Copyright 2004 American Medical Association. All Rights Reserved. Applicable FARS/DFARS Restrictions Apply to Government Use.2004
Influenza vaccination effectively reduces influenza-related morbidity in children but is underused. The pediatric emergency department is a potential intervention point for increasing influenza vaccination in children.
To assess the effectiveness of a pediatric emergency department–based influenza vaccination program.
A prospective, randomized, controlled clinical trial.
We recruited subjects from persons seeking pediatric emergency care at a large tertiary care hospital during influenza vaccination season in 2002. Eligible subjects were themselves or had a household member at increased risk for influenza complications, according to Advisory Committee on Immunization Practices guidelines.
We randomly assigned subjects by family to the “vaccine offered” group or the “education only” group. We educated both groups about influenza illness, the risks and benefits of influenza vaccination, and how to receive influenza vaccination outside the study. We offered influenza vaccination to the “vaccine offered” group and later obtained telephone reports of the postinfluenza-season vaccination status of both groups.
Four hundred thirty-seven subjects completed the study (238 from the “vaccine offered” group, 199 from the “education only” group). At follow-up, the percentage of influenza-vaccinated subjects in the “vaccine offered” group was greater than in the “education only” group. This was true for pediatric patients (57% vs 36%) (relative risk, 1.59; 95% confidence interval, 1.16-2.16) as well as for their accompanying adult and child family members (75% vs 34%) (relative risk, 2.19; 95% confidence interval, 1.64-2.92).
A pediatric emergency department–based influenza vaccination program increases vaccination of targeted children and family members who use the emergency department during influenza vaccination season.
Influenza may result in serious disease, especially among chronically ill children.1- 3 Inactivated influenza vaccination has been shown to be safe and effective at preventing influenza and its complications among pediatric patients.4- 6 For this reason, the American Academy of Pediatrics7 and the Advisory Committee on Immunization Practices8 recommend influenza vaccination for children, who are at an “increased risk” for complications from influenza. Specifically mentioned are those children with disorders of the pulmonary and cardiovascular systems, metabolic disease, renal disease, immune deficiency, hemoglobinopathy, and those receiving chronic aspirin therapy.8 Nevertheless, pediatric influenza vaccination is underused among these children.8 For example, during any given vaccination season, only 10% to 25% of children with asthma receive the vaccine.9- 11
The benefits of influenza immunization are not limited to chronically ill children. Among otherwise well children, excess hospitalization, increased outpatient visitation, and increased antibiotic use during influenza season are clinically relevant events.3 This is especially true of children younger than 2 years of age, for whom the incidence of influenza-related hospitalization is higher than it is in many adult groups considered to be “high risk.”2,3,8 Previously, the American Academy of Pediatrics7 and the Advisory Committee on Immunization Practices8 have “encouraged” vaccination of children 6 months to 23 months of age. Then, in October 2003 and March 2004, the Advisory Committee on Immunization Practices12 and the American Academy of Pediatrics13 voted to make influenza vaccination a routine recommendation for children in this age group beginning fall 2004. These groups also recommend vaccination of household members of individuals considered to be at “increased risk”8 because household contacts represent a source of influenza infection to those at high risk for influenza complications.
There are many possible reasons for failure to receive influenza vaccination. Lack of a primary care office visit during the traditional influenza vaccination period may account for almost half of the nonadherence to current influenza vaccine recommendations.9 An emergencydepartment (ED) visit may offer a unique opportunity for increasing use of influenza vaccination in the children who need it because almost three quarters offrequent pediatric ED users have chronic illness.14Additionally, pediatric ED users represent an undervaccinated population,15 so an ED-based influenza vaccination program might reach children who have increased barriers to vaccination.
The concept of using the pediatric ED for increasing childhood vaccination is not novel. The 1993 “Standards for Pediatric Immunization Practices” set forth by the US Department of Health and Human Services included the ED visit as a point of intervention for screening immunization status and administering vaccines.16 Although there have been many other proponents of vaccination in the pediatric ED setting,15,17,18 significant potential problems also have been identified.19 More recently, the National Vaccine Advisory Committee’s “Standards for Child and Adolescent Immunization Practices” de-emphasized vaccination by health care providers other than the child’s primary care physician, citing the benefits of a “medical home.”20 However, the standards continued to support vaccination of some children by other health care providers when the child’s main contact with the health care system is not in a primary care office.
In theory, a pediatric ED–based influenza vaccination program might be expected to increase vaccination of children for whom the vaccine is indicated. On the other hand, contraindications and precautions to vaccination because ofintercurrent illness or parental preference to have children vaccinated by the primary care physician might offset any potential benefit of offering influenza vaccination in the ED.
The objective of this study was to assess the effectiveness of a pediatric ED–based influenza immunization program.
The study was a prospective, randomized clinical trial comparing an intervention with a control.
The intervention took place in a tertiary care hospital in Rochester, NY, during a 14-week period from September 24, 2002, to December 31, 2002. The start date corresponded to the first date that influenza vaccine was available through the hospital pharmacy during the influenza vaccination season. Trained enrollers approached all families who arrived with a child (younger than 19 years of age) at the pediatric ED or fast track area between 9 AM and midnight. Families were encouraged to read a double-sided page on risk factors for influenza-related morbidity and mortality, influenza illness, influenza vaccine risks and benefits, and how to receive the influenza vaccine outside the study. Families were then asked if they were eligible for enrollment (ie, if anyone living at the patient’s address was in an increased risk category for influenza-related morbidity and mortality) (Figure 1). Only family members present in the ED were eligible to participate. Family members choosing to enroll signed institutional review board–approved consent or assent forms or had consent forms signed by a guardian, depending on the subject’s age. Willingness to be vaccinated was not a requirement of study participation.
Criteria used for defining individuals as being at “increased risk.”
Family members who were not present in the ED and persons who had previously enrolled were not eligible to enter the study.
In this study, the family, not the individual patient, was the unit of randomization. All enrolled families received educational materials regarding influenza illness, indications and contraindications for influenza vaccination, and how and where influenza vaccination could be obtained outside the study. Enrolled families filled out health questionnaires to determine whether each family member had current indications or contraindications to influenza vaccination. After these steps, families were then randomly assigned to 1 of 2 study arms: the “vaccine offered” (VO) intervention group or the “education only” (EO) control group. The assignment of each family was in a sealed envelope distributed by order of enrollment. Each envelope contained an assignment that had been randomly generated with a block of 4 restriction. All those randomized to the EO group received no further intervention. Influenza vaccination was not offered in the ED outside the study. Each participating member of a family randomized to the VO group was offered influenza vaccination if they did not have a contraindication or precaution (Figure 2).8,21
Criteria used for defining contraindications and precautions to vaccination. This study did not provide repeat vaccinations for children younger than 9 years who had recently received their first influenza vaccine. Pregnant women were instructed to see their obstetrician regarding the timing of vaccination.
For those who chose to be vaccinated, signed informed vaccine consent forms were obtained prior to the vaccination. Vaccination procedure followed standard guidelines for administration with the split antigen inactivated influenza vaccine Fluzone (Aventis Pasteur, Swiftwater, Pa). Subjects were provided a copy of the vaccination record and instructed to give a copy to their primary care physician. We recommended that subjects younger than 9 years of age who were receiving influenza vaccination for the first time follow up with their primary care physician to receive a second influenza vaccination at least 4 weeks after the first. Subjects were informed that the cost of vaccination would be billed to the subject vaccinee or to the subject vaccinee’s insurance company.
During April and May 2003, one investigator not blinded to the study arm (D.P.) attempted to make telephone contact with enrolled families from both study arms. An adult family member was asked to detail whether or not each subject had been vaccinated against influenza. For both the VO group and the EO group, the outcome measure was the stated vaccination status of the subjects based on the telephone follow-up interview. This study did not differentiate a full vs partial vaccination status for those subjects younger than 9 years of age who might or might not have received 2 vaccinations.
This study was approved by the University of Rochester research subject’s review board.
We used χ2 for evaluating the significance of differences between the VO and the EO groups with regard to sex, age, insurance, contraindications to vaccination, and vaccination status at enrollment. Relative risks with 95% confidence intervals were calculated for comparison of vaccination status between the groups. Additionally, logistic regression using SAS System for Windows 8.2 (SAS Institute Inc, Cary, NC) was performed to identify and control for potentially confounding independent variables, modeling for vaccination status as the main outcome variable. Stepwise selection was used to construct the preliminary models with P ≤ .01 for entry and P > .05 for removal. The following independent variables were studied: (1) study arm assignment (VO vs EO); (2) sex; (3) insurance status; (4) age younger than 6 months; (5) age 6 to 23 months; (6) asthma; (7) heart disease; (8) sickle cell disease; (9) diabetes; (10) unspecified serious chronic illness; (11) immune deficiency; (12) high-risk family member; (13) younger than 18 years of age and receiving long-term aspirin therapy; (14) influenza vaccination prior to enrollment; (15) current moderate, severe, or febrile illness; (16) history of Guillain-Barré syndrome; (17) allergy to latex, chicken egg, or thimerosal; (18) previous bad reaction to influenza vaccine; and (19) pregnancy.
During the enrollment period, enrollers identified 1183 eligible families; 846 families (72%) refused to participate. Among the 337 families participating in the study, 171 were randomized to the VO group and 166 to the EO group. Because our primary focus was the vaccination of children, families who enrolled only adult family members were not included in the telephone follow-up or data analysis (39 in the VO group, 34 in the EO group). During the follow-up period, we were able to contact all but 22 families in the VO group (106/128, 83%) and all but 31 families in the EO group (101/132, 77%). Three families (all from the VO group) were excluded from data analysis because study protocol was not followed. One family (VO group) withdrew from the study.
Vaccination data from telephone follow-up were available for 207 families (106 from the VO group, 101 from the EO group). These families represented a total of 437 subjects (238 in the VO group, 199 in the EO group). Figure 3 provides additional details regarding the flow of subjects from recruitment to completion of the study.
Enrollment allocation follow-up analysis.
Figure 4 displays the indications for influenza vaccination for those children who completed the study. Many children had more than 1 indication. Among adults, 91% had an increased-risk family member, 18% had asthma, 11% had an unspecified serious chronic disease, 10% had diabetes, and 3% or less reported having 1 of the following: heart disease, immune deficiency, sickle cell disease, or pregnancy.
Indications for influenza vaccination. Many children had more than 1 qualifying condition.
Table 1 describes the characteristics of the subjects completing the study. Although a larger proportion of subjects in the VO group was insured than in the EO group, the difference was smaller and not statistically significant when only children were considered (94% vs 88%) (χ2 = 2.61, P = .10). A smaller proportion of individuals randomized to the VO group than to the EO group reported having received influenza vaccination prior to enrollment.
More children than adults had vaccine contraindications and precautions at the time of enrollment (43 children, 15 adults). These included current moderate or severe illness: 32 children (22 VO, 10 EO), 1 adult (VO); allergy to latex, chicken egg, or thimerosal: 7 children (2 VO, 5 EO), 6 adults (1 VO, 5 EO); previous bad reaction to the influenza vaccine: 1 child (EO), 2 adults (1 VO, 1 EO); and pregnancy: 3 children (1 VO, 2 EO), 6 adults (3 VO, 3 EO).
During the telephone follow-up interview, some interviewees reported a vaccination status that contradicted a previously reported vaccination status or evidence of vaccination obtained as part of the study. Of 319 subjects not vaccinated in the pediatric ED, 5 subjects (all VO group) reported receiving influenza vaccination in the pediatric ED, while 314 (98%) correctly reported that they had not been vaccinated in the ED. Of the 120 subjects who were vaccinated in the pediatric ED, 8 subjects (all VO group) reported never receiving influenza vaccine, while 112 (93%) correctly reported being vaccinated in the pediatric ED. Of 72 subjects enrolled in the study with a stated status of already having been vaccinated against influenza during the 2002-2003 season, 13 (5 VO group, 8 EO group) reported never having been vaccinated at telephone follow-up, while 59 (82%) did not alter their reported vaccination status. As noted above, for both the VO and the EO groups, the outcome measure was the stated vaccination status of family members based on the telephone follow-up interview.
At the end of the influenza season, vaccination status as assessed by telephone report was analyzed both by family unit and by subject. For the analysis by family unit, the outcomes were 1 or more family members vaccinated and 50% or more family members vaccinated. A greater proportion of families had 1 or more family members vaccinated in the VO group (80/106; 75%) as compared with the EO group (56/101; 55%) (relative risk, 1.36; 95% confidence interval, 1.11-1.67). Similarly, there was a greater proportion of families with 50% or more members vaccinated in the VO group (49/106; 46%) as compared with the EO group (33/101; 33%) (relative risk, 1.41; 95% confidence interval, 1.00-2.00).
Table 2 shows the vaccination frequency for individual subjects by study arm as assessed by telephone report after influenza season. Subject groups are shown as a whole and by subgroup. In every category, subjects in the VO group reported higher vaccination rates than did subjects in the EO group. For the 80 vaccinated children in the VO group, 53 (66%) reported vaccination in the pediatric ED rather than in another setting.
Logistic regression analysis of independent variables is shown in Table 3. Bivariate analysis of the identified covariates against study arm revealed no significant associations except for the previously reported association with influenza vaccination prior to enrollment (Table 1). Age younger than 6 months was present in12 (5.0%) of 238 in the VO group and 9 (4.5%) of 199 in the EO group (χ2 = 0.06, P = .80). Heart disease waspresent in 6 (2.5%) of 238 in the VO group and 9 (4.5%) of 199 in the EO group (χ2 = 1.31, P = .30). Unspecified serious chronic illness was present in 26 (10.9%) of 238 in the VO group and 46 (23.1%) of 153 in the EO group (χ2 = 0.06, P = .80).
This study has demonstrated a clinically important increase in the influenza vaccination rates of targeted children seeking care in a pediatric ED, and their accompanying family members, when an influenza vaccination program is in place. Previous publications have addressed ED-based influenza vaccination of adults and routine childhood vaccination in the pediatric ED. Many of these studies have been uncontrolled feasibility studies. This study addresses the efficacy of a pediatric ED–based influenza vaccination program in a prospective, randomized, controlled manner.
Previous studies of influenza vaccination in an ED setting have focused on adult patients in adult EDs. In 1987 and 1988, Polis et al22,23 published the results of questionnaire interviews that demonstrated a willingness of “high risk” seekers of adult emergency care to accept influenza vaccination in the ED. Later, in 1993, Rodriguez and Baraff24 were able to vaccinate 50% of influenza-unvaccinated adults in their ED study model. Most recently, in 1998, Slobodkin et al25 demonstrated 61% vaccination of “high risk” adult emergency care seekers who had no contraindications when a nursing incentive program was instituted. None of these studies compared vaccination frequency attained through their vaccination programs with a control group. Thus, the possible effectiveness of these programs can only be inferred with relation to the low background rate of influenza vaccination.
Similarly, Cunningham26 demonstrated that pediatric ED vaccination with routine childhood vaccines was “feasible”; during a 2-year period, 2514 of 9321 children seen in a pediatric ED were given 6482 vaccinations. However, without controls, it could not be established that overall vaccination frequency was increased in an important way. It may have been that those who were not up-to-date would soon have been vaccinated at an office visit with the primary care provider. When routine childhood vaccination in the pediatric ED was studied in a prospective manner by Rodewald et al27 (randomized, controlled clinical trial) and Szilagyi et al28 (cohort), there was very little or no improvement in overall vaccination frequency.
The value of a pediatric ED–based influenza vaccination program must be measured in relation to the proportion of children who would not otherwise have been vaccinated. In this study, all intervention subgroups demonstrated an important increase in vaccination frequency when compared with their control groups. However, a pediatric ED–based influenza vaccination program appears to be of particular value to those targeted children who have not been influenza-vaccinated by December for the current season. October and November are the optimal times for individuals at increased risk for influenza-related morbidity and mortality and their household contacts to be vaccinated.8 Targeted children unvaccinated by December appear to be particularly unlikely to be vaccinated. In our study, in the absence of a pediatric ED–based influenza immunization program, only 10% of children unvaccinated in December went on to be vaccinated. For this group, explorational analysis revealed an almost 7-fold increase in influenza vaccination by the end of influenza season when the vaccine was offered during the ED visit as compared with when education alone was provided.
While it may be debated whether the better policy is to wait until December to start a pediatric ED–based influenza vaccination program out of respect for a patient’s medical home, some children will benefit from initiating such a program as soon as influenza vaccine becomes available.
There are some limitations to this study. This study was performed before the recommendation to routinely vaccinate children aged 6 to 23 months was made. During the study period, the Advisory Committee on Immunization Practices guidelines stated that influenza vaccination in this age group was “encouraged when feasible.”8 The benefits of a pediatric ED–based influenza vaccination program may not be as compelling when the new guidelines become active.
Another concern is that the proportion of participants relative to the number of eligible participants (28%) leaves room for self-selection bias. The individual reasons for refusal to enter the study are not known. However, nonparticipating families may have included a substantial proportion who would not choose to use the ED for influenza vaccination. Although a limitation, this low rate of participation may not be unusual. In a survey of urban ED patients and visitors, Wilets et al29 found that only 72% of potential subjects were willing to take the survey, and of this select group, only 47% stated a willingness to participate in medical center–sponsored research. The low participation rate raises concerns about the feasibility of such a program. However, if a pediatric ED–based influenza vaccination program were adopted as standard of care, we speculate that the refusal rate would be much lower.
The nature of our data allowed an understanding of some of the recall error present in our data related to subject report of vaccination status. The proportion of recall errors is within the range reported by other studies of influenza vaccination status awareness and recall. Bedford and Howell30 reported a 97% accuracy comparing questionnaire results with general practitioner office records for an elderly population in Ireland. A Canadian survey of influenza vaccination status validated by comparison with office records found a sensitivity of 93% and a specificity of 98%.31 On the other hand, a large survey in Switzerland revealed that only 72% of adults were aware of their vaccination status.32
The random assignment of subjects led to a higher proportion of subjects in the EO (control) group who were already vaccinated against influenza before the study began. This inequity, however, would bias our results toward the null. Random assignment also led to a greater proportion of insured subjects in the VO group than in the EO group. This difference, although small (93% vs 87%), might have biased toward showing an effect, assuming the insured subjects were more likely to seek and receive preventive health care such as influenza vaccination.
Another limitation is that our outcome measure, reported vaccination status, did not differentiate between those first-time vaccine recipients under 9 years of age who received only 1 dose of influenza vaccine instead of following the 2-vaccine-dose schedule. If vaccination status were measured only in terms of full vaccination according to the 2-dose schedule when appropriate, it is possible that our numbers would overestimate the proportion vaccinated. This overestimation would be expected to be relatively similar in both the VO and the EO groups. However, we cannot discount the possibility that those children vaccinated with a first dose in a primary care office might have a second-dose follow-up arranged at the time of the first vaccination. The follow-up in this case might be expected to be superior to the verbal and written directions for follow-up provided by the pediatric ED–based influenza vaccination program model that we used.
Finally, this study does not address whether a pediatric ED–based influenza vaccination program would be expected to increase vaccination of targeted children in the population at large. Although 74% of children who pass through a pediatric ED have chronic illnesses that would target them for influenza vaccination,14 it is not known what proportion of targeted children pass through an ED during any given influenza vaccination season. It is also not known what proportion of families would accept vaccination if it were not part of research but routine operating procedure in the ED.
In summary, influenza vaccination is an underused means of decreasing influenza-related morbidity and mortality in children. Efforts need to be made to increase influenza vaccination of targeted children and their families. Although influenza vaccination in the office of the primary care physician remains the goal for influenza vaccine–targeted children, this study suggests that a pediatric ED–based vaccination program is one means of increasing influenza vaccination in targeted children and their families.
Vaccination is the primary means of preventing serious morbidity and mortality from influenza disease. Annual influenza vaccination rates in the 10% to 20% range per vaccination season for asthmatic children suggest that the vaccine is underused. Effective ways to increase vaccination of targeted children are necessary.
This study demonstrates that a pediatric ED–based influenza immunization program is one means of increasing influenza vaccination in the select population of targeted children who visit an ED during influenza vaccination season.
Correspondence: Dante Pappano, MD, 601 Elmwood, Box 655, Rochester, NY 14642 (email@example.com).
Accepted for Publication: July 15, 2004.
Funding/Support: This study was supported in part by the Ronald McDonald House Charities, Oak Brook, Ill, whose contributions were without constraint on the design, implementation, interpretation, or publication of the study.
Pappano D, Humiston S, Goepp J. Efficacy of a Pediatric Emergency Department–Based Influenza Vaccination Program. Arch Pediatr Adolesc Med. 2004;158(11):1077-1083. doi:10.1001/archpedi.158.11.1077