Incidence of cytomegalovirus infection among the general population and pregnant women in the United States
© Colugnati et al; licensee BioMed Central Ltd. 2007
Received: 17 January 2007
Accepted: 02 July 2007
Published: 02 July 2007
Cytomegalovirus (CMV) is a common opportunistic infection among HIV-infected individuals, a major source of serious complications among organ-transplant recipients, and a leading cause of hearing loss, vision loss, and mental retardation among congenitally infected children. Women infected for the first time during pregnancy are especially likely to transmit CMV to their fetuses. More children suffer serious disabilities caused by congenital CMV than by several better-known childhood maladies such as Down syndrome or fetal alcohol syndrome
Using CMV seroprevalence data from the nationally representative Third National Health and Nutrition Examination Survey, we estimated CMV incidence among the general United States population and among pregnant women. We employed catalytic models that used age-specific CMV seroprevalences as cumulative markers of past infections in order to derive estimates of three basic parameters: the force of infection, the basic reproductive rate, and the average age of infection. Our main focus was the force of infection, an instantaneous per capita rate of acquisition of infection that approximates the incidence of infection in the seronegative population.
Among the United States population ages 12–49 the force of infection was 1.6 infections per 100 susceptible persons per year (95% confidence interval: 1.2, 2.4). The associated basic reproductive rate of 1.7 indicates that, on average, an infected person transmits CMV to nearly two susceptible people. The average age of CMV infection was 28.6 years. Force of infection was significantly higher among non-Hispanic Blacks (5.7) and Mexican Americans (5.1) than among non-Hispanic Whites (1.4). Force of infection was significantly higher in the low household income group (3.5) than in the middle (2.1) and upper (1.5) household income groups. Based on these CMV incidence estimates, approximately 27,000 new CMV infections occur among seronegative pregnant women in the United States each year.
These thousands of CMV infections in pregnant women, along with the sharp racial/ethnic disparities in CMV incidence, are compelling reasons for accelerating research on vaccines and other interventions for preventing congenital CMV disease. Nevertheless, the relatively low force of infection provides encouraging evidence that modestly effective vaccines and rates of vaccination could significantly reduce CMV transmission.
Cytomegalovirus (CMV) is a common opportunistic infection among human immunodeficiency virus (HIV)-infected individuals, a major source of serious viral complications among organ-transplant recipients, and a leading cause of hearing loss, vision loss, and mental retardation among congenitally infected children. In fact, more children suffer serious disabilities caused by congenital CMV than by several better-known childhood maladies such as Down syndrome or fetal alcohol syndrome .
Like other herpesviruses, primary CMV infection is followed by the establishment of lifelong latent infection from which periodic reactivation is common [2, 3]. Symptoms are usually absent during primary infection and reactivation, but CMV can be shed in various bodily secretions, particularly urine and saliva . CMV is transmitted person-to-person via close non-sexual contact, sexual activity, breastfeeding, blood transfusions, and organ transplantation . For pregnant women, important sources of infection include sexual activity and contact with the urine or saliva of young children, especially their own children [5–7].
Congenital CMV infection is most likely to occur following a primary infection in the mother during pregnancy . However, maternal CMV reactivation or reinfection with a different CMV strain can also lead to fetal infection . Approximately 10 percent of congenitally infected infants are symptomatic at birth, and of the 90 percent who are asymptomatic, 10–15 percent will develop symptoms over months or even years .
Incidence of primary CMV infections has been estimated only in small or specialized populations, such as pregnant women or day care providers. The most comprehensive study of CMV incidence was carried out by Griffiths and colleagues in the United Kingdom , in which they estimated that more than three seronegative women per 100 seroconvert each year. However, their study was limited to pregnant women and was hospital-based rather than population-based. Robust, nationally representative estimates of CMV incidence are essential for 1) assessing the burden of primary CMV infection in the United States population, especially among pregnant women; 2) examining whether there are racial/ethnic disparities in primary maternal infection rates, which might be responsible for racial/ethnic disparities in congenital infection rates; and 3) evaluating how effective a vaccine or other intervention must be in order to reduce the incidence of congenital CMV disease. To obtain estimates of CMV incidence in the United States, we employed mathematical models that used age-specific CMV seroprevalences from the Third National Health and Nutrition Examination Survey (NHANES III).
Study population and design
NHANES III was conducted from 1988 to 1994 and provides nationally representative estimates of the health and nutritional status of the civilian, noninstitutionalized population of the United States. In order to produce population-representative estimates, NHANES III used a multistage, stratified, clustered sample design and generated sample weights proportional to the probability of participant selection. All our analyses used the NHANES III sample weights and sample design variables to correct the CMV seroprevalence point estimates for population representativeness and the interval estimates for the multistage complex sample design. The study protocol was approved by the authors' institutional review board. More details about NHANES III can be found in the official documentation . Serologic testing for CMV immunoglobulin G (IgG) was conducted as described previously .
The main focus of our models of CMV incidence was the age range 12–49 years. Over 90 percent of participants in this age range had sera available for CMV testing (N = 11,859) so that seroprevalence estimates were representative of the United States population. More importantly, this age range included women of childbearing age and so has key relevance for congenital CMV disease. Although surplus sera was only available for approximately 70 percent of 6–11 year-olds (N = 2,679), we also ran models in this age group to assess whether incidence rates differed by race/ethnicity. Nationally-representative CMV seroprevalence estimates were not available for children less than six years old.
Description of models
Here we give an overview of the models of CMV incidence. A more detailed description is provided in the Appendix. We employed catalytic models [13, 14] that used age-specific CMV seroprevalences as cumulative markers of past infections in order to derive estimates of three basic parameters: the force of infection, the basic reproductive rate, and the average age of infection. The force of infection is the instantaneous per capita rate of acquisition of infection  and will be expressed in this article as the number of primary CMV infections per 100 seronegative persons per year. The basic reproductive rate is a function of the force of infection and is the average number of secondary infections produced when one infected individual is introduced into a host population where everyone is susceptible. The average age of infection is also a function of the force of infection and is the age at which an individual in a given population typically acquires a specific infection. We considered parameter differences to be statistically significant when corresponding confidence intervals did not overlap.
The models made the following assumptions: CMV infection does not affect the mortality rate; seroprevalence in newborns equals zero; the death rate is type I, meaning everyone survives until a specific age, after which the survival probability is zero; and every person in the population is equally susceptible (i.e., homogeneous mixing).
We estimated the model parameters for the entire United States population and for specific population groups stratified by sex, race/ethnicity, and/or household income. Race/ethnicity was a self-reported variable that consisted of non-Hispanic Whites, non-Hispanic Blacks, Mexican Americans and Others. As recommended in NHANES III documentation , Others was excluded from the analyses because the sample size was too small and encompassed a diverse mix of race/ethnicity. Household income was divided into low, medium, and high, as defined previously [11, 12].
Estimating risk of CMV infection during pregnancy
We estimated risk of CMV infection for seronegative women during pregnancy as risk = 100 × [1 - e(-rate×time )], where rate was the force of infection per 100 women per year and time was the duration of pregnancy . We multiplied this risk by the proportion of women who are CMV seronegative to obtain risk of CMV infection during pregnancy for the entire population (i.e., seronegative and seropositive) of women. We then multiplied the risk of infection in the entire population of women by the average number of live-birth pregnancies per year for the years 1988–1994 . This product represented the estimated annual number of women with a primary CMV infection during pregnancy.
CMV force of infection, basic reproductive rates, and average age of infection among persons 12–49 years old in the United States.
Force of Infection (95% CI)*
Basic reproductive rate (95% CI)
Average age of infection in years (95% CI)
Entire U.S. population
Income per family size
Risk and frequency of CMV primary infection during pregnancy in the United States.
Risk among seronegative women/100 pregnancies*
Risk for all women/100 pregnancies
No. live-birth pregnancies (100's)†
No. women with primary infection during live-birth pregnancies
Robust estimates of the frequency of new CMV infections are essential for understanding and preventing viral transmission. This study provides the first estimates of CMV incidence that are based on population-representative data. We found that among CMV-seronegative individuals aged 12–49 in the United States, nearly one in 60 seroconverts each year.
This relatively low force of infection indicates that CMV is less easily transmitted than some other infections, such as measles or rubella. For these infections, high vaccine efficacy and coverage are required in order to interrupt transmission . In contrast, a CMV vaccine would not need to have such high efficacy and coverage to substantially prevent CMV transmission. Griffiths et al. , who estimated forces of CMV infection of 3.1–3.5/100 persons/year in the United Kingdom, showed that modest rates of vaccination (~60 percent) would be able to eradicate CMV infection from the human population. Our estimates, which are similar but even lower overall (force of infection = 1.6/100 persons/year), provide further evidence that modestly effective vaccines and rates of vaccination could significantly reduce CMV transmission.
Our models identified large racial/ethnic disparities in the frequencies of new CMV infections. The force of infection for CMV was considerably higher in non-Hispanic Blacks and Mexican Americans than in non-Hispanic Whites. The nearly three-fold differences in risk of primary CMV infection among seronegative women could be responsible for much of the racial/ethnic disparities in rates of infants born with congenital CMV . Racial/ethnic differences were especially pronounced among adolescent girls (ages 12–19 years), among whom primary infection was 50 times more likely in seronegative non-Hispanic blacks and 15 times more likely in seronegative Mexican Americans than in non-Hispanic whites. These higher forces of infection (i.e., incidence in seronegative individuals) suggest that CMV is circulating more frequently in these racial/ethnic groups. Thus, seropositive, pregnant non-Hispanic blacks and Mexican Americans may be at a higher risk of suffering re-infection with a different strain of CMV, which also places their infants at risk of symptomatic congenital CMV . These disparities indicate that interventions, such as vaccines or education campaigns, may need to be tailored to meet the needs of different racial/ethnic groups and different age groups.
In addition to race/ethnicity, low household income was a risk factor for CMV infection. People with low household income may be more likely to have a larger family and experience crowding, thus facilitating CMV transmission via close contact. However, because force of infection was more strongly associated with race/ethnicity than with household income, high-risk racial/ethnic groups may have a higher prevalence of additional factors related to CMV transmission, such as increased exposure to CMV while caring for young children. A more detailed analysis of risk factors for CMV infection in NHANES III can be found in Staras et al. .
Among women ages 20–49 years, force of infection appeared to be independent of age, suggesting that risk of infection during pregnancy is fairly constant during these ages, and that interventions to prevent congenital CMV must target all women of childbearing age. CMV had a higher force of infection than infections transmitted primarily via sex or injection drug use, such as herpes simplex virus type 2 (HSV-2) or hepatitis B virus (HBV). This suggests either that CMV is more easily transmissible through such behaviors  or, more likely, that CMV is transmitted via other, additional routes. Given that CMV has been shown to be transmitted via urine or saliva during close, non-sexual contact, it is likely that this sort of transmission plays a major role in the dynamics of CMV infection .
We estimated that each year in the United States more than 27,000 pregnant women experience primary CMV infection and are thus at high risk of giving birth to a child with congenital CMV infection. This estimate does not include any fetal losses that may have been caused by primary CMV infection, nor does it include the many pregnancies affected by CMV reactivation or reinfection among seropositive women. The burden of primary CMV infections during pregnancy falls disproportionately on disadvantaged women--those of low income and racial/ethnic minorities. Furthermore, teenaged minority women are at especially high risk of primary CMV infections during pregnancy, due to their high prevalence of susceptibility, high force of infection, and high pregnancy rates.
The risk of primary CMV infection during pregnancy among seronegative women is similar to previous estimates . For seronegative women, CMV infection represents one of the highest risks for fetal damage that they experience during pregnancy . Because CMV transmission is potentially preventable , CMV antibody screening prior to or near the beginning of pregnancy should be evaluated as a means of identifying women at high risk for having congenitally infected infants. Studies should pursue whether knowledge of high risk status is a useful motivational tool for modifying behaviors, such as hand hygiene, for reducing risk of infection . Such screening may also lead to the administration of CMV hyperimmuneglobulins or antiviral drugs for prevention or therapy of fetal infection and disease [23, 24].
In this study the modeling assumptions appeared to have been reasonably satisfied. On a population level, CMV infection does not contribute significantly to mortality among infected individuals. Nearly all members (=99 percent) of the population are susceptible at birth, and infection is believed to induce life-long immunity. The type I death-rate cut-off was chosen as 70 years to approximate the U.S. life expectancy during the years that NHANES III was conducted, but modifying the cut-off had little effect on the model results. The assumption of homogeneous mixing is unlikely to be completely true, but because CMV infection is common and has multiple transmission modes, susceptible individuals are likely to have similar risks of exposure to CMV.
An important limitation of our models was that the data were from a single, cross-sectional study so that time trends were not able to be addressed. Thus, high CMV seroprevalence in cohorts of older people might not reflect current incidence and could cause the models to overestimate the force of infection . We sought to minimize this potential bias by focusing most of our analyses on a limited age range (12–49 years). It is also important to note that our younger, age-specific force of infection estimates (i.e., for ages 6–11 and 12–19 years) were imprecise, with wide confidence intervals. Furthermore, the models implicitly assumed that seroprevalence was monotonically increasing with age, as if this cross-sectional study were a cohort study in which seroprevalence was measured at various ages of follow-up. However, this assumption was violated for some of the younger subpopulations. As a result, we occasionally obtained negative estimates for the force of infection (Figure 2), although these estimates were not statistically different from zero.
The calculations of risk of primary infection during pregnancy required several assumptions, one of which was that the force of infection was the same for pregnant and non-pregnant women. Women who are pregnant may have fewer sex partners (and thus lower risk of exposure to CMV) during pregnancy; on the other hand, pregnant women may be more likely than non-pregnant women to be exposed to young children (a group that frequently sheds CMV). Pregnant women may also have a higher risk of acquiring infections because of pregnancy-induced immune depression .
Based on our models, we would estimate that more than one million United States women have experienced primary CMV infections during pregnancy since CMV was first isolated 50 years ago [26, 27]. A substantial proportion of these infections would have led to congenital infections, leaving thousands of children with lifelong disabilities. Children from disadvantaged racial/ethnic groups are likely to have been disproportionately impacted. These many affected children are a compelling argument for accelerating research on vaccines and other interventions for the prevention of congenital CMV .
Each year, thousands of CMV infections occur in pregnant women in the United States, putting numerous unborn babies at risk for serious disabilities. Incidence of CMV infection in pregnant women is not distributed evenly, but exhibits sharp racial/ethnic disparities, especially affecting non-Hispanic blacks and Mexican Americans. Because of the magnitude of the problem and its associated health disparities, there is an urgent need to accelerate research on vaccines and other interventions for preventing congenital CMV disease. Nevertheless, the low incidence of CMV infection relative to other vaccine-preventable infections provides encouraging evidence that modestly effective vaccines and rates of vaccination could significantly reduce CMV transmission.
where L = 70 is the threshold age for the type I death rate.
When estimating force of infection for different subgroup categories, one category was chosen to be the referent category and the others were represented by indicator variables and were included in the models with interaction for age. For example, in the case of race/ethnicity, which had 3 categories and White as the referent category, the model was:
ln(S -(a)) = -(β 0 + λ 0 a + β 1 δ[White - Black] + β 2 δ[Mexican] + λ 1 δ[White - Black]a + λ 2 δ[Mexican]a),
where δ [X] = 1 if X and 0 otherwise.
Comparison of force of infection for different viruses for selected* age ranges.
Force of infection (per 100 persons per year)
Lit. review – misc. sources
Lit. review – misc. sources
Lit. review – misc. sources
3.1 and 3.5
We thank Kay Radford, Ashley Schoenfisch and Minal Amin for serologic testing and Geraldine McQuillan for technical assistance. This study was supported in part by funding from the National Vaccine Program Office. Fernando Colugnati received support from the Brazilian Centro Nacional de Pesquisa while he was a guest researcher at the CDC. This research was supported in part by an appointment to the Research Participation Program at the Centers for Disease Control and Prevention administered by the Oak Ridge Institute for Science and Education through an interagency agreement between the U.S. Department of Energy and CDC (stipend support for Stephanie A.S. Staras).
- Cannon MJ, Davis KF: Washing our hands of the congenital cytomegalovirus disease epidemic. BMC Public Health. 2005, 5: 70-10.1186/1471-2458-5-70.View ArticlePubMedPubMed CentralGoogle Scholar
- Alford CA, Stagno S, Pass RF, Britt WJ: Congenital and perinatal cytomegalovirus infections. Rev Infect Dis. 1990, 12 Suppl 7: S745-S753.View ArticlePubMedGoogle Scholar
- Brown HL, Abernathy MP: Cytomegalovirus infection. Semin Perinatol. 1998, 22: 260-266. 10.1016/S0146-0005(98)80014-1.View ArticlePubMedGoogle Scholar
- Stagno S: Cytomegalovirus. Infectious diseases of the fetus and newborn infant. Edited by: Remington JS and Klein JO. 2001, Philadelphia, W.B. Saunders Company, 389-424.Google Scholar
- Pass RF, Hutto C, Ricks R, Cloud GA: Increased rate of cytomegalovirus infection among parents of children attending day-care centers. N Engl J Med. 1986, 314: 1414-1418.View ArticlePubMedGoogle Scholar
- Adler SP: Cytomegalovirus and child day care: risk factors for maternal infection. Pediatr Infect Dis J. 1991, 10: 590-594.View ArticlePubMedGoogle Scholar
- Fowler KB, Pass RF: Risk factors for congenital cytomegalovirus infection in the offspring of young women: exposure to young children and recent onset of sexual activity. Pediatrics. 2006, 118: e286-e292. 10.1542/peds.2005-1142.View ArticlePubMedGoogle Scholar
- Boppana SB, Rivera LB, Fowler KB, Mach M, Britt WJ: Intrauterine transmission of cytomegalovirus to infants of women with preconceptional immunity. N Engl J Med. 2001, 344: 1366-1371. 10.1056/NEJM200105033441804.View ArticlePubMedGoogle Scholar
- Stagno S, Whitley RJ: Herpesvirus infection of pregnancy. N Engl J Med. 1985, 313: 1270-1274.View ArticlePubMedGoogle Scholar
- Griffiths PD, McLean A, Emery VC: Encouraging prospects for immunisation against primary cytomegalovirus infection. Vaccine. 2001, 19: 1356-1362. 10.1016/S0264-410X(00)00377-7.View ArticlePubMedGoogle Scholar
- Plan and operation of the Third National Health and Nutrition Examination Survey, 1988-94. Series 1: programs and collection procedures. Vital Health Stat 1. 1994, (32): 1-407.
- Staras SAS, Dollard SC, Radford KW, Flanders WD, Pass RF, Cannon MJ: Seroprevalence of cytomegalovirus infection in the United States, 1988-1994. Clin Infect Dis. 2006, 43: 1143-1151. 10.1086/508173.View ArticlePubMedGoogle Scholar
- Anderson RM, May RM: Infectious diseases of humans--dynamics and control. 1991, Oxford, United Kingdom, Oxford Scientific PublicationsGoogle Scholar
- Farrington CP: Modelling forces of infection for measles, mumps and rubella. Stat Med. 1990, 9: 953-967. 10.1002/sim.4780090811.View ArticlePubMedGoogle Scholar
- Centers for Disease Control and Prevention: Analytic and Reporting Guidelines: The Third National Health and Nutrition Examination Survey, NHANES III (1988-1994). 1996, Hyattsville, MD, National Center for Health StatisticsGoogle Scholar
- Ades AE: Methods for estimating the incidence of primary infection in pregnancy: a reappraisal of toxoplasmosis and cytomegalovirus data. Epidemiol Infect. 1992, 108: 367-375.View ArticlePubMedPubMed CentralGoogle Scholar
- Ventura SJ, Mosher WD, Curtin SC, Abma JC, Henshaw S: Trends in pregnancy rates for the United States, 1976-97: an update. Natl Vital Stat Rep. 2001, 49 (4): 1-9.Google Scholar
- Edmunds WJ, Gay NJ, Kretzschmar M, Pebody RG, Wachmann H: The pre-vaccination epidemiology of measles, mumps and rubella in Europe: implications for modelling studies. Epidemiol Infect. 2000, 125: 635-650. 10.1017/S0950268800004672.View ArticlePubMedPubMed CentralGoogle Scholar
- Fowler KB, Stagno S, Pass RF: Maternal age and congenital cytomegalovirus infection: screening of two diverse newborn populations, 1980-1990. J Infect Dis. 1993, 168: 552-556.View ArticlePubMedGoogle Scholar
- Stover CT, Smith DK, Schmid DS, Pellett PE, Stewart JA, Klein RS, Mayer K, Vlahov D, Schuman P, Cannon MJ: Prevalence of and risk factors for viral infections among human immunodeficiency virus (HIV)-infected and high-risk HIV-uninfected women. J Infect Dis. 2003, 187: 1388-1396. 10.1086/374649.View ArticlePubMedGoogle Scholar
- Cannon MJ, Pellett PE: Risk of congenital cytomegalovirus infection. Clin Infect Dis. 2005, 40: 1701-1702. 10.1086/430172.View ArticlePubMedGoogle Scholar
- Adler SP, Finney JW, Manganello AM, Best AM: Prevention of child-to-mother transmission of cytomegalovirus among pregnant women. J Pediatr. 2004, 145: 485-491. 10.1016/j.jpeds.2004.05.041.View ArticlePubMedGoogle Scholar
- Nigro G, Adler SP, La Torre R, Best AM: Passive immunization during pregnancy for congenital cytomegalovirus infection. N Engl J Med. 2005, 353: 1350-1362. 10.1056/NEJMoa043337.View ArticlePubMedGoogle Scholar
- Adler SP, Nigro G, Pereira L: Recent advances in the prevention and treatment of congenital cytomegalovirus infections. Semin Perinatol. 2007, 31: 10-18. 10.1053/j.semperi.2007.01.002.View ArticlePubMedGoogle Scholar
- Yip L, McCluskey J, Sinclair R: Immunological aspects of pregnancy. Clin Dermatol. 2006, 24: 84-87. 10.1016/j.clindermatol.2005.10.022.View ArticlePubMedGoogle Scholar
- SMITH MG: Propagation in tissue cultures of a cytopathogenic virus from human salivary gland virus (SGV) disease. Proc Soc Exp Biol Med. 1956, 92: 424-430.View ArticlePubMedGoogle Scholar
- ROWE WP, HARTLEY JW, WATERMAN S, TURNER HC, HUEBNER RJ: Cytopathogenic agent resembling human salivary gland virus recovered from tissue cultures of human adenoids. Proc Soc Exp Biol Med. 1956, 92: 418-424.View ArticlePubMedGoogle Scholar
- Ross DS, Dollard SC, Victor M, Sumartojo E, Cannon MJ: The epidemiology and prevention of congenital cytomegalovirus infection and disease: activities of the Centers for Disease Control and Prevention Workgroup. J Womens Health (Larchmt ). 2006, 15: 224-229. 10.1089/jwh.2006.15.224.View ArticleGoogle Scholar
- Analysis of Complex Surveys. Edited by: Skinner CJ, Holt D and Smith TMF. 1989, Chichester, John Wiley & SonsGoogle Scholar
- Gidding HF, MacIntyre CR, Burgess MA, Gilbert GL: The seroepidemiology and transmission dynamics of varicella in Australia. Epidemiol Infect. 2003, 131: 1085-1089. 10.1017/S0950268803001134.View ArticlePubMedPubMed CentralGoogle Scholar
- Armstrong GL, Schillinger J, Markowitz L, Nahmias AJ, Johnson RE, McQuillan GM, St Louis ME: Incidence of herpes simplex virus type 2 infection in the United States. Am J Epidemiol. 2001, 153: 912-920. 10.1093/aje/153.9.912.View ArticlePubMedGoogle Scholar
- Coleman PJ, McQuillan GM, Moyer LA, Lambert SB, Margolis HS: Incidence of hepatitis B virus infection in the United States, 1976-1994: estimates from the National Health and Nutrition Examination Surveys. J Infect Dis. 1998, 178: 954-959.View ArticlePubMedGoogle Scholar
- The pre-publication history for this paper can be accessed here:http://www.biomedcentral.com/1471-2334/7/71/prepub