The cost of infertility diagnosis and treatment in Canada in 1995

Human Reproduction vol.12 no.5 pp.951 958, 1997 The cost of infertility diagnosis and treatment in Canada in 1995 J.A.Collins 1,2,4, D.Feeny 3,2 and J.Gunby 1 treatment. This information is needed to provide a context for the cost of advanced infertility treatment, to aid in the Departments of 1 Obstetrics and Gynecology, 2 Clinical Epidemiology and Biostatistics and 3 Centre for Health Economics development of public policy, to strengthen clinical decisionand Policy Analysis and Department of Economics, McMaster making, and to assist in the formulation of policy on research University, Hamilton, ON L8N 3Z5, Canada and development for infertility. 4 To whom correspondence should be addressed The question could be addressed in a cautious manner by reviewing and commenting on the scant literature available. The objectives of this study were to estimate the direct More challenging is an alternative strategy with the following cost of infertility management, including diagnosis and objectives: (i) to estimate from various sources of existing treatment, in Canada during 1995, and the relative cost information the overall direct cost of infertility management per live birth by treatment category. The analysis was in a single country (in this case, Canada) for a single year based on the following estimates: the prevalence of (1995) and (ii) to estimate average cost outcome ratios for the infertility in Canada in 1995; the volume and distribution treatments that are pertinent to each infertility diagnostic of infertility services; and the effectiveness and cost of category. specific infertility treatments. In 1995 there were ~330 000 The estimate for objective (i) will include costs covered by couples experiencing infertility in Canada. It is estimated provincial health care plans and drug insurance plans, and also that 50% (150 000) sought medical advice or treatment costs for treatment paid by the patients receiving the services. during that year. A total of 13 diagnostic and treatment Not included are costs to the patient such as transport to the categories account for nearly all of the treatments received, clinic, accommodation or time off work, medical costs due to and these categories form the treatment model. The cost complications of treatments (e.g. ovarian hyperstimulation of treatment per live birth ranges from Cdn$650 for syndrome) or costs to the patient or the health care system clomiphene treatment of unexplained infertility to accrued during pregnancy, delivery or the neonatal period. It Cdn$41 000 for in-vitro fertilization. For a hypothetical is accepted that infertility treatment may result in a higher group of 100 couples, the annual cost of diagnosis and than average proportion of multiple births. However, the treatment would be Cdn$77 000 and Cdn$200 000 respect- additional costs of premature births resulting from multiple ively for a total of Cdn$277 000, or an average of Cdn$2770 gestation are not included here. The final estimate of infertility per couple. After 1 year of treatment, it is expected that management costs will be compared with the total annual 26 of these 100 couples would achieve a live birth. The health care expenditure in Canada to put into perspective the total annual cost of infertility management in Canada, issue of infertility treatment cost. estimated to be ~Cdn$415 million, is 0.6% of the annual For the second objective, we will determine the direct cost cost of health care. of conceiving a pregnancy which results in a live birth in Key words: cost analysis/effectiveness/infertility treatment/ couples who have consulted a physician regarding infertility. utilization This cost will vary depending on the treatment received, which in turn depends on the reason for infertility. The analysis assumes that for fertile couples there is no cost involved in conceiving a child. Therefore any medical assistance provided Introduction to infertile patients or any expenditure by the patients themselves In recent years, public attention has been focused on issues is an extra cost to society. To make decisions about surrounding assisted reproductive technology for the treat- infertility services it is necessary to be aware not only of their ment of infertility, and in particular the cost of such treatment. costs but also of their usefulness in leading to the goal of In-vitro fertilization (IVF) and other new health technologies infertility treatment, i.e. the live birth of a healthy infant. are expensive because they require highly trained personnel, The information required for the analysis includes estimates costly drugs, expensive equipment and frequent monitoring of: (i) the prevalence of infertility in Canada in 1995; (ii) the (US Congress, Office of Technology Assessment, 1988). The volume and distribution of infertility services; (iii) the costs cost of IVF in Canada ranges from Cdn$1800 to $4600 per of specific infertility diagnostic and treatment protocols; and cycle, plus the cost of medications which average Cdn$1500 (iv) the effectiveness of specific infertility treatments. Estimates 2500 (Fluker and Tiffin, 1996). of the cost of medical services that are not based on actual Few studies, however, address the overall direct cost of expenditures necessarily entail assumptions concerning cost, infertility management, which includes both diagnosis and effectiveness and utilization. Uncertainties about the level of European Society for Human Reproduction and Embryology 951

J.A.Collins, D.Feeny and J.Gunby Table I. Distribution and costs of infertility diagnostic services Service Percentage No. of tests Cost per couple Cost per 100 of couples per couple tested (Cdn$) couples (Cdn$) Assessment by family physician/gynaecologist 100 1.0 48.20 4820 Consultation by specialist 82 1.0 54.80 4412 Repeat visit to specialist 82 1.0 18.70 1533 Semen analysis 87 1.6 20.00 1740 Hysterosalpingogram 69 1.0 81.21 5603 Laparoscopy 63 1.0 792.61 49 934 Mid-luteal serum progesterone 60 1.5 31.02 1861 Other serum hormones 49 3.0 93.87 4600 Endometrial biopsy 34 1.2 73.23 2490 Total cost (Cdn$) 76 993 expenditure are to be expected, and the potential impact of such uncertainties can be consequential. Until empirical data are available, the resulting estimates should be viewed as no more than guestimates of the real expenditures. make the volume of service calculation below, we need to know the proportion of women who seek medical advice or treatment for infertility, information which is available from the USA data but not from Canadian sources. Number of couples Materials and methods In the 1991 Canadian census there were 3.6 million couples of whom the female partner was aged 18 44 years and who had been The prevalence of infertility in Canada in 1995 married or co-habiting for at least 1 year. From 1991 to 1995 the Proportion of couples total population of Canada increased by 8.2%; thus the estimate of The Royal Commission on New Reproductive Technologies (RCNRT) such couples for 1995 would be 3.9 million. The estimate of the estimated the prevalence of infertility in Canada from three telephone number of infertile couples in Canada in 1995 is 8.5% of this figure, surveys conducted across Canada in late 1991 and early 1992, which or ~330 000 couples. included a total of 1412 respondents. The denominator was all women in Canada aged 18 44 years who had been married or co-habiting The volume and distribution of infertility services for at least 1 year. A woman was classified as infertile if she responded Volume of services that she had not used contraception in the past 12 months, that neither The proportion of infertile couples in Canada who seek medical care she nor her partner had been surgically sterilized, and that she had has not been assessed in a formal manner. Although care-seeking not been pregnant in the past 12 months. All three surveys yielded behaviour may not be similar in Canada and the USA, the proportion similar results, which is an indication that the estimate is reliable. of infertile couples is highly comparable, and this may excuse the The combined results showed that 8.5% of such couples were use of estimates from the USA. The NSFG survey estimated that experiencing infertility at the time of the survey. The 95% confidence 2.3% of all American women of reproductive age receive medical interval for the estimate was 7.0 9.9% (RCNRT, 1993). This was not advice or treatment for infertility in a given year (Mosher and Pratt, an estimate of the proportion of couples who would experience 1991). The female population of Canada aged 15 44 years was 6.5 infertility in their lifetime, but rather an estimate of the proportion million in 1995, of which 2.3% is ~150 000. Assuming that the of couples of reproductive age who were currently aware that they number of women seeking infertility services who are not in the cowere having difficulty conceiving, and who may potentially seek habiting, 18 44 year old age group is negligible, this would mean medical assistance to achieve pregnancy. that 45% of infertile couples sought medical assistance to achieve Comparison with estimates in the USA pregnancy in 1995. The prevalence of infertility in the Canadian RCNRT survey was virtually the same as the estimate in the USA, although the Distribution of diagnostic services methodology was slightly different. Using the methods of the Of those couples who seek medical assistance for their infertility, the National Study of Family Growth (NSFG) in the USA, the 1 year proportion who would be expected to receive the most commonly figure for Canada in 1991 1992 was 8.0% of currently married or provided diagnostic services over the course of a year are shown in co-habiting couples with a female partner aged 18 44 years (RCNRT, Table I. These estimates are based on data from 2198 couples seen 1993). For the USA in 1988, the 1 year estimate for the prevalence in Canadian fertility clinics from 1984 to 1991 during the Canadian of infertility was 7.9% of currently married couples with a female Infertility Treatment Evaluation Study (CITES; Collins et al., 1993). partner aged 15 44 years (Mosher and Pratt, 1990). The distribution of diagnostic services in 1995 is not expected This comparison of estimates for Canada and the USA is useful to differ greatly from the CITES figures. It is assumed that all for two reasons. First, to project the 1991 Canadian prevalence patients seeking medical assistance first see their family physician or estimate to 1995 we assume that it is stable over time. Because the gynaecologist, who would carry out only basic diagnostic assessments. RCNRT survey was the first to estimate reliably the prevalence of Because 82% of CITES patients underwent either a diagnostic infertility in Canada, data to support this assumption are not available. laparoscopy or a treatment usually available only in a fertility clinic, However, the estimates of infertility for the USA were stable from this figure was used, in the absence of more precise data, as the 1965 to 1988 (Mosher and Pratt, 1990). Comparable prevalence estimate for the proportion of couples who are referred to a fertility estimates in 1991 for the USA and Canada support the assumption specialist for further work-up. Couples having diagnostic tests would that the prevalence of infertility is also stable in Canada. Second, to return at least once to the specialist to discuss test results and plan care. 952

Cost of infertility treatment in Canada Table II. Distribution of infertility treatments Treatment Percentage of couples Diagnostic group CITES 1995 actual projected Advice or diagnosis only (self-treated) 40 40 All Clomiphene citrate 19 20 10 unexplained infertility, 10 ovulation defect Surgery 8 10 8 tubal factor, 2 endometriosis In-vitro fertilization 6 6 3 tubal factor, 3 persistent infertility c Therapeutic donor insemination 6 6 Male factor Intrauterine insemination (IUI) b 4 4 2 unexplained infertility, 2 male factor Ovulation suppression 3 2 Endometriosis Human menopausal gonadotrophin (HMG) a 2 2 Ovulation defect Male treatment 2 2 Male factor Other 10 0 HMG/IUI 0 8 Persistent infertility CITES Canadian Infertility Treatment Evaluation Study. a Or follicle stimulating hormone. b Partner s spermatozoa. c All in-vitro fertilization treatment is classified under persistent infertility in the following tables. Distribution of treatment services The distribution of treatments among couples using fertility services was also estimated from CITES. This study reflected the treatment choices of couples with an average 3.5 years duration of infertility and a mean female age of 29.5 years, all of whom were attending specialized clinics (Collins et al., 1993). Table II shows the distribution of their initial treatment choices. Because the CITES treatment choices in 1984 1991 may not reflect the choices in 1995, in particular the use of human menopausal gonadotrophins (HMG) for ovulation induction with intrauterine insemination (IUI) of partner s spermatozoa, which has recently become more widely available, the latter treatment replaces the other treatment category for the projected 1995 distribution of treatments (Table II and Figure 1). The infertility diagnostic category that would apply to each treatment is also given in Table II. During a single year, some couples will undergo more than one treatment. In the CITES data, 6.5% of patients received a second treatment within 1 year. The effect of these additional treatments on live birth is inconsequential, because couples will only move on to a second treatment if the first has been unsuccessful, but the effect on costs may be considerable. Thus, the cost calculations below account for the cost of a second treatment within 1 year by adding 5% to the cost of each treatment. The effectiveness of infertility treatment For the purpose of estimating effectiveness, the commonly used treatments shown in Table II were grouped by primary diagnosis Figure 1. Comparison of the percentage of infertility patients in because the reason for the infertility has an effect on success rate. each diagnosis/treatment group (hatched bars) and the cost per live birth for each group (solid bars). CC clomiphene citrate; The group of couples who received only advice and/or diagnosis Ovul ovulation defect; UI unexplained infertility; were considered to be untreated or self-treated, because it is assumed Endo endometriosis; OvSup ovulation suppression; that they continued to attempt pregnancy, and would include those Tubl tubal factor; Male male factor; PI persistent infertility. with all diagnoses of infertility. Effectiveness was estimated with See also footnote to Table III. $ are Cdn$. respect to live birth rate, which is the most relevant outcome. However, in most published studies pregnancy rate is reported. These rates were converted to live birth rates using the proportion of live treatment protocol to yield the live birth rates per patient per annum. births observed among pregnancies in CITES (86%; Collins et al., Typical durations of therapy for cyclical treatments were drawn from 1995). Table III specifies the diagnosis and treatment groupings and a description of treatment in CITES (Collins et al., 1993). For summarizes the published evidence on effectiveness. example, in the clomiphene trials for unexplained infertility, the The live birth rates given in Table III are the aggregates of the aggregate pregnancy rate per cycle was 6.8%, which converts to a pregnancy rates per cycle in the reported trials, multiplied by the live 6.8 0.86 5.8% live birth rate per cycle of treatment. The typical birth factor (0.86). These were multiplied by the duration of each duration of this treatment was 6 months, taking into account any 953

J.A.Collins, D.Feeny and J.Gunby Table III. Description of data sources for infertility treatment effectiveness and live birth rates Diagnosis group Treatment Level of evidence Number of Live birth Number of Live birth patients rate per cycles per rate per (cycles) cycle (%) treatment treatment (%) All diagnoses Self-treated Prospective cohort a 2198 12 14 Ovulation defect Clomiphene Randomized controlled trial (RCT) b 24 5.7 6 34 HMG Aggregate of four case series b (3713) 14.2 3 43 Unexplained infertility Clomiphene Meta-analysis of four RCT b (738) 5.8 6 35 IUI Meta-analysis of two RCT c (178) 5.2 5 26 Endometriosis Ovulation suppression Meta-analysis of nine RCT b 357 6 30 Surgery Prospective cohort d 48 25 Tubal factor Surgery Aggregate of 14 case series b 1275 21 Male factor IUI Meta-analysis of six RCT b (895) 2.8 5 14 TDI Aggregate of five RCT e 380 12.9 5 65 Clomiphene Meta-analysis of four RCT b 145 6 16 Persistent infertility HMG/IUI Meta-analysis of 13 RCT b (625) 12.4 3 37 IVF Registry report for Canada f (3707) 14.1 2 28 HMG human menopausal gonadotrophin; IUI intrauterine insemination; IVF in-vitro fertilization; TDI therapeutic donor insemination. a Collins et al. (1995). b ESHRE Capri Workshop (1996). c Taylor and Collins (1992). d Adamson and Pasta (1994). e Byrd et al. (1990); Odem et al. (1991); Patton et al. (1992); Robinson et al. (1992); Subak et al. (1992). f International Working Group for Registers on Assisted Reproduction (1995). Table IV. Costs of infertility treatment per cycle, per treatment and per live birth Diagnostic group Treatment Number of Cost per Cost per Live birth Cost per cycles per cycle or treatment rate per live birth treatment month (Cdn$) (Cdn$) treatment (Cdn$) All diagnoses Self-treated 12 20 240 0.14 1714 Ovulation defect Clomiphene 6 35 229 0.34 674 HMG 3 1600 4800 0.43 11 163 Unexplained Clomiphene 6 35 229 0.35 654 infertility IUI 5 204 1020 0.26 3923 Endometriosis Ovulation suppression 6 300 1844 0.30 6147 Surgery 4000 0.25 16 000 Tubal factor Surgery 4000 0.21 19 048 Male factor IUI 5 204 1020 0.14 7286 TDI 5 354 1770 0.65 2723 Clomiphene 3 90 304 0.16 1900 Persistent infertility HMG/IUI 3 1530 4890 0.37 13 216 IVF 2 5700 11 400 0.28 40 714 See Table III for abbreviations. shorter duration of use due to conceptions and any longer use among some patients. Thus the expected probability of live birth for this therapy per annum was 5.8 6 35.0%. In the case of non-cyclical therapy, such as surgery or ovulation suppression, live birth rates were estimated for the first year after completion of the intervention. Table III summarizes the live birth rates per cycle, the typical number of cycles per treatment and the expected live birth rate per patient per year. The cost of infertility services Only direct costs were included in the analysis, as discussed in Introduction. When nationally representative, published cost estimates were not available, known costs were used from Chedoke-McMaster Hospital (CMH), Hamilton, Ontario, Canada. With respect to costs to consumers, Hamilton is near the median among Canadian cities. Therefore it is perhaps plausible that the same is true for health care costs. The costs were derived for individual diagnostic tests and treatments as described below, and these are summarized in Tables I and IV. Published cost estimates were projected to 1995 Canadian 954 dollars with the use of the inflation factor for health care costs in Canada for December 1995 (Statistics Canada, 1996). Cost of advice and diagnosis These costs are summarized in Table I. The costs for physician visits and diagnostic tests were taken from the 1995 Ontario fee schedule. The average cost of advice and diagnosis per patient is estimated to be Cdn$770. Self-treatment costs No published estimate could be found for the amount spent by patients who decide not to undergo treatment after being assessed. Expenditures are routinely made, however, to buy thermometers or test kits, for example. Lacking any information on the extent of these costs, it was assumed that couples who are not undergoing treatment protocols spend Cdn$20 per month or Cdn$240 per annum on infertility-related items. Clomiphene costs No published estimate could be found. The estimate shown (Cdn$229) was based on the average prescription charge in Central West Ontario

Cost of infertility treatment in Canada ($35 per cycle for six cycles $210) plus one follow-up visit ($19). per annum and a live birth rate after six cycles of treatment of 35%; No provision was made for the cost of tests used to monitor the cycle therefore the cost per live birth was $229/0.35 $654. The cost of because these are optional. diagnosis was not included in this figure. HMG costs A published estimate of the 1994 cost of HMG/IUI was US $1200, or Cdn$1680 per treatment cycle (Peterson et al., 1994). This cost Results included clinic visits, procedures, monitoring and medications, and The cost of diagnosis for 100 couples was calculated to be applied to women of all diagnostic categories. About half the women ~$77 000 (Table I). A summary of treatment costs for 100 with ovulation defects who are treated with HMG are more difficult couples is shown in Table V. The cost per treatment from to stimulate and were considered to require three additional ampoules Table IV was multiplied by the proportion of patients receiving of medication and two additional clinic visits ($250). Subtracting IUI that treatment from Table II, and these were summed to costs (estimated below), the HMG costs for women with ovulation defects would be $1600 per cycle. estimate the total annual cost of infertility treatment in a hypothetical group of 100 couples. When the 5% adjustment IUI costs for repeated treatment courses within 1 year was added, the No published estimate could be found. The estimate of $204 per total annual cost for 100 couples came to ~$200 000. Thus, cycle is the sum of CMH laboratory charges ($100), a luteinizing hormone (LH) test kit cost ($50), procedure fees ($24) and ambulatory the total cost of infertility management for 100 couples was facility charges ($30). $77 000 $200 000 $277 000. The average cost of infertility management per couple was therefore $2770. The Ovulation suppression costs No published estimate could be found. Gonadotrophin-releasing annual cost in Canada for the 150 000 couples estimated hormone agonist costs range from as low as $200 to $500 per month, to have sought medical advice or treatment in 1995 was depending on product and dosage. The average was taken to be $300 $415 million. per month, and the final estimate included one physician visit ($19) The cost per live birth for the various treatments is shown and five injection visits ($5 each), for a total of $1844 over 6 months. in Table IV and Figure 1. This cost ranged from $654 for Tubal and endometriosis surgery costs clomiphene treatment of couples with unexplained infertility No published estimate could be found. The total shown ($4000) is to $40 714 for IVF treatment of couples with persistent the cost at CMH for tubal reanastomosis, including hospital costs, infertility. The mean cost per live birth was $7745. Excluding surgeons fees, anaesthetists fees, and pre- and post-operative visits. IVF treatment and births, the mean cost per live birth would The same estimate was used for the cost of laparotomy for have been $5311. endometriosis. In Table V the live birth rate per patient for each treatment Therapeutic donor insemination costs from Table III was multiplied by the percentage of patients For this service it was assumed that the insemination would be receiving that treatment from Table II to estimate the number intrauterine with sperm preparation. From a recent survey of Canadian of live births for each treatment in a hypothetical group of fertility clinics, the median fee for therapeutic donor insemination 100 couples. The overall number of live births per 100 couples was $250 (Fluker and Tiffin, 1996), which included the costs of during 1 year was just under 26. Figure 2 shows the percentage screening and testing donors and laboratory charges for freezing, of total births that resulted from each treatment within a thawing and processing spermatozoa. The estimate of $354 per cycle diagnostic group. The live birth rate per patient was only 14% was the sum of this fee, the LH test kit cost ($50), procedure fees ($24) and ambulatory facility charges ($30). for self-treatment (Table III), but 40 of the 100 couples were in this category, so it accounted for 0.14 40 5.6 births per Male treatment costs annum, which was 22% of the 26 live births in the hypothetical No published estimate could be found. Clomiphene (25 mg/day) appears to be the most frequent treatment, usually for 3 months. cohort. In contrast, although the live birth rate was much Typical costs were $95 per month for drugs and dispensing fees, and higher (43%) for three cycles of HMG treatment for ovulation one physician visit ($19), giving a total of $304. defects, only two of the 100 couples received this treatment, so it accounted for only 0.8 or 3% of the live births per annum IVF costs Charges for IVF in Canadian centres in 1995 ranged from $1800 to in 100 patients. $4620 per cycle (Fluker and Tiffin, 1996). The median cost of $3800 Similarly, the cost of each treatment in a group of 100 (including clinic visits, injections, medications, monitoring, physician infertile couples divided by the total cost ($200 000) gave the fees and laboratory and hospital charges), combined with an estimated percentage of the total cost attributable to each treatment average cost of $1900 for medications, provided the cost estimate within a diagnostic group (Figure 2). for one cycle of IVF of $5700. Cost per live birth The costs of infertility treatment for the various diagnostic categories and treatments are summarized in Table IV. The cost per patient per annum was derived from the cost per cycle or month multiplied by the typical number of months of treatment together with the additional costs that accrue during the course of treatment. The cost per live birth was the cost of treatment per patient divided by the proportion of cycles that resulted in live births. For example, clomiphene treatment for unexplained infertility had a cost of $229 per patient Discussion In this model of infertility diagnosis and treatment in Canada in 1995, 26% of infertile couples seeking medical advice or treatment have live births associated with services costing Cdn$415 million. This per annum live birth rate is higher than the 31% live birth rate during a mean 26 months follow-up among 2198 couples in the CITES (Collins et al., 1995). It is also higher than the rates in similar published follow-up 955

J.A.Collins, D.Feeny and J.Gunby Table V. Number of births and costs of infertility treatment for 100 hypothetical couples Diagnostic group Treatment No. of couples Live birth Number of Cost per Total cost of receiving rate per live births patient treatment treatment treatment (Cdn$) (Cdn$) 5% (Cdn$) a All diagnoses Self-treated 40 0.14 5.6 240 10 080 Ovulation defect Clomiphene 10 0.34 3.4 229 2405 HMG 2 0.43 0.8 4800 10 080 Unexplained Clomiphene 10 0.35 3.5 229 2405 infertility IUI 2 0.26 0.5 1020 2142 Endometriosis Ovulation suppression 2 0.30 0.6 1844 3872 Surgery 2 0.25 0.5 4000 8400 Tubal factor Surgery 8 0.21 1.7 4000 33 600 Male factor IUI 2 0.14 0.3 1020 2142 TDI 6 0.65 3.9 1770 11 151 Clomiphene 2 0.16 0.3 304 638 Persistent infertility HMG/IUI 8 0.37 3.0 4890 41 076 IVF 6 0.28 1.7 11 400 71 820 Total 25.8 199 811 See Table III for abbreviations. a Adjustment factor for multiple treatments per patient per annum. Figure 2. Comparison of the percentage of the total number of births (hatched bars) and the percentage of the total cost of infertility treatment (solid bars) attributable to each diagnosis/ treatment group in a group of 100 hypothetical couples. Abbreviations are defined in the legend to Figure 1. studies, but all of these reports were based on patients attending specialized clinics (Collins et al., 1984; Dunphy et al., 1989; Bostofte et al., 1993; Eimers et al., 1994). Such patients have a longer duration of infertility and lower fertility than the average infertile couple. Therefore the model appears to be consistent with published empirical observations. The live birth rates for the various treatments used in the model were derived primarily from trial results, and thus are the most accurate estimates available. However, the untreated live birth rate may be underestimated because it was derived 956 from patients with a long duration of infertility (Collins et al., 1995). Compared with the estimates for live birth rates, more assumptions about costs were necessary because of missing information, and this could have affected the validity of the cost calculations. The costs attributed to the self-treated group are notional, and the costs of many treatments have not been formally assessed. It was also necessary to use local costs for some treatments such as IUI. Nevertheless, for the most expensive therapies (HMG and IVF) the cost estimate could be based on recently published figures (Peterson et al., 1994; Fluker and Tiffin, 1996). With respect to the IVF data, we elected to make use of published figures for IVF success rates with fresh embryo transfer only, because the detailed cost analysis was based on this treatment. By 1995, embryo cryopreservation was available in all Canadian centres. This technique was forecast to increase live birth rates per stimulation cycle by an estimated 1.8% (Daya et al., 1995). IVF costs per cycle and per live birth, including embryo cryopreservation, were given by Daya et al. (1995). The cost per live birth in that paper (Cdn$35 000) was lower than the present estimate because of this inclusion of cryopreservation successes and the effects of inflation. Natural cycles were not considered here because our centre was virtually the only one in Canada that provided this treatment and, since the Ontario government s payment policy changed in 1994, we no longer do so. We did not provide separate estimates for intracytoplasmic sperm injection (ICSI) because data on utilization in Canada and success rates are not yet available. Nevertheless, the live birth rates and costs for ICSI are probably similar to those for IVF, and would be included in our analysis as treatment for persistent infertility. It can be argued that an analysis of the cost of infertility treatment should include costs of the complications and consequences of that treatment, such as ovarian hyperstimulation syndrome, ectopic pregnancy, miscarriage and premature delivery due to multiple gestation. Neumann et al. (1994) concluded that these factors added 21% to the base cost of an IVF cycle, which for our estimate would be

Cost of infertility treatment in Canada ~Cdn$1200. Taking into account treatment-related maternal alternative model could assume that the proportion of patients complications only, they estimated a 2% increase in cycle cost. receiving IVF is 4%, as reported for the USA in the NSFG, If we added either of these increments to the cycle cost of rather than 6%, and that the proportion receiving HMG/IUI IVF, we would have to similarly increase the cost of all the treatment is 4% rather than 8% (Wilcox and Mosher, 1993). other treatments. However, because the data are not available The 6% of couples who would not be allocated in this model for all treatments, we have chosen not to include these extra could be assumed to receive a variety of other treatments costs in our analysis. On the other hand, neither have we having an aggregate of the average cost per patient for included possible savings and benefits to society attributable all treatments except HMG/IUI and IVF. In this alternative to fertility treatment. model the total cost of treatment for 100 couples, adjusted Costs and outcomes are important factors that influence for repeated treatments, would be ~Cdn$162 000 rather than clinical decisions, and it would be inappropriate to isolate Cdn$200 000, a reduction of 19%. The overall cost of infertility costs from effectiveness. The overall cost of infertility treatment diagnosis and treatment in Canada would be Cdn$359 million was derived from estimates of the cost of each treatment and rather than Cdn$415 million. With the treatment distribution the number of infertile couples making use of each service. In assumptions for this model, the aggregate birth rate would be a parallel fashion, the overall success of infertility treatment reduced by only 0.6% to 25.2%. It seems clear that the model was derived from estimates of the effectiveness of each presented in Results, based as it is on the treatment experience treatment and the number of infertile couples making use of in advanced infertility clinics, yields a cost estimate which is each service. Thus it is appropriate to compare the proportion at the higher end of the likely range of true costs for infertility of cost and the proportion of births associated with a given treatment in Canada. Differences in the order of Cdn$56 treatment (Figure 2). It is obvious that the proportion of costs million arising from a change in the assumptions for a single does not always correspond to the proportion of live births. diagnostic category should not distract attention from the For example, both clomiphene and IUI are used for the usefulness of the original model, if only because there is little treatment of unexplained infertility, contributing about equally information about the overall cost of infertility management to the cost of infertility treatment (1.2 and 1.1% respectively). in the public domain. However, clomiphene treatment results in 14% of live births, The estimates of cost per live birth provided in Table IV while IUI results in only 2%. Self-treatment is associated are quite crude and are not a substitute for a careful economic with just 5% of total costs but contributes 22% of live births, evaluation of each of the services listed. At best these are in part because it is the largest category. On the other hand, estimates of the average cost per live birth. These average cost HMG/IUI and IVF treatments for persistent infertility together figures do not provide information comparing the incremental account for 57% of the total treatment costs but result in just cost of one clinically relevant treatment plan relative to the over 18% of the live births, in part because patients in this cost of a clinically relevant alternative plan for the same category are the most difficult to treat. indication. In addition, the estimates do not take into account The distribution of treatment services used in the model any systematic difference among categories in the quality of was based on the treatment choices of CITES couples, who the live births, such as the incidence of multiple pregnancies may not be typical of infertile couples in general. However, and prematurity. The descriptive information provided in Table the proportion who visited a physician and, after discussion IV is not meant to be used as if it were evaluative information and perhaps some diagnostic tests, decided not to have any such as the information available for IVF (Goeree et al., 1993; treatment was similar in CITES (40%) and in the American Neumann et al., 1994). NSFG survey (43%), which included visits to family physicians What is the context for health expenditures in the order of (Wilcox and Mosher, 1993). A recent unpublished survey of Cdn$400 million? Total health care spending in Canada is 1500 Canadian family physicians found that clomiphene was estimated to be Cdn$74 billion for 1995, of which ~72% was prescribed on a regular basis by 11% of respondents and publicly funded. The cost of infertility management is ~0.5 occasionally by 34%, but other types of infertility treatment 0.6% of the Cdn$74 billion total. In comparison, health were rarely used. A companion survey of 500 Canadian care spending for the treatment and care of patients with obstetricians/gynaecologists found that 88% may prescribe osteoporosis in Canada amounts to ~Cdn$1.3 billion annually clomiphene for their infertile patients, but only 15 20% offered or 1.8% of total health care costs (Goeree et al., 1996). The more complex treatments such as ovulation induction with average annual cost per patient treated for osteoporosis is HMG or artificial insemination with partner s or donor ~Cdn$17 000, compared with Cdn$2700 per couple for infertilspermatozoa (S.Daya, personal communication). Therefore, ity management. it seems that the majority of infertility treatments are provided Unquestionably, infertile couples and their clinicians put a by fertility clinics or specialists, justifying the use of the high value on infertility diagnosis and treatment services. The CITES figures as the best available data for estimating the funding of health programmes, however, is also influenced by distribution of treatment services. societal attitudes and the views of politicians and health Any misjudgement about the proportion of couples undergoing administrators who may not be as sensitive to the burden of HMG/IUI or IVF treatment would not alter the high infertility as a health disorder. One implication of the results cost outcome ratio for these categories, but it could have a of this study is that the majority of the cost of infertility marked effect on the total cost per 100 couples and on the treatment arises from the diagnostic categories that are the estimate of the overall costs in Canada. For example, an most difficult to treat. 957

J.A.Collins, D.Feeny and J.Gunby References Adamson, G.D. and Pasta, D.J. (1994) Surgical treatment of endometriosisassociated infertility: meta-analysis compared with survival analysis. Am. J. Obstet. Gynecol., 171, 1488 1505. Bostofte, E. et al. (1993) Fertility prognosis for infertile couples. Fertil. Steril., 59, 102 107. Byrd, W. et al. (1990) A prospective randomized study of pregnancy rates following intrauterine and intracervical insemination using frozen donor sperm. Fertil. Steril., 53, 521 527. Collins, J.A. et al. (1984) A proportional hazards analysis of the clinical characteristics of infertile couples. Am. J. Obstet. Gynecol., 148, 527 532. Collins, J.A., Burrows, E.A. and Willan, A. (1993) Infertile couples and their treatment in Canadian academic infertility clinics. In Royal Commission on New Reproductive Technologies, Treatment of Infertility: Current Practices and Psychosocial Implications. Ministry of Supply and Services Canada, Ottawa, Canada, pp. 233 340. Collins, J.A., Burrows, E.A. and Willan, A.R. (1995) The prognosis for live birth among untreated infertile couples. Fertil. Steril., 64, 22 28. Daya, S. et al. (1995) Natural cycles for in-vitro fertilization: cost-effectiveness analysis and factors influencing outcome. Hum. Reprod., 10, 1719 1724. Dunphy, B.C. et al. (1989) Female age, the length of involuntary infertility prior to investigation and fertility outcome. Hum. Reprod., 4, 527 530. Eimers, J.M. et al. (1994) The prediction of the chance to conceive in subfertile couples. Fertil. Steril., 61, 44 52. ESHRE Capri Workshop (1996) Guidelines to the prevalence, diagnosis, treatment and management of infertility, 1996. Hum. Reprod., 11, 101 132. Fluker, M.R. and Tiffin, G.J. (1996) Assisted reproductive technologies a primer for Canadian physicians. J. Soc. Obstet. Gynecol. Can., 18, 451 465. Goeree, R., Labelle, R. and Jarrell, J.F. (1993) Cost-effectiveness of an in-vitro fertilization program and the costs of associated hospitalizations and other infertility treatments. In Royal Commission on New Reproductive Technologies, New Reproductive Technologies and the Health Care System: The Case for Evidence-Based Medicine. Ministry of Supply and Services Canada, Ottawa, Canada, pp. 569 599. Goeree, R. et al. (1996) An assessment of the burden of illness due to osteoporosis in Canada. J. Soc. Obstet. Gynecol. Can., 18 (Suppl.), 15 24. International Working Group for Registers on Assisted Reproduction (1995) World Collaborative Report 1993. International Federation of Fertility Societies, Montpellier, France. Mosher, W. and Pratt, W. (1990) Fecundity and Infertility in the United States, 1965 1988. US Government Printing Office, Hyattsville, MD, USA. Mosher, W.D. and Pratt, W.F. (1991) Fecundity and infertility in the United States: incidence and trends. Fertil. Steril., 56, 192 193. Neumann, P.J., Weinstein, M.C. and Gharib, S.D. (1994) The cost of a successful delivery with in vitro fertilization. N. Engl. J. Med., 331, 239 243. Odem, R.R. et al. (1991) Therapeutic donor insemination: a prospective randomized study of scheduling methods. Fertil. Steril., 55, 976 982. Patton, P.E. et al. (1992) Intrauterine insemination outperforms intracervical insemination in a randomized, controlled study with frozen, donor semen. Fertil. Steril., 57, 559 564. Peterson, C.M. et al. (1994) Ovulation induction with gonadotropins and intrauterine insemination compared with in vitro fertilization and no therapy: a prospective, non-randomized, cohort study and meta-analysis. Fertil. Steril., 62, 535 544. Robinson, J.N. et al. (1992) A randomized prospective study to assess the effect of the use of home urinary luteinizing hormone detection on the efficacy of donor insemination. Hum. Reprod., 7, 63 65. Royal Commission on New Reproductive Technologies (1993) Proceed with Care: Final Report of the Royal Commission on New Reproductive Technologies. Ministry of Government Services Canada, Ottawa, Canada, Vol. 1. Statistics Canada (1996) Consumer Prices and Price Indexes: October December 1995. Minister of Industry, Ottawa, Canada. Subak, L.L., Adamson, D. and Boltz, N.L. (1992) Therapeutic donor insemination: a prospective randomized trial of fresh versus frozen sperm. Am. J. Obstet. Gynecol., 166, 1597 1606. Taylor, P.J. and Collins, J.A. (1992) Unexplained Infertility. Oxford Medical Publications, New York, NY, USA. US Congress, Office of Technology Assessment (1988) Infertility: Medical and Social Choices. US Government Printing Office, Washington, DC, USA. Wilcox, L.S. and Mosher, W.D. (1993) Use of infertility services in the United States. Obstet. Gynecol., 82, 122 127. Received on October 3, 1996; accepted on March 3, 1997 958