Should I take a statin? I am in my 50’s and I have high cholesterol. I’m otherwise healthy, exercise and am not overweight. I’m also a doctor so I should be able to find a few randomized controlled trials, maybe a meta-analysis and have an answer shortly, right? Wrong, what we’re talking about here is the primary prevention of something bad happening such as a myocardial infarction (MI) or death by me taking a statin.
A meta-analysis is usually a good place to start looking for information on a medical topic. It is a process which mathematically combines the results of clinical trials. After many hours on the Internet, it turns out there may be somewhere around 24 meta-analyses of this topic - not all reaching the same conclusions. The range of studies in the meta-analysis was anywhere from 6 to 27 trials so there are as many meta-analysis out there as there are original clinical trials. There is even a systemic review of the meta-analyses.
Taking an initial look at some of the meta-analyses on this topic I find some saying there is a decreased overall mortality and some saying there is not. The question that comes to mind is how could so many people look at this data and reach different conclusions? There are lots of numbers being thrown around in these analysis like relative risk, risk reduction, absolute risk, number needed to treat and even median postponement of death. I would like to know what is already known about this topic, but how can I find that out in an unbiased way? Once I figure out if I should take a statin, it should be possible to apply the results to people in other risk groups.
It’s apparent I need to look at this data myself. It’s time for some Treatment Scores™. My goal is to find the net treatment benefit, if there is one, in taking a statin for primary prevention in someone similar to me. We’ll go to the Treatment Scores™ diagnosis tool and start with a primary diagnosis of prevention of a major cardiac event by statins. I am starting with major cardiac events since this was the primary outcome in most of the studies. We’d also like to know the overall effect on mortality and this will be included in the analysis. We’re looking at primary prevention in a relatively health non-smoker with a high total cholesterol.
Now we’ll skip ahead to the StarBlocks™ as we need to organize the literature. Going to Pubmed, it’s not that easy to find the actual studies. Searching under statins and primary prevention generates 3220 results. Finding the actual studies that are relevant takes some doing. Eventually, I find the large randomized controlled studies that seem most relevant and get them entered into my StarBlocks™. These are the studies that are commonly used in the meta-analysis.
Before we start our analysis, here are a couple of observations. Most of these studies are randomized controlled studies, which is great. However, even randomized controlled trials are not perfect and can have biases and potential errors. Another observation is that some of the studies were stopped early, which can lead to some exaggerated effects. This is similar to how too many meta-analyses can also be misleading. There were a lot of secondary outcomes studied in these papers and sometimes this too can lead to some spurious results. We’ll talk about these issues in future blogs.
First, we are looking at primary prevention. Taking statins after a myocardial infarction (MI) is secondary prevention and is a whole different issue. In these studies, the primary outcome was prevention of major cardiac events. This usually consisted of having an MI, a fatal MI or need for cardiac revascularization procedures. As we will discuss in another blog, the primary outcome is usually the most accurate so we will use this as our main statistic. Secondary outcomes consisted of rates of cerebral vascular accidents (CVA’s) and overall mortality rates. We’ll use 5 years as our follow up period although as mentioned previously several studies were stopped earlier. The follow up durations range from 1.9 to 5.3 years. What kinds of patients were studied? As it turns out, not really patients like me. Most of them had high total cholesterol levels as expected, but some were elderly with angina and had high risks of coronary events. Many of them were hypertensive, diabetic or smokers or some combination of all three. The majority were men.
It appears we have a range of patients in these studies: some being at a “high risk” of having a major cardiac event and some being low risk. It appears not everyone responds to a statin in the same way. Some people have a large decrease in their cholesterol levels and some not so much. However, we don’t know ahead of time into which group a particular patient will fall so all of the patients will be analyzed the same way.
What we are trying to do is to quantify the actual benefit, if there is one, of taking a statin. The relative risk reduction for preventing a major cardiac event varies between 14% and 44% in these studies. However, this statistic can be misleading. Its clinical significance depends on the overall incidence of the event. This is something else we’ll discuss in future blogs.
Let’s look at the absolute risk reduction. This is just the difference in the risk of one group having an outcome minus the risk of the other group having an outcome. For instance, if 5% of the patients in the placebo group have an MI and 3% of patients in the treatment group have an MI, the absolute risk reduction is 2%. If 15% of the patients in the placebo group have an MI and 13% of patients in the treatment group have an MI, the absolute risk reduction is 2%. In my StarBlocks™, the absolute risk reduction for a major coronary event is listed in the Statistic We Have box. These range between 1.2% and 2.6% with an average of 1.91%.
Another way this is commonly looked at is the Number Needed to Treat or NNT. This is the reciprocal of the absolute risk difference. In this case, if we take the reciprocal of the average, it is 1/.0191 which would be 52. In other words, 52 patients would have to be treated to prevent one major adverse cardiac event.
If we look at overall mortality, the average absolute risk difference is 0.3%. This gives an NNT of 300. In other words, 300 people would have to take a statin to prevent one death in these patients. The average number for risk difference for preventing a stroke is the same: 0.3%, again yielding an NNT of 300.
I am going to use the absolute risk reduction as the gross treatment benefit on my 0 to 100 scale. This is something easy to understand. For example, if I have a 10% chance of having a heart attack in the next 10 years and drug A lowers that to 8 percent. The benefit is 2% or 2 on a 0-100 scale. We will have to subtract something from this value to get a final Treatment Score™ because of the potential side effects of drug A.
Thus, before getting our final Treatment Score™ for statins, we’ll also consider side effects. It’s hard to get 100% accurate numbers for side effects. For instance, some of the large trials found that there was an increased incidence of some kinds of cancers and others did not. When you are looking at a lot of subgroups it’s possible to get some spurious results just by chance. There are some useful numbers in the PDR regarding side effects and also some large cohort studies. Putting it all together (and a lot of this evidence is weak), here’s what I came up with. Statins can increase Liver Function Tests in some patients but it will resolve if the drugs are stopped. Some patients get myopathies which will resolve if the drugs are stopped. There is a small chance of rhabdomyolisis which is a breakdown of muscle tissues. There probably is some increase in the chance of developing diabetes. The Treatment Scores™ given in the Treatment Score™ calculator below are my best estimates. There are some reports of memory loss when statins are started, but there may be some protection against dementia. Again, the evidence is weak so I gave both dementia prevention and memory loss Treatment Scores™ of 0. There is also the monetary cost of the drugs to consider, but this in not included in this Treatment Score™ analysis.
I am in a relatively low risk group, so to get my Treatment Score™, I need to know if higher risk patients get more benefit from taking a statin than lower risk patients. Looking at the data, this does seem like a reasonable assumption. The studies with the lower risk patients in general have a lower absolute risk reduction than the patients in the higher risk groups.
The original goal of this example was to find the Treatment Score™ of someone like me. This will be the net treatment benefit on a 0 to 100-point scale. Let’s analyze major cardiac events first. Since I am in the lower end of the risk spectrum, I’ll start with an absolute risk reduction for a major cardiac event of 1.2%. This was the absolute risk reduction in what appears to be the lowest risk group studied. This study did have a significant number of patients who were smokers or diabetic so my absolute risk reduction is probably less. We’ll be generous here and decrease the 1.2% to 1.1%. We need to subtract from this value because there are side effects which will decrease the net treatment benefit. Again, being generous we will subtract another 0.1% to come up with a final Treatment Score™ of 1.
Therefore, in my analysis the maximum Treatment Score™ or net benefit of me taking a statin for preventing a major adverse cardiac event is 1 on a 0 to 100-point scale and could be even lower.
If we look at preventing overall mortality, the overall risk reduction was .3%. In someone who is lower risk like me the Treatment Score™ effectively becomes 0 after taking into account side effects. The same would be true for prevention of a CVA. Treatment Scores™ makes it apparent that taking a statin will not prevent me from having an early death.
It’s not unexpected that the Treatment Score™ for low risk patients taking a statin is low. I won’t be taking a statin, but for someone with more risk factors, the Treatment Score™ would be higher and they may reach a different conclusion. It is very likely that many of the therapies we use now will have low Treatment Scores™. That doesn’t mean they shouldn’t be used. But at least with Treatment Scores™, I now know what I know. I can see the major studies and see what the benefits and potential harms are. I won’t have to rely on a panel of experts, who may have their own biases, to make a medical decision. There is no reason why doctors and patients should not be able to understand one number used for all treatments. It will then be possible to personalize medicine. For instance, the 60-year-old diabetic smoker may decide taking a statin is in their best interest even though the Treatment Score™ may be somewhere between 2 to 2.5 for preventing a major cardiac event.
Eventually, as Treatments Scores™ gets to a beta stage, my Treatment Score™ will be combined with the Treatment Scores™ of other health professionals. This should allow for even more accurate analysis using crowd wisdom. Also, we will be able to build on the work already done by others and not have to spend hours just to get the literature available to us in a usable form.
As a final note, I wanted to mention meta-analyses. They do have a place in medicine, but unfortunately, they are usually taken as the final word on a topic. As this example shows, 27 meta-analyses of one topic reaching different conclusions makes it apparent that meta-analyses have a downside. Treatment Scores™ are not meant to replace meta-analyses. They are, however, meant to get the medical literature in a simple, unbiased, usable form for everyone. It truly is time for Treatment Scores™.
Follow this blog by entering your email address in the box at the top right. Then you will automatically receive all new posts. (See the disclaimers below.)
Follow Treatment Scores:
Twitter:
https://Twitter.com/TreatmentScores
Facebook:
https://Facebook.com/TreatmentScores
AngelList:
https://angel.co/treatment-scores
Blog:
http://TreatmentScoresBlog.com
Website:
http://TreatmentScores.com
DISCLAIMERS:
You must consult your own licensed physician, or other licensed medical professional, for diagnosis, treatment, and for the interpretation of all medical statistics including Treatment Scores. Treatment Scores are for educational purposes only. Treatment Scores may be incomplete, inaccurate, harmful, or even cause death if used for treatment instead of consulting a licensed medical professional. No medical advice is being given. We DO NOT CLAIM to cure, treat, or prevent any illness or condition. Nor do our services provide medical advice or constitute a physician patient relationship. Contact a physician or other medical professional if you suspect that you are ill. Call emergency services (call 911 if available) or go to the nearest emergency room if an emergency is suspected. We are not responsible for any delays in care from using our website, our services, or for any other reason. We are not responsible for any consequential damages of any nature whatsoever. We make no warranties of any kind in connection with our writings or the use of TreatmentScoresBlog.com or TreatmentScores.com. Treatment Scores are about what happened to patients studied in the past; they do not predict the future.
Copyright
Copyright © 2016 Treatment Scores, Inc.