GLP-1 drugs have become widely prescribed and used; their original use was to help people, originally those with Type II diabetes, lose weight. People who take them often notice they experience less “food noise,” or internal chatter. As the use of these drugs has become widespread, users have also noticed that they are less interested in smoking, drinking, and taking recreational drugs.
The study found that people who were taking GLP1s were less likely to report nicotine dependence in the following few years. Does the study support a causal claim (i.e., the GLP1s cause people to give up smoking?)
Photo credit: PeopleImages/Shutterstock
At first, the observations were anecdotal, but before long, scientists studied it experimentally. Now one of those scientists, Ziyad Al-Aly, has described their work in The Conversation (an online platform where researchers write about their work for a general public). The article, for example, describes some animal experiments showing that GLP-1s reduce animals’ intake of alcohol, nicotine, and cocaine.
What about in humans? Here’s a description of Dr. Al-Aly’s study (purple text is directly quoted):
To find out whether these drugs have a similar effect on people, we turned to the electronic health records of more than 600,000 patients with Type 2 diabetes at the U.S. Department of Veterans Affairs – one of the largest health care databases in the world.
We designed a study that applied the rigor of randomized controlled trials – the gold standard in medicine – to real-world data. We compared people who started GLP-1 drugs to people who did not, adjusting for differences in health history, demographics and other factors, and followed both groups for three years.
What we found was striking. In the group already struggling with addiction, there were 50% fewer deaths due to substance use among those taking GLP-1 drugs compared with those who were not. We also found 39% fewer overdoses, 26% fewer drug-related hospitalizations and 25% fewer suicide attempts. Over three years, this translated to roughly 12 fewer serious events in total per 1,000 people using GLP-1 drugs – including two fewer deaths.
The drugs also appeared to prevent addiction from developing in the first place. Among people with no prior substance use disorder, those taking GLP-1 drugs had an 18% lower risk of developing alcohol use disorder, a 25% lower risk of opioid use disorder and an approximately 20% lower risk of cocaine and nicotine dependence. Over three years, this translated to roughly six to seven fewer new diagnoses per 1,000 GLP-1 users.
Questions:
a) Re-read the quoted text above and identify the study’s variables (I count eight of them). As you list them, indicate whether they are manipulated or measured.
b) Is this study a correlational study or an experiment? Explain your answer.
c) Recall that to support a causal claim such as “GLP1 drugs can prevent addiction from developing” we need to establish covariance, temporal precedence, and internal validity. This study does show covariance because the result showed that those taking GLP1s (A) had an 18% to %25 lower risk of developing a dependence on alcohol, opioids, or nicotine (B).
That means A <–> B (A covaries with B).
The study also supports temporal precedence, because the method established that A–> B Can you say why?
d) The third criterion for causation is internal validity. In this study, it sounds like the researchers statistically controlled for health history, demographics such as gender and SES. That means we can rule out those particular alternative explanations, or confounds. But statistical controls cannot rule out every possible confound. What are some other variables (“C” variables) that might be associated both with (variable A) deciding to start GLP-1s (or not) and also with (variable B), developing an addiction to alcohol or other drugs? List the variable and explain its connection with A and B.
e) Why do you think the author claims that this study “applied the rigor of a randomized controlled trial”? (An RCT is an experiment)? Do you agree that their study, at least in terms of how it is described in this post, is as rigorous as an RCT?
f) If you were to conduct a study like this as an RCT, how would you do it? What would you manipulate, and what would you measure? Why is the RCT method superior to the correlational approach used here?