Hum. Reprod. Advance Access originally published online on June 21, 2008
Human Reproduction 2008 23(9):2174-2175; doi:10.1093/humrep/den236
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Letters to the Editor |
Reply: Can assisted reproductive technologies help offset population ageing?
RAND Europe, Westbrook Centre, Milton Road, Cambridge CB4 1YG, UK
1 Correspondence address. Tel: +44-1223-353329; Fax: +44-1223-358845; E-mail: hoorens{at}rand.org
We warmly welcome scrutiny of our work and the open debate that it has prompted. However, we believe Te Velde et al. misrepresent our thesis and misstate our definitions and assumptions.
In our exploratory work, we tested the hypothesis that assisted reproductive technologies (ART) could be part of a population policy mix and concluded that ‘ART does have potential to contribute to the total fertility rate (TFR)’ (Grant et al., 2006
; Hoorens et al., 2007). In reaching this conclusion, we have repeatedly stressed a number of caveats and limitations to our analysis and the need for further research. Indeed, the RAND Europe report to which Te Velde et al. refer is titled ‘A preliminary assessment of the demographic and economic impact of Assisted Reproductive Technologies’ [our italics]. At no time have we made ‘recommendations’ to government about adopting ART as a part of a population policy as Te Velde et al. claim.
Te Velde et al. dispute two parts of our analysis—the infertility definition and the cost–benefit assumptions. First, we stress that the definition of infertility has no material impact on our central finding that the TFR in the UK would increase by 0.04, if Danish ART practice was adopted. We only used the proportion of women with conception difficulties by age to provide an upper bound estimate on the impact of ART on TFR (0.20 in the UK). We do not, as Te Velde et al. suggest, use a 1 year period of unprotected intercourse as a threshold value for the infertility definition. As explained in our paper, there are intrinsic difficulties with such an estimate.
Unfortunately, the literature does not present a coherent and consistent definition of infertility. In the paper, we showed that there is a significant variation between different estimates of infertility. We agree that conservative definitions should be used in order not to overestimate the impact and recommended that future studies should address this issue in more detail. Therefore, we decided to use an average of the estimates with a 2 year threshold produced by Van Balen et al. (1997), Bongaarts (1982) and Dunson et al. (2004).
We appreciate Te Velde et al.'s comment about the accuracy of our assumption that all couples who do have a child after ART would have been otherwise childless. We acknowledge in our paper the non-negligible probability of spontaneous pregnancy for couples undergoing ART. Evers et al. (1998), for example, found a 12 month cumulative pregnancy rate between 2.4% and 6.6% for patients with ‘severe reproductive disorders’ on the waiting list. Nonetheless, a recent paper by Sobotka et al. (2008) using a different methodology resulted in very similar estimates for the ‘net’ impact of assisted reproduction on fertility rates as those resulted from our study. In their analysis, Sobotka et al. do compensate for the probability of spontaneous pregnancy.
It is important to emphasize that our definition of infertility is a classification for the purpose of implementing a forecast model, not a diagnosis. Distinctions must be made somewhere; infertility is spectral, but policy decisions (e.g. ART indications or eligibility) are discrete. Moreover, infertility is distributed both biologically and in terms of outside observers' estimates. Observed experience (e.g. time-to-pregnancy) may increase the accuracy of infertility estimates, but they remain imperfect—even taking into account, as a clinician would, the frequency and timing of intercourse and other observable factors such as smoking, stress, physical condition, etc. Hence, any rules, including those of Te Velde et al., produce errors.
The discussion of classifications is really a matter of thresholds, and not just diagnostic precision. Thresholds are tied to policy and medical decisions and thus must balance costs and benefits. In addition to natural and imperfectly observable heterogeneity in pair fertility across the population, there is a general drift downward with age. Even without cost considerations, setting a threshold represents what statisticians call a ‘bandit’ problem—a choice under uncertainty where gathering more information (e.g. by undergoing or not undergoing ART) changes both beliefs and the true state of the underlying stochastic process. Moreover, infertility itself (i.e. as distinct from clinicians' posterior estimates) is multi-dimensional; it includes the infertility of the pair and the individuals, ‘infertility-without ART’, ‘infertility-under-ART’ and even ‘infertility without ART but after prior ART’, etc. The definition(s) we used are meant to establish working clarity, not guide clinical practice or solve the complex learning and choice problem. Any such definitions serve in policy analysis to create central estimates and establish parameters for sensitivity analysis, not to give precise forecasts.
We have some sympathy with Te Velde et al.'s comments on the cost–benefit analysis. That said, the paper and report stress the limitations of the analysis and, by necessity, some of underlying assumptions. In contrast to Te Velde's claim, we acknowledge the health effects associated with the higher probability of multiple births associated with multiple embryo transfers (De Sutter et al., 2002; Schieve et al., 2004), and make note of the crude assumption that ART babies are identical (biologically and from a socio-economic point of view) to non-ART babies. More importantly, we emphasize the need for modelling the impact of moving to single-embryo transfer on both fertility rates and associated cost–benefits. Further, we note here that if post-natal costs associated with prematurity are included at 
136% of the costs of ART treatment (Collins and Graves, 2000), the ratio would still be of a similar order of magnitude when compared with social interventions such as child-benefits, and hence our conclusion would not be affected.
However, even if ART-related multiple and/or premature births were medically the same as naturally conceived births for the same parental age, etc., cost and clinical outcomes may yet differ due to incentive and selection effects. The ART process focuses attention and thus can change the behaviour of parents and the medical system, and selection for ART treatment can produce a sample that differs from the non-ART population in other (non-fertility) ways correlated with health-care costs and clinical outcomes. Moreover, multiple and premature births may have other cost impacts that should be taken into account where possible. These range from economies of scale in child-rearing to possible parental income benefits of concentrating the child-rearing years and, at least in principle, could partially offset any underestimation of health-care costs.
Thus, the conclusion that our work ‘grossly overestimates’ the impact of ART on fertility is itself in danger of a gross misstatement of our analysis.
References
Bongaarts J. Infertility after age 30: a false alarm. Fam Plan Perspect (1982) 14:75–78.[CrossRef][Web of Science][Medline]
Collins J, Graves G. The economic consequences of multiple gestation pregnancy in assisted conception cycles. Hum Fertil (2000) 3:275–283.[CrossRef]
De Sutter P, Gerris J, Dhont M. A health-economic decision-analytic model comparing double with single embryo transfer in IVF/ICSI. Hum Reprod (2002) 17:2891–2896.
Dunson DB, Baird DD, Colombo B. Increased fertility with age in men and women. Obstet Gynecol (2004) 103:51–56.[CrossRef][Web of Science][Medline]
Evers JL, de Haas HW, Land JA, Dumoulin JC, Dunselman GA. Treatment-independent pregnancy rate in patients with sever reproductive disorders. Hum Reprod (1998) 13:1206–1209.
Grant J, Hoorens S, Gallo F, Cave J. Should ART be part of a population mix? In: A Preliminary Assessment of the Demographic and Economic Impact of Assisted Reproductive Technologies (2006) Cambridge: RAND Europe.
Hoorens S, Gallo F, Cave JAK, Grant JC. Can assisted reproductive technologies help to offset population ageing? An assessment of the demographic and economic impact of ART in Denmark and UK. Hum Reprod (2007) 22:2471–2475.
Schieve LA, Rasmussen SA, Buck GM, Schendel DE, Reynolds MA, Wright VC. Are children born after assisted reproductive technology at increased risk for adverse health outcomes? Obstet Gynecol (2004) 103:1154–1163.[CrossRef][Web of Science][Medline]
Sobotka T, Hansen MA, Kold Jensen T, Pedersen AT, Lutz W, Skakkebæk NE. The contribution of assisted reproduction to completed fertility: an analysis of Danish data. Popul Dev Rev (2008) 34:79–101.[CrossRef][Web of Science]
Van Balen F, Verdurmen JE, Ketting E. Age, the desire to have a child and cumulative pregnancy rate. Hum Reprod (1997) 12:623–627.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||