Effectiveness of the 2003–2004 Influenza Vaccine Among Children 6 Months to 8 Years of Age, With 1 vs 2 Doses
http://www.100md.com
《小儿科》
Clinical Research Unit Department of Preventive Medicine, Kaiser Permanente Colorado, Denver, Colorado
National Immunization Program, Centers for Disease Control and Prevention, Atlanta, Georgia
ABSTRACT
Objective. To evaluate the effectiveness of 1 and 2 doses of the 2003–2004 influenza vaccine in preventing medically attended influenza-like illness (ILI) among children 6 to 23 months and 6 months to 8 years of age.
Design and Methods. Outpatient and emergency department visits and immunization records were used to conduct a retrospective cohort study among children 6 months to 8 years of age. ILI and pneumonia and influenza (P&I) outcomes were defined on the basis of International Classification of Diseases, Ninth Revision, codes. Influenza vaccine effectiveness (VE) was calculated as (1 – hazard rate ratio) x 100.
Results. A total of 29726 children were included in the analyses; 17.3% were 6 to 23 months of age. By November 19, 2003, the start of peak influenza activity, 7.5% and 9.9% of children 6 months to 8 years were fully or partially vaccinated against influenza, respectively. For fully vaccinated children 6 to 23 months of age, VE against ILI and P&I was 25% and 49%, respectively. No statistically significant reduction in ILI or P&I rates was observed for partially vaccinated children 6 to 23 months of age (–3% and 22%, respectively). For fully vaccinated children 6 months to 8 years of age, VE against ILI and P&I was 23% and 51%, respectively. For partial vaccination, VE was significant only for P&I (23%).
Conclusions. Despite a suboptimal match between the influenza vaccine and predominant circulating strains, influenza vaccination provided substantial protection for fully vaccinated children and possibly some protection for partially vaccinated children <9 years of age. These findings support vaccinating targeted children even when the vaccine match is suboptimal, and they highlight the need to vaccinate previously unvaccinated children with 2 doses for optimal protection.
Key Words: influenza vaccine effectiveness pediatrics unvaccinated
Abbreviations: ILI, influenza-like illness P&I, pneumonia and influenza HRR, hazard rate ratio VE, vaccine effectiveness ICD-9, International Classification of Diseases, Ninth Revision RSV, respiratory syncytial virus KPCO, Kaiser Permanente Colorado TCH, the Children’s Hospital CI, confidence interval ACIP, Advisory Committee on Immunization Practices
Influenza-related illnesses result in the hospitalization of 1 to 2 children per 1000 children <5 years of age annually in the United States, with children <2 years of age being at particularly high risk.1–4 Beginning in 2004–2005, yearly influenza vaccination was recommended for all children 6 to 23 months of age and for out-of-home caregivers and household contacts of children <2 years of age; influenza vaccination of children 6 months of age with certain medical conditions has been recommended for many years.5 Concerns about logistic challenges in vaccinating millions of US children during the autumn and early winter months have been raised, particularly considering that children <9 years of age who have not been vaccinated previously against influenza are recommended to receive 2 doses of inactivated influenza vaccine 1 month apart for the first year.6–8 Therefore, many children 6 to 23 months of age would need 2 doses of vaccine the first year of vaccination for optimal protection. Limited data are available on the effectiveness of 1 vs 2 doses of vaccine among children <9 years of age; the recommendation for the age cutoff was based principally on serologic data.9 In addition, few studies have assessed the effectiveness of the inactivated influenza vaccine among children 6 to 23 months of age during years with and without an optimal antigenic match between the vaccine and circulating strains.10,11 The overlap in circulation of influenza, respiratory syncytial virus (RSV), and other viruses and low vaccination rates have limited the ability of previous studies to evaluate the effectiveness of the influenza vaccine against non–laboratory-confirmed outcomes.1,2,10–12
The 2003–2004 influenza season in Colorado was unusual in 2 particular respects. First, influenza activity started earlier than usual (November 1, 2003, to January 17, 2004) and overlapped little with RSV activity (December 13, 2003, to April 17, 2004).13,14 Because of the distinct times for these disease entities, International Classification of Diseases, Ninth Revision (ICD-9) respiratory diagnoses were much more likely to be influenza-related during November and December of 2003. Second, the predominant circulating influenza A virus strain was antigenically different from the vaccine strain. Using the electronic medical records and immunization tracking system of Kaiser Permanente Colorado (KPCO), we sought to capitalize on the unusual, early, 2003–2004 influenza season to provide estimates of vaccine effectiveness (VE) among children. The primary objective of this study was to evaluate the effectiveness of influenza vaccine in preventing medically attended influenza-like illness (ILI) among children 6 to 23 months of age enrolled in KPCO during the 2003–2004 influenza season and the combined 6-month to 8-year age group. We examined VE for both partially vaccinated and fully vaccinated children, allowing for varying vaccination status over time for a given child depending on when shots were completed and the lag time needed for immunity.
METHODS
Study Population
Children 6 months through 8 years of age on October 1, 2003, who lived in the Denver/Boulder metropolitan area and were enrolled continuously in KPCO from October 1, 2003, through December 2003 were eligible for inclusion in the study population. The Denver/Boulder metropolitan market area of KPCO has 380000 members. To ensure that children were using KPCO as their primary source of health care, enrolled children were required to have had 1 health care visit to the health maintenance organization in the past 8 months for <1-year-old children, 12 months for 1- to 2-year-old children, 2 years for 2- to 4-year-old children, and 2.5 years for 5- to 8-year-old children. Overall, 29726 children were included in the analyses.
Vaccination Status
For many years, KPCO has identified members at high risk for influenza and encouraged influenza vaccinations. The KPCO Immunization Task Force recommended that all children 6 to 23 months of age, in addition to the other high-risk groups, be immunized with the influenza vaccine beginning in the 2002–2003 influenza season, before the full recommendation made by the Advisory Committee on Immunization Practices (ACIP). In addition to high-risk members, KPCO identified proactively and sent mailings to families with children 6 to 23 months of age, to encourage influenza vaccination. The influenza vaccination status of the cohort for years 2002 and 2003 was determined with the KPCO Immunization Tracking System. Immunizations completed during clinic visits were entered into the KPCO electronic medical record by health care providers in the pediatrics, family medicine, primary care, urgent care, and trauma departments and then transferred immediately to the Immunization Tracking System. Large numbers of influenza shots were also provided during designated influenza clinics organized during the autumn. Pediatric vaccinations given during influenza clinics were entered from consent forms directly into the electronic medical record. The accuracy and timeliness of this system are high priorities, because the system is the final repository of all immunization data and generates all vaccine reports used by KPCO. Full vaccination status was defined as having 2 influenza shots a minimum of 14 days before the ILI event date. Children vaccinated in the prior year had to have only 1 shot completed in the autumn of 2003 to meet the full vaccination definition. Children were considered partially vaccinated if they had only 1 influenza shot in the autumn of 2003 at least 14 days before the ILI event and no prior vaccination recorded in the previous year. Children vaccinated only in the previous year were considered unvaccinated.
Presence of Chronic Medical Conditions
To adjust for variations in health status and the prevalence of chronic conditions that could influence the likelihood of having a medically attended ILI during the period of observation, KPCO’s asthma and diabetes registries, as well as a pharmacy-based risk adjustment system called RxRisk, were used to identify patients with 1 relevant condition.15–18 The RxRisk system is a clinically validated algorithm that classifies children into chronic disease categories on the basis of drug prescriptions filled.19 Conditions deemed relevant for inclusion in the final models to be used for comorbidity adjustment included allergic rhinitis, asthma, diabetes mellitus, history of gastrointestinal disorders, cognitive or behavioral health disorders, and an aggregate category denoting a history of malignancy, liver or renal disease, and transplants. Underlying cardiac disease was not included in the model because <0.05% of the study population was identified as having this condition.
Respiratory Illness Outcomes
Study Period
To minimize misclassification attributable to non–influenza-related respiratory ILI, the outcome period was defined as November 19 through December 7, 2003, the peak weeks of influenza activity based on surveillance data from the Children’s Hospital (TCH) Department of Epidemiology and Pathology, the Colorado Department of Public Health and Environment-sponsored Influenza Surveillance Program, and the Centers for Disease Control and Prevention-sponsored Bioterrorism Surveillance Demonstration Program.13,14,21 The primary data source used to define this period was the laboratory isolate data derived from the TCH Department of Epidemiology and Pathology. This source of data represents a subset of all laboratory isolates of confirmed influenza cases collected and reported by the Colorado Department of Public Health and Environment, but these data are specific to children in the Denver metropolitan area and TCH is the primary contract hospital for KPCO pediatric patients. We chose the peak period of November 19 through December 7 for the following reasons: laboratory-confirmed cases increased from 50 during the week ending November 8 to >200 during the week ending November 15, total cases reported by TCH peaked at >350 during the week ending November 22, and the number of reported cases of influenza fell below 100 by December 13, whereas the number of cases of RSV was increasing. During the period of November 19 through December 7, TCH reported <15 positive tests each for RSV, parainfluenza, rhinoviruses, and pertussis.
Statistical Analyses
The association between influenza vaccination and subsequent medically attended ILI events was modeled with Cox proportional-hazards models.22,23 Survival analyses were used to account for changes in vaccination status during the period of interest; influenza vaccination status was a time-varying variable. Time to event used a count of days elapsed since the start of the selected time interval. The proportional-hazards assumption was tested by using interaction terms for fully and partially vaccinated variables with time. Both interaction terms were nonsignificant (P > .10).
When multiple ILI visits were found for a child, only the first ILI was modeled. The limited time frame for these analyses (19 days) made multiple events very rare. The vast majority of children had no ILI event (N = 27375 [92.1%], an additional 2105 children (7.1%) had only 1 ILI event, and >1 ILI event occurred for only 246 children in this sample (<1%). Rates of ILIs among vaccinated and unvaccinated children could be modeled with multiple ILI events by using Poisson regression models. Preliminary models for this sample compared vaccination associations with ILI outcomes from Poisson regression and survival analyses. These comparison models produced very similar results in the estimated magnitude of association between vaccination status and ILI outcomes. We selected survival analysis for all remaining analyses because these models could account more readily for daily changes in vaccination status through the use of time-varying variables.
We estimated models for all children 6 months to 8 years of age. Given the recent ACIP recommendations, we also estimated separate models for children 6 to 23 months of age. In addition to estimating the hazard rate ratio (HRR) of an ILI outcome according to vaccination status, we also expanded the models to include age, gender, and dichotomous comorbidity flags denoting the presence of a chronic condition and to assess their contribution to having a medically attended ILI. Separate models were also estimated for all children 6 months to 8 years of age that were identified in KPCO’s asthma disease management registry. Influenza VE was estimated to be (1 – HRR) x 100.
RESULTS
Vaccination Status and Prevalence of Chronic Medical Conditions
Source of ILI Events
A total of 2351 children (7.9%) had an ILI event during the study period, across all pediatric age categories. Approximately 92% of these events were derived from outpatient visits at a KPCO facility. The remaining events originated from emergency department visits. Of patients 6 to 23 months of age, 774 (15.1% of the population) had an ILI visit. Less than 3% of the total population had a P&I visit, with 221 of these events involving children 6 to 23 months of age (4.3% of the 6- to 23-month-old population). A slightly higher proportion of these P&I events, 9.5%, were derived from emergency department visits.
Estimates of VE
The results of the Cox proportional-hazards models for the broad ILI outcome and the narrower P&I outcome for children 6 months to 8 years of age are described in Table 2. For the ILI outcome, models with only vaccination status indicators demonstrated a significant protective effect, with a HRR of 0.85 (95% confidence interval [CI]: 0.73–0.99) and VE of 15% for all children 6 months to 8 years of age. With adjustment for age, gender, and chronic conditions, the fully vaccinated HRR was more protective (HRR: 0.77; 95% CI: 0.66–0.90), corresponding to a VE estimate of 23%. For the P&I outcome, this finding was even more pronounced for the fully vaccinated, with estimated HRR values of 0.56 (95% CI: 0.42–0.76; VE: 44%) for the univariate model and 0.49 (95% CI: 0.36–0.67; VE: 51%) for the fully adjusted model. In the adjusted model for the P&I outcome, a protective effect was also demonstrated among partially vaccinated members (HRR: 0.77; 95% CI: 0.62–0.95; VE: 23%).
Age was inversely associated with both the ILI and P&I outcomes: patients 6 to 11 months of age were 4 to 5 times more likely to present with an ILI, compared with 7- and 8-year-old patients. Asthma, allergic rhinitis, and a history of receiving a prescription for a gastroprotective agent all had HRR values of >1 and were statistically significant for the broad ILI outcome. These findings suggested that patients with a history of these conditions were more likely to present with an ILI outcome.
VE model estimates for children 6 to 23 months of age are shown in Table 3. The less-specific ILI univariate model showed a significant protective VE of 34% (HRR: 0.66; 95% CI: 0.50–0.88). The adjusted HRR estimate, although similar in magnitude to the all-age model, did not reach statistical significance in this smaller sample (HRR: 0.75; 95% CI: 0.56–1.00). A significant protective effect against P&I for fully vaccinated patients was found in both unadjusted (HRR: 0.53; 95% CI: 0.30–0.93; VE: 47%) and adjusted (HRR: 0.51; 95% CI: 0.29–0.91; VE: 49%) models. For the less-specific ILI outcome, both age (in months) and the presence of a gastrointestinal comorbidity were associated significantly with an ILI visit. Patients at younger ages (HRR: 0.66) and those with a history of receiving a prescription for a gastrointestinal illness (HRR: 1.39) were more likely to have an ambulatory encounter for ILI. None of the chronic condition variables was found to be associated statistically with a medically attended P&I.
A subanalysis was performed with a sample of 2229 children 6 months to 8 years of age who were identified in the asthma registry. Two hundred (9%) experienced an ILI event and 70 (3%) experienced a P&I event; 28% were fully vaccinated and 18% were partially vaccinated by December 7, 2003. Estimates derived from univariate and multivariate adjusted models for fully vaccinated children did not demonstrate a protective effect of vaccination against ILI (unadjusted HRR: 0.93; 95% CI: 0.67–1.30; adjusted HRR: 0.95; 95% CI: 0.68–1.33). However, a significant protective effect against P&I for fully vaccinated patients was found in both unadjusted (HRR: 0.30; 95% CI: 0.13–0.69) and adjusted (HRR: 0.32; 95% CI: 0.14–0.74) models, with VE estimated to be 70% and 68%, respectively. No protective effect of partial vaccination was found (adjusted HRR: 1.02; 95% CI: 0.55–1.89).
DISCUSSION
To our knowledge, this study is the largest study evaluating the effectiveness of influenza vaccine among children 6 to 23 months of age and the effectiveness of 1 vs 2 doses among children 6 months to 8 years of age. Our findings suggest that both doses are needed for previously unvaccinated children 6 months to 8 years of age, to provide maximal protection against influenza. In addition, despite the suboptimal antigenic match between the influenza vaccine and the circulating virus strain in the autumn of 2003, influenza vaccination was found to be effective in preventing medically attended ILIs among children 6 to 23 months of age, as well as all children 6 months to 8 years of age.
The estimates found in this study are consistent with the findings of other studies of influenza VE against medically attended ILI among children. Many of the previously published estimates of the efficacy of inactivated trivalent influenza vaccine among young children were based on randomized control trials with an end point of laboratory-confirmed influenza.10,20,24–27 In a recent review article, Zangwill and Belshe27 identified 7 studies, all randomized controlled trials, that reported efficacy data associated with the trivalent inactivated influenza vaccine for children <9 years of age. The authors pooled data from 5 of the studies that included children <9 years of age and estimated efficacy to be 63%. However, for studies with children <5 years of age, reported efficacies ranged from 12% to 83%. In a study evaluating data for children 6 to 23 months of age by Hoberman et al,10 efficacy ranged from 66% in the first year, when the attack rate was high, to –7% (not statistically significant) in the second study year, when the attack rate was low. In a study assessing the 1996–1997 trivalent inactivated influenza vaccine among children attending day care, Hurwitz et al11 found VE estimates of 45% associated with the reduction in laboratory-confirmed cases of influenza A and 31% for influenza B.
Several smaller studies demonstrated VE in the presence of a suboptimal match between the circulating influenza virus strain and the vaccine strain. In an examination of 137 patients with moderate/severe asthma during the 1992–1993 influenza season, when a drifted strain predominated, Sugaya et al28 found VE estimates of 67.5%. Three other studies24,25,29 also reviewed by Zangwill and Belshe27 demonstrated reduced protection against heterotypic influenza strains, although the VE estimates overlapped with those from studies conducted during years in which the circulating and virus strains were well matched.
This study showed relatively high vaccination rates and demonstrated the effect of proactive recommendations for influenza vaccinations for all children <9 years of age. Although ACIP’s explicit recommendations do not include vaccination for all children 2 years of age, studies suggested that the vaccination of school-aged children can reduce the spread of influenza in the community.30,31 Given the recent influenza vaccine shortages, and before broader recommendations are made, findings from vaccination studies with school-aged children, like the one currently underway in Texas, and benefits of the universal vaccination program in Ontario, Canada, should be evaluated to inform policymakers regarding to the societal benefits of vaccinating all school-aged children.32
Our study found that a large proportion of young children had been given a gastroprotective agent (5.5%) and use of such medications was associated significantly with an ILI visit. Reflux is more common among infants with a history of prematurity, a condition that was not controlled for directly in our analyses and that is associated with increased health services utilization. Although little research is available describing the direct association between medical visits for ILI and use of gastroprotective agents, some children diagnosed as having chronic cough or asthma might have reflux and intermittent aspiration.33 A recently published study examining the association between the use of acid-suppressing drugs and the occurrence of community-acquired pneumonia among adults found that the current use of gastric acid-suppressing therapy was associated with an increased risk of community-acquired pneumonia.34 Additional research is needed to understand more clearly whether reflux or the use of acid-suppressing therapy is associated with an increased risk of medically attended respiratory illness among young children.
Asthma was the predominant childhood disease in this sample, and influenza vaccination had been recommended for asthmatic children by the ACIP for many years. For children identified in KPCO’s asthma registry, our study found a large, significant, protective effect for the P&I outcome and a small nonsignificant estimate of a protective effect for the broader ILI outcome. Children were identified in KPCO’s asthma disease registry if they had either 1 prescription for a -receptor agonist (eg, albuterol or salmeterol) or 1 inpatient or outpatient diagnosis of asthma (ICD-9 code 493). No adjustment for asthma severity was made. This is particularly important because children with severe asthma are more likely to be vaccinated against influenza and they are more likely to have respiratory illness episodes than are children with mild asthma.15,16 Therefore, any study using non–influenza laboratory-confirmed outcomes among asthmatic children is likely to underestimate the true VE. Our non–laboratory-confirmed outcomes might have had more misclassification among asthmatic patients than among otherwise healthy children because of other causes of respiratory symptoms in this subgroup. This misclassification might have been worse for the broad ILI outcome, which would explain in part the lack of a significant protective effect for this outcome. However, we cannot rule out coding bias related to physician knowledge of vaccination status, which might have strengthened the protective effect for the P&I outcome. In the literature, results remain mixed for studies examining influenza VE in preventing ILI among asthmatics.35–38 Despite the current need for additional studies, the safety of the vaccine among asthmatic patients, the evidence that influenza infection leads to asthma exacerbation, and the results of our study all support vaccination for this group.
Our findings confirm the advantage of full (2 shots) versus partial (1 shot) vaccination in protecting against influenza among previously unvaccinated children <9 years of age and particularly among children 6 to 23 months of age. It is possible that the protective effect of partial vaccination for all children 6 months to 8 years of age found in the P&I outcome model is subject to limitations in interpretation, because vaccination status data were captured only over a 2-year period. Specifically, a child >2 years of age who received an influenza vaccination in 2003 but not 2002 was denoted as partially vaccinated. Some subset of these children might have received an influenza vaccination in 2001 or earlier and therefore should have been assigned the status of fully vaccinated. However, it is very unlikely that this type of misclassification is an issue for children 6 to 23 months of age, as included in these analyses. This misclassification, if it occurred, would have overestimated the effect among older, partially vaccinated children by including fully vaccinated patients in the same group and, if corrected, could only strengthen the recommendation that 2 shots are needed for VE.
This study has several limitations. It relied on administrative data, and the respiratory outcomes were not laboratory confirmed. We used a very short observation period of only 19 days. However, the lack of substantial overlap in RSV and influenza activity during the study period minimized the misclassification of illness. Also, because this was an observational study, vaccination status was not randomized. Therefore, irrespective of the adjustment variables used, selection bias may exist with respect to which patients obtained vaccination. However, the size of the population, the ability to adjust for the presence of underlying illness, and supporting virologic surveillance data to define precisely the period of peak influenza activity are important strengths of this study. In addition, there is a very low likelihood that we did not record all sources of utilization associated with KPCO children seeking care for ILI. Data used in these analyses included all utilization derived from KPCO medical office encounters, including all after-hours and urgent care visits, as well as all utilization that took place in emergency departments, including TCH (the only children’s hospital in the Denver metropolitan area). In addition, the recording of emergency department cases should account for the majority of patients admitted subsequently as inpatients for treatment for more severe disease.
The early occurrence of the 2003–2004 influenza season, coupled with intense media coverage of pediatric morbidity and death in the later weeks of the outbreak, could have generated significant variation in the coding of cases by KPCO physicians. For example, physicians might be more likely to code influenza in the later weeks of the outbreak and viral infection not otherwise specified, fever, or cough in the earlier weeks. We limited our study to a 19-day period that represented the peak weeks of influenza activity; influenza first appeared in early November and abated in early January 2004.13 Limiting our analysis to the peak weeks would minimize the coding misclassification by clinicians that could be expected during the first and last days of the seasonal influenza epidemic. Finally, it is possible that physicians assigning visit diagnoses might have been less likely to code influenza for vaccinated children, although the widespread publicity regarding the suboptimal antigenic match and vaccine failures likely would have lessened this potential coding bias. However, it is unlikely that a child presenting with fever, cough, or other upper respiratory tract illness symptoms would have an ICD-9 diagnosis code other than those used for defining an ILI (Appendix 1) and therefore would not be identified in our study.
CONCLUSIONS
Influenza vaccine was shown to prevent medically attended ILI among children 6 to 23 months of age, as well as older children, even with a suboptimal match. This study provides supportive data regarding the benefits of vaccination for children and the need for 2 doses among previously unvaccinated young children <9 years of age. The logistics of instituting the influenza vaccine recommendations for young children pose a formidable challenge to private and public health care systems. Additional strategies for fully vaccinating children before the onset of influenza activity should be evaluated.
APPENDIX 1. ILI AND P&I Definitions
ACKNOWLEDGMENTS
We acknowledge the funding contributions of the Centers for Disease Control and Prevention (contract 254-2004-M-06362, CFDA 93.283) and the Kaiser Permanente Colorado Clinical Research Unit.
FOOTNOTES
Accepted Mar 15, 2005.
These data were presented in part to the Advisory Committee on Immunization Practices; June 23, 2004; Atlanta, GA.
No conflict of interest declared.
REFERENCES
Neuzil KM, Mellen BG, Wright PF, Mitchel EF Jr, Griffin MR. The effect of influenza on hospitalization, outpatient visits, and courses of antibiotics in children. N Engl J Med.2000; 342 :225 –231
Izurieta HS, Thompson WW, Kramarz P, et al. Influenza and rates of hospitalization for respiratory disease among infants and young children. N Engl J Med.2000; 342 :232 –239
O’Brien MA, Uyeki TM, Shay DK, et al. Incidence of outpatient visits and hospitalizations related to influenza in infants and young children. Pediatrics.2004; 113 :585 –593
Iwane MK, Edwards KM, Szilagyi PG, et al. New Vaccine Surveillance Network: population-based surveillance for hospitalizations associated with respiratory syncytial virus, influenza virus, and parainfluenza viruses among young children. Pediatrics.2004; 113 :1758 –1764
Centers for Disease Control and Prevention. Prevention and control of influenza: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep.2004; 53(RR-6) :1 –40
Humiston SG, Szilagyi PG, Iwane MK, et al. The feasibility of universal influenza vaccination for infants and toddlers. Arch Pediatr Adolesc Med.2004; 158 :867 –874
Glazner JE, Beaty BL, Pearson KA, Berman S. The cost of giving childhood vaccinations: differences among provider types. Pediatrics.2004; 113 :1582 –1587
Szilagyi PG, Iwane MK, Schaffer S, et al. Potential burden of universal influenza vaccination of young children on visits to primary care practices. Pediatrics.2003; 112 :821 –828
La Montagne JR, Noble GR, Quinnan GV, et al. Summary of clinical trials of inactivated influenza vaccine, 1978. Rev Infect Dis.1983; 5 :723 –736
Hoberman A, Greenberg DP, Paradise JL, et al. Effectiveness of inactivated influenza vaccine in preventing acute otitis media in young children: a randomized controlled trial. JAMA.2003; 290 :1608 –1616
Hurwitz ES, Haber M, Chang A, et al. Studies of the 1996–1997 inactivated influenza vaccine among children attending day care: immunologic response, protection against infection, and clinical effectiveness. J Infect Dis.2000; 182 :1218 –1221
Clements DA, Langdon L, Bland C, Walter E. Influenza A vaccine decreases the incidence of otitis media in 6- to 30-month-old children in day care. Arch Pediatr Adolesc Med.1995; 149 :1113 –1117
Kramarz P, DeStefano F, Gargiullo PM, et al. Influenza vaccination in children with asthma in health maintenance organizations: Vaccine Safety Datalink Team. Vaccine.2000; 18 :2288 –2294
Kramarz P, Destefano F, Gargiullo PM, et al. Does influenza vaccination prevent asthma exacerbations in children J Pediatr.2001; 138 :306 –310
Hak E, Verheij TJ, Grobbee DE, Nichol KL, Hoes AW. Confounding by indication in non-experimental evaluation of vaccine effectiveness: the example of prevention of influenza complications. J Epidemiol Community Health.2002; 56 :951 –955
Hak E, Wei F, Nordin J, Mullooly J, Poblete S, Nichol KL. Development and validation of a clinical prediction rule for hospitalization due to pneumonia or influenza or death during influenza epidemics among community-dwelling elderly persons. J Infect Dis.2004; 189 :450 –458
Fishman PA, Shay DK. Development and estimation of a pediatric chronic disease score using automated pharmacy data. Med Care.1999; 37 :847 –883
Neuzil KM, Dupont WD, Wright PF, Edwards KM. Efficacy of inactivated and cold-adapted vaccines against influenza A infection, 1985 to 1990: the pediatric experience. Pediatr Infect Dis J.2001; 20 :733 –740
Yih WK, Caldwell B, Harmon R, et al. The National Bioterrorism Syndromic Surveillance Demonstration Program. MMWR Morb Mortal Wkly Rep.2004; 53(suppl) :43 –46
Cox DR. Regression models and life tables (with discussion). J R Stat Soc Ser B.1972; 34 :187 –220
Hosmer DW, Lemeshow S. Applied Survival Analysis: Regression Modeling of Time to Event Data. New York, NY: John Wiley & Sons;1999
Gruber WC, Taber LH, Glezen P, et al. Live attenuated and inactivated influenza vaccine in school-age children. Am J Dis Child.1990; 144 :595 –600
Clover RD, Crawford S, Glezen WP, et al. Comparison of heterotypic protection against influenza A/Taiwan/86 (H1N1) by attenuated and inactivated vaccines to A/Chile/83-like viruses. J Infect Dis.1991; 163 :300 –304
Heikkinen T, Tuuskanen O, Waris M, et al. Influenza vaccination in the prevention of acute otitis media in children. Am J Dis Child.1991; 145 :445 –448
Zangwill KM, Belshe BR. Safety and efficacy of trivalent inactivated influenza vaccine in young children: a summary for the new era of routine vaccination. Pediatr Infect Dis J.2004; 23 :189 –200
Sugaya N, Nerome K, Ishida M, Matsumoto M, Mitamura K, Nirasawa M. Efficacy of inactivated vaccine in preventing antigenically drifted influenza type A and well-matched type B. JAMA.1994; 272 :1122 –1126
Piedra PA, Glezen WP. Influenza in children: epidemiology, immunity, and vaccines. Semin Pediatr Infect Dis.1991; 2 :140 –146
Reichert TA, Sugaya N, Fedson DS, et al. The Japanese experience with vaccinating school children against influenza. N Engl J Med.2001; 344 :889 –896
Monto AS, Davenport FM, Napier JA, Francis T Jr. Modification of an outbreak of influenza in Tecumseh, Michigan, by vaccination of schoolchildren. J Infect Dis.1970; 122 :16 –25
Glezen WP. New vaccines for old diseases: trivalent cold-adapted influenza vaccine. Pediatr Ann.2004; 33 :545 –550
Thomson F, Masters IB, Chang AB. Persistent cough in children and the overuse of medications. J Paediatr Child Health.2002; 38 :578 –581
Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Stricker BH, Jansen JB. Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs. JAMA.2004; 292 :1955 –1960
Cates CJ, Jefferson TO, Bara AI, Rowe BH. Vaccines for preventing influenza in people with asthma. Cochrane Database Syst Rev.2004; (2) :CD000364
Christy C, Aligne CA, Auinger P, Pulcino T, Weitzman M. Effectiveness of influenza vaccine for the prevention of asthma exacerbations. Arch Dis Child.2004; 89 :734 –735
Bueving HJ, Bernsen RM, de Jongste JC, et al. Influenza vaccination in children with asthma: randomized double-blind placebo-controlled trial. Am J Respir Crit Care Med.2004; 169 :488 –493
Smits AJ, Hak E, Stalman WA, van Essen GA, Hoes AW, Verheij TJ. Clinical effectiveness of conventional influenza vaccination in asthmatic children. Epidemiol Infect.2002; 128 :205 –211(Debra P. Ritzwoller, PhD,)
National Immunization Program, Centers for Disease Control and Prevention, Atlanta, Georgia
ABSTRACT
Objective. To evaluate the effectiveness of 1 and 2 doses of the 2003–2004 influenza vaccine in preventing medically attended influenza-like illness (ILI) among children 6 to 23 months and 6 months to 8 years of age.
Design and Methods. Outpatient and emergency department visits and immunization records were used to conduct a retrospective cohort study among children 6 months to 8 years of age. ILI and pneumonia and influenza (P&I) outcomes were defined on the basis of International Classification of Diseases, Ninth Revision, codes. Influenza vaccine effectiveness (VE) was calculated as (1 – hazard rate ratio) x 100.
Results. A total of 29726 children were included in the analyses; 17.3% were 6 to 23 months of age. By November 19, 2003, the start of peak influenza activity, 7.5% and 9.9% of children 6 months to 8 years were fully or partially vaccinated against influenza, respectively. For fully vaccinated children 6 to 23 months of age, VE against ILI and P&I was 25% and 49%, respectively. No statistically significant reduction in ILI or P&I rates was observed for partially vaccinated children 6 to 23 months of age (–3% and 22%, respectively). For fully vaccinated children 6 months to 8 years of age, VE against ILI and P&I was 23% and 51%, respectively. For partial vaccination, VE was significant only for P&I (23%).
Conclusions. Despite a suboptimal match between the influenza vaccine and predominant circulating strains, influenza vaccination provided substantial protection for fully vaccinated children and possibly some protection for partially vaccinated children <9 years of age. These findings support vaccinating targeted children even when the vaccine match is suboptimal, and they highlight the need to vaccinate previously unvaccinated children with 2 doses for optimal protection.
Key Words: influenza vaccine effectiveness pediatrics unvaccinated
Abbreviations: ILI, influenza-like illness P&I, pneumonia and influenza HRR, hazard rate ratio VE, vaccine effectiveness ICD-9, International Classification of Diseases, Ninth Revision RSV, respiratory syncytial virus KPCO, Kaiser Permanente Colorado TCH, the Children’s Hospital CI, confidence interval ACIP, Advisory Committee on Immunization Practices
Influenza-related illnesses result in the hospitalization of 1 to 2 children per 1000 children <5 years of age annually in the United States, with children <2 years of age being at particularly high risk.1–4 Beginning in 2004–2005, yearly influenza vaccination was recommended for all children 6 to 23 months of age and for out-of-home caregivers and household contacts of children <2 years of age; influenza vaccination of children 6 months of age with certain medical conditions has been recommended for many years.5 Concerns about logistic challenges in vaccinating millions of US children during the autumn and early winter months have been raised, particularly considering that children <9 years of age who have not been vaccinated previously against influenza are recommended to receive 2 doses of inactivated influenza vaccine 1 month apart for the first year.6–8 Therefore, many children 6 to 23 months of age would need 2 doses of vaccine the first year of vaccination for optimal protection. Limited data are available on the effectiveness of 1 vs 2 doses of vaccine among children <9 years of age; the recommendation for the age cutoff was based principally on serologic data.9 In addition, few studies have assessed the effectiveness of the inactivated influenza vaccine among children 6 to 23 months of age during years with and without an optimal antigenic match between the vaccine and circulating strains.10,11 The overlap in circulation of influenza, respiratory syncytial virus (RSV), and other viruses and low vaccination rates have limited the ability of previous studies to evaluate the effectiveness of the influenza vaccine against non–laboratory-confirmed outcomes.1,2,10–12
The 2003–2004 influenza season in Colorado was unusual in 2 particular respects. First, influenza activity started earlier than usual (November 1, 2003, to January 17, 2004) and overlapped little with RSV activity (December 13, 2003, to April 17, 2004).13,14 Because of the distinct times for these disease entities, International Classification of Diseases, Ninth Revision (ICD-9) respiratory diagnoses were much more likely to be influenza-related during November and December of 2003. Second, the predominant circulating influenza A virus strain was antigenically different from the vaccine strain. Using the electronic medical records and immunization tracking system of Kaiser Permanente Colorado (KPCO), we sought to capitalize on the unusual, early, 2003–2004 influenza season to provide estimates of vaccine effectiveness (VE) among children. The primary objective of this study was to evaluate the effectiveness of influenza vaccine in preventing medically attended influenza-like illness (ILI) among children 6 to 23 months of age enrolled in KPCO during the 2003–2004 influenza season and the combined 6-month to 8-year age group. We examined VE for both partially vaccinated and fully vaccinated children, allowing for varying vaccination status over time for a given child depending on when shots were completed and the lag time needed for immunity.
METHODS
Study Population
Children 6 months through 8 years of age on October 1, 2003, who lived in the Denver/Boulder metropolitan area and were enrolled continuously in KPCO from October 1, 2003, through December 2003 were eligible for inclusion in the study population. The Denver/Boulder metropolitan market area of KPCO has 380000 members. To ensure that children were using KPCO as their primary source of health care, enrolled children were required to have had 1 health care visit to the health maintenance organization in the past 8 months for <1-year-old children, 12 months for 1- to 2-year-old children, 2 years for 2- to 4-year-old children, and 2.5 years for 5- to 8-year-old children. Overall, 29726 children were included in the analyses.
Vaccination Status
For many years, KPCO has identified members at high risk for influenza and encouraged influenza vaccinations. The KPCO Immunization Task Force recommended that all children 6 to 23 months of age, in addition to the other high-risk groups, be immunized with the influenza vaccine beginning in the 2002–2003 influenza season, before the full recommendation made by the Advisory Committee on Immunization Practices (ACIP). In addition to high-risk members, KPCO identified proactively and sent mailings to families with children 6 to 23 months of age, to encourage influenza vaccination. The influenza vaccination status of the cohort for years 2002 and 2003 was determined with the KPCO Immunization Tracking System. Immunizations completed during clinic visits were entered into the KPCO electronic medical record by health care providers in the pediatrics, family medicine, primary care, urgent care, and trauma departments and then transferred immediately to the Immunization Tracking System. Large numbers of influenza shots were also provided during designated influenza clinics organized during the autumn. Pediatric vaccinations given during influenza clinics were entered from consent forms directly into the electronic medical record. The accuracy and timeliness of this system are high priorities, because the system is the final repository of all immunization data and generates all vaccine reports used by KPCO. Full vaccination status was defined as having 2 influenza shots a minimum of 14 days before the ILI event date. Children vaccinated in the prior year had to have only 1 shot completed in the autumn of 2003 to meet the full vaccination definition. Children were considered partially vaccinated if they had only 1 influenza shot in the autumn of 2003 at least 14 days before the ILI event and no prior vaccination recorded in the previous year. Children vaccinated only in the previous year were considered unvaccinated.
Presence of Chronic Medical Conditions
To adjust for variations in health status and the prevalence of chronic conditions that could influence the likelihood of having a medically attended ILI during the period of observation, KPCO’s asthma and diabetes registries, as well as a pharmacy-based risk adjustment system called RxRisk, were used to identify patients with 1 relevant condition.15–18 The RxRisk system is a clinically validated algorithm that classifies children into chronic disease categories on the basis of drug prescriptions filled.19 Conditions deemed relevant for inclusion in the final models to be used for comorbidity adjustment included allergic rhinitis, asthma, diabetes mellitus, history of gastrointestinal disorders, cognitive or behavioral health disorders, and an aggregate category denoting a history of malignancy, liver or renal disease, and transplants. Underlying cardiac disease was not included in the model because <0.05% of the study population was identified as having this condition.
Respiratory Illness Outcomes
Study Period
To minimize misclassification attributable to non–influenza-related respiratory ILI, the outcome period was defined as November 19 through December 7, 2003, the peak weeks of influenza activity based on surveillance data from the Children’s Hospital (TCH) Department of Epidemiology and Pathology, the Colorado Department of Public Health and Environment-sponsored Influenza Surveillance Program, and the Centers for Disease Control and Prevention-sponsored Bioterrorism Surveillance Demonstration Program.13,14,21 The primary data source used to define this period was the laboratory isolate data derived from the TCH Department of Epidemiology and Pathology. This source of data represents a subset of all laboratory isolates of confirmed influenza cases collected and reported by the Colorado Department of Public Health and Environment, but these data are specific to children in the Denver metropolitan area and TCH is the primary contract hospital for KPCO pediatric patients. We chose the peak period of November 19 through December 7 for the following reasons: laboratory-confirmed cases increased from 50 during the week ending November 8 to >200 during the week ending November 15, total cases reported by TCH peaked at >350 during the week ending November 22, and the number of reported cases of influenza fell below 100 by December 13, whereas the number of cases of RSV was increasing. During the period of November 19 through December 7, TCH reported <15 positive tests each for RSV, parainfluenza, rhinoviruses, and pertussis.
Statistical Analyses
The association between influenza vaccination and subsequent medically attended ILI events was modeled with Cox proportional-hazards models.22,23 Survival analyses were used to account for changes in vaccination status during the period of interest; influenza vaccination status was a time-varying variable. Time to event used a count of days elapsed since the start of the selected time interval. The proportional-hazards assumption was tested by using interaction terms for fully and partially vaccinated variables with time. Both interaction terms were nonsignificant (P > .10).
When multiple ILI visits were found for a child, only the first ILI was modeled. The limited time frame for these analyses (19 days) made multiple events very rare. The vast majority of children had no ILI event (N = 27375 [92.1%], an additional 2105 children (7.1%) had only 1 ILI event, and >1 ILI event occurred for only 246 children in this sample (<1%). Rates of ILIs among vaccinated and unvaccinated children could be modeled with multiple ILI events by using Poisson regression models. Preliminary models for this sample compared vaccination associations with ILI outcomes from Poisson regression and survival analyses. These comparison models produced very similar results in the estimated magnitude of association between vaccination status and ILI outcomes. We selected survival analysis for all remaining analyses because these models could account more readily for daily changes in vaccination status through the use of time-varying variables.
We estimated models for all children 6 months to 8 years of age. Given the recent ACIP recommendations, we also estimated separate models for children 6 to 23 months of age. In addition to estimating the hazard rate ratio (HRR) of an ILI outcome according to vaccination status, we also expanded the models to include age, gender, and dichotomous comorbidity flags denoting the presence of a chronic condition and to assess their contribution to having a medically attended ILI. Separate models were also estimated for all children 6 months to 8 years of age that were identified in KPCO’s asthma disease management registry. Influenza VE was estimated to be (1 – HRR) x 100.
RESULTS
Vaccination Status and Prevalence of Chronic Medical Conditions
Source of ILI Events
A total of 2351 children (7.9%) had an ILI event during the study period, across all pediatric age categories. Approximately 92% of these events were derived from outpatient visits at a KPCO facility. The remaining events originated from emergency department visits. Of patients 6 to 23 months of age, 774 (15.1% of the population) had an ILI visit. Less than 3% of the total population had a P&I visit, with 221 of these events involving children 6 to 23 months of age (4.3% of the 6- to 23-month-old population). A slightly higher proportion of these P&I events, 9.5%, were derived from emergency department visits.
Estimates of VE
The results of the Cox proportional-hazards models for the broad ILI outcome and the narrower P&I outcome for children 6 months to 8 years of age are described in Table 2. For the ILI outcome, models with only vaccination status indicators demonstrated a significant protective effect, with a HRR of 0.85 (95% confidence interval [CI]: 0.73–0.99) and VE of 15% for all children 6 months to 8 years of age. With adjustment for age, gender, and chronic conditions, the fully vaccinated HRR was more protective (HRR: 0.77; 95% CI: 0.66–0.90), corresponding to a VE estimate of 23%. For the P&I outcome, this finding was even more pronounced for the fully vaccinated, with estimated HRR values of 0.56 (95% CI: 0.42–0.76; VE: 44%) for the univariate model and 0.49 (95% CI: 0.36–0.67; VE: 51%) for the fully adjusted model. In the adjusted model for the P&I outcome, a protective effect was also demonstrated among partially vaccinated members (HRR: 0.77; 95% CI: 0.62–0.95; VE: 23%).
Age was inversely associated with both the ILI and P&I outcomes: patients 6 to 11 months of age were 4 to 5 times more likely to present with an ILI, compared with 7- and 8-year-old patients. Asthma, allergic rhinitis, and a history of receiving a prescription for a gastroprotective agent all had HRR values of >1 and were statistically significant for the broad ILI outcome. These findings suggested that patients with a history of these conditions were more likely to present with an ILI outcome.
VE model estimates for children 6 to 23 months of age are shown in Table 3. The less-specific ILI univariate model showed a significant protective VE of 34% (HRR: 0.66; 95% CI: 0.50–0.88). The adjusted HRR estimate, although similar in magnitude to the all-age model, did not reach statistical significance in this smaller sample (HRR: 0.75; 95% CI: 0.56–1.00). A significant protective effect against P&I for fully vaccinated patients was found in both unadjusted (HRR: 0.53; 95% CI: 0.30–0.93; VE: 47%) and adjusted (HRR: 0.51; 95% CI: 0.29–0.91; VE: 49%) models. For the less-specific ILI outcome, both age (in months) and the presence of a gastrointestinal comorbidity were associated significantly with an ILI visit. Patients at younger ages (HRR: 0.66) and those with a history of receiving a prescription for a gastrointestinal illness (HRR: 1.39) were more likely to have an ambulatory encounter for ILI. None of the chronic condition variables was found to be associated statistically with a medically attended P&I.
A subanalysis was performed with a sample of 2229 children 6 months to 8 years of age who were identified in the asthma registry. Two hundred (9%) experienced an ILI event and 70 (3%) experienced a P&I event; 28% were fully vaccinated and 18% were partially vaccinated by December 7, 2003. Estimates derived from univariate and multivariate adjusted models for fully vaccinated children did not demonstrate a protective effect of vaccination against ILI (unadjusted HRR: 0.93; 95% CI: 0.67–1.30; adjusted HRR: 0.95; 95% CI: 0.68–1.33). However, a significant protective effect against P&I for fully vaccinated patients was found in both unadjusted (HRR: 0.30; 95% CI: 0.13–0.69) and adjusted (HRR: 0.32; 95% CI: 0.14–0.74) models, with VE estimated to be 70% and 68%, respectively. No protective effect of partial vaccination was found (adjusted HRR: 1.02; 95% CI: 0.55–1.89).
DISCUSSION
To our knowledge, this study is the largest study evaluating the effectiveness of influenza vaccine among children 6 to 23 months of age and the effectiveness of 1 vs 2 doses among children 6 months to 8 years of age. Our findings suggest that both doses are needed for previously unvaccinated children 6 months to 8 years of age, to provide maximal protection against influenza. In addition, despite the suboptimal antigenic match between the influenza vaccine and the circulating virus strain in the autumn of 2003, influenza vaccination was found to be effective in preventing medically attended ILIs among children 6 to 23 months of age, as well as all children 6 months to 8 years of age.
The estimates found in this study are consistent with the findings of other studies of influenza VE against medically attended ILI among children. Many of the previously published estimates of the efficacy of inactivated trivalent influenza vaccine among young children were based on randomized control trials with an end point of laboratory-confirmed influenza.10,20,24–27 In a recent review article, Zangwill and Belshe27 identified 7 studies, all randomized controlled trials, that reported efficacy data associated with the trivalent inactivated influenza vaccine for children <9 years of age. The authors pooled data from 5 of the studies that included children <9 years of age and estimated efficacy to be 63%. However, for studies with children <5 years of age, reported efficacies ranged from 12% to 83%. In a study evaluating data for children 6 to 23 months of age by Hoberman et al,10 efficacy ranged from 66% in the first year, when the attack rate was high, to –7% (not statistically significant) in the second study year, when the attack rate was low. In a study assessing the 1996–1997 trivalent inactivated influenza vaccine among children attending day care, Hurwitz et al11 found VE estimates of 45% associated with the reduction in laboratory-confirmed cases of influenza A and 31% for influenza B.
Several smaller studies demonstrated VE in the presence of a suboptimal match between the circulating influenza virus strain and the vaccine strain. In an examination of 137 patients with moderate/severe asthma during the 1992–1993 influenza season, when a drifted strain predominated, Sugaya et al28 found VE estimates of 67.5%. Three other studies24,25,29 also reviewed by Zangwill and Belshe27 demonstrated reduced protection against heterotypic influenza strains, although the VE estimates overlapped with those from studies conducted during years in which the circulating and virus strains were well matched.
This study showed relatively high vaccination rates and demonstrated the effect of proactive recommendations for influenza vaccinations for all children <9 years of age. Although ACIP’s explicit recommendations do not include vaccination for all children 2 years of age, studies suggested that the vaccination of school-aged children can reduce the spread of influenza in the community.30,31 Given the recent influenza vaccine shortages, and before broader recommendations are made, findings from vaccination studies with school-aged children, like the one currently underway in Texas, and benefits of the universal vaccination program in Ontario, Canada, should be evaluated to inform policymakers regarding to the societal benefits of vaccinating all school-aged children.32
Our study found that a large proportion of young children had been given a gastroprotective agent (5.5%) and use of such medications was associated significantly with an ILI visit. Reflux is more common among infants with a history of prematurity, a condition that was not controlled for directly in our analyses and that is associated with increased health services utilization. Although little research is available describing the direct association between medical visits for ILI and use of gastroprotective agents, some children diagnosed as having chronic cough or asthma might have reflux and intermittent aspiration.33 A recently published study examining the association between the use of acid-suppressing drugs and the occurrence of community-acquired pneumonia among adults found that the current use of gastric acid-suppressing therapy was associated with an increased risk of community-acquired pneumonia.34 Additional research is needed to understand more clearly whether reflux or the use of acid-suppressing therapy is associated with an increased risk of medically attended respiratory illness among young children.
Asthma was the predominant childhood disease in this sample, and influenza vaccination had been recommended for asthmatic children by the ACIP for many years. For children identified in KPCO’s asthma registry, our study found a large, significant, protective effect for the P&I outcome and a small nonsignificant estimate of a protective effect for the broader ILI outcome. Children were identified in KPCO’s asthma disease registry if they had either 1 prescription for a -receptor agonist (eg, albuterol or salmeterol) or 1 inpatient or outpatient diagnosis of asthma (ICD-9 code 493). No adjustment for asthma severity was made. This is particularly important because children with severe asthma are more likely to be vaccinated against influenza and they are more likely to have respiratory illness episodes than are children with mild asthma.15,16 Therefore, any study using non–influenza laboratory-confirmed outcomes among asthmatic children is likely to underestimate the true VE. Our non–laboratory-confirmed outcomes might have had more misclassification among asthmatic patients than among otherwise healthy children because of other causes of respiratory symptoms in this subgroup. This misclassification might have been worse for the broad ILI outcome, which would explain in part the lack of a significant protective effect for this outcome. However, we cannot rule out coding bias related to physician knowledge of vaccination status, which might have strengthened the protective effect for the P&I outcome. In the literature, results remain mixed for studies examining influenza VE in preventing ILI among asthmatics.35–38 Despite the current need for additional studies, the safety of the vaccine among asthmatic patients, the evidence that influenza infection leads to asthma exacerbation, and the results of our study all support vaccination for this group.
Our findings confirm the advantage of full (2 shots) versus partial (1 shot) vaccination in protecting against influenza among previously unvaccinated children <9 years of age and particularly among children 6 to 23 months of age. It is possible that the protective effect of partial vaccination for all children 6 months to 8 years of age found in the P&I outcome model is subject to limitations in interpretation, because vaccination status data were captured only over a 2-year period. Specifically, a child >2 years of age who received an influenza vaccination in 2003 but not 2002 was denoted as partially vaccinated. Some subset of these children might have received an influenza vaccination in 2001 or earlier and therefore should have been assigned the status of fully vaccinated. However, it is very unlikely that this type of misclassification is an issue for children 6 to 23 months of age, as included in these analyses. This misclassification, if it occurred, would have overestimated the effect among older, partially vaccinated children by including fully vaccinated patients in the same group and, if corrected, could only strengthen the recommendation that 2 shots are needed for VE.
This study has several limitations. It relied on administrative data, and the respiratory outcomes were not laboratory confirmed. We used a very short observation period of only 19 days. However, the lack of substantial overlap in RSV and influenza activity during the study period minimized the misclassification of illness. Also, because this was an observational study, vaccination status was not randomized. Therefore, irrespective of the adjustment variables used, selection bias may exist with respect to which patients obtained vaccination. However, the size of the population, the ability to adjust for the presence of underlying illness, and supporting virologic surveillance data to define precisely the period of peak influenza activity are important strengths of this study. In addition, there is a very low likelihood that we did not record all sources of utilization associated with KPCO children seeking care for ILI. Data used in these analyses included all utilization derived from KPCO medical office encounters, including all after-hours and urgent care visits, as well as all utilization that took place in emergency departments, including TCH (the only children’s hospital in the Denver metropolitan area). In addition, the recording of emergency department cases should account for the majority of patients admitted subsequently as inpatients for treatment for more severe disease.
The early occurrence of the 2003–2004 influenza season, coupled with intense media coverage of pediatric morbidity and death in the later weeks of the outbreak, could have generated significant variation in the coding of cases by KPCO physicians. For example, physicians might be more likely to code influenza in the later weeks of the outbreak and viral infection not otherwise specified, fever, or cough in the earlier weeks. We limited our study to a 19-day period that represented the peak weeks of influenza activity; influenza first appeared in early November and abated in early January 2004.13 Limiting our analysis to the peak weeks would minimize the coding misclassification by clinicians that could be expected during the first and last days of the seasonal influenza epidemic. Finally, it is possible that physicians assigning visit diagnoses might have been less likely to code influenza for vaccinated children, although the widespread publicity regarding the suboptimal antigenic match and vaccine failures likely would have lessened this potential coding bias. However, it is unlikely that a child presenting with fever, cough, or other upper respiratory tract illness symptoms would have an ICD-9 diagnosis code other than those used for defining an ILI (Appendix 1) and therefore would not be identified in our study.
CONCLUSIONS
Influenza vaccine was shown to prevent medically attended ILI among children 6 to 23 months of age, as well as older children, even with a suboptimal match. This study provides supportive data regarding the benefits of vaccination for children and the need for 2 doses among previously unvaccinated young children <9 years of age. The logistics of instituting the influenza vaccine recommendations for young children pose a formidable challenge to private and public health care systems. Additional strategies for fully vaccinating children before the onset of influenza activity should be evaluated.
APPENDIX 1. ILI AND P&I Definitions
ACKNOWLEDGMENTS
We acknowledge the funding contributions of the Centers for Disease Control and Prevention (contract 254-2004-M-06362, CFDA 93.283) and the Kaiser Permanente Colorado Clinical Research Unit.
FOOTNOTES
Accepted Mar 15, 2005.
These data were presented in part to the Advisory Committee on Immunization Practices; June 23, 2004; Atlanta, GA.
No conflict of interest declared.
REFERENCES
Neuzil KM, Mellen BG, Wright PF, Mitchel EF Jr, Griffin MR. The effect of influenza on hospitalization, outpatient visits, and courses of antibiotics in children. N Engl J Med.2000; 342 :225 –231
Izurieta HS, Thompson WW, Kramarz P, et al. Influenza and rates of hospitalization for respiratory disease among infants and young children. N Engl J Med.2000; 342 :232 –239
O’Brien MA, Uyeki TM, Shay DK, et al. Incidence of outpatient visits and hospitalizations related to influenza in infants and young children. Pediatrics.2004; 113 :585 –593
Iwane MK, Edwards KM, Szilagyi PG, et al. New Vaccine Surveillance Network: population-based surveillance for hospitalizations associated with respiratory syncytial virus, influenza virus, and parainfluenza viruses among young children. Pediatrics.2004; 113 :1758 –1764
Centers for Disease Control and Prevention. Prevention and control of influenza: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR Recomm Rep.2004; 53(RR-6) :1 –40
Humiston SG, Szilagyi PG, Iwane MK, et al. The feasibility of universal influenza vaccination for infants and toddlers. Arch Pediatr Adolesc Med.2004; 158 :867 –874
Glazner JE, Beaty BL, Pearson KA, Berman S. The cost of giving childhood vaccinations: differences among provider types. Pediatrics.2004; 113 :1582 –1587
Szilagyi PG, Iwane MK, Schaffer S, et al. Potential burden of universal influenza vaccination of young children on visits to primary care practices. Pediatrics.2003; 112 :821 –828
La Montagne JR, Noble GR, Quinnan GV, et al. Summary of clinical trials of inactivated influenza vaccine, 1978. Rev Infect Dis.1983; 5 :723 –736
Hoberman A, Greenberg DP, Paradise JL, et al. Effectiveness of inactivated influenza vaccine in preventing acute otitis media in young children: a randomized controlled trial. JAMA.2003; 290 :1608 –1616
Hurwitz ES, Haber M, Chang A, et al. Studies of the 1996–1997 inactivated influenza vaccine among children attending day care: immunologic response, protection against infection, and clinical effectiveness. J Infect Dis.2000; 182 :1218 –1221
Clements DA, Langdon L, Bland C, Walter E. Influenza A vaccine decreases the incidence of otitis media in 6- to 30-month-old children in day care. Arch Pediatr Adolesc Med.1995; 149 :1113 –1117
Kramarz P, DeStefano F, Gargiullo PM, et al. Influenza vaccination in children with asthma in health maintenance organizations: Vaccine Safety Datalink Team. Vaccine.2000; 18 :2288 –2294
Kramarz P, Destefano F, Gargiullo PM, et al. Does influenza vaccination prevent asthma exacerbations in children J Pediatr.2001; 138 :306 –310
Hak E, Verheij TJ, Grobbee DE, Nichol KL, Hoes AW. Confounding by indication in non-experimental evaluation of vaccine effectiveness: the example of prevention of influenza complications. J Epidemiol Community Health.2002; 56 :951 –955
Hak E, Wei F, Nordin J, Mullooly J, Poblete S, Nichol KL. Development and validation of a clinical prediction rule for hospitalization due to pneumonia or influenza or death during influenza epidemics among community-dwelling elderly persons. J Infect Dis.2004; 189 :450 –458
Fishman PA, Shay DK. Development and estimation of a pediatric chronic disease score using automated pharmacy data. Med Care.1999; 37 :847 –883
Neuzil KM, Dupont WD, Wright PF, Edwards KM. Efficacy of inactivated and cold-adapted vaccines against influenza A infection, 1985 to 1990: the pediatric experience. Pediatr Infect Dis J.2001; 20 :733 –740
Yih WK, Caldwell B, Harmon R, et al. The National Bioterrorism Syndromic Surveillance Demonstration Program. MMWR Morb Mortal Wkly Rep.2004; 53(suppl) :43 –46
Cox DR. Regression models and life tables (with discussion). J R Stat Soc Ser B.1972; 34 :187 –220
Hosmer DW, Lemeshow S. Applied Survival Analysis: Regression Modeling of Time to Event Data. New York, NY: John Wiley & Sons;1999
Gruber WC, Taber LH, Glezen P, et al. Live attenuated and inactivated influenza vaccine in school-age children. Am J Dis Child.1990; 144 :595 –600
Clover RD, Crawford S, Glezen WP, et al. Comparison of heterotypic protection against influenza A/Taiwan/86 (H1N1) by attenuated and inactivated vaccines to A/Chile/83-like viruses. J Infect Dis.1991; 163 :300 –304
Heikkinen T, Tuuskanen O, Waris M, et al. Influenza vaccination in the prevention of acute otitis media in children. Am J Dis Child.1991; 145 :445 –448
Zangwill KM, Belshe BR. Safety and efficacy of trivalent inactivated influenza vaccine in young children: a summary for the new era of routine vaccination. Pediatr Infect Dis J.2004; 23 :189 –200
Sugaya N, Nerome K, Ishida M, Matsumoto M, Mitamura K, Nirasawa M. Efficacy of inactivated vaccine in preventing antigenically drifted influenza type A and well-matched type B. JAMA.1994; 272 :1122 –1126
Piedra PA, Glezen WP. Influenza in children: epidemiology, immunity, and vaccines. Semin Pediatr Infect Dis.1991; 2 :140 –146
Reichert TA, Sugaya N, Fedson DS, et al. The Japanese experience with vaccinating school children against influenza. N Engl J Med.2001; 344 :889 –896
Monto AS, Davenport FM, Napier JA, Francis T Jr. Modification of an outbreak of influenza in Tecumseh, Michigan, by vaccination of schoolchildren. J Infect Dis.1970; 122 :16 –25
Glezen WP. New vaccines for old diseases: trivalent cold-adapted influenza vaccine. Pediatr Ann.2004; 33 :545 –550
Thomson F, Masters IB, Chang AB. Persistent cough in children and the overuse of medications. J Paediatr Child Health.2002; 38 :578 –581
Laheij RJ, Sturkenboom MC, Hassing RJ, Dieleman J, Stricker BH, Jansen JB. Risk of community-acquired pneumonia and use of gastric acid-suppressive drugs. JAMA.2004; 292 :1955 –1960
Cates CJ, Jefferson TO, Bara AI, Rowe BH. Vaccines for preventing influenza in people with asthma. Cochrane Database Syst Rev.2004; (2) :CD000364
Christy C, Aligne CA, Auinger P, Pulcino T, Weitzman M. Effectiveness of influenza vaccine for the prevention of asthma exacerbations. Arch Dis Child.2004; 89 :734 –735
Bueving HJ, Bernsen RM, de Jongste JC, et al. Influenza vaccination in children with asthma: randomized double-blind placebo-controlled trial. Am J Respir Crit Care Med.2004; 169 :488 –493
Smits AJ, Hak E, Stalman WA, van Essen GA, Hoes AW, Verheij TJ. Clinical effectiveness of conventional influenza vaccination in asthmatic children. Epidemiol Infect.2002; 128 :205 –211(Debra P. Ritzwoller, PhD,)