 |
|
 |
|
 |
 |
|
 |
 |
 |
 |
 |
 |
 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
* Department of Pediatrics, Medical University of South Carolina Children’s Hospital, Charleston, South Carolina
Center for Health Policy and Research, University of Massachusetts Medical School, Worcester, Massachusetts
 |
ABSTRACT |
|---|
 TOP ABSTRACT METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods. We performed an incremental cost-effectiveness analysis of the 3 strategies compared with current practice. We used a decision analytic model and a spreadsheet to estimate the direct costs and outcomes, including the savings resulting from prevented kernicterus, for an annual cohort of 2 800 000 healthy term newborns who are eligible for early discharge. We used a modified societal perspective and 2002 US dollars. With each strategy, the test and treatment thresholds for hyperbilirubinemia are lowered compared with current practice.
Results. With the base-case assumptions (current incidence of kernicterus 1:100 000 and a relative risk reduction [RRR] of 0.7 with each strategy), the cost to prevent 1 case of kernicterus was $10 321 463, $5 743 905, and $9 191 352 respectively for strategies 1, 2, and 3 listed above. The total annual incremental costs for the cohort were, respectively, $202 300 671, $112 580 535, and $180 150 494. Sensitivity analyses showed that the cost per case is highly dependent on the population incidence of kernicterus and the RRR with each strategy, both of which are currently unknown. In our model, annual cost savings of $46 179 465 for the cohort would result with strategy 2, if the incidence of kernicterus is high (1:10 000 births or higher) and the RRR is high (
0.7). If the incidence is lower or the RRR is lower, then the cost per case prevented ranged from $4 145 676 to as high as $77 650 240.
Conclusions. Widespread implementation of these strategies is likely to increase health care costs significantly with uncertain benefits. It is premature to implement routine predischarge serum or transcutaneous bilirubin screening on a large scale. However, universal follow-up may have benefits beyond kernicterus prevention, which we did not include in our model. Research is required to determine the epidemiology, risk factors, and causes of kernicterus; to evaluate the effectiveness of strategies intended to prevent kernicterus; and to determine the cost per quality-adjusted life year with any proposed preventive strategy.
Key Words: kernicterus • cost-effectiveness • newborn • infant • jaundice • cost • screening • bilirubin
Abbreviations: CDC, Centers for Disease Control and Prevention • AAP, American Academy of Pediatrics • RRR, relative risk reduction
Recently, there has been concern about a resurgence of kernicterus in the United States,1 with case reports of extremely high serum bilirubin levels occurring in term or near-term infants. Although there are no epidemiologic studies of kernicterus and the population incidence of kernicterus is unknown,2,3 concern that its incidence might be rising has led to a sentinel event alert issued by the Joint Commission on Accreditation of Healthcare Organizations,4 a report of 4 cases of kernicterus by the Centers for Disease Control and Prevention (CDC) in its Morbidity and Mortality Weekly Report,5 and a statement by the American Academy of Pediatrics (AAP)2 intended to bring the issue of kernicterus to the attention of the pediatric community. In the sentinel event alert, the Morbidity and Mortality Weekly Report and the AAP statement, several potential causes and risk factors for the occurrence of kernicterus were discussed. These documents also listed several potential risk reduction strategies to ensure the early detection and timely treatment of hyperbilirubinemia and thereby decrease the risk of kernicterus. Some of the listed strategies include adherence to the AAP Practice Guidelines for Management of Hyperbilirubinemia in the Healthy Term Newborn,6 either universal follow-up by a physician or a pediatric nurse of all newborns within 24 to 48 hours of hospital discharge or discharge and follow-up strategies based on risk assessment, and plotting of predischarge serum or transcutaneous bilirubin values on a percentile-based nomogram to identify and follow up infants who are identified to be at risk for severe hyperbilirubinemia. A recent AAP guideline about the management of hyperbilirubinemia in the newborn infant who is 35 weeks of gestation7 has also endorsed these strategies.
Currently, health care providers and health systems that are responsible for the care of newborn infants before and immediately after hospital discharge are under pressure to implement strategies such as the ones listed above. However, although these strategies seem logical and make sense intuitively, their effectiveness in reducing kernicterus is unproved. Several of these strategies will increase health care costs as a result of increased bilirubin testing, a greater number of office or home nurse visits, and an increased number of infants treated (a consequence of a lowered treatment threshold). Three of the strategies listed above are particularly likely to involve increased spending on health care resources—universal follow-up after hospital discharge, routine predischarge serum bilirubin testing with selective follow-up, and routine predischarge transcutaneous bilirubin testing with selective follow-up. An assessment of the benefits, risks, and costs of strategies to prevent kernicterus therefore is necessary before implementing nationwide policy changes. We undertook an incremental cost-effectiveness analysis of these 3 kernicterus prevention strategies compared with the current pattern of practice in the care of newborn infants.
|
METHODS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Structure of the Model
We constructed a decision analytic model to evaluate 3 different strategies that are intended to prevent kernicterus and compared them with the current practice of prevention and management of jaundice in newborn infants, with each strategy to be applied before the standard office visit at 2 weeks of age. We used DATA 3.5 (TreeAge Software, Inc, Williamstown, MA) to structure the decision tree and estimate the total costs of each strategy. One branch of this tree is depicted in Fig 1. The branch structures for the other strategies modeled are identical to the depicted branch; however, the transitional probabilities for each strategy are different and shown in Table 1. We used a Microsoft Excel spreadsheet to calculate the cost per case of kernicterus prevented.
|
|
Under each strategy, within 2 days of early discharge, each infant would receive a follow-up office visit, a home visit by a nurse, or no follow-up (either because of parental noncompliance or because the first visit is intentionally scheduled at 2 weeks of life, with telephone support in the interim). The early office visit or home nurse visit is in addition to the standard 2-week visit that is recommended by the AAP. The home visit by the nurse would, in some cases, result in a referral to the physician’s office for evaluation by the primary care physician before the 2-week visit.
At the postdischarge office visit, the history and physical examination, including visual assessment of skin color, is used to determine the need for serum bilirubin testing and other laboratory tests such as a hemogram, blood type, Coombs test, and reticulocyte count. In addition, in all 3 strategies, the threshold for laboratory testing is lowered, because authorities have emphasized the unreliability of visual estimation of bilirubin levels and the need for a low threshold for measuring the serum bilirubin.1,2,4,5,7 In addition, the knowledge of the infant’s predischarge bilirubin percentile value (derived from the serum or transcutaneous bilirubin value) induces a lower threshold for laboratory testing and additional follow-up. The serum bilirubin value will guide decisions about the need for reassessment by the health care provider in the subsequent day or 2, repeat serum bilirubin testing to track a borderline bilirubin value (this is included under "laboratory testing" in the model), and the need for treatment with phototherapy. The availability of resources, comorbid conditions in the infant (eg, dehydration), and local practice patterns would guide decisions about whether phototherapy was provided at home or in the hospital. Because the frequency of exchange transfusion is low, we did not include it in the model.
Probabilities
The baseline estimates for probabilities used in the model are listed in Table 1. For strategy 1, current practice, we based the probabilities listed in Table 1 of an infant’s being seen in the office or during a home visit on publications by Galbraith,10 who reported, using population-based data, that 32% of early-discharged infants were seen by a health care provider within 2 days of discharge, and on other publications from single institutions that reported that one third to two thirds of early-discharged newborns do not receive the AAP-recommended follow-up visits.11,12 Sixty percent of healthy newborns are reported to develop jaundice,6 and on a recent survey,13 pediatricians reported checking serum bilirubin levels in 55% of jaundiced newborns who are seen in the office. Therefore, we used 0.33 (0.6 * 0.55) for the probability of laboratory testing during an office visit. This estimate is consistent with a report that 17% to 52% of infants in a large health maintenance organization had at least 1 bilirubin level checked.14 Meara et al15 reported that 0.47% of all infants discharged early were rehospitalized for jaundice. Madden et al16 reported that 1.8% to 2.4% of all such infants were treated with hospital or home phototherapy. Therefore, we adjusted the probabilities of laboratory testing and receiving phototherapy treatment in our model so that 
2% of the cohort infants under the "current practice" strategy received phototherapy. We obtained the remaining probabilities from interviews with pediatricians who represented 4 group practices in Burlington, VT, about their practice patterns.
We estimated some of the probabilities for strategies 3 and 4 from the original articles describing the hour-specific bilirubin nomogram8 and the correlation between the transcutaneous bilirubinometry readings and serum bilirubin.9 The remaining probabilities are estimates on the basis of the first author’s knowledge about neonatal jaundice and its treatment and on discussions with pediatric colleagues about likely pediatric practice changes in response to promulgated prevention strategies. We assumed that in all 3 prevention strategies, the threshold for treatment would be lowered, because policies that promote aggressive screening for jaundice and for risk of jaundice in combination with published alerts from authorities about a resurgence of kernicterus are likely to cause many health care providers to treat earlier in the course of disease and at lower levels of bilirubin than with current practice. The percentage of all cohort infants who received phototherapy in the 4 pathways is as follows: current practice, 2.3%; universal follow-up, 8.1%; predischarge serum bilirubin, 5.6%; and predischarge transcutaneous bilirubin, 7.8%.
Costs
The baseline estimates for costs used in the model are listed in Table 2. The costs are obtained by summing the individual costs of the branches for each unique pathway in the decision analysis model. We estimated costs from a modified societal perspective, for the total number of liveborn infants per year in the United States who would be eligible for the 3 preventive strategies. We estimated all costs in 2002 dollars. We obtained provider charges for laboratory tests for bilirubin, hemogram, blood typing, and Coombs test from the central laboratory at Fletcher Allen Health Care hospital (Burlington, VT). To this we added the costs for supplies for blood sampling ($1.00 per infant) and 20 minutes of nursing time (at an hourly rate of $24). We obtained the costs of transcutaneous bilirubinometry from the company (Respironics, Murrysville, PA) that markets Bilichek, the instrument tested by Bhutani et al.9 The cost of the bilirubinometer was $3995. We assumed that 2 such instruments would be used for transcutaneous bilirubin tests in 8000 infants before becoming nonfunctional or outdated by a newer model. The cost of the disposables (Lensette tips) for each infant tested was $6.80. Therefore, the total cost for each infant tested was $7.80. We obtained the charges for an office visit ($50 per visit) from the charges for pediatric office visits in Burlington, VT, by talking to local pediatricians. We obtained the charges for home visits by skilled nurses from the Visiting Nurses Association of Vermont ($95 per visit).
|
When we obtained charges instead of direct costs, we derived the costs by applying a cost-to-charge ratio of 0.58, the Medicare statewide average operating cost-to-charge ratio for an urban hospital in Vermont.17 For the home nurse visit, the home phototherapy, and the transcutaneous measurement of bilirubin, we assumed that the charge and the cost were identical. All costs are expressed in 2002 dollars. We included only direct costs and did not include indirect costs such as work-loss costs for the parents.
We based our estimate of the savings resulting from the prevention of kernicterus on published data from the CDC on the average lifetime direct and indirect costs per person of cerebral palsy ($921 000 in 2003 dollars, $900 738 in 2002 dollars) and mental retardation ($1 014 000 in 2003 dollars and $991 692 in 2002 dollars) discounted at 3%.18 For purposes of comparison, these more recent estimates are somewhat different from lifetime costs for cerebral palsy ranging from $644 846 (at a 5% discount rate in 2002 dollars) to $1 533 272 (at a 2% discount rate in 2002 dollars) reported by Waitzman et al19 in 1992. We used $900 000 for the lifetime cost for a child with kernicterus, assuming that the lifetime cost for a child with kernicterus would be similar to that of a child with cerebral palsy or mental retardation.
Outcomes
As our primary outcome, we estimated the cost to prevent 1 case of kernicterus, by implementing each one of strategies 2, 3, or 4. We assumed that the effectiveness of each strategy in preventing kernicterus was similar. In the absence of data to determine the relative risk reduction (RRR) with each strategy, we assumed an RRR of 0.7 (ie, with each strategy, 70% of the cases of kernicterus occurring with current practice would be prevented). We estimated the incremental cost for the entire cohort with each of these strategies by subtracting, from the cost of a given strategy, the cost of current management as well as the savings resulting from kernicterus cases prevented ($900 000 multiplied by the number of kernicterus cases prevented). We divided this incremental cost by the estimated number of cases of kernicterus prevented per year to obtain the cost per case prevented.
Sensitivity Analysis
The true population incidence of kernicterus is unknown.2,3 Therefore, in modeling the cost per case of kernicterus prevented, we performed 1-way sensitivity analyses by varying the incidence of kernicterus from 1:10 000 healthy term live births to 1:500 000 such births and by varying the RRR from 1.0 to 0.1.
|
RESULTS |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
With each of these strategies, we modeled the pathways of decision making and probabilities that are depicted in Fig 1. These are only some of the pathways possible, and several other patterns of testing, treating, and following up are possible. For example, pediatricians and nurses could use transcutaneous bilirubinometers to assess infants in their offices or at home in addition to the in-hospital use of this technique. In some areas, the nurse performing the home visit could, if necessary, draw a serum bilirubin sample. A clinical risk assessment using demographic factors, features in the maternal and perinatal history, and physical examination such as described by Newman et al7,20 could be performed as a substitute for or in addition to 1 of the strategies modeled by us. One appealing strategy that is likely not to lead to increased costs is the use of the mandatory early follow-up visit as a replacement for the 2-week visit instead of being an additional visit. However, the impact of this strategy on kernicterus as well as on other outcomes such as breastfeeding success, hypernatremic dehydration, and other early neonatal problems has to be assessed before implementing it. The likely compliance of parents with this strategy also has to be assessed.
Our study has several limitations. Although we used published data as far as possible to obtain the transitional probabilities for current practice and the 3 strategies modeled, some of the probabilities used in the model were derived from interviews with pediatricians in Burlington, VT, and pediatric practice patterns probably vary widely across the United States. The change in practice pattern in response to any strategy suggested by the AAP is also likely to vary widely across the country. More rigorous, nationwide data on the likely patterns of surveillance and management of hyperbilirubinemia by pediatricians and family practitioners in response to the AAP guidelines are required to strengthen the reliability of our model. The cost per case of kernicterus prevented reported in our article may be an underestimation because of 2 reasons. First, we did not include indirect costs in our estimation of costs (except in the estimate of lifetime costs of kernicterus). Second, some of our cost estimates may be lower than actual nationwide costs. A more rigorous determination of costs is required to strengthen the reliability of the costs component of the model. Finally, we did not model other benefits that might result from follow-up office or home nurse visits, such as improved lactation and decrease in dehydration. Quantifying these benefits and including them in the model would alter the cost-effectiveness reported in our study.
A key problem with any attempt to determine the cost-effectiveness of strategies to prevent kernicterus is the unknown population incidence of kernicterus.2,3 In data from a large health maintenance organization published by Newman et al,14 1 infant in 10 000 had a serum bilirubin of 30 mg/dL, but none of the 11 infants identified in the study had kernicterus on follow-up.21 In data from the kernicterus registry published by Johnson,1 90 cases were collected over 16 years. If we assume that only ine quarter of all kernicterus cases were reported to this registry, with 4 million births per year in the United States, this yields an estimated incidence of 1 case of kernicterus per 178 000 live births. In our base-case estimate (Table 3), we used an annual incidence of 1 case of kernicterus per 100 000 healthy term live births. Because the true incidence of kernicterus may be lower or higher than this, we also performed a 1-way sensitivity analysis using a range of estimates from 1:10 000 to 1:500 000 healthy term live births (Table 4). In a recent report of preliminary data from New Jersey,22 the CDC reported an incidence of 82 cases of kernicterus per 100 000 live births (identified using International Classification of Diseases, Ninth Revision codes in all infants, not just healthy term infants), an incidence that is encompassed within our range of estimates. Once these preliminary data are confirmed and restricted to healthy term newborns, they will be valuable in refining the cost-effectiveness of kernicterus prevention strategies.
No data are available on the efficacy or effectiveness of any of the suggested strategies in preventing kernicterus. The ideal study design to assess efficacy for each strategy would be a randomized, controlled trial of screening and follow-up, such as the one done for neonatal cystic fibrosis screening.23 In the absence of such evidence of efficacy, we assumed an optimistic relative risk reduction of 0.7 (ie, 70% of the cases of kernicterus currently occurring would be prevented by each of the strategies). We emphasize that this RRR refers to the prevention strategy as applied to a population of newborns, not to the treatment of jaundice in an individual patient. Although kernicterus in an individual healthy term infant is completely preventable, a prevention program applied to a population of newborns is unlikely to eliminate all cases of kernicterus in that population because the effectiveness of such an intervention is the product of its efficacy, compliance, and penetration,24 and there are often deficiencies in compliance and penetration. For example, some of the cases of kernicterus that currently occur are ascribed to deficiencies in health services after hospital discharge, such as delays in obtaining an office appointment for a jaundiced infant, failure to check a bilirubin level in a jaundiced infant, failure to recognize risk factors for severe hyperbilirubinemia, reliance on exposure to sunlight as a treatment for jaundice, and failure to treat hyperbilirubinemia at recommended levels.1,2,4,5 If such deficiencies persist after implementing predischarge prevention strategies, then cases of kernicterus would continue to occur. Increased bilirubin testing before hospital discharge is unlikely to prevent such cases and would represent wasted health care dollars.
Also, implementing a surveillance or screening strategy before discharge may induce a false sense of reassurance and thus alter the behavior of health care providers, perhaps by making them less vigilant. The false-negative rate (sensitivity) of the percentile-based prediction method recommended by Bhutani et al8 is unknown. This might result in the development of kernicterus in some infants who, with current practice, would have been detected and treated in time. Therefore, with each of these strategies, there is a possibility that the degree of risk reduction may not be as high as 0.7, and there may possibly even be an increase in risk, with an increase in the number of cases of kernicterus compared with the current incidence. We also made an assumption that each of the strategies modeled would be equally effective in achieving this RRR. The effectiveness and, consequently, the cost-effectiveness of each of these strategies may vary from each other in different settings across the country.
With our baseline assumptions and estimates and keeping in mind the limitations of the data, we found that the cost to prevent 1 case of kernicterus using the 3 strategies that we modeled ranged from $5 700 000 to $10 000 000, depending on the strategy used. Considerable cost savings ($46 000 000 for the cohort annually) would result from routine predischarge serum bilirubin screening if the incidence of kernicterus were high (1:10 000 births) and the preventive program were highly effective (RRR of 0.7). However, with lower incidence rates and lower RRR estimates, the cost to prevent 1 case ranged from $4 100 000 to as high as $78 000 000. Two key drivers of these wide ranges of costs are the uncertainty in the degree of risk reduction (ie, the uncertainty about the number of cases of kernicterus prevented with each strategy) and the population incidence of kernicterus. Our results provide a framework with which to evaluate the benefits, costs, and risks of 3 potential preventive strategies for kernicterus and that can be applied to other potential strategies as well. When allocating resources to improve health care outcomes, policy makers should keep these costs and uncertainties in mind.
We emphasize that our results do not suggest that attempts should not be made to eliminate kernicterus or that kernicterus is not a disease worth preventing. They do suggest, however, that it is premature to implement large-scale routine bilirubin screening (either serum or transcutaneous) before hospital discharge because of the potential for high costs, uncertain effectiveness, and the uncertain population incidence of kernicterus. Before widespread implementation, the benefits and the lack of risks of such screening first should be confirmed by rigorously testing these strategies on a smaller scale. Because universal follow-up within 48 hours of early discharge may have benefits in addition to kernicterus prevention, such as improved breastfeeding and prevention of dehydration, we speculate, pending quantification of these benefits, that this strategy might still prove to be cost-effective. It is also reasonable to implement other preventive strategies that are recommended by the AAP and other authorities,1,2,4,5 even in the absence of rigorous proof of effectiveness, as they are unlikely to increase health care costs significantly and are unlikely to be harmful. These include a predischarge risk assessment that is based on clinical risk factors20; introduction of policies and procedures that allow nurses to order bilirubin testing for jaundiced newborns and that specifically cover the nurse’s role, documentation, charting requirements, and monitoring of jaundice predischarge; provision to parents of adequate verbal and written information about newborn infants and jaundice; and provision of adequate equipment to test for and treat jaundice.
Additional research is required to understand the epidemiology of kernicterus, identify causes and risk factors for kernicterus, and determine the efficacy of proposed strategies to decrease kernicterus. Research is also required to determine the long-term use of health care resources and costs of caring for children with kernicterus as well as the quality of life in infants who survive kernicterus. With such information, a more comprehensive measure of cost-effectiveness, such as the cost per quality-adjusted life year, can be calculated.
|
FOOTNOTES |
|---|
Reprint requests to (G.S.) MUSC Children’s Hospital, Room 664, Neonatal Division, 165 Ashley Ave, PO Box 250917, Charleston, SC 29425. E-mail: suresh{at}musc.edu
This study was presented in part at the Pediatric Academic Societies Annual Meeting; May 4–7, 2002; Baltimore, Maryland.
|
REFERENCES |
|---|
|
TOPABSTRACTMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. E. Burgos, S. K. Schmitt, D. K. Stevenson, and C. S. Phibbs Readmission for Neonatal Jaundice in California, 1991-2000: Trends and Implications Pediatrics, April 1, 2008; 121(4): e864 - e869. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Maisels and A. F. McDonagh Phototherapy for Neonatal Jaundice N. Engl. J. Med., February 28, 2008; 358(9): 920 - 928. [Full Text] [PDF]
|
|
|
|
 |
|
 S. Wainer, J. Rabi, and M. Lyon Coombs' testing and neonatal hyperbilirubinemia Can. Med. Assoc. J., March 27, 2007; 176(7): 972 - 973. [Full Text] [PDF]
|
|
|
|
|
|
M. Sgro, D. Campbell, and V. Shah Coombs' testing and neonatal hyperbilirubinemia Can. Med. Assoc. J., March 27, 2007; 176(7): 973 - 973. [Full Text] [PDF]
|
|
|
|
|
|
M. Sgro, D. Campbell, and V. Shah Incidence and causes of severe neonatal hyperbilirubinemia in Canada. Can. Med. Assoc. J., September 12, 2006; 175(6): 587 - 590. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A Kanjilal, P L Prasad, and R Keren Clinical assessment of neonatal hyperbilirubinaemia * Author's reply Arch. Dis. Child., November 1, 2005; 90(11): 1202 - 1202. [Full Text] [PDF]
|
|
|
|
 |
|
 D Manning American Academy of Pediatrics guidelines for detecting neonatal hyperbilirubinaemia and preventing kernicterus Arch. Dis. Child. Fetal Neonatal Ed., November 1, 2005; 90(6): F450 - F451. [Full Text] [PDF]
|
|
|
|
 |
|
 The New Hyperbilirubinemia Guidelines: Controversy and Cost Journal Watch Pediatrics and Adolescent Medicine, November 29, 2004; 2004(1129): 10 - 10. [Full Text]
|
|
|
|
|
|
N. A. Holtzman Management of Hyperbilirubinemia: Quality of Evidence and Cost Pediatrics, October 1, 2004; 114(4): 1086 - 1088. [Full Text] [PDF]
|
|
Read all P3Rs
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 |
||
|
||
Home | My Pediatrics | Journal Information | Current Issue | Past Issues | Subscriptions & Services | Contact Us Click here for faster international access © 2004 American Academy of Pediatrics. All rights reserved. |
||
|
||
