Health technology The use of inhaled nitric oxide (ino) in the treatment of neonatal respiratory failure was investigated.

Transcription

1 Cost-effectiveness of inhaled nitric oxide in the treatment of neonatal respiratory failure in the United States Angus D C, Clermont G, Watson R S, Linde-Zwirble W T, Clark R H, Roberts M S Record Status This is a critical abstract of an economic evaluation that meets the criteria for inclusion on NHS EED. Each abstract contains a brief summary of the methods, the results and conclusions followed by a detailed critical assessment on the reliability of the study and the conclusions drawn. Health technology The use of inhaled nitric oxide (ino) in the treatment of neonatal respiratory failure was investigated. Type of intervention Treatment. Economic study type Cost-utility analysis. Study population The study population comprised a hypothetical cohort of term and near-term newborns with hypoxic respiratory failure. Setting The study setting was secondary care. The economic study was carried out in the USA. Dates to which data relate The effectiveness data were derived from studies dating from 1993 to The price year was Source of effectiveness data The effectiveness data were obtained from a review and synthesis of completed studies, with the authors supplementing the results from the review with their own assumptions for the reference case. Modelling The authors constructed a standard decision model for mechanically ventilated newborns with hypoxic respiratory failure. The model compared two identical cohorts, one that received no ino and one that received ino. The authors incorporated decisions about which newborns were transferred to centres providing extra corporeal membrane oxygenation (ECMO), and where ino was started. Under the base-case, the authors assumed that ino was administered to newborns only after transfer to an ECMO centre. Under the "real world" reference case, the authors assumed that ino was administered to newborns at the local hospital and not at the ECMO. This case incorporated two decisions. More specifically, the decision to transfer the infant to an ECMO centre after the initiation of ino and, if the infant was transferred, the decision to start ECMO therapy. The duration of the clinical and economic effects was modelled for one year. The outcomes included alive and healthy, alive with neurologic disability, alive with major respiratory disease, alive with both disabilities, and dead. For both the base-case and reference case, the authors used Monte Carlo simulations to generate 1,000 cohorts. Page: 1 / 7

2 Outcomes assessed in the review In the base-case, the outcomes assessed were the ECMO rate and the death rates in ECMO recipients and those not receiving ECMO. In the reference case, the outcomes assessed were the death rates in ECMO recipients and those not receiving ECMO. There were also long-term outcomes that were relevant to both cases. These outcomes were the proportion of survivors with neurodevelopmental abnormalities, the proportion of survivors with chronic pulmonary impairments, and the average utility for a survivor with any of these abnormalities. Study designs and other criteria for inclusion in the review Transition probabilities for the base-case and the proportion of individuals with any of the two abnormalities mentioned above were derived from pooling the results of two randomised controlled trials (RCTs). The average utilities were derived from a general population cohort. Sources searched to identify primary studies Not reported. Criteria used to ensure the validity of primary studies Not reported. Methods used to judge relevance and validity, and for extracting data Not reported. Number of primary studies included Approximately four primary studies were included in the review. Methods of combining primary studies Transition probabilities for the base-case were derived by pooling the results of the NINOS and CINRGI trials (see Other Publications of Related Interest). This pooling appears to have been performed in the form of a meta-analysis. Investigation of differences between primary studies The authors did not report how differences between the NINOS and CINRGI trials were investigated. However, in both these trials over 90% of newborns were transferred to ECMO centres before the study gas (ino or placebo) was initiated. Results of the review In the base-case: the ECMO rate was 38.3% (+/- 3.1) for those receiving ino and 59.3% (+/- 3.2) for those not receiving ino; in ECMO recipients, the death rate was 17.4% (+/- 4.0) for those receiving ino and 13.9% (+/- 2.9) for those not receiving ino; in those not receiving ECMO, the death rate was 6.8% (+/- 2.1) for those receiving ino and 13.1% (+/- 3.4) for those not receiving ino. In the reference case, the death rates in ECMO and non-ecmo recipients were the same as those in the base-case. The proportion of survivors with neurodevelopmental abnormalities was 35.4% (+/- 5.2) for those receiving ino and Page: 2 / 7

3 29.3% (+/- 4.9) for those not receiving ino. The average utility for these survivors was 0.7. The proportion of survivors with chronic pulmonary impairments was 13.6% (+/- 2.3) for those receiving ino and 20.0% (+/- 3.8) for those not receiving ino. The average utility for these survivors was 0.7. Methods used to derive estimates of effectiveness The authors made assumptions to supplement the results derived from the review of the literature, in order to obtain estimates for the reference case. Estimates of effectiveness and key assumptions The authors made the following assumptions. First, all newborns at local hospitals who had been transferred to an ECMO would receive ino at the hospital. For the no-ino arm, 0% of newborns received ino. Second, when ino is started at local hospitals, rapid clinical improvement would occur in the infants, thus avoiding transfer to ECMO centres. Thus, the ino transfer rate was assumed to be 100% minus the absolute reduction in ECMO by ino (21.0%), that is, 79.0%. It was assumed that the no INO transfer rate was 100%. Third, the death rate for newborns not transferred to the ECMO centre was the same as the ino arm death rate in those not receiving ECMO in the base-case (i.e. 6.8% +/- 2.1). The authors also assumed that for those not receiving ino, there were no deaths at local hospitals as all newborns were transferred. Fourth, the ECMO rate in the ino arm was equal to the base-case ECMO rate (38.3%) divided by the transfer rate (79.4%). Thus, in the reference case, the ECMO rate was assumed to be 48.2% (+/- 4.4). The no-ino arm ECMO rate was the same as that used in the base-case (59.3%) Finally, when infants suffered from both neurodevelopmental abnormalities and chronic pulmonary impairments the effect would be multiplicative. Hence, the utility associated with this health state would be 0.49 (i.e. 0.7 multiplied by 0.7). Measure of benefits used in the economic analysis The measure of benefits used was the quality-adjusted life-years (QALYs) gained. The utilities were derived from the average utility of adult patients living with chronic illness in the Beaver Dam general population cohort. Direct costs The costs and the quantities were reported separately. The direct costs of the health service and those of relatives were included in the analysis. These covered the acute care costs at both ECMO centres and local hospitals, post-discharge costs (e.g. mechanical ventilation) and transportation costs. To generate hospital costs at ECMO centres, the authors multiplied the number days at each different type of hospital ward for patients in the CINRGI trial by estimates of the daily costs. These costs were obtained from a study consisting of 260 term or near-term newborns referred to one of 4 ECMO centres. The authors multiplied all individual charges in these billing records by hospital and departmentspecific Centres for Medicare and Medicaid Services cost-to-charge ratios, then summed all costs incurred each day to generate daily costs. The authors generated separate cost estimates for the first day on ECMO and the first day on mechanical ventilation, as the initial day of invasive therapy was more expensive than subsequent days. In the reference case, for those infants not transferred to ECMO centres, the authors assumed that their hospital costs were a fraction of the costs that would have been incurred at an ECMO centre. This fraction was estimated at 0.8, following analysis of discharge abstracts of newborns in 1994 in California. The cost of ino therapy was derived from current pricing in the US market. The duration of therapy was based on patient-level observations from the CINRGI trial. Discounting was not relevant since all the costs were incurred during one year and, appropriately, it was not performed. The study reported the average costs. The price year was Page: 3 / 7

4 Statistical analysis of costs The total costs were treated stochastically. The 95% confidence ellipses around point estimates were generated using the method described by Mullahy and Manning (see Other Publications of Related Interest). Indirect Costs The indirect costs included were those due to lost wages of the infants' relatives. The sum of lost wages was calculated as the average daily wage multiplied by the total hospital length of stay, plus half-day wage multiplied by the number of outpatient visits (as obtained from Pennsylvania Medicaid claims data). The price year was Discounting was not performed because the costs were incurred during less than one year. Currency US dollars ($). Sensitivity analysis Both one-way and two-way sensitivity analyses were conducted on the reference case. The authors varied the assumptions about clinical effects, health care costs and the population treated. For the population treated, the authors explored the inclusion of two additional types of patients. More specifically, those patients not meeting the entry criteria for the clinical trials, and patients who would have met the entry criteria but never had access to ECMO centres. Estimated benefits used in the economic analysis In the base-case (where ino is initiated at ECMO centres), the number of QALYs gained per patient was 0.77 (95% confidence interval, CI: ) in the ino arm and 0.74 (95% CI: ) in the no-ino arm. This yielded an additional QALYs in the first year of life (95% CI: QALY lost to QALY gained) per treated infant. In the reference case (where ino is initiated at local hospitals), the number of QALYs gained per patient was 0.77 (95% CI: ) in the ino arm and 0.74 (95% CI: ) in the no-ino arm. This yielded an additional 0.03 QALYs (95% CI: QALYs lost to QALYs gained) at one year per treated infant. Cost results In the base-case, the mean cost per patient in the first year was $73,200 (95% CI: 69,140-77,390) in the ino arm and $75,080 (95% CI: 70,950-79,240) in the no-ino arm. This generated cost-savings of $1,880 (95% CI: $7,420 cheaper to $3,550 more expensive). In the reference case, the mean cost per patient in the first year was $75,800 (95% CI: 70,100-80,890) in the ino arm and $80,200 (95% CI: 76,200-85,000) in the no-ino arm. This generated cost-savings of $4,400 (95% CI: $11,040 cheaper to $2,410 more expensive). Synthesis of costs and benefits The costs and benefits were combined by calculating a cost-utility ratio (the additional cost required per QALY gained). Compared with no-ino therapy, ino therapy was found to be dominant (i.e. reduced health care costs and improved quality-adjusted survival) in the base-case, although there was considerable uncertainty around these means, with only 60.3% probability that ino was dominant. However, there was 84.6% probability that the incremental cost-utility ratio of using ino as opposed to not was lower than $100,000 per QALY, and only 6.3% probability that ino was dominated (i.e. higher costs and worse outcomes than a strategy of no INO). In the reference case, initiating ino at local hospitals also reduced the health care costs and improved quality-adjusted survival. Again, uncertainties around the means were considerable, with a 71.6% probability of ino being dominant, a 91.6% probability of ino having an incremental cost-utility ratio lower than $100,000 per QALY, and only 2.6% Page: 4 / 7

5 probability of it being dominated. In the reference case, the results were insensitive to the incidence of complications and their impact on quality of life. They were also insensitive to assumptions about differences in cost between local and ECMO hospitals and the impact on ECMO mortality. The results were moderately sensitive to the survival benefit of ino in newborns who did not receive ECMO, and also to including patients who would not benefit from ino. The results were sensitive to the price of the drug and hospital length of stay. Authors' conclusions Despite its high acquisition costs, inhaled nitric oxide (ino) was found to have a favourable cost-effectiveness profile for the treatment of hypoxic respiratory failure in term or near-term newborns. Under the best circumstances ino both reduced the costs and improved outcomes, independent of whether it was initiated at a local hospital before transfer or on arrival at an ECMO centre. CRD COMMENTARY - Selection of comparators No explicit justification was given for using a no-ino strategy as the comparator. The use of this comparator is justified as it was not until the end of 1999 that the US Food and Drug Administration approved ino as a treatment for severe hypoxic respiratory failure in newborns. You should decide if the comparator represents current practice in your own setting. Validity of estimate of measure of effectiveness The authors did not report the sources searched to identify the primary studies. They also failed to report whether a systematic review of the literature had been undertaken to identify relevant research and to minimise biases. However, transition probabilities for the base-case were derived from two RCTs, whose results were combined using a metaanalysis of the measures of effectiveness on 483 patients. Further, the authors reported measures of variance alongside the mean values. Despite the fact that the authors did not report how differences between the two RCTs were investigated, it would appear that both of these studies were very similar and investigated the same patient population. For the reference case, the authors supplemented the results of the review with their own assumptions. These assumptions were clearly detailed and would appear to be robust. In addition, the estimates from the review o the literature and the authors' assumptions were investigated in the sensitivity analysis. Validity of estimate of measure of benefit The estimation of benefits was modelled. The authors obtained utility values from a general population cohort. However, this study was based on utility levels of adults and it is unclear if these values can be generalised to an infant population. Validity of estimate of costs All the categories of cost relevant to the perspective adopted were included in the analysis. However, some relevant costs such as mortality costs (i.e. foregone productivity due to the infant dying) were not included in the analysis. Other costs omitted by the authors were the costs to parents associated with the care of a sick newborn (although the indirect costs of lost wages were included) and the ongoing health care costs of managing specific pulmonary or neurologic morbidities. It would appear that these omissions are likely to bias in favour of the no-ino strategy. The costs and the quantities were reported separately, which will enhance the generalisability of the authors' results to other settings. The results were derived from published studies. Sensitivity analyses on the average length of stay were conducted, using what appear to have been appropriate ranges. The unit costs were also derived from published sources and were varied in the sensitivity analysis. Medicare and Medicaid charges were used to proxy prices in some instances. Discounting was unnecessary since all the costs were incurred during one year and, therefore, was not performed. The date to which the prices related was reported, which will ease any future reflation exercises. Page: 5 / 7

6 Other issues The authors did not make appropriate comparisons of their results with the findings of similar studies. The issue of generalisability to other settings was addressed in the sensitivity analysis, although the authors warned readers that as ECMO use and cost vary substantially between the USA and other countries, the relevance of their findings to other countries remains unclear. The authors do not appear to have presented their results selectively and their conclusions reflected the scope of the analysis. The authors reported a number of further limitations to their study. First, the relatively small sample sizes of the two RCTs led to considerable uncertainty around the point estimates. However, results of the Monte Carlo simulations demonstrated that ino appeared to be favourable. Second, the time horizon was restricted to the first year of life. This conservatively assumed that all the costs and effects of ino would disappear at one year. Further research is needed to determine if ino differentially affects sequelae of severe neonatal illness. Third, quality of life was derived from an adult population since it is difficult to evaluate in infants. However, the authors' conclusions were robust to varying quality of life attributed to infants. Finally, because data on the decision-making process that leads to transfers of a newborn to an ECMO facility are lacking, the authors assumed that experienced clinicians at local hospitals would assess a newborn's responsiveness to ino and initiate transfer to an ECMO centre appropriately. Implications of the study The authors reported that ino can be disseminated into the community and used in a reasonably large proportion of all ventilated, term infants with confidence that it will be cost-effective. However, the authors warned that the costeffectiveness will degenerate if use is too indiscriminate. The authors also recommended subsequent study of the longterm effects of perinatal ino to understand fully the societal consequences of this therapy. Source of funding Supported in part by a grant from INO Therapeutics Inc., Clinton (NJ), USA. Bibliographic details Angus D C, Clermont G, Watson R S, Linde-Zwirble W T, Clark R H, Roberts M S. Cost-effectiveness of inhaled nitric oxide in the treatment of neonatal respiratory failure in the United States. Pediatrics 2003; 112(6): PubMedID Other publications of related interest Neonatal Inhaled Nitric Oxide Study Group. Inhaled nitric oxide in full term and nearly full term infants with hypoxic respiratory failure. New England Journal of Medicine 1997;336: Clark RH, Kueser TJ, Walker MW, et al. Low-dose nitric oxide therapy for persistent pulmonary hypertension of the newborn. New England Journal of Medicine 2000;342: Javitt J, Dei CR, Chiang YP. Cost-effectiveness of screening and cryotherapy for threshold retinopathy of prematurity. Pediatrics 1993;91: Mullahy J, Manning WG. Statistical issues in cost-effectiveness analyses. In: Sloan F, editor. Valuing health care: costs, benefits, and effectiveness of pharmaceuticals and other medical technologies. New York (NY): Cambridge University Press; p Indexing Status Page: 6 / 7

7 Powered by TCPDF ( Subject indexing assigned by NLM MeSH Anoxia /economics /therapy; Cost-Benefit Analysis; Decision Trees; Extracorporeal Membrane Oxygenation /economics; Hospital Costs; Hospitals, General /economics; Hospitals, Special /economics; Humans; Infant, Newborn; Length of Stay; Monte Carlo Method; Morbidity; Nitric Oxide /administration & dosage /economics /therapeutic use; Patient Transfer /economics; Probability; Quality-Adjusted Life Years; Respiratory Insufficiency /economics /therapy; Survival Analysis; Treatment Outcome; United States; Vasodilator Agents /administration & dosage /economics /therapeutic use AccessionNumber Date bibliographic record published 30/09/2004 Date abstract record published 30/09/2004 Page: 7 / 7