An economic evaluation of single dose systemic methotrexate and laparoscopic surgery for the treatment of unruptured ectopic pregnancy

Transcription

1 British Journal of Obstetrics and Gynaecology February 2001, Vol. 108, pp. 204±212 An economic evaluation of single dose systemic methotrexate and laparoscopic surgery for the treatment of unruptured ectopic pregnancy Martin C. Sowter*, Cindy M. Farquhar, Guy Gudex Objective To compare the direct and indirect costs of single dose systemic methotrexate with laparoscopic surgery for the treatment of unruptured ectopic pregnancy. Design A cost minimisation study undertaken alongside a randomised trial. Setting Departments of Obstetrics and Gynaecology in three hospitals in Auckland, New Zealand. Participants Sixty-two women with an ectopic pregnancy randomised to treatment with either a single dose of methotrexate (50mg/m 2 ) or laparoscopic surgery. Main outcome measures Direct and indirect costs based on the results of the randomised trial. Results Direct costs per case were signi cantly lower in the methotrexate group (mean $NZ 1470) than in the laparoscopy group (mean $NZ 3083) with a mean difference of $NZ 1613 (95% CI $NZ $NZ 2061). These signi cant differences existed under a wide range of alternative assumptions about unit costs. The difference in direct costs in favour of methotrexate was greatest for women presenting with low pretreatment serum b-hcg concentrations. Mean indirect costs were also signi cantly lower in the methotrexate group (mean $NZ 1141) than in the laparoscopy group (mean $NZ 1899) with a mean difference of $NZ 758 (95% CI $NZ $NZ 1240). For women presenting with pretreatment serum b-hcg concentrations of over 1500 IU/ L this difference in indirect costs is lost due to the prolonged follow up required and a higher rate of surgical intervention in women receiving methotrexate. Conclusion This economic evaluation shows that treating suitable women with an ectopic pregnancy using systemic methotrexate therapy results in a signi cant reduction in direct costs. The indirect costs borne by the woman and her carers are only likely to be reduced in women with pretreatment serum b-hcg concentrations under 1500 IU/L. INTRODUCTION In most developed countries the incidence of ectopic pregnancy has increased considerably over the last 20 years and now accounts for 1%-2% of all pregnancies 1± 4. The great majority of women with an ectopic pregnancy are currently treated surgically, usually by laparotomy. Randomised trials have shown laparoscopic surgery to be a safe and effective alternative to laparotomy with signi cant reductions in the costs of treatment, although there are still considerable costs associated with operative laparoscopy, both for the health-care provider and the women themselves 5±8. A number of case series published over the last ve years have demonstrated that with single dose systemic methotrexate, a proportion of women with an ectopic pregnancy can be treated on an ambulatory basis without the need for either diagnostic or operative laparoscopy 9±12. Economic evaluations based University Department of Obstetrics and Gynaecology, National Women's Hospital, Auckland, New Zealand * Correspondence: Dr M. Sowter, Fertility Associates Ltd., Ascot Integrated Hospital, Private Bag 28910, Auckland, New Zealand. q RCOG 2001 British Journal of Obstetrics and Gynaecology PII: S (00) on ef cacy data from these uncontrolled studies have concluded that considerable savings in both direct and indirect costs could be made if all suitable women were treated medically 13±18. However, to date, no economic evaluation comparing single dose methotrexate therapy with surgical treatment has been based on prospectively collected data from a randomised trial. In this economic evaluation, the rst to be conducted alongside a randomised trial 19, we have compared the direct and indirect costs associated with the treatment of women with unruptured ectopic pregnancies using either single dose systemic methotrexate or laparoscopic surgery. METHODS The clinical trial on which this economic evaluation is based was an open pragmatic randomised trial undertaken at three Auckland hospitals (National Women's hospital, North Shore hospital and Middlemore hospital). Women were recruited from 28 July 1997 to 27 September Women were allocated to either out-patient treatment with single dose methotrexate or to laparoscopy, which was undertaken with the aim of performing a laparoscopic salpingotomy if possible. A detailed

2 ECONOMIC EVALUATION OF SINGLE DOSE SYSTEMIC METHOTREXATE 205 description of the design of the trial and treatment outcomes is reported on pages 192±203 of this issue 19. Data were collected prospectively for each woman from her entry to the trial until the completion of follow up. Data on the use of resources prior to randomisation were also collected as some women had been admitted overnight or reviewed clinically a number of times before a de nitive diagnosis was made. Suf cient data on the use of resources were collected to permit the calculation of the cost of investigations undertaken; the costs of initial and follow up visits to the gynaecology assessment unit; the costs of all drugs used; operative and anaesthetic costs; in-patient hotel costs; and any costs associated with additional treatments and complications. Direct costs The costing of health service resources used was based on 1998 New Zealand unit costs and are expressed in New Zealand dollars ($NZ 1.00 ˆ 0.31 sterling). Although these data were collected for all women in the trial the value of these resources was based on unit costs at one hospital (National Women's hospital). It is likely that there is some variation in the unit costs between each hospital but accurate and reliable unit costs could be most easily obtained for National Women's Hospital, and these unit costs have been used to value the use of resources by all women in the trial. All consumables used have been costed using market prices including general service tax (12.5 %). Individual cost components were costed as follows: Investigative costs included all investigations undertaken during the diagnostic process and during treatment follow up. The costs of initial liver function and renal function tests were excluded in the laparoscopy group, as these were tests speci c to the trial undertaken to determine trial eligibility. Unit costs are based on those charged to other hospital departments by Auckland Healthcare laboratory services and the hospital ultrasound department (Table 1). Unit costs of drugs used were based on the purchase price negotiated with suppliers by Auckland Healthcare. Dispensing and other pharmacy operational costs were billed separately to each hospital department and have been included in the relevant department's overhead costs. The hourly cost of time spent in the gynaecology assessment unit (analogous to an early pregnancy unit) was based on the departmental accounts, the number of women seen and the annual hours of occupancy. This unit cost included all staff costs, ancillary costs, and corporate overheads. Operative and anaesthesia costs included a xed cost per case, comprising of the capital costs of a laparoscopy Table 1. Use of resource and unit costs for women in the trial. Values are given as the number of units used (n) or mean [range]. Resource use Unit cost ($NZ) Methotrexate (n ˆ 34) Laparoscopy (n ˆ 28) Investigation (cost per test) Full blood counts Liver function tests Renal function tests HCG assays Ultrasound scans Histology Therapeutic Operative laparoscopy 4 29 Laparotomy 1 0 Total theatre time (min) Theatre time (min) 76 [50±107] a 68 [40±101] Doses of methotrexate per dose 46 2 Patient in hospital Total time in hospital (days) per day 0.82 [0±5] 1.77 [0.4±4.7] Post-op. time (days) 1.24 [0.2±3.0] Follow up Follow up consultations per visit 4.4 [1±8] 2.9 [1±6] Working days lost by employed women 6.0 [0±25] 11.1 [1±28] Days unable to perform household tasks b 4.3 [2±6] 8.7 [5±12] Working days lost by employed partners c 2.2 [0±11] 3.3 [0±13] Working days lost by other carers 16 [12 female; 4 male] 26 [23 female; 3 male] Distance travelled during follow up (km) [2±300] 78 [4±270] a b Mean (range) for the 5 women operated on during follow up. To avoid double-counting, only the days that women not in paid employment were unable to perform household tasks were included in the calculation of indirect costs. c Number of women compared with number of men, respectively.

3 206 M.C. SOWTER ET AL. tower, laparoscopy instruments, some anaesthesia costs, and consumables used for every case; and a variable cost per case, dependent on the duration of each procedure, comprising theatre overheads, other anaesthesia costs, and staff costs. Fixed equipment costs per case were calculated from their annual rates of use and an estimated annual equivalent cost based on the 1998 purchase price. An annual discount rate of 6% was used, and it was assumed that the equipment has a useful life of ve years. Total xed theatre costs for equipment and disposables per laparoscopy were $NZ Fixed anaesthesia costs were $NZ Variable theatre costs per case included the cost per minute of anaesthetic equipment used, departmental overheads, staff costs and the costs of any additional drugs and consumables used. The costs per minute of anaesthetic equipment used were based on their annual equivalent cost and the annual logged minutes of operating per anaesthetic machine undertaken in the department. Departmental overheads per minute were calculated from the theatre and anaesthetic department accounts for the nancial year 1997/1998 and the total logged minutes of operating within the same period. Theatre overheads were calculated to be $NZ 7.01 per minute and anaesthesia costs per minute were calculated to be $NZ Staff costs were based on the gross hourly pay for nursing, technical and medical staff present in theatre. It was assumed that a surgeon and anaesthetist of both specialist and registrar grade were present in theatre. Total staff costs per minute were calculated to be $NZ Details of any additional drugs and consumables used were recorded for each case. For in-patient hotel costs the calculation of use of resources was based on the number of hours spent by each woman on the gynaecology ward. Unit costs were calculated from the departmental accounts for the 1997/8 nancial year and the total hours of bed occupancy for the same period. Included were all nursing, medical, and support staff costs, consumables, and corporate overheads. An hourly cost of $NZ (equivalent to $ NZ per day) was calculated. Follow up costs were based on the number of follow up visits for each woman was recorded. The unit cost of each follow up visit has been based on the hourly overheads and medical staff costs in the gynaecology assessment unit. It was assumed that each follow up visit lasted 30 minutes and these were calculated to cost $NZ A modi ed human capital±cost approach to the calculation of indirect costs was used 20. At trial entry details were obtained of the woman's and her partner's employment; distance from the hospital; and of other friends or family members who might care for the woman following treatment. Women recorded how many days they and their partners or other carers had been unable to work because of their treatment. They were also asked for how many days following treatment they were unable to undertake normal domestic tasks and how much they had spent on medical or sanitary supplies, and visits to private specialists, and complementary therapists. This information was used to calculate indirect costs based on 1. the reduction in paid production due to each woman's treatment de ned as the lost income due to the absence from work of women, their partners and other carers in paid employment. The mean total daily income for men ($NZ ) and women ($NZ ) in the Auckland area was used, based on the latest available labour market statistics (Quarterly Employment Survey May 1998, Statistics New Zealand, Wellington); 2. the reduction in unpaid production due to each woman's treatment de ned as the time women were unable to perform normal domestic activities. This was valued at the same rate as the mean total daily income for women in Auckland ($NZ per day). To avoid double counting, these costs were only included for women not in paid employment. No allowance was made for unpaid production lost by partners or other carers not in paid employment; 3. additional indirect costs (transport, other medical and non-medical expenses), which were accrued by the woman due to travel costs and the purchase of sanitary products, medication and healthcare from other providers. A transport cost of $NZ 0.71 per km was used (New Zealand Automobile Association 1998). Sensitivity and scenario analyses Some uncertainty concerning the unit costs used in this study is inevitable. Because of this, a non-probabilistic, one way sensitivity analysis was undertaken for the estimation of total direct costs. It was assumed that for each cost component (e.g. gynaecology assessment unit costs, theatre costs, hotel costs, investigation costs, drug costs, and follow up costs), unit costs were 50%, 150% and 200% of base-case unit costs. This sensitivity analysis was not undertaken for indirect costs, because the value of all components of lost production was based on the same unit costs. In the randomised trial on which this economic evaluation is based, women treated with methotrexate were more likely to require prolonged follow up and multiple doses of methotrexate or surgical intervention at higher pretreatment serum b-hcg concentrations. A sensitivity analysis was undertaken to determine if pretreatment serum b-hcg concentrations would affect total direct and indirect costs. A scenario analysis was also undertaken to estimate what the overall potential reduction in treatments costs per ectopic pregnancy would be if all eligible women in our total woman population were treated with methotrex-

4 ECONOMIC EVALUATION OF SINGLE DOSE SYSTEMIC METHOTREXATE 207 ate. The likely effects on treatment costs of using an upper limit for pretreatment serum b-hcg concentration of 1500 IU/L and 1000 IU/L for medical therapy were also determined. This was done because although using methotrexate only in women with a low pretreatment serum b-hcg concentration would increase the cost savings made for each woman treated medically (Table 4), the proportion of women with an ectopic pregnancy eligible for medical therapy would also be reduced and so potentially would the overall reduction in treatment costs for our total woman population. An attempt was made to collect data on the use of resources for women with an ectopic pregnancy not entering the randomised trial 18 on which this economic evaluation has been based. These data were collected in order to determine if treatment costs for women having the same treatment as part of the randomised trial were similar to those for women not eligible for or declining randomisation. These data also permitted the determination of direct costs for women undergoing treatment not compared within the trial including laparotomy either after diagnostic laparoscopy or as their primary treatment as well as direct costs for women managed by observation alone. Statistical methods Analysis of all outcomes in the trial was by intentionto-treat. For the economic evaluation, data on all randomised women were included. Where appropriate, results are given as mean values with 95% con dence intervals or a range. All statistical analysis was undertaken using SPSS Version7.0 (SPSS Corporation, Chicago, Illinois, USA) and Con dence Interval Analysis software 21. RESULTS Trial outcomes Seventy-nine women were eligible for trial entry and 62 were recruited from a total woman population of 218 wo-men presenting with an ectopic pregnancy. Thirtyfour women were randomised to methotrexate and 28 women to laparoscopy. In the methotrexate group 22 women (65%) were successfully treated with a single dose of methotrexate and nine women (26%) required more than one dose. Eight (24%) women were readmitted with abdominal pain after treatment with methotrexate. Four of these women were observed only, and four underwent surgery. One woman receiving methotrexate requested surgical treatment the day after randomisation. In the laparoscopy group all 28 women were treated by laparoscopy: Seventeen women had a salpingotomy or tubal abortion and required follow up with weekly serum b-hcg measurements; Eleven women women underwent salpingectomy and did not require follow up serum b-hcg measurements: and two women (7%) had persistent trophoblast, one of whom required a second laparoscopy. Information on the use of resources necessary to calculate direct costs was available for all 62 randomised women. The calculation of indirect costs was based on the use of resources in 32 of the 34 women (94%) randomised to methotrexate, and 26 of the 28 women (93%) randomised to laparoscopy. The four women who failed to complete their follow up questionnaires experienced no treatment complications or hospital readmission following treatment. Their indirect costs are likely to be similar to the costs of the women who were successfully followed up. Direct costs The main elements of the use of resources for each group and the unit costs used to value these resources are shown in Table 1. The direct costs per woman are shown in Table 2. The mean total direct costs per woman were signi cantly lower in the methotrexate group ($NZ 1470) than in the laparoscopy group ($NZ 3083), with a mean difference in total direct costs of $NZ 1613 (95% CI of difference 1166 to 2061). This is equivalent to a 52% reduction in direct costs per woman treated with methotrexate compared with laparoscopic management. The mean (range) number of working days lost in each treatment group which was used to calculate lost production by women, their partners and other carers, is shown Table 1. Thirty women in the methotrexate group and 21 women in the laparoscopy group were in paid employment. In the methotrexate group 28 women had a partner in paid employment and 27 identi ed other potential carers. In the laparoscopy group 22 women had a partner in paid employment and 20 women had other potential carers. Additional indirect costs consisted predominately of travel costs and expenditure on sanitary products, over the counter analgesics and complementary therapies. The indirect costs per woman are shown in Table 2. Indirect costs were signi cantly lower in the methotrexate group ($NZ 1141) compared with the laparoscopy group ($NZ 1899), with a mean difference in total mean indirect costs of $NZ 758 (95% CI of difference $NZ 277 to $NZ 1240). This is equivalent to a 40% reduction in indirect costs for women treated with methotrexate. Most of this difference was due to a signi cantly lower number of working days being lost by women treated with methotrexate (Table 1). Sensitivity and scenario analysis When a sensitivity analysis was undertaken to take account of uncertainties in the estimation of direct

5 208 M.C. SOWTER ET AL. Table 2. Direct and indirect costs per woman in the trial. Figs. are in $NZ. Mean difference (95% CI) is the mean cost of methotrexate therapy minus the mean cost of laparoscopic surgery. Methotrexate Laparoscopy Cost component Mean Median (range) Mean Median (range) Mean difference (95% CI) Direct costs Gynaecology assessment unit (114 to 456) (122 to 456) -6 (-46 to 34) Operative (0 to 1991) (966 to 2225) (-1455 to -987) Hotel (0 to 2285) (136 to 2293) -479 (-709 to -250) Investigations (214 to 788) (155 to 559) 4 (-46 to 54) Drugs (24 to 103) (17 to 80) 13 (3 to 23) Follow up (52 to 416) (52 to 416) 76 (17 to 135) Total (634 to 4976) (1717 to 4530) (-2061 to -1166) Lost paid production by patient (0 to 2832) (0 to 3172) -577 (-977 to -178) Lost unpaid production by patient (0 to 567) (113 to 3172) -774 (-2718 to -1170) Lost paid production by partner (0 to 1680) (0 to 1985) -162 (-410 to 87) Lost paid production by other carers 73 0 (0 to 1133) (0 to 1133) -80 (-217 to 57) Personal medical expenditure 24 2 (0 to 210) 18 0 (0 to 160) 6 (-8 to 19) Other expenses 12 0 (0 to 200) 12 0 (0 to 85) 0 (-22 to 22) Travel costs (2 to 213) (3 to 192) 22 (-8 to 52) Total (17 to 3963) (294 to 4267) -758 (-1240 to -277) costs, the differences in mean total direct costs were most sensitive to changes in the value of theatre and hotel costs. However, in all cases the mean differences in total direct costs were signi cantly higher in the laparoscopy group (Table 3), even when the unit costs of individual cost components were varied by 50%, 150% and 200% of base-case unit costs. When a sensitivity analysis was undertaken to determine if initial serum b-hcg concentrations would affect total direct and indirect costs the difference in direct costs was greatest at lower (under 1500 IU/L) serum b-hcg concentrations (Table 4). This was primarily due to reduced theatre and hotel costs in women treated with methotrexate. Methotrexate therapy was also associated with a greater reduction in indirect costs at lower serum b-hcg concentrations. However, at pretreatment serum b-hcg concentrations above 1500 IU/L there was no reduction in the indirect costs borne by women and their carers. A scenario analysis was also undertaken to determine the overall reduction in treatment costs per woman if all eligible women in our patient population of 218 women were treated with methotrexate, and also the effect of using an upper limit of 1500 IU/L and 1000 IU/L for medical therapy. In the randomised trial on which this economic evaluation was based 19 36% (79 of 218 women) of the total woman population were eligible for trial entry. If methotrexate rather than operative laparoscopy was used to treat all 79 of these women, then a potential reduction in direct costs of $NZ 580 for each of the 218 woman presenting with an ectopic pregnancy could be made (based on the mean difference in costs in Table 4). If an upper limit for pretreatment serum b-hcg of 1500 IU/L and 1000 IU/L was used, only 24% (53 women) and 20% (43 women) of our total woman population, respectively, would have been eligible for methotrexate therapy and the reduction in direct costs would have decreased to $NZ 422 and $NZ 386 per ectopic pregnancy, respectively. For indirect costs, the overall reduction in treatment costs per ectopic pregnancy would be $NZ 273, $NZ 264 and $NZ 248 for each scenario. The costs of treatment were available for 149 out of 156 women who were either not eligible for trial entry or declined randomisation. These women were treated by operative laparoscopy (97 women); laparoscopy followed by laparotomy (n ˆ 20); laparotomy without prior laparoscopy (n ˆ 20); methotrexate (n ˆ 7) or observation only (n ˆ 5). Major components of cost (theatre time and hospital in-patient stay) and mean total direct costs are shown in Table 5. DISCUSSION One of the principle driving forces behind moves away from laparotomy and towards laparoscopic and medical methods of treating ectopic pregnancy is a perception that there are economic gains to be made. From a provider's perspective, a shorter hospital stay or outpatient treatment has clear monetary bene ts. There may also be bene ts to the community from shorter convalescence, a more rapid return of women to work, and possibly greater potential for the preservation of future fertility 22. This study, conducted alongside a randomised trial 19, compared the economic consequences of using either a laparoscopic or a medical method treatment of unruptured ectopic pregnancy. We found that methotrexate

6 ECONOMIC EVALUATION OF SINGLE DOSE SYSTEMIC METHOTREXATE 209 Table 3. A sensitivity analysis showing changes in mean total cost differences where it is assumed for each cost component that costs are 50%, 150% and 200% of base-case costs. Figures are shown as mean differences (95% CI) in $NZ, that is, the mean total cost of methotrexate therapy minus the mean total cost of laparoscopy. Gynae assessment unit Operative Hotel Investigations Drugs Follow up 50% of base case (-2060 to -1160) (-1362 to -644) (-1706 to -1041) (-2053 to -1178) (-2064 to -1175) (-2104 to -1197) 150% of base case (-2062 to -1170) (-2770 to -1677) (-2427 to -1277) (-2069 to -1153) (-2058 to -1156) (-2019 to -1133) 200% of base case (-2064 to -1174) (3486 to -2184) (-2801 to -1385) (-2079 to -1139) (-2054 to -1146) (-1977 to -1097)

7 210 M.C. SOWTER ET AL. Table 4. A sensitivity analysis showing the effect of initial serum b-hcg concentrations on total direct and total indirect costs. Figures are in $NZ. Women have been divided into those with an initial serum b-hcg concentration below 1000 IU/l, below 1500 IU/l, and above 1500 IU/l. Methotrexate Laparoscopy Initial b-hcg level (IU/l) Mean Median (Range) Mean Median (Range) Mean difference (95% CI) Direct costs, 1000 (n ˆ 34) (634 to 2191) (1717 to 530) (-2402 to -1459), 1500 (n ˆ 44) (634 to 3196) (1717 to 4530) (-2215 to -1302) (n ˆ 18) (707 to 4976) (2655 to 3918) (-2358 to -184) All women (634 to 4976) (1717 to 4530) (-2061 to -1166), 1000 (n ˆ 31) (97 to 2393) (294 to 4267) (-1860 to -577), 1500 (n ˆ 40) (97 to 2393) (294 to 4267) (-1621 to -577) (n ˆ 18) (241 to 3963) (661 to 3325) -2 (-1052 to 1050) All women (97 to 3963) (294 to 4267) -758 (-1240 to -277) therapy was associated with a 52% reduction in direct costs and 40% reduction in indirect costs per woman treated, compared with laparoscopy in the base-case analysis. The main differences in costs within the trial that accounts for these reductions are reduced theatre usage and hospital stay. Duplicating these reductions in normal clinical practice would depend on a low rate of surgical intervention in women treated with methotrexate and the ability to continue their treatment entirely within an outpatient setting. Our sensitivity analysis demonstrated that differences in treatment costs between methotrexate and laparoscopy are greatest in women presenting at low pretreatment serum b-hcg concentrations, where there is a lower chance of requiring hospital readmission or surgical intervention. Both direct and indirect treatment costs are greater in women with higher pretreatment serum b-hcg concentrations. Larger numbers of women than were recruited to our study would be needed to determine reliably at what serum b-hcg concentration all savings in direct costs are lost. However, the marked differences in direct costs for our base-case analysis (when 15% of women treated with methotrexate required surgery and 9% suffered subsequent tubal rupture) suggest that the direct costs of the two treatments would only be equivalent at a clinically unacceptable rate of treatment failure for methotrexate. One other economic evaluation comparing methotrexate therapy with laparoscopy has attempted to stratify costs by initial serum b-hcg concentration 23. This study, based on a randomised trial comparing a multiple dose methotrexate regime with laparoscopy 24, concluded that because of an increasing rate of hospital readmission, medical therapy would become as expensive as operative laparoscopy for women with an initial serum b-hcg concentration above 3000 IU/L. Of concern in our study is the nding that for women with initial serum b-hcg concentrations above 1500 IU/L there were no savings in indirect costs. For these women the increased chance of hospital readmission, surgical intervention and a greater number of follow up visits, appear to eliminate any bene ts associated with medical therapy compared with operative laparoscopy. In our scenario analysis we estimated the effect on the cost of treating our total woman population if methotrexate therapy were restricted to women with an initial serum b-hcg concentration below 1500 IU/L or 1000 IU/L. This would potentially increase the cost savings per woman treated medically rather than surgically, but would also reduce the proportion of our total population eligible for medical therapy and consequently the overall potential savings. However, this reduction in overall potential savings needs to be weighed against the lower Table 5. Use of resources (theatre time and hospital stay only) and total direct costs per woman not in the trial. Resource use data were available for 149/156 women not in the trial. Values are given as mean (95% CI). Treatment a Theatre time (min) Total hospital stay (days) Post-op stay b (days) Total direct cost ($ NZ) Laparoscopy (n ˆ 97) 79 (74-83) 1.62 ( ) 1.56 ( ) 3083 ( ) Laparoscopy plus 93 (74-111) 3.66 ( ) 3.23 ( ) 4175 ( ) laparotomy (n ˆ 20) Laparotomy (n ˆ 20) 66 (56-75) 3.93 ( ) 3.57 ( ) 3976 ( ) Methotrexate (n ˆ 7) ± ± 0.88 ( ) ± ± 1759 ( ) Observation (n ˆ 5) ± ± 1.99 ( ) ± ± 1235 ( ) a b n ˆ number of women for whom data were available. Post-op stay is the time from surgery to discharge.

8 ECONOMIC EVALUATION OF SINGLE DOSE SYSTEMIC METHOTREXATE 211 chance of medically treated women requiring surgical intervention and the greater predictability of treatment outcome when medical therapy is restricted to women with lower initial serum b-hcg concentrations. A treatment that can be used ef ciently and safely in 20 to 25 percent of women with an ectopic pregnancy is more likely to become than one which is used in a larger proportion of women, but with the need for frequent surgical rescue. The ndings of this trial are similar to other economic evaluations 13±18 that have compared single dose systemic methotrexate with operative laparoscopy, although our study is the rst to based on a randomised trial and the rst to compare indirect costs. This study and all other economic evaluations published to date have been concerned only with relatively short term outcomes. Longer term economic outcomes after the treatment of tubal pregnancy, such as the need for assisted conception, also need to be considered 25,26. Tubal patency rates in excess of 80% have been reported in uncontrolled studies of single dose methotrexate therapy 9,27,28. However, the only randomised trial to report tubal patency rates after methotrexate and laparoscopic salpingotomy found no difference between medical and surgical treatment 24. Subsequent cumulative spontaneous pregnancy rates in this trial (available for 34 women after methotrexate and 40 women after salpingotomy) at two years post-treatment were also identical 29. As more data on long term outcomes after medical and surgical therapy become available the relative costs of each therapy may need to be reassessed, taking account of the costs of treatment for subsequent infertility. CONCLUSION This study demonstrates that 20% to 25% of women with ectopic pregnancies could be reliably treated using methotrexate with considerable reduction in both direct and indirect costs. Treating a larger proportion of women would be associated with a higher rate of hospital readmission and surgical intervention. Direct cost savings would still be made, but the increase in indirect costs means that methotrexate is only likely to be a preferable treatment if the serum b-hcg concentration is below 1500 IU/L. Acknowledgements We would like to thank Mr C. Foy, for statistical advice and the Finance Department at Auckland Healthcare for providing advice on the collection of data on the use of resources and the calculation of unit costs, and Associate Professor T. Ashton and Ms A. Heynes for advice on the design of this economic evaluation and analysis of the data collected. References 1. MaÈkinen J. Ectopic pregnancy in Finland : a massive increase. BMJ 1987;294:740± Coste J, Job-Spira N, Aublet-Cuvelier B. Incidence of ectopic pregnancy. First results of a population based register. Hum Reprod 1994;9:742± Department of Health. Report on Con dential Enquiries into Maternal Deaths in the United Kingdom London: HMSO, Centers for Disease Control. Ectopic Pregnancy in the United States Morb Mortal Wkly Rep 1995; 44: Gray DT, Thorburn J, Lundorff P, Strandell A, Lindblom B. A costeffectiveness study of a randomised trial of laparoscopy versus laparotomy for ectopic pregnancy. Lancet 1995;345:1139± Vermesh M, Silva PD, Rosen GF, Stein AL, Fossum GT, Sauer MV. Management of unruptured ectopic gestation by linear salpingostomy: a prospective, randomized clinical trial of laparoscopy versus laparotomy. Obstet Gynecol 1989;73:400± Brumsted J, Kessler C, Gibson C, Nakajima S, Riddick DH, Gibson M. A comparison of laparoscopy and laparotomy for the treatment of ectopic pregnancy. Obstet Gynecol 1988;71:889± Maruri F, Azziz R. Laparoscopic surgery for ectopic pregnancies: technology assessment and public health implications. Fertil Steril 1993;59:487± Stovall TG, Ling FW. Single dose methotrexate: an expanded clinical trial. Am J Obstet Gynecol 1993;168:1759± Henry MA, Gentry WL. Single injection of methotrexate for treatment of ectopic pregnancies. Am J Obstet Gynecol 1994;171:1584± Stika CS, Anderson L, Frederiksen MC. Single dose methotrexate for the treatment of ectopic pregnancy: Northwestern Memorial Hospital three-year experience. Am J Obstet Gynecol 1996;174:1840± Lipscomb GH, Bran D, McCord ML, Portera JC, Ling FW. Analysis of three hundred fteen ectopic pregnancies treated with single dose methotrexate. Am J Obstet Gynecol 1998;178:1354± Creinin MD, Washington AE. Cost of ectopic pregnancy management: surgery versus methotrexate. Fertil Steril 1993;60:963± Alexander JM, Rouse DJ, Varner E, Austin Jr JM. Treatment of the small unruptured ectopic pregnancy: a cost analysis of methotrexate versus laparoscopy. Obstet Gynecol 1996;88:123± Yao M, Tulandi T, Kaplow M, Smith AP. A comparison of methotrexate versus laparoscopic surgery for the treatment of ectopic pregnancy: a cost analysis. Hum Reprod 1996;11:2762± Hidlebaugh D, O'Mara P. Clinical and nancial analyses of ectopic pregnancy management at a large health plan. J Am Assoc Gynecol Laparosc 1997;4:207± Robin F, Lecuru F, Bernard JP, et al. Methotrexate provides signi cant cost savings for the treatment of unruptured ectopic pregnancy. Clin Drug Invest 1998;15:405± Morlock RJ, Lafata JE, Eisenstein D. Cost-effectiveness of single dose methotrexate compared with laparoscopic treatment of ectopic pregnancy. Obstet Gynecol 2000;95:407± Sowter MC, Farquhar CM, Petrie KJ, Gudex G. A randomised controlled trial comparing single dose systemic methotrexate and laparoscopic surgery for the treatment of unruptured ectopic pregnancy. Br J Obstet Gynaecol 2000;108:192± Van Roijen L, Essink-Bot M, Koopmanschap MA. Labor and health status in economic evaluation of health care. Int J Technol Assess Health Care 1996;12:405± Gardner M, Gardner S, Winter PD. Con dence Interval Analysis (CIA) Computer Programme Manual. London: British Medical Journal Publications, Sau AK, Auld BJ, Sau M. Current status of management of ectopic pregnancy. Gynaecol Endoscopy 1999;8:73± Mol BWJ, Hajenius PJ, Engelsbel S, et al. The treatment of ectopic pregnancy in the Netherlands: an economic evaluation of systemic methotrexate and laparoscopic salpingostomy. Am J Obstet Gynecol 1999;181:945±951.

9 212 M.C. SOWTER ET AL. 24. Hajenius PJ, Engelsbel S, Mol BWJ, et al. Randomised trial of systemic methotrexate versus laparoscopic salpingostomy in tubal pregnancy. Lancet 1997;350:774± Rulin MC. Is salpingostomy the surgical treatment of choice for unruptured tubal pregnancy? Obstet Gynecol 1995;86:1010± Mol BWJ, Hajenius P, Engelsbel S, et al. Is conservative surgery for tubal pregnancy preferable to salpingectomy? Br J Obstet Gynaecol 1997;104:834± SchaÈfer D, Pfuhl J-P, Baumann R. Systemic treatment of ectopic pregnancies with single dose methotrexate. J Am Assoc Gynecol Laparosc 1994;1:213± Corsan GH, Karacan M, Qasim S, Bohrer MK, Ransom MX, Kemmann E. Identi cation of hormonal parameters for successful systemic single dose methotrexate therapy in ectopic pregnancy. Hum Reprod 1995;10:2719± Dias Pereira G, Hajenius PJ, Mol BWJ, Ankum WM, Hemrika DJ, Bossuyt PMM, Van der Veen F. Fertility outcome after systemic methotrexate and laparoscopic salpingostomy for tubal surgery. Lancet 1999;353:724±725. Accepted 1 November 2000