Economic Evaluation of Inhaled Nitric Oxide in Preterm Infants Undergoing Mechanical Ventilation
OBJECTIVE: In the previously reported Nitric Oxide for Chronic Lung Disease (NO CLD) trial, ventilated preterm infants who received a course of inhaled nitric oxide (iNO) between 7 and 21 days of life had a significant improvement in survival without bronchopulmonary dysplasia (BPD), as well as a shorter duration of admission and ventilation. However, the price for the drug may be a barrier to widespread use. We sought to estimate the incremental cost-effectiveness of iNO therapy to prevent BPD in infants of <1250 g birth weight.
METHODS: We used patient-level data from the NO CLD randomized trial. The study took a third-party payer perspective and measured costs and effects through hospital discharge. We applied previously reported hospital per-diem costs stratified according to intensity of ventilatory support, nitric oxide costs from standard market prices, and professional (physician) fees from the Medicare fee schedule. We compared log transformed costs by using multivariable modeling and performed incremental cost-effectiveness analysis with estimation of uncertainty through nonparametric bootstrapping.
RESULTS: The mean cost per infant was $193125 in the placebo group and $194702 in the iNO group (adjusted P = .17). The point estimate for the incremental cost per additional survivor without BPD was $21297. For infants in whom iNO was initiated between 7 and 14 days of life, the mean cost per infant was $187407 in the placebo group and $181525 in the iNO group (adjusted P = .46). In this group of early treated infants, there was a 71% probability that iNO actually decreased costs while improving outcomes.
CONCLUSIONS: Despite its higher price relative to many other neonatal therapies, iNO in this trial was not associated with higher costs of care, an effect that is likely due to its impact on length of stay and ventilation. Indeed, for infants who receive nitric oxide between 7 and 14 days of life, the therapy seemed to lower costs while improving outcomes.
In recent years, several clinical studies have explored the efficacy of inhaled nitric oxide (iNO) for prevention of death as well as pulmonary disability in preterm infants.1–5 In the most recent of these reports, the Nitric Oxide for Chronic Lung Disease (NO CLD) study group showed that ventilated infants with birth weight 500 to 1250 g who received a minimum of 24 days of nitric oxide had improved survival without bronchopulmonary dysplasia (BPD), as well as shorter lengths of stay and requirement for supplemental oxygen.1,2
Studies such as the NO CLD study mark a new line of investigation for nitric oxide, with potentially important resource implications. Previous investigations had shown the utility of the drug as rescue therapy for term or near-term infants with persistent pulmonary hypertension.6 In that population, the relatively expensive therapy was highly targeted to a small group of infants for fairly short treatment courses, and was shown to be efficacious and cost-effective.7–9 In contrast, a new indication for use in ventilated preterm infants at high risk of BPD would involve treatment of much larger numbers of patients and for substantially longer courses. It has been unclear to what extent the drug cost for this indication would be offset by improved patient outcomes or decreased resource use. We undertook a retrospective economic evaluation using patient-level data from the NO CLD trial to determine the value for money of treatment of ventilated extremely low birth weight infants with nitric oxide.
Overview of the Clinical Trial
The NO CLD trial enrolled 582 infants from 21 NICUs in the United States between 2000 and 2005.1,2 Infants were 500 to 1250 g, with gestational age at birth of ≤32 weeks, and required either mechanical ventilation or continuous positive airway pressure (CPAP) between 7 and 21 days of age. Infants were randomly assigned to receive either placebo or iNO at weekly decreasing doses, beginning at 20 ppm, for a minimum of 24 days. Gas was continued after extubation by using either nasal CPAP or nasal cannula. The study had 80% power to detect a 12.5% improvement in its primary composite outcome of survival without BPD. The trial noted an improvement in survival without BPD, from 36.5% to 43.9%, in infants treated with iNO (P = .03), as well as a shortening in days of oxygen support (P = .006) and duration of initial hospital admission (P = .04). Treated infants were also significantly more likely to have been discharged or off respiratory support at 40 weeks (P = .01) and 44 weeks (P = .03) postmenstrual age,1 and significantly fewer treated infants required treatment with pulmonary medications in the first year of life.10 In posthoc analyses, there was a significant interaction between treatment and age at study enrollment: survival without BPD at 36 weeks for infants who enrolled in the study between 7 and 14 days of life was 27% in the placebo arm and 49.1% in the iNO arm (P = .0004). Approval for the study was obtained from institutional review boards of participating hospitals and for subsequent analyses from the University of California, San Francisco institutional review board.
Framing of Economic Evaluation
We undertook a retrospective, incremental cost-effectiveness analysis by using patient-level data from the original clinical trial. We used a third-party payer perspective, in which direct medical costs relevant to a payer were considered. Because of data limitations, we did not include costs relevant to parents, such as out-of-pocket expenses or work absences. We considered costs and benefits through the first discharge from the hospital. In light of this time horizon, we did not discount costs or effects. All analyses were undertaken by using an intention to treat approach.
Sources for Resource Tallies, Costs, and Effects
Information on use of medical resources was collected in the case report forms for the clinical trial. These data included days in hospital on mechanical ventilation, CPAP, oxygen, or room air. During the transition from mechanical ventilation to CPAP, an infant needed to remain extubated for at least 3 days to be considered successfully extubated. Subsequent brief periods on ventilation, for example for herniorrhaphy, were not included. High flow nasal cannula was considered to be CPAP. Information on ventilatory support was collected daily for the first 28 days of the trial, then weekly.
We obtained per-diem hospital costs for each level of respiratory acuity (mechanical ventilation, CPAP, oxygen, or no support) from a separate database of all daily charges for a similar group of patients in 1 tertiary-care NICU. Although stratified by level of ventilation, these per-diem averages considered all charges for patient care, including both personnel and nonpersonnel charges, as well as allocated values for hospital overhead expenses. Charges were converted to costs by using cost center-specific federal cost-to-charge ratios.11,12 Such ratios use a standardized, federally mandated methodology to ensure that they represent the true costs of hospital care, and are available for both adult and pediatric hospitals. The resulting daily costs were $1986 for infants on mechanical ventilation, $1694 for infants on CPAP, $765 for infants on nasal cannula, and $656 for infants in room air. We calculated physician fees ($478 for critical infants after the first day of life and $170 for noncritical infants <1500 g) by using resource-based relative value unit for critical care and intensive care, converted to costs by using the Medicare conversion factor for 2005.13 We used the market price for nitric oxide, which was $12000 for a course extending to 30 days at the time the trial was undertaken. All costs are reported in 2006 US dollars.
To maintain consistency with the clinical trial, we used survival without BPD as the effectiveness measure for the economic evaluation.
We compared differences in resource use by using the log-rank test.
Where initial descriptive analyses indicated that cost data were skewed, we performed log-transformation to achieve a normal distribution.14 Because siblings share genetic and prenatal characteristics related to the occurrence of chronic lung disease, their data are considered clustered. To account for such nonindependence of data from multiple gestation infants, we compared total daily costs by using general estimating equations, in which the dependant variable was log-transformed total daily cost, and independent variables included study arm, study site, gestational age, and birth weight.2
We calculated the incremental cost-effectiveness ratio by dividing the difference in mean cost per patient in the placebo and intervention arms by the difference in the mean effect between the study arms. The variance of this incremental cost-effectiveness ratio must consider the cost and effect distributions simultaneously and cannot, therefore, be accurately calculated by using traditional methods. To assess statistical uncertainty in the joint distribution of costs and effects, we used nonparametric bootstrapping, in which we chose 1000 samples of 582 infants, with replacement, from the original study data set.14–16 For each sample, we calculated mean cost, mean effect, and the incremental cost-effectiveness ratio. We determined the proportion of these cost-effectiveness values below thresholds of $0 to $300000 per case of death or BPD avoided, corresponding to the probability that the therapy would be considered appealing to decision-makers with these willingness to pay thresholds.17
Finally, we assessed uncertainty in parameter values by using sensitivity analysis, in which we recalculated the cost-effectiveness after varying the input values for stratified daily hospital cost and physician fees from 50% to 150% of the baseline. We performed similar sensitivity analysis by varying the cost of iNO from $1200 per course to $24000 per course.
Because posthoc analysis of the clinical trial had shown greater efficacy for infants enrolled early, we performed all analyses separately for the full cohort and for infants enrolled on days 7 to 14 of life.
Resource Use Comparison
The study population for the economic evaluation was identical to that for the previously reported clinical trial. As shown in Table 1, there were no baseline differences in gestational age, birth weight, gender, or race for either the entire cohort or for the 7 to 14 day cohort.
Table 2 shows resource use before hospital discharge. Survival curves for the proportion of infants requiring ventilatory support or ongoing hospitalization are presented in Figs 1 and 2, respectively. For the full cohort, as well as for the 7 to 14 day group, infants in the nitric oxide group had significantly fewer days in the hospital. The full cohort also had significantly fewer days on either CPAP or ventilatory support, and were more likely to be discharged alive and in room air.
A disaggregated description of mean costs is provided in Table 3. Nitric-oxide–treated infants in the full cohort showed trends toward lower hospital and physician costs, but differences did not meet statistical significance. Similarly, infants who received nitric oxide beginning between 7 and 14 days of life showed trends toward lower hospital, physician, and total costs, but these were also not statistically significant.
In multivariable analyses adjusting for hospital site, gestational age, birth weight, and age at enrollment, costs were statistically similar between the nitric oxide and placebo groups for both the full cohort (P = .17) and the 7 to 14 day cohorts (P = .46).
Point estimates for the calculation of the incremental cost-effectiveness ratio are shown in Table 4. For the full cohort, the incremental cost was $21 297 for each additional iNO-treated survivor without chronic lung disease. For infants enrolled between 7 and 14 days of age, the point estimate for total costs was lower by $5882 for nitric oxide–treated infants, whereas the proportion of infants surviving without chronic lung disease was substantially better. Because nitric oxide reduces costs while improving outcomes, it is considered “dominant” or cost-saving in this population.
In Fig 3, we have plotted the mean cost and mean effectiveness for each of the 1000 bootstrap replications, reflecting the statistical uncertainty in our joint estimate of cost per survivor without chronic lung disease, for infants enrolled between 7 and 14 days of life. In this figure, the right lower quadrant represents therapies with lower costs and improved effectiveness, whereas the left upper quadrant represents therapies with higher costs and lower effectiveness. Therapies in the right upper quadrant cost more than placebo but are more effective. In our analysis, 71% of the bootstrap replications are in the lower right quadrant, indicating a 71% probability in repeated trials that this therapy would actually reduce costs while improving outcomes. The corresponding probability for the full cohort was 43%.
Because most health policy decision-makers would adopt a therapy even if it did not save money, provided that the cost per outcome were in a reasonable range, we also generated the corresponding cost-effectiveness acceptability curve, shown in Fig 4. This curve plots the probability that an intervention would be considered economically desirable by a decision-maker who was willing to pay between $0 and $300 000 per survivor without chronic lung disease. On this chart, 95% of the replications were either cost saving or were below an expenditure of $67 000 per additional survivor without chronic lung disease for infants who were enrolled between 7 and 14 days of life. This would correspond to a traditional “statistically significant” difference in the proportion of infants meeting this threshold at the P = .05 level.
Because there is only 1 supplier of pharmaceutical grade nitric oxide in the United States, its market price may not accurately reflect its price of production. We, therefore, repeated the above analyses after varying the price of iNO from $1200 per course to $24 000 per course. As shown in Fig 5, the cost-effectiveness depends on the price for a course of treatment. However, the point estimate for the cost per survivor without BPD remains below 0 (ie, it continues to be cost-saving) through a cost of approximately $10 000 for all infants, or approximately $17 000 for infants who were enrolled between 7 and 14 days of life.
Because the unit prices for medical care may vary substantially between locales and institutions, Table 5 presents the results of sensitivity analysis of costs other than the study drug. Higher unit costs for either a day of hospital care or physician fees improved the cost-effectiveness, because iNO shortened admission and time on more expensive ventilatory support. Conversely, if actual unit prices were lower than our base case estimates, cost-effectiveness is worse. In the latter scenario, lowering all non-iNO costs by 50% resulted in a cost of $95 610 per additional survivor without BPD.
In this retrospective economic evaluation, we used patient-level data from a multi-center randomized, controlled trial to document the value-for-money, through hospital discharge, of iNO for prevention of death and BPD. Contrary to previous concerns that the intervention would increase the cost of care, we found that infants treated with iNO had statistically similar costs to those in the placebo group, because the cost of the drug was offset by a decrease in days of ventilation and hospital admission. For infants enrolled between 7 and 14 days of life, who were subject to posthoc analysis in the original clinical trial report, we documented a 95% probability that the intervention cost less than $67 000 per additional survivor without BPD, and a 71% probability that it actually saved money while improving outcomes.
Because most formal economic evaluations alongside neonatal randomized trials have used different measures of effect, it is difficult to make direct comparisons to other therapies. The cost-effectiveness of iNO in our preterm population does seem to compare favorably to that of extracorporeal membrane oxygenation (£13 385 per year of life gained)18; iNO for respiratory failure and persistent pulmonary hypertension in term infants ($33 234 per life gained)9; and universal, rather than selective, hearing screening ($44 000 per case of deafness detected by 6 months).19 Although this may seem surprising given the high price, the daily cost of neonatal intensive care is also quite high, so an intervention that shortens length of stay without incurring other adverse resource consequences is likely to be economically appealing.
It is important to note that our results were sensitive to the price of a course of iNO. Infants who received the drug would have had lower costs and better outcomes with drug costs up to $10 000 per course for all infants and $17 000 for infants who enrolled between 7 and 14 days of life. Unfortunately, the market price of a drug for which there is only 1 supplier is unlikely to reflect its price of production. Until competition drives prices down to the cost of production plus a normal profit, we cannot determine if the true cost-effectiveness is actually better than that estimated here. Conversely, the pricing structure in the United States includes a ceiling on the total charge. If current market prices were continued on an hourly basis for a full 24 days, there is the potential for a substantial decrement in cost-effectiveness.
We undertook this economic evaluation after the results of the randomized, controlled trial were reported. The retrospective nature of the study gives rise to certain limitations that should be considered in interpreting the results. First, because the trial case report forms did not include any information on parental out-of-pocket expenditures or work absence, we were unable to undertake a “societal perspective,” in which all costs to all parties are considered. The narrower, “third-party payer” perspective, which considers costs accruing to the supplier of medical services only, might miss important costs borne by families and other nonmedical participants in a patient's care.
Similarly, because we did not collect extensive resource use information after discharge, our time horizon for the economic evaluation was limited to hospital discharge, in parallel with the primary outcome of the clinical trial. Because 36-week, 40-week, and 44-week outcomes were all better in the iNO group,1 1-year pulmonary outcomes were improved,10 and 24-month neurodevelopmental outcomes were statistically similar,20 it is likely that resource use in the iNO patients after discharge would be less that that for placebo patients.
Our per-diem cost estimates are consistent with daily costs of prematurity derived from total hospitalization costs previously reported in the literature,21 but may seem conservative to readers who are familiar with neonatal charges and reimbursements.22 It is therefore important to note that, because the economic benefits of nitric oxide seem to be because of shorter lengths of stay and ventilation, the use of higher per-diem cost inputs would result in lower (improved) cost-effectiveness estimates. This is borne out in sensitivity analyses in which the total cost savings improved substantially when hospital or physician cost inputs were assumed to be 50% higher than in the base case.
The cost-effectiveness of any therapy depends not only on such resource considerations but on the measured effectiveness. In this regard, our most notable result was in the subsample of patients enrolled between 7 and 14 days of age, for whom the efficacy estimate was derived in posthoc analysis. We recognize the potential for bias inherent in posthoc analyses; however, because adoption of iNO is under consideration on the basis of these data, we believe the associated economic outcomes should still have utility for clinicians and policy maker. The NO CLD trial also used a different treatment protocol compared with other trials of iNO in preterm infants, by starting iNO later and treating for longer with higher total doses. The conclusions of this evaluation cannot be generalized to these other therapeutic regimens.
On the basis of efficacy estimates from the NO CLD randomized trial, iNO for the prevention of death and BPD in preterm infants seems economically sound, particularly for those infants in whom it is started in the second week of life. We recommend that future trials of this or similarly expensive therapies routinely include formal, prospective economic evaluations so decisions regarding adoption consider the optimal use of our constrained resources.
This work was supported by grants from the National Institutes of Health (U01-HL62514, P50-HL56401, P30-HD26979, P30-MRDDRC, and P30-HD26979) and the General Clinical Research Centers Program (M01-RR00240, M01-RR00084, M01-RR00425, M01-RR001271, M01-RR00064, and M01-RR00080).
- Accepted June 12, 2009.
- Address correspondence to John A.F. Zupancic, MD, ScD, Department of Neonatology, Beth Israel Deaconess Medical Center, 330 Brookline Ave, Rose Building Room 318, Boston, MA 02215. E-mail:
Ikaria (formerly INO Therapeutics) provided study gas and masked delivery systems for the primary NO CLD trial. Dr Zupancic received reimbursement for participation in an Ikaria expert advisory panel and internal presentation; Drs R.A. and P.L. Ballard received support from Ikaria to fund completion of 24-month follow-up and data analysis; and Dr Hibbs received reimbursement for travel to investigators meetings as part of these grants. Ikaria did not play any role in the design, analysis, interpretation, or reporting of the study.
What's Known on This Subject:
iNO administered to ventilated preterm infants with birth weights of <1250 g between 7 and 21 days of age improves survival without BPD. The cost-effectiveness of nitric oxide used in this population was not known.
What This Study Adds:
Despite its higher price relative to many other neonatal therapies, iNO does not seem to be associated with higher costs of care, an effect that is likely due to its impact on length of stay and ventilation.
- ↵Finer NN, Barrington KJ. Nitric oxide for respiratory failure in infants born at or near term. Cochrane Database Syst Rev.2006;(2):CD000399
- ↵Angus DC, Clermont G, Watson RS, Linde-Zwirble WT, Clark RH, Roberts MS. Cost-effectiveness of inhaled nitric oxide in the treatment of neonatal respiratory failure in the United States. Pediatrics.2003;112 (6 pt 1):1351– 1360
- ↵Lorch SA, Cnaan A, Barnhart K. Cost-effectiveness of inhaled nitric oxide for the management of persistent pulmonary hypertension of the newborn. Pediatrics.2004;114 (2):417
- ↵Centers for Medicare and Medicaid Services. Hospital cost report, fiscal years 1996 to current. Available at: www.cms.hhs.gov/CostReports/02_HospitalCostReport.asp. Accessed January 10, 2007
- ↵American Academy of Pediatrics. RBRVS: what is it and how does it affect pediatrics? Available at: http://practice.aap.org/content.aspx?aID=1652. Accessed December 20, 2006
- ↵Glick HA, Doshi JA, Sonnad SS, Polsky D. Economic Evaluation in Clinical Trials. New York, NY: Oxford University Press; 2007
- ↵Petrou S, Bischof M, Bennett C, Elbourne D, Field D, McNally H. Cost-effectiveness of neonatal extracorporeal membrane oxygenation based on 7-year results from the United Kingdom Collaborative ECMO Trial. Pediatrics.2006;117 (5):1640– 1649
- ↵Keren R, Helfand M, Homer C, McPhillips H, Lieu TA. Projected cost-effectiveness of statewide universal newborn hearing screening. Pediatrics.2002;110 (5):855– 864
- ↵Walsh M, Hibbs A, Martin R, et al. Neurodevelopmental outcomes at 24 months for extremely low birth weight neonates in the NO CLD trial of inhaled nitric oxide (iNO) to prevent bronchopulmonary dysplasia [abstract]. Pediatric Academic Societies 2008; E-PAS. 2008:4080.8 . Available at: www.pas-meeting.org
- ↵Zupancic JAF. A systematic review of costs associated with preterm birth. In: Behrman R, Stith-Butler A, eds. Preterm Birth: Causes, Consequences and Prevention. Washington, D.C.: National Academies Press; 2006
- Copyright © 2009 by the American Academy of Pediatrics