THE NEW VACCINE SURVEILLANCE NETWORK DESIGN AND METHODOLOGY
The New Vaccine Surveillance Network (NVSN) was established by the Centers for Disease Control and Prevention to evaluate the burden of selected illnesses for which new vaccines either are available or are expected to be developed and to measure the impact of these vaccines, once licensed, on disease rates and related health outcomes. The NVSN currently includes 3 sites that are engaged in active population-based surveillance for acute respiratory viral infections in children aged younger than 5 years: Vanderbilt University (Nashville/Davidson County); University of Rochester (Rochester/Monroe County); and Cincinnati Children's Hospital Medical Center (Cincinnati/ Hamilton County). Currently 140,000 children in these 3 counties are under surveillance, representing nearly 1% of children younger than 5 years of age in the United States. The race/ethnicity of these children is similar to the rest of the United States, but all 3 sites have more blacks and fewer Hispanics than the total U.S. population (Table 1).
Most prior studies of the burden of acute respiratory illnesses (ARIs) in young children were not prospective or population-based and did not take advantage of the more sensitive molecular viral diagnostic techniques such as polymerase chain reaction assay (PCR). The recently expanded recommendations for inactivated influenza vaccine in young children,1 the licensure of the live attenuated intranasal influenza vaccine and continuing efforts at vaccine development for respiratory syncytial virus (RSV) and parainfluenza virus render the NVSN both timely and important.
To date, we have performed prospective, year-round population-based surveillance on children younger than 5 years of age who were hospitalized with fever and/or respiratory symptoms since October 2000 in Nashville and Rochester and since 2003 in Cincinnati. County residents are enrolled 4 days per week at hospitals in each site that together include >95% of county children hospitalized for ARI. After parent/guardian informed consent, nasal and throat swabs are obtained for viral culture and PCR, the caretaker is interviewed; after discharge, the medical record is reviewed. Hospitalizations per 10,000 children are expressed as the weighted number of ARI or pathogen-specific hospitalizations, divided by the number of children in the county population multiplied by 10,000. Rates were calculated with SAS software by weighting the observed number of enrolled hospitalizations to account for sampling 4 days per week and eligible patients who were not enrolled.
To date, we have performed surveillance on children younger than 5 years of age with fever and/or respiratory symptoms at selected emergency departments and outpatient clinics in Nashville and Rochester (since 2002), and in Cincinnati (since 2003). In each county, 1 emergency room and 1–4 outpatient practices participated in surveillance activities during RSV and influenza seasons. With the use of systematic sampling, children at each of these sites were enrolled, and similar information was collected as for inpatients.
EPIDEMIOLOGY OF VIRAL DISEASES IN THE SURVEILLANCE POPULATION
During the first year of inpatient surveillance, 592 children hospitalized with ARI/fever were enrolled, representing 73% of those eligible. Since then, enrollment of eligible children has been >80%. First year results were recently published by Iwane et al.2 There were 180 ARI/fever hospitalizations per 10,000 children younger than 5 years of age, with >60% of admissions in children younger than 1 year of age. One-third of children had high risk conditions, including 24% with asthma, 4% with other lung disease, 3% with heart disease and 2% with sickle-cell disease. One-third of children had 1 of the 3 major viral pathogens of study interest identified by PCR and/or culture including RSV (20%), influenza (3%) and parainfluenza (7%). Other respiratory viruses were identified in 36% of children and included adenovirus, human metapneumoviruses (hMPV) and picornaviruses, the latter of which were primarily rhinoviruses. In 39% of children, no virus was identified. Infants younger than 6 months of age accounted for >50% of RSV hospitalizations and >40% of hospitalizations because of influenza and parainfluenza. More than 80% of hospitalizations associated with these 3 pathogens were in children younger than 2 years of age.
hMPV was identified by PCR in ∼4% of hospitalized children as recently summarized by Mullins et al.3 Children with this pathogen were more likely to be older (73% were 6–23 months of age), than those with RSV (36%), or the rest of the study cohort (31%). There was also a higher prevalence of high risk conditions (54%) compared with those with RSV (29%) or the rest of the study cohort (31%).
During the first year of surveillance, cultures were performed locally and reverse transcription-PCR was performed at the Centers for Disease Control and Prevention. A child was considered to have RSV, influenza or parainfluenza infection if there was a positive culture for the virus or reverse transcription (RT)-PCR was positive on 2 consecutive tests. Weinberg et al4 examined the yield of these 2 diagnostic methods during the first 13 months of surveillance. The use of PCR resulted in considerable increase in viral identification, especially for RSV. The ratio of specimens positive by RT-PCR to those positive by culture was 2.6:1 for RSV, 1.6:1 for influenza A and B and 1.5:1 for parainfluenza 1, 2 and 3. Of the 193 isolates positive for 1 of these 3 groups of pathogens, only 8 (4%) were identified by culture alone, whereas 104 (54%) were identified only by PCR. The yield for both PCR and culture peaked at days 3–4 of symptoms (∼40% positive for PCR); whereas on day 1 of symptoms, the yield for both PCR and culture was <10% (Fig. 1).
UTILITY OF NVSN DATA TO POLICYMAKERS AND VACCINE EVALUATION
In the past few years, policymakers have been discussing appropriate influenza vaccine recommendations for children. There has been a growing awareness that influenza hospitalizations in young children are substantial and that influenza can also result in serious outcomes, including encephalopathy5 and death,6 in previously healthy children. In 2003–2004, there was extensive media coverage about influenza in children and a number of influenza deaths. The NVSN provided a means to assess the impact of influenza in children younger than 5 years of age compared with previous years and to assess whether 2003–2004 was an unusual year for influenza in children, or whether the increased attention was more likely related to an increased recognition of influenza in children aided by more widespread use of rapid diagnostic testing.
It is important to know the actual burden of influenza illness to determine appropriate targets for immunization and to try to measure the impact of new vaccination policies. In addition, because influenza epidemics vary by year, it is important to assess the burden of influenza during a number of different seasons. The NVSN has contributed longitudinal data on rates of influenza-associated illness and assessment of these rates at 3 different sites. Data from the NVSN have been presented to the Advisory Committee for Immunization Practices on several occasions to help address some of these questions.
Figure 2 presents the variability of influenza rates over time and by site for children younger than 5 years of age. The overall mean rate was 9 hospitalizations per 10,000 children younger than 5 years of age, and the rate varied from 3 per 10,000 in Monroe County (Rochester, NY) in 2002–2003 and 2003–2004 to 22 per 10,000 in Hamilton County (Cincinnati, OH) in 2003–2004. Mean rates during the 4 study years were 43, 9 and 4 per 10,000 for children younger than 6 months, 6–23 months and 24 months to younger than 5 years of age, respectively. Of those 6–23 months of age hospitalized with influenza, 70% had no identifiable high risk conditions for which the vaccine was recommended. These and other incidence data were useful to the Advisory Committee for Immunization Practices, which has recently recommended annual influenza immunization of all children 6–23 months of age.1
Our preliminary look at influenza data through the 2003–2004 season suggests that influenza hospitalization rates were relatively high in 2003–2004, compared with the previous 3 influenza seasons, which were relatively mild in the United States (Fig. 2). In Davidson County (Nashville, TN), influenza hospitalization rates were highest in 2003–2004 (20 per 10,000) but were only 1.5-fold greater than during the 2001–2002 season (13 per 10,000). In Monroe County, in Rochester, NY, influenza hospitalization rates appeared to be lower in 2003–2004 than in the previous 2 seasons. We examined discharge diagnoses for children hospitalized during 3 influenza seasons from 2000–2001 through 2002–2003 compared with those hospitalized in 2003–2004 (Fig. 3). There appeared to be a shift in discharge diagnoses from fever/sepsis (decreased from 27% to 14% over time) to influenza (increased from 22% to 35%); other diagnoses showed little change. Thus the attention to influenza in children during 2003–2004 was likely the result of both a real, although not remarkable, increase in disease in some areas, as well as an increased recognition of febrile illness in children as being caused by influenza. The latter may have been the result of increased use of rapid diagnostic tests, an increase in media attention to influenza because of the early appearance of influenza in some geographic locations,6,7 ongoing deliberations about expanding influenza vaccine recommendations in children and the reports of pediatric influenza-associated deaths and encephalopathy (Fig. 4). 5,6 Data from 2003 to 2004 also clearly demonstrated that the impact of influenza varied by location, even within a single season. Analysis is ongoing, and these early results will be further investigated.
NVSN FUTURE DIRECTIONS
Studies are planned for the 2003–2004 and 2004–2005 influenza seasons to assess the effectiveness of 1 and 2 doses of influenza vaccine against medically attended influenza illness in healthy children 6–23 months of age who are now recommended to receive influenza vaccine. The NVSN is also evaluating risk and protective factors related to severity of influenza illness, including maternal immunization for children younger than 6 months of age who have high rates of influenza hospitalizations and for whom there is no licensed vaccine. The network will also be working to promote influenza vaccination in the 3 counties and to follow the adherence to the new recommendation to vaccinate children 6–23 months of age and household contacts of children younger than 2 years of age.
Although there are no licensed RSV vaccines, effectiveness of RSV monoclonal antibody prophylaxis in high risk children could be evaluated using NVSN data. RSV causes 2–3 times the number of hospitalizations in young children as influenza, parainfluenza and hMPV. However, the last 2 years of outpatient data suggest that influenza may result in similar or greater numbers of outpatient visits than RSV. Further analyses of the outpatient burden of these viruses are in progress.
Parainfluenza virus infections also are an important cause of pediatric hospitalizations. Recent NVSN studies have added to the precision of estimates of hospitalization rates (9 per 10,000 children younger than 5 years of age), which may help guide decisions on vaccine development.9
Only one-third of ARI/fever hospitalizations in young children are associated with RSV, influenza and parainfluenza viruses. Further work on the role of hMPVs, picornoviruses (especially rhinoviruses) and coronaviruses are in progress. The single year of data on hMPVs showed a hospitalization rate similar to that for influenza and parainfluenza viruses; however, data for additional years are needed.3 Further NVSN studies on hMPV, rhinovirus and coronavirus incidence rates among hospitalized children are under way. Specimens collected as part of the NVSN studies could also be used to help identify previously unrecognized viral causes of pediatric respiratory infections. Finally the collection of specimens from healthy controls during well child outpatient visits will help to determine the background carriage rates of these viruses in asymptomatic children.
Respiratory viruses cause a major share of illness in young children both in the United States and globally. Population-based surveillance, coupled with new viral diagnostic tests as done in the NVSN, can define the burden of this illness more precisely and help determine the impact of prophylactic and therapeutic measures to limit morbidity and mortality from these agents.
Question: Do you think children with asthma exacerbations or wheezing are overrepresented overall in your rhinovirus data?
Marie R. Griffin, MD: We have just started collecting data on rhinoviruses, and we have not really looked at their epidemiology compared with that of others. We are in the process of doing that, and we are very interested to see if they are weighted by children with asthma and what the distribution of diagnoses or presenting symptoms is with rhinoviruses. We think it is also important to look at healthy children as well, but the number of children admitted with rhinovirus certainly seems higher than you would expect.
Question: One of the findings in other studies with rhinovirus is that there is a fairly high frequency of detection in presumably asymptomatic or minimally symptomatic children with background rates as high as 10–20%. Do you have any data on controls?
Marie R. Griffin, MD: When we did our outpatient surveillance, we enrolled healthy children who were being seen in the outpatient setting but had no respiratory symptoms. Therefore we will have data on controls. We have the rhinovirus from year 1 and the healthy children from year 4. We are hoping to have concurrent data where we have the rhinoviruses for year 4 when we have the healthy children. This would be ideal, but we are aware that there is some background.
1. Harper SA, Fukuda K, Uyeki TM, Cox NJ, Bridges CB. Prevention and control of influenza
2. Iwane MK, Edwards KM, Szilagyi PG, et al. Population-based surveillance
for hospitalizations associated with respiratory syncytial virus
virus, and parainfluenza viruses among young children
3. Mullins JA, Erdman DD, Weinberg GA, et al. Human metapneumovirus infection among children
hospitalized with acute respiratory illness. Emerg Infect Dis
4. Weinberg GA, Erdman DD, Edwards KM, et al. Superiority of reversetranscription polymerase chain reaction to conventional viral culture in the diagnosis of acute respiratory tract infections in children
. J Infect Dis
5. Request for information about acute encephalopathy associated with influenza
virus infection in U.S. children
6. Update: influenza
activity—United States, 2003 to 2004 season. MMWR
7. Update: influenza
activity—United States, 2003 to 2004 season. MMWR
8. Weekly Report: Influenza
summary update. Available at: http://www.cdc.gov/flu/weeklyarchives2002-2003/weekly20.htm
The New Vaccine Surveillance Network Study Group includes: CDC: Larry Anderson, Carolyn Bridges, John Copeland, James A. Mullins, Benjamin Schwartz, David Shay; Vanderbilt University: Amondrea Blackwell, Ann Clay, Jennifer Doersam, Kathryn Edwards, Diane Kent, Haijing Li, Jody Peters, Yi-Wei Tang, Nayleen Whitehead, Yuwei Zhu; University of Rochester: Linda Anderson, Richard Barth, Caroline Hall, Gerry Lofthus, Andrea Marino, Stanley Schaffer, Kenneth Schnabel, Laura Shone, Peter Szilagyi; University of Cincinnati: Rachel Akers, David Bernstein, Emilie Grube, Linda Jamison, Ardythe Morrow, Joel Mortensen, Vanessa Wimmer, David Witte.
9. Weinberg GA, Hall CB, Walker FJ, et al. Parainfluenza virus infection of young children
: burden of hospitalization and seasonal pattern of infection [abstract]. Pediatr Res. 2004;55:321A. Abstract 1824.