PEDIATRICS Vol. 121 No. 1 January 2008, pp. 28-36 (doi:10.1542/peds.2007-0633)
ARTICLE |
CoolSim: Using Industrial Modeling Techniques to Examine the Impact of Selective Head Cooling in a Model of Perinatal Regionalization
a Department of Neonatology
b Division of Clinical Computing, Beth Israel Deaconess Medical Center, Boston, Massachusetts
c Division of Newborn Medicine, Harvard Medical School, Boston, Massachusetts
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
OBJECTIVE. A selective head-cooling device for the treatment of moderate to severe hypoxic-ischemic encephalopathy has been approved by the Food and Drug Administration for use in the United States. To reflect the complexity of health care delivery at the systems level, we used the industrial modeling technique of discrete event simulation to analyze the impact of various deployment strategies for selective head cooling and its partner technology, amplitude-integrated electroencephalography.
METHODS. We modeled the course through the perinatal system of all births in Massachusetts over a 1-year period. Cohort and care characteristics were drawn from existing databases. Results of a recently published trial were used to estimate the effects of selective head cooling. One thousand cohort replications were conducted to assess uncertainty. Several policy alternatives were examined, including no use of selective head cooling and scenarios that altered the number and placement of selective head-cooling and amplitude-integrated electroencephalography units throughout the state. Patient-level outcome and cost data were assessed.
RESULTS. For all scenarios tested, the use of amplitude-integrated electroencephalography/selective head cooling resulted in better outcomes at lower cost. However, substantial differences in transfer rates, failure-to-cool rates, and total costs were seen across scenarios. Optimal decision-making regarding the number and placement of devices led to a 16% improvement in cost savings and a 10-fold decrease in failure-to-cool rates, compared with other deployment scenarios. These results were insensitive to significant changes in model inputs.
CONCLUSIONS. On the basis of currently available data, the package of amplitude-integrated electroencephalography and selective head cooling seems to be an economically desirable intervention. Quantifiable techniques to assess system-wide technology performance provide a powerful approach to informing decisions regarding the structure and function of health care systems.
Key Words: infant • newborn • therapeutic hypothermia • hypoxia-ischemia • perinatal regionalization • discrete event simulation
Abbreviations: SHC—selective head cooling • HIE—hypoxic-ischemic encephalopathy • DES—discrete event simulation • aEEG—amplitude-integrated electroencephalography • NDI—neurodevelopmental impairment
In December 2006, the US Food and Drug Administration granted premarket approval of the Olympic Cool-Cap device (Olympic Medical Corporation, Seattle, WA) to provide selective head cooling (SHC) for the treatment of moderate to severe hypoxic-ischemic encephalopathy (HIE) in term infants. This device represents a commercially available version of the cooling device studied by Gluckman et al.1 Despite the action of the Food and Drug Administration regarding the Cool-Cap device, ongoing bedside and research questions exist regarding the appropriate initiation and management of therapeutic hypothermia. Moreover, little attention has been paid to identifying effective deployment strategies for this technology across regional systems of care. System-wide costs and benefits of varying approaches to adoption have not been studied systematically. Unfortunately, health care systems are complex entities, and studying operating or resource policies such as these deployment decisions is challenging.
Other industries, such as manufacturing, airport and port shipping, banking, retail, and hospitality facilities, have successfully applied computer-based, industrial modeling techniques to similar types of decision-making. These approaches have been successful not only in decreasing costs and improving efficiency but also in yielding expanded insight into relevant operational processes and allowing testing of new concepts before implementation. The ability of computer-based modeling techniques to achieve these results in compressed time without the need to disturb existing systems2 holds great promise for health care operations and outcomes research.
This project was designed as a demonstration of the use of industrial modeling techniques to estimate the impact and value for money of several approaches to the deployment of SHC devices. We used the technique known as discrete event simulation (DES) to perform an incremental cost-effectiveness analysis of the use of SHC in a simulated cohort of all newborns born and cared for in Massachusetts during a 1-year period. The use of DES allowed us to consider factors such as institutional capabilities to treat existing comorbidities and complications of therapy, the need for postnatal transport, device expenses, and the narrow time window for instituting treatment.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Approach
We performed an incremental cost-effectiveness analysis of SHC for the treatment of moderate to severe HIE by using the entry criteria, clinical management, and outcome definitions used in the randomized, controlled trial of this treatment reported by Gluckman et al.1 DES,3,4 which is widely used in industrial applications but only recently has been applied to health care, was used to create a high-fidelity simulation of all births at hospitals with delivery services in Massachusetts over a 1-year period. By using publicly available data sources, this simulation created study cohorts that were similar to the Massachusetts birth cohort in terms of timing and location of births, as well as clinical characteristics (eg, gestational age and Apgar scores). It modeled the need not only for cooling and other intensive care services but also interhospital transport for individual infants. Finally, it allowed the availability of SHC within individual NICUs to be modified on the basis of the deployment strategy being examined.
We synthesized information from primary databases and from a systematic literature review to identify the costs, mortality rates, and morbidity rates experienced by a term birth cohort. Estimates of SHC efficacy and adverse event rates were drawn from the report by Gluckman et al.1 Because both the protocol used in that report and the Food and Drug Administration approval of the SHC device required amplitude-integrated electroencephalography (aEEG) results as an entry criterion for SHC, we also examined varying approaches to aEEG device deployment.
Policy Alternatives Examined
The policy alternatives examined in our model differed in the extent to which SHC and aEEG devices were placed in hospitals with differing baseline capabilities for neonatal care (Table 1). Hospitals were categorized as described by the American Academy of Pediatrics.5 Level 1 facilities provide care to asymptomatic term and near-term infants. Level 2 facilities provide intermediate levels of care but not prolonged mechanical ventilation. Level 3 (subspecialty) facilities provide mechanical ventilation and other aspects of neonatal critical care. We designated level 3 regional referral centers that accept postnatal transfers as level 4 facilities (not included in the American Academy of Pediatrics guidelines). To reflect the distribution of neonatal services in Massachusetts, we modeled 23 level 1, 17 level 2, 5 level 3, and 5 level 4 units.
|
In the first scenario ("status quo"), neither aEEG screening nor SHC was available at any hospital. The baseline comparator was thus the current status of care in Massachusetts. In the no-obstacles scenario, there were no restrictions (such as availability of timely transport or cooling equipment) to starting SHC within the recommended 6 hours after birth. In addition, the sensitivity and specificity of aEEG were set at 100%. The no-obstacles scenario therefore likely represents an overestimate of the impact that might be expected of the technology in real-world applications. The remaining alternatives were intermediate between these extremes. In those alternatives, the sensitivity and specificity of aEEG screening were set at 91% and 86%, respectively,6,7 and placement of screening and cooling devices was as described in Table 1.
Simulation Model
A customized model of a statewide perinatal system was implemented by using Arena 10.0 (Rockwell Automation, Milwaukee, WI). The model simulated the experience of infants born at Massachusetts hospitals with delivery, nursery, and transport services, customized according to data on each hospital. Outcome attributes were assigned to infants at the conclusion of their simulated hospital course. Supplemental data analysis was performed with SAS 9.0 (SAS Institute, Cary, NC).
Generation of Birth Cohort
Volume, mode of delivery, and gestational ages of births were estimated for each hospital by using publicly available data.8 Five-minute Apgar scores were assigned to each infant probabilistically, on the basis of gestational age and Massachusetts-specific Apgar score distributions.9 To distribute HIE cases across hospitals, we used Massachusetts birth rates and Apgar score distributions to calculate Apgar score-specific HIE rates that were applied to individual infants. In these calculations, we assumed that the distribution in Massachusetts of moderate to severe HIE across Apgar score categories (ie, 0–3, 4–6, and 7–10) would be similar to that reported by Gluckman et al.1 Our primary analyses were based on a presumed population incidence of moderate to severe HIE of 1 case per 1000 live births.
Nursery Care
Decisions regarding aEEG and SHC services were simulated for individual patients on the basis of the published trial protocol.1 In the status quo scenario, aEEG was never performed. In the no-obstacles scenario, all infants with gestational age of 
36 weeks with moderate to severe HIE underwent aEEG and were referred for cooling. In the remaining scenarios, all such infants and those with Apgar scores of 
3 underwent aEEG, for which test characteristics are listed in Table 2. Infants with abnormal aEEG results were referred for SHC. SHC was modeled as a constrained resource, with the number of SHC devices and their distribution as described in Table 1. Cooling required 3 days to complete, in accordance with current practice. Because data on length of stay were not reported by Gluckman et al,1 we used a pooled estimate from other trials of therapeutic hypothermia (Table 2).
|
Transport
Infants were transported when either aEEG or SHC was indicated and was unavailable at the current location. In the status quo scenario, to account for infants with HIE needing critical care services, all infants with moderate to severe HIE born in level 1 hospitals and 50% of such infants born in level 2 hospitals were transported to level 4 hospitals. Finally, certain infants were transferred for reasons unrelated to HIE, on the basis of hospital- and gestational age-specific rates of transfer.8 Although these infants were not directly involved in comparisons of aEEG/SHC policies, they placed demands on the transport system that affected its availability for HIE transfers.
Information regarding transport destination was drawn from existing referral patterns. Time required for travel between the originating and receiving hospitals was estimated on the basis of data from a geographic information system database.10 Each of 5 receiving hospitals had 1 transport team except in the no-obstacles scenario, in which 4 transport teams were modeled to prevent competition for that resource.
Outcomes
Infants were assigned death, severe neurodevelopmental impairment (NDI), other complications (eg, bradycardia or thrombocytopenia), and length of stay based on their underlying HIE status and the therapy they received earlier in the model. To model the outcomes of infants who had moderate/severe HIE and did not receive aEEG/SHC, we used the outcomes of the infants in the control group of the trial reported by Gluckman et al.1 Similarly, the outcomes of the infants who had HIE and were cooled were based on those of the infants in the intervention arms of that trial. Infants who did not have HIE but were cooled (false-positive aEEG results) received treatment complication rates of the intervention group, and mortality and NDI rates of typical NICU infants. A comprehensive cost tally, including per diem hospital, physician, transport, equipment, and long-term adverse outcome costs, was calculated for each patient.
Model Validation
We validated the model in 3 steps. Face validity was established in presentations to clinicians. Technical validity was verified by entering extreme value inputs for which the output was obvious and predictable. These included, for example, 0 or 1.00 probabilities for HIE, mortality, NDI, or test characteristics and unavailability of any aEEG or SHC equipment. Because the technology has not yet been introduced in large systems of care, predictive validity testing was more limited. However, the incidence of HIE and the HIE-specific rates of death and morbidity did closely match literature estimates.
Data Sources for Model Inputs
Sources and point estimates for economic inputs to the model are shown in Table 2. We used previously calculated daily rates for level 2 or 3 care from summed total daily hospital costs for a population of non–ventilation-treated or ventilation-treated term infants, respectively, in the first 21 days after admission to one tertiary-level NICU.11 These rates were converted to costs by using cost center-specific cost/charge ratios calculated from the Center for Medicare and Medicaid Services Hospital Cost Report. Costs for electrocardiograms and platelet transfusions were calculated in the same manner. Costs for electrocardiograms and platelet transfusions were obtained from the same source. Costs of physician services were derived from the 2006 Medicare resource-based relative value scale fee schedule. The cost range for the postnatal stay for well term newborns was extracted from several estimates in a systematic review of the literature performed for the Institute of Medicine.12
We used the market price for the aEEG monitor and the cooling equipment sold by Olympic Medical. We amortized the costs of capital equipment over a 5-year period at 3% and distributed these fixed costs evenly across the birth cohort.
We obtained transport costs from base and per-mile rates published by the Centers for Medicare and Medicaid Services in the ambulance fee schedule.13 These were specific to Massachusetts special care transports and were keyed to the location of the referral hospital.
We used estimates of long-term costs of cerebral palsy and mental retardation as surrogates for the cost of NDI after HIE.14 In keeping with our societal perspective, literature estimates for long-term costs of NDI included productivity losses attributable to death or inability to work, as well as direct nonmedical costs such as out-of-pocket expenses for caregivers and special education.
All costs were expressed in 2006 US dollars. Where appropriate, currencies were converted between time periods by using the medical care component of the Consumer Price Index and the Hospital Producer Price Index.15,16
Analyses and Statistical Considerations
We analyzed the impact of the alternative policies along several axes, including total costs, cases of NDI averted, cases of death averted, and life-years and quality-adjusted life-years saved. For each comparison of policy alternatives, we calculated the incremental cost-effectiveness ratio, that is, the ratio of the difference in mean costs between scenarios to the difference in mean effects. In addition, we examined the number of infants with HIE not receiving SHC and transport volume among different deployment strategies.
The analyses used a societal perspective, in which all costs and effects were considered regardless of the parties to whom they accrued. A lifespan time horizon captured the long-term costs and effects of NDI. Because individuals have been shown to have a preference for deferring expenditures and for receiving benefits earlier, a discount rate of 3% was applied to both costs and effects.
Sensitivity Analyses
Uncertainty regarding model inputs was assessed by using 2 approaches. For variables that were available only as point estimates, we used deterministic sensitivity analyses, in which input parameters were varied through a plausible range and the effect on outcomes was assessed. The ranges for sensitivity analyses are given in Table 2. To reflect the statistical uncertainty associated with our estimates of efficacy and adverse events, we used stochastic sensitivity analysis.17 The model performed 1000 independent replications representing single, full-year, birth cohorts, each of which resulted in a single estimate of cost and effect.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
Population Characteristics
Within the 1000 birth cohorts, the annual number of deliveries ranged from 75873 to 76953 deliveries per year. The gestational age distribution was as follows: <32 weeks, 1.5%; 32 to 35 weeks, 4.1%;
36 weeks, 94.1%; unknown, 0.3%. The proportions delivered at level 4, 3, 2, and 1 facilities were 17.1%, 28.1%, 34.5%, and 20.1%, respectively. These characteristics are similar to those represented in the Massachusetts Department of Public Health data set.8 The annual incidence rate of moderate to severe HIE among infants with gestational ages of 36 weeks was 1.0 case per 1000 live births (range: 0.6–1.3 case per 1000 live births).
Impact of aEEG and SHC
Table 3 presents economic and noneconomic outcomes of the scenarios tested. Under the no-obstacles assumptions, there was a substantial increase in lives saved as well as NDIs prevented. Moreover, despite substantial screening and treatment costs, total costs actually decreased, mainly because of a decrease in costs associated with NDIs. SHC was thus a dominant option; that is, it improved outcomes while decreasing total expenditures.
|
When the no-obstacles assumptions were removed, the number of HIE-affected infants who were not cooled increased, as did the number of false-positive and false-negative aEEG results. However, we continued to observe a decrease in costs with an increase in lives saved and NDIs averted for all scenarios, compared with the status quo (Table 3). Figure 1 is a scatter plot of changes in costs and effects for one such comparison (aEEG4-SHC4 versus status quo) across 1000 model replications. In this comparison, 686 of the replications are in the lower right quadrant of the cost-effectiveness plane, corresponding to a 68.6% probability that the therapy costs less and is more effective than the status quo. Of note, an additional 18.9% of replications are in the right upper quadrant, below a desirable cost-effectiveness ratio of $50000 per quality-adjusted life-year saved. Similar patterns were observed for all other scenarios, compared with the status quo. In these comparisons, the probabilities of lower costs and improved outcomes ranged between 68% and 73%, and the probabilities of a desirable cost-effectiveness ratio ranged between 88% and 92%.
|
Effects of Device Diffusion on Impact
In other comparisons, we examined the effects of diffusion of the aEEG-SHC technology package from transport (level 4) centers outward to all level 3 or level 2 units (Table 4). As expected, as the diffusion of aEEG systems increased, we observed a decrease in number of transports and associated costs. The cost-effectiveness of moving from placement of aEEG in level 4 centers to placement in level 3 centers (aEEG4-SHC4 to aEEG3-SHC4) was dominant. In contrast, transition from placement in level 3 centers (aEEG3-SHC4) to placement in level 2 centers (aEEG2-SHC4) cost an additional $84211 per quality-adjusted life-year saved.
|
Diffusion of SHC from level 4 units to level 3 units, with aEEG available in level 3 NICUs, resulted in lower costs and improved outcomes. Diffusion of the entire aEEG/SHC package from level 4 NICUs to level 3 NICUs produced a similar result (Table 4).
Effects of Institutional Capacity on Impact
The scenarios described above examined diffusion of the technology from more-centralized to less-centralized deployment. In that process, the overall number of devices in the state increased. Therefore, we tested an increase in institutional capacity (that is, the number of devices in a particular hospital) while holding diffusion constant. Table 4 shows the impact of doubling the number of cooling devices in each hospital in which they were installed. This increase in system capacity was dominant, regardless of whether the devices were placed in level 3 or level 4 units. Notably, increasing statewide capacity by increasing institutional capacity resulted in better economic and health outcomes than did increasing capacity through diffusion to all level 3 hospitals.
Effects of Changes in System Structure on Operational Failures
Our modeling approach also allowed examination of the frequency and causes of instances in which cooling was indicated but was not initiated. Causes of these failures to cool were late arrival (ie, at >6 hours of age) at an institution with the capacity to perform SHC and lack of SHC device availability because of use by other infants. In this model based in a small geographic area, most cases of failure to cool were attributable to unavailability of the SHC device, rather than late arrival (Fig 2). Our models predicted dramatic decreases in rates of failure to cool attributable to device unavailability when 2 devices rather than 1 were placed in each location. For example, adding a second SHC device to level 4 units reduced the rate from 13% to 1.3%. Similarly, adding a second device to both level 3 and level 4 units decreased the rate from 6.7% to 0.3%.
|
Sensitivity Analyses
In deterministic sensitivity analyses, we examined the effects of modifying several aspects of our model unrelated to decisions regarding deployment. We repeated our analyses with higher Apgar score cutoff values for initiation of aEEG screening, thus more than doubling the number of screening tests performed. Next, we lowered aEEG sensitivity and specificity rates to 60%, which affected referrals for cooling. Finally, we evaluated the impact of varying the case incidence rate of HIE between 50% and 150% of the original estimate. None of these modifications affected the dominance of the various implementation scenarios over the status quo. Moreover, this pattern held in the presence of changes in estimated costs of NDI, equipment, and inpatient care of as much as ±50%.
Sensitivity analyses also allowed us to explore factors that might contribute to operational failures. As noted previously, in Massachusetts, where most births occur in close proximity to tertiary NICUs, late arrival played a relatively minor role in contributing to rates of failure to cool. However, late arrival may play a larger role in locales where transport times are greater than those seen in Massachusetts. In our models, increasing transport times by 50% resulted in a fivefold increase in infants arriving late for SHC.
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
This study used industrial modeling techniques to assess the impact of SHC combined with aEEG in the management of HIE within a regionalized system of care. Use of the DES method allowed us not only to examine clinical and economic outcomes in a statewide cohort of newborns but also to gain insight into factors that may influence these outcomes. This method represents a powerful approach to informing decisions regarding the structure and function of health care systems.
We found that the use of SHC leads to better outcomes at lower costs across a wide range of deployment scenarios. Depending on the deployment scenario, this technology may be expected to result in 3.7 to 4.4 million dollars in annual societal cost savings in Massachusetts while reducing the burden of HIE-related NDIs by 19.8% to 23.3%.
To our knowledge, CoolSim is the first application of DES to technology assessment in neonatal care. DES affords great flexibility in modeling complex, time-dependent systems such as the statewide coordination of neonatal care. It has found wide application for understanding industrial processes but has been used infrequently in medical applications. When paired with hospital-specific data on transport rates and infant characteristics, it allowed our analyses to include high-fidelity modeling of every delivery service in Massachusetts. In contrast to other modeling techniques, DES allows time, clinical history, and competition for constrained resources to be considered easily and simultaneously. It thus can simulate geographic aspects of medical infrastructure and interactions among different elements of the system, including patients, devices, personnel, and transport teams.
CoolSim capitalizes on these advantages of DES to test explicitly the effects of different approaches to device deployment. Specifically, we examined changes in costs and effects when the location (diffusion) and number (capacity) of SHC and aEEG units were varied. We found that diffusion of SHC from a small number of transport centers outward to all level 3 centers resulted in an increase in life-years and quality-adjusted life-years while lowering overall costs for the simulated cohorts. Despite these benefits of diffusion of the aEEG/SHC technology package, it is important to realize that selection of deployment on the basis of diffusion characteristics alone would lead to erroneous conclusions. For example, expanding placement of cooling devices from transport centers to all level 3 facilities seems reasonable, whereas doubling the number of units in transport centers is a more economically desirable approach.
Our approach also allowed examination of changes in system performance. For example, although late arrival (ie, >6 hours of age) because of transport was not a common cause of failure to initiate SHC therapy in Massachusetts, doubling transport times showed that this type of system failure would be 5 times more frequent in a locale with greater population dispersion. In contrast, unavailability of cooling equipment was common when only 1 cooling device was available at NICUs performing SHC and was reduced dramatically by increasing the number of cooling devices available statewide.
Collectively, our analyses demonstrate that optimal decision-making regarding the number and placement of aEEG and cooling units within the statewide system can lead to a 10-fold decrease in rates of failure to cool because of late arrival or device unavailability, as well as a 16% improvement in cost savings, compared with other deployment scenarios. The interactions between technological impact and factors such as geographic features, transport, resource placement, and availability are complex. We think that quantitative methods enabled by techniques such as DES allow rigorous examination of these interactions.
Several limitations of this study merit attention. First, our analyses are based on the specific geographic features and care structures present in Massachusetts. The state's small size and exceptionally high density of hospitals lead to specific results that may have limited generalizability to other systems of care. For example, failure to cool an infant because of delay in initiation of therapy is less likely to occur in Massachusetts than in locales with longer transport times. In addition, it should be noted that, in our deployment scenarios, we limited the availability of the aEEG/SHC technology package to units that are capable of providing the wide range of diagnostic, therapeutic, and follow-up services needed by this group of high-risk infants. Second, we used currently available data from a multicenter, randomized, controlled trial for our estimates of clinical outcomes associated with HIE and SHC. Other trials are ongoing and could yield efficacy estimates that differ from those used in our models. As in the initial trial reported by Gluckman et al,1 our analyses focused on the primary outcome of death or severe disability. Therefore, milder forms of disability were not directly considered in our models. However, at least 2 conditions would need to be met for these types of disability to change the central conclusions regarding the cost-effectiveness of the aEEG/SHC technology package. First, an increased rate of mild to moderate disabilities would need to be seen in the treatment group, compared with the control group. Second, this increase would need to be substantial, because the cost implications and effects on life expectancy would be lower than those seen with severe disabilities. Data to support such assumptions do not currently exist. In considering these potential limitations, it is reassuring that extensive sensitivity analyses suggested that even substantial changes to input data would yield incremental cost-effectiveness ratios well within the acceptable range.
Our analyses also did not consider variations in quality of care among the different hospitals. Any intervention that has been shown to be efficacious in a randomized trial will move into real-world practice with the potential for an initial decrease in efficacy early in its adoption. Unfortunately, the magnitude of such an effect is unknown; we refrained from estimating it in the absence of data. It is reassuring that the CoolCap trial efficacy estimates are averaged over multiple institutions, with the median number of infants who were cooled at each site being just 4, with a range of 0 to 11. It is possible that this variability in experience was also associated with site-level differences in the efficacy rates, which would be reflected in the overall average estimate. Once an intervention is incorporated into routine practice, an extensive body of literature demonstrates that variability in outcomes is experienced across institutions, but relevant data with respect to HIE and cooling are currently unavailable. It should be noted that the CoolSim architecture provides flexible interfaces for easy modification of model parameters, thus allowing each of these limitations to be addressed as more information becomes available.
Currently 2 approaches, SHC and whole-body cooling, are available for providing therapeutic hypothermia for the treatment of HIE. Both have been shown in large, randomized, controlled trials to be beneficial in improving 18-month outcomes.1,23,25 Our analyses are centered on the protocol used to provide SHC. Although the 2 approaches differ in several respects, it should be noted that the similarity in both the populations studied and the outcomes seen in these 2 trials suggest that the fundamental conclusions regarding the cost-effectiveness of induced hypothermia would not be substantially different for the 2 approaches, compared with the status quo of not using these therapies.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
We found that, on the basis of currently available data, the aEEG/SHC technology package is likely to be an extremely cost-effective intervention. Furthermore, we demonstrated that industrial modeling techniques provide a powerful approach to understanding regionally distributed systems of care. They are likely to prove useful in policy-making for paradigms such as neonatal care, emergency services, and health services in emerging medical systems, where distribution of limited resources occurs across time and geographic regions. In coordinated systems of care, the highly granular information available from techniques such as DES should enable improved rational decision-making We propose that the ideal distribution of any technology in such systems of care is best determined through quantifiable techniques, rather than through expert judgment or market forces.
|
ACKNOWLEDGMENTS |
|---|
This study was funded by a grant from the Institute for Health Technology Studies.
We thank our expert advisory panel, Drs Don Goldmann, Jeffrey Horbar, Russell Jennings, Jerry Lucey, and Marie McCormick, for their guidance and thoughtful review. We are grateful to Amaris Keiser for assistance with our review of the literature.
|
FOOTNOTES |
|---|
Accepted Jun 12, 2007.
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: jzupanci{at}bidmc.harvard.edu
The authors have indicated they have no financial relationships relevant to this article to disclose.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONCONCLUSIONSREFERENCES |
|---|
- Gluckman PD, Wyatt JS, Azzopardi D, et al. Selective head cooling with mild systemic hypothermia after neonatal encephalopathy: multicentre randomised trial. Lancet. 2005;365 :663 –670[Web of Science][Medline]
- Chung CA. Simulation Modeling Handbook: A Practical Approach. Boca Raton, FL: CRC Press; 2004
- Kelton WD, Sadowski RP, Sturrock DT. Simulation With Arena. 3rd ed. New York, NY: McGraw-Hill; 2004
- Caro JJ. Pharmacoeconomic analyses using discrete event simulation. Pharmacoeconomics. 2005;23 :323 –332[CrossRef][Web of Science][Medline]
- Stark AR. Levels of neonatal care.
Pediatrics. 2004;114
:1341
–1347
[Abstract/Free Full Text] - al Naqeeb N, Edwards AD, Cowan FM, Azzopardi D. Assessment of neonatal encephalopathy by amplitude-integrated electroencephalography.
Pediatrics. 1999;103
:1263
–1271
[Abstract/Free Full Text] - Toet MC, Hellstrom-Westas L, Groenendaal F, Eken P, de Vries LS. Amplitude integrated EEG 3 and 6 hours after birth in full term neonates with hypoxic-ischaemic encephalopathy.
Arch Dis Child Fetal Neonatal Ed. 1999;81
:F19
–F23
[Abstract/Free Full Text] - Massachusetts Department of Public Health. Massachusetts Community Health Information Profile (MassChip). Boston, MA: Massachusetts Department of Public Health; 2006. Available at: http://masschip.state.ma.us/. Accessed July 1, 2006
- Hamilton BE, Martin JA, Ventura SJ, Sutton PD, Menacker F. Births: preliminary data for 2004. Natl Vital Stat Rep. 2005;54 :1 –17. Available at: www.cdc.gov/nchs/data/nvsr/nvsr54/nvsr54_08.pdf.[Medline]
- MapQuest. Available at: www.mapquest.com/directions/. Accessed June 6, 2006
- McBride J, Parad R, Davis J, Allred L, Zupancic J. Cost analysis on the use of recombinant human superoxide dismutase (rhSOD) at birth in preterm infants to improve pulmonary outcome. Pediatr Res. 2005;57 :1232
- 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, DC: National Academies Press; 2006
- Centers for Medicare and Medicaid Services. Ambulance Fee Schedule public use file, 2006. Available at: www.cms.hhs.gov/AmbulanceFeeSchedule. Accessed July 1, 2006
- Centers for Disease Control and Prevention. Economic costs associated with mental retardation, cerebral palsy, hearing loss, and vision impairment: United States, 2003. MMWR Morb Mort Wkly Rep. 2004;53 :57 –59[Medline]
- Bureau of Labor Statistics. Producer Price Index: general medical and surgical hospitals, 2006. Available at: www.bls.gov/ppi/#data. Accessed July 1, 2006
- Bureau of Labor Statistics. Consumer Price Index: medical care, all urban consumers, northeast urban, 2006. Available at: www.bls.gov/cpi. Accessed July 1, 2006
- Doubilet P, Begg CB, Weinstein MC, Braun P, McNeil BJ. Probabilistic sensitivity analysis using Monte Carlo simulation: a practical approach.
Med Decis Making. 1985;5
:157
–177
[Free Full Text] - Hutton JL, Pharoah PO. Life expectancy in severe cerebral palsy.
Arch Dis Child. 2006;91
:254
–258
[Abstract/Free Full Text] - Arias E. United States life tables, 2003. Natl Vital Stat Rep. 2006;54 :1 –40[Medline]
- Furlong W, Feeny D, Torrance GW, et al. Multiplicative Multi-attribute Utility Function for the Health Utilities Index Mark 3 (HUI3) System: A Technical Report. Hamilton, Canada: McMaster University Center for Health Economics and Policy Analysis; 1998. Center for Health Economics and Policy Analysis Working Paper 98–11
- Akisu M, Huseyinov A, Yalaz M, Cetin H, Kultursay N. Selective head cooling with hypothermia suppresses the generation of platelet-activating factor in cerebrospinal fluid of newborn infants with perinatal asphyxia. Prostaglandins Leukot Essent Fatty Acids. 2003;69 :45 –50[CrossRef][Web of Science][Medline]
- Battin MR, Penrice J, Gunn TR, Gunn AJ. Treatment of term infants with head cooling and mild systemic hypothermia (35.0°C and 34.5°C) after perinatal asphyxia.
Pediatrics. 2003;111
:244
–251
[Abstract/Free Full Text] - Shankaran S, Laptook AR, Ehrenkranz RA, et al. Whole-body hypothermia for neonates with hypoxic-ischemic encephalopathy.
N Engl J Med. 2005;353
:1574
–1584
[Abstract/Free Full Text] - Gray JE, McCormick MC, Richardson DK, Ringer S. Normal birth weight intensive care unit survivors: outcome assessment.
Pediatrics. 1996;97
:832
–838
[Abstract/Free Full Text] - Wyatt JS, Gluckman PD, Liu PY, et al., Determinants of outcomes after head cooling for neonatal encephalopathy.
Pediatrics. 2007;119
:912
–921
[Abstract/Free Full Text]
PEDIATRICS (ISSN 1098-4275). ©2008 by the American Academy of Pediatrics
CiteULike 
Connotea 
Del.icio.us 
Digg 
Facebook 
Reddit 
Technorati 
Twitter What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
