Much of what we learn about medicine comes from the newspapers. Unfortunately, things get reported in ways destined to grab our attention. And while the headlines are generally credible, results of studies can be presented in ways that are misleading. Making sense out of the numbers you read can be confusing, especially since math is not everyone’s forte. Statistics, in particular, is a branch of math
that is particularly formidable. So it is important to have a little understanding about how statistics work when it comes to evaluating the benefits and risks of a medical treatment, especially regarding hormone replacement therapy (HRT).
After the Women’s Health Initiative (WHI) was published in 2003, guidelines came out that discouraged women from taking HRT. This was because this famous study concluded that the risks of taking hormones after menopause outweighed the benefits. This caused many women, and their doctors, to decide that estrogen replacement was not safe. Millions of women quit their hormones and millions since then have not been placed on them. This is unfortunate because today experts in menopause believe that the benefits of starting HRT at the time of menopause outweigh the risks.
The reason for this turnaround is many-fold. We now know that there are serious shortcomings regarding the way the WHI’s conclusions have been applied. (See my article, The WHI and Why Women Need to Know About It). In addition the way the study’s findings were communicated to the public played a pivotal role. They were reported in a way that grossly over-exaggerates the risks of taking hormones. This happened because there are various statistical ways numbers can be presented – making it appear that the risks of a treatment are very great when they are actually very small.
This was the case in the WHI. Unfortunately many doctors did not explain this to their patients and simply advised them not to take HRT based on the conclusions of the study. However, had women been given a truer picture of the results, I think they would have been better informed about whether or not they feel comfortable taking HRT. This is where a knowledge of statistics would have helped them.
Before going into further detail about how the WHI results were reported, let me give you a hypothetical example to illustrate what I am talking about.
Let’s make up a study. The purpose is to test if eating chocolate causes skin cancer. We recruit 2,000 people to be in the study. We then have half of the people eat one Hershey bar a day for five years. The other half are not allowed to eat any chocolate for all those years. As time goes on, we check for skin cancer in all of the volunteers. At the end of the study, the number of people with skin cancer in each group is counted.
Here are the results:
- There were six people who developed skin cancer in the group of 1,000 who ate a Hershey bar daily
- There were three people who developed skin cancer in the group of 1,000 who were not allowed to eat chocolate.
How do you report these results?
One could easily say that chocolate eaters are twice as likely to get skin cancer compared to non-eaters, since six is twice as much as three. Or, put another way, there were 50% more cases of cancer in the chocolate group, which is also the same as saying that eating chocolate every day increases the risk of skin cancer by 50%!
All of these conclusions sound pretty bad, don’t they? It sounds like it means that 50% of people eating chocolate will get skin cancer, or that if you eat chocolate you have a 50% chance of getting skin cancer.
This is far from what this study shows, which is that only three more people out of 1,000 would be at risk of skin cancer if they ate chocolate every day. This is only a 0.3% risk of developing skin cancer. This is not a very high risk.
You should also realize that the other 997 people who ate chocolate every day did not get skin cancer. So, if you were deciding whether you should give up chocolate, you would have to decide if you are more likely going to be one of those 997 people who didn’t get skin cancer or you are more likely to be one of those three that did. Based on this, you then decide if it is worth giving up chocolate.
Personally, I think the odds are pretty good that I would be one of those 997 who didn’t get skin cancer from chocolate and I, for one, would be willing to take this 0.3% risk rather than giving up chocolate for five years!
By looking at the results this way, I am looking at my absolute risk, which 3 out of 1,000. Describing one’s absolute risk is a much more meaningful way of presenting the results than telling someone that eating chocolate will increase their risk of skin cancer by 50%. The 50% calculation is based on the statistical measure called relative risk. The relative risk is derived by comparing the number of people who got skin cancer in each group to each other, without taking into account how many total people were in the study.
Hopefully this makes sense.
Now let’s talk about the WHI and its results.* When the WHI conclusions came out, we were told that HRT increased the risk of a heart attack by 25%; increased the risk of breast cancer by 26%, and that it doubled the risks of a blood clot. This news was certainly alarming, and based on this report you can see why women abandoned their hormones. However many people didn’t understand that these numbers represented one’s relative risk.
Women should have been presented with the absolute risk numbers. This reveals that out of 10,000 women who took HRT, there were only 8 more women who developed breast cancer and 7 more who had a heart attack.** This equates to an absolute risk of 0.08% and 0.07% respectively. Or put another way, there is less than a one in 1,000 chance of being diagnosed with one of these adverse events if you took HRT. This certainly seems like a much lower risk than having a 25% higher risk of a negative outcome.
Furthermore, once armed with this knowledge, women can assess how the absolute risks compare to the benefits. And while the purpose of this blog is to enlighten you about the various ways results can be statistically presented, it needs to be emphasized that taking HRT leads to significant benefits. These were also quantified in the WHI and included the finding that for the 10,000 women taking HRT there were 6 fewer cases of colon cancer and 5 fewer fractures when compared to the women not taking HRT. And even more importantly, taking HRT did not increase the risk of death during the study.
In conclusion, when a woman is considering whether or not to take HRT, it is important that she receives thorough counseling on the benefits and risks, taking into account her unique health situation, symptoms, and preferences. As part of this discussion, women should be informed of the absolute risk of any consequence because this is much more meaningful than using relative risk figures.
*Note: there were two parts to the WHI study. Women who still had a uterus were treated with both Premarin and medroxyprogesterone and were in the first study. The results I describe in this blog pertain to the women in this part of the study. The second part of the study included women who had a hysterectomy and therefore were only given Premarin. Women in this group actually had a lower relative and absolute risk of getting breast cancer or a heart attack. In fact their risk was less than the women who did not take HRT!
**These were approximately 8,000 women in the treatment and placebo groups in this section of the WHI study. To make the numbers easier to interpret, they are rounded up to 10,000 women in each group.
Be the first to reply