Help understanding stats?

ColleenS80

I took a stats class in college, but clearly I’ve slept since then.

Can anyone help me understand the stats listed in the linked article? Now I do understand that the article is a few years old AND these women were treated a long time ago, but I just have a hard tIme with the stats...

The first two sets seem pretty low, but then the third, where they talk about cumulative risk, they seem really high. I know “cumulative” changes things but I can’t remember how exactly. Anyone help? Thank you!!!!

https://www.mdedge.com/acssurgerynews/article/105843/breast-cancer/significant-risk-relapse-remains-er-positive-breast#

gemmafromlondon

I found this a very difficult article because I could not discover what the percentages meant statistically. I know it was for reccurence of bc but After 10 - 15 years as well as the bc deaths, surely some patients had died of other causes, thus the sample size would have been reduced, and the percentage rate different from that achieved in the early years? If it was a percentage of survival one needs to know the number of deaths from other causes - unless the investigators considered that there was some underlying aetiology connected with bc.

For example 1 - 5 years with sample - 4000 30 die = 0.75%. 5 - 10 years sample now 3970 if 30 die = 0.76% BUT if 20 had died of other causes sample - 3950 and deaths = 1.26% .

I am not sure what the cumulative risks involved - weight?age/comorbidity?. or was it just the addition of the time factor? Or positive nodes? .A higher number of positive nodes usually indicates a more advanced cancer. It would seem that breast cancer recurrence is far less likely than other cancers (if that is what they mean by distal).

In all - it is very opaque but I am no expert

Gemma

ColleenS80

Thanks Gemma...I’m finding it very difficult as well

hapa

I have quite a bit of experience with statistics and it is really hard to understand what this article is talking about. This article is actually just a summary of an article in a medical journal. It's possible the author didn't understand statistics enough to summarize it well.

The first two sets of numbers are average risk per year, or "annualized risk". So in the first paragraph, the risk of recurrence is 9.9% on average per year, for ER+'s, every year, from years 1 through 5 (in reality, the risk is higher in years 1-2 and drops every year after that, but they average it out over 5 years). Then it drops to 5.4% per year from years 5 through 10, meaning for people who didn't have recurrence after 5 years, they had a 5.4% chance every year of having recurrence for the next five years. It comes out to something like a cumulative 67% chance of recurrence over 20 years if you do the math. That seems like a lot to me, but I found the original article and it appears to be correct. But here they are talking about both local and distant recurrence.

In the third set of stats, they are talking about distant recurrence only, and they are giving cumulative stats. So after 5 years, 27.1% of ER+ people had distant recurrence. After 10 years, 31.9% had distant recurrence but this includes all the people who had it in the first five years as well as the folks who had it in years 5-10.

The original article is here: http://ascopubs.org/doi/full/10.1200/JCO.2015.62.3.... It is easier to understand and has charts. It also has stats for local, regional, contralateral, and distant recurrence.

One more thing I want to point out, and only because I didn't realize this until recently. When you're looking at an "overall survival" statistic, that includes people who died from other causes. When you're looking at overall survival on a 20-year study, A LOT of those people will have died of other causes. If you consider the fact that the average age of diagnosis for breast cancer is 61-62 years old, and the risk of death over 20 years for the average 62 year old is 35% as of 2014, you can imagine that for studies that started 20 years ago even more of those women would have died of something else. I get my stats on odds of dying from this table: https://www.finder.com/life-insurance/odds-of-dying and just keep them in mind whenever I'm looking at overall survival statistics. It is really helpful if the study gives average age. For the large herceptin studies for example, the average age of participants was 52.

I hope this helps!

ColleenS80

thanks hapa! Lots of useful information!

I certainly don't like that 67%number either-yuck.

This excerpt from the study is frustrating:

“Long-term follow-up is an orphan item that is not given sufficient priority or attention in the clinical trials community."

Since I'm just 38, late recurrence is something that can really keep me up at night. I guess all I can do is continue to hope and pray that if I have a recurrence, it will be far enough off that we will have better ways to control it by then.

Thanks again, I really appreciate your input!

FarAwayToo

hapa, thank you for providing a link to the original study. As grim as they sound, I don't think the stats cited are relevant to women treated in the 21st century.

Women in the study were treated between 1978 and 1985, which means they weren't treated with taxanes (if they were treated with chemotherapy) or AIs. This is especially important when considering late recurrence in ER+ cancers. I don't have links to studies off hand, but I remember several which had proven that benefit from AI vs Tamoxifen for DFS holds for much longer than 5 years that AIs were taken. I see that both of you, hapa and Colleen, are taking AIs with ovarian suppression. I also chose AI + OS (and recently oophorectomy) to lower my chances of recurrence. It used to be that women our age (I was DXed at 40) weren't even considered for AI therapy, but SOFT and TEXT trial results are changing the practices rapidly.

ColleenS80

All encouraging points, FarAway. Thanks so much for sharing!