A review of the medical literature was necessary to absorb the increasing volume of published information. Meta-analysis seeks to answer a certain clinical question using a well-defined strategy to locate relevant evidence, assess the available studies using clear methodological criteria and formally summarise the results. This method consists of an analytical approach in which different, independent studies are integrated and the results are combined into a single common result. Compared with narrative reviews, meta-analysis has the great advantage of being less influenced by a reviewer's personal opinion, thus providing impartial conclusions. Additionally, all the results can easily be recalculated and compared with the conclusions stated by the authors. Even when it does not produce definitive conclusions about the usefulness of a treatment or technique, a meta-analysis may indicate the need for further randomised studies on the subject. Furthermore, previous meta-analyses allow the identification of the most important issues to be analysed in future research studies. Based on positive results in animals, several RCTs evaluating the effects of gamete and embryo cultures at low O2 concentrations (~ 5%) on clinical outcomes have been published, especially recently [24, 28, 35, 36, 38–40]; however, the conclusions of these studies have not always been consistent. Therefore, given the clinical potential of this method, a review of this subject was deemed useful.
Despite the apparent positive trend with low O2 levels, the results showed similar rates of fertilisation, implantation (generally and in the subgroup with transfer at day 2 or 3), and ongoing pregnancy. Hence, despite some positive results, it seems premature to recommend the use of low O2 tension for oocyte and embryo culture. To achieve a low oxygen concentration, an incubator should have sensors for both CO2 and O2; however, the commonly used units only have sensors for CO2. In addition, nitrogen is initially used to purge O2 from the incubator. Thus, the cost of supplying these three gases is greater than that for a typical incubator because each time the incubator is handled (i.e., when the door is opened), the interior atmosphere must be re-equilibrated with a further injection of nitrogen or a balanced mixture of gases to maintain a low O2 concentration . One option would be exchanging the typical large incubators for low-volume units (e.g., benchtop incubators) that use continuous gas flow [46, 47]. In either case, the decision to use low O2 concentrations implies necessary changes in laboratory practices with important economic repercussions. Therefore, we believe that additional randomised controlled trials are still necessary before evidence-based recommendations can be provided.
However, in the present meta-analysis, embryonic culture at low O2 concentrations displayed favourable effects The implantation rate was highly significant for blastocyst transfers only (P = 0.006). With only 4 studies evaluated for implantation rate and ongoing pregnancy rate, the P-values were 0.06 and 0.051 for these parameters. This being two of the three defined end-points for this study suggests that low-oxygen culture may be beneficial. Similar positive results excluded from this meta-analysis (e.g., data reported by only one study or data presented in a way making calcualtions impossible) have been reported by some RCTs (increases in biochemical [37, 39] and cumulative  pregnancy rates and higher rates of live births ). In addition, other non-randomised studies have also reported increased implantation  and pregnancy [21, 48, 49] rates. On the basis of these data, it may seem attractive to consider low O2 tension as a means to improve clinical outcomes.
Meta-analysis also presents problems, such as the quality of the primary studies or the form in which data are reported and the dependence on a sufficient number of eligible studies to justify the statistical analysis. Methodological problems caused by clinical heterogeneity and insufficient power (low sample size) cause difficulty in drawing inferences from the meta-analysis. In this study, heterogeneity was found only for the general implantation rate, which may be related to the calculations being carried out with the results of transfer of embryos in the cleavaged stage and in the blastocyst stage together. No heterogeneity was found for the other outcomes. This reflects a relative agreement among the trials regarding the studied parameters and has particular importance when statistically significant outcomes were found. Nevertheless, it should be stressed that tests of heterogeneity among studies have low power. Therefore, where there is a non-significant test for heterogeneity among studies, it may be that a relationship is present but remains undetected.
This meta-analysis failed to show any statistically significant differences in the most clinically relevant outcome: the ongoing pregnancy rate. This observation may be related to the small cumulative sample size (i.e., insufficient power). With respect to the ongoing pregnancy rates, based on the results obtained with OC~5 and OC~20 embryo sets in the general population (34.6%, 31.6% and 484/1, 398, 450/1, 423, respectively), the subgroup with transfers at day 2 or 3 (31.1%, 29.2% and 349/1, 121, 338/1, 159, respectively) and the subgroup with transfers at the blastocyst stage (48.7%, 135/277 and 42.4%, 112/264, respectively), the ability to detect a difference of 5% with a power of 80% would require approximately 4000 (general population), > 4, 000 (day 2 or 3) and 2, 000 patients blastocyst) respectively, to reach definitive conclusions. Thus, for more consistent conclusions, this meta-analysis suggests that researchers should wait for the results of new RCTs that provide more information on clinical parameters. In addition, although smaller studies conducted on diverse populations may better reflect the natural heterogeneity of treatment effectiveness as found in daily practice, large studies may produce a more precise answer. Thus, further RCTs with larger sample size will be helpful for the corroboration of these results.
In many animal studies, the favourable effect of low oxygen concentration has been reported in the late stages of embryonic (the blastocyst stage) and even in foetal development [6, 8–10, 12, 15, 16, 50–54]. For this reason, the necessity of reducing oxygen levels has been associated with blastocyst culture. In fact, more recent studies evaluating the transfer of human blastocysts have reported significantly better results with the use of low concentrations of oxygen (increased embryo quality [24, 28, 36, 37] and higher implantation , pregnancy and live birth [24, 37] rates). Unfortunately, the differences in embryonic evaluation criteria and in the method of presenting the data among these studies precluded the assessment of embryo quality in this meta-analysis. Although this meta-analysis identified higher implantation rates in the group of embryos cultured at ~5% O2 than those grown at atmospheric O2 levels with respect to blastocyst transfer, it failed to identify any difference in the rates of ongoing pregnancies. Thus, despite the studies suggesting a direct action on the results, the clinically relevant beneficial effects of low O2 concentration were not confirmed. However, it is important to note that we cannot rule out that these negative results may be related to the small number of studies that met the inclusion criteria (only 2), leading to a cumulative sample size insufficient to detect a significant difference. Again, future controlled trials will help to clarify this issue.
It has been argued that deleterious effects of high oxygen levels can have an effect on embryos during the early developmental stages despite the later onset of symptoms that occur during or subsequent to the blastocyst stage, which may affect clinical outcomes [6, 24, 46, 50]. Recently, Nannasy et al.  reported that human embryos cultured at ~20% oxygen concentration for days 1-2 and subsequently at 5% oxygen for days 3-5 did not display better implantation, pregnancy, or blastulation rates than embryos maintained at 20% oxygen for the entire time period (0-3/5 days). The authors speculated that, among the possible explanations, embryos cultured in low oxygen on days 3-5 may not be able to overcome defects induced during days 1-2 of culture in atmospheric oxygen. Nonetheless, human studies examining O2 tensions throughout the culture period and performing embryo transfers at day 2 or 3 [21, 24, 35, 38, 40] did not identify, in most cases, beneficial clinical outcomes associated with reduced oxygen levels, and these data were confirmed by the calculations of this meta-analysis. Future controlled trials will help to clarify this issue and should include results from later clinical stages, such as live birth rate, in the analysis.
In addition, we should consider that several factors, such as differences in patient structure, embryo transfer policy, duration of cultivation, and culture media, could contribute to the observation of the contradictory outcomes of published studies on low oxygen. The possible more/less protective effect of various culture media against ROS and different concentrations of protective molecules has to be highlighted. Problems with the maintenance of stable low O2 atmosphere in the incubators during manipulation of embryos should also be stressed. Such differences must be considered in future trials or meta-analysis.
In conclusion, the findings of this meta-analysis demonstrate that, despite some promising results, it seems too early to conclude that low O2 culture conditions has an effect on IVF outcome. Additional randomised controlled trials are necessary before evidence-based recommendations can be provided. It should be emphasised that the present meta-analysis does not provide any evidence that low oxygen concentration is unnecessary.