Note: None of this is medical advice. This is me rambling about statins and recommendations in general. Don’t go trying to buy statins online for your parents or whatever based on this.
Guidelines are great. Learning and remembering all of the medicine is sort of hard  , and it’s helpful to have heuristics to allow for easier decision-making; a few that come to mind are Wells’ Criteria for PE/DVT risk and CHADS-VASc scores for Afib-related stroke risk. A more prickly attending might respond with something along the lines of “medicine is an art and/not a science” and that you can’t just distill years of clinical training into a risk-calculator on your phone. Your more savvy medical podcaster might say something like “it’s a useful tool to supplement your clinical gestalt” or whatever, which is probably way closer to being the correct answer. I’ve found myself plugging patients into the Atherosclerotic Cardiovascular Disease (ASCVD) Risk Calculator both because it’s handy to stratify patients into neat categories and because if you mention this in your charting, the insurance company will give you a gold star.
From what I can gather, the calculator is from the 2013 American Heart/American Cardiology Assosiation’s guidelines on Primary Prevention. As far as I know, this is more or less the last word on this sort of thing, and I’d roughly estimate that 60% of physicians follow it to a tee. Another thing one might notice when looking at the ASCVD score, is that it tells you it’s for patients between 40 and 75. Even in the most recent 2019 AHA/ACC guidelines, at age 75 the recommendations change from (I paraphrased here) “use the risk calculator and prescribe statins accordingly” to “do a clinical assessment and have a risk discussion”. This isn’t quite tantamount to saying that people older than 75 shouldn’t be on statins, but it implies that clinicians should at least be thinking in that direction. A lower bounds of 40 makes enough sense, since CV risk is pretty low in that category so unless there is a specific reason to think otherwise, it can be assumed that the vast majority of people have low CV-related mortality risk and that pretty much any spending on preventative care in that cohort would probably just be burning money and statins.
The upper bounds though, is where it gets weird. People above 75 are more, not less, predisposed to CVD than the younger folks. So you’d expect the recommendations to be at least as strong (and possibly more so) in the older population. Wait. I mean. They are aren’t they? I suppose it’s possible there’s some sort of bias in the data, where people who make it to 75 and haven’t already developed CVD are just for whatever reason (genetic, environmental, lifestyle etc) just as likely to continue not having CVD after that age cutoff. Or maybe it’s that those who develop CVD in that 40-75 cohort just end up dying before they can color the sample populations after 75.
It turns out that neither of those is true (or true enough) and that these guidelines are (spoilers) probably going to get filled in the near future. So why is it that there aren’t hard recommendations on statin use in people over 75 yet? Vamos a ver.
I think a natural place to begin thinking about this, would be to identify things that give you pause when considering prescribing statins to an elderly person. Here is a short list that I came up with:
- Side effects that either increase in relevance, significance, magnitude and/or prevalence with advancing age.
- Somewhat related but technically different, polypharmacy
- Cost effectiveness
- Do statins even work This is of course, not an exhaustive list but one that captures some salient points and (more importantly) points that I am intellectually capable of addressing.
Statins are pretty safe (I think). To add a little perspective:
- Within the realm of sane dosing regimens, a statin’s ability to exert its effects is dose-dependent only to a point, and notably less than something like metoprolol where (within the bounds of normal prescribing) one can precipitate hypotension pretty easily; it’s very unlikely you’d even be able over-lower your patient’s cholesterol.
- The LD50 (amount needed to kill about half of the people in a sample size) of a statin like Atorvastatin is something like 5000mg/kg which is orders of magnitude higher than typical dosing(10-20mg/day), compared to something like APAP (acetaminophen) where the LD50 is within a single-digit multiplication of its typical dosing(500mg, 2000mg/kg). It’s way harder to kill your patient with statins than it is with lots of other drugs.
Now this isn’t to say that Metoprolol and APAP are super dangerous and that statins have no side effects; clearly, well monitored rate/BP control is much more beneficial than the risks of hypotension alone and Tylenol is OTC. That’s just where statins sit in my brain in terms of safety profile: almost as pretty of a picture as you’re likely to get.
That having been said, the side effect that garners the most attention (rightfully) is myalgia/arthralgia which is/are mediated through my favorite mechanism of action: unclear/multifactorial. ‘Algias are the most important side effect because they’re both the most common side effect reported, occurring in ~9%1 of patients and accounts for ~60% of patients who discontinued their statin. Being in, even mild, pain is annoying and generally enough of a reason to not want to do a treatment anymore; but myalgia/arthralgia can be absolutely debilitating in frail elderly folk and can discourage the modest amounts of physical activity that already limited to. Related to the ‘algias is the risk of rhabdomyolysis, which is quite serious, but thankfully exceedingly rare. The incidence of both the ‘algias and rhabdo is thought to be dose-dependent, given that drug-drug interactions (CYP mediated, because of course it is) play a significant role in reported cases; but this has yet to be demonstrated definitively in the literature.
Next on the list is hepatic dysfunction, which is way less scary than it sounds. Something like 3% of patients develop elevated hepatic transaminases. I was taught to follow up on patients started on statins to check for transaminitis2 but I never quite understood why we did it if we never did anything with the information when we got it, and I was glad to learn that the CDC dropped that recommendation back in 2012. As it turns out, nearly all of these patients remain asymptomatic, with severe sequelae being ever rarer than rhabdomyolysis; the Cochrane review barely mentions liver-related side effects.
The side effect that I personally found the most interesting was the increased risk of DM2, which I didn’t realize had accumulated enough evidence to merit actual concern. The oft-cited JUPITER trial and an older trial (with a way worse acronym: AFCAPS) showed a small increase in the incidence of newly diagnosed Diabetes among patients receiving statin therapy which Cochrane summarized as having a RR of 1.18 (95% CI 1.01 - 1.39). This is scary enough to prompt further investigations, and thankfully there existed enough industrious folks to run a few trials which were at least designed to look at DM2 specifically. This study from the BMJ is the most definitive sounding paper I could find on the matter (which used standardized diagnostic criteria for diabetes and bothered to follow up on continued statin therapy; the above mentioned trials did not) finds a slightly more concerning effect (HR 1.31). By their own admission, the investigators were unable to fully control for factors related to statin prescriptions (many of which overlap with diabetes), and they weren’t able to stratify on the intensity of statin therapy (which other studies had indicated might be related to diabetes risk). The authors said they tried controlled for both of these somehow and still found an association, but I’m never sure what to make of things when investigators say they ‘statisitcally control for’ a factor; so I remain agnostic.
I love buzzwords about as much as I love guidelines. Polypharmacy is probably the biggest of them in Geriatrics for good reason. Eldery patients accumulate quite impressive med lists, which is quite a burden physiologically, mentally, and economically on an already vulnerable population. The push to de-pharmacize the elderly has probably been a net positive in preventing complications, and improving quality of life through decreasing side-effect burden. Since so many older patients are prescribed statins, they’re going to be on the list of drugs to cut when a long med list comes across the desk of a savvy prescriber. I’m not sure how to prove this other than in a survey, but I’d guess that a good many providers take a look at the AHA/ACC guidelines, see that the over-75 recommendations are squishy, and decide that statins are ‘a maybe’ for those patients. For polypharmacy reasons alone, this is pretty rational. As I mentioned before, statins have real side effects which range from worrying to potentially debilitating. A lot of their side effects are mediated through CYP pathways, so there are real concerns about its interactions with a lot of other medications in both directions. In its use as a preventative measure, the NNT hovers in the 90-200 range, so in patients who have about as long to live as the 10 year mortality window where we measure success, the decision seems pretty wishy washy. All of the online tools (because of course there are) I found gave pretty soft suggestions for discontinuing statins when given with a list other common medications (beta blockers, ASA, meds with anticholinergic effects etc) so maybe this is much ado about nothing.
So I was going to really get into this when I first listed this as a reason to be wary of statin use in the eldery. I had grand plans to gather and collate tables of statin prices (they’re pretty much all generic now!), figure out QALY costs, debate the merits of QALY, and even try to do some back-of-the-envelope calculations for the costs of the side effect burden. But, alas, people much smarter than I already did this to a level of detail that would have likely taken me weeks. The good folks over at Circulation (top 4 journal name, fight me) ran an analysis on this sort of recently, and found that statin use is overwhelmingly positive in terms of cost-savings, and QALY gained. The study noted that even blind prescribing of statins to anyone over 45 results in cost savings and that more careful and personalized selection criteria (they used 20% framingham risk, 7.5% AHA/ACC risk) would greatly reduce costs from side effects and pill burden and almost certainly tip the scales even more in favor of statin use. Long story short: statins are cheap and the events they prevent are prevalent and super expensive; the math checks out. The only reason I even bothered to bring this point up was because someone in a NYT article I can no longer find mentioned it in (what appeared to be) a serious rebuttal to putting statins in the water supply.
So it’s been explained to me, through various attendings’ musings, that LDL-C’s correlation with CVD risk is controversial in that levels of LDL-C don’t correlate directly with outcomes. One attending went so far as to say that CV risk had nothing to do with cholesterol at all; but to be fair he also talked about the illuminati a bunch so I don’t think his view was representative of a medically rigorous consensus. This article (or more accurately: this rambling screed) argues that:
- Lowering LDL-C doesn’t reduce mortality
- All the statin/LDL-C trials use RRR and are thus suspect
- Statins haven’t been shown to reduce mortality
- Statins don’t exhibit their benefits through CV benefits
I think the weakest argument, if taken at face value without evaluating the claims themselves, is the 4th. Because, if statins do reduce mortality, I personally don’t really care if it’s through CV benefits or not, other than out of general curiosity. As for the other points, a study published by the BMJ concludes exactly the opposite of this; showing clear associations between LDL-C and mortality, and even using fun statistics which aren’t RRR like Hazard Ratios. On top of that, the authors also conclude that patients over 75 stand to benefit more from LDL-C reduction than younger patients. But the real crux of the issue here (I think) is that smart people are still piecing together how cholesterol and ASCVD risk are related in the first place.
This article from JAMA I think, explains it better than I could ever hope to but I’ll give it a go. VLDL, IDL and LDL are remnant particles of apoB complexes. As more triglycerides and cholesterol esters are cleaved from VLDL particles, they become less dense and are classified instead as IDL and then later as LDL. Once an apoB complex (be if VLDL, IDL, or LDL at its point in the journey) reaches a certain size threshold, it becomes sterically capable of crossing vascular endothelium, and can become trapped there either starting or exacerbating an atherosclerotic plaque. LDL-C is a measure of the cholesterol content of LDL particles which we then use as an estimate of total LDL concentration. So the goal, of any measure aimed at preventing ASCVD progression is actually aimed at reducing the concentration of apoB containing VLDL remnant particles in circulation; it’s just that in a vast majority of patients, LDL particles constitute something like 90% of circulating VLDL remnant particles. So, this is a long winded way of saying that you’re bound to run into problems if you look at LDL-C reduction as an endpoint in your analyses because (although it is a workable approximation) it’s not actually what we’re trying to reduce.
I apologize in advance for this terrible analogy. Imagine you’re in charge of freight trains. The longer your train is, the slower it goes; this is bad. Each cart carries random amounts of corn; they can be full, half full, 1/4 full etc. Unfortunately, your only means of measuring the train is the gross weight of the corn being transported and the only lever you have reduces the amount of corn on the train. The degree to which this actually reduces the length of your train can vary depending on how full each cart is. Carts that had little grain to begin with get emptied and are removed from the train, but fuller carts that still have grain left remain attached to the train. The end result is that in some cases you pull your lever and the train gets significantly shorter, goes faster, and you get that performance bonus to pay for timmy’s operation. In some other cases, you pull your lever but the train barely gets any shorter, doesn’t go much faster and you’re left scratching your head and getting upset at big pharma.
Fortunately, as always, there are a bunch of smart people who have already investigated this, or are currently doing it.
A 2018 AHA study took a look at the PALM registry and zoomed in on patients older than 75, which ended up being something over 1000 individuals. They found that patients in the registry who met criteria for a statin were prescribed one at more or less the same rate regardless of their age, and that these two groups (either side of 75) reported side effects (like myalgia) at about the same rates. Both of these are great news and provide some signal that side effect burden doesn’t seem to scale much with age. The study did note that older patients were more likely to be prescribed less-intensive statin therapy than younger patients, and that this provided an opportunity to prevent more CVD adverse events by targeting “under-treated” patients in that cohort. Thankfully, the authors also took care to note that even among the more intensely treated 75+ patients, side effect rates remained similar to their younger counterparts. Not a huge sample, and it’s a retrospective cohort, but all of the conclusions go in the right direction so I’d chalk this one up as a win.
A 2020 JAMA study at least solves the sample size problem though: a whopping 300K cohort whittled down to 50K who recently started statin therapy with a median age of 81 were followed for a median of 6.8 years. The study showed a 19.5 all-cause deaths prevented per 1000 person-years benefit (78.7 vs 98.2) and a 3.1 CV deaths prevented per 1000 person-years advantage (22.6 vs. 25.7) advantage for statin users over non-statin users. I know! Great numbers for sure, but all-cause mortality benefits were better than CV-related mortality benefits! Go figure. Now, there’s gripes to be had about this study: it was like 90% men and 90% white (VA data). But again, all the data points in the right direction, and decidedly so. It certainly raises questions about where the benefits of statins come from, and we can certainly investigate that particular result further, but it’s hard to argue that the results aren’t encouraging.
There are also a bunch of ongoing trials3 that are looking at statin use in the elderly which will hopefully answer a bunch of the questions we all still have. A group of investigators at Duke are going to try to randomize primary preventative treatment with Atorvastatin to about 20,000 participants and track outcomes, which hopefully will back up what we’ve learned in the retrospective cohort studies I mentioned above. A group of Korean investigators has a more modest 1800 participants and is comparing low and high intensity statin therapy in the elderly for primary prevention of CV outcomes. A more sober group of Australians is investigating a group of 18000 participants to see if the CV and mortality benefits of statins continue past 70, and if those benefits outweigh the risks of side effects. I think the sum of these studies, which directly address the deficiencies in the data we have to work with and we’ll hopefully have a more concrete set of guidelines for statin use for people over 75.
People over 75 should probably be getting statins at rates similar to their younger counterparts. The lack of hard guidelines represents a lack of hard evidence regarding effectiveness and side effect profiles for patients of advanced age as opposed to evidence that statins would be actively harmful. Obviously, you’d have to consider side effect burden, drug-drug interactions, and polypharmacy when ultimately making the decision to either start or continue a statin in someone over 75; but it should be prefaced with the same reassurances that every other cohort gets: statins confer significant mortality benefits, and (barring obvious reasons not to) you should err on the side of prescribing them.
In the literature it ranges from 0.1% to somewhere around 10%. Cochrane’s (the Daria of medical review journals) review of statins cite the rate of myalgia to be around 9.4%, which I will accept for argument’s sake, although this doesn’t seem to match my own experience. ↩
I hate and love this term. It’s such a terrible description (my liver transaminases are…inflamed?), but it does just roll off of the tongue doesn’t it?) ↩
Some of which are RCTs and not retrospective cohorts!, and they all have fun acronyms: STAREE, SCOPE-75, and PREVENTABLE) ↩