Musician and all around good guy, Henry Rollins, quips, “What goes best with a cup of coffee? Another cup.”

We love our coffee. So when new research suggests coffee is good for our health, there is much rejoicing in the land. That’s exactly what happened this month when Circulation published research connecting coffee consumption with a 15 percent decrease in mortality. Amazing news! Fill ‘er up!

This is where I step in and ruin your day as I have done before. More than once. This latest coffee research pulled information from multiple ongoing studies of health professionals. It looked at coffee consumption patterns and death from different causes. The study was observational. This means it looked at behaviors chosen without influence from researchers. Subjects weren’t told whether or not to consume coffee. They just did what they wanted and researchers looked at how that played out.

Observational studies make it possible to get information from extremely large groups of people. This particular study used information from around 200,000 individuals. But we can only show an association with this type of research. We can’t show causation. That’s a big drawback.

We know from this study that drinking three to five cups of coffee is associated with a lower death rate. Maybe the coffee itself is responsible. But it’s also possible that coffee drinkers engage in other healthy habits that explain the decreased death rate. We simply can’t name the exact cause from an observational study.

Mainly we use these types of studies to provide insight into health issues that need to be studied further. It’s a mistake to treat their results as the last word.. One of the best bad examples of this involved hormone replacement therapy (HRT) in post-menopausal women.

The Nurses' Health Study has followed over 100,000 women since 1976 to help determine risk factors for cardiovascular disease. It provides large amounts of observational data for researchers. Back in 1991, researchers found a decreased risk of death due to cardiovascular disease in women who used HRT. HRT became standard medical care after this research was published. It was hard to find a post-menopausal woman who wasn’t on HRT by the late 90’s.

However, another study published in 1991 contradicted the Nurses’ Health Study. In this study, women who received HRT had a higher risk of heart attacks and death from heart attacks than those who received placebo. These results were the exact opposite of those from the observational Nurse’s Health Study. They were so clear-cut that the study was stopped early. It would have been unethical to continue giving HRT to the women who were receiving it.

Why did these results override what had already been learned in the Nurses’ Health Study? This study was called the Women’s Health Initiative (WHI). It was a double blind, randomized, placebo-controlled trial. This type of trial is considered the gold standard in medical research. It allows us to establish causation. Let’s break it down into its components.


The WHI was double-blind, meaning neither patients nor their physicians knew if they were receiving HRT or placebo. This stops bias from creeping in, but it’s not always possible to blind a study. For example, surgery is difficult to perform without either the surgeon or the patient figuring it out. A study on coffee consumption would also be difficult to blind, as most people know when they are drinking a cup of coffee.

Double-blinding works well with drugs. In the WHI, placebo pills matching the hormone pills were distributed to participants in coded bottles. There wasn’t a good way to figure out who was receiving placebo or hormones without access to the computer database.


The other important component of the study is randomization. By randomizing patients to receive a treatment or placebo, researchers minimize the effects of personal habits and characteristics. In the Nurses’ Health Study, women may have had fewer cardiac events on HRT because women who went on HRT at the time were more naturally inclined to take good care of themselves.

But in the WHI, randomization ensured both groups had a mix of women with good and bad health habits. This aspect really allows us to nail the causation piece.


Finally, we like these studies to be placebo-controlled. Rather than treating all the patients in the study, some receive placebo. This gives researchers a standard for comparing treatment benefits.

It’s easy to understand why the standard of care made a quick about-face when the WHI results were released. Today HRT is used mainly on a case-by-case basis to manage severe hot flashes or to help counter accelerated bone loss.

As you would expect, it is always easier to find observational data than randomized, controlled trial data. So it’s important for the healthcare consumer to dig into the research behind the headlines to avoid making the mistakes made with HRT. In the meantime, it will be a while before we get a randomized trial for coffee consumption. Who would want to be in the placebo group for that one?