Saturday, January 28, 2012

Numbers and research

Numbers and statistics are the backbone of medical research analysis. I am not a statistician by any stretch. I took a statistics class in college to meet a distribution requirement. After college I took a summer class at the little college down the street, Harvard, where it didn't matter if we had the numbers right (we took a vote - if everyone thought that 47.3% of 6,329,145 was around 3,000,000, we used 3,000,000) because it was the principle of the process. We needed to focus on learning the importance of how the numbers are determined and how to figure out their significance.

Another big part of statistical analysis is ensuring the numbers that are being analyzed are what you think they are - that if you think you have a pool of people who were born west of the Mississippi between 1950-1960 and now live in Florida, that you don't have anyone from Connecticut in it as that would skew your data. And that the same information was collected from all of them so you are not comparing apples and oranges. You want to compare red apples and red apples, not red apples and slightly yellow apples.


Here's an example of data being reanalyzed and people saying well maybe not but maybe yes. In 1995, the million woman study was started in Great Britain to compare diet and lifestyle and health factors such as cancer. One significant result was the link was seen between Hormone Replacement Therapy and breast cancer. Then everyone stopped HRT and breast cancer rates went down so the proof is there. But now there are claims of shortcomings and lack of causality. The new claims are: 'HRT may or may not increase the risk of breast cancer, but the MWS did not establish that it does...'

But still '"This report would not change how I counsel women, as multiple studies, including a randomized trial, have shown an increased risk of breast cancer from combination hormone therapy," Kathy Helzlsouer, MD, director of the prevention and research center at Mercy Medical Center in Baltimore...'

So where does this leave us? Somewhat doubting this one study but based on the findings of other studies and the drop in breast cancer diagnoses, we can still state it was correct.

No comments: