Hum. Reprod. Advance Access originally published online on June 21, 2007
Human Reproduction 2007 22(8):2344-2345; doi:10.1093/humrep/dem160
Letters to the Editor |
Pregnancy is predictable: a large-scale prospective external validation of the prediction of spontaneous pregnancy in subfertile couples
Department of Obstetrics and Gynaecology, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands Department of Medical Affairs, Radboud University Nijmegen Medical Centre, PO Box 9101, 6500 HB Nijmegen, The Netherlands
1 Correspondence address. Tel.: +31-24-361-4748; Fax: +31-24-366-8597; E-mail: j.smeenk{at}obgyn.umcn.nl
Models are nowadays being put forward as an essential tool for clinical decision-making and in counselling patients. They are introduced as a necessary tool to prevent unnecessary treatment. In fact that is what a clinician wants: a crystal ball to predict pregnancy in order to counsel patients in a reliable way.
Previously, in a paper by Hunault et al. (2004)
, a synthesis model was introduced based upon the original data of three previously published studies to predict spontaneous pregnancy in a subfertile population (Eimers et al., 1994; Collins et al. 1995; Snick et al., 1997).
In the recent paper by Van der Steeg et al. (2007), the dream finally seems to come true, since a prospective validation study resulted in a paper bearing the title ‘pregnancy is predictable...’. A prospective validation study is described, in which a consecutive cohort of 3021 subfertile couples was studied. Our centre was among the 38 centres that included patients.
An essential step in the validation of the model of Hunault et al. (2004) was performed: the authors validated the model by applying it to another population than the one in which it was developed. This is important because prediction models tend to be overoptimistic when applied in the population in which they are developed compared with when applied in other populations (Stolwijk et al., 1996), although the authors state differently in the manuscript.
From the calibration plot in the paper, where the cohort was split up into 10 groups based on the deciles of the calculated probabilities, it becomes clear that the majority of the patients has a predicted probability which is around the mean predicted cumulative ongoing pregnancy rate (32%). As this percentage will be the prior probability of a spontaneous pregnancy without any information about the patient, we feel that a prognostic model is only useful if it changes this prior probability in an accurate way.
Furthermore, it was stated that in 1104 couples (36%) the probability of a spontaneous pregnancy was over 40%, being the cut-off percentage for expectant management. In the calibration plot, however, only two deciles corresponding with 611 couples with a probability of over 40% were described. This leads to speculation about the remaining couples.
Lastly, by validating the model that was also used in the decision whether or not to treat patients, a bias towards better calibration was introduced.
In contrast with the authors, we conclude from the paper that for most patients pregnancy cannot be predicted in a reliable way. The discriminative capacity, expressed by the c-index (which is comparable to the area under the ROC-curve) was reported to be 0.59. (One has to remember that a perfect model has a c-index of 1.0 and that a c-index of 0.5 means that prediction is no better than the flip of a coin.)
Previously, another model was validated by our group and since a c-index of 0.63 was found, we concluded that for the majority of the women the model did not give any more certainty (Smeenk et al. 2000).
Van der Steeg et al. (2007) state in the discussion that it is more important to accurately estimate whether a couple has a high or low chance to conceive than to know exactly which couple will conceive and which will not. However, this is the problem in counselling, as the couple wants an individual approach. In fact, the couple does want to know whether they will become pregnant or not and a perfect model would indicate a pregnancy chance of 0 or 100%. The high percentage of couples being treated within 6 months after the workup, sometimes despite their high prognosis and expectant study design, underlines this issue.
In summary, models to predict pregnancy can well be used to identify different prognostic groups in a population and thereby serve as a tool in patient counselling and shared decision-making. The discriminative capacity however is very limited, meaning that models provide limited information about who will or will not be pregnant.
References
Collins JA, Burrows EA, Willan AR. The prognosis for live birth among untreated inferile couples. Fertil Steril (1995) 64:22–28.[Web of Science][Medline]
Eimers JM, te Velde ER, Gerritse R, Vogelzang ET, Looman CW, Habbema JD. The prediction of the chance to conceive in subfertile couples. Fertil Steril (1994) 61:44–52.[Web of Science][Medline]
Hunault CC, Habbema JDF, Eijkemans MJC, Collins JA, Evers JLH, te Velde ER. Two new prediction rules for spontaneous pregnancy leading to live birth among subfertile couples, based on the synthesis of three previous models. Hum Reprod (2004) 19:2019–2026.
Smeenk JMJ, Stolwijk AM, Kremer JAM, Braat DDM. External validation of the templeton model for predicting success after IVF. Hum Reprod (2000) 15:1065–1068.
Snick HK, Snick TS, Evers JL, Collins JA. The spontaneous pregnancy prognosis in untreated subfertile couples: the Walcheren primary care study. Hum Reprod (1997) 12:1582–1588.
Stolwijk AM, Zielhuis GA, Hamilton CJ, Straatman H, Hollanders JM, Goverde HJ, van Dop PA, Verbeek AL. Prognostic models for the probability of achieving an ongoing pregnancy after in-vitro fertilization and the importance of testing their predictive value. Hum Repod (1996) 11:2298–2303.
Van der Steeg JW, Steures P, Eijkemans MJC, Habbema JDF, Hompes PGA, Broekmans FJ, van Dessel HJHM, Bossuyt PMM, van der Veen F, Mol BWJ. Pregnancy is predictable: a large-scale prospective external validation of the prediction of spontaneous pregnancy in subfertile couples. Hum Reprod (2007) 22:536–542.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||