When you search “egg freezing utilization rate” the top result is this study from the UK1 which states that only 16% of patients returned to use their frozen eggs. And this study from Sweden2 found a similar utilization rate of 15%. Plus, in both of these studies some patients did more than one retrieval cycle - dividing the number of women who used their eggs by the total number of retrieval cycles results in an even lower rate of utilization *per cycle* at 15% and 11% respectively. But I expect that both of these *severely* underestimate where the utilization rate will end up with more time.

The UK study uses data on egg freezing cycles done between January 2016 and March 2022 and calculates the utilization rate as the percentage of those women who had returned to use their eggs by March 2023. Similarly, the Swedish study uses data on egg freezing cycles done between August 2011 and August 2017 and calculates the utilization rate as the percentage of those women who had returned to use their eggs by November 2018.

In both cases they’re counting patients who froze their eggs only a year or two ago and hadn’t yet returned to use them as “not utilizing their eggs”. Technically true for now… but we should expect many of those women to return for their eggs in future years. When freezing eggs for the purpose of fertility extension most women don’t expect to use them within only a year or two. In the Swedish study the women who returned for their eggs “had a mean age of 38.7 years (36-42 years) at freezing and 42.7 years (38-45 years) at warming” giving an average time of 3.9 years between freezing and usage while the average time between freezing and retrieval in the UK study was 4.4 years. Both of these are similar to the median time between freeze and thaw of 4.2 years found in this large study3.

On top of that, the number of egg retrievals done for the purpose of fertility extension has increased over time. The UK study included a bar chart showing the number of cycles done over time but it wasn’t broken down to the individual year and the bar heights didn’t seem to quite sum to the total…

I couldn’t find the number of egg freezing cycles done by year in the Swedish study, but this study from the UK4 includes a figure showing the number of egg freezing and thawing cycles from 2010 to 2016.

I eyeballed the numbers of egg freezing cycles per year from this figure, put them into excel (file is here, tab is named “freeze and thaw cycles in UK”) and got the percentage increase in number of freezing cycles done per year. For years after 2016 I assumed the growth rate was equal to the geometric average of the annual growth from 2010 to 2016 which was 37%.

To estimate the number of egg freezing cycles per year in the Swedish study I assumed the annual growth rates were the same as those in the above table. I also made the assumption that all utilizations happened 4 years after the eggs were frozen, which means I assumed all retrievals happened in the last two years of the study. I then assumed that the trend in utilizations exactly matched that of the egg freezing cycles but with a 4 year lag. Finally, I divided my modeled utilization numbers per year by the modeled freezing numbers per year from 4 years prior (same file, data in column J on tab named “modeled utilization”). This resulted in an estimated utilization rate per freezing cycle of 72%!

I did something similar for the UK study except I used the numbers I eyeballed from the bar chart shown in figure 2 to estimate the growth in number of freezing cycles for years 2018-2022 and used those in the table above for years 2016-2017. My estimate for the utilization rate per freezing cycle in this case was also very high at 64% (same file, data in column K on tab named “modeled utilization”).

Now, of course these estimates involved a lot of assumptions. But even if I assume the freezing cycles were spread out evenly over the 6 year periods in each study and that the retrievals were spread evenly over the final 2 years I’d get utilization rates per cycle of 34% from the Swedish study and 44% from the UK study (same file, see tab “modeled utilization no growth”).

So, overall I think we should expect utilization rates to be quite high. Probably closer to 50% than 15%. But if anyone has an idea for how I could improve this model or access to more relevant data please let me know in the comments!