Observed Effectiveness of Palivizumab for 29–36-Week Gestation Infants

Health plan claims data were extracted from the computerized databases of 9 Texas Medicaid managed care programs (Texas Children’s Health Plan, Houston; Community Health Choice, Houston; Parkland Community Health Plan, Dallas; Aetna Better Health of Texas, Dallas; Driscoll Health Plan, Corpus Christi; Community First Health Plans, San Antonio; Seton Health Plan, Austin; FirstCare Health Plans, Austin; and Superior Health Plan, Austin). A claim is a request for payment for a health care service. For a health care provider or hospital to be paid for a service, a claim with diagnosis codes supporting the claim must be submitted to the health plan. For a pharmacy to be paid for medication dispensed to a health plan member, a claim must be submitted to the health plan with the National Drug Code (NDC) and amount of the medication dispensed.

Data were extracted for health plan members who were ≤6 months at the start of RSV season (on or after April 1) and born before December 31 of their first year’s RSV season in 2012, 2013, and 2014. One health plan provided data for the eligible infants in the RSV seasons starting in 2013 and 2014 only. Consistent with the epidemiology of RSV in Texas, we generously defined the start of the RSV season in Texas as October 1 and end as March 30.² GA was determined from International Classification of Diseases, Ninth Revision (ICD-9) codes on claims data. If a member had claims with >1 GA code, the gestational code that had the greatest number of claims was assigned. In the case of an equal number of claims with conflicting GAs, the code with the lowest GA was assigned. Members were excluded if they had GA <29 weeks or >36 weeks, diagnosis codes for chronic respiratory disease, hemodynamically significant congenital heart disease, pulmonary hypertension, hematopoietic stem cell or other transplantation, or severe genetic syndrome. Members with a pharmacy claim for medications used for heart failure, chronic lung disease of prematurity, and pulmonary hypertension were excluded (Supplementary Table 6). Members were also excluded if they had <3 months of health plan eligibility during the RSV season in their first year of life.

An RSV hospitalization during the infant’s first RSV season was defined as a hospital claim with an ICD-9 code for RSV infection (079.6, 466.11, and/or 480.1), with date of admission ≥7 days after the date of birth (DOB) and before April 30 of the year after the child’s birth. Rationale for these criteria is to limit outcomes analyses to the time period the child would reasonably be expected to benefit from palivizumab (if administered) and to exclude those infants in whom the RSV infection complicated the initial nursery course. A bronchiolitis without an RSV diagnosis hospitalization was defined as a hospital claim with an ICD-9 code of 466.19 (Bronchiolitis, unspecified), no codes for RSV infection, and date of admission ≥7 days after the DOB and before April 30 of the year after the child’s birth. The number of hospital days per hospitalization was defined as (date of discharge – date of admission).

Palivizumab doses and amount dispensed were determined from outpatient pharmacy claims. Any palivizumab doses administered while the infant was hospitalized were not captured. The number of “eligible” doses was determined as the number of months between either October 1 or the date of the child’s neonatal hospital discharge (whichever came later) and March 1 of the RSV season of interest. Consistent with the 2009 AAP Recommendations,⁴ the maximum number of “eligible” doses was 5. Note that although the AAP 2009 modified recommendations advised no more than 3 doses up until 90 days of age for otherwise healthy infants with a GA of 32 weeks 0 days to 34 weeks 6 days, for the purposes of our analyses we considered 5 doses as the number of potentially eligible doses, as many of the infants of 32 to 36 weeks GA in our sample received >3 doses of palivizumab.

Infants of 29 to 32 weeks’ gestation and 33 to 36 weeks’ gestation were analyzed separately. The χ² test was used to determine statistical significance of differences in bivariate analyses of categorical variables. For analysis of between-group differences in number of hospital days, the Wilcoxon rank-sum test was used. Multivariable analyses were performed by using logistic regression procedures. Age was reduced to categorical data as age group for inclusion in the multivariable logistic regression models, with age group 1 as DOB October 1 to December 31, age group 2 as DOB July 1 to September 30, and age group 3 as DOB April 1 to June 30. Statistical significance was accepted as 2-tailed P < .05.

The study protocol was approved by the Baylor College of Medicine Institutional Review Board (protocol number H-36406) and by the relevant institutional review boards of the participating institutions.

We identified 14 097 eligible prematurely born infants, 631 (4.5%) with GA of 29 to 30 weeks, 1400 (9.9%) with GA of 31 to 32 weeks, 3617 (25.7%) with GA of 33 to 34 weeks, and 8449 (59.9%) with GA of 35 to 36 weeks. Of these infants, 1285 had paid claims for the dispensing of ≥1 doses of palivizumab; 843 (41.5%) of the infants 29 to 32 weeks GA and 442 (3.7%) of the infants of 33 to 36 weeks GA. The infants who received palivizumab were younger than those who did not receive palivizumab, with the mean age for those dispensed ≥1 doses of palivizumab being 17 days younger for the infants of 29 to 32 weeks GA and 35 days younger for the infants of 33 to 36 weeks GA (Table 1).

A total of 588 (4.2%) infants had a paid hospital claim with a diagnosis code for RSV during the RSV season beginning in their first year of life, including 85 (4.2%) of the infants of 29 to 32 weeks GA, and 503 (4.2%) of the infants of 33 to 36 weeks GA(Table 2). For the infants of 33 to 36 weeks’ gestation, RSV hospitalizations were lower for patients born April 1 to September 30 compared with those born October 1 to December 31 (3.1% vs 6.0%, P < .001) (Table 3). Failure to obtain sufficient palivizumab to allow for monthly dosing was substantial; of those infants 29 to 32 weeks' gestation who had any palivizumab dispensed, 320 (38%) had ≤50% of recommended doese dispensed (Table 3).

A total of 319 of the infants (2.3%) had a hospitalization for bronchiolitis without an RSV diagnosis during the RSV season beginning in their first year of life, including 51 (2.5%) of the infants 29 to 32 weeks’ gestation and 268 (2.2%) of the infants 33 to 36 weeks’ gestation (Table 4).

There were fewer RSV hospitalizations among the infants 29 to 32 weeks GA who were dispensed ≥1 doses of palivizumab; 5.0% of those without palivizumab dispensing versus 3.1% of those with ≥1 dispensing (P = .04). Most of this difference was accounted for by those with 80% to 100% of recommended palivizumab doses dispensed (adjusted odds ratio [aOR] 0.30, 95% confidence interval [CI] 0.12–0.78 with no palivizumab dispensed as the reference group). The dose response (by fraction of eligible doses dispensed) was statistically significant (P for trend = .009). For the infants of 33 to 36 weeks GA, the post neonatal RSV hospitalization rate was not different for those with 0 palivizumab dispensings compared with those with ≥1 dispensings (4.5% vs 4.2%, P = .7) (Tables 2 and 3).

Hospitalizations for bronchiolitis without an RSV diagnosis were greater for those infants 29 to 32 weeks GA who had palivizumab dispensed than for those who did not receive palivizumab (3.3% vs 1.9%, P = .05). Most of the difference was accounted for by those who received 80% or more of recommended palivizumab doses (aOR 2.91, 95% CI 1.44–5.92 with no palivizumab dispensed as the reference group). The dose response (by fraction of eligible doses dispensed) was statistically significant (P for trend = .004). There was no significant difference in hospitalization rates for bronchiolitis without an RSV diagnosis by palivizumab dispensing for the infants 33 to 36 weeks GA (2.9% vs 2.2%, P = .3) (Tables 4 and 5). Analyses considering only hospitalizations with a length of stay (LOS) ≥1 day demonstrated similar between-group differences for both hospitalizations with an RSV diagnosis and for bronchiolitis without an RSV diagnosis (Supplemental Tables 7, 8, 9, and 10).

There were a total of 316 RSV hospital days for the 1188 infants of 29 to 32 weeks GA who did not receive palivizumab compared with 138 RSV hospital days for the 843 infants of 29 to 32 weeks GA who did have ≥1 palivizumab doses dispensed (0.27 days/infant vs 0.16 days/infant, P = .04). There were a total of 71 hospital days for bronchiolitis without an RSV diagnosis among the 1188 infants of 29 to 32 weeks GA who did not receive palivizumab and 109 hospital days for bronchiolitis without an RSV diagnosis among the 843 infants of 29 to 32 weeks GA who did receive palivizumab (0.06 days/infant vs 0.13 days/infant, P < .001).

For the 843 members of 29 to 32 weeks’ gestation, 3020.5 mL palivizumab was dispensed. With a retail of $2800/1 mL,⁷ the cost of the palivizumab administered would be $8 457 400 ($10 033/infant started on palivizumab). By using the 2015 Texas Medicaid price of $2537/1 mL, the cost would be $7 663 009 ($9090/infant started on palivizumab).

In pooled data from 9 Medicaid managed care programs in Texas, among infants born at 29 to 32 weeks GA without significant chronic illness, we found that palivizumab administration was associated with both a decreased RSV hospitalization rate and an increased bronchiolitis without RSV diagnosis hospitalization rate. The reduction in number of hospital days for RSV infection associated with palivizumab is offset by the increased number of hospital days for bronchiolitis without an RSV diagnosis. Given the relatively low prevalence of both RSV hospitalization and hospitalization for bronchiolitis without an RSV diagnosis in this lower-risk population, the absolute differences in prevalence are small, even though odds ratios are substantial.

Among infants 33 to 36 weeks’ gestation without significant chronic illness, there were no significant differences associated with palivizumab administration in hospitalization rates for either RSV infection or for bronchiolitis without RSV diagnosis. At 3.6%, palivizumab utilization in this group of older premature infants was low.

Plausible explanations for our findings include potential for false-negative tests for RSV infection and increased rates of other respiratory infections among infants receiving palivizumab. Higher antibody titers (such as from palivizumab administration) may decrease viral shedding and/or neutralize the virus and thus impact sensitivity of tests for RSV.^8,9 The requirement for a physician’s office visit for palivizumab administration may lead to greater exposure to respiratory viruses, such as from sick children in the waiting room with viral contamination of toys and surfaces.¹⁰ It is possible that some of the hospitalizations for bronchiolitis without an RSV diagnosis were patients in whom an RSV diagnostic test was not performed.

There were 2 relevant randomized controlled clinical trials of palivizumab that we identified. The Impact-RSV trial randomized 1502 children ≤35 weeks GA and <6 months of age. The race/ethnicity of the sample was predominantly (58%) white and the mean GA of participants was 29 weeks. The rate of RSV hospitalizations associated with palivizumab administration was substantially lower in the premature infants without bronchopulmonary dysplasia and without congenital heart disease (1.8% vs 8.1%, P < .001). Although non–RSV-positive respiratory hospitalizations were not reported, they did report that there were no differences in upper respiratory tract infection symptoms or cough between placebo and palivizumab groups.¹¹ The relative difference in rates of RSV hospitalization that we observed comparing no palivizumab dispensed to ≥80% of eligible doses dispensed were similar to that from the Impact-RSV trial. However, the absolute difference was smaller due to a lower baseline hospitalization rate, as the lower limit of GA for inclusion in our study was 29 weeks.

A clinical trial in the Netherlands randomized 429 otherwise healthy infants 33 to 35 weeks GA and ≤6 months of age at the start of RSV season.¹² This study also found lower rates of RSV-related hospitalization (0.9% vs 5.1%, P = .01) as well as a decrease in wheezing days ≥2 months after palivizumab prophylaxis ended (1.1% of days vs 3.9% of days, P < .001). We did not observe differences by palivizumab administration status in the infants of 33 to 36 weeks GA. In the Netherlands study, subjects randomized to placebo had slightly greater maternal smoking (17% vs 15%), lower breast-feeding rates (24% vs 29%), and greater day care attendance (53% vs 48%). Although these differences were not statistically significant, it is possible that this imbalance influenced the observed differences as maternal smoking and day care attendance increase RSV hospitalization and wheezing risk, whereas breast-feeding decreases this risk.^13,–16 

Our population differed from those included in these trials. Our population was a Medicaid-insured population. Medicaid is designed for low-income families who would not otherwise be able to afford health insurance for their children.

Our findings of increased rate of hospitalization for bronchiolitis without an RSV diagnosis associated with palivizumab administration has not been reported by previous studies. Krilov et al,¹⁷ in an analysis of Medicaid claims data, found results that appear to conflict with these findings; compared with those who completed their recommended palivizumab doses, those who did not receive all of the recommended palivizumab doses had higher rates of hospitalization for RSV-related respiratory illness but no code for RSV. There are important differences between our study and the Krilov et al¹⁷ study. In contrast to our study, Krilov et al¹⁷ excluded infants who did not receive any palivizumab and included infants with chronic lung disease, hemodynamically significant congenital heart disease, and GA <29 weeks. Krilov et al¹⁷ received research support from MedImmune; 3 of the coauthors served as consultants to MedImmune, and 2 of the coauthors were MedImmune employees, whereas our study received no pharmaceutical industry support.

Our study has a number of important limitations. Our dataset captured palivizumab doses dispensed for outpatient administration only. Palivizumab administered as part of a hospitalization was not captured. Our dataset documented paid pharmacy claims for the dispensing of palivizumab, it did not document palivizumab administration. It is possible that doses of palivizumab were dispensed and paid for, but not administered. We have data only on patients for the time during which they were members of the participating health plans. We did not have access to virology laboratory data; the RSV diagnosis was determined from the hospital discharge diagnoses as submitted to the health plans. Data on hospital admission and discharge was available only as date of admission and date of discharge as submitted on claims for payment to the respective health plans. Time of admission and time of discharge were not available in our dataset. Nonadherence to recommended doses of palivizumab was common in our sample. As this was an observational study, it is possible that the adherent patients were different in important ways from the nonadherent.

Our study was not a randomized controlled clinical trial, and thus lacks the benefit of randomization to adjust for unmeasured differences between treatment groups. However, as a real-world observational study, our results may better reflect outcomes achieved in clinical practice. Individuals who choose to participate in clinical trials may differ in important ways from the general population.

Our findings cannot be generalized to infants <29 weeks’ gestation, those with chronic lung disease of prematurity, those with pulmonary hypertension, and those with hemodynamically significant congenital heart disease, as those infants were specifically excluded from our dataset.

Palivizumab administration is associated with a reduced rate of RSV-positive hospitalization but an increased rate of hospitalization for bronchiolitis without an RSV diagnosis among infants 29 to 32 weeks’ gestation without other significant chronic illness. Among the infants 33 to 36 weeks GA, we found no significant differences associated with palivizumab administration. Our findings support the recommendations of the 2014 AAP Clinical Practice Guideline, which advises against the use of palivizumab for otherwise healthy infants ≥29 weeks 0 days’ gestation. Further research on palivizumab efficacy should examine the effect of palivizumab administration on respiratory morbidity from other than RSV respiratory viral illness.

AAP

American Academy of Pediatrics

aOR

adjusted odds ratio

CI

confidence interval

DOB

date of birth

GA

gestational age

ICD-9

International Classification of Diseases, Ninth Revision

LOS

length of stay

NDC

National Drug Codes

RSV

respiratory syncytial virus

There was no external funding for this study. In-kind support for data extraction was provided by the participating health plans, and we are grateful for the efforts of the information technology specialists at each participating health plan who extracted the claims data from their computerized databases for this study. E. O'Brian Smith, PhD, was consulted on statistical methods. Sheldon Kaplan, MD, reviewed the manuscript and provided helpful suggestions.