Introduction

IVF programs seem to be easy to compare, because, uniquely for medicine, the treatment outcome can be quantified: patient will either achieve childbirth or not. On the other hand, one can argue that IVF programs performance is influenced by patient’s selection and therefore the direct comparison is not providing an accurate picture.

Indeed, if an IVF program is overly concerned about its success rates, it may refuse service to the poor prognosis patients and thus its success is not a true reflection of better service, but mere the result of preselecting for treatment better prognosis patients.

Donor egg recipients (DER) represent a group of patients that would be expected to have relatively low variability in prognosis between the IVF programs, since all programs pre-select donors using similar criteria. Therefore, it could be assumed that comparing programs by outcomes in this group would be fairer. Yet, some factors, such as for example, the level compensation, may introduce a bias in direct comparison between IVF programs for outcomes in DER.  

Even though for the reasons above, a direct comparison of IVF programs by using a delivery rates in any group of patients is questionable, the enrollment of DER and regular patients in the same program creates a rare opportunity to address one of the most intriguing and sensitive questions in assisted reproduction: whether the difference in pregnancy rates between programs is due to the difference in therapy or patients selection.  

Indeed, it would be safe to assume that, despite some bias, due to generally similar pre-selection criteria, the groups of DER is overall more uniformed (less variable) between programs than regular patients. Therefore, if the treatment outcome is influenced by patient’s selection, we would expect less variability in outcomes among IVF programs in DER, than in regular patients.

Material and methods

The results for the last available 3 years (2008, 2009 and 2010) were pulled together from CDC database to increase the study group size. Only programs that performed at least 50 donor eggs cycles over 3 years period were included.  This limited the dataset to one hundred clinics that performed a total of 18,439 donor cycles, 73517 cycles in patients under 35 years and 40,414 cycles in patients 38-40 years of age. To asses the variability between IVF programs we used Leven’s test for coefficient of variation (CV).  

Results

1. Comparison of CVs of deliveries per cycle between DER and patients under 35 years of age. 

2. Comparison of CVs fo deliveries per cycle between DER and patients 38-40 years of age

Discussion

Our data show that the variability in outcomes between IVF programs for younger patients is not just lower, but significantly lower than for DER (p<0.001). It would be very hard to argue that random patients, who have a variety of conditions leading to infertility, represent a more uniformed group in their treatment prognosis than pre-selected egg donors. Since that is indeed highly unlikely, it is reasonable to conclude that the difference in pregnancy rates between programs in younger patients is not due to variability in patients populations, but more so to the level of service provided to the patients. It has to be noted though, that due to the study design, only relatively large programs were included into analysis and the conclusion may not hold true for small programs.

Curiously, our calculations also show that the variability in outcomes for older patients (38-40 years of age) is higher than variability in DER outcomes. This may be due to the known fact that in this age group, patients are considerably more diverse in their prognosis. Also, differences   in the programs thresholds for discouraging patients from using their own eggs, if, for example, tests indicate low ovarian reserve, and use donor eggs instead.