Abstract
BACKGROUND. Infants born preterm are at increased risk of adverse health and developmental outcomes. Mortality and morbidity after preterm birth impose a burden on finite public sector resources. This study considers the economic consequences of preterm birth from birth to adult life and compares the costs accruing to those born preterm with those born at term.
METHODS. A decision-analytic model was constructed to estimate the costs to the public sector over the first 18 years after birth, stratified by week of gestational age at birth. Costs were discounted and reported in UK pounds at 2006 prices. Probabilistic sensitivity analysis was used to examine uncertainty in the model parameters and generate confidence intervals surrounding the cost estimates.
RESULTS. The model estimates the costs associated with a hypothetical cohort of 669601 children and is based on live birth and preterm birth data from England and Wales in 2006. The total cost of preterm birth to the public sector was estimated to be £2.946 billion (US $4.567 billion), and an inverse relationship was identified between gestational age at birth and the average public sector cost per surviving child. The incremental cost per preterm child surviving to 18 years compared with a term survivor was estimated at £22885 (US $35471). The corresponding estimates for a very and extremely preterm child were substantially higher at £61781 (US $95760) and £94740 (US $146847), respectively.
CONCLUSIONS. Despite concerns about ongoing costs after discharge from perinatal services, the largest contribution to the economic implications of preterm birth are hospital inpatient costs after birth, which are responsible for 92.0% of the incremental costs per preterm survivor.
The problems of the etiology, epidemiology, management, and sequelae of preterm birth, delivery <37 weeks' gestation, in developed countries have been recently reviewed.1–3 In high-income countries, the incidence of preterm birth is reported to be between 5% and 12%,4,5 with 7.2% of all live births in England born preterm in 2006.6 Infants born preterm are at increased risk of adverse neonatal outcomes, including bronchopulmonary dysplasia, intraventricular hemorrhage, retinopathy of prematurity, and neonatal sepsis, which result in more intensive and longer stays in neonatal care.2,3,7–15 In the longer term, preterm infants are at increased risk of morbidity and disability, including respiratory problems, motor and sensory impairment, learning difficulties, and social and behavioral problems.3,5,10,14,16–21
Although the health sequelae of preterm birth are well documented, relatively little is known about its economic consequences. Several studies estimate the costs of preterm birth during the neonatal period,22–25 although few consider the economic impact over the longer term.5,26,27 Two systematic reviews of the long-term costs of preterm birth highlight the variable methodologic quality of the literature.5,26 The reviews reveal that preterm birth can result in substantial costs to the health sector after the infant's initial discharge from the hospital. It can also impose a substantial burden on special education and social services, on families and caregivers of the infants and, more broadly, on society.
We are aware of 1 other attempt to model the costs of preterm birth throughout childhood: a study by the US Institute of Medicine Committee on Preterm Birth: Causes, Consequences, and Prevention.5 The annual societal economic burden associated with preterm birth in the United States was estimated to be at least $26.2 billion, or $51600 per infant born preterm (2005, US$). However, the authors acknowledge the limitations of their study, including limited data available on the costs associated with disability, unsubstantiated assumptions made about the provision of special education services, and the failure to perform sensitivity analysis.
The objective of the current study was to estimate the economic consequences of preterm birth in England and Wales throughout childhood by using decision-analytic modeling techniques.
METHODS
A Markov model28 was constructed to estimate the costs of preterm birth over the first 18 years of life. The model was designed to estimate the costs associated with a hypothetical cohort of children, the size of which was set at 669601 to reflect the number of live births in England and Wales in 2006.29 Costs were estimated from a public sector perspective, which encompasses health, social, and education services, expressed in pounds sterling and valued at 2006 prices.
Model Structure
The model structure is presented in Fig 1. A cohort enters the model on live birth. Movement between health states is determined by assigning gestation-specific transitional probabilities. The initial distribution of the birth cohort reflects the probability of live birth by week of gestation (from 23 completed weeks onward), and the costs incurred by gestational age categories are then tracked through the model. After the health state “live birth” a small proportion of the cohort die in the delivery room or during transfer, some are “admitted to neonatal care” and the remainder are routinely discharged from the hospital. Of those children “admitted to neonatal care,” a proportion will die although the majority survives to discharge. After discharge from the hospital, children enter a health state defined by the time period “from discharge to 2 years of age.” Survivors at 2 years are then allocated to 1 of 4 states that describe their overall level of disability: none, mild, moderate, or severe. The 4 functional ability states provide an overall assessment of their motor, sensory, and cognitive abilities in line with the criteria provided in Table 1. For each subsequent year of childhood, it is possible for a child to remain in the same disability state, move to another disability state, or to die.
Model Parameters: Transitional Probabilities
A literature search was undertaken to identify data on transitional probabilities between the model health states, differentiated by week of gestational age at birth. The search was limited to studies since 1990 given developments in perinatal practices over the past 2 decades. The scope, methods, and selection criteria for the literature search are reported in detail elsewhere.30 Descriptions and sources for all transitional probabilities are presented in Appendix 1.
Gestation-specific estimates for the probability of live birth were derived from 2005 to 2006 maternity statistics for England6 and supplementary data from the Department of Health obtained through personal communication. Gestation-specific transitional probabilities for death in the delivery room or during transfer and for admission to neonatal care were primarily derived from 2 cohort studies: the EPICure cohort of infants born at <26 weeks' gestation in the United Kingdom and Ireland in 199531 and the EPIPAGE cohort of infants born at <33 weeks' gestation in 9 regions of France in 1997.32 Values for term admissions to neonatal care were from a Swedish study from 1998 to 2001.22
In the absence of national data on neonatal survival compatible with the model structure, individual level data for infants born at <33 weeks' gestation who were admitted to a neonatal unit in the former Trent region of England between 2002 and 2005 were used to estimate the gestation-specific probability of being discharged alive from neonatal care. The Trent Neonatal Network provides recent data and has outcomes that are consistent with available national perinatal data and findings in other preterm birth cohorts.3,29,33 Estimates for the probability of a child dying between hospital discharge and 2 years of age were derived by using patient level data from 3 cohort studies: (1) the EPICure cohort;33 (2) the 1991 to 1992 cohort of the Victorian Infant Collaborative Study Group (VICSG) of infants born in the Victoria state of Australia at 26 or 27 weeks' gestation14; and (3) the Oxford Record Linkage Study for infants born in Oxfordshire or Berkshire between 1990 and 1993 at 28 weeks' gestation or later.34–36 Where there were no data available for the moderately preterm gestational weeks, model parameters were extrapolated from the aforementioned sources by using exponential functions.
The VICSG 1991 to 1992 cohort provided the probabilities of no, mild, moderate, and severe disability at 2 years of age for surviving children born at 23, 24, 25, 26, and 27 gestational weeks and at term.14 For the remaining gestational ages, the probabilities of no, mild, moderate, and severe disability at 2 years of age were estimated by fitting a multinominal logit model to the VICSG data and predicting disability outcomes. Annual transitional probabilities for movements between disability states beyond 2 years of age were derived from analyses of individual level data for the VICSG 1991 to 1992 cohort and its follow-up studies at 2,37 5,38 and 819 years of age. Adjustments were required to account for the annual probability of death for disability groups by using data from life tables.39,40
Model Parameters: Cost of Health States
Cost estimates for the health states were largely derived from primary sources, although supplemented where necessary with secondary data. The literature search was limited to studies published since 2000 to augment 2 systematic reviews on the costs of preterm birth and low birth weight.5,27 Also, changes in health care practices and relative prices of resource inputs limit the generalizability of older cost data to the current clinical context. The literature search is reported elsewhere,30 and the detail of the costs parameters are contained in Appendix 2.
To estimate costs associated with live birth, gestation-specific England and Wales counts on method of delivery, defined as spontaneous delivery, instrumental delivery, elective cesarean section, or emergency cesarean section, were obtained for 2006 from the Department of Health by personal communication. Costs were then estimated by weighting the cost of each method of delivery41 by the likelihoods of each method.
The costs associated with neonatal care were estimated by using gestation-specific England and Wales counts of finished consultant episodes and episode duration (in days) for different levels of neonatal care (special, high dependency, intensive) for 2006 from the Department of Health (by personal communication). Costs were then estimated by summing the combined average lengths of stay and per diem costs for each level of neonatal care.41
Gestation-specific costs between discharge and 2 years were estimated by prospectively collecting resource use data for 290 children in the greater London, Oxford, Portsmouth, and Bristol areas of England. These children had acted as a control group for a prospective cohort study of infants with group B Streptococcus disease.42 The children had no clinical evidence of sepsis during the first 7 days of life and were matched for birth weight and time of birth to the group B Streptococcus cases. Questionnaires were sent to the parents of each infant at 12 and 24 months, recording the number, type, and duration of hospital readmissions and community health and social service contacts over the previous 12 months. Questionnaires were also sent to each child's general practitioner and health visitor at 24 months to obtain data on the child's attendances and referrals to hospitals and other community health and social service providers over the previous 2-year period. Data were crosschecked to ensure validity and completeness. The resource use data were combined with unit costs obtained from primary and secondary sources to generate gestation-specific costs per child over the period between hospital discharge and 2 years.
Costs were estimated for each disability state by using detailed resource use and outcomes data collected as part of the EPICure study. Parents of children in the EPICure study and their term-born controls completed questionnaires about their child's hospital inpatient admissions and outpatient visits, contact with community health and social care professionals, and use of education services during the sixth year of life.43 These data were combined with unit costs to obtain a net cost per child during the sixth year of life.43 Mean annual costs and their respective distributions were estimated for each disability state on the basis of the children's overall disability classifications at 6 years of age. Each child was classified as having no, mild, moderate, or severe disability by using the criteria contained in Table 1. These criteria are consistent with the disability definitions applied to the VICSG cohorts,14 ensuring consistency in the disability criteria used for estimating costs and transitional probabilities.
The costs associated with a child's death were based on observational research and encompassed post mortem examination and associated procedures and the cost of counseling parents.44
Analytical Methods
A model was used to generate total costs for an annual cohort of children born in England and Wales, as well as costs stratified by week of gestational age at birth and preterm category (extremely preterm, very preterm, all preterm). In addition, average costs per survivor and the additional, or incremental, cost per surviving child born preterm compared with term were reported. Cost estimates were further disaggregated into cost categories (hospital inpatient, hospital outpatient, community health and social care, education) and periods of life (delivery, neonatal, discharge to 2 years, 2–5 years, 5–11 years, and 11–18 years). Costs accruing after the first year of life were discounted by using an annual rate of 3.5%.45,46
Probabilistic sensitivity analysis was used to examine the uncertainty in model parameter estimates and generate confidence intervals (CIs) surrounding cost estimates.27 One-way sensitivity analyses were performed to examine the impact on costs of 2 assumptions. First, a linear model, rather than a multinomial logit model, based on data from the VICSG 1991 to 1992 cohort study was used to estimate disability outcomes at 2 years of age for children born between 28 and 36 weeks' gestation. Second, it was assumed that were no movements between disability states beyond 2 years of age. Analyses were performed by using Excel (Microsoft Corp, Redmond, WA) and Stata (Stata Corp, College Station, TX) software.
RESULTS
The health outcomes predicted by the model are summarized in Table 2 for the entire cohort and for each preterm category. The total cost of preterm birth to the public sector was estimated at £2.946 billion over childhood (Table 3). Of this, 34% is attributable to those born very preterm and 8% to those born extremely preterm. An inverse relationship was identified between gestational age at birth and the average public sector cost per surviving child. The average cost per surviving child born at term to 18 years was estimated at £41907, whereas the corresponding values for the preterm, very preterm, and extremely preterm categories were 1.54, 2.46, and 3.24 times higher, respectively. The incremental cost per preterm child surviving to 18 years compared with a term survivor was estimated at £22764. The incremental cost substantially increased for the very and extremely preterm groups at £61509 and £94190, respectively.
Estimates of the incremental cost per preterm survivor are disaggregated into cost categories and periods of life in Table 4. Preterm birth is on average associated with increased costs in every cost category and for every period of childhood. The table highlights the importance of hospital inpatient costs borne during the neonatal period in explaining the additional public sector costs of preterm birth. Costs borne during the neonatal period are responsible for 92.0% of the incremental cost per preterm survivor.
Finally, the 1-way sensitivity analyses showed that our assumptions on the disability outcomes at 2 years of age for children born between 28 and 36 weeks' gestation and the movements between disability states beyond 2 years of age had a minimal impact on total public sector costs and on the incremental cost per preterm survivor.
DISCUSSION
This is, to our knowledge, the first attempt to model the long-term costs of preterm birth throughout childhood in a UK setting. The total cost of preterm birth to the public sector was estimated at £2.946 billion. Although 7.2% of all births are preterm, 10.2% of all costs are incurred by this group and, therefore, a child born preterm imposes an additional cost to the public sector. The incremental cost per preterm child surviving to 18 years when compared with a term survivor was estimated at £22885 throughout childhood, and increased for the very and extremely preterm child, being 2.70 and 4.14 times higher than for a preterm child, respectively. These cost estimates are broadly in line with the other modeling study of the costs of preterm birth, which had estimated the incremental lifetime costs of a preterm birth in the United States at $51600 (2005 $US).5 This suggests that our cost estimates are generally applicable to other developed countries, and we expect the marginally higher costs in the United States to reflect differences in the health care system and the longer time horizon of the model.
Disaggregated total cost estimates highlight the costs associated with neonatal care and education, whereas the incremental cost of preterm birth are largely explained by admission to neonatal care. These findings reflect the low differential rate in the need for special education and the lack of unit cost data for special education services provided in mainstream schools.
Strengths and Limitations
The strengths of the model include disaggregation of cost estimates by week of gestation and by broad category of preterm birth, the application of several validated sources of primary data on health outcomes, resource use and unit costs, and the application of the latest analytical methods within the decision-analytic modeling. The study does, however, have some limitations. First, a paucity of published data on the effects of preterm birth into adulthood restricted the time horizon for our analyses to the childhood years. Second, by adopting a public sector perspective our study has excluded the broader societal impacts of preterm birth. For example, future analyses could include costs borne by local authorities and voluntary organizations, or by families and informal caregivers, such as the additional costs of travel, child care, and accommodation.
Moderate Preterm Birth
Our study revealed that of the £2.946 billion annual economic burden of preterm birth to the public sector, £1.956 billion (66.4%) is attributable to moderate preterm birth. Although data on the consequences of moderate preterm birth are sparse, they consistently suggest that infants born at between 33 and 36 weeks of gestation may be at increased risk of adverse growth,47,48 neuropsychological,47,49,50 educational,51 and behavioral52 outcomes. We are not aware of any UK prospective studies focusing on this population. There is, therefore, no means of accurately quantifying neonatal morbidity and long-term outcomes for this population or for identifying risk factors contributing to these outcomes.
Costs of Neonatal Care
The study highlighted the costs experienced during the neonatal period. In absolute terms, almost one third of the total public sector economic burden of preterm birth is borne during this period, whereas 92.1% of the incremental cost per preterm survivor is represented by neonatal care. To disaggregate the costs of neonatal care by gestational age, we performed a cost analysis by using Health Episode Statistics data on the count of finished consultant episodes and episode duration by level of neonatal care and gestational age (by personal communication), as well as unit cost data derived from 2006 National Health Service reference costs.41 There is a need, however, for more formal accounting studies that calculate detailed costs of each level of neonatal care.
Clinical and Policy Implications
Our study has important clinical and policy implications. In addition to informing the planning of services, the broad economic aggregates can provide a basis for assessing competing research strategies. The model can also be used as a basis for simulating the costs and consequences of interventions aimed at preventing preterm birth or alleviating its effects. For example, in additional analyses, we simulated the consequences of a hypothetical intervention that delayed preterm birth by 1 week across all gestational categories: the total public sector cost of preterm birth (excluding any intervention costs) fell from £2.946 billion to £1.952 billion.
CONCLUSIONS
The results of this study should facilitate the effective planning of perinatal and pediatric services and inform the development of future economic evaluations of interventions aimed at preventing preterm birth or alleviating its effects. Additional research is required that identifies, measures, and values the broader and longer-term economic impact of preterm birth in a valid and reliable manner.
Parameter Values for Decision-Analytical Markov Model: Transitional Probabilities
Parameter Values for Decision-Analytical Markov Model: Cost of Health States
Structure of the Markov model estimating the childhood economic costs of preterm birth.
Classification of Functional Ability States Values were Deleted from the Abstract Per Journal Style. Also, Per Journal Style, All Abstracts Must Be Less than 300 Words Long (Note also that no Abbreviations can be Used). If Your Abstract Extends Beyond 300 Words, Please Edit as Necessary (If Changes are Extensive, Include or e-mail an Electronic Text File with the New Abstract)
Health Outcomes at 18 Years of Age
Public Sector Costs Associated With Preterm Birth (£2006 Prices)
Incremental Cost per Preterm Survivor Disaggregated by Cost Category and Period of Life (£2006 Prices)
Acknowledgments
We thank Tommy's, The Baby Charity, for funding the work reported here. The National Perinatal Epidemiology Unit was funded by the Department of Health for England. The Health Economics Research Centre was funded by the National Coordinating Centre for Research Capacity Development, England. Dr Petrou was supported by a Medical Research Council senior nonclinical research fellowship. The views expressed by the authors do not necessarily reflect those of the funding bodies.
We thank Jane Henderson for contribution to the literature reviews and colleagues at the Information Centre for Health and Social Care and the Department for Children, Schools, and Families for providing some of the parameter inputs for the decision-analytic model. We also thank members of the advisory group (Professor Peter Brocklehurst, Professor Andrew Wilkinson, Professor Dieter Wolke, Professor Martin Knapp, Dr Edward Wozniak, Ms Jane Brewin, Ms Anita Charlesworth, and Ms Michelle Deans), which met 3 times and provided an invaluable steer to the study.
Footnotes
- Accepted October 20, 2008.
- Address correspondence to Lindsay Mangham, MA, MBA, London School of Hygiene and Tropical Medicine, Health Policy Unit, Keppel Street, London WC1E 7HT, United Kingdom. E-mail: lindsay.mangham{at}lshtm.ac.uk
The authors have indicated they have no financial relationships relevant to this article to disclose
What's Known on This Subject
Relatively little is known about the long-term costs associated with preterm birth. There has been 1 earlier modeling study that estimated the lifetime costs in the United States, but limited detail of the model is known.
What This Study Adds
This is the first study to estimate the long-term costs of preterm birth in childhood in England and Wales. The results of the study are presented by week of gestation and differentiated into various cost categories.
REFERENCES
- Copyright © 2009 by the American Academy of Pediatrics
In this issue
Jump to section
Related Articles
Cited By...
- Economic evaluation alongside the Speed of Increasing milk Feeds Trial (SIFT)
- Systematic review and network meta-analysis with individual participant data on Cord Management at Preterm Birth (iCOMP): study protocol
- Cost-effectiveness of strategies preventing late-onset infection in preterm infants
- Systematic review and network meta-analysis with individual participant data on cord management at preterm birth (iCOMP): study protocol
- The impact of level of neonatal care provision on outcomes for preterm babies born between 27 and 31 weeks of gestation, or with a birth weight between 1000 and 1500 g: a review of the literature
- Impact of preterm birth on brain development and long-term outcome: protocol for a cohort study in Scotland
- Inconsistent outcome reporting in large neonatal trials: a systematic review
- Cost of preterm birth during initial hospitalization: A care providers perspective
- Optimising neonatal service provision for preterm babies born between 27 and 31 weeks gestation in England (OPTI-PREM), using national data, qualitative research and economic analysis: a study protocol
- Economic consequences of preterm birth: a systematic review of the recent literature (2009-2017)
- Hypothermia for perinatal asphyxia: trial-based resource use and costs at 6-7 years
- Open randomised trial of the (Arabin) pessary to prevent preterm birth in twin pregnancy with health economics and acceptability: STOPPIT-2--a study protocol
- Designing the landscape for technological development in neonatal neurocritical care
- Continuing Education Module--Kangaroo Mother Care 2: Potential Beneficial Impacts on Brain Development in Premature Infants
- Ambient temperature as a trigger of preterm delivery in a temperate climate
- A pharmacoeconomic approach to assessing the costs and benefits of air quality interventions that improve health: a case study
- Recent guidance on antenatal corticosteroids in prematurity
- Evidence, Quality, and Waste: Solving the Value Equation in Neonatology
- Economic Benefits and Costs of Human Milk Feedings: A Strategy to Reduce the Risk of Prematurity-Related Morbidities in Very-Low-Birth-Weight Infants
- Prevention of preterm births: are we looking in the wrong place? The case for primary prevention
- Making GRADE accessible: a proposal for graphic display of evidence quality assessments
- Association between thyroid autoantibodies and miscarriage and preterm birth: meta-analysis of evidence
- A structured review of the recent literature on the economic consequences of preterm birth
- Functional cardiac MRI in preterm and term newborns
- End-of-life decisions as bedside rationing. An ethical analysis of life support restrictions in an Indian neonatal unit