Abstract
Objectives. To compare the cumulative use and cost of hospital inpatient services to 5 years of age by individuals divided into 4 subgroups by gestational age at birth.
Design. Costs applied to the hospital service utilization profile of each infant born in 2 areas covered by the Oxford Record Linkage Study during 1970–1993.
Setting. Oxfordshire and West Berkshire, southern United Kingdom.
Subjects. 239 694 individuals divided into 4 subgroups by gestational age at birth: <28 weeks, 28 to 31 weeks, 32 to 36 weeks, ≥37 weeks.
Main Outcome Measures. Number and duration of hospital admissions during the first 5 years of life and costs, expressed in £ sterling and valued at 1998–1999 prices, of hospital inpatient services.
Results. The total duration of hospital admissions for infants born at <28 and at 28 to 31 gestational weeks was 85 and 16 times that for term infants, respectively, once duration of life had been taken into account. Hospital inpatient service costs were significantly higher for preterm infants than for term infants, with the cost differences persisting throughout infancy and early and mid-childhood. Over the first 5 years of life, the adjusted mean cost difference was estimated at £14 614 (US $22 798) when infants born at <28 weeks gestational age were compared with term infants and £11 958 (US $18 654) when infants born at 28 to 31 weeks gestational age were compared with term infants. Independent contributions to total cost came from being born: small for gestational age, a multiple, during the 1970s and early 1980s, to a woman of extreme maternal age or who was hospitalized antenatally, and from experiencing extended survival or childhood disease. However, preterm birth remained the strongest predictor of high cost.
Conclusions. Preterm birth is a major predictor of how much an individual will cost hospital service providers during the first 5 years of life.
The high rates of morbidity and mortality arising from preterm birth impose a considerable burden on finite health care resources. Preterm infants are at increased risk of a range of adverse neonatal outcomes including chronic lung disease,1 severe brain injury,2 retinopathy of prematurity,3 necrotizing enterocolitis,4 and neonatal sepsis.5 In later life, preterm infants are at increased risk of motor and sensory impairment,6,7 learning difficulties,8–12 and behavioral problems.13–16
Although the health sequelae of preterm birth are well-documented, relatively little is known about its economic consequences. The Organization for Economic Cooperation and Development has provided estimates of the cost of a range of conditions originating in the perinatal period, including preterm birth, to the acute health services of several industrialized countries.17 However, these cost estimates were derived from widely differing national data systems with a varying degree of methodologic rigor. In addition, a number of studies have been conducted with the specific aim of estimating the costs of preterm birth during the neonatal period.18,19 Although informative, these studies have tended to overlook the condition’s economic impact over the longer term.
Assessments of the economic consequences of preterm birth throughout childhood could provide a framework for identifying priorities for research and development, inform future economic evaluations of prevention and treatment strategies, and act as an invaluable resource to clinical decision-makers and budgetary and service planners. The goal of our study was to conduct a comprehensive economic assessment of the hospital inpatient service utilization and costs of a large, geographically defined cohort of infants during the first 5 years of life. The study also set out to identify clinical and sociodemographic predictors of hospital inpatient costs during the first 5 years of life. The specific hypothesis tested was that preterm birth is associated with significantly increased hospital inpatient service utilization and costs throughout infancy and early and mid-childhood.
METHODS
Oxford Record Linkage Study (ORLS)
Data from the ORLS formed the basis of the investigation. The ORLS is a collection of linked, anonymized birth registrations, death certificates, and statistical abstracts of National Health Service (NHS) hospital inpatient and day case admissions for part of southern United Kingdom.20 Data collection in the ORLS covered Oxfordshire and West Berkshire from 1966 and increased its population coverage to include 6 of the 8 districts of the former Oxford region from 1975 and the whole of the former Oxford region from 1984.20 The ORLS maternity and perinatal data were derived from the ORLS’s own data collection systems, which covered Oxfordshire and West Berkshire only until 1989 and from maternity hospital episodes statistics thereafter.21
Study Population
The study population comprised all infants born to women who both lived and delivered in Oxfordshire or West Berkshire during the period from January 1, 1970, to December 31, 1993. These time limits arose because, before 1970, much of the relevant perinatal information had not been routinely extracted from the patient records, and a delivery cutoff point of December 31, 1993 was required for follow-up to cover the first 5 years of life. For the purpose of analysis, the data were stratified into 4 subgroups according to the infants’ gestational age at birth: <28 weeks; 28 weeks to 31 completed weeks inclusive; 32 weeks to 36 completed weeks inclusive; and ≥37 weeks. The data were also stratified into four 6-year time periods according to their date of birth (1970–1975, 1976–1981, 1982–1987, and 1988–1993) to capture the impact of alterations in health care practices during the study period. Over the study period, ∼6% of births to residents of Oxfordshire and West Berkshire took place outside of these 2 areas. These were not included in the analyses. In addition, it was not possible to trace the admissions of the study population to hospitals outside the areas covered at the relevant times by the ORLS.
Hospital Service Utilization
A record of hospital inpatient service utilization between birth and 5 years of age was compiled for all study infants. The data extracted from the ORLS included the date of each hospital admission, the duration of hospital stay, specialty on admission, operative procedures performed, and diagnoses recorded on discharge from hospital based on codes from the eighth and ninth revisions of the International Classification of Diseases.22,23 Each day-case admission was counted as a full 24-hour period for the purposes of this study. We calculated the total time spent in the hospital by each individual by summing the lengths of stay of each individual’s successive admissions. In addition, estimates of service utilization were calculated for all infants who were alive at the start of the period of life of interest (initial birth admission, consecutive years of life, and first 5 years), with censoring for deaths.
Hospital Service Costs
Hospital inpatient service costs were calculated for each hospital admission regardless of the diagnoses recorded on discharge. The cost of each hospital admission was estimated by multiplying the length of stay by the per diem cost of the respective specialty. The specialty-based per diem costs were based on the English Department of Health’s NHS Trust Financial Returns for 1997–1998 and 1998–1999, which had been averaged over these 2 financial years to eliminate any random fluctuation in the data. These returns incorporate short-run current average revenue costs plus revenue and capital overheads and are widely accepted as reliable indicators of hospital service costs.24 For the hospital records with an unknown or incorrect specialty code, the per diem medical or surgical cost was applied, depending on the approximate ORLS code range. No subspecialty coding was available for psychiatric hospitalizations, so per diem psychiatric costs were applied to each day of a psychiatric admission. All costs are expressed in constant £ 1998–1999 sterling by using the NHS Hospital and Community Health Services pay and price deflators provided by the English Department of Health.
Statistical Analysis
The number of deaths over the first 5 years of life among live births was examined by using survival analysis. A multivariate Cox regression model was used to compare the number of deaths in the 4 gestational age subgroups and over the 4 time periods. Hazard ratios (HRs) and 95% confidence intervals (CIs) for the independent effects of gestational age and time period were calculated. The probabilities of early neonatal deaths (0–7 days), late neonatal deaths (8–28 days), postneonatal deaths (29–365 days), and deaths during years 1 through 5, conditional on being alive at the start of the period, were estimated by using the life-table method.
The total duration of hospital admissions, including the initial birth admission, during the first 5 years of life was compared between gestational age subgroups and time periods of birth by using a multivariate negative binomial regression.25 Data on all study infants were incorporated into this analysis. Relative rates and 95% CIs for the number of days of hospitalization are reported for the multivariate negative binomial regression, accounting for and not accounting for duration of life.
Cost differences between the 4 gestational age subgroups that occurred during the initial birth admission and during each of the first 5 years of life were tested by using simple linear regression, with only children alive at the start of each period included within cost estimates. In addition, the effect of gestational age on the total 5-year costs of all study infants was analyzed, first in a simple linear regression and then in a multiple regression, taking account of clinical and sociodemographic factors that were considered to influence costs.19 Confounding factors considered in the regression analyses included whether the infant was small for gestational age (<10th percentile of birth weight),26 year of birth (1970–1975, 1976–1981, 1982–1987, or 1988–1993), sex, survival period (<8 days, 8–28 days, 29 days–1 year, 1–5 years, or >5 years), maternal age at time of delivery (<20 years, 20–35 years, or >35 years), multiplicity of birth (singleton or multiple), social class based on the male partner’s occupation (I, II, III, IV, or V),27 maternal hospitalized days during the antenatal period (none, 1–10, 11–20, or ≥21), and the presence at any time point during the study period of childhood diseases included in the International Classification of Diseases.22,23 The size of the study sample (n = 233 082 in the multiple regression) was sufficiently large to expect robust parameter estimates, and therefore alternative methods such as bootstrapping techniques were not applied.28
For all statistical analyses, differences were considered significant if P values were .001 or less. All analyses were performed with a microcomputer using SAS 8.0 software (SAS Institute Inc, Cary, NC).
RESULTS
Study Population
A total of 281 212 infants were born to women who both lived and delivered in Oxfordshire or West Berkshire during the study period, 278 683 of whom were born alive. Information on gestational age at birth was available for 239 694 (86.0%) infants, which closely resembled the total population of live births in terms of clinical and sociodemographic characteristics (data available on request). The number of births, deaths and mortality rates among infants for whom information on gestational age at birth was available is shown in Table 1. When compared with term infants (born at ≥37 gestational weeks), the number of deaths over the first 5 years of life was 146 times higher (HR: 145.7; 95% CI: 127.1–166.9) among infants born at <28 gestational weeks, 33 times higher (HR: 33.3; 95% CI: 28.9–38.2) among infants born at 28 to 31 gestational weeks, and 5 times higher (HR: 5.2; 95% CI: 4.6–5.8) among infants born at 32 to 36 gestational weeks. Mortality differences between the gestational age groups were greatest during the early neonatal period. When compared with term infants, the numbers of early neonatal deaths were 368 (HR: 368.5; 95% CI: 310.7–437.0), 85 (HR: 84.6; 95% CI: 70.8–101.0), and 11 times higher (HR: 10.6; 95% CI: 8.9–12.4) among infants born at <28, 28 to 31, and 32 to 36 gestational weeks, respectively. The number of deaths during the first 5 years of life declined over the 4 time periods. Compared with the period 1970–1975, the number of deaths decreased by 20% (95% CI : 12%–28%), 48% (95% CI: 42%–54%), and 68% (95% CI: 64%–72%) during 1976–1981, 1982–1987, and 1988–1993, respectively.
Births, Deaths, and Mortality Rates Among Infants Born to Women Living and Delivering in Oxfordshire or West Berkshire by Gestational Age
Hospital Service Utilization
The median number (and interquartile range) of admissions during the first 5 years of life, for infants born at <28 gestational weeks, at 28 to 31 gestational weeks, at 32 to 36 gestational weeks, and at term were 2 (1, 2), 2 (1, 3), 1 (1, 2), and 1 (1, 2), respectively. The total duration of hospital admissions during the first 5 years of life for infants born at <28 gestational weeks and for infants born at 28 to 31 gestational weeks was almost 8 times that for term infants (Table 2). Taking into account duration of life, the total duration of admissions for infants born at <28 gestational weeks and for infants born at 28 to 31 gestational weeks increased to 85 and 16 times that for term infants, respectively. The total duration of admissions for infants born at 32 to 36 gestational weeks was 2.6 times that for term infants without accounting for duration of life and 3 times that for term infants once duration of life had been accounted for.
Total Duration of Hospital Admissions (Days) During the First 5 Years of Life by Gestational Age and Time Period of Birth
The median number (and interquartile range) of admissions during the first 5 years of life for infants born between 1970 and 1975, 1976 and 1981, 1982 and 1987, and 1988 and 1993 were 1 (1, 1), 1 (1, 2), 1 (1, 2), and 1 (1, 1), respectively. The total duration of admissions for infants born during the first 3 time periods was between 1.17 and 1.30 times that for infants born during 1988–1993 without accounting for duration of life and between 1.21 and 1.33 times once duration of life had been accounted for (Table 2).
Cost Differences Between Gestational Age Subgroups
For each gestational age subgroup, the largest component of cost was incurred during the infant’s first year of life (Table 3). Relative to term infants, initial birth admission costs were significantly higher for all preterm subgroups (P < .0001): £5818 (95% CI: £5730–5905) higher among infants born at <28 gestational weeks; £5948 (95% CI: 5895–6002) higher among infants born at 28 to 31 gestational weeks; and £1170 (95% CI: £1151–1188) higher among infants born at 32 to 36 gestational weeks. Among infants who survived the initial birth admission, the cost of first year readmissions was £10 333 (95% CI: £9991–10 675) higher for infants born at <28 gestational weeks compared with term infants, £6976 (95% CI: £6802–7149) higher for infants born at 28 to 31 gestational weeks, and £1720 (95% CI: £1663–1776) higher for infants born at 32 to 36 gestational weeks. The cost differences between the preterm subgroups and the term reference group persisted through the subsequent 4 years of life (P < .0001) with the exception of the £44 cost difference in fifth-year readmission costs between infants born at <28 gestational weeks and those born at term (P = .32).
Mean Cost of Hospital Inpatient Services (£ UK 1998–1999) During Infancy and Early and Mid-childhood by Gestational Age
Clinical and Sociodemographic Characteristics in Relation to Total Cost
Relationships between the clinical and sociodemographic characteristics of the study population and total hospital inpatient service costs during the first 5 years of life are shown in Table 4. The simple linear regression revealed gestational age at birth to be the strongest predictor of total costs during the first 5 years of life (R2: 0.093). Other factors significantly associated with total costs during the first 5 years of life included small for gestational age (R2: 0.007), year of birth (R2: 0.003), sex (R2: 0.001), survival period (R2: 0.020), multiplicity of birth (R2: 0.008), maternal age at time of delivery (R2: 0.001), maternal hospitalized days before birth (R2: 0.007), and the presence of all categories of childhood disease (R2: 0.002–0.033). Although social class was also significantly associated with total costs, it was excluded from additional analyses, partly as a consequence of its low explanatory value (R2: 0.0004) and partly because social class data were not collected in the hospital systems after 1990.
Clinical and Sociodemographic Factors Predicting Cumulative Hospital Inpatient Service Costs (£ UK 1998–1999) by Age 5 Determined by Simple Linear Regression (n = 239 694)
Clinical and Sociodemographic Predictors of Total Cost
The results of the multiple regression analysis are shown in Table 5. Preterm birth was significantly associated with increased hospital inpatient service costs during the first 5 years of life (P < .0001). The adjusted mean cost difference was estimated at £14 614 (95% CI: £14 268–14 960) when infants born at <28 weeks gestational age were compared with term infants and £11 958 (95% CI: £11 756–12 160) when infants born at 28 to 31 weeks gestational age were compared with term infants. All other clinical and sociodemographic factors incorporated within the multiple regression were independently associated with total hospital inpatient service costs during the first 5 years of life with the exception of the sex of the infant (P = .032). On average, costs were greater for infants who were born small for gestational age (P < .0001) during the 1970s and early 1980s (P < .0001) and survived for >28 days (P < .0001). In addition, being born a multiple (P < .0001), to mothers aged <20 years or >35 years (P = .0009) or to mothers who were hospitalized antenatally (P < .0001) was associated with significantly increased costs. The presence of all categories of childhood disease at any time point during the study period was also associated with significantly increased costs (P < .0001).
Clinical and Sociodemographic Factors Predicting Cumulative Hospital Inpatient Service Costs (£ UK 1998–1999) by Age 5 Determined by Multiple Regression (n = 233 082*)
DISCUSSION
A search of the published and unpublished medical and health economics literature by us revealed that relatively few studies have estimated the long-term economic impact of preterm birth.18,19,29 Moreover, the studies that have been conducted to date were based on relatively small samples of infants and frequently violated current methodologic requirements of health economic evaluation.19 Our study generated a mean cost difference of £14 614 (US $22 798) when infants born at <28 weeks gestational age were compared with term infants and £11 958 (US $18 654) when infants born at 28 to 31 weeks gestational age were compared with term infants. Moreover, these cost differences had been calculated after a multivariate analysis controlled for a wide range of clinical and sociodemographic confounding factors including birth weight for gestational age, year of birth, sex, survival period, maternal age and morbidity, multiplicity of birth, and a comprehensive profile of childhood diseases.
The major strengths of this study are that it was based on a large cohort of infants in a geographically defined area and included a comprehensive and validated record of hospital inpatient service data.30,31 The availability of linked data enabled us to relate gestational age at birth to subsequent admissions and to sum lengths of stay of individuals’ successive admissions to calculate their total time spent in hospital. This provided a reliable basis for estimating the economic implications of preterm birth. The study-cost accounting was comprehensive and included all significant hospital inpatient cost items. Furthermore, the study sample size was large enough to detect statistically significant differences in the cost of hospital inpatient care between infants of varying gestational age at birth. Therefore, we would argue that the study was sufficiently sized to arrive at conclusions that are both meaningful and relevant to decision-makers.
This study does have limitations that should be borne in mind. By focusing on the cost of hospital inpatient services provided to infants, this study has adopted a narrow health service perspective. A small number of studies has shown that survivors of low gestational age and birth weight consume significantly greater family practitioner services,32 education services,33 and social services34 than infants born at term or at normal birth weight. Moreover, it is likely that preterm birth has other long-term consequences that require evaluation from an economic perspective.
A second limitation is the 5-year time frame on which the economic study is based. Although the effects of preterm birth are acutely felt during the first 5 years postpartum,1–7 it is likely that the condition has longer-term consequences in terms of health status and health service utilization over the child’s lifetime6 and of the child’s educational8–12 and social13–16 requirements. If this is the case, then longer-term research is required to provide a complete assessment of the condition’s economic implications.
A third limitation of this study is that the specialty-based per diem costs applied to each hospital admission may not have captured subtle differences in the care provided to infants with varying diagnoses. The English Department of Health has compiled an alternative data set of NHS reference costs,35 which is based on categories of acute care interventions that are clinically distinct and have similar implications for resources. However, these costs have been criticized for having improbably wide ranges of values for the same health care resource groups,36 and it was decided therefore that the NHS Trust Financial Returns provided the most rigorous values for our calculations.
A fourth limitation of this study is that the data set did not include admissions of the study population to hospitals outside the former Oxford region. As a result, we might have underestimated the absolute levels of hospital service utilization and costs for our study population. Nevertheless, there is no evidence to suggest that the mean difference in care costs between the gestational age subgroups and, consequently, the costs that can be attributed to preterm birth are affected by the level of migration.
A final limitation is that the generalizability of the study findings are constrained by differences in care policies across industrialized nations as well as differences in the relative price structures of resource inputs across health care systems. However, evidence of similar care practices for preterm infants across industrialized nations37 suggests that the geographic generalizability of our results may be extensive, although country-specific estimates of unit costs are required to quantify likely cost estimates in local settings.
Despite the limitations of our study, the results have important clinical and policy implications. Given recent evidence of an increasing incidence of preterm birth,37 it is imperative that clinical decision-makers and budgetary and service planners recognize the overall economic impact of the condition in their service planning as well as the potential contribution of clinical and sociodemographic factors to future health care costs.
In addition to informing the planning of services, the broad economic aggregates that we have estimated can provide a basis for assessing competing strategies for research and prevention. However, it should be noted that cost data alone cannot identify the most efficient allocation of finite health care resources. Rather, it is information on incremental costs and incremental health gains attributable to particular health care interventions that can identify the combination of human and material inputs that maximize health benefits. This can be achieved through the general framework of economic evaluation. A number of interventions are effective at ameliorating the morbidity and mortality associated with preterm birth.38 In contrast, relatively few interventions have proved effective at preventing or delaying preterm birth.38 For finite resources in this area to be allocated in an efficient manner, it is imperative that economic evaluations of effective interventions are conducted. Our data should be of interest and use to researchers planning to evaluate new or existing interventions from an economic perspective, particularly those wishing to incorporate within a decision-analytic framework the long-term economic impact of preterm birth and the cost effectiveness of prevention and treatment strategies.
Acknowledgments
This study was funded by Wellcome Trust Health Services Research grant 068516/Z/02/Z. The National Perinatal Epidemiology Unit is core-funded by the Department of Health, United Kingdom. Data collection for the ORLS was funded by the former Oxford Regional Health Authority. The Unit of Health Care Epidemiology is funded by the Department of Health and Social Care (South), United Kingdom.
We thank colleagues at the National Perinatal Epidemiology Unit, University of Oxford, who commented on successive drafts and have, as always, given helpful advice.
Footnotes
- Received November 22, 2002.
- Accepted February 18, 2003.
- Reprint requests to (S.P.) National Perinatal Epidemiology Unit, Institute of Health Sciences, University of Oxford, Old Rd, Headington, Oxford OX3 7LF, United Kingdom. E-mail: stavros.petrou{at}perinatal-epidemiology.oxford.ac.uk
The views expressed by the authors do not necessarily reflect those of the funding bodies.
REFERENCES
- Copyright © 2003 by the American Academy of Pediatrics