virus and did not develop measles disease. Several different terms have been used to describe measles virus infections in persons without classical symptoms, including subclinical measles [7, 9], asymptomatic measles [2, 4, 5, 6], modified measles [1, 4, 8], mild measles [1, 5, 6], and secondary immune response [2, 3, 4, 6]. Herein we use the term “inapparent infection” to indicate serological evidence of measles virus infection in any person who does not meet the clinical case definition of measles (the clinical case definition of measles is a case of generalized maculopapular rash lasting for ⭓3 days, with fever of ⭓38.3C, and cough, coryza, or conjunctivitis [11]). Inapparent measles virus infections could be epidemiologically important if infected persons are capable of transmitting measles virus. Many countries, includShedding and Inapparent Infection • JID 2004:189 (Suppl 1) • S165 ing the United States, have the elimination of measles as a public health goal [12]. Recent mathematical models have suggested that eliminating measles would be impossible if measles virus can be transmitted in the absence of classical measles disease [13]. However, there is very little information available regarding the infectivity of persons with inapparent measles virus infections. There are no studies of persons with inapparent measles virus infections that document respiratory shedding of measles virus, which is a presumed prerequisite for transmitting the virus to other people. In 1999, Vardas et al. [6] reported that measles virus was isolated from urine of an asymptomatic contact of a person with measles. The nucleotide sequence of the measles virus isolated from the contact was identical to that of the virus isolated from the individual with measles. Although genotypically identical viruses were recovered, the authors could not rule out cross-contamination. To estimate the potential for measles virus transmission from persons with inapparent infections, we studied a cohort of people exposed to confirmed measles cases. We collected blood, urine, and respiratory specimens to identify those with inapparent infections and to determine whether any of the exposed cohort had detectable measles virus. METHODS Study population. Persons exposed to confirmed measles cases during outbreaks in the fall of 1998 in Anchorage [14] and Phoenix were contacted to identify those who met the following study criteria: contact with a person with confirmed measles for ⭓1 h in the same room during the infectious period (from 3 days before rash onset through 3 days after rash onset), availability for specimen collection at the times defined by the protocol, and no receipt of measles vaccine and no diagnosis of measles within 2 months before specimen collection. Eligible contacts were invited to participate in the study. Written consent was obtained from participants, and the protocol was approved by the Institutional Review Board of the Centers for Disease Control and Prevention (CDC). We collected demographic information; previous history of measles, rubella, or chickenpox; and vaccination history. We also collected details of the exposure and symptoms experienced after exposure. We obtained samples of blood and urine and throat swabs from the participants 12–16 days after their exposure to measles. Viral shedding is most intense in a classical measles case during this period [12]. We estimated that this period would give us the highest likelihood of detecting measles virus in this cohort. Two days later, throat swabbing was repeated to increase the probability of detecting pharyngeal shedding of measles virus. About 14 days after the collection of the initial blood sample, a follow-up blood sample was drawn. All specimens were shipped immediately in a cold box with ice S166 • JID 2004:189 (Suppl 1) • Lievano et al. packs to Atlanta and processed immediately upon arrival at the Measles Laboratory, CDC. Laboratory analysis. All serum specimens were tested for the presence of IgM antibodies as evidence of recent infection by means of EIA capture test [15–17]. Plaque reduction neutralization (PRN) assays were done on paired serum samples as described elsewhere [18]. We used the initial PRN antibody titer as an estimate of baseline immunity. Lymphocytes were separated from whole blood specimens by using Lymphocyte Separation Medium (ICN Biochemicals). Lymphocytes were stored at ⫺70C in cryoprotective medium. Before being used for RNA extraction or virus isolation, lymphocytes were washed with 1 mL of sterile PBS (pH 7.5) and resuspended in 200 mL of Dulbecco’s modified Eagle medium (Gibco BRL Life Technologies). Urine and throat swab eluates were centrifuged at 770 g for 15 min at 4C, and the pellet was resuspended in 1 mL of Dulbecco’s modified Eagle medium and stored at ⫺70C. Virus isolation was done by inoculating 24-well plates containing B95a cells [19]. Duplicate wells were infected with 100 mL of each specimen and observed microscopically for the presence of viral cytopathic effect. Infected cells were passaged at least 3 times. If no cytopathic effect was observed after the third passage, the sample was considered negative. RNA was extracted from the specimens by use of the guanidinium acid–phenol procedure, and the presence of measles virus RNA was detected by reverse transcriptase–polymerase chain reaction (RT-PCR) as described elsewhere [19]. The primer sequences were the same as described elsewhere [20], except that the primers were 5-labeled with a fluorescent dye (FAM). The RT-PCR reactions were performed with the GeneAmp EZ rTth RNA PCR system (Perkin-Elmer), and PCR products were analyzed on an ABI PRISM 310 Genetic Analyzer (Applied Biosystems). The sensitivity of this procedure was determined to be 104 molecules of RNA. RESULTS A total of 44 confirmed cases of measles occurred during the 2 outbreaks. We studied 6 case patients, all of whom had laboratory-confirmed measles that occurred during the study period and agreed to provide contact information. Of the 364 exposed contacts of these 6 case patients, 153 met eligibility criteria, and of these, 133 (87%) consented to participate. Among the 133 cohort members, 12 (9%) were not vaccinated, 36 (27%) were previously vaccinated with 1 dose, and 85 (64%) were previously vaccinated with ⭓2 doses. Twelve participants (9%) reported previous history of measles. One hundred four persons (78%) were exposed for 1–2 h and 29 (22%) for ⭓3 h. The median age of the 133 participants was 16 years (range, 13–55 years). Eighty-nine participants (67%) Table 1. Patient no. 1 Description of source measles cases in Anchorage and Phoenix in 1998. Age, years Place of exposure Doses of measles-containing vaccine Time since last dose of MMR, years Positive for IgM antibody by EIA capture 25 Cruise ship 1 120 Yes No. of people exposed No. of people in study F + R + C1 + C2 21 15 Symptoms 2 17 High school 2 11 Yes F + R + C1 60 13 3 14 High school 1 14 Yes F + R + C1 + C2 + C3 60 8 4 16 Home 1 15 Yes F + R + C1 + C2 + C3 3 1 5 17 High school 0 NA Yes F + R + C1 + C2 + C3 110 55 6 45 High school 0 NA Yes F + R + C1 + C2 + C3 110 41 NOTE. C1, cough; C2, coryza; C3, conjunctivitis; F, fever; MMR, measles-mumps-rubella vaccine; NA, not available; R, rash. were female, 110 (83%) were white, and 112 (84%) were students. The 6 patients with measles exposed the study participants in different settings. Contacts of 4 patients were exposed in the high school setting. One case patient exposed contacts on a cruise ship and 1 exposed household contacts (table 1). None of the 133 study participants developed classical measles symptoms. Among initial and follow-up blood samples, 11 (8%) of 133 tested positive for measles-specific IgM, indicating recent measles virus infection (95% confidence interval, 0%– 29%). No measles virus was isolated from any of the initial or follow-up serum samples or from lymphocytes, urine, or throat swabs (table 2). The observed virus isolation rate was 0 among all contacts. The upper 95% confidence intervals for this rate were calculated as 2.2% among all 133 contacts, 10% among Table 2. Contacts of measles patients described in table 1 who had any positive result of testing for measles but who did not meet the clinical case definition for measles. Contact no. Exposed Virological to case Initial Convalescent 4-fold results of no. IgM IgM increase throat swab 1 1 2 ⫺ + ⫺ ⫺ 2 3 ⫺ + ⫺ ⫺ 3 5 + + ⫺ ⫺ 4 5 + ⫺ ⫺ ⫺ 5 5 + ⫺ ⫺ ⫺ 6 5 + + ⫺ ⫺ 7 6 + + ⫺ ⫺ 8 5 ⫺ + ⫺ ⫺ 9 1 ⫺ + ⫺ ⫺ 10 1 ⫺ ⫺ + ⫺ 11 1 ⫺ ⫺ + ⫺ 12 1 ⫺ ⫺ ⫺ …a 5 7 2 1 Total NOTE. No viruses were identified in urine, lymphocytes, or in the second throat swabs. +, positive; ⫺, negative. a Nucleotide sequence (D3) did not match outbreak sequence (D6). the 29 contacts exposed for ⭓3 h, and 29% among the 11 with serological evidence of infection. A throat swab from one contact without serological evidence of infection yielded positive results by PCR. However, the nucleotide sequence obtained (D3) was not consistent with the genotype associated with the outbreak (D6) [21], suggesting that the result was due to a laboratory artifact. Genotype D3 has been detected only 4 times in the United States since the 1989–1991 resurgence and, in each case, was the result of importation of virus. A genotype D3 virus was used to produce the positive control used in the RT-PCR assays [22]. We examined hours of exposure to a measles case, years since last dose of measles vaccine, baseline immunity (as estimated by PRN), presence of any symptom, measles history, and vaccination status as potential risk factors for inapparent infection. As seen in table 3, only hours of exposure was found to be statistically significantly associated with inapparent infection. There were 4 persons (4%) with serological evidence of infection among 104 persons exposed for 1–2 h, compared with 7 (24%) of 29 exposed for ⭓3 hours (relative risk, 6.0; 95% confidence interval, 1.9–19.2). Among those with initial PRN titers 11052 (n p 82), 5 developed inapparent infection; of those with titers !1052 (n p 48), 6 developed inapparent infection (relative risk, 2.0; 95% confidence interval, 0.7–6.4). No association was found for any of the above-described variables when an analysis was conducted stratifying the cohort by hours of exposure. DISCUSSION Measles virus was not detected in respiratory, urine, or blood specimens collected from persons exposed to measles, despite serological evidence of measles virus infection in 8% of the exposed cohort. Studies showing serological evidence of measles virus infections in persons without classical measles symptoms have raised the concern that transmission of measles virus in persons without clinically apparent measles may be an impediment to global measles elimination and eradication. The rate Shedding and Inapparent Infection • JID 2004:189 (Suppl 1) • S167 Table 3. Risk factors for inapparent measles virus infections. Percentage with positive serological results Relative risk 103 4 Ref. 30 23 6.0 !15 95 6 Ref. ⭓15 24 13 2.0 Variable N 95% confidence interval Hours of exposure !3 ⭓3 1.9–19.2 Years since last MCV 0.5–7.4 Postexposure immunity (PRN titer) ⭓1052 82 6 Ref. !1052 48 13 2.0 0.7–6.4 Symptoms Yes 81 6 Ref. No 51 12 1.9 No 105 8 Ref. Yes 10 10 1.2 0.6–5.8 Measles history 0.2–8.7 MCV received Yes 121 8 Ref. No 12 8 1.0 NOTE. MCV, measles-containing vaccine; PRN, plaque reduction neutralization; Ref., referent. of these serological responses among exposed persons has been estimated in a range from 4% to 42% by statistical and laboratory methods (table 4). The use of different diagnostic methods and different degrees of exposure may have affected the calculated rate of inapparent measles virus infections in highly vaccinated populations. This was the first attempt to detect measles virus in a highly vaccinated group of people who had been exposed to measles. The collection of multiple specimens for virus detection (throat swabs, urine, and lymphocytes), repeated on separate days during the period when viral shedding was most likely, was a unique approach to evaluate the potential for spread of measles from persons with inapparent measles virus infections [20]. Addition of highly sensitive RT-PCR testing to standard virus Table 4. 0.1–7.2 isolation techniques maximized the likelihood of detecting measles virus RNA present in the specimens [22]. The absence of detectable measles virus among persons with clinically inapparent measles in this study suggests that infectious transmission from such cases is rare, if it occurs at all, and is unlikely to be an impediment to measles elimination and eradication efforts. Our study suggests that the intensive exposure to persons with symptomatic measles is required for a substantial proportion of previously immune exposed contacts to develop inapparent infections, which reduces the likelihood that these infections will be epidemiologically important. If persons with inapparent infections are capable of transmitting measles virus, they are likely much less infectious than classical cases, on the Studies of susceptibility to inapparent measles virus infections. Author, year [reference] Chen 1990 [8] Ozanne 1992 [4] Huiss 1997 [2] No. of persons studied Estimated and observed susceptibility to inapparent infections, no. (%) Age range, years 18 7 (39) 18–22 496 20 (4) 7–44 44 4 (9) 26–50 Helfand 1998 [5] 44 10 (23) 18–68 Whittle 1999 [9] 102 43 (42) 7 months–7 years S168 • JID 2004:189 (Suppl 1) • Lievano et al. Assay used Plaque reduction neutralization Country United States Complement fixation Canada EIA, neutralization, hemagglutination inhibition Luxemburg EIA capture United States Hemagglutination inhibition Senegal basis of the absence of detectable measles virus in respiratory specimens in our study. Therefore, the likelihood that exposure to an inapparently infected person would lead to another inapparent infection is probably 0. Measles surveillance data from the United States are consistent with an extremely limited role of inapparent infections in measles virus transmission. There are only a few cases of confirmed measles each year in which epidemiological linkage to a classical measles case is not detected [23–25]. Our study had limited power to detect viral shedding, because we were able to identify only 11 people with serological evidence of infection for study of possible viral shedding among the pool of subjects certainly exposed for ⭓1 h to measles virus. However, we consider that this study closely approximates the most common exposure settings likely to occur, given the current low measles incidence in United States. We used our best judgment about timing of collecting the specimens to determine measles virus infection. Because the IgM response is quick and transient in the secondary immune response, we may have missed some instances of serological evidence of infection. Likewise, in some persons in the cohort, the IgG titer may have already begun to rise before our initial serum sample, limiting our ability to detect a 4-fold increase in titer. Thus, there may have been many more inapparent infections in this cohort that we did not detect. However, we did not detect evidence of measles virus in any member of the exposed cohort, including those whose inapparent infections we might have missed. We also used our best judgment about the timing of specimen collection to identify evidence of viral shedding. It is possible that viral shedding occurs earlier in persons with inapparent infections than in persons with classical measles. However, if that is the case, the duration of shedding would likely be very transient, because an early and more vigorous immune response would be expected to shut off viral proliferation earlier, leaving less time for virus to replicate. Therefore, we consider it very unlikely that altering the timing of the collection of specimens for virus isolation would yield different results. We cannot quantify the relation between detection of measles virus and infectiousness. However, because we could not detect the viral type related to the outbreak in any of the specimens, using the very sensitive RT-PCR and direct isolation, it is unlikely that the case person would be infectious. Our findings are not applicable to populations with high prevalence of HIV. Persons with AIDS may develop severe measles without showing classical symptoms such as rash [26, 27] and may be infectious. Studies of the rates of undiagnosed “rashless” measles in patients with AIDS and of shedding of measles virus and infectiousness in such cases should be undertaken to better characterize their frequency in these populations. However, it appears that these patients do not play an important role in the epidemiology and control of measles, because countries in which HIV is hyperendemic in Southern Africa have recently documented excellent measles control during at least 5 years [28]. Although inapparent measles virus infections did occur among some persons exposed to symptomatic measles in our study, intensive exposure was required in this highly vaccinated population, and we were unable to detect shedding of measles virus among exposed persons who did not develop classical measles. These data suggest that inapparent measles virus infections are not likely to be highly contagious and are probably not important in measles virus transmission in immune populations. Inapparent infections should not be considered an impediment to measles control and elimination, at least in areas with low prevalence of HIV infection. Acknowledgments We thank study participants, parents, and teachers for their collaboration; staff of the dental clinic in Anchorage for their cooperation; Susan Baum and Tracey Lynn for their critical assistance in gathering subjects for the study; the Alaska Department of Health and Social Services, the Arizona Department of Health and Social Services; the Arctic Investigations Program, Anchorage; and Jay Butler, Helen Peters, and Marilyn Getty for their collaboration. References 1. Edmoston MB, Addiss DG, McPherson JT, Berg JL, Circo SR, Davis JP. Mild measles and secondary vaccine failure during a sustained outbreak in a highly vaccinated population. JAMA 1990; 263:2467–71. 2. Huiss S, Damien B, Schneider F, Muller CP. Characteristics of asymptomatic secondary immune responses to measles virus in late convalescent donors. Clin Exp Immunol 1997; 109:416–20. 3. Muller CP, Huiss S, Schneider F. Secondary immune responses in parents of children with recent measles. Lancet 1996; 348:1379–80. 4. Ozanne G, D’Halewyn MA. Secondary immune response in a vaccinated population during a large measles epidemic. J Clin Microbiol 1992; 30:1778–82. 5. Helfand RF, Kim DK, Gary HE, et al. Nonclassic measles infections in an immune population exposed to measles during a college bus trip. J Med Virol 1998; 56:337–41. 6. Vardas E, Kreis S. Isolation of measles virus from a naturally-immune, asymptomatically re-infected individual. J Clin Virol 1999; 13:173–9. 7. Pedersen IR, Mordhorst CH, Glikmann G, von Magnus H. Subclinical measles infection in vaccinated seropositive individuals in arctic Greenland. Vaccine 1989; 7:345–8. 8. Chen RT, Markowitz LE, Albretch P, et al. Measles antibody: reevaluation of protective titers. J Infect Dis 1990; 162:1036–42. 9. Whittle HC, Aaby P, Samb B, Jensen H, Bennett J, Simondon F. Effect of subclinical infection on maintaining immunity against measles in vaccinated children in West Africa. Lancet 1999; 353:98–102. 10. Reyes MA, De Borrero MF, Roa J, Bergonzoli G, Saravia NG. Measles vaccine failure after documented seroconversion. Pediatr Infect Dis J 1987; 6:848–51. 11. Centers for Disease Control and Prevention. Measles. Pink book. 5th ed. Atlanta: Centers for Disease Control and Prevention, 1999:135–59. 12. Redd SC, Markowitz LE, Katz SL. Measles vaccine. In: Plotkin SA, Shedding and Inapparent Infection • JID 2004:189 (Suppl 1) • S169 13. 14. 15. 16. 17. 18. 19. Orenstein WA, eds. Vaccines. 3d ed. Philadelphia: WB Saunders, 1999: 222–66. Mossong J, Nokes DJ, Edmunds WJ, Cox MJ, Ratnam S, Muller CP. Modeling the impact of inapparent measles transmission in vaccinated populations with waning immunity. Am J Epidemiol 1999; 150: 1238–49. Centers for Disease Control and Prevention. Transmission of measles among a highly vaccinated school population—Anchorage, Alaska 1998. MMWR Morb Mortal Wkly Rep 1999; 47:1109–11. Erdman DD, Anderson LJ, Adams DR, Stewart JA, Markowitz LE, Bellini WJ. Evaluation of monoclonal antibody-based capture enzyme immunoassays for detection of specific antibodies to measles virus. J Clin Microbiol 1991; 29:1466–71. Ratnam S, Tipples G, Head C, Fauvel M, Fearon M, Ward BJ. Performance of indirect immunoglobulin M (IgM) serology tests and IgM capture assays for laboratory diagnosis of measles. J Clin Microbiol 2000; 38:99–104. Hummel KB, Erdman DD, Heath J, Bellini WJ. Baculovirus expression of the nucleoprotein gene of measles virus and utility of the recombinant protein in diagnostic enzyme immunoassays. J Clin Microbiol 1992; 30:2874–80. Albrecht P, Herrmann K, Burns GR. Role of virus strain in conventional and enhanced measles plaque neutralization test. J Virol Methods 1981; 3:251–60. Kobune F, Sakata H, Sugira A. Marmoset lymphoblastoid cells as a sensitive host for isolation of measles virus. J Virol 1990; 64:700–5. S170 • JID 2004:189 (Suppl 1) • Lievano et al. 20. Rota PA, Khan AS, Durigon E, Yuran T, Villamarzo YS, Bellini WJ. Detection of measles virus RNA in urine specimens from vaccine recipients. J Clin Microbiol 1995; 33:2485–88. 21. Lynn TV, Beller M, Funk EA, et al. The incremental effectiveness of 2 doses of measles-containing vaccine compared with 1 among high school students during an outbreak. J Infect Dis 2004; 189(Suppl 1): S86–90. 22. Rota PA, Liffick SL, Rota JS, et al. Molecular epidemiology of measles viruses in the United States: 1997–2001. Emerg Infect Dis 2002; 8: 902–8. 23. Centers for Disease Control and Prevention. Measles—United States, 1997. MMWR Morb Mortal Wkly Rep 1998; 47:273–6. 24. Centers for Disease Control and Prevention. Epidemiology of measles—United States, 1998. MMWR Morb Mortal Wkly Rep 1999; 48: 749–53. 25. Centers for Disease Control and Prevention. Measles—United States, 1999. MMWR Morb Mortal Wkly Rep 2000; 49:557–60. 26. Moss WJ, Cutts F, Griffin D. Implications of the human immunodeficiency virus epidemic for control and eradication of measles. Clin Infect Dis 1999; 29:106–12. 27. Cutts FT, Henao-Restrepo AM, Olive JM. Measles elimination: progress and challenges. Vaccine 1999; 17:S47–52. 28. Biellik R, Madema S, Taole A, et al. First 5 years of measles elimination in southern Africa: 1996–2000. Lancet 2002; 359:1564–8.