Chronic lung disease (CLD) diagnosis: International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 770.7 at any time during the 24 months of age or an inpatient or outpatient claim with ICD-9-CM code 496 occurring between birth and 28 days of age. CF indicates cystic fibrosis; CHD, congenital heart disease; and SCID, severe combined immunodeficiency.
Predicted risk of hospital admissions for RSV per 1000 patient season-months for healthy term infants (T; black line) and preterm infants with CLD (P; orange line) by age. Incidence rates are averaged across all modeled covariates (sex, race/ethnicity, plurality, palivizumab exposure, season-month, season-year, and state). The 95% CI bands (dotted lines; black dotted line indicates healthy term infants and orange dotted line indicates preterm infants with CLD) are very narrow for term infants owing to a large sample size. The age at which the predicted risk of hospitalization for RSV approximates the risk of healthy term infants at 1 month of age is 18.5 months (blue horizontal and vertical lines).
eAppendix. Codes for the Identification of Palivizumab Exposures
eTable. Sensitivity Analysis for the Effect of Missed Respiratory Syncytial Virus Hospitalizations on Age Threshold Estimates (for Primary Analysis)
Customize your JAMA Network experience by selecting one or more topics from the list below.
Winterstein AG, Choi Y, Meissner HC. Association of Age With Risk of Hospitalization for Respiratory Syncytial Virus in Preterm Infants With Chronic Lung Disease. JAMA Pediatr. 2018;172(2):154–160. doi:10.1001/jamapediatrics.2017.3792
At what age does the risk of hospitalization for respiratory syncytial virus among preterm infants with chronic lung disease approximate the risk of healthy, 1-month-old term infants?
In this cohort study of 1 018 593 healthy term and 5181 preterm infants with chronic lung disease, the risk of hospitalization for respiratory syncytial virus among preterm infants with chronic lung disease became similar to that of 1-month-old term infants at an age of 18.5 months when other risk factors were controlled.
The identified age threshold supports the current guideline, which recommends limiting immunoprophylaxis for respiratory syncytial virus up to the second year of life among infants with chronic lung disease.
It is unknown whether the age threshold (≤24 months) for preterm infants with chronic lung disease (CLD) to receive immunoprophylaxis for respiratory syncytial virus (RSV) as currently recommended by American Academy of Pediatrics guidelines correctly identified infants at higher risk for hospitalization for RSV.
To determine the age when the risk of hospitalization for RSV among preterm infants with CLD becomes equivalent to the risk for healthy, 1-month-old term infants who do not qualify for immunoprophylaxis.
Design, Setting, and Participants
A retrospective cohort study was conducted of 1 018 593 healthy term infants and 5181 preterm infants with CLD using Medicaid billing records (Medicaid Analytic eXtract files) from January 1, 1999, to December 31, 2010, linked to Florida and Texas birth and death certificates.
Age-trend discrete time logistic regression models within a survival analysis framework were developed, adjusting for covariates including the use of immunoprophylaxis, to compare the risk of hospitalization of preterm infants (<32 weeks’ gestational age) with CLD at 3 through 34 months of age with the risk of hospitalization of term infants (37-41 weeks’ gestational age) at 1 month of age.
Main Outcomes and Measures
Age at which risk of hospitalization for RSV among preterm infants with CLD equals the risk for healthy term infants at age 1 month.
The study cohort included 1 018 593 healthy term infants and 5181 preterm infants with CLD; because patients could reenter the cohort for a second or third season, the total study cohort consisted of 1 880 531 healthy term infant-seasons (926 206 girls and 954 325 boys; mean [SD] age at first season entry, 12.6 [9.6] months) and 8680 CLD infant-seasons (3519 girls and 5161 boys; mean [SD] age at first season entry, 15.1 [9.1] months). Among term infants with siblings, the risk of hospitalization for RSV averaged across all covariate strata was 9.0 (95% CI, 8.4-9.6) per 1000 patient season-months at 1 month of age. The risk of hospitalization for RSV among preterm infants with CLD became similar to that of 1-month-old term infants at an age of 18.5 months (95% CI, 15.6-22.8).
Conclusions and Relevance
The age threshold at which the risk of hospitalization for RSV among qualifying preterm infants with CLD approximates that of healthy term infants supports the current American Academy of Pediatrics practice guideline recommending RSV prophylaxis until a maximum of 24 months of age.
Respiratory syncytial virus (RSV) is one of the most common causes of respiratory tract infections among children,1,2 and manifestations may require hospitalization and intensive care.3 Preterm infants with chronic lung disease (CLD) of prematurity have an increased risk of hospitalization for RSV compared with healthy term infants, but the duration of this elevated risk remains incompletely defined.4 Passive RSV immunoprophylaxis is available based on the results of a randomized clinical trial involving 782 infants with CLD, which showed a significant reduction of hospitalization for RSV.5
Chronologic age has consistently been shown to be one of the strongest risk factors for hospitalization for RSV in healthy and high-risk infants.6-8 This risk declines during the first year of life owing to several factors as infants mature, including greater diameter of the bronchioles as well as maturation of the intrinsic and adaptive immune response.9
Current guidelines put forward by the American Academy of Pediatrics Committee on Infectious Diseases include age thresholds for each of the proposed indications for RSV immunoprophylaxis.10 For infants with CLD, the guidelines suggest prophylaxis with palivizumab up to 24 months of age if they required at least 28 days of supplemental oxygen after birth and continued to require medical treatment (supplemental oxygen or chronic corticosteroid or diuretic therapy) within 6 months before the RSV season. The goal of this study was to determine the age at which the risk of hospitalization for RSV among infants with CLD drops to the level of risk of healthy term infants in their first month of life.
We conducted a retrospective cohort study comparing the age-specific risk of hospitalization for RSV of infants with CLD at 3 through 34 months of age with the risk of hospitalization for RSV of healthy term infants at 1 month of age. Healthy term infants were defined as not having any medical condition that would qualify for immunoprophylaxis according to the American Academy of Pediatrics guidelines.10 We extracted data for the study cohort from the Medicaid Analytic eXtract files for Medicaid beneficiaries for Florida and Texas from January 1, 1999, to December 31, 2010 (12 seasons), which include billing records for inpatient and outpatient encounters including diagnoses and procedures, as well as pharmacy dispensing records. Infant billing records were linked to birth and death certificate data for both Florida and Texas.11,12 This study was approved by the institutional review boards and privacy boards of the University of Florida, the Centers for Medicare & Medicaid Services, and the Florida and Texas Departments of Health with a waiver for informed consent and Health Insurance Portability and Accountability Act of 1996 authorization.
All infants were required to have at least 1 month of Medicaid enrollment during the core RSV season, which was defined as November through February for both states according to infection rates published by the Centers for Disease Control and Prevention.13
We excluded infants with other known or proposed risk factors for hospitalization for RSV, including congenital heart disease, cystic fibrosis, immunodeficiency (acquired or congenital), or Down syndrome diagnosed at any time during follow-up using diagnoses from inpatient and outpatient visits and pharmacy dispensing data.6
The CLD cohort included only infants with a gestational age less than 32 weeks (ascertained from the clinical estimate or, if not available, calculated gestational age on birth certificates) who had an inpatient or outpatient visit with a diagnosis of CLD (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] code, 770.7) at any time of follow-up or with a diagnosis with an ICD-9-CM code of 496 (chronic airway obstruction, not elsewhere classified) between birth and 28 days of age. The healthy term counterparts had a gestational age of more than 36 weeks and 6 days to less than 42 weeks. For the primary analysis, we limited healthy term infants to those with at least 1 sibling aged younger than 5 years, which presents an additional risk factor for RSV that does not qualify for immunoprophylaxis on its own.10 In a sensitivity analysis, we included all healthy term infants regardless of the presence of siblings.
Healthy term infants entered the cohort at their first month of age after at least 1 month of Medicaid enrollment, the beginning of the RSV season, and after a minimum of 30 days in ambulatory care, whichever occurred last. Presence in ambulatory care was required to capture recent exposure to palivizumab because inpatient drug administration is not discernable from capitated hospital charges. Infants with CLD entered the cohort at the beginning of the RSV season after having CLD treatment in the past 6 months based on pharmacy charges (corticosteroids or diuretics) or medical charges (oxygen) or documentation of the need for assisted ventilation on birth certificates and after a minimum of 90 days of age. The higher age threshold for cohort entry was chosen because most infants with CLD had longer hospital stays, and only a small portion were available for analysis during their first weeks of life.
Infants were followed up until the end of the study-defined RSV season, 34 months of age, hospital admission for non-RSV reasons, loss of Medicaid enrollment, or death, whichever occurred first. Infants who were censored because of hospitalization could reenter the cohort after 30 days in ambulatory care. Infants were also allowed to reenter the cohort for a second or third season as long as they met all other inclusion criteria. For computational efficiency, follow-up time was segmented into 15-day blocks.
For each 15-day block, we ascertained hospitalizations for RSV using inpatient claims with primary or secondary diagnoses for RSV-related pneumonia (ICD-9-CM code, 480.1), RSV bronchiolitis (ICD-9-CM code, 466.11), or other RSV infections (ICD-9-CM code, 079.6).
Based on dosing guidelines for palivizumab, we considered infants to have immunoprophylaxis for 30 days following a medical procedure indicating administration of palivizumab or a prescription for palivizumab filled (eAppendix in the Supplement). For palivizumab prescriptions filled, we also required an office visit claim within 10 days to ensure administration.6 For each 15-day block, we concluded exposure to palivizumab prophylaxis if at least 7 days overlapped with the above-defined palivizumab exposure periods. We used the first day of each 15-day block to set infant age and season-month. We established the presence of a sibling younger than 5 years using the date of the last live birth or plurality variable on the birth certificate.
We developed age-trend discrete time logistic regression models within a survival analysis framework14 to evaluate the association of age with risk of hospitalization for RSV between premature infants with CLD and healthy term infants. We included baseline covariates (sex, race/ethnicity, multiple births, state, and RSV season-year) and time-dependent covariates (age, calendar month, siblings <5 years of age, and palivizumab exposure status), as well as interaction terms between state and calendar month and between state and siblings to adjust for confounding. Siblings were adjusted for only in the sensitivity analysis. We used restricted cubic spline terms to model independent age trend curves for risk of hospitalization for RSV during 15-day intervals in preterm infants with CLD and healthy term infants. We generated age risk curves for preterm infants with CLD and healthy term infants using β coefficients from regression models and fixing covariates to mean values across the study population. We equated the CLD age curve risk function from our fitted model with the age curve risk function for healthy term infants evaluated at 1 month of age and solved for the CLD age with the same level of risk.
To estimate 95% CIs for our CLD age estimates, we generated 1000 bootstrap samples of the original data set and refitted the original age trend regression model to obtain a sample of 1000 CLD age estimates.15 We then used the 2.5th and 97.5th percentiles from this sample to define an empirical 95% bootstrap CI. Details of the statistical methods have been published in a previous study.16
In the first sensitivity analysis, we exchanged the primary comparator group of healthy term infants with those with or without siblings. The second set of sensitivity analyses was aimed at estimating the consequence of outcomes misclassification using a misclassified outcomes modeling framework.17 Specifically, we assumed that some hospitalizations for RSV were misclassified as infections without identification of the pathogen. We defined several scenarios in which this presumed false-negative rate for hospitalization for RSV was varied, either nondifferentially (ie, across the entire study cohort) or differentially within strata of interest, assuming lower false-negative rates in premature infants with CLD and/or for younger infants (aged ≤6 months). All analyses were conducted with SAS, version 9.4 (SAS Institute Inc), and graphs were created with R, version 3.1.3 (R Foundation for Statistical Computing).
The study cohorts were assembled from 4.6 million infants in Texas and 2.7 million infants in Florida who were enrolled in Medicaid during January 1, 1999, through December 31, 2010 (Figure 1). Linkage to Vital Statistics birth records allowed retention of 42% of these infants for whom Social Security numbers were available in both data sets. A total of 1 018 593 healthy term infants without an indication for RSV immunoprophylaxis and 5181 infants with CLD and a gestational age less than 32 weeks formed the study cohorts. Because patients could reenter the cohort for a second or third season, the total study cohort consisted of 1 880 531 healthy term infant-seasons and 8680 CLD infant-seasons.
On average, preterm infants with CLD entered the study cohort slightly later because we required at least 90 days of age for cohort entry. As expected, preterm infants with CLD had higher rates of palivizumab use (3704 [42.7%] vs 1309 [0.1%]) and hospitalizations for RSV (230 [2.7%] vs 15 079 [0.8%]) and were more often born as multiplets (1822 [21.0%] vs 51 202 [2.7%]) compared with healthy term infants (Table 1). Quiz Ref IDMost infants with CLD (6433 [74.1%]) were born between 24 and 29 weeks of gestation.
After controlling for risk factors, we observed a pronounced decreasing trend in risk of hospitalization for RSV with increasing infant age. In healthy term infants, the odds of hospitalization for RSV dropped by 82% at 12 months of age (odds ratio [OR], 0.18; 95% CI, 0.17-0.20) and by 95% at 24 months of age (OR, 0.05; 95% CI, 0.04-0.05) compared with 1 month of age. Compared with preterm infants with CLD at 3 months of age, subsequent age groups initially showed a statistically insignificant increase in hospitalization for RSV, followed by a consistent decline to about half the odds at 30 months of age (OR, 0.51; 95% CI, 0.29-0.88). Preterm CLD status significantly increased the risk of hospitalization for RSV at any considered age, with larger ORs at the older age compared with term infants, who had a more rapidly decreasing risk of hospitalization for RSV (Table 2). Male infants were more likely to have hospitalizations for RSV (OR, 1.22; 95% CI, 1.18-1.26) than females.
Figure 2 shows incidence curves derived from our multivariate model averaged across all covariates such as RSV season, sex, multiplets, race/ethnicity, and palivizumab use. Quiz Ref IDAmong term infants, the predicted risk of hospitalization for RSV was 9.0 (95% CI, 8.4-9.6) per 1000 patient season-months at 1 month of age (10.8 per 100 season-years; 95% CI, 10.1-11.5; or 3.6 per 100 patient-years assuming that hospitalizations for RSV occur only during the 4 studied season-months).Quiz Ref ID The risk of hospitalization for RSV among preterm infants with CLD was similar to that of 1-month-old term infants at an age of 18.5 months (95% CI, 15.6-22.8).Quiz Ref ID At 9 months of age, the hospitalization rates of preterm infants with CLD reached a peak, estimated at 15.3 (95% CI, 11.9-19.8) per 1000 patient season-months (18.4 per 100 patient season-years; 95% CI, 14.3-23.8). At 18 months, risk had dropped to 9.0 (95% CI, 7.0-11.7) per 1000 season-months, approximating the risk of 1-month-old term infants.
Among term infants with or without siblings under 5 years of age, the predicted risk of hospitalization for RSV was 6.5 (95% CI, 6.2-6.8) per 1000 patient season-months at 1 month of age. The risk of hospitalization for RSV among preterm infants with CLD was similar to that of 1-month-old term infants at an age of 25.5 months (95% CI, 18.0-35.0). Because the upper 95% CI limit was not precisely attainable during our study follow-up, we assigned 35 months.
Varying assumptions about outcomes misclassification resulted in minor changes in the age threshold (18.2 to 18.5 months of age [eTable in the Supplement]), indicating when the risk of hospitalization for RSV among preterm infants with CLD approximated that of healthy term infants at 1 month of age.
To our knowledge, this is the first study that addresses the age-dependent risk of hospitalization for RSV in infants with CLD in comparison with healthy term infants. It confirms the current American Academy of Pediatrics guideline for infants with CLD, supporting the use of palivizumab during the second year of life.10 We chose to require the presence of siblings for healthy term infants as the principal benchmark because this constellation of risk factors defines the highest risk of hospitalization for RSV that is not considered sufficient to warrant immunoprophylaxis.18 In other words, if a healthy term infant with young siblings at 1 month of age is not recommended for immunoprophylaxis, then an equitable age threshold for palivizumab use in preterm infants with CLD is 18.5 months of age. If the presence of young siblings is not considered (and balanced across cohorts via multivariate adjustment), the age threshold increases to 25.5 months.
Healthy term infants exhibited a rapid decrease in the risk of hospitalization for RSV, with less than half the risk at an age of 6 months when compared with 1-month-old infants. In contrast, the risk of hospitalization for RSV among patients with CLD declined more slowly, with a decrease to about half the risk at the age of 2 years when compared with 3-month-old infants (OR, 0.58; 95% CI 0.38-0.88). Thus, infants with CLD exhibited a more sustained susceptibility to severe RSV infections, but the risk had decreased to a similar level as that of 1-month-old term infants before they had reached their second birthday.
The 2014 American Academy of Pediatrics guidelines restrict recommendations for immunoprophylaxis to infants with CLD who require continuous supplemental oxygen at least 28 days after birth and medical treatment during the 6-month period before the start of RSV season in the second year of life.10 In this study, we were able to ascertain the need for continuous oxygen supplementation at birth, but we found that our data do not provide granular information on duration of oxygen use during hospitalization. As a substitute, we required medical treatment within 6 months prior to the RSV season during both the first and second year of life. This limitation may result in the inclusion of infants with slightly less severe CLD than intended in the current guidelines. However, the fact that infants continued a need for treatment past discharge and more than three-quarters of infants with CLD were very or extremely preterm suggests more severe CLD at birth. It is also possible, but clinically less plausible, that our definition excluded some infants with less severe CLD at the beginning of the season if they required oxygen at birth but stabilized so quickly that medical treatment could be discontinued before 1 year of age.
Incidence rates of hospitalizations for RSV among healthy term infants at 1 month of age—9 per 1000 patient season-months—were higher than those of a laboratory-confirmed report.8 Differences may be inherent in the source population, with our cohort representing children eligible for Medicaid benefits, who may have greater risk of respiratory infection compared with the New Vaccine Surveillance Network cohort, which represents a sample of US counties surrounding Nashville, Rochester, and Cincinnati. We also restricted analysis to the 4 core season months that consistently showed the highest rates of hospitalization for RSV in both Florida and Texas, while the surveillance network considered 6 months. Last, we had to rely on the designation of the infectious pathogen on billing records (via ICD-9-CM coding), which may have attributed some RSV infections to other or unknown pathogens and vice versa.
We found that the overall risk of hospitalizations for RSV decreased over the study period, which is consistent with previous reports of reductions in hospitalizations for bronchiolitis.19,20 To ensure that the observed decrease was not disproportionate between the 2 infant groups, we examined the interaction between term infants vs preterm infants with CLD and risk of RSV in each season and found no statistically significant differences. This finding provides some assurance regarding the extrapolation of our findings to more recent study years. Specifically, if changes (eg, reductions) in risk of hospitalization for RSV were proportional, the reported age threshold would remain accurate. If scientific or clinical advances further attenuated the risk of RSV in infants with CLD, the resulting age thresholds would decrease.
The age threshold provided in this study will facilitate decisions by clinicians and advisory groups regarding use of immunoprophylaxis for infants with CLD. Cost-benefit evaluations for prophylaxis should consider the relatively small risk difference and large numbers needed to treat in the second year of life.21-23 Such differences would be attenuated even further if the risk of hospitalization for RSV continued to decline.
This study has several strengths and limitations. First, we included a large number of infants with CLD, which allowed risk calculation in 2-week increments of increasing age. Second, using external data linkages, we applied restrictions for risk conditions and adjusted for risk factors identifiable in claims data and birth certificates. For instance, we used gestational age from birth certificates considered as the criterion standard in a validation study of ICD-9-CM gestational age codes, which inadequately identified preterm infants.24 Nevertheless, there are environmental risk factors for hospitalizations for RSV that could not be operationalized. One unmeasured and unadjusted risk factor, daycare attendance, deserves some discussion.10 Given the severity of disease, it is conceivable that few infants with CLD attended daycare during follow-up, while their healthy term counterparts would show increasing enrollment in daycare, resulting in higher risk of RSV. However, because our comparison referred to healthy infants at 1 month of age, we assume that daycare attendance played no role in the determination of the presented age threshold.
Finally, we ascertained our study outcome of hospitalizations for RSV from Medicaid billing records, raising concerns about misclassification. Respiratory syncytial virus positivity detected by reverse transcription polymerase chain reaction has been high (87%) among infants with RSV-specific ICD-9-CM codes; however, only 35% of RSV-positive patients had RSV-specific ICD-9-CM codes in 1 validation study.25 To quantify the effect of such misclassification bias on the reported age threshold, we varied assumptions about false-negatives. As in a previous study for moderate preterm infants, the age threshold of 18.5 months was robust even when we assumed that 75% of RSV hospitalizations were missed.16
The age at which the risk of hospitalization for RSV in preterm infants with CLD approximates that of healthy term infants at 1 month of age supports the current guideline, which recommends immunoprophylaxis for RSV up to the second year of life in infants with CLD.
Corresponding Author: Almut G. Winterstein, PhD, Department of Pharmaceutical Outcomes and Policy, College of Pharmacy, University of Florida, 1225 Center Dr, Gainesville, FL 32611 (email@example.com).
Accepted for Publication: August 15, 2017.
Published Online: December 4, 2017. doi:10.1001/jamapediatrics.2017.3792
Author Contributions: Dr Winterstein had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: Winterstein, Choi.
Drafting of the manuscript: Choi, Meissner.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Winterstein, Choi.
Obtained funding: Winterstein.
Administrative, technical, or material support: Winterstein, Choi.
Study supervision: Winterstein, Meissner.
Conflict of Interest Disclosures: None reported.
Funding/Support: This study was funded in part by grant MED152 from the Florida Agency for Healthcare Administration.
Role of the Funder/Sponsor: The funding source had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: This article reflects the views of the authors and should not be construed to represent the views or policies of the Florida Department of Health or the Texas Department of State Health Services.
Additional Contributions: We thank the Public Health Statistics, Office of Vital Statistics, Florida Department of Health and the Texas Department of State Health Services for the provision of birth certificate data.
Create a personal account or sign in to: