 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
ARTICLE |
a Pediatric Research Consortium
b Pediatric Generalist Research Group
c Center for Outcomes Research at Children’s Hospital of Philadelphia, and Departments of Pediatrics and Anesthesiology and Critical Care Medicine, University of Pennsylvania School of Medicine and Children’s Hospital of Philadelphia, Philadelphia, Pennsylvania
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
METHODS. Retrospective cohort data from an electronic health record were used to evaluate the association between demographic, clinical, and immunization variables on immunization delay at 24 months. Patients 2 to 5 years old as of May 31, 2003, with an office visit between May 31, 2002, and May 31, 2003, were selected (N = 5464). Univariate and multivariable models were developed to predict vaccination delay at 24 months per the Advisory Committee on Immunization Practices guidelines.
RESULTS. Overall up-to-date immunization rates at 3, 7, 13, and 24 months were 75%, 45%, 82%, and 71%. Multivariable models using electronic health record data showed that early immunization status was the strongest predictor of immunization delay at 24 months. Multivariate analysis revealed that children who were inadequately immunized at 3 months of age were more than 4.5 times as likely to be immunization delayed at 24 months. In this analysis, patient and caregiver factors associated with immunization delay included insurance status and nonparent caregiver. Children who were premature were less likely to be delayed.
CONCLUSIONS. Using an electronic health record with information entered at the point of care, we found that early immunization status is a strong predictor of immunization delay for young children that can be identified as early as 3 months of age. Electronic health records may prove useful to clinicians and health systems in identifying children at high risk for immunization delay.
Key Words: immunization assessment • medical informatics • quality improvement • urban
Abbreviations: DTP—diphtheria-tetanus-pertussis • DTaP—diphtheria-tetanus-acellular pertussis • MMR—measles-mumps-rubella • Hib—Haemophilus influenzae type b • 4:3:1:3:3—4 DTP/DTaP, 3 polio, 1 MMR, 3 Hib, and 3 hepatitis B vaccines • EHR—electronic health record • CHOP—Children’s Hospital of Philadelphia • OR—odds ratio • CI—confidence interval
Immunization rates have been designated as 1 of 10 leading health indicators for the nation by Healthy People 2010.1 Immunization rates serve as excellent measures of the quality of pediatric care, because immunization schedules are standardized, widely accepted, and optimized to protect children from life-threatening illness. However, levels of vaccination have not universally reached goals of 90% coverage, with children in minority populations at greatest risk of delay.2
Despite the recognized importance of immunizations, multiple studies have found discrepancies in immunization rates between different groups of children. Previous work has demonstrated that mothers living in poverty were more than twice as likely to have incompletely vaccinated children,3 and baseline rates among inner-city children were nearly 20% below those in more affluent suburbs.4 Immunization rates for 4 diphtheria-tetanus-acellular pertussis (DTP/DTaP); 3 polio; 1 measles-mumps-rubella (MMR); 3 Haemophilus influenzae type b (Hib); and 3 hepatitis B vaccines (4:3:1:3:3) at 24 months of age were >13% lower among black children than white children in the 2003 National Immunization Survey for Philadelphia.2 Given these disparities, continued work is needed to devise strategies to best identify young children at risk.
With rapid changes in information technology occurring in US health care systems, an opportunity exists to improve patient outcomes through the application of medical information gathered as part of routine visits. Increasingly prevalent electronic health records (EHRs), which internally maintain immunization records for each child, represent a promising tool in the characterization and reduction of barriers to immunization. The objective of this study was to use data entered at the point of care within an EHR to identify factors that predispose children in an inner-city population to immunization delay. By studying the predictive strength of factors known early in life, this study tests the hypothesis that children at increased risk of immunization delay at 24 months of age can be identified using EHR data by 3 months of age.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
120 pediatric residents. All practices in the network began using the EHR EpicCare (Epic Systems Corporation, Verona, WI) between June 2001 and June 2002. Demographic and financial information for patients seen in the health system before this time was entered into a previous EHR (IDX; General Electric Company, Fairfield, CT) and transferred into EpicCare on implementation of the system within each practice. Clinical information for patients seen before EpicCare implementation was abstracted into the EHRs by clinicians for complex patients or by trained abstractionists for other patients and confirmed during the subsequent office visit. Since implementation, EpicCare has been the sole source of documentation for all health care encounters and telephone calls to each practice. The EHR includes patient demographics, birth, social, family, and past medical history including problem lists with International Classification of Diseases, Ninth Revision codes, medication orders, and laboratory and radiology results.
The immunization database within the EHR provides a comprehensive record of all vaccines received by a patient. To ensure that the desired vaccine product is administered, all orders are dictionary driven and based on Current Procedural Terminology codes, allowing the desired immunization to be selected from a menu of specific options. Furthermore, all ordered immunizations must be recorded at the point of care for charts to be completed. Before EpicCare implementation, immunizations were entered at the point of care via another EHR, TDS 7000 (Eclipsys Corporation, Boca Raton, FL). These records were then transferred electronically to EpicCare. When patients were identified as missing vaccine doses during routine clinical care, records of previous vaccinations were obtained from a citywide immunization database and entered by nursing staff or providers into the EHR. This procedure was followed at both sick and well visits. For new patients from within or outside Philadelphia County, vaccine records were also requested directly from families and their former medical providers and then entered into the electronic record.
Design and Patient Population
This retrospective cohort study included all patients between 24 and 60 months of age as of May 31, 2003, with at least 1 primary care network office visit between May 31, 2002, and May 31, 2003 (N = 5464).
The main outcome measure was up-to-date immunization status at 24 months of age for DTP/DTaP, polio, MMR, Hib, and hepatitis B (4:3:1:3:3). The principal independent variable of interest was immunization delay at 3 months of age. Immunization delay at 7 or 13 months was also considered. Delay was defined according to the recommended childhood immunization schedule and remained consistent throughout the period under study.5–7 To be up-to-date at 3 months of age a child required at least 1 DTP/DTaP, polio, Hib, and hepatitis B vaccine, and to be up-to-date at 7 or 13 months of age a child required at least 3 DTP/DTaP and 2 polio, Hib, and hepatitis B vaccines.
Other independent variables studied included patient and caregiver characteristics. Patient factors included gender, race (categorized as white, non-Hispanic black, Hispanic, Asian, and other), chronic medical problems, and insurance type as of the last office visit during the study period (categorized as commercial, Medicaid, self-pay, or other). The chronic conditions considered were defined by a previously published list, identified from the EHR problem list, and include sickle cell disease, cerebral palsy, prematurity, congenital heart disease, and congenital or chromosomal anomalies.8 Patients with each of these conditions command increased attention in outpatient pediatric practice, and each problem is present at birth and, therefore, precedes any vaccination. Caregiver characteristics studied included relation to the patient as of the last visit (defined as mother, father, or other and captured by the office registration staff), age (
20 or >20 years of age), and income derived from census tract data based on 5-digit zip codes and divided into lowest quartile, middle 50%, and highest quartile for Philadelphia. For the income scale, the lowest group had family incomes of or less than $21329, the middle group had incomes between $21329 and $30099, and the highest group had incomes of more than $30099.
Statistical Analysis
Demographic, clinical, and immunization status variables were described by using proportions. In the first phase of the analysis, the relationship between individual independent variables and immunization status at 24 months of age was determined by using unadjusted relative risks. To specify the absolute increase in risk of immunization delay at 24 months for each independent variable, the attributable risk of immunization delay was calculated. The attributable risk was defined as the risk of delay at 24 months in those with the variable of interest minus the risk in all others or, where specified, minus the risk in the reference group (for variables such as race and insurer).
We next developed 4 logistic-regression models to predict vaccination delay at 24 months of age. A model that included all demographic and clinical variables, as well as immunization status at 3 months, was the central focus of the analysis. For comparison, we developed a base model with all demographic and clinical variables but no previous immunization status, as well as 2 additional models that added immunization status at either 7 or 13 months to the base model. The log-likelihood ratio statistic was used to evaluate the overall significance of the models, and the C statistic was calculated to explain the discriminative ability of each model. To ensure that minimal bias was introduced by the implementation of the EHR during the study period, a limited analysis was performed by using these models with only patients <36 months of age.
To validate the multivariable models, patients were randomly assigned to a derivation or validation data set. The multivariable models were initially built by using the derivation data set and then applied to the validation set. After testing for differences in model parameters between data sets and not finding significant ones (P value for likelihood ratio test = .40), the 2 data sets were combined for the estimation of the final models.
The study was approved by the institutional review board of CHOP.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Immunization Rates
Overall up-to-date immunization rates for the entire study sample at 3, 7, 13, and 24 months were 75%, 45%, 81%, and 71%, respectively. If the population was limited to children born between June 1, 2000, and May 31, 2001 (the youngest patients in the cohort), rates at 3, 7, 13, and 24 months were similar (78%, 44%, 81%, and 74%, respectively).
Predictors of Immunization Delay at Age 24 Months
In the univariate analysis (Table 1), early immunization delay was most strongly associated with delay at 24 months of age. In particular, children with immunization delay at 3 months of age were 2.6 times more likely to be delayed and had a 34% absolute increase in risk of immunization delay at 24 months of age. Patient and caregiver characteristics were overall less strongly associated with immunization status at 24 months of age. However, children in the "other" race category, with sickle cell disease, Medicaid, no insurance, or "other" insurance, or a nonparent as the primary caregiver were significantly more likely to have immunization delay at 24 months. Prematurity was associated with improved immunization rates.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
In this study, among a broad range of patient-, caregiver-, and immunization-specific predictors, one factor (early immunization delay) was the most important factor predicting being up-to-date at 24 months. Even adjusting for all factors considered, children with delay at 3 months were >4.5 times more likely than other children to be delayed at 24 months. Because delayed early vaccination status is easily identified and has proven to be predictive of incomplete immunization in multiple settings, it could be used as a practical marker for providers to target outreach to patients at highest risk.9–14 Beginning these interventions at an early age is especially important, because visits are most frequent during the first year of life. More broadly, efforts to improve immunizations rates may have far-reaching consequences in improving the overall delivery of preventive services to young children.3,15,16
Supporting the validity of the EHR as an immunization research tool, the results obtained from this study match those of previous studies that have used a variety of retrospective and prospective strategies to identify predictors of immunization delay in high-risk populations. Consistent with previous studies, we found that insurance status and family structure are associated with immunization status.3,11,17,18 In our study, having a nonparent as the primary caregiver for a child compared with the mother increased the risk of immunization delay at 2 years of age. Children without commercial insurance were more likely to be delayed. In addition, although multiple medical conditions present from birth were considered, only prematurity and sickle cell disease were found to be significantly associated with immunization status. This pattern is consistent with earlier work that did not show an association between many chronic medical conditions and immunization status.18
In relying on EHR data, our study has several strengths and limitations. EHR data are readily available on large patient populations, do not require the time or expense associated with prospective data collection, and are derived from the actual patient chart without translation. However, during the implementation of an EHR, the potential exists for information to be omitted. The electronic translation of immunization records, as well as demographic and insurance information directly from other electronic systems in the practices studied, has been shown to result in more complete data.19 For clinical information, the dual processes of abstraction and physician review used in the health system studied ensured that providers had the opportunity to review the information in the EHR. Finally, the consistency of results between the youngest patients in the cohort, those most likely to have received care using the current EHR before 24 months of age, and the overall group suggests that results were not biased by the implementation process.
As with any patient chart, the completeness of EHR data for children joining practices after the neonatal period depends on obtaining patient records. For immunizations, a lack of early vaccine records for patients transferring into the practices studied could have potentially increased the measured effect of not being vaccinated by 3 months of age on immunization delay at 24 months of age, because these children may not have been caught up on immunizations by 24 months and did not have shots in our health system by this early age. To address this concern, we again considered the association of early immunization delay and delay at 24 months for only patients <36 months of age at the end of the study period, the subset of the population most likely to have received care in our health system at an early age. The persistence of the highly significant association of lack of immunization by 3 months and immunization delay at 24 months in this group suggests that this association did not arise from bias.
Despite these limitations, the EHR-based approach used for this study is particularly effective in tracking immunization status, the main outcome considered. Strongly supporting the validity of the immunization record in the EHR, the rate for 4:3:1:3:3 found in our study at 24 months of age of 71% up-to-date very closes matches the overall rate for Philadelphia throughout the study period. According to the National Immunization Survey, rates at 24 months were 68% in 1998, 69% in 2000, and 70% in 2003.2,20,21 Further supporting the validity of the EHR for immunization research, previous studies have found that immunization databases like the one we used are less likely than other systems to have missing doses because information is entered at the point of care.18 By conducting an electronic review of charts through the EHR, the process of manual chart review, subject to human error and involving considerable expense and time, was eliminated. Also limiting bias from missing vaccine doses, children with immunization delay at any of the 4 centers involved in this study had their vaccinations checked against the Philadelphia Kids Immunization Database/Tracking System Registry, a citywide immunization database, during sick or well visits when delay was recognized. Records were then updated directly within the EHR. The use of such registries to supplement data in patient charts has previously been shown to provide a better estimate of the number of children who are truly up-to-date.22 As a result, although doses administered at nonstudy practices may have been missed, immunization data available reflects vaccination delivery both inside and beyond the health system considered.
EHRs that include immunization records, as well as clinical and demographic data from across multiple practices or entire health systems, represent a remarkable new resource to identify and study characteristics of children at risk of immunization delay at an early age. Using EHRs, groups of children can readily be identified by such factors as birth date, age at first or last visit, insurance type, zip code, or a specific diagnosis. Because information is collected prospectively as part of routine care, recall bias, even in a retrospective study, is minimized. Furthermore, although information in paper charts often requires interpretation because of issues of handwriting or a lack of strict matching of diagnoses to diagnostic codes, all immunizations and medical problems within the EHR are numerically coded and unique. In addition, the collection of clinical, demographic, and billing information in 1 system offers the opportunity to better understand the relationship of factors in each of these domains on the delivery of immunizations.
The ability to identify children at high risk of immunization delay from an early age is a key step in meeting the national goal of improved immunization rates. To meet this objective, the discussion of immunization practices has shifted increasingly to practical approaches for quality improvement.15,23–25 The most dramatic benefit of the EHR may be the ability to collect data, which has occurred in this project, and then provide clinical alerts to providers which influence care in the office or trigger contact with families to encourage appointment scheduling. On the basis of EHR data, providers may also be educated regarding their own performance on such measures as vaccination rates. However, although work has begun to show the promise of office-based systems in improving preventive care, the potential impact of EHRs on the quality of care has not been widely studied in pediatrics.23,24,26 Through the electronic review of >5000 charts within an EHR, we have found that immunization delay by as early as 3 months of age is strongly predictive of incomplete immunization at 2 years of age. As EHRs become increasingly prevalent, electronically captured clinical and demographic data may be easily used to guide pediatric care and facilitate rapid cycle quality improvement efforts for the health of children.
|
ACKNOWLEDGMENTS |
|---|
We thank Jonathan Crossette, MPH, Robert Grundmeier, MD, and Orit Even-Shoshan for their helpful comments and technical insights during the course of this research. In addition, we recognize Barbara Watson, MD, for thoughtful review and Susan Healis for administrative support.
|
FOOTNOTES |
|---|
Address correspondence to Alexander G. Fiks, MD, Primary Care Center at University City, Children’s Hospital of Philadelphia, St Leonard’s Court, Suite 110, 39th and Chestnut Streets, Philadelphia, PA 19104. E-mail: fiks{at}email.chop.edu
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2006 American Academy of Pediatrics. All rights reserved. |
||
|
||
