Download PDF
Semin Reprod Med 2003; 21(1): 039-048
DOI: 10.1055/s-2003-39993
Copyright © 2003 by Thieme Medical Publishers, Inc., 333 Seventh Avenue, New York, NY 10001, USA. Tel.: +1(212) 584-4662

Characteristics of the Best Prognostic Evidence: An Example on Prediction of Outcome after Clomiphene Citrate Induction of Ovulation in Normogonadotropic Oligoamenorrheic Infertility

M. J.C. Eijkemans1 , J. D.F. Habbema1 , B. C.J.M. Fauser2
  • 1Department of Public Health, Erasmus Medical Center, Rotterdam, The Netherlands
  • 2Center of Reproductive Medicine, Erasmus Medical Center, Rotterdam, The Netherlands
Further Information

Publication History

Publication Date:
12 June 2003 (online)

    Permissions and Reprints

ABSTRACT

The standard first-line treatment for normogonadotropic anovulatory infertile patients [referred to as World Health Organization group 2 (WHO 2)] is ovulation induction using clomiphene citrate (CC) in incremental doses. Twenty to 25% of women show clomiphene-resistant anovulation (CRA), that is, they remain anovulatory even after multiple attempts with increased doses of CC. About 50% of the ovulatory CC patients conceive within six CC-induced cycles. Given the heterogeneous nature of the group, the individual prognosis (i.e., the chance of success) will vary considerably between patients. In the event an individual prognosis of each patient would be available before the start of the treatment, the overall efficiency of ovulation induction could be improved. Prognostic evidence at an individual level should use multiple patient variables, including results from previous treatments (if any). When variables are interdependent, a statistical model can be used to relate individual characteristics with the predicted outcome. Such a model will provide estimates of prognosis for individualized patient profiles, allowing new patients to profit from the experience of the cohort of previous patients used to build the model. This paper discusses the prediction of time to pregnancy following induction of ovulation with CC. This prediction was broken down in two steps, leading to two separate prognostic models. The first model predicts an intermediate outcome, the chance that the patient will be CRA (i.e., no ovulation in response to CC medication); the second model predicts the final outcome (time until pregnancy) in women who do ovulate. The CRA model was based on a prospective cohort study of 201 patients with normogonadotropic oligoamenorrheic infertility, 45 of whom were CRA (22%). It contained four predictor variables all related to the diagnosis of PCOS within the group of WHO 2: Increased free androgen index (FAI; hyperandrogenemia), elevated body mass index (BMI; obesity), greater mean ovarian volume (as an ultrasound feature of polycystic ovaries), and amenorrhea were all predictive for CRA. The second model was based on the non-CRA patients and contained two prognostic variables: increased age and oligomenorrhea were predictive for longer time to pregnancy after first ovulation with CC. Using the example of the prediction of time to pregnancy following induction of ovulation with CC, we present and discuss characteristics of good prognostic evidence for clinical use, focusing on study design, statistical analysis, evaluation, and presentation of results.

KEYWORDS

Prognosis - treatment outcome - anovulation - ovulation induction

REFERENCES

  • 1 Pache T D, Chadha S, Gooren L J. et al . Ovarian morphology in long-term androgen-treated female to male transsexuals: a human model for the study of polycystic ovarian syndrome?.  Histopathology . 1991;  19 445-452
    PubMedSearch in Google Scholar
  • 2 van Dessel J H, Schipper I, Pache T D, van Geldorp H, de Jong H F, Fauser B C. Normal human follicle development: an evaluation of correlations with oestradiol, androstenedione and progesterone levels in individual follicles.  Clin Endocrinol (Oxf) . 1996;  44 191-198
    CrossrefPubMedSearch in Google Scholar
  • 3 Laven J S, Imani B, Eijkemans M J, de Jong H F, Fauser B C. Absent biologically relevant associations between serum inhibin B concentrations and characteristics of polycystic ovary syndrome in normogonadotrophic anovulatory infertility.  Hum Reprod . 2001;  16 1359-1364
    CrossrefPubMedSearch in Google Scholar
  • 4 Fauser B C, Pache T D, Lamberts S W, Hop W C, de Jong H F, Dahl K D. Serum bioactive and immunoreactive luteinizing hormone and follicle-stimulating hormone levels in women with cycle abnormalities, with or without polycystic ovarian disease.  J Clin Endocrinol Metab . 1991;  73 811-817
    PubMedSearch in Google Scholar
  • 5 Pache T D, Wladimiroff J W, Hop W C, Fauser B C. How to discriminate between normal and polycystic ovaries: transvaginal US study.  Radiology . 1992;  183 421-423
    PubMedSearch in Google Scholar
  • 6 van Santbrink J E, Hop W C, van Dessel J T, de Jong H F, Fauser B C. Decremental follicle-stimulating hormone and dominant follicle development during the normal menstrual cycle.  Fertil Steril . 1995;  64 37-43
    PubMedSearch in Google Scholar
  • 7 Fauser B C, Van Heusden M A. Manipulation of human ovarian function: physiological concepts and clinical consequences.  Endocr Rev . 1997;  18 71-106
    CrossrefPubMedSearch in Google Scholar
  • 8 Imani B, Eijkemans M J, te Velde R E, Habbema J D, Fauser B C. Predictors of patients remaining anovulatory during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility.  J Clin Endocrinol Metab . 1998;  83 2361-2365
    CrossrefPubMedSearch in Google Scholar
  • 9 Imani B, Eijkemans M J, te Velde R E, Habbema J D, Fauser B C. Predictors of chances to conceive in ovulatory patients during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility.  J Clin Endocrinol Metab . 1999;  84 1617-1622
    CrossrefPubMedSearch in Google Scholar
  • 10 Imani B, Eijkemans M J, de Jong H F. et al . Free androgen index and leptin are the most prominent endocrine predictors of ovarian response during clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility.  J Clin Endocrinol Metab . 2000;  85 676-682
    CrossrefPubMedSearch in Google Scholar
  • 11 Imani B, Eijkemans M J, te Velde R E, Habbema J D, Fauser B C. A nomogram to predict the probability of live birth after clomiphene citrate induction of ovulation in normogonadotropic oligoamenorrheic infertility.  Fertil Steril . 2002;  77 91-97
    CrossrefPubMedSearch in Google Scholar
  • 12 Imani B, Eijkemans M J, Faessen G H, Bouchard P, Giudice L C, Fauser B C. Prediction of the individual follicle-stimulating hormone threshold for gonadotropin induction of ovulation in normogonadotropic anovulatory infertility: an approach to increase safety and efficiency.  Fertil Steril . 2002;  77 83-90
    PubMedSearch in Google Scholar
  • 13 Templeton A, Morris J K, Parslow W. Factors that affect outcome of in-vitro fertilisation treatment.  Lancet . 1996;  348 1402-1406
    CrossrefPubMedSearch in Google Scholar
  • 14 Eimers J M, te Velde R E, Gerritse R, Vogelzang E T, Looman C W, Habbema J D. The prediction of the chance to conceive in subfertile couples.  Fertil Steril . 1994;  61 44-52
    PubMedSearch in Google Scholar
  • 15 Collins J A, Burrows E A, Wilan A R. The prognosis for live birth among untreated infertile couples.  Fertil Steril . 1995;  64 22-28
    PubMedSearch in Google Scholar
  • 16 Snick H K, Snick T S, Evers J L, Collins J A. The spontaneous pregnancy prognosis in untreated subfertile couples: the Walcheren primary care study.  Hum Reprod . 1997;  12 1582-1588
    CrossrefPubMedSearch in Google Scholar
  • 17 Stolwijk A M, Hamilton C J, Hollanders J M, Bastiaans L A, Zielhuis G A. A more realistic approach to the cumulative pregnancy rate after in-vitro fertilization.  Hum Reprod . 1996;  11 660-663
    PubMedSearch in Google Scholar
  • 18 Cross S S, Harrison R F, Kennedy R L. Introduction to neural networks.  Lancet . 1995;  346 1075-1079
    CrossrefPubMedSearch in Google Scholar
  • 19 Baxt W G. Application of artificial neural networks to clinical medicine.  Lancet . 1995;  346 1135-1138
    CrossrefPubMedSearch in Google Scholar
  • 20 Breiman L, Friedmand L, Olshen R, Stone C J. Classification and regression trees. Belmont, CA: Wadsworth; 1984
    Search in Google Scholar
  • 21 Pearl J. Probabilistic reasoning in intelligent systems: networks of plausible inference. Palo Alto, CA: Morgan Kaufmann Publishers; 1988
    Search in Google Scholar
  • 22 Heckerman D E, Horvitz E J, Nathwani B N. Toward normative expert systems, part I: the Pathfinder project.  Methods Inf Med . 1992;  31 90-105
    PubMedSearch in Google Scholar
  • 23 Shwe M A, Middleton B, Heckerman D E. et al . Probabilistic diagnosis using a reformulation of the INTERNIST-1/QMR knowledge base I: the probabilistic model and inference algorithms.  Methods Inf Med . 1991;  30 241-255
    PubMedSearch in Google Scholar
  • 24 Harrell Jr E F, Lee K L, Matchar D B, Reichert T A. Regression models for prognostic prediction: advantages, problems, and suggested solutions.  Cancer Treat Rep . 1985;  69 1071-1077
    PubMedSearch in Google Scholar
  • 25 Steyerberg E W, Eijkemans M J, Harrell Jr E F, Habbema J D. Prognostic modelling with logistic regression analysis: a comparison of selection and estimation methods in small data sets.  Stat Med . 2000;  19 1059-1079
    CrossrefPubMedSearch in Google Scholar
  • 26 Miller A J. Subset selection in regression. London: Chapman & Hall; 1990
    Search in Google Scholar
  • 27 Hurvitch C M, Tsai C L. The impact of model selection on inference in linear regression.  Am Stat . 1990;  44 214-217
    PubMedSearch in Google Scholar
  • 28 Steyerberg E W, Eijkemans M J, Habbema J D. Stepwise selection in small data sets: a simulation study of bias in logistic regression analysis.  J Clin Epidemiol . 1999;  52 935-942
    CrossrefPubMedSearch in Google Scholar
  • 29 Little R JA. Regression with missing X's: a review.  J Am Stat Assoc . 1992;  87 1227-1237
    PubMedSearch in Google Scholar
  • 30 Efron B, Tibshirani R. An introduction to the bootstrap. New York: Chapman & Hall; 1993
    Search in Google Scholar
  • 31 Hosmer D W, Lemeshow S. Applied logistic regression. New York: John Wiley; 1989
    Search in Google Scholar
  • 32 Copas J B. Regression, prediction and shrinkage (with discussion).  J Royal Stat Soc, Series B . 1983;  45 311-354
    PubMedSearch in Google Scholar
  • 33 Van Houwelingen C J, Le Cessie S. Predictive value of statistical models.  Stat Med . 1990;  9 1303-1325
    PubMedSearch in Google Scholar
  • 34 Harrell Jr E F, Lee K L, Califf R M, Pryor D B, Rosati R A. Regression modelling strategies for improved prognostic prediction.  Stat Med . 1984;  3 143-52
    PubMedSearch in Google Scholar