Home-based care after a shortened hospital stay versus hospital-based care postpartum: an economic evaluation
Dr S. Petrou, National Perinatal Epidemiology Unit, Institute of Health Sciences, University of Oxford, Old Road, Headington, Oxford OX3 7LF, England, UK.
Objectives To compare the cost effectiveness of early postnatal discharge and home midwifery support with a traditional postnatal hospital stay.
Design Cost minimisation analysis within a pragmatic randomised controlled trial.
Setting The University Hospital of Geneva and its catchment area.
Population Four hundred and fifty-nine deliveries of a single infant at term following an uncomplicated pregnancy.
Methods Prospective economic evaluation alongside a randomised controlled trial in which women were allocated to either early postnatal discharge combined with home midwifery support (n= 228) or a traditional postnatal hospital stay (n= 231).
Main outcome measures Costs (Swiss francs, 2000 prices) to the health service, social services, patients, carers and society accrued between delivery and 28 days postpartum.
Results Clinical and psychosocial outcomes were similar in the two trial arms. Early postnatal discharge combined with home midwifery support resulted in a significant reduction in postnatal hospital care costs (bootstrap mean difference 1524 francs, 95% confidence interval [CI] 675 to 2403) and a significant increase in community care costs (bootstrap mean difference 295 francs, 95% CI 245 to 343). There were no significant differences in average hospital readmission, hospital outpatient care, direct non-medical and indirect costs between the two trial groups. Overall, early postnatal discharge combined with home midwifery support resulted in a significant cost saving of 1221 francs per mother–infant dyad (bootstrap mean difference 1209 francs, 95% CI 202 to 2155). This finding remained relatively robust following variations in the values of key economic parameters performed as part of a comprehensive sensitivity analysis.
Conclusions A policy of early postnatal discharge combined with home midwifery support exhibits weak economic dominance over traditional postnatal care, that is, it significantly reduces costs without compromising the health and wellbeing of the mother and infant.
The postnatal length of stay following ‘normal’ delivery has declined in most industrialised countries in recent years, largely as a consequence of increased efforts to control health care costs and a popular trend towards demedicalising aspects of childbirth.1 The relative benefits and risks associated with early postnatal discharge policies have been evaluated by eight randomised controlled trials to date,2–9 of which only one included an economic evaluation.6 That study compared an early discharge policy after unplanned caesarean delivery, accompanied by a minimum of two home visits and 10 telephone calls, with standard care in hospital without follow up at home. The early discharge policy resulted in substantial savings to health care providers. However, application of hospital charges rather than costs is likely to have resulted in an over-estimation of the real savings that can be attributed to early discharge.
Several economic analyses based on cohort studies or case series have concluded that the economic value of the resources released by early postnatal discharge are not offset by increased costs to other sectors of the health service or the wider economy.10–20 However, in addition to basing assessments of the efficacy of early postnatal discharge policies on observational evidence, these economic analyses were characterised by a number of methodological limitations. These included the adoption of a narrow economic perspective,10–13,15–20 a failure to provide disaggregated information on reported costs10,12–16,18–20 and limited sensitivity analyses.10–12,14–20
In this article, we report the results of a prospective economic evaluation that was conducted in the context of a relatively large randomised controlled trial, designed to overcome the methodological deficiencies of earlier studies. The economic evaluation tested the hypothesis that a policy of early discharge from hospital combined with home midwifery support reduces costs compared with a traditional postnatal hospital stay, without compromising the health and wellbeing of the mother and the infant. The economic evaluation has been written in conjunction with the clinical results of the randomised controlled trial reported in this issue.21
A pragmatic randomised controlled trial was conducted to evaluate the benefits and risks associated with a policy of early postnatal discharge combined with home midwifery support. Nulliparous and multiparous women delivering a single infant at term (≥37 weeks of gestation) following an uncomplicated pregnancy were randomised in the trial. All women were recruited from the sole public hospital in Geneva, Switzerland, the University Hospital of Geneva (HUG), between November 1998 and October 2000. Women allocated to the home-based care group were scheduled for hospital discharge at 24–48 hours after a vaginal delivery or 72–96 hours after a caesarean section. A midwife then visited these women at home during the first 10 days postpartum, with the number of visits and the interval between visits determined by the needs of the family. Women allocated to the hospital-based care group were scheduled for hospital discharge at four to five days after a vaginal delivery or six to seven days after a caesarean section, without subsequent home midwifery support unless clinically indicated. The primary outcomes of the trial included the proportion of women continuing breastfeeding beyond 28 days postpartum; total duration of breastfeeding; women's satisfaction with the care received by themselves and their infants; maternal and neonatal safety; and health care and societal costs. All analyses and comparisons were performed on an intention-to-treat basis. The trial had been approved by the relevant research ethics committees. Further details of the design and conduct of the trial are reported in the accompanying clinical paper.21
The appropriate form of economic evaluation was determined by the clinical and psychosocial results of the randomised controlled trial.21 Although provision was made in the study for conducting alternative forms of economic evaluation, the appropriate form is a cost minimisation analysis,22 given that home-based care did not have a statistically significant effect on any of the pre-specified clinical or psychosocial outcomes.22 Moreover, the trial had been sized to detect a difference of one-fourth of the standard deviation of cost estimates.21,23
The economic evaluation was conducted from a societal perspective and covered the cost of hospital and community health and social services, the costs borne by women themselves and their informal carers, as well as the costs of lost production. The time horizon for the economic evaluation covered the period between discharge from the delivery suite and 28 days postpartum.
Data were collected about all significant resource inputs attributable to the mother and infant during the study period. Resource inputs were divided into six main categories: (1) postnatal care, (2) hospital readmissions, (3) hospital outpatient care, (4) community health and social care, (5) direct non-medical resource inputs and (6) absences from work.
All resource inputs attributable to the mother's and infant's hospital inpatient care were collected using a computerised hospital information system developed by the HUG.24 The information system combined top-down and bottom-up accounting methods, which had been previously validated in the fields of cardiology and hepatology. Medical and nursing staffing inputs, the use of equipment and their associated revenue and capital overheads were estimated for each inpatient ward using simultaneous equation allocation methods. These resource inputs were time-dependent and declined on each successive day of the inpatient stay, reflecting declining staff contact with women and their infants. These resource inputs were additionally dependent upon the number of occupied bed-days in each inpatient ward, as occupancy rates affected the level of support that medical and nursing staff could provide to individual patients. Estimates of each of these resource inputs were then allocated to each study subject according to the timing and duration of their inpatient stay at each level of care (postnatal, intensive, intermediate, general, operative and post-operative). In addition, the computerised hospital information system provided a record of the number and type of laboratory and radiological procedures performed, and the type and dose of medications administered, following each contact with the study subjects. A similar combination of top-down and bottom-up accounting methods was used to estimate the resource inputs attributable to each subject's use of hospital outpatient care.
Postal questionnaires completed by the women at 7 and 28 days postpartum recorded the number, type and duration of community health and social care contacts made by themselves and their infants during the period following initial hospital discharge. These questionnaires also recorded the direct non-medical resource inputs attributable to the health care process, for example, child care support of siblings and distances travelled to health care providers, as well as any time that the woman's partner had to take off work during or following the birth. Telephone contact was made by the research assistant with each woman on the nearest working day to 7 and 28 days postpartum. This provided an opportunity to resolve any misunderstanding about the questions and to remind women to return their questionnaires.
All resource use data were entered directly from the research instruments into a purpose built data collection programme with in-built safeguards against inconsistent entries.
Unit costs for resources used by trial participants were obtained from a variety of sources. All unit costs employed followed recent guidelines on costing resource inputs as part of economic evaluation.22,25 They were expressed in Swiss francs and valued at 2000 prices.
The unit costs of hospital-based resource inputs were calculated from first principles using established accounting methods26 and then fed into the computerised hospital information system. These unit costs incorporated short-run current average revenue costs, plus revenue and capital overheads.
The unit costs of community health and social services were calculated by contacting the relevant community providers and obtaining their financial accounts for the calendar year 2000. Total costs of service provision including staff salaries, employer on-costs and the cost of travel, training, administration and revenue and capital overheads were divided by the annual number of client contact hours by each category of staff. This allowed us to estimate a productive cost per client contact hour that incorporated indirect activities, such as travelling and paper work, for each area of community service provision.
Travel by women and their informal carers to health care providers was valued using the average cost of public transport in Geneva during the year 2000. All other direct non-medical resource inputs were valued using the costs provided by the women themselves in the 28-day postpartum questionnaires. Work absences by the woman's partner were valued using gender-specific median salaries in the canton of Geneva.
Unit costs were combined with resource volumes to obtain a net cost per trial participant during the study period. The primary analysis was of total costs, but results are also given by individual resource use and cost components and by cost category. All results are reported as mean values with standard deviations, and mean differences in costs with 95% confidence intervals (CIs) where applicable. We tested for differences between the trial groups using Student's t tests, assuming equality of variances, and considered those differences significant if two-tailed P values were 0.05 or less. As the data for costs were skewed, we used bootstrap estimation to derive 95% CIs of mean cost differences between the groups.27 Each of these confidence intervals was calculated using 1000 bias-corrected bootstrap replications. All analyses were performed with a microcomputer using the Statistical Package for the Social Sciences (SPSS) (version 11.5; SPSS, Chicago, Illinois) and SAS (SAS Institute, Cary, North Carolina, USA) software.
A series of multiway sensitivity analyses was undertaken to explore the implications of alternative study assumptions.28 Changes in five key variables were considered and the resulting effects on the mean cost differences between the trial groups were estimated.
- (a)Staff costs: The mean level of medical and nursing staffing support for each inpatient ward and outpatient department was reduced and increased by 25% in order to reflect the variations in the level of staffing inputs that might be observed in routine practice.
- (b)Occupied bed-days: The postnatal care costs of each mother–infant dyad were reduced and increased by 20% in order to reflect variations in the monthly number of occupied bed-days during the course of the study.
- (c)Community service utilisation: Three alternative scenarios of community service utilisation were tested in response to a tendency, on the part of participants in health economic studies, to under-report numbers of community service contacts.29 In scenario 1, community service utilisation by the mother–infant dyads was assumed to be 10% greater than reported by the women. In scenario 2, community service utilisation by the mother–infant dyads was assumed to be 20% greater than reported by the women, while in scenario 3, community service utilisation was assumed to be 30% greater.
- (d)Home midwifery support: The number of home visits was varied to the lower and upper 95% confidence limits around the mean of each trial group to reflect variation in the level of domiciliary support that one might expect in routine practice.
- (e)Productivity losses: The economic value of each day of work absence by the woman's partner was varied to the 25th and 75th centiles of the male Genevese income distribution in order to capture the impact that the socio-demographic profile of the wider population might have on the study results.
Resource use, clinical effectiveness and psychosocial data were collected for 459 women who were randomly allocated to either home-based care (n= 228) or hospital-based care (n= 231). The clinical and psychosocial results of the trial are presented and discussed in detail in the main clinical paper.21 In brief, there were no significant differences between the groups with respect to parity, smoking status, years of education, marital status, living arrangements, family income, work status of partner and mode of delivery at the time of random assignment. There were no statistically significant differences between the groups with respect to any of the pre-specified clinical or psychosocial outcomes, highlighting the need for a full assessment of costs within an economic evaluation framework.
Table 1 shows the use of resources by the trial groups and the unit costs of each resource item. The home-based care policy was successful in reducing the mean duration of the mother's hospital stay by 41 hours (P < 0.001) and in increasing the mean number of home midwifery visits by 2.9 visits (P < 0.001). The additional 2.1 km travelled by women allocated to the home-based care group to health care providers reached statistical significance (P < 0.001). Otherwise, there were no statistically significant differences between the trial groups in terms of hospital readmissions made by the mother or infant, their utilisation of hospital outpatient care and community care services, other direct non-medical resource inputs and time taken off work by the woman's partner.
Table 1. Resource use (means [SD] unless otherwise indicated) and unit costs (Swiss francs, 2000 prices) of resource items.
|Maternal postnatal stay, hours||106 ||65 ||Variation by warda|
|Infant postnatal stay, hours||109 ||68 ||Variation by warda|
|Maternal readmissions/emergency visits, n (%)||4 (1.7)||4 (1.8)||Variation by warda|
|Duration, hours||0.6 (5.0)||1.3 (15)||Variation by warda|
|Infant readmissions/emergency visits, n (%)||18 (7.8)||20 (8.8)||Variation by warda|
|Duration, hours||3.5 (17)||3.6 (13)||Variation by warda|
|Paediatrician contacts||1.0 [1.0]||1.0 [0.8]||79.0 per contacta|
|Gynaecologist contacts||0.2 [0.5]||0.2 [0.5]||121.4 per contacta|
|Midwifery contacts||1.8 [2.1]||4.7 [2.3]||103.4 per contactb|
|Physiotherapist contacts||0.1 [0.8]||0.0 [0.1]||79.0 per contacta|
|Community nurse contacts||0.1 [0.5]||0.1 [0.6]||55.0 per contact hourc|
|Other community medical care contacts||0.2 [0.6]||0.1 [0.4]||Variation by carerd|
|Family/home help contacts||0.8 [6.6]||0.4 [1.5]||37.5 per contact houre,f|
|Other community social care contacts||0.1 [0.5]||0.0 [0.3]||Variation by carerd|
|Babysitter contacts||0.2 [1.6]||0.0 [0.3]||15.0 per hourd|
|Distance travelled, km||9.2 [5.1]||11.3 [6.9]||1.0 per kmd|
|Time taken off work, days||7.2 [6.2]||6.8 [5.4]||240.1 per dayg|
Table 2 presents the mean cost per mother–infant dyad through the duration of the study according to category of cost and trial group. A policy of home-based care reduced postnatal care costs by an average of 1554 francs (bootstrap mean difference 1524; P < 0.001) compared with hospital-based care. Most of this cost difference can be explained by the differences between the two groups in the duration of maternal and infant postnatal hospital stays, as indicated in Table 1. The increased use of midwifery services by women allocated to home-based care contributed to a significant difference in community care costs of 294 francs (bootstrap mean difference 295; P < 0.001). There were no significant differences in average hospital readmission, hospital outpatient care, direct non-medical and indirect costs between the two trial groups. The net societal cost per mother–infant dyad in the two groups were 9019 francs in the hospital-based care group and 7798 francs in the home-based care group, representing a societal cost saving per mother–infant dyad of 1221 francs (bootstrap mean difference 1209; P= 0.017).
Table 2. Mean costs [SD] and mean cost differences by cost category (Swiss francs, 2000 prices).
|Postnatal hospital care||6772 ||5218 ||1554||0.001||1524 (675 to 2403)|
|Hospital readmissions||136 ||301 ||−166||0.158||−169 (−424 to 46)|
|Hospital outpatient care||111 ||103 ||8||0.397||8 (−9 to 26)|
|Community care||234 ||528 ||−294||<0.001||−295 (−245 to −343)|
|Direct non-medical costs||20 ||13 ||7||0.353||7 (−4 to 25)|
|Total direct costs||7273 ||6164 ||1110||0.024||1130 (151 to 2020)|
|Indirect costs||1746 ||1635 ||111||0.422||112 (−161 to 404)|
|Total costs||9019 ||7798 ||1221||0.017||1209 (202 to 2155)|
We performed sensitivity analyses on key economic parameters to determine the impact that plausible variation of those parameters might have on total mother–infant dyad costs and cost differences between the trial groups (Table 3). A 25% reduction and increase in the mean level of medical and nursing staffing levels for each inpatient ward and outpatient department had the effect of reducing and increasing the cost difference between the trial groups by 369 francs (bootstrap values 362, 365). Similarly, a 20% reduction and increase in postnatal care costs, reflecting variations in bed occupancy, had the effect of reducing and increasing the cost difference between the trial groups by 311 francs (bootstrap values 303, 307). Assuming that community service utilisation by the mother–infant dyads was greater than reported by the women had the effect of reducing the cost difference between the trial groups, although the cost difference remained statistically significant. The cost difference fell by 30 francs (bootstrap value 11) when community service utilisation was assumed to be 10% greater than reported by the women. The cost difference fell by 59 francs (bootstrap value 48) when community service utilisation was assumed to be 20% greater than reported by the women and by 89 francs (bootstrap value 77) when community service utilisation was assumed to be 30% greater. Finally, variations in the level of home midwifery support and the income level at which work absences were valued had a marginal effect on the study results. Simultaneous variation of the key economic parameters did not significantly affect the baseline study results.
Table 3. Sensitivity analysis of mean costs [SD] and mean cost differences (Swiss francs, 2000 prices).
|Baseline analysis||9019 ||7798 ||1221||0.017||1209 (202 to 2155)|
|(a) Staff costs:|
| 1) 25% reduction in staff inputs||7501 ||6649 ||851||<0.001||847 (444 to 1233)|
| 2) 25% increase in staff inputs||10,537 ||8947 ||1590||<0.001||1574 (1182 to 1950)|
|(b) Occupied bed-days:|
| 1) 20% reduction in postnatal care costs||7664 ||6755 ||910||0.034||906 (118 to 1761)|
| 2) 20% increase in postnatal care costs||10,373 ||8842 ||1531||0.010||1516 (315 to 2697)|
|(c) Community service utilisation:|
| 1) 10% greater than reported||9042 ||7851 ||1191||0.020||1198 (195 to 2176)|
| 2) 20% greater than reported||9066 ||7904 ||1162||0.023||1161 (107 to 2160)|
| 3) 30% greater than reported||9089 ||7957 ||1132||0.027||1132 (113 to 2082)|
|(d) Home midwifery support:|
| 1) Lower point of 95% CI||8991 ||7767 ||1223||0.018||1227 (153 to 2234)|
| 2) Upper point of 95% CI||9047 ||7829 ||1218||0.018||1207 (244 to 2167)|
|(e) Productivity losses:|
| 1) 25th centile of income distribution||8146 ||6981 ||1165||0.019||1166 (124 to 2127)|
| 2) 75th centile of income distribution||9892 ||8616 ||1276||0.017||1311 (281 to 2314)|
This study demonstrated that, for women delivering a single infant at term following an uncomplicated delivery, a policy of early postnatal discharge combined with home midwifery support is significantly less costly than traditional postnatal care, with average savings to Swiss society of over 1200 francs per mother–infant dyad. Moreover, this conclusion remained relatively robust following variations in the values of key economic parameters performed as part of a comprehensive sensitivity analysis. When viewed in conjunction with the clinical and psychosocial results from the trial,21 the economic evaluation provides rigorous evidence in favour of a policy of early postnatal discharge combined with home midwifery support. Indeed, the policy exhibits weak economic ‘dominance’ over traditional postnatal care, that is, it significantly reduces costs without compromising the health and wellbeing of the mother and infant.23
The strengths of the trial upon which this economic evaluation was based are that it was randomised and controlled, included a relatively large sample, avoided many of the selection biases that characterised earlier studies2–9 and provided a comprehensive set of resource use and clinical effectiveness data. These provide a reliable basis for estimating the economic efficiency of a policy of early postnatal discharge combined with home midwifery support. The study cost accounting was comprehensive and included all significant cost items, whose values were calculated according to established principles in economic theory.
The sample size calculations for the trial was informed by what the investigators considered to be a meaningful and achievable reduction in costs, at conventional levels of statistical significance.21 The sample size that was subsequently achieved was sufficiently large to detect significant differences in health care and societal costs between the home-based and hospital-based care groups. Therefore, we would argue that the economic evaluation was sufficiently sized to arrive at conclusions that are both meaningful and relevant to decision-makers.
Readers and decision-makers considering the policy implications of the study results should bear in mind a number of limitations. First, the generalisability of the study findings are constrained by differences in postnatal discharge policies across industrialised nations, as well as differences in the relative price structures of resource inputs across health care systems. Geneva is a prosperous city, characterised by well-resourced maternity services and relatively short distances that women have to travel to care providers. However, the average postnatal stay experienced by Genevese women mirrors that of many industrialised nations.30 This suggests that the geographic generalisability of our results may be extensive, although country-specific estimates of unit costs are required to quantify likely cost savings in local settings. In contrast, the average postnatal stay in some industrialised nations, notably the United Kingdom and the United States, already mirrors that experienced by women allocated to our intervention group, and the cost savings that can be achieved by earlier postnatal discharge in these nations are likely to be limited. The study does, however, provide a barometer for the likely economic changes that can be anticipated as a result of changes in postnatal care that might follow clinical, organisational or legislative pressures.
A second limitation of the study is the relatively limited time frame of the economic evaluation. Although the economic evaluation involved a rigorous assessment of the costs and benefits of home-based care during the first 28 days postpartum, it is possible that the policy has longer term consequences in terms of health status and health service utilisation over the mother's and infant's lifetime. Indeed, preliminary evidence from the six-month follow up of trial participants suggests that infants of mothers randomised to home-based care may be more likely to be readmitted to hospital during the first six months postpartum.21 An additional analysis that included the costs of these subsequent hospital readmissions within the economic evaluation did not significantly alter the study results. Nevertheless, the relative cost effectiveness of a policy of early postnatal discharge combined with home midwifery support is likely to depend, in part, upon the time frame of interest.
A third limitation of the study arises from the absence of a broad economic approach to the valuation of outcomes of the alternative policies. It is possible that women valued a number of attributes of postnatal care, such as the location of care and the provision of information, reassurance and confidence building, independently of their impact upon the trial outcomes. If this were the case, then a broader economic approach to the measurement of outcomes, such as contingent valuation31 or stated preference discrete choice modelling,32 might have provided more subtle information to decision-makers.
If it is accepted that a policy of early postnatal discharge combined with home midwifery support is cost effective, then decision-makers must consider whether the study results can be generalised to their own settings and must decide how to organise postnatal services. The opportunity costs of the freed postnatal beds that would follow increased implementation of early postnatal discharge will depend upon their alternative uses. In the long term, it is conceivable that the freed postnatal beds will be filled by new obstetric patients, the care of whom might increase overall hospital expenditure. Alternatively, the freed postnatal beds might be closed, releasing staff and support services for other areas of obstetric care, although the value of any financial savings that might ensue must be counterbalanced by the fixed costs attributable to postnatal wards. Similarly, decision-makers must consider the implications that a shift in practice would have in terms of increased pressure placed on community midwifery services. Increased use of these services might make more efficient use of existing professionals. Alternatively, the demand for community midwifery services might increase to such a degree that new teams are required, the opportunity costs of which might include a whole range of managerial, training and support costs.
In conclusion, our study provides rigorous evidence of the economic advantages of a policy of early postnatal discharge combined with home midwifery support, both from a health service and wider societal perspective. It is incumbent upon decision-makers to decide whether the study results are applicable to their own settings and, if so, what appropriate configuration of postnatal services should be implemented.
The authors would like to thank colleagues at the National Perinatal Epidemiology Unit, University of Oxford, and Geneva University Hospitals who have commented on successive drafts and have, as always, given helpful advice. The study received financial support from the Swiss Fonds National de la Recherche Scientifique (FNRS) (ref: 32-52954.97) and from the Quality of Care Programme, Geneva University Hospitals. The National Perinatal Epidemiology Unit is core funded by the Department of Health, England. The views expressed by the authors do not necessarily reflect those of the funding bodies.