REV I EW ARTI C L E

PharmacoEconomin 3 (n 362-373. 1993 11m. 7690/93/00J~~362/S06.00!O e Adi s Internation al limiled. An ri!!.hts reserved

""e1 ;!09

Cost Effectiveness of Surfactant Replacement in Pre term Babies

Miranda Mu.gford and Sarah Howard National Perinatal Epidemiology Unit. Radcliffe Infirmary, Oxford, England

Contents

302 J63 364 J65 367 368 3&' 369 369 369 370 371 371 371

Summary

Summary 1. Methods for Cost-Effectiveness Evaluation 2. Sources of Evidence About Effectiveness and Costs 3. Policies for Surfactant Replacement 4. Effects of Different Policies for Administration of Surfactant 5. Costs of Hospital Care

5.1 Costs to Parents 5.2 Long Term Costs

6. Cost Effectiveness of Different Policies for the Use of Surfactant for Preterm Babies 6.1 Early or Prophylactic Administration of Surfactant 6.2 Surfactant Given as Treatment 6.3 Early Surfactant Versus Later Selective Administration 6.4 Reduction of Effects of RDS - Dosage Requirw

7. Conclusions

Surfactant replacement in preterm babies has been shown in recent years in randomised con­trolled trials to br an efTe<:tive treatment for respiratory distress syndrome (RDS). It is ex.pensive and, because it increases survival, it has implications for the costs of neonatal services. We used evidence about resource use obtained from trials of surfactant and other studies on the economics of surfactant to assess the cost effectiveness of different policies for its use. For the smallest babies, surfactant is likcly to increase overall costs of neonatal care, but also to reduce the rat io of costs to survival. whether surfactant is given prophylactically or as a treatment for established ROS. h is less clear what the optimal policy should be for babies of more than around 3 I weeks' gestation. Comparison of the relative cost effectiveness of policies of early prophylaclic suriactant and surfactant for later Ifeatment of ROS, and of different dosage policies. is currently being conducted in the contex.t of 2 large multicenl re trials. No policy for surfactant use should be considered in isolation from thc availability of effective obstetric interventions which have been shown to reduce the risk of RDS in prcterm babies and which will therefore reduce the need for surfactant.

Cost Effectiveness of Surfactant Replacement in Prelerm Babi~ 363

Neonatal intensive care has been shown to be one of the more costly healthcare program mes in comparison to benefits gained (Torrance & Zipur. sky 1984). The ratio of benefits to costs is associ· ated with birthweight. Higher costs and poorer out­comes arc associated with smaller, less mature babies (Mugford 1988a). The ratio of benefits to costs might be altered by a treatment for preterm babies which improves outcomes while also re­ducing the need for such costly care. In recent years, several interventions have been shown to do this, and one of these is to give exogenous lung surfac­tant to preterm newborn babies.

One of the commonest causes o f death in a neo­natal unit is respiratory distress syndrome (RDS) which occurs when the immature lungs of preterm babies lack the surfactant that is necessary for lung expansion after birth and therefore for breathing. The risk of RDS is inversely related to gestational age: babies born before 32 weeks' gestation are at the highest risk, and the incidence of RDS in this group is over 50% (Halliday el al. 1989). Sinee RDS accounts for a large proportion of deaths of pre­term babies and for a large proportion of the high cost of neonatal care, new treatments that might reduce either the risk of RDS or the severity of the disease and its long term effects in survivors are of particular econom ic interest.

Exogenous surfactants include a variety of formulations, including phospholipids and surfac­tant proteins, that are biochemically similar to hu· man lung surfactant. Surfactant is either manufac· tured artificially or extracted from animal sources. It is administered through the endotracheal tube that is used to intubate the baby for ventilation. Surfactant is an expensive product. In the UK, col­foseeril palmitate (synthetic surfactant) is mar­keted at £314.29 for a 10Smg (Sml) vial. Porcine lung surfactant (Curosurf) is likely to be priced at about £800 for 3ml and £400 for l o5ml. The dosage required varies with a baby's weight and the prod­uct, but is around Sml per 1000g for colfoseeril­palmitate and between 1.25 and 205ml per lOOOg for porcine lung surfactant. The protocol for treat­ment with Exosurf NeonatalTN (colfosceril palmi­tate) requires 2 doses in the first 12 hours. so a

minimum cost of a course of surfactant fo r a baby under 1000g will be around £600 to £700. The pro­tocol from the CU ROSURF4 multicentre random· ised trial requires a minimum I dose and up to 2 further doses, at a cost of£800 to £1200 fo r babies under l OOOg (Holliday & Tarnow-Mordi 1990). Costs for babies over 1000g weight, or those re­quiring longer courses of treatment, would ob­viously be higher.

Health purchasers and providers need evidence that there is sufficient gain in clinical outcome to justify any additional costs that surfactant entails. A considerable body of literature (Soli & McQueen 1992) is emerging on the effectiveness of surfactant in different circumstances, and several surfactant products now have product licences in the US and in European countries.

We discuss which economic questions can be addressed, outline which policies for surfactant use have been considered, review sources of evidence about costs and effectiveness, and then summarise what can be said about the cost effectiveness of giving surfactant to preterm babies.

I. Mdhods for Cost-Effectiv~"~ss Evaluation

The term 'cost effectiveness' begs the question of what costs and what outcomes are important. The answer will vary according to the viewpoint of different decisionmakers. Full economic evalu­ation of different approaches to healthcare may re­quire information about long term outcomes and about the relative values of these outcomes, and about the consequences for resource use both in the short and the long term, whether they are fi· nanced by health providers, service users or other agencies (Drummond et al. 19S7). From the point of view of hospital managers. short term costs and healthy survival to discharge from neonatal units would be the key issues. For parents, health policy· makers or purchasers, however, the longer term costs and outcomes after hospital discharge would also be important. 80th policymakers and caregiv. ers should also be aware of the additional costs to parents both during a baby's stay in a neonatal unit and later, at home. Although policymakers might

364

lind il helpful to compare the costs of alternative health programmes in terms of quality of life gained, methods for doing this arc various and are still being developed. In addition, as surfactant is such a new product, long term follow-u p data 3rc still limited. As a result, we will discuss the evi­dence about cost effectiveness of surfactant in terms of the shan term costs of hospital care prior to discharge from hospital. We also consider evidence concerning morbidity at hospital discharge.

To assess cost effectiveness of different policics, it is necessary 10 find out how much change there would be in the outcomes that are considered im­ponan! and in the costs of care both in terms of numbers of inputs and monetary values.

Methods for measuring hospital costs of neo­natal care vary. In some health systems, such as in the US, individual charges for healthcare arc gen­erated as pan of the reimbursement process, and these charges have been used as proxy for costs in many studies of the economics of neonatal care. In countries with less direct charging for healthcare, such as the UK, individual patient costs are not usually recorded routinely, and it is therefore nec­essary to estimate costs as a separate exercise. Methods used for costing hospital care can be ei ther patient-specific, where costs are estimated from in­dividual records of care, or based on ward ex­penditures, which are ailocated according to in­patient days. However, such a measure does not take account of variations in the intensity of care provided. In the U K, neonatal clinicians have ar­rived at a broad categorisation oflevels of intensity of care, and this classification has been adopted as a national standard by the British Paediatric As­sociation (1984, 1991). A number of economic studies have confirmed that these levels of care re­flect considerable differences in the daily costs of care. A funher lime and motion study by the Nonhern Regional Health Aut~ority in the UK has shown that these levels are indicated almost en­tirely by the venti latory requirements of babies, with few exceptions (Nonhern Regional Network Study, in press). A baby who requires anificial ven­tilation is in the 'intensive' category, and this may be funher subdivided by the amount of oxygen re-

PhormacoEconomics J (5) 199J

quired. A baby who has supplemental oxygen but can breathe unaided would be in the next most costly category of care, sometimes referred to as 'special care' or 'high dependency care,' while a baby able to breathe air unaided, but slill requiring care in a neonatal unit. would be in the lowest cost category of care referred to as normal nursery care. This finding is being tested in a UK study of the costs of economic care (Tarnow-Mordi et a!. 1992). Clearl y there arc variations in the in puts requi red at each level of care, and so precision in obtaining representative costing is difficult without large numbers of observations.

2. Sources of Eridence About E/fectireness and Costs

Because comparison of different strategies of heaJthcare using randomised controlled trials pro­vides the least biased evidence of differences be­tween policies, we have concentrated on this source of data about the effectiveness of policies for sur­factant administration. We arc fonunate that this source of evidence has already been thoroughly and recently reviewed (Dechant & Faulds 1991; Soil 1992a.b,c,d; Soil & McQueen 1992; Speer et al. 1992). Two further large international multicentre trials have recenly been completed (Grant & EI­bourne 1990; Halliday & Tarnow-Mordi 1990; OSIRIS \992). These trials will considerably in­crease the precision of the estimates of effects of different policies for surfactant use, when their re­sults have been added to the meta-analyses of all trials reported to date.

Although the value of incorporating economic questions into trial designs has been recognised for some time (Drummond & Stoddart 1984), few trials published so far provide enough evidence about costs for cost-effectiveness analysis, and overviews have not so far reviewed this aspect. We have therefore reviewed trials again for their economic content, and have also reviewed observational studies for additional data about the costs of dif­ferent levels of neonatal care.

We reviewed published evidence about the costs of neonatal care in some detail elsewhere (Howard

COSI EtTecliveness of Surfactant Replacement in Preterm Babies 365

& Mugford 1992; Mugford [988a, [988b), and re­ported 5 studies of the econom ics of surfactant use (8ackhouse [992; Maniscalco et al. 1989; Mugford et al. 1991; Phibbs et al. 1990; Shennan et al. 1989).

We have also studied the reports of randomised controlled trials of surfactant treatment policies in­cluded in the Oxford Database of Perinatal Trials (Chalmers 1992) for any evidence of economic out­comes. Although Soli quotes over 30 published trials that have evaluated d ifferent policies for sur­factant use, only 17 included data about some as­pect of the amount and intensity of care, as meas­ured by length of time on oxygen and on assisted ventilation, and length of stay in the neonatal unit. Only 4 of these 17 trials induded enough infor­mation 10 assess cost effectiveness of alternative policies. They measured variables about ventila­tion and oxygen therapy for all babies entered into the trial, and also measured length of stay in the neonatal unit (Dunn et al. 1991; Enhorning et al. 1985; Raju et a1. 1987; Shennan et a1. 1989). Al­though 3 of the trial reports did not include any economic analysis of the results, I evaluated the monetary costs of alternative policies (Shennan et a1. 1989).

In addition to these trials, 3 other economic analyses of surfactant treatment have been based on data from babies taking part in trials (Manis­calco e1 aL 1989; Phibbs et al. 1990; Tubman et aL 1990). These analyses did not use data for the whole randomised cohort. Maniscalco and colleagues (1989) compared experiences on babies from 2 sep­arate trials, whereas the studies of Phibbs and Tub­man and associates measured costs for babies from a single centre who were taking part in larger multieentrc trials. Finally, one study took data aboul cffeels of alternative policies from overviews of trials, and data about costs from an observa­tional study in a unit where none of the policies was currently in practice (Mugford et al. 1991). Some of the results of these studies are discussed in lhe following sections.

Where possible, we have quoted monetary costs, converted to current UK prices, of different poli­cies of care as estimated in original studies. Where monetary costs were not estimated in the original

articles, we have estimated the costs of care byap­plying UK costs of neonatal care to the evidencc about the amount of care given in the studies eval­uating surfactant policies. For each policy of care where there is evidence of differences in cost ef­fectiveness, we diseuss how the policy affects over­all neonatal care costs, and how the ratio of total costs to the number of survivors changes. The lat­ter measure is commonly referred to as 'cost per survivor' and should not be confused with the av­erage cost of care for babies who survive, which excludes the costs of caring for babies who subse­quently die. We also calculated the incremental costs per additional survivor implied by the dif­ferent policies. All prices were convened to £ ster­ling using purchasing power parities (OECD 1990), and then to current price levels using the UK Health Services Price Index (Department of Health 1992).

3. Policies for Surfactant Replacement

Exogenous surfactant was first developed as a treatment for RDS. It is important, when consid­ering policies for surfactant usc, to take account of other fonns of care that could prevent RDS. Figure 1 illustrates the points at which clinical interven­tion could alter the outcomes for a preterm baby. Administration of conicosteroids to women who are expected to give birth prematurely has been shown to be very effective in reducing neonatal mortality and the incidence of RDS (Crowley et al. 1990). Introduction of a policy of giving women corticosteroids prior to anticipated preterm deliv­ery, taking account of the fact that this would not always be possible, reduces the incidence of RDS and mortality, thereby reducing the average costs of care for all babies of less than 35 weeks' gesta­tion by 10%. Average costs for babies of less than 31 weeks' gestation are increased by 7% because the increased survival of such preterm babies re­sults in a greater requirement for neonatal care than would be the case for morc mature babies. Costs per survivor would be reduced by 14% and by 9%, respectively, by this intervention (Mugford et a1. 1991).

RDS cannot be prevented completely by ob-

366 PharmacoEconomics j (5) /993

Pretenn birth L Alltenatal corticosteroids? I ,nlk;IPlIled

Yo.".

f-- - - - - - - - - - - -- - - - - - - - - --

Delivery

1-- - - - - - - - - - - -- - - - - - - - - - -

<12 hour. L Prophylactic sul1actant? I

y..". -- - - - - - -- - - - - - - - - - - - - --

Neonatal ~~

I No

ADS?

y"

I Surtactant treatmer11? I ,.". Sing/a Multiple - -

I"""'" "" I"""'" "" I Survive "" 1""- ""

Fig. 1. Interventions to prevent or treat respiratory distress syndrome (RDS) during and after preterm birth.

sletric interventions such as administration of corticosteroids and so surfactant remains import­ant. We consider evidence for comparative cost­effectiveness of:

(I) Surfactant given early after delivery, pro­phylactically, to prevent the onset of RDS, com­pared with placebo control treatment and subse­quent conventional management;

(2) Surfactant given as a treatment for estab­lished RDS, compared with conventional manage­ment of RDS without surfactant;

(3) Early or prophylactic surfactant compared with subsequent treatmenl with surfactant only if RDS occurs;

(4) Giving only I or 2 doses of surfactant com­pared with further repeated doses.

Cost Effectiveness of Surfactant Replacement in Preterm Babies 367

4. EffectJ of Different Po/icits for AdminiJtration of Surfaclotlt

The primary outcome measure in most of the randomised trials looking at the effectiveness of different policies for surfactant use is 'survival at hospital discharge without chronic lung disease'. Chronic lung disease or bronchopulmonary dys­plasia (BPO) is measured by oxygen dependence either at the twenty-eighth day after delivery, o r at the expected date of delivery on w hich the baby would have been born if it had been born al fu ll term. that is, at 40 weeks' gestation.

Table J gives a brief summary, taken from over­views of randomised trials, of the effects of these policies on mortality, and on the incidence of ROS and long term oxygen dependence. Results are shown in terms of odds ratios and confidence in­tervals. The odds ratios show the odds of a partic­ular outcome for the experimental group in the trial compared to the control group in the trial, and the confidence intervals show in what range the true odds ralio for that outcome is likely to lie with 95%

confidence. For adverse outcomes, such as death, odds ratios of less than one suggest that the ex­perimental 'treatment' is more effective than the control 'treatment', and 95% confidence intervals that exclude one indicate that this effect is statis­tically significant al the 5% level. The typical odds ratio and confidence interval summarise the infor­mation from several comparable trials.

It is clear from table I that all types of currently available surfactant reduce mortality, whether they are given prophylactically or as treatment. These overviews also suggest that the reduction in mor­tality is not accompanied by any increase in the number of babies surviving with long term oxygen dependence measured by the degree of BPO. It is not clear from the overview of the 3 trials so far published, what relative advantage early prophy­lactic use of surfactant has when compared with later administration. Speer and colleagues (1992), in an analysis of 3 trials of different policies for use of porcine lung surfactant, found that early prophylactic use reduced the odds of neonatal death, severe RDS and intraventricular haemor-

Table I. Effectiveness 01 differem policies tor pl'lvenlion and traatment 01 rlSpir.1ofy dlsu.8. syndrome (RDS) using ,urfactant

'''''' """""" Typical odds 95% oonticlence Releten08 ratio Interval (no. 01 lrials)

Surfaet8n' fOl' trntment of RDS n control

Synthetic Death betOl'a d ischarg.e 0.65 0.50-0.82 Son 1992a

(5'

"'a 0.68 0.46-0.99 Natural Death befOl'e discharge 0.45 0.26-0.78 SoIl 1992b

(12)

"'a 1.01 0.81 -1.27

PropI'IyIKtic aurfaet8nt ... control (el aurfKUMII) .... surfactant. Reported mortatity 0.62 0.50-0.77 SolI 1992c

(14)

"'a 0." 0.77-1.15 Pfophy&ectic: wrfKtMlt n treatment wilt! ..... rfaet8n' fE.tty va late) Death before discharge 0.78 0.56-1.08 SOIl 1992d

(3)

"F'!) 1.10 0.81-1.63 Single dose ~, mu~pIe dose Death before discharge 0.59 0.34-1.02 SoIl 1992e

(2,

"'a 1.12 0.59-2.12

Abbr.viatiOns: ePD • bronchopulmonary dysplasl • .

.168

rhage. Results of the OSIRIS trial comparing the use of synthetic surfactant under different usage policies have now been published (OSIRIS 1992). These results have not been included in the over­views of trials, but the trials showed that early administration of surfactant led to a reduction of about 6% in the combined risk of death or chronic lung disease, compared with a policy of delayed, selective administration of a surfactant.

It is not yet clear from the overview whether multiple doses of surfactant would lead 10 better outcomes than restriction to I or 2 doses, but the OSIRIS trial could not detect a difference in out­come even with over 3000 babies in each group.

Although the overviews consider the effects of surfactant on BPD or death, they do not specifi­cally analyse how the different policies affect the number of babies who survive to discharge but have chronic lung disease. As this may be an important indicator for further health service requirements after discharge, we have estimated from the data in the overviews the numbers who had BPD but did nOI die before discharge from hospital. We have not perfomled a formal meta-analysis of these data, but we nOle that there is not a consistent pattern for all the trials. The results do not suggest to us that any of the policies for surfactant use would result in a large increase in numbers of babies dis­charged alive but having chronic lung disease.

5, COSIS of Hospital Care

Most of the published work on economics of neonatal care has concentrated on heaJthcarc costs. Most studies have been based on costs in individ­ual neonatal units although a recent study (Ford­ham ct al. 1992) based costs on 17 neonatal units in one English health region.

The hospital costs of care for preterm babies are influenced by the intensity of care, determined principally by ventilation and oxygen require­ments, and by length of stay in the neonatal unit. The daily cost of intensive care for a baby being artificially ventilated can be between 2 and 3 limes that of special or high dependency neonatal care, which, in turn, is about twice the cost of normal

PharmacoEconolllics 3 (5) 1993

nursery care for 'well' babies without any medical need who are cared for in a neonatal unit (Howard & Mugford 1992). These ratios are fairly consist­ent, regardless of the costing method used. Based on our review, we have assumed the average cur­rent daily costs of neonatal care in the UK to be £540 for the most intensive care, £220 for high­dependency care such as would be required by a baby receiving supplemental oxygen, and £140 for the lowest level of care given in a neonatal unit.

Babies experience different patterns of intensity of care and length of hospital stay depending on their gestational age at binh, whether they have RDS, and whether they survive to discharge. The evidence suggests that babies who survive without having had RDS are likely to have care costing half that which would have been necessary if they had RDS (M ugford et al. 1991: Phibbs et al. 1981). Pre­term babies who die are more likely to have had intensive care, but have lower average costs than those who survive, mainly because of the shorter time they are in neonatal care.

The trial reports which did include data about resource utilisation by babies give cost estimates for care which are consistent with the evidence from observational studies of neonatal costs. In panie­ular, costs are higher for babies of lower gestations, and cost estimates from the US arc much higher than those from the UK, reflecting the different healthcare systems and costing methods. The eco­nomic results from these studies are presented in tables II and III , and arc discussed below.

5.1 Costs to Parents

Relatively few studies have looked at costs to families while their baby is in neonatal care. Those that have been published show that these costs are considerable. Not surprisingly, they are likely to be related to the length of stay of the baby and dis­lance from hospital to home (Giacioa et a!. 1985; Smith & Baum 1983). In 1983, Smith and Baum estimated that the mean cost over a 2-month pe­riod for parents travelling by car to visit babies re­ceiving neonatal care in the Oxford region of the UK was £41 for parents of babies born at the re-

Cost Effectiveness of Surfactant Replacement in Preterm Babies 369

gional maternity unit, and £123 for parents of ba­bies born elsewhere. Giacoia and associates found that parents who had to travel over 50 miles to the hospilal visited less often than parents who lived nearer. There are other important costs to families such as the loss of earnings if the baby is in the neonalal unit for a long period and also the cost of exira help at home, especially if there are other children in the family (0 look after, but there has so far been very linle research on this important subject.

5.2 Long Term Costs

The reported trials of surfactant have already demonstrated that surfactant increases the number of prelerm babies who survive to go home. This in itself is important for health planners to know, as prelerm infants have additional care require­ments at home (Botting et al. 1990; Couric! & Dav­ies 1988) and are more likely to be rehospitalised during the early years of life (Mutch et al. 1992).

6. Cost Effectiveness of Different Policies for the Use of Surfactant for Preterm Babies 6.1 Early or Prophylactic Administration of Surfactant

If surfactant is used, the timing of its use may affect both how many babies reeeive it and also its overall effectiveness. Giving surfactant at birth or very soon afterwards, when babies are still only at risk of developing RDS, has been shown to reduce mortality and also the incidence of RDS, com­pared to a placebo control treatment (table I). Table II gives details of 3 studies that have compared resource implications of these policies. The gesta­tional age groups differed among the 3 studies, so they are not directly comparable, as the risk ofRDS and eosl of care varies with gestational age, In comparisons between groups of babies it is im­portant 10 ensure that initial disease severity is taken into account.

In table II, international differences in neonatal

care costs are highlighted in the comparison of the results from Maniscalco and colleagues., (1989) who based costs on US hospital charges, the study by Shennan and colleagues (1989) who estimated total cost by level of care in Canada, and a comparison of UK neonatal costs (Mugford ct al. 1991). The 3 studies used different costing methods. Maniscalco and colleagues did not include the cost of surfac­tant and at £1000 per baby this would increase the cost for 100 babies in the surfactant group to £4870000 and the cost per survivor in this group to £52 935. We converted all costs to 1992 sterling values. Table II shows that costs of neonatal care for babies having prophylactic surfactant are not consistently higher or lower than for those who did not receive surfactant. There is a difference in the pattern of care intensity between policies. Shennan and colleagues (1989), who showed that total costs of care are less with prophylactic use of surfactant, also found that even though survival is increased, babies receiving surfactant have less time on the ventilator and on oxygen, and also that the overall length of stay and cost of care was reduced, as was the 'cost per survivor'. Maniscalco and his col­leagues (1989) also found a reduction in average daily cost, but the longer average hospital stay for the surfactant group meant that there was only a very small difference in total charges. This study also found that prophylactic surfactant would re­sult in a reduction in the 'cost per survivor'. This would not necessarily be the case when babies of greater gestational ages are included. Mugford et al. (1991) showed that the overall cost of care for babies under 35 weeks would be increased, and that prophylactic surfactant would decrease costs per survivor if given to babies under 31 weeks' gesta­tion, but would increase the costs per survivor by 12% if given to all babies under 35 weeks.

Incremental analysis of these results based on the estimates of Manisealco and colleagues (1989) and Mugford et a!. (1991) show an additional cost per additional survivor of £7000 to £13 000, while the estimate of Shennan and associates (1989) im­plies net savings from the policy of prophylactic surfactant, compared with a policy of care without any surfactant.

370 Pharmacol::conomii's j (5) /993

Table II . Studies comparing prophylactic administration 01 surfactant with placeoo cOI"Itroi or no surfactant (1992 prices)

Relerence Gestation No. 01 babies % survival Total cost per 100 Cost per survivor' Additional

iwee~s) babies (. (1000) (-£1000) cost

surfactant COf1lrol surfactant ~tr. surtactant control surfactant contrOl (savings)

'" a!kljliooal

survivor

Maniscalco at 25·29 38 " " " 48701> .,,, " " £7143

al. (1989) Shennan 8t al ." " 60 93 " 1712 2259 " " (£91 000) (1989) Mugford &1 aL <35 " " " .. 50' '" , , £12875 (1991)

<" " " " '" "" "" " " £9'()()

, Total cost fOf all bab<as divided by number of survivors.

b Including the oosl 01 surfactant at £1000 per baby (not included in original cost estimates).

6.2 Surfactant Given as Treatment

Table III includes results from three studies. For two (Raju el al. 1987; Tubman et al. 1990) the costs quoted have been estimated using the authors' data about levels of care and our estimates of current UK neonatal care costs. The impact on costs of surfactant used as a treatment for established RDS is not clear from the studies quoted in table III. This partly reflects the small number of babies for whom cost data are available. Table III shows that the three studies all showed increased survival, which is consistent with the overview quoted in table I. In the trials reported in table III, babies are

categorised by birthweight rather than gestational age. Although birth weight and gestational age dis· tributions do not exactly correspond, birth weights of under 1000g approximately correspond to ges­tational ages under 28 weeks, and those under ISOOg to gestational ages under 32 weeks. Table 1JI shows how costs are affected by surfactant treatment for RDS. Backhouse estimated costs separately for two birthweight groups, and fo und that although total costs of care were increased by surfactant for ba­bies in the lower birth weight group, who are at higher risk of RDS, overall costs were reduced for the babies with birth weights over 1250g. Two OUI

of the three studies found that treatment of RDS

Table III. Surfactant fOf treatment 01 respiratory distress syndrome versus placebo control treatment

Reference Birthweighl No. of babies % survival Total cost per 100 Cost per survi~ Additional (9) babies (, £1000) (. £1000) cost (saYings)

surfactant control surfactant oontroj surfactant ~tr" surfactant control per additional

survivor

Raju at al. 751-1750 17 13 " 46 1745 1961 21 43 (£6000) (1987)

Tubman et 700-2000 19 ,. 79 36 1743 '" 22 17 £26721 al. (1990)

Backhouse 700-1350 100 100 85 71 5189 4661 61 " £37714 (1992)

;1>1250 100 100 " 91 ,,,, 3989 37 44 (£178667)

, Total cost for all babies diyidad by number of survivors.

Cost Effectiveness of Surfactant Replacement in Prc lerm Babies 371

with surfactant for babies with lower birthweights led to a reduction in 'cost per survivor'. Although Tubman and colleagues found an increase in 'cost per survivor', the additional cost that they calcu­lated of achiev ing an additional survivor would be £ 16 596 when updated to 1992 prices. which, as they observe, is not more tha n is typically spent on neonatal care for very preterm infants. Our es­timate based on different neonatal costs shows this additional cost to be over £26 000. However, table III shows that fo r babies of birth weight between about 700 and 1500g the additional cost per ad­ditional survivor could range from a net savi ng of £6000 to a cost of £3 7 700.

6.3 Earl y Surfactant Versus Later Selective Administration

II is not possible to compare directly Ihe evi­dence fo r cost effectiveness of surfactant used in these different ways from the studies referred to above. because the trials upon which they were based included babies in different gestational age or birthweight groups, and disease severity was not measured systematicall y in allihese studies. For all these reasons the different studies included babies with different levels of risk of RDS. However, sur­factant treatment has been compared, within ran­domised tria ls. with surfactant used in a preventive mO:de, and one study (Phibbs el al. 1990) included economic measurements. The authors estimated that proph ylaxis increased the average COSIS of care by $US4 IQO per baby (1992 prices). No improve­ments in outcomes were observed in this study, which was too small to detect important differ­ences in mortality. We have already mentioned one much larger trial com paring these different surfac­tant policies which has now reported results. A multicentre economic evaluation is also in prog­ress in collaboration with the UK centres taking part in two multieentre surfactant trials (Tarnow­Mordi el al. 1992).

6.4 Reduction of Effects of RDS ­Dosage Required

Apart from timing, the optimum dose of sur­factant is also important. Given the high unit cost of surfactant, longer courses could only be eco-

nomically justified if there is a significant addi­tional benefit to be gained. The current overview of this aspect of surfactant usc leaves the question unresol ved. If there is no improvement in out­come, then it is clearly not cost elTecti ve to give additional doses of surfactant. The OSIRIS trial (OSIRIS 1992) com pared giving two doses with giving up to four doses of surfactant. The CUROSU RF4 trial (Halliday & Tarnow-Mordi 1990) compared giving a standard regi men of 2 or 3 doses with the option of giving additional doses of surfactant. The results of the OSIRIS trial do not suggest that there would be additional benefit from multiple doses of surfactant.

7. Conclusions

Surfactant is effective in reducing respiratory ill­ness and mortality in preterm babies. Because it increases survival. and may thus increase the length of stay in neonatal units, and because of its ex­pense, surfactant can increase the overall cost of carc. The balance between these changes is likely to vary with mortality risks and severity of disease: in the most preterm babies who have the highest mortality risk. and the longest need for care if they survive, the costs of longer hospital care may out­weigh the savings from reduced ventilator and oxy­gen requi remen ts. From the small number of stud­ies so far published it is difficult to assess the size and direction of this effect on costs with confi­dence. This problem is particularly clear for analy­sis of incremental costs. There is a need for further work on pooling data about costs and combining them with the overview data on elTecti veness of different pol icies for surfactant use. However. it appears that surfactant is likely to reduce the ratio of costs to survival for neonatal care. especially for surfactant given as Ireatment, and for babies born at lower gestational ages.

There are sti ll man y unresolved questions about the economics of surfactant use. Man y of these would have been answered if more of the man y trials so far published had included bener indica­tors of resource utilisation, sueh as days at different levels of intensity of care, or some equivalent. The

372

overviews which we have quoted in th is paper d o not include a nalysis o f the effects on continuous variables. such as length of time o n a ventila tor. This approach is now bei ng developed, and we plan further analyses of those trials which have incl uded economic data. This will allow u s to estimate con­fi dence intervals for the data we have presented here.

Further work is also planned, in collaborat ion with the coordinators of t he OSIRIS and CU ROSURF4 t rials o n the econom ics of early ver­sus delayed selecti ve use of surfactant and on the number of doses that arc g iven. This includes a detailed survey of the costs o f neonatal care in 62 neonatal units in the UK. Final results 3rc likely to be available lale in 1994.

Purchasers of healthcarc s hould be aware that a lthough surfactant can reduce the cost per survi­vor o f neonatal care, this ratio can be decreased further. and overall COSts of neonatal care may even be reduced if effective obstetric care was given t o more mothers before p reterm del ivery. In part ic­ular, compared with the p rophylactic or more se­lecti ve use of surfactant, a policy of care that in­cludes intention to give corticosteroids before preterm deli very is more cost effective. There is no doubt that admi nistration of cort icostero ids prior to preterm delivery is more cost effective than sur­facta nt alone, and that it wou ld rcduce the overall need fo r surfacta nt. It would n ot-replace the need for surfactant because there are some situations where binh takes place too soon after the woman has made c ontact with care givers. However, in the UK, there is still scope for much wider use of the policy (Donaldson 1992; Khanna & Richmond 199 1). Therefore, there is a nced for further analy­sis of the impact on requ irement for surfactant and the relative economic advantage of early versus late administration when cort icosteroids arc used more routinely.

Acknowledgements

We wish t o thank manyeolieagues for their comments o n drafts of this paper. Our research is funded by the Department o f Health (England). and .... 'e also aeknow.

Pharmaco£amQmics j (j) /993

ledge the support of the Medical Research Council (U K) for the p roject 'ai nical and economic evaluation of exogenous surfactant treatment for neonata l resp ira tory dislr"Css syndrome'.

References

Backhouse ME. Cost-effectiven~~, of bosun neonatal rescue therapy in infants with ROO. Advances in Paediatrics Pam­phlet. p. 8, January. 1992

Bolling BJ , Macfarlane AJ. Price FV {Edsl. Thre~. four and more: a study of tri plet and higher order births. Her Maj(1ty's Sta· tionary OtT.c-e, London, 1990

Brit ish Paediatric Associa tion and British Associa tion for Peri­nalal Paediatrics. Catesorics of babies Kqui ring neonatal ca re. Bri tish Paediatric A~sociation and BritiSh Association for Peri· natal Paediatrics, London, t984

Bri tish Paediatric Associa tion and Briti sh A~sociation for Peri· na lal Medicine. Categories of babies requiring nconatal can:. British Paediatric Association and British Association for Peri­natal Medicine. london. 1991

Chalmers I (Ed.). Oxford database of perinatal trials. Version 1.3. Disk issue 8, Au tumn, O~rord University Press., Oxford. 1992

Couriel JM. Davies P. Costs and benefi ts of a community specia l can: baby service. British Medical Journal 296: 1043·1046. 1988

Crowlfoy P. Chalmers I, Keirse MJNC. The effects of ronlcoslcr· oid ad ministration before pn'term cklivery. an o"en'iew of the evidence from controlled trial$. British Journal of ObstetriC$ and Gynaecolosy 97: ] ]-23. 1990

Crowley P. Conicosteroids prior to p«tcnn dcli>·ery. In Chalmers (Ed.) Oxford dat.a.1J;aS(" of perinatal trials.. Vc~ J.J. Disk ,SSt!( II, Reeord 29SS, O~ford University Press., O~ford. 1992

Dtthant K L, Faulds D. Colfaseeri l palmitate. A review of the therapeutic effiCllCY and clinicaltolcrability of I synthctit sur­factant P«'P3.ration (EllosurP Neonatal''') in neonatal rcspi· ratO<)' distress syndrome. Drugs 42: 877.g94. 1991

Department of Health. Financial matters. National Health Serv· icc management ueeutive. Referen~ 7/92. Department of Health, London, UK. 1992

DonaldloOn U. Mai ntaining u~lIenc-e: the preservation and de· velopment ofspeciali5ed services. British Medical Journal lOS: 1280.1283, t992

Drummond MF, Stoddart GL. Economic anai)"sis Ind clinical trials. Contro lled ainical Trials 3: 113-128. 1984

Drummond MF. Stoddart GL. Torrance GW. Methods for the evaluation of hc-atth care provammcs. Od ord University Press., O~ford. 1987

Dunn MS, Shennan AT, Zayack D. Pournayer F. Bovine surfac­lant re placement therapy in neoMteS of 1(1$ than 30 weeks' gestation: a randomized contrOlled trial of prophyluis versus t reatment. Pediatrics 87: 377-386, 1991

Enhorn ing G, Shennan A, Possrnaytr F. Dunn M. Chen CPo et al. Prevention of neonatal respira tory distrcSli syndrome by tracheal insti llation of surf.ctant: a randemised clinical trial. Pediatrics 76: 145-1S3. 1985

f ordham R, Field 0 , HCJd&e$ S. Normand C, Mason E. ct al. Cost of neonatal care across. rqional health authority. Journal of Public Health Meditine ]4: 127-130. ]992

Giaroia GP. Rutledce D. West K. Factors affecti", visitation of sid: newborns. ainical PedialriC$ 24: 2S9-262. 198~

Grant A. Elboume D. OSIRIS: open study or inr.nts at hiab risk of or wi th l"C$piratory inwfliciency - Iht role of surfactant. Back&round information and protocol. Naliona! Perinatal Epi· ckmiolosy Unit, O~rord, ]990

Halliday HL. McOurc G. Reid M. Handbook of neonatal inten­sive care. 3rd ed .. BailltTc Tindall. London. 1989

Cost Effectiveness of Surfactant Replacement in Preterm Babies 3J3

Halliday HL. Tarno ..... -Mordi WOo CU ROSU RF' 4: protocol of 4th collaborative multirentre study ofCurosurf. Royal Mater­nity Hospilat. Belfast . 1990

Howard SE. Mugford M. The costs ofneonalat intensive care and the economies of surfaClanl. National Perinatal Epidemiology Unit. Oxford, 1992

Khan na R. Richmond S. Antenatal treatment to enhance lung maturity. Provisional repon on perinalal and late neonalal death in the nonhern region. pp. 27-30. Nonhem Regional Health Authority. New.;astle. 199t

Maniscalco WM. Kendig JW. Shapiro DL. Surfactant replace­ment therapy: impact on hospital charges for premature in­fants with respiratory distress syndrome. Pediatrics 83: 1-6. 1989

Mugford M. A review of the economics of care for sick newborn infants. Community Medicine 10: 99-111. 1988a

Mugford M. A review of the economies of care for sick ne ..... born infants. Com:spondence. Community Medicine 10, 1988b

Mugford M. Piercy J. Chalmers I. Cost implications of different approaches to the pre"ention of respiratory distress syndrome. Archives of Disease in Chi ldhood 66: 757_764. 1991

Mutch L. Ashurst H. Macfarlane A. Binhweight and hospital ad­mission before the age of two years. Archi,'es of Disease in Childhood 57: 900-904. 1992

Nonhem Neonatal Network Study. measuring neonatal nursing workload. Archives of Disease in Childhood. in press

Organisation for Economic Cooperation and [kvelopment. Health ca re systems in transition. The search for efficiency. OECD Social Policy Studies No, 7. Paris. 1990

OSIRIS Collaborat ive Group. Early versus delayed selective neo­natal surfaclant - the judgement of OSIRIS. Lancet 340: 1363-1369. 1992

Phiblr.; CS. Phibbs RH. Tooley WH, Wakely A. Sniderman S. The cost..effectiveness of synthetic surfactant therapy for neonatal respiratory distTesli syndrome. Oinical Research 38: !78A. 1990

Phibbs CS. Williams RL, Phibbs RH. Newborn risk factors and costs of neonatal in tensive care. Pediatrics 68: 313-320. 1981

Raju TNK. Bhal R. McCulloch KM. Maeta H. Furner S. Double­blind controlled trial of singlc-<losc treatment with bovine sur­factant in ~vere hya line membrane disease, Lancet 1: 65J-6S6. 1987

Shennan A. Dunn M. Possmaycr F. Cost..effecti"eness of single

dose surfactant prophylaxis in infants of Icss than 30 weeks geslation. Pediatric Research 25: 23 1A. 1989

Smith MA, Saum JD, Costs of visiting babies in special care baby units. Archives of Disease in Childhood 58: 56-59. 1983

Soll RF. Synthetic surfactant treatment of ROS. In Chalmers (Ed.) Oxford database of perinatal trials, Version 1.2. Disk issue 8. Record 5252. Autumn. 1992a

Soil RF. Natural surfactant treatment of RDS. In Chalmers (Ed.) Oxford database of perinatal tria ls. Venion 1.2. Disk issue 8. Record 5206. Autumn. 1992b

Soli RF. Prophylactic administrat ion of any surfactant. In Chal. mers {Ed.) Oxford database ofpcrinalaltrials. Version 1.2. Disk issue 8. Record 5664. Autumn. 1992c

Soll RF. Prophylactic surfactant versus treatment with surfactant. In Chalmers (Ed.) Oxford database of perinatal trials. Version 1.2, Disk issue 8. Record 5675. Autumn. 1992d

Son RF. Single versus multiple dose surfactant for severe ROS. In Chalmcl"l; (Ed.) Oxford database of perinatal trials. Version 1.2, Disk issue 8. Record 7091. Autumn. 1992e

Soli RF. McQueen Me. Respiratory distress syndrome In Sinclair & Bracken. Effective can: of the newborn infant. pp. 325-358. Oxford Univel"l;ity Press. Oxford. 1992

Speer CPo Roben$On B. Curstedt T. Halliday HL, Compagnanc D. Randomiled European multicentre IriaT of surfactant reo placement therapy for severe respiratory d istress syndrome: single venus mlll1iplc doses of Curosurf. Paediatrics 89: 13-20. 1992

Tarnow-Mordi WOo Halliday H. Hey E. Malek M. Mugford M. et a\. The economics of neonatal surfacwntthcrapy. MRC News 57: 16-17.1992

Torrance GW. Zipursky A. Cost..effectiveness of antepanum Rh immunisation Clinics in Perinatology I J: 267-281. 1984

Tubman TR, Halliday H L. Normand e. Cost of surfactant re­placement treatment for severe neonawl respiratory distress syndrome: a randomiled control tria!. British Medical Journal 301: 842-845. 1990

Com:spondence and n:prints: Miranda Mutiford. National Peri­natal Epidemiology Unit. Radcliffe Infirmary. Oxford OX2 6HE, England.