UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Tubal subfertility and ectopic pregnancy. Evaluating the effectiveness of diagnostictestsMol, B.W.J.

Link to publication

Citation for published version (APA):Mol, B. W. J. (1999). Tubal subfertility and ectopic pregnancy. Evaluating the effectiveness of diagnostic tests

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: http://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 27 May 2018

7. Cost-effectiveness analysis of hysterosalpingography, laparoscopy, and Chlamydia antibody testing in subfertile couples

Ben W.J. Mol, Fulco Van der Veen, John A. Collins, Elizabeth A. Burrows, and Patrick MM. Bossuyt Submitted for publication

Abstract

Objedive: To compare strategies that can be used to detect tubal pathology in subfertile couples with respect to live birth rates, number of cycles of IVF-ET and total costs. Methods: Baseline characteristics of > 2,000 subfertile couples collected in the Canadian Infertility Treatment Evaluation Study (CITES) were used. Expectant management, i.e., no diagnosis or treatment, was considered to be the reference strategy. In strategy 2 and 3, IVF-ET was either offered immediately or after 2V2 years. In strategy 4, the decision to offer or delay treatment was based on the spontaneous conception chances of the couple, based on female age, duration of subfertüity, previous pregnancies and regularity of the menstrual cycle. Moreover, nine strategies were evaluated incorporating combinations of Chlamydia Antibody Testing (CAT), CA-125 measurement, hysterosalpingography (HSG) and laparoscopy. For each strategy, we calculated expected live birth rates, expected number of IVF-ET cycles and expected total costs. Expected spontaneous live birth rates were obtained from CITES. Expected IVF-ET success rates were obtained from a cohort study in the UK. Results: Without treatment, the 3-year expected cumulative live birth rate was 13%, whereas it varied between 34% and 49% for other strategies. Costs of these strategies varied between US$ 19,800 and US$ 27,500 per couple. The strategy in which the decision to perform laparoscopy either immediately or after one year, depending on the result of CAT, was the most cost-effective strategy, whereas the strategy in which this decision was based on the result of HSG was almost as cost-effective. Sensitivity analysis showed that the strategy starting with CAT was the most cost-effective in couples in which 3-year conception chances were > 14%, whereas the strategy starting with HSG was the most cost-effective in couples with worse fertility prospects. Use of serum CA-125 measurement was only cost-effective in case fertility prospects were very poor. Conclusion: The diagnostic work-up to detect tubal pathology in subfertile couples should start with CAT in couples with relatively good fertility prospects, whereas couples with relatively poor fertility prospects benefit from a strategy starting with immediate HSG.

69

Chapter 7

7.1 Introduction

Tubal pathology is a main cause of subfertility. Laparoscopy and hysterosalpingography (HSG) are commonly used to visualize tubal pathology. Since pelvic inflammatory disease is a major cause of tubal pathology, Chlamydia Antibody Testing (CAT) is also used to detect women with tubal pathology.1 2 An obvious disadvantage of CAT is that it provides no information on the severity of tubal pathology, since it does not visualize the fallopian tubes. In contrast, HSG and laparoscopy can discriminate between proximal tubal occlusion, distal tubal occlusion and hydrosalpinx of bodi tubes separately.3

Laparoscopy with chromopertubation is considered to be the most reliable tool in the diagnosis of tubal disease, and is therefore often considered as the reference test in the evaluation of the diagnostic performance of other tests. Sensitivity and specificity of HSG in the diagnosis of tubal occlusion are 65% and 83%, respectively, as compared to laparoscopy {chapter J).4 The discriminative performance of CAT for the diagnosis of any tubal pathology is comparable to that of HSG {chapter 4).lz With respect to prognosis, double-sided tubal occlusion detected at HSG hampers fertility prospects severely, whereas one-sided tubal occlusion at HSG has only limited impact, as was shown in chapter 5} Fertility prospects of women with one-sided and two-sided tubal padiology detected at laparoscopy are even worse as compared to those of women with comparable HSG findings. However, in chapter 6 it was demonstrated that even double-sided occlusion at laparoscopy is not a perfect predictor of infertility.6

In addition to its diagnostic impact, HSG with oil-soluble contrast media is also considered to improve fertility prospects by hydrotubation of the fallopian tubes. In a meta­analysis, Watson et aL showed HSG with oil-soluble contrast media to improve fertility prospects compared to HSG with water-soluble contrast.7 The only randomized study reporting on fertility after HSG using oil-soluble contrast media as compared to fertility in patients not undergoing HSG showed an almost 3-fold increase of the pregnancy rates after HSG with oil-soluble contrast media.8 We are not aware of randomized studies reporting on fertility in women who had HSG using water-soluble contrast media as compared to women who had no HSG done. It is possible that HSG with water-soluble contrast media also has a beneficial effect on the fertility prospects. Similarly, laparoscopy might have a beneficial effect, but studies comparing 'treatment independent' fertility rates after laparoscopy and HSG are also lacking.

In contrast to HSG, laparoscopy allows diagnosis and treatment of mild and moderate endometriosis. Ablation of endometriotic leasions almost doubles the probability of spontaneous pregnancy in women with this disease.9 Since treatment of endometriosis is important, early detection of this disorder with serum CA-125 measurement might be useful.10

In view of diese data, the question arises which strategy should be recommended to assess tubal pathology in subfertile women. From a therapeutic perspective, IVF-ET is the treatment of choice in women with double-sided tubal pathology. Among women suspected of having tubal pathology, some have a low chance of spontaneous pregnancy and might profit from immediate IVF-ET, whereas others still have a reasonable chance

70

Cost-effediveness of diagnosis of tubal pathology

to conceive spontaneously. Thus, various diagnostic tests, such as HSG, laparoscopy or CAT could be used to distinguish couples that might benefit from immediate IVF-ET from couples for which expectant management is more appropriate, followed by IVF-ET when conception does not occur. The purpose of the present study was to compare both costs and effectiveness of various strategies in the work-up of subfertile couples suspected of having tubal pathology.

7.2 Materials and methods

The analysis was performed using the prognostic profile of 2,167 couples with complete data that were collected in CITES.11 The expected number of live births and the expected total costs of 13 strategies were calculated. The strategies are summarized in Table 1. In strategy 1 the couple was managed expectantly, i.e., no diagnosis or treatment was performed. This strategy was considered to be the reference strategy, to which all otiier strategies were compared. In strategy 2, three cycles IVF-ET were performed immediately in each couple. In strategy 3, each couple was managed expectantly for 2Vi years, and IVF-ET was subsequently offered to couples that had not conceived spontaneously within that period.

In strategy 4, IVF-ET was delayed for 2Vz years in case the expected 3-year treatment-independent conception rate resulting in the five birth of a child exceeded 15%, whereas it was offered immediately in all other cases. The expected treatment independent conception rate was based on female age, duration of subfertility, primary or secondary sub fertility and regularity of the menstrual cycle.

In strategy 5, HSG was performed in all women. The results of HSG were classified as normal, one-sided abnormal or two-sided abnormal. The expected treatment-independent conception rate resulting in live birth was then recalculated, taking into account the issues considered in strategy 4 and the results of HSG. Again, IVF-ET was delayed for 2Vz years in case the expected 3-year treatment-independent conception rate resulting in the live birth of a child exceeded 15%, whereas it was offered immediately in all other cases.

In strategy 6, laparoscopy was performed in all women. Tubal pathology at laparoscopy was classified as normal, one-sided abnormal or two-sided abnormal. Endometriosis was scored according to the classification of the American Fertility Society: no endometriosis, endometriosis grade I/11 and endometriosis grade III/IV.1 2 Endometnotic lesions grade I / I I were supposed to be treated with ablation.9 Furthermore, tubal abnormalities were supposed to be treated in order to improve the IVF-ET success rates.13 The expected treatment-independent conception rate resulting in live birth was recalculated as described in strategy 4, but using die results of laparoscopy in stead of HSG. Similarly, IVF-ET was delayed for 2Vz years in case the expected 3-year treatment-independent conception rate resulting in the live birth of a child exceeded 15%, whereas it was offered immediately in all other cases.

In strategy 7, the diagnostic work-up started always with CAT, whereas HSG was performed only in case the probability of tubal pathology exceeded 15%. The probability of tubal pathology was calculated from a logistic regression model incorporating female age,

71

Chapter 7

duration of subfertility, primary or secondary subfertility, regularity of the menstrual cycle and the result of CAT. In case HSG had been performed, the results were incorporated in the calculations of the expected pregnancy rates. In case HSG had not been performed, the expected pregnancy rates were solely determined from female age, duration of subfertility, primary or secondary subfertihty, regularity of the menstrual cycle and CAT. In vitro fertilization and embryo-transfer was delayed for "2}/z years in case the expected 3-year treatment-independent conception rate resulting in the live birth of a child exceeded 15%, whereas it was offered immediately in all other cases.

In strategy 8, the diagnostic work-up started always with CAT, whereas laparoscopy was performed in case the probability of tubal pathology exceeded 15%. The probability of tubal pathology was again calculated from a logistic regression model incorporating female age, duration of subfertihty, primary or secondary subfertility, regularity of the menstrual cycle and the result of CAT. Subsequent decision making was based on the expected pregnancy rates after expectant management, as described in strategies 4 to 7. In case laparoscopy had been performed, the result of this test was incorporated in the calculations of the expected pregnancy rates. In case laparoscopy had not been performed, the expected pregnancy rate was solely determined from female age, duration of subfertility, primary or secondary subfertihty, regularity of the menstrual cycle and the result of CAT. In patients in whom laparoscopy had been performed, endometriosis was supposed to be treated with ablation, whereas tubal abnormalities were supposed to be corrected in order to improve the success rates of IVF-ET.13

In strategy 9, all patients underwent serum CA-125 measurement, and laparoscopy was offered to patients in whom the risk of endometriosis was > 15%. The probability of endometriosis was calculated from a logistic regression model incorporating female age, duration of subfertihty, primary or secondary subfertihty, regularity of the menstrual cycle and serum CA-125 measurement. In case laparoscopy had been performed, the findings were incorporated in die calculations of the expected pregnancy rates. Otherwise, the expected pregnancy rates were solely determined from female age, duration of subfertihty, primary or secondary subfertihty, regularity of the menstrual cycle and serum CA-125 level. In vitro fertilization and embryo-trans fer was delayed for 2Vi years in case the expected 3-year treatment-independent conception rate resulting in live birth was > 15%, whereas it was offered immediately in case this treatment independent conception rate was < 15%. In patients in whom laparoscopy had been performed, endometriosis was supposed to be treated with ablation, whereas tubal abnormalities were supposed to be corrected in order to improve the IVF-ET success rates.13

In strategy 10, HSG was performed in each patient Laparoscopy was performed in case the probability of tubal pathology after HSG exceeded 15%. Subsequent decision making was again based on the expected pregnancy rates after 3 years of expectant management. Again, endometriosis and tubal pathology detected at laparoscopy were supposed to be treated surgically.913

In each of the strategies 8, 9, and 10, laparoscopy was only performed at the start of the diagnostic work-up. In clinical practice, however, laparoscopy is often delayed after normal

72

Cost-effediveness of äagnosis of tubal pathology

Table 1: The 13 evaluated strategies. Strategy Diagnosis Intervention

1. No treatment (reference) 2. Immediate IVF-ET 3. Delay IVF-ET for three

years 4. Immediate or delayed IVF-

ET, based on baseline characteristics

5. HSG and IVF-ET

None None None

None

HSG

6. Laparoscopy and IVF-ET Laparoscopy

7. CAT, HSG, and IVF-ET CAT, HSG in case probability of tubal pathology > 0.15

8. CAT, immediate CAT, laparoscopy in case laparoscopy, and IVF-ET probability of tubal pathology

>0.15

9. CA-125, immediate CA-125, laparoscopy in case laparoscopy, and IVF-ET probability of endometriosis >

0.15

10. HSG, immediate HSG, laparoscopy in case laparoscopy, and IVF-ET probability of tubal pathology

>0.15

11 .CAT, immediate or delayed laparoscopy, and IVF-ET

12. CAT, CA-125, immediate or delayed laparoscopy, and IVF-ET

13.HSG, immediate or delayed laparoscopy, and IVF-ET

CAT, laparoscopy in case probability of tubal pathology > 0.15 or in case pregnancy does not occur within 1 year CAT, CA-125, laparoscopy in case probability of tubal pathology or endometriosis > 0.15 or in case pregnancy does not occur within 1 year HSG, laparoscopy in case probability of tubal pathology

> 0.15 or in case pregnancy does not occur within 1 year

None Immediate 3 cycles IVF-ET in all couples 3 cycles IVF-ET in case successful pregnancy does not occur in three years Immediate 3 cycles IVF-ET in all couples in case expected live birth rate < 0.15, otherwise 3 cycles IVF-ET after three years Immediate 3 cycles IVF-ET in case in case expected live birth rate after HSG < 0.15, otherwise IVF-ET after 2Vz years Immediate 3 cycles IVF-ET in case in case expected live birth rate after laparoscopy < 0.15, otherwise IVF-ET after 2Vi years Immediate 3 cycles IVF-ET in case expected live birth rate after HSG < 0.15, otherwise 3 cycles IVF-ET after 2lA years. Immediate 3 cycles IVF-ET in case expected live birth rate after laparoscopy < 0.15, otherwise 3 cycles IVF-ET after 2Vz years. Surgical treatment of endometnosis. Immediate 3 cycles IVF-ET in case expected live birth rate after laparoscopy < 0.15, otherwise 3 cycles IVF-ET after 2V4 years. Surgical treatment of endometriosis Immediate 3 cycles IVF-ET in case expected live birth rate after laparoscopy < 0.15, otherwise 3 cycles IVF-ET after XA years. Surgical treatment of endometriosis. Immediate 3 cycles IVF-ET in case expected live birth rate after laparoscopy < 0.15, otherwise 3 cycles IVF-ET after 21/2 years. Surgical treatment of endometriosis. Immediate 3 cycles IVF-ET in case expected live birth rate after laparoscopy < 0.15, otherwise 3 cycles IVF-ET after 2Vz years. Surgical treatment of endometriosis.

Immediate 3 cycles IVF-ET in case expected live birth rate after laparoscopy < 0.15, otherwise 3 cycles IVF-ET after 2'/2 years. Surgical treatment of endometriosis.

HSG or normal CAT, but it still will be performed in case pregnancy has not occurred within a reasonable period. We therefore also analyzed adjusted versions of strategies 8, 9 and 10, in which laparoscopy was offered 1 year after start of the diagnostic work-up to those couples who did not have a laparoscopy done immediately and who were not pregnant after 1 year.

73

Chapter 7

In strategy 11, CAT was always performed, and laparoscopy was performed immediately in those patients in whom the probability for tubal pathology was > 15%. In case laparoscopy was not performed immediately, it was offered to those women who were not pregnant at one year after start of the diagnostic work-up. Decision making on the performance of IVF-ET was comparable to that of previous strategies, i.e., IVF-ET was performed in case the probability for live birth was < 15%. In strategy 12, CA-125 measurement and CAT were performed in all patients, and laparoscopy was offered immediately to those patients in whom the probability for tubal pathology and or endometriosis was > 15%. In case laparoscopy was not performed immediately, it was offered to those women who were not pregnant one year after start of the diagnostic work­up and decision making on IVF-ET was comparable to that of previous pregnancies.

In strategy 13, HSG was performed in all patients, and laparoscopy was performed immediately in those patients in whom the probability for tubal pathology was > 15%. In case laparoscopy was not performed immediately, it was offered to those women who were not pregnant one year after start of the diagnostic work-up and decision making on IVF-ET was comparable to that of previous strategies.

Analysis

The aim of the analysis was to calculate the expected number of live births after 3 years, the expected number of cycles of IVF-ET and the expected total costs for each of the 13 strategies. We used the prognostic profile of 2,167 couples that were included in CITES.11

For each couple data were available on female age, duration of subfertility, previous pregnancies, regular menstrual cycle and tubal status as assessed at HSG and/or laparoscopy, and endometriosis, as assessed at laparoscopy.

Tubal status at HSG and laparoscopy was classified as either normal, one-sided tubal pathology or two-sided tubal pathology. Endometriosis at laparoscopy was classified according to the classification of the American Fertility Society: no endometriosis, endometriosis grade I / I I or endometriosis grade III/TV.12

The results of CAT- and CA-125 measurement were not available in the CITES data-set These results were generated from the findings at laparoscopy and from conditional distributions reported in the literature. Results of CAT for patients with two-sided tubal pathology, for patients with one-sided tubal pathology and for patients without tubal pathology were drawn from the distribution of results published recendy by Land et aL2

Results of CA-125 measurement of patients with endometriosis grade III/TV, endometriosis grade I / I I and patients without endometriosis were drawn from the distribution of results reported in a recent meta-analysis assessing the performance of CA-125 in the diagnosis of endometriosis.6 The random number generator in the SAS® statistical software package was used to sample values from the respective distributions.

For calculation of the probability of a spontaneous live birth within the 36 upcoming months, we used a prognostic model for spontaneous live birth that has been published previously.14 We also calculated for each couple the expected live birth rate for 3 consecutive cycles IVF-ET, either performed immediately, or after 2Vi years, using the prognostic model published by Templeton et alls In the study of Templeton, tubal

74

Cost-effectiveness of àagnosis of tubal pathology

pathology reduced the success rates of IVF-ET. If in one of the strategies laparoscopy had been performed prior to IVF-ET, tubal pathology was supposed to be treated, and success-rates of IVF-ET were supposed to be identical to diose of comparable couples without tubal pathology.13

Hysterosalpingography was supposed to be performed with oil-soluble contrast. In case laparoscopy was performed, the tubes were supposed to be flushed with oil-soluble contrast media after the chromopertubation with Methylene Blue. As soon as HSG or laparoscopy had been performed, fertility prospects were supposed to increase with 30% as compared to the situation in which patency tests had not been performed.7 Treatment of endometriosis grade I / I I during laparoscopy was supposed to increase the fertility prospects with 80%.9

The costs of HSG, laparoscopy, CAT- and CA-125 measurement were obtained from a recent study on the cost of infertility diagnosis and treatment in Canada, and converted into US dollars.16 They were US$ 81 for a HSG, US$ 793 for laparoscopy, US$ 20 for CAT and US$ 20 for CA-125 measurement Costs of IVF-ET were calculated by adding up the costs of one cycle IVF-ET, the cost of hospital stay due to pre-term labor and costs of neonatal intensive care. The costs of IVF-ET were US$ 8.000 per cycle17, the expected average costs of hospitalization for preterm labor and subsequent neonatal care were US$ 9,850 and US$ 38,000 per woman for women with singleton and twin pregnancies, respectively.18 Costs of handicap care due to preterm delivery were not taken into account. The probability of a twin pregnancy in case of IVF-ET was supposed to be 25%.19

For the strategies 4 to 10, i.e., the strategies in which the decision to delay IVF-ET or to offer it immediately was based on the prognostic profile of the patient, we calculated the percentage of couples in which the decision to delay IVF-ET was changed into immediate offering of IVF-ET. Subsequently, the mean total costs to establish one additional live birth were calculated for each of the other 12 strategies, using expectant management as the reference strategy. Additionally, we calculated the number of cycles IVF-ET that had to be performed to establish one additional live birth. Twin deliveries were considered to be a single count

There is broad agreement that - in cost-effectiveness analysis - all future costs and health consequences should be expressed m terms of their "present value".20 For health care expenditures, future costs should be discounted: costs made in the future should be valued less than costs made in the present. The discount rate for costs was set on 5% per year in the initial analysis, in agreement with recent guidelines for reporting cost-effectiveness analyses.20 Similarly, a conception resulting in the live birth of a child within 9 months will be valued higher than a conception resulting in the live birth in the future, and the occurrence of live birth should therefore also be discounted. However, the individual variation in this discount rate is likely to be stronger than the variation of the discount rate for costs. In other words, whereas one couple desperately wants a child now, another couple might be virtually indifferent between a child now or a child in 3 years. Since there is no standard solution for this dilemma, we discounted the live birth of a child with 10% per year in the initial analysis.

75

Chapter 7

Table 2: Impact on the decision to perform IVF-HT immediately or delay itfor2'/2 years it t the 10 evaluated strategies

Strategy % couples % couples in % couples in whom IVF- % couples that are offered

that is whom ET is delayed in strategy immediate IVF-ET in

immediately IVF-ET is 4, but that are offered strategy 4, but in whom

offered IVF- delayed immediate IVF-ET IVF-ET is delayed

ET after additional testing after additional testing

1. No treatment (reference) 0 0 - -2. Immediate IVF-ET 100 0 - -3. Delay IVF-ET for three years 0 100 - -4. Immediate or delayed IVF-ET, 48 52 - -

based on baseline charactenstics" 5. HSG and IVF-ET 53 47 21 16 6. Laparoscopy and IVF-ET 49 51 20 19 7. CAT, HSG and IVF-ET 49 51 17 16 8. CAT, laparoscopy, and IVF-ET 49 51 17 16 9. CA-125, laparoscopy, and IVF-ET 35 65 4 17 10. HSG, laparoscopy and, IVF-ET 48 52 17 17 * Baseline characteristics considered were female age, duration of subfertility, primary or secondary

subfertility, regularity of the menstrual cycle.

Sensitivity analysis The initial analysis calculated the mean cost-effectiveness for all couples in the CITES

study. To evaluate if specific characteristics affected the cost-effectiveness of the evaluated strategies, we also calculated the mean cost-effectiveness for different categories of maternal age, for different categories of duration of subfertility, for patients with and without a regular menstrual cycle, and for patients with and without a previous pregnancy. Furthermore, we calculated for each couple the expected 3-year cumulative conception rate resulting in live birth, and then calculated the mean cost per additional live birth as a function of the baseline fertility chances.

7.3 Results

Table 2 shows the consequences of strategies 1 to 10 on the decision either to offer IVF-ET immediately or to delay treatment for at least T-h years. In strategy 4, where the decision to offer or withhold IVF-ET is based on the prognosis for spontaneous conception, IVF-ET would be immediately offered to 48 % of the couples, due to the fact that in these couples the probability of spontaneous conception resulting in live birth is < 15%.

In case decision making is based on the prognostic profile of the patient and the results of HSG (strategy 5), IVF-ET is immediately offered to 53% of the couples. The difference with strategy 4 is explained by the fact that 2 1 % of the couples, in whom IVF-ET was delayed in strategy 4, are now offered IVF-ET immediately, whereas in 16% of the couples, in whom IVF-ET would have been performed immediately, IVF-ET would be now delayed. For strategy 6, laparoscopy in all patients, these percentages are comparable.

In strategy 9, in which the decision for laparoscopy is based on the probability for endometriosis taking into account the result of CA-125 measurement, IVF-ET would be

76

Cost-effectiveness of'diagnosis of tubal pathology

offered to only 35% of the couples. In strategy 7, 8 and 10, CAT and HSG, CAT and laparoscopy and HSG and laparoscope IVF-ET would immediately be offered to 49%, 49% and 48% of the couples, respectively.

Table 3 shows the expected mean number of live births, the expected mean number of cycles IVF-ET and the expected mean costs for all 13 strategies, as well as the expected mean number of cycles IVF-ET and the expected mean costs per additional live birth for the strategies 2 to 13, using strategy 1 (expectant management) as the reference strategy. N o treatment at all would result in a live birth rate of 12.5 per 100 couples. Immediate IVF-ET in all couples would result in a 3 year cumulative live birth rate of 49.1 per 100 couples, whereas a 2Vi year delay of IVF-ET would result in live birth rate of 34.1 per 100 couples. Expected live birth rates in the strategies 4 to 13 varied between 41.2 and 47.0 per 100 couples.

The expected number of cycles IVF-ET per 100 couples would be 241 in case IVF-ET had been performed immediately in all couples, and 165 in case IVF-ET had been delayed for 2Vz years. In the strategies 4 to 13 the expected number of cycles IVF-ET would have varied between 192 and 206 per 100 couples. For the expected total costs a similar pattern was observed: US$ 27,500 per couple in case rVF-ET would be offered to all couples, US$ 19,800 per couple in case IVF-ET would be delayed for three years, and between US$ 22,600 and US$ 23,400 per couple in the strategies 4 to 13.

Table 3. Strategy Expected Expected Mean costs per Mean no of Mean costs per

live births no of "cycles 100 couples Cycles TVF- additional live

after 3 year rVF-ETper (USS100.000) ETper birth

per 100 couples 100 couples additional Eve birth

(US S 1.000)

1. No treatment (reference) 12.5 0 0 Reference Reference

2. Immediate IVF-ET 49.1 241 27.5 6.6 75.1

3. Delay IVF-ET for 2Vi years 34.1 165 19.8 7.6 91.7

4. Immediate or delayed IVF-ET, 41.4 199 23.4 6.9 81.0 based on baseline characteristics

5. HSG & IVF-ET 43.3 199 22.6 6.5 73.4

6. Laparoscopy & IVF-ET 43.7 192 22.6 6.2 72.4 7. CAT, HSG, & IVF-ET 42.4 201 23.4 6.7 78.3 8. CAT, immediate laparoscopy, & 42.3 199 23.4 6.7 78.5

IVF-ET 9. CA-125, immediate laparoscopy, 41.8 199 23.3 6.8 79.5

&IVF-ET 10. HSG, immediate laparoscopy, & 41.2 197 23.2 6.9 80.8

IVF-ET 11. CAT, immediate or delayed 46.7 209 22.8 6.1 66.7

laparoscopyl & IVF-ET 12. CAT, CA-125, immediate or 41.6 206 22.6 7.1 111

delayed laparoscopy, & IVF-ET 13. HSG, immediate or delayed 47.0 204 23.2 5.9 67.2

laparoscopy, & IVF-ET

77

Chapter 7

Immediate IVF-ET would result in 6.6 cycles per additional live birth, and delayed IVF-ET would result in 7.6 cycles per additional live birth. All strategies incorporating a test or a combination of tests required fewer cycles IVF-ET per additional live birth as compared to delay of IVF-ET in all patients. However, only strategy 11 and strategy 13 required less IVF-ET cycles per additional pregnancy as compared to immediate IVF-ET.

In terms of costs per additional live birth, immediate IVF-ET would cost US$ 75,100 per additional live birth, and delay of IVF-ET in all patients would cost US$ 91,700 per additional live birth. Again, each strategy using diagnostic tests was more cost-effective than expectant management. However, only strategies 11 and 13 were more cost-effective than strategy 2 (immediate IVF-ET for all patients). Strategy 11 was the most cost-effective treatment, with US$ 66,700 per additional live birth. Strategy 13 was almost as cost-effective, US$ 67,200 per additional live birth.

Sensitivity analysis Table 4 shows the results of the sensitivity analysis. The expected cost per additional

depended strongly on female age in each of the 13 strategies. In women under 34 years of age, strategy 11 (CAT with either immediate laparoscopy or laparoscopy after 1 year) was the most cost-effective strategy. In women over 34 years of age, strategy 13 (HSG with either immediate laparoscopy or laparoscopy after 1 year) was the most cost- effective strategy. Duration of subfacility had limited impact on the cost-effectiveness of the strategies, although the cost per additional live birth slightly increased in case the duration of subfertility increases.

Each strategy was less cost-effective in couples with primary subfertility as compared to couples with secondary subfertility. For couples with primary subfertility, HSG widi either immediate laparoscopy or laparoscopy after one year (strategy 13) was the most cost-effective strategy, whereas for couples with secondary subfertility CAT with either immediate laparoscopy or laparoscopy after one year (strategy 11) is the most cost-effective

Table 4: Sensitivity analysis evaluating the impact oj'female age, duration of subfertiüty, primary or secondary subfertility

and regularity of the menstrual cycle on the mean costs per additional üve birth.

Expected mean costs per additional live birth (US $ 1.000) Strategy Female age Duration of subfertility Primary Regular

(years) (months) subfertility cyclus <30 30-3435-38 >39 <24 25-36 37-48 >48 yes no yes no

2. Immediate IVF-ET 75 72 78 101 74 71 77 80 88 52 77 70 3. Delay IVF-ET for 3 years 82 85 111 181 87 92 90 98 111 57 93 88 4. Baseline characteristics 80 77 84 109 81 81 79 81 101 52 80 83 5. HSG 69 70 80 105 69 72 77 78 85 48 75 69 6. Laparoscopy 68 69 80 105 67 71 75 80 85 47 73 72 7. CAT & HSG 77 75 83 104 78 78 78 80 96 52 78 81 8. CAT & laparoscopy 77 75 82 103 77 78 78 81 96 52 78 81 9. CA-125 & laparoscopy 78 76 83 104 79 79 79 81 98 52 79 82 10. HSG & laparoscopy 80 77 84 108 81 81 80 81 100 52 80 83 11. CAT & laparoscopy 63 64 73 92 60 64 69 78 80 42 68 63 12. CAT, CA-125 & laparoscopy 71 68 86 126 74 73 75 84 89 47 79 71 13. HSG & laparoscopy 64 65 72 87 64 64 68 74 77 47 68 64

78

Cost-effectiveness of diagnosis of tubal pathology

strategy. Presence or absence of a regular cycle had limited impact on the cost-

effectiveness of the strategies. Figure 1 shows the impact of differences in the baseline prognosis on the cost-

effectiveness of strategies 2 to 4 and 11 to 13. The figure shows that the strategy using CAT to decide whether laparoscopy should be performed either immediately or should be delayed with 1 year, was expected to be the most cost-effective strategy for couples in whom the 3 year cumulative probability for pregnancy resulting in live birth was > 14%. For couples with a baseline prognosis between 5% and 14%, die strategy using HSG to decide if laparoscopy should be performed either immediately or should be delayed with one year, was expected to be the most cost-effective strategy. Addition of CA-125 measurement was only useful for couples in whom die probability for live birth was < 5%. Immediate IVF-ET was as cost-effective as strategies incorporating immediate laparoscopy in couples in whom the probability for live birth is < 2%.

Table 5 shows the effect of variation of the most important parameters. An increase of the discount rate for costs made strategies in which IVF-ET was delayed more cost-effective. At a discount rate for costs of 20%, delay of IVF-ET would be almost as cost-effective as strategies 11 and 13, in which the decision for laparospcopy was based on CAT and HSG, respectively.

The effect of variation of the discount rate of pregnancy was rather the opposite way. If the discount rate for live birth was 0%, the difference in cost-effectiveness between the stra-

Figure 1: The cost per additional live birth for six strategies as function of the baselinefertility prognosis

110-

a7°-£ 60

"3 50-

S 30-o u

20-

10-

0-0%

strategy 2; Immediate IVF-ET strategy 3; Delay IVF-ET for 3 years strategy 4; Decision based on baseline characteristics

• strategy 11; Decision based on CAT & laparoscopy • strategy 12; Decision based on CAT, CA-125 & laparoscopy • strategy 13; Decision based on HSG & laparoscopy

I 5% 10% 15% 20% 25%

Baseline probablity fot treatment independent conception resulting in live birth within 3 years

->—'—r~ 30%

79

Chapter 7

tegies decreases, and the additional value of diagnostic tests appeared to be limited. In contrast, if a pregnancy resulting in live birth in one year was valued less than a pregnancy resulting in live birth now, the cost-effectiveness of all strategies in which treatment was delayed, decreased. At a discount rate for live birth of 20%, immediate IVF-ET was the most cost-effective strategy.

Variation of the direct therapeutic effect of HSG and laparoscopy by Pertubation of the fallopian tubes had some impact on the cost-effectiveness of strategies incorporating HSG and/or laparoscopy. However, even in absence of such a therapeutic effect, strategies 11 and 13, in which the decision for laparoscopy was based on CAT and HSG, would remain the two most cost-effective strategies.

7.4 Discussion

This study assesses the cost-effectiveness of various strategies to detect tubal pathology m subfertile couples. In the population included in the CITES study, the strategies in which either CAT or HSG was used to decide whether laparoscopy should be performed immediately or postponed for one year in case ongoing pregnancy had not occurred, were the most cost-effective ones. The expected 3-year cumulative live birth rate was 47% in both strategies, resulting in a cost-effectiveness of approximately US $ 67,000 per additional live birth, if expectant management was used as the reference strategy. The sensitivity analysis indicated that changes in prognostic parameters female age, duration of subfertility, previous pregnancies and a regular menstrual cycle did not make another strategy more cost-effective. However, the strategy in which the decision to perform or postpone laparoscopy was based on HSG-results was more cost-effective in couples with poor spontaneous conception chances, whereas the strategy in which this decision was based on CAT was more cost-effective in couples in whom the spontaneous conception chances were relatively high.

Table 5: Sensitivity analysis evaluating the impact of variation of "discounting of costs, discounting of üve birth and

effectiveness of the therapeutic impact of HSG. The bold numbers represent the results of the baseüne analysis.

Strategy Expected mean costs per additional live birth (US $ 1.000) Discount rate costs Discount rate live birth RR HSG/lapararoscopy'

2. Immediate IVF-ET 3. Delay IVF-ET for 3 years 4. Baseline characteristics 5. HSG 6. Laparoscopy 7. CAT & HSG 8. CAT & laparoscopy 9. CA-125 & laparoscopy 10. HSG & laparoscopy 11. CAT & laparoscopy 12. CAT, CA-125, & laparoscopy 13. HSG & laparoscopy *RR HSG/laparoscopy = Relative Risk of HSG or laparoscopy on live birth using oil-soluble contrast media.

80

0% 10% 20% 0% 5% 15% 20% 1.0 1.1 1.2 75.9 74.7 73.4 75.6 75.4 75.0 75.0 75.1 75.1 75.1

106.3 78.3 55.6 67.5 78.3 108.1128.9 91.7 91.7 917 86.8 75.4 65.9 71.9 76.3 85.4 90.1 81.0 81.0 81.0 77.9 69.1 61.7 64.8 69.0 77.6 82.3 80.9 78.2 75.7 77.5 68.3 60.7 64.0 68.3 77.4 82.1 81.1 78.2 75.4 85.4 74.4 65.2 70.8 75.2 84.2 88.8 80.9 80.4 80.0 84.3 73.8 65.1 70.3 74.5 83.2 87.6 80.7 80.1 79.4 86.9 75.0 65.2 70.9 75.7 85.8 91.0 81.9 81.5 81.1 84.0 73.4 65.0 70.2 74.3 82.9 87.3 80.6 80.0 79.3 70.4 63.0 56.0 58.8 62.6 71.0 75.9 71.7 70.0 68.2 81.8 73.1 65.0 70.2 73.8 81.1 84.7 85.2 82.5 79.9 71.1 63.0 55.0 59.8 63.2 71.2 75.7 73.6 71.3 69.1

Cost-effectiveness of diagnosis of tubal pathology

We used previously published prognostic models to compare several strategies in the work-up for subfertility, and applied them on an existing cohort of subfertile couples.111+

15 Alternatively, we could have assessed costs and effects in theoretical cases. Such an analysis would have had several disadvantages. First, the number of possible combinations of patient profiles is very large, thereby hampering a report on possible combinations. Second, the cost-effectiveness reported in the present study is based on patient characteristics observed in real patients, and the frequency distribution of the couples with different prognostic profiles is therefore concordant with the one encountered in clinical practice.

Unfortunately, results of CA-125 and CAT measurement were not available in the CITES study. We therefore used the association between laparoscopic findings on one hand and CA-125- and CAT levels on the other hand, to generate hypothetical results of CA-125 and CAT measurements in couples included in CITES. Since the association between the results of CA-125 and CAT measurements and other characteristics, such as a history of dysmennorhea or salpingitis, is not quantified, the diagnostic value of these features could not be evaluated in our analysis. Future studies providing data on these associations are required to assess more complicated diagnostic strategies that incorporate data from the medical history.

There is considerable debate about the validity of prognostic models for spontaneous conception or IVF-ET.21 Most models have not been validated in other populations, and the few models that have been validated externally performed disappointingly.21 The models used in the present analysis are by our knowledge the best available sofar. Although our model on spontaneous conception was developed in the CITES-data, from which we also derived the prognostic profiles, there is no risk for overfitting, since the prediction of the model was not related to the real fertility outcome as observed in CITES. In contrast, the models to predict presence of tubal pathology and endometriosis, were developed and applied in the same data, thereby inducing the risk of overestimating their performance by overfitting.

The fertility prognosis was supposed to improve with a relative risk of 1.3 after HSG and laparoscopy, due to pertubation of the tubes. A meta-analysis of four studies comparing oil- and water-soluble contrast media for HSG revealed a relative risk of 2.0 in favor of oil-soluble contrast media.7 Unfortunately, in two of these studies the method of randomization was not reported, whereas one other study was not truly randomized.22"24

The remaining study reported a relative nsk of 1.3 in favor of oil-soluble contrast medium.25 The only randomized study reporting on fertility after HSG using oil-soluble contrast media as compared to fertility without a HSG showed an almost 3-fold increase of the pregnancy rates after H S G 8 Thus, the relative risk of 1.3 for the direct therapeutic effect of HSG on fertility outcome seems to be a conservative estimate. Although data on the effect of chromopertubation at lapaoroscopy on fertility are lacking, it seems reasonable to assume that chromopertubation of the tubes at laparoscopy also has a positive effect on fertility outcome. Moreover, the sensitivity analysis in which the therapeutic impact of HSG and laparoscopy was limited showed that even a relative nsk

81

Chapter 7

of one, i.e., no therapeutic effect of HSG and laparoscopy, would not alter the outcome

of the analysis. Tubal surgery was not taken into account in the present study, since the effectiveness

of tubal surgery could not be obtained from well performed prognostic studies on tubal surgery, let alone randomized clinical trials. Apart from this lack of evidence on the effectiveness of tubal surgery, the number of patients in which tubal surgery might be indicated is usually limited. However, if one believes that tubal surgery is an effective treatment in patients with tubal pathology, the indication for this treatment should be made at laparoscopy. The strategy in which laparoscopy is performed in all patients would in that case become more cost-effective.

With respect to tubal surgery, it is also important to emphasize that the present analysis was limited to live birth of the first child, whereas the wish for a second child was not taken into account. If a subfertile couple wants more than one child, tubal surgery might become more attractive than TVF-ET. A second consequence of the wish for multiple children might be that expectant management in women with a relatively old age would become less attractive as compared to immediate treatment.

In conclusion, the present analysis showed that the use of CAT, HSG and/or laparoscopy is cost-effective in most subfertile couples. The diagnostic work-up to detect tubal pathology in subfertile couples should start with CAT in couples with relatively good fertility prospects, whereas couples with relatively poor fertility prospects benefit from immediate HSG.

7.5 References

1. Mol BWJ, Dijkman B, Wertheim P, Lijmer JG, Van der Veen F, Bossuyt PMM. Chlamydia antibody titers in the diagnosis of tubal pathology; a meta-analysis. Fertil Steril 1997;67:1031-7.

2. Land JA, Evers JLH, Goossens VJ. How to use Chlamydia antibody testing in subfertility patients? Hum Reprod 1998;13:1094-8.

3. Mol BWJ, Swart P, Bossuyt PMM, Van Beurden M, Van der Veen F. Reproducibility of the interpretation of hysterosalpingography in the diagnosis of tubal pathology. Hum Reprod 1996;11:1204-8.

4. Swart P, Mol BWJ, Van der Veen F, Van Beurden M, Redekop WK, Bossuyt PMM. The accuracy of hysterosalpingography in the diagnosis of tubal pathology, a meta-analysis. Fertil Stenl 1995;64:486-91.

5. Mol BW), Swart P, Bossuyt PMM, Van der Veen F. Is hysterosalpingography an important tool in predicting fertility outcome? Fertil Steril 1997;67:663-9.

6. Mol BWJ, Collins JA, Burrows EA, Van der Veen F, Bossuyt PMM. Comparison of hysterosalpingography and laparoscopy in predicting fertility outcome. Hum Reprod In press

7. Watson A, Vandekerckhove P, Lilford R, Vail A Brosens I, Hughes E. A meta-analysis of the therapeutic role of oil-soluble contrast media at hysterosalpingography: a surprising result? Fertil Stenl 1994;61:470-7.

8. Ogata R, Nakamura G, Uchiumi Y, Yokoyama M, Watanabe Y, Nozaki M Sano M, Nakano H. Therapeutic efficacy of hysterosalpingography (HSG) in infertility, a prospective, randomized, clinical study.jpnj Fertil Stenl 1993;38:91-4.

9. Marcoux S, Maheux R, Bérubé S. Laparoscopic surgery in infertile women with minimal or mild endometriosis. N Engl J Med 1997;337:217-22.

82

Cost-effectiveness of diagnosis of tubal pathology

10. Mol BWJ, Bairam N, Lijmer JG, Wiegerinck MAHM, Bongers MY, Van der Veen F, Bossuyt PMM. The accuracy of CA-125 in the diagnosis of endometriosis: a meta-analysis. Fertil Steril 1998^70:1101-

8. 11. Collins A, Burrows EA, Willan AR. The prognosis for live birth among untreated infertile couples.

Fertil Steril 1995;64:22-8. 12. The American Fertility Society. Revised American Fertility Society classification of endometriosis:

1985. Fertil Steril 1985;43:351-2.

13. Nackely AC, Muasher SJ. The significance of hydrosalpinx in in vitro fertilization. Fertil Steril

1998;69:73-84. 14. Mol BWJ, Collins JA, Van der Veen F, Bonsel GJ, Wiegerinck MAHW, Bossuyt PMM. Cost-

effectiveness of in-vitro fertilisation and embryo-transfer. Submitted. 15. Templeton A, Morris JK, Parslow W. Factors that affect outcome of in-vitro fertilisation treatment.

Lancet 1996;348:1402-6.

16. Collins JA, Feeny D, Gunby J. The cost of infertility diagnosis and treatment in Canada in 1995. Hum

Reprod 1997;12:951-8.

17. Neumann PJ, Gharib SD, Weinstein M C The cost of a successful delivery with in vitro fertilization.

N Engl J Med 1994;331:239-43. 18. Callahan TL, Hall JE, Ettner SL, Chnstiansen CL, Greene MF, Crowley WF. The economic impact of

assisted-reproduction techniques to their incidence. N Engl J Med 1994;331:244-9. 19. Wolner-Hanssen P, Rydhstroem H. Cost-effectiveness of in-vitro fertilization: estimated costs per

successful pregnancy after transfer of one or two embryos. Hum Reprod 1998;13:88-94. 20. Gold MR, Siegel JE, Russell LB, Weinstein M C Cost-effectiveness in health and medicine. New York,

Oxford. Oxford University Press 1996; pages 259 and 262-3. 21. Stolwijk AM. Human fecundity under natural conditions and during in vitro fertilization. Nijmegen.

Thesis. 1997. 22. Schwabe MG, Shapiro SS, Haning RV Jr. Hysterosalpingography with oü contrast medium enhances

fertility in patients with infertility of unknown etiology. Fertil Steril 1983;40:604-6. 23. De Boer AD, Vemer HM, Willemsen WNP, Sanders FMB. Oü or aqueous contrast media for

hysterosalpingography: a prospective, randomized clinical study. Eur J Obstet Gynaecol Reprod Biol 1988;28:65-8.

24. Rasmussen F, Lindequist S, Larsen C, Justesen P. Therapeutic effect of hysterosalpingography: oil- versus water soluble contrast media - a randomized, prospective study. Radiology 1991;179:75-8.

25. Alper MM, Gamer PR, Spence JEH, Quarrington A. Pregnancy rates after hysterosalpingography with oil- and water soluble contrast media. Obstet Gynecol 1986;68:6-9.

83