A cost analysis of early discharge and domiciliary visits versus standard hospital care for low-risk obstetric clients Anthony Scott Centre for Health Economics Research and Evaluation, Westmead Hospital, Sydney

Abstract: This paper presents an economic analysis o f three obstetric early discharge schemes operating in western Sydney. The primary objective was t o examine the net econoniic value of health service resources released due to each early discharge scheme. Conventional postnatal obstetric care comprised an inpatient stay of four to five days. Early discharge provided an alternative location. Clients who chose to participate in the scheme were discharged t w o t o three days earlier than usual and visited at home by a midwife for up t o seven days. Other studies of early discharge schemes have shown that resulting maternal and infant morbidity were no higher than with conventional care. This study t o o k the form of a cost analysis. The early discharge schemes at Blacktown and Westmead Hospitals used more resources than they released, whereas the scheme at Auburn Hospital released resources in excess of those it used. Sensitivity analysis showed that the results for Westmead and Auburn Hospitals were sensitive to changes in key assumptions used in the analysis. The number o f domiciliary visits per client was an important determinant of the ability of schemes to release resources in excess of their costs. The existence of early discharge provides an extra choice to clients in the location of postnatal care. The question for decision makers is whether this choice is worth the extra cost of early discharge schemes. ( A w f J Public Healfh 1994; 18: 96-100)

iven the increasing demands on health-care systems, improving the efficiency of health G care provision has become a major preoccu-

pation o f the governments of most developed countries. I Australia is no exception. The Medicare Incentive Package was an initiative of the Australian Commonwealth Government in 1988-89. The pack- age provided funding for state health departments to establish schemes to improve the efficiency of the public hospital system in selected areas. A commit- ment to the evaluation of such schemes was a con- dition of funding.

Obstetric early discharge was established by the Western Sydney Area Health Service. It was intended t o reduce the use of obstetric beds by reducing postnatal length of stay, and was expected to release resources which may have had more beneficial alternative uses. The early-discharge schemes oper- ated at three hospitals. Westmead Hospital is a large teaching hospital and Blacktown and Auburn are smaller district hospitals. Mothers without medical complications who chose early discharge were sent home within three days of the baby’s birth. Postnatal care at home for the first week after the birth was provided by a Domiciliary Midwifery Program (DMP) based at each hospital. Women without medical com- plications who were discharged after a conventional postnatal stay received no home visits. Women book- ing into hospital were asked to indicate their preference for early discharge or a hospital postnatal stay, with the option of changing their decision after the birth.

Previous evaluations o f obstetric early discharge have found no differences in maternal and infant morbidity. Yanover et al. found no increased mor- bidity among early-discharge clients where they were visited at home by nurse practitioners.2 Norr et al. studied the impact on low-income mothers and their infants of early discharge with home visiting, which

Correspondence to Anthony Scott. CHERE, Department of Com- munity Medicine. Westmead Hospital. Westmead, NSW 2145. Fax (02) 689 1049.

was compelled by bed shortages.:+ There was no dif- ference between the study group and the two control groups in terms of maternal and infant morbidity. The women who were discharged early scored highest on maternal-infant attachment assessment, had the lowest number of concerns, and were more satisfied. The impact of early discharge without home support on outcomes for primiparous women and their infants was studied by Lemmer who examined maternal and neonatal morbidity and postpartum concerns of mothers.” The results demonstrated no difference in maternal concerns between the groups and the only difference in physical factors was more infants with jaundice in the early discharge group, which may have been overcome by home visiting. Burnell et al., in a study of early discharge with home visiting, measured maternal and infant physical prob- lems, feeding problems, mother’s attitude and emo- tional state..s The study concluded that no harm resulted from early discharge and that there were no benefits from a longer postnatal hospital stay.

The early discharge scheme at Westmead Hospital had been operating since 1984 and was evaluated during its first year of operation.6 Maternal and infant morbidity and postnatal adjustment were measured. N o increased maternal or infant mor- bidity was found for those women who chose early discharge. The General Health Questionnaire was used to detect depressive illness, with fewer mothers in the early-discharge group having depressive illness than their hospital counterparts.’ The safety of early- discharge programs has therefore been adequately demonstrated in these and other studies, provided there is home follow-up by a midwife. However, there have been no published economic analyses of obstetric early discharge schemes.

This study took the form of a cost analysis.x The objective was to ascertain, from the point of view of the hospital and the area health service, the economic cost of two different regimens of postnatal care for women eligible for early discharge. These were: con- ventional hospital postnatal care; and a combination of shorter stay hospital care and domiciliary postnatal

96 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 voi 18 NO 1

COST OF OBSTETRIC EARLY DISCHARGE

care. Costs to patients and their fanlilies and patient satisfaction were also measured but are not reported here because of lack of spare.q

Methods Two groups o f obstetric clients were compared at each o f the hospitals. The progrum group at each hos- pital comprised women, screened as medically eli- gible, who chose early postnatal discharge. The hospikzl group comprised women screened as medi- cally eligible for early discharge during I.he same period, but who did not choose early discharge.

Screening was based on medical criteria which defined those women who experienced an uncomplicated pregnancy and childbirth as those eli- gible for early discharge. Screening was conducted prospectively at each hospital on one or two days per week between 2 1 May and 23 July 199 1 . For the pro- &Tam group, screening was conducted ;is usual by midwives on each DMP. For the hospital group, screening was conducted by the area coordinator of the DMP.

Data were collected by self-administered question- naire. The number of questionnaires distributed at each hospital reflected the relative sizes of their respective domiciliary midwifery programs and obstetric units. Program and hospital clients at Auburn hospital received 40 and 5 1 questionnaires respectively, while those at Blacktown received 60 and 72, anti at Westmead 50 and 77. The women in the program group were given the questionnaire on their next-to-last home visit by the domiciliary mid- wife and were given the option of returning it with the midwife on her last visit or posting it. The women in the hospital group were asked t o complete the questionnaire before they left hospital.

The questionnaire obtained information on the length of postnatal stay for both groups at each hos- pital and the number of domiciliary visits received by clients in the program groups. Information on these variables was not routinely available from hospital records at the time the study was conducted. Demo- graphic characteristics were compared for survey participants in the program and hospital groups using chi-square and Fisher’s exact tests of statistical significance. Socioeconomic status for each group was measured using the Index of Relative Socio- economic Disadvantage and the Index o f Education and Occupation, available from the Australian Bureau of Statistics. l o These indices were derived from a combination of socioeconomic variables for postcodes, collected in the 1986 census. Thus, the indices refer to the aggregate socioeconomic status of the client’s postcode of residence.

Hospital resource w e Nursing costs: Nursing staff levels for each postnatal ward were obtained from the nursing administration at each hospital. The DMPs at Auburn and Westmead Hospitals both used nursing staff from the postnatal wards to cover for annual leave and sick leave and so the staff establishment of each postnatal ward in these hospitals was reduced accordingly. Staff inputs were valued at industrial award rates (for 1989-90) adjusted for oncosts of superannuation (3 per cent), paid holiday (2 per cent) and penalty rates (18 per

cent). The annual ward nursing cost was divided by the annual number of occupied bed-days on the ward to obtain an average nursing cost per bed-day. This figure was then weighted according to the dependency of women elighle for early discharge, relative to the average dependency o f clients on the ward. Dependency weights were obtained from the ward dependency system which is based on the approximate numher o f nursing staff hours needed to care for each type o f patient for each day they are in hospital. These weights declined on each success- ive postnatal day, reflecting the declining staff input. The weighted average nursing cost per day t turefore declined as length of stay increased. Non-nursing costs: Non-nursing costs included: ward costs attributed directly to the ward, such as goods and services, ward clerk, ward space and equipment; and costs attributed to other hospital departments, such as client meals, domestic services, linen services, adniinistrdtion, medical records and power. The costs of goods and services and ward clerk were obtained from the ward. Ward space and equipment were valued by calculating an annual equivalent cost of their replacement value.x Costs originating from other hospital departments were measured b y allocating a proportion of each department’s annual total cost to the postnatal ward. The annual total cost o f the department was niulti- plied by the ward’s annual utilisation of the depart- ment, expressed as a proportion of total hospital utilisation of the department. In cases where ward utilisation was not measurable, costs were allocated pro rata to occupied bed-days. Data on the ward’s usage of each department were obtained from the relevant department for the financial year 1989-90. The allocation process provided an annual ward cost for the ward’s use of each hospital department. Aver- age non-nursing cost per bed-day was then found by dividing the sum of annual ward and non-ward costs by the annual number of postnatal ward occupied bed-days. Using the average assumed that all clients on the ward used the same proportion of the services of each hospital departnlent each day of their stay.

L)omic.ilinry midwifery program resource use The total annual cost comprised each program’s share of area health service support (program coordi- nator, car, c-omputers and office accommodation), office space used by midwives, staff, cars, travel, equipment, and goods and services. Annual equiv- alent costs were calculated for office space, cars and equipment. Program-activity statistics provided data on travelling costs. Annual goods and services expen- diture was obt.ained froni the nursing unit managers of each program. Given that each DMP caters for other client types (for example, caesarean section clients), the total annual cost was multiplied by the proportion of clients who were eligible for early dis- charge, to give a total annual cost for this group of clients. This was the figure used in subsequent calcu- lations. For each program, the average cost per domi- ciliary visit was obtained by dividing the total annual cost of the program by the total number of postnatal visits (based on current activity). The DMP cost per client was calculated by multiplying the average cost per visit at current activity levels by the average num- ber of visits per DMP client.

AUSTRALlAN JOURNAL Of PUBLIC HEALTH 1994 v o t . 18 NO. 1 97

SCOTT

Cost per episode of postnatal raw For women in the hospital group, the cost per epi- sode of postnatal care comprised hospital costs only. These were calculated by adding together the total cost of each successive day they spent in hospital (where total cost per day equals the weighted average nursing cost plus average non-nursing cost). For women in the program group, the cost per episode of postnatal care comprised hospital and DMP costs. The hospital cost was calculated as for women in the hospital group, and added to the DMP cost per cli- ent. All cost figures were for the financial year 1989-90.

Net resource sauings A program generated net resource savings if the net economic value of resources released was positive. The net economic value o f resources released per episode o f postnatal care was defined as the value of the bed-days released from reductions in length of stay, minus the additional cost of domiciliary visiting. This was calculated by subtracting the cost per epi- sode of postnatal care in the program group from the cost per episode of postnatal care in the hospital group. Annual figures were also calculated. The eco- nomic value of resources released from reductions in length of stay were valued on the assumption that freed bed-days were filled by other patients.

sensitivity analysis Sensitivity analysis was used to test the robustness of the result (in this case the net economic value of resources released) to changes in key assumptions in the analysis and to factors which may have been vari- able over time. For each assumption which was changed, all other assumptions remained the same. Best and worst scenarios were estimated from varvinc

mates for each program were obtained from one standard deviation around the mean. Length of postnatal stay: this measured the number of bed-days released at each hospital. Length o f stay for the hospital group and then the program group was varied using upper and Lower estimates of one standard deviation around the mean.

Results Response rate The response rates in each hospital were between 90 per cent and 91 per cent for the program groups, and between 74 per cent and 80 per cent for the hospital groups. In total, 136 out of 150 clients in the pro- gram group responded (91 per cent) and 153 out of 200 (77 per cent) in the hospital group responded. Clients in the program group completed the ques- tionnaire an average of 5.7 days after delivery com- pared to 3.1 days for the hospital group. More nonrespondents than respondents were having their first baby and were born in a country where the first language was not English. The average age of nonrespondents was not significantly different from that of respondents.

Demographic chrurteristics More program women than those in the hospital group were socioeconomically disadvantaged, but the difference was not statistically significant. There was no significant difference between the two groups for language spoken, country of birth or mean age. Parity was the only characteristic where the differ- ence between the two groups was statistically signifi- cant (x' = 11.48, P = 0.003). The hospital group contained a higher proportion of women who were having their first baby.

i c l

each assumption. The robustness of the results of the study was tested using the following assumptions:

Ward staffing: the actual establishment varied from month to month at each hospital, The mean estab-

,sewice resource use

gives estimates Of length bed-daYs released and the number of DMP visits per client in

~.

a

lishment (in full-time equivilents) was used as the baseline with one standard deviation as upper and lower bounds. Occupied bed-days: the annual number of occu- pied bed-days for the postnatal ward varied from year to year in line with the local birth rate. Esti- mates were k 1 0 per cent of actual bed-days for 1989-90. This was a proxy for the technical efficiency of the ward, that is, as annual bed-days increased then average cost per bed-day fell (assuming total ward costs remained the same). DMP activity (postnatal visits per month): this was highly variable from month to month and affected the technical efficiency of each DMP, that is, if DMP activity rose then average cost per domiciliary visit was reduced (assuming total DMP costs remained the same). Upper and lower estimates of activity were taken from the highest and lowest actual monthly activity of each program at current staffing levels. Home visits per client: the average number of visits per DMP client was not set by clinical precedent or on evidence of health benefit but was discretionary and determined by workload. If the DMP was busy then visits per client fell. Upper and lower esti-

both program and hospital groups. The total cost per bed-day for each hospital and the average cost per DMP visit are given in Table 2. Table 3 gives costs per episode of postnatal care, for both hospital and pro- gram groups. This shows that each scheme releases resources due to reductions in length of stay, but once the extra cost of the DMP has been taken into account then the early-discharge schemes at Westmead and Blacktown Hospitals cost more than the value of resources released. The scheme at Auburn Hospital released resources valued more highly than its costs, but the difference was small and the result was sensitive to some of the assumptions used in the analysis. Table 4 gives annual figures of the value of resources released, the total cost of each DMP, and the net economic value of resources released.

Sensitivity analysis The results of the sensitivity analysis are shown in Table 5. The result for Auburn Hospital was extremely sensitive to all of the assumptions used in the analysis. Although the study results show an annual saving of $3072, the robustness of thi$ result is limited. The result for Westmead Hospital was also

98 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 VOL 18 NO 1

COST OF OBSTETRIC EARLY DISCHARGE

Table 1 : Length of stay, bed-days released and visits per client for mothers having postnatal care in hospital and

those in the Domiciliary Midwifery Progrum

Auburn Westmeod Blocktown

Hospitol group

Domiciliary Midwifery Progrorn group Average length of postnatal stoy 2.06 1.71 1.83

Bed-days releosed per episode 2.40 2.30 2.13 Annual bed-dovs released 1178" 1273b 1870'

Average length of postnotol stoy 4.46 4.01 3.95

Average number of visits by midwife 4.28 5.1 1 5.98

Table 2: Cost per bed-day and cost per visit by the domiciliary midwife

Auburn Westmead Blocktown __ Totot cost per bed-dayb 1 doy after delivery 144 170 129 2 days after delivery 136 155 118 3 days after delivery 136 155 118 4 or more days oher delivery 128 141 107

Average cost per visit by dorniciliory midwife 72 74 71

Nofees: Notes: (a) based on overage level of activity January-June 1991 (491 program

(b) based on avemge Iewl of adivity, Januaty-June 1991 (553 program

(c) based on awmge law1 of activity, June-October 199'1 (678 progmm

(a) These figures haw been rounded to the nearest dollar. (b) Total cost per day equals weighted average nuning cost per day

clients annually)

clients annually)

clients annually)

(which declines on each successiw day) plus average non-nursing cost per day (which i s constant).

Table 3: Economic w l u e of resources released per episode of postnatal care ($)

Auburn Westmead Blocktown Hospital Program Total Hospital Program Totol Hospital Program Totol

cost cost cost cost cost cost cost cost cost

Cost per episode of postnatal care

Progrom group 307 308 615 312 379 691 238 423 661 Hospital group 622 0 622 654 0 654 479 0 479

Economic value of resources releosed 314 -308 6 342 -379 -36 240 -423 -183

Note: (a) Rounded to the nearest dollar

Table 4: Economic w l u e of resources released per year ($) due to reductions in length of stay

Auburn Westmeod Blocktown

Economic volue of resources releosed 154 291 189 294 211 075

Totol program cost 151 219 209490 211 075 Net economic value of

resources releosed per year 3 072 -20 196 -1 60 477

sensitive to some of the assumptions of the study. This result is robust with respect to changes in ward staff and the technical efficiency of the DMP (that is, postnatal visits per month). For Blacktown DMP, no matter which assumption was varied, the program cost more than the value of resources it released.

Discussion The cost of the DMP at Blacktown and Westmead Hospitals exceeded the value of resources released due to reductions in length of stay. The scheme at Auburn Hospital released resources valued more highly than its costs, but the difference was small and sensitive to the assumptions used in the analysis. The results of the evaluations in all three hospitals indi-

cated that the ability of early discharge (combined with the current level of domiciliary visits) to gener- ate net resource savings was limited.

The most important factor determining whether the net economic value of resources released at the three hospitals was positive or negative was the num- ber of domiciliary visits per client. These were dis- cretionary in nature and fluctuated in line with DMP activity. There seems little information on what should be the optimal number of postnatal visits. This will depend on the outcome of each visit compared to its costs. If additional visits add nothing to the wel- fare of women or their babies, then the number of visits per client could be reduced without loss of effectiveness. I f reductions in the number of visits per client were accompanied by an increased uptake of clients or a reduction in staffing, then the ability of each scheme to release resources in excess of its costs would be improved.

The method used to assign a value to the resources released by early discharge requires clarification. Opportunity cost in economics refers to the value of the benefits which are given up if resources are used one way rather than another. Thus the opportunity cost of the freed resource depends on the alternative

Table 5: Sensitivity analysis (in $1

Assumptions

Ward staff Occupied bed-days Domiciliary mldwife activity:

postnatol visits per month Home visik per client Length of rtoy: hospital group Length of stay: program group

-

Auburn Westmeod Blocktown Best Study result Worst Best Study result Worst - Best Study result Worst

3960 3072 -1320 -14176 -20196 -31025 -151470 -160477 -169485 14 233 3 072 -5 365 279 -20 196 -36554 -142254 -160477 -175391

20263 3072 -10935 -3828 -20 196 -38990 -140701 -160477 -175391 128 094 3 072 -44 745 91 284 -20 196 -71 370 -45 639 -160 477 -215 478 88434 3072 -82290 31 462 -20 196 -71 370 -45639 -160477 -215478 42 972 3 072 -36 828 53 147 -20 196 -92 526 -83 005 -160 477 -237 950

~~~~~ ~ ~ ~____ ___ ~

AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 VOL 18 NO. 1

~

99

SCOT1

use to which it would otherwise be put. For the pur- poses of this study, it was assumed that freed bed- days would be used to admit other patients. The total cost per bed-day was used as a proxy for the econ- omic cost of each bed-day released on the assump- tion that this best reflected its value to patients newly admitted. This assumption seems reasonable, given constraints on obstetric bed supply owing to the growing population in western Sydney coupled with a relatively fixed number of obstetric beds.”

The estimate of the economic value of resources released is not equivalent to savings in hospital expenditure. If the bed-days are filled by new patients then hospital expenditure will rise, firstly because the early discharge scheme is being financed in addition to the bed-days, and secondly, because the increased throughput of patients means that more expensive bed-days are substituted for less expensive bed-days (that is, those at the end of the period of postnatal care) and would involve additional antenatal clinics, labour ward and obstetric theatre costs. Alternatively, if the freed beds are closed, the existence of fixed costs will ensure that the value of any financial savings realised is less than the economic value of resources released.

Several aspects concerned with study design should be noted. The first relates to differences in length of stay. The two groups of women were self- selected into hospital and program groups. Self- selection was imposed on the study design as client choice was an integral part of the scheme. Conse- quently, the hospital group contained a larger pro- portion of first-time mothers than the program group. This introduced a bias which inflated the aver- age length of stay of the hospital group relative to the program group and so exaggerated any reduction in length of stay achieved by the scheme. Taking this into account through direct adjustment of length of stay for parity, the reduction in length of stay across all three hospitals was 2.22 days, compared to 2.25 days (unadjusted). This was not expected to alter sig- nificantly the results of the analysis.

More generally, lengths of stay are falling for reasons independent of early-discharge schemes. The potential for such schemes to generate reductions in hospital stays (over and above those that would occur anyway) will be increasingly restricted as lengths of stay continue to fal1.l2

The second aspect of study design of concern was that screening of patients into low-risk groups was carried out by the DMP coordinator for the hospital group and by DMP midwives for the program group. Although the criteria used to judge eligibility for early discharge were the same for both groups, they still may have interpreted the criteria differently. The direction of this bias is, however, difficult to predict.

The literature demonstrated that the effectiveness of such schemes is not likely to be different from con- ventional care, provided clients are followed up at home. However, effects on patient welfare may encompass more than just effects on health status. One benefit of the early-discharge scheme was the increased choice it provided women over the location and form of postnatal care. Other dimensions of obstetric care which patients may value indepen-

dently of health status, such as the ability to choose their location of delivery, the provision of infor- mation, reassurance and confidence-building, have not been considered in the literature.

The results of the study should be interpreted with these issues in mind. When early-discharge schemes cost more than the value of resources they release, this does not mean that they should be discontinued. A value judgment must be made by decision makers whether the increased choice available to women is worth the additional cost.

Conclusion This study has shown that early discharge schemes may not always generate net resource savings. More attention should be devoted to assessing the many dimensions of benefit generated by such schemes before conclusions about their cost-effectiveness are reached. As the management of pregnancy by nonmedical staff gains credence, alternatives for the care of mothers and infants will grow. These changes must be evaluated thoroughly in order to ensure that the health, satisfaction and involvement in decision making of mothers is maximised, given the available resources.

Acknowledgments The study was funded by a grant from Western Syd- ney Area Health Service. Thanks also go to Sue Psaila and the staff of each program for their assistance in data collection. Thanks go to Alan Shiell, Jane Hall and Stephen Leeder in providing invaluable com- ments on earlier drafts of this article. The assistance of Patsy Kenny, Sue Cameron, Madeleine King and David Newell is also acknowledged for assistance with study design, data collection and analysis.

References 1.

2.

3.

4.

5.

6.

7.

8.

9.

10.

1 1 .

12.

Organisation for Economic Cooperation and Development. Health care system zn transitimr. Paris: OECD, 1990. Yanover MJ, Jones D, Miller MD. Perinatal care of low-risk mothers and infants. N Engl J Med 1976; 294: 702-5. Norr KF, Nacion KW, Abramson R. Early discharge with home follow-up: impacts on low-income mothers and infants. J Obstel Cjnecol Neonatal NUTS 1989; 18: 133-4 1. Lemmer C. Early discharge: outcomes of primiparas and their infants. J Obstet Gynecol Neonatal NUTS 1987; 16:

Burnell I , McCarthy M, Chamberlain GVP, Hawkins DF, et al. Patient preference and postnatal hospital stay. J Obstet (jnoecol 1982; 3: 43-7. James ML, Hudson CN, Gebski VJ. Browne 1.H. et al. An evaluation of planned early postnatal transfer home with nursing support. Med J Aust 1987; 147: 434-8. Nott PN, Cutts S. Validation o f the 30-item General Health Questionnaire in postpartum women. Psycho1 Med 1982; 12: 409-1 3. Drummond MF, Stoddart GL, Torrance GW. Methdsfor the economic eualuation of health care programmes. Oxford: Oxford Medical Publications. 1987. Kenny P, King M. Cameron S. Shiell A. Satisfaction with postnatal care-the choice of home or hospital. Midwijety 1993; 9: 146-53. McLennan W. Socio-economic illdexes for area. Cat. no. 1356.0. Canberra: Australian Bureau o f Statistics, 1990. Shearman R. Matentity smices in New South Wales. Sydney: New South Wales Department of Health. 1989. Clavarino L. Gibberd R. New South Wales hospital separations: 1979-1988/89 and projections for 2001. Newcastle: New South Wales Health Services Research Group, University of Newcastle. 1990.

230-6.

100 AUSTRALIAN JOURNAL OF PUBLIC HEALTH 1994 VOL. 18 NO. 1