|Home | About | Journals | Submit | Contact Us | Français|
To determine both practice and child characteristics and practice strategies associated with receipt of influenza vaccine in young children during the 2004–2005 influenza season, the first season for the universal influenza vaccination recommendation for all children aged 6–23 months.
Clinical and demographic data from randomly selected children aged 6–23 months were obtained by chart review from a community-based cohort study in three U.S. counties. The proportion of children vaccinated by April 5, 2005 in each practice was obtained. To assess practice characteristics and strategies, sampled practices received a self-administered practice survey. Practice and child characteristics predicting complete influenza vaccination were determined using multinomial logistic regression.
Forty-six (88%) of 52 sampled practices completed the survey and permitted chart reviews. Of 2384 children aged 6–23 months who were studied, 27% were completely vaccinated. The proportion of children completely vaccinated varied widely among practices (0–71%). Most practices (87%) implemented ≥ 1 vaccination strategy (year-round discussion with parents about influenza vaccine, evening/weekend influenza vaccine clinics, standing orders, or saving a second dose for children who had received the first of two recommended doses). Complete influenza vaccination was associated with three practice characteristics-- suburban location, lower patient volume, and vaccination strategies of evening/weekend vaccine clinics; and with child characteristics of younger age, existing high-risk conditions, ≥ 6 well visits to the practice by age 3 years, and any practice visit from October through January.
Modifiable factors associated with increased influenza vaccination coverage include October-January practice visits and evening/weekend vaccine clinics.
Although influenza vaccine has been recommended for children with certain high risk conditions for decades,1 recent studies have documented the substantial burden of infection in all children..2–6 For this reason, the Advisory Committee on Immunization Practices (ACIP) expanded the universal influenza vaccine recommendations to children aged 6–23 months in 2004–2005,7 to children aged 6–59 months in 2006–2007,8 and to children aged 6 months-18 years in 2009–2010.9
The administration of influenza vaccine presents unique challenges to healthcare providers since it requires yearly administration during a relatively short time period. Vaccine supply delays and shortages can occur, making it difficult for practices to predict when vaccination campaigns might begin and whether adequate supplies will be available. Because of poor immunogenicity of a single dose of vaccine in young children, two doses are required for children aged <9 years who previously have not been completely vaccinated.9 Most studies evaluating methods to increase vaccination rates in young children were reported prior to the universal influenza vaccination recommendations for young children.10–14 Since the majority of childhood vaccines are administered in primary care settings,15 we sought to determine both practice and child characteristics and practice strategies that were associated with influenza vaccine coverage among children aged 6–23 months in 2004–2005, the first season that influenza vaccine was universally recommended.
This study was approved by the institutional review board of the Centers for Disease Control and Prevention (CDC) and of each participating institution in the three communities.
The CDC-funded New Vaccine Surveillance Network (NVSN) conducted in 2004–2005 a community-based cohort study of influenza vaccination coverage in three U.S. counties—Hamilton County, Ohio (Cincinnati), Davidson County, Tennessee (Nashville) and Monroe County, New York (Rochester). This community-based cohort study was designed to estimate influenza vaccine effectiveness in young children over two study years.16 Because these data provided a practice-level assessment of influenza vaccination rates, we surveyed the practices to assess their strategies for influenza vaccination delivery. The combined practice-level chart reviews and practice surveys provided the needed information to assess how practice-level strategies were associated with influenza vaccination rates.
We used two sampling frameworks from a prior study of influenza vaccine effectiveness to identify the study population (Figure 1). The first involved a random sample of primary care practices in each county (n=52 practices from the 3 counties) for practice-based assessment of influenza vaccination rates. The second sampling framework involved nine practices who had participated in the prior study to assess vaccine effectiveness for outpatient visits.16 We utilized probability proportional to size sampling17 from county practices (i.e. more children from larger practices) to select a random sample of the county’s population.6 First, we identified the practices by randomly selecting 30 practice clusters with replacement among the pediatric and family medicine practices in each county that provided care for >30 newborns per year and agreed to participate. Then for each cluster, we randomly selected 30 county children younger than 5 years of age from the patient list generated by that practice,18 as necessitated by the original study for which this cohort was created.16 Each participating practice had between one and eight clusters, depending on practice size. The end result was a sample that represented each county.
Additionally, we randomly selected 249 additional children from 9 of the same practices for chart review to represent children enrolled in their clinic and participating in outpatient influenza surveillance (Figure 1).6 Because we wanted to assess variation among practices and to account for clustering within practices in the analysis, we included all randomly selected children whose charts were reviewed from these practices.
A standardized chart review of the randomly selected children was completed by trained research assistants in the summer of 2005. Data collected included birth date, gender, date of both well and sick visits from birth through May 2005, presence of high-risk medical conditions outlined in the ACIP influenza vaccine recommendations,7 and dates of documented influenza vaccinations. We determined the age of each child on October 1, 2004 and excluded children aged <6 months on December 31, 2004. Race was not recorded on most charts that we reviewed and hence could not be included. To assess the level of preventive care each child received, we measured the number of well-child visits from birth until 3 years of age.
Influenza vaccination status was determined by chart review with inclusion of vaccines obtained outside the practice, if available. Children who received one dose of influenza vaccine in 2004–2005 and had no previous documented doses were considered partially vaccinated. Children were considered completely vaccinated if they received 1) two doses of influenza vaccine at least 24 days apart in 2004–2005 or 2) one dose of influenza vaccine in 2004–2005 and at least one previous dose of influenza vaccine in earlier years, consistent with the 2004–2005 recommendations.7 Children who did not receive any influenza vaccine in 2004–2005 were considered unvaccinated.
A practice survey was created by specialists in pediatrics, infectious diseases, epidemiology and vaccine delivery at the sites and CDC and based on the conceptual framework proposed by the Task Force for Community Preventive Services. From this framework, we selected relevant factors from the literature and immunization strategies recommended by this Task Force.10–14,19 The survey was pilot tested in a convenience sample of pediatric providers and their office managers. The self-administered survey included 22-items concentrating on practice patterns for administering vaccine to children aged 6–23 months. The survey was mailed in early November 2005 to the office manager of each participating practice with an explanatory letter detailing its purpose; a maximum of three reminders were sent to non-responding practices. Office managers were asked to consult with practice physicians to complete the survey; one survey was allotted per practice. The survey included multiple-choice responses with open text field options to characterize the practice demography, the practice’s influenza vaccine supply (2004–2005 had only one half the anticipated vaccine supply),20 the impact of vaccine shortage on vaccine delivery, characteristics of those children who were routinely recommended for influenza vaccine in the practice, and strategies used to identify and administer influenza vaccine to young children.
The following questions ascertained strategies used by the practices. Three yes/no questions included: Did you save a second dose of influenza vaccine for previously unvaccinated children? Did your practice use a standing order system for the 2004–2005 influenza vaccine? Did your practice utilize flu shot clinics? For the latter question, affirmative responders were to check all the applicable times—morning, lunch, afternoon, evening and weekends. Another question asked when providers usually discussed influenza vaccination with parents. Responses included at well visits during influenza season, at sick visits during influenza season, at visits year-round, or not discussed.
Patient-level characteristics obtained from the chart review included age, gender, insurance status (public or private), high-risk medical conditions, health care visits between October-January, dates of well visits from birth until 3 years, and dates of influenza vaccinations. Practice-level characteristics obtained from the survey included practice type, residential setting, practice volume, proportion of children participating in the Vaccines For Children (VFC) program and the adequacy of the practice’s 2004–2005 influenza vaccine supply. Each vaccination strategy was captured. The survey asked about systems to identify vaccine-eligible children and about willingness to vaccinate at all visits; these variables were not analyzed because >90% of practices responded affirmatively to each one. The primary outcome for practice level analysis was complete influenza vaccination.
Chi-squared analyses were used to compare characteristics and vaccination status of individual children. The proportion of children completely or partially vaccinated in each practice was calculated. For practices, we estimated the median proportion and the 95% confidence intervals (CI) of children completely vaccinated by practice characteristics and practice strategies. To evaluate practice and patient characteristics, a multinomial logistic regression analysis accounting for the three potential outcomes—complete, partial or no vaccination—was performed with the child as the unit of analysis, including random effects of site and practices since children within practices were clustered within three study sites.
A question in the practice survey stated: “Did your practice use a recall system in 2004–2005?” Responses included no, yes, or space for an alternative response. We discovered that some practices interpreted a recall system as personalized mailings or phone calls for specific children requiring vaccine, whereas others interpreted a recall system as generalized electronic or postal mailings about influenza vaccine at the practice (i.e. a notice that the vaccine was available). Since this question was understood differently, we excluded recall/reminder systems from analysis.
Statistical analyses were performed using STATA 8.1 (STATACorp), SAS (SAS Institute Inc.) and R version 2.7.2 (R Foundation for Statistical Computing).
Forty-six (88%) of 52 practices responded to the survey and permitted chart reviews on a total of 2384 children aged 6–23 months. The number of children included per practice ranged from 18 to 275. Although influenza vaccine was administered from August 31, 2004 through April 5, 2005 in the practices surveyed, most doses (95%) were administered from October through January.
Among practices, the proportion of children who was completely vaccinated ranged from 0% to 71% with a median of 23%; the proportion who was partially vaccinated in 2004–2005 ranged from 6% to 62% with a median of 24% (Figure 2).
Overall, half of the practices sampled were from Cincinnati and one-quarter each were from Rochester and Nashville. Most practices described themselves as community-based, suburban primary care practices with a mean practice volume of 19 children per doctor per day, with ≤25% of children in the practice receiving vaccine through VFC (Table 1). Eighteen practices (39%) reported an influenza vaccine shortage that limited their ability to vaccinate children aged 6–23 months.
Practice characteristics varied among the three counties with more Cincinnati and Rochester, than Nashville, practices being described as suburban (83%, 83% and 36%, p=0.01). Suburban practices reported having fewer children participating in VFC than urban practices (23% versus 85%, p<0.001) and being less likely to save a second dose of vaccine for children who received the first of two recommended doses (18% versus 46%, p=0.05).
Most practices (87%) reported using at least one strategy to facilitate influenza vaccination, with an average of two strategies per practice. The median complete influenza vaccine coverage increased from 12% to 32% (p=0.29) for practices reporting from none to all four strategies, respectively (Figure 3).
The study children were equally divided by age groups (6-<12 months, 12–17 months and 18–23 months) and by gender with 1735 well-child and 2910 follow-up or acute care office visits occurring from October through January. Most children had private (58%) or public (37%) health insurance; 1% had no insurance and 4% had unknown insurance status. Overall, 9% of children had a high-risk medical condition for which influenza vaccine was specifically recommended.
More children with well-visits from October through January, than without such visits, were completely vaccinated. Although proportions of completely vaccinated children were similar across counties, the proportions of unvaccinated children differed. by location (p=0.03).
A multinomial logistic regression analysis with random effects was performed to identify those practice and child characteristics and practice strategies that predicted complete, partial, or no vaccination during 2004–2005 (Table 2). With the child as the unit of analysis, the model included random effects of sites and practices. Patient characteristics associated with complete influenza vaccination included (a) younger age, (b) high-risk condition for which influenza vaccine is specifically recommended, and (c) at least 6 well-child visits during the first 3 years of life. Practice characteristics associated with complete influenza vaccination included (a) any visit from October through January, (b) suburban practices (with a 3-fold higher vaccination coverage than urban practices), and (c) implementing evening/weekend influenza vaccine clinics (with a 4-fold higher complete vaccination coverage).
Previous studies have examined interventions to improve influenza vaccination in children,10–14 but this study is the first, to our knowledge, to simultaneously assess practice and child characteristics and practice strategies associated with influenza vaccination coverage in geographically diverse sites. We found that two modifiable factors were associated with increased complete influenza vaccinations--any practice visit during October through January and evening/weekend influenza vaccine clinics. In the 2004–2005 study year, the first year when all children aged 6–23 months were recommended to receive vaccine and a year with a vaccine shortage, influenza vaccination coverage varied widely among practices with a range from 0% to 71% for complete vaccination.
Overall, we found that 27% of children aged 6–23 months were completely vaccinated; these coverage rates were comparable to the upper range of those reported in the 2004–2005 National Immunization Survey and from sentinel vaccination sites.21,22 Our estimates were slightly higher than those reported by the Behavioral Risk Factor Surveillance System and a network of practices in New York City and slightly lower than that from Northern California Kaiser Permanente.23–25 Many factors likely influenced the variation in influenza vaccination estimates, including methodology for determining vaccination coverage.23–25
Confirmed by earlier publications, several factors were associated with increased influenza vaccinations, including any practice visit from October through January.26 Also, children aged 6–11 months were more likely to be vaccinated than children aged 18–23 months.27 Whether this difference was due to providers more frequently reviewing vaccination status in younger than older children or more frequent visits in younger children is unknown. Children with high-risk conditions were prioritized during this vaccine shortage.28 Finally, evening/weekend vaccination clinics were associated with increased complete vaccination coverage as was previously predicted.27,29 Discussing influenza vaccine year-round was a parental education strategy associated with a trend to increased complete vaccination. This strategy may be more effective if it was coupled with other strategies, such as year-round scheduling of influenza vaccination visits for the fall.30
One practice strategy, standing orders, is recommended by experts31–33 and has been associated with increased vaccine coverage in adult studies;34–37 however, in our study, it was not associated with increased numbers of complete vaccinations. This paradoxical result likely reflects the inconsistent interpretation of the definition of standing orders,38 the ineffective coupling with other strategies, and/or inability of standing orders to overcome the uncertainty about how many doses each child needed to be completely vaccinated. For young children who need two vaccine doses, standing orders would need to also be coupled with strategies to ensure receipt of the second dose. Further study is needed on standing orders in children.34–37
This study has several limitations. Survey responses may have differed from actual practices.39–41 Variation in vaccination coverage among practices likely reflects a combination of differences among practices, patient populations, and parental priorities.42 All vaccinations received were based on chart review, however, children might have concurrently obtained care from multiple providers and thus we may have missed some influenza vaccinations. The six non-participating practices may have systematically differed from those that participated, but with a practice response rate of 88%, this is unlikely to have affected the overall findings. Since this study was performed in urban and suburban settings, the generalizability of the results to rural counties is unknown. Race is frequently an important variable in vaccination status,18,36,43 however there is a paucity of data about racial/ethnic differences in influenza vaccination status among children. As in many practice-based studies, race was not obtainable from most charts and thus was not included in this study. Additionally, we did not measure operational factors such as leadership, modification of current processes, measurement, reporting, and removal of barriers from practices, which probably influenced the effectiveness of all implemented strategies.44–46
Another limitation is that the 2004–2005 influenza season was the initial season when influenza vaccine was universally recommended for all children aged 6–23 months.7 The season was complicated by a limited supply of vaccine,47 making it difficult for some parents to have their child vaccinated48 and affecting 45% of pediatricians nationally with variation by region, practice type and size.49 We measured the self-reported presence of a shortage within the practice, however, the dynamics of vaccine supply at the practice level, including which practices were impacted, to what extent it impacted public and/or private supplies and for how long, were not fully captured. Additionally, some practices may have pooled their orders, and larger orders were generally filled earlier during this shortage. Thus, these results may not be representative of other influenza seasons. We studied several commonly used practice strategies for vaccine implementation but not all, such as year-around scheduling.30 Because of variations in interpretation of the reminder-recall question, we could not assess this practice strategy, although this intervention has already been shown by other studies to increase influenza vaccination coverage among high-risk children.10,14,50 The strategies we investigated have not been well studied for high-risk children and this is the first study to explore their effectiveness for influenza vaccination of all young children. Further, we lacked the power to assess multi-component strategies that are currently recommended.9
The expansion of influenza vaccine recommendations for children has created a challenge for pediatric providers. We found that influenza vaccine coverage for children aged 6–23 months in 2004–2005 varied widely among practices. Complete vaccination coverage was strongly associated with having a practice visit from October through January and with evening/weekend influenza vaccine clinics, suggesting that practice access is a key component. Our results suggest that high volume practices should be cognizant of the need to implement strategies to better vaccinate children such as scheduling evening/weekend influenza vaccine clinics at least 24 days apart or systematically scheduling second visits for those needing two doses. As physicians work to improve influenza vaccine coverage and as the vaccine recommendations expand to include all children aged 6 months-18 years,9 continued assessment of practice-based strategies to enhance influenza immunization as well as parental preferences and priorities are needed.
As pediatric influenza vaccine recommendations have expanded, coverage for young children has increased but remains sub-optimal. Several practice strategies are recommended to enhance vaccination, including influenza vaccine clinics, reminder-recall systems, year-round discussions with parents, and standing orders.
This study assesses those practice and child characteristics and practice strategies that predict complete influenza vaccination among children 6–23 months of age. It highlights the importance of improving access to influenza vaccine in primary care practices with several strategies.
This project was supported by Cooperative Agreement Number 1 U01 IP000022 from the Centers for Disease Control and Prevention (CDC) and through cooperative agreements U38/CCU217969, U38/CCU417958 and U38/CCU522352. Dr. Poehling received support from Robert Wood Johnson Generalist Physician Faculty Scholar Program, National Institute of Allergy and Infectious Diseases (K23 AI065805) and Wachovia Research Fund. Dr. Edwards receives research funding from the National Institutes of Health, the Centers for Disease Control and Prevention, Wyeth, Sanofi-pasteur, Novartis, and CSL Limited. Dr. Staat receives research funding from the Centers for Disease Control and Prevention, MedImmune, Merck and GlaxoSmithKline. She serves on the advisory board for MedImmune, Merck and GlaxoSmithKline. Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the CDC.
We thank all the practices who participated and made this study possible. We also thank many others who provided invaluable support and contributions. They include Jim Alexander, MD, Carolyn Bridges, MD, John Copeland, SB, MS, Charmaine Coulen, MPH, Aaron Curns, MPH, Marika Iwane, MPH, PhD, Jennifer Reuer, MPH, Ben Schwartz, MD, Ranee Seither, MPH, Frances Walker, MSPH, and John Zhang, PhD at the Centers for Disease Control and Prevention in Atlanta, GA; Nancy Back, BSN, MPH, Michol Holloway, MPH, Harrison Jackson, BS, Marilyn Rice, MS and Meredith Tabangin, MPH at Cincinnati Children’s Hospital Medical Center in Cinicinnati, OH; Carol Ann Clay RN, Diane Kent RN, Erin Keckley, RN, Nayleen Whitehead, and Marie R. Griffin, MD MPH at Vanderbilt University Medical Center in Nashville, TN, Christina Albertin MPH, Geraldine Lofthus, PhD, Caroline B. Hall, MD, and Geoffrey A. Weinberg, MD at University of Rochester School of Medicine and Dentistry in Rochester, NY.
We also want to thank the anonymous reviewers whose suggestions significantly enhanced this manuscript.