Human Reproduction Update Advance Access originally published online on January 28, 2008
Human Reproduction Update 2008 14(2):195-196; doi:10.1093/humupd/dmm045
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Reply: Comment on: Is progesterone elevation on the day of human chorionic gonadotrophin administration associated with the probability of pregnancy in in vitro fertilization? A systematic review and meta-analysis. By Venetis et al (2007)
Unit for Human Reproduction,
1st Department of Obstetrics and Gynaecology,
Papageoriou General Hospital
Aristotle University of Thessaloniki,
Nea Efkarpia, Peripheral Road,
Thessaloniki,
Greece
We would like to thank Dr Bosch for his interest in our work (Venetis et al., 2007
). Dr Bosch raises certain issues regarding the interpretation of the results of this systematic review and meta-analysis, which deserve commenting.
(i) The concept of trend
In the systematic review and meta-analysis by Venetis et al. (2007), a clear research question was asked: is progesterone elevation on the day of hCG administration associated with the probability of pregnancy, in women undergoing ovarian stimulation using gonadotrophins and GnRH analogues for IVF? The results of the quantitative data synthesis, which was performed using a methodology described in detail in the published manuscript, failed to reject the null hypothesis at an alpha error level of 0.05. The conclusion derived from these data, as stated in the manuscript, was that ‘the best available evidence does not support an association between progesterone elevation on the day of hCG administration and the probability of clinical pregnancy in women undergoing ovarian stimulation with GnRH analogues and gonadotrophins for IVF’. The concept of ‘trend’ to which Dr Bosch refers to, is difficult to define and is not compatible with hypothesis testing and statistical inference. Therefore, it cannot be the result of a meta-analysis, in which the null hypothesis is either accepted or rejected.
(ii) Clear clinical relevance of a non-statistically significant result
Failing to reject the null hypothesis might be attributed to lack of power, however, it could also reflect the absence of a true difference.
Thus, it is not clear how differences that do not reach statistical significance can be of ‘clear’ clinical relevance, as Dr Bosch suggests in his letter.
(iii) Analysis of ‘crude data’
Such an approach, known as ‘treat-as-one-trial’, is considered statistically inappropriate, since it disregards study-to-study variation and it is prone to biased results and the well-known issue of Simpson's Paradox (Bravata and Olkin, 2001; Altman and Deeks, 2002). For these reasons, combining the results from individual studies, where feasible, is performed with the use of proper meta-analytic models (Deeks et al., 2001). The application of such models in the current meta-analysis failed to support the presence of statistically significant differences in terms of pregnancy rates in patients with and those without progesterone elevation.
(iv) Association between effect size and threshold level of progesterone used to classify patients as those with and those without progesterone elevation
Analyzing the potential association between progesterone thresholds and effect size, might be of interest. However, it would require a much larger number of studies, in which various progesterone thresholds should have been used. This task was not feasible in the present systematic review and meta-analysis due to the limited number of eligible studies.
(v) Analysis of secondary outcomes
Prediction of the occurrence of progesterone elevation on a certain patient population might be performed by using multivariate logistic regression analysis, as Dr Bosch suggests. However, this was not the aim of the current systematic review and meta-analysis, and even in the context of a secondary, exploratory analysis, it would not have been feasible, due to the limited number of studies.
(vi) In his letter Dr Bosch states "In conclusion, the available evidence suggests that the elevation of circulating P concentrations at the end of the follicular phase has a negative impact on GnRH antagonist cycles."
We do not share Dr Bosch's certainty regarding the negative effect of elevated progesterone on the day of hCG on pregnancy rates in GnRH antagonist cycles. Clearly, based on the results of the current systematic review and meta-analysis (Venetis et al., 2007), such a statement, cannot be supported.
 |
References |
|---|
 TOP References |
|---|
Altman D, Deeks J. Meta-analysis, Simpson's paradox, and the number needed to treat. BMC Med Res Methodol (2002) 2:3.[CrossRef][Medline]
Bravata DM, Olkin I. Simple pooling versus combining in meta-analysis. Eval Health Prof (2001) 24:218–230.
Deeks JJ, Altman DG, Bradburn MJ. Statistical methods for examining heterogeneity and combining results from several studies in meta-analysis. In: Systematic Reviews in Health Care Meta-analysis in Context—Egger M, Davey Smith G, Altman GD, eds. (2001) London: BMJ Publishing Group. 285–312.
Venetis CA, Kolibianakis EM, Papanikolaou E, Bontis J, Devroey P, Tarlatzis BC. Is progesterone elevation on the day of human chorionic gonadotrophin administration associated with the probability of pregnancy in in vitro fertilization? A systematic review and meta-analysis. Hum Reprod Update (2007) 13:343–355.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||