Statins and cholesterol: what the evidence shows

Statins are among the most studied drugs in medicine, and the arguments about them rarely turn on a shortage of data. The trials are large, long, and mostly point the same direction: lowering LDL cholesterol with a statin reduces heart attacks, strokes, and procedures to open blocked arteries. The harder questions are about who benefits enough to make the tradeoff worthwhile, how big that benefit is in plain numbers, and what to make of the side effects patients report. This piece walks through what the major trials and meta-analyses found, and where the evidence is strong, weak, or mixed.

The core finding: lower LDL, fewer events

The single most cited result comes from the Cholesterol Treatment Trialists’ Collaboration, a group that pools patient-level data from the big statin trials. In its 2010 analysis of about 170,000 people across 26 randomized trials, each 1 mmol/L reduction in LDL cholesterol (roughly 39 mg/dL) was associated with about a 21 percent lower rate of major vascular events (Cholesterol Treatment Trialists’ Collaboration, Lancet, 2010). That covered heart attacks, coronary deaths, strokes, and revascularization procedures.

Two details matter. The benefit was proportional to how far LDL fell, so larger reductions bought larger relative reductions in risk. And the effect accrued over time, growing in the later years of treatment rather than appearing all at once. A follow-up analysis in 2012 extended the picture to people at low baseline risk and found the proportional benefit was still present in that group (Cholesterol Treatment Trialists’ Collaboration, Lancet, 2012), which became a central point in later debates about who to treat.

A useful way to read all of this: the trials tell you the relative reduction in risk is fairly steady across different kinds of patients. They do not tell you the absolute benefit, because that depends entirely on how much risk a person started with.

Secondary prevention: the strongest case

Secondary prevention means treating people who already have cardiovascular disease, such as a prior heart attack or known coronary artery disease. This is where the evidence is most convincing, because these patients face a high risk of another event, so a given percentage reduction translates into a large number of events prevented.

The landmark trial is the Scandinavian Simvastatin Survival Study, known as 4S. It enrolled 4,444 patients with coronary heart disease and elevated cholesterol and randomized them to simvastatin or placebo. Over a median of about 5.4 years, all-cause mortality fell from roughly 12 percent to about 8 percent, a relative reduction near 30 percent (Scandinavian Simvastatin Survival Study Group, Lancet, 1994). In absolute terms, that is close to one death avoided for every 30 patients treated over that period. For a mortality endpoint, that is a large effect, and 4S is often described as the trial that settled the question for high-risk patients.

Later secondary-prevention and high-risk trials reinforced the pattern, and the CTT meta-analyses showed the benefit extended to strokes and revascularization as well, not just coronary events.

Primary prevention: real benefit, smaller in absolute terms

Primary prevention means treating people who have not yet had a cardiovascular event. The relative risk reduction looks similar to secondary prevention, but because these people start at lower risk, the absolute benefit is smaller and the number needed to treat is larger.

Two trials anchor this discussion. The West of Scotland Coronary Prevention Study (WOSCOPS) gave pravastatin to men with high cholesterol and no history of heart attack, and reduced coronary events by about 31 percent (Shepherd et al., NEJM, 1995). Years later, JUPITER tested rosuvastatin in men and women who had normal LDL but elevated high-sensitivity C-reactive protein, a marker of inflammation. It reported a large relative reduction in a combined cardiovascular endpoint, with a hazard ratio near 0.56, and the trial was stopped early for benefit (Ridker et al., NEJM, 2008).

JUPITER also illustrates why absolute numbers matter. Because the participants were at moderate risk and the trial was short, the number of people you had to treat to prevent one event was substantial, and critics and supporters read the same data differently depending on whether they emphasized the relative or the absolute figure. Both readings are defensible. The relative reduction was real; the absolute event rate in a lower-risk group was low to begin with.

Older adults

Whether the benefit holds in older people has its own evidence base. PROSPER enrolled adults roughly 70 to 82 years old, a mix of people with existing vascular disease and people at elevated risk, and gave them pravastatin or placebo. It reduced the primary combined endpoint of coronary death, nonfatal heart attack, and stroke by about 15 percent, a hazard ratio near 0.85 (Shepherd et al., Lancet, 2002). The coronary benefit was clear; the effect on stroke in that trial was less pronounced.

For the oldest patients starting a statin for primary prevention, the evidence thins out. Trials enrolled relatively few people over 75 with no prior cardiovascular disease, which is one reason guideline bodies stop short of a firm recommendation in that group. This is a place where the data are genuinely limited rather than contradictory.

Side effects and the nocebo question

Muscle aches are the most common reason people stop statins, and untangling drug effect from expectation has been difficult. Observational reports and everyday clinical experience suggest muscle symptoms are common. Blinded trials tell a more complicated story, because many people on placebo report the same symptoms.

The SAMSON trial addressed this directly with an n-of-1 design. Patients who had stopped statins because of side effects took, in random order and without knowing which was which, a statin, a placebo, and no pill across a series of months, rating their symptoms daily. The symptom burden during statin months was almost the same as during placebo months. The researchers calculated that about 90 percent of the symptom intensity reported on the statin was also present on placebo (Wood et al., NEJM, 2020). This is described as a nocebo effect, where the expectation of harm produces real symptoms. SAMSON was small, and it does not prove no one ever has a true statin-related muscle problem. Serious muscle injury from statins is real but rare. What SAMSON suggests is that a large share of the everyday aches attributed to statins are not caused by the drug itself.

The diabetes signal is a separate matter and is better established. Pooled trial data show statins modestly raise the risk of new-onset type 2 diabetes, by roughly 9 percent in one large meta-analysis (Sattar et al., Lancet, 2010). In JUPITER and other trials this worked out to a small number of extra diagnoses relative to the cardiovascular events prevented. The broad reviews of statin safety concluded the cardiovascular benefit outweighs this risk for most people who meet treatment criteria (Collins et al., Lancet, 2016), though that is a population-level judgment, not a statement about any one person.

What the guidelines say now

In 2022 the US Preventive Services Task Force updated its primary-prevention recommendation. For adults aged 40 to 75 who have at least one cardiovascular risk factor (such as high LDL, high blood pressure, diabetes, or smoking) and an estimated 10-year risk of a cardiovascular event of 10 percent or higher, the Task Force recommended a statin, a grade B recommendation (US Preventive Services Task Force, JAMA, 2022). For people in that age range with a 10-year risk between 7.5 and 10 percent, it said clinicians may offer a statin selectively, a weaker grade C recommendation reflecting smaller absolute benefit.

For adults 76 and older starting a statin for primary prevention, the Task Force concluded the evidence was insufficient to recommend for or against, again pointing to the shortage of trial data in that group. The guideline is built around estimated risk precisely because the trials show the same lesson: the relative benefit is fairly constant, so the absolute benefit rises and falls with a person’s underlying risk.

This article reports what the trials and guidelines found. It is not medical advice, and it does not tell you whether to start, continue, or stop a statin, or what your cholesterol number should be. Those decisions depend on your own risk, your history, and tradeoffs that only you and your clinician can weigh together. Bring these findings to that conversation rather than treating them as a prescription.

Common questions

What is the difference between primary and secondary prevention?

Secondary prevention treats people who already have cardiovascular disease, such as a past heart attack or stroke. Primary prevention treats people who have not yet had an event, to lower the odds of a first one. The relative benefit of a statin is similar in both, but the absolute benefit is larger in secondary prevention because those patients start at much higher risk.

Why do relative and absolute risk numbers differ so much?

Relative risk reduction describes the percentage drop in a group’s event rate, and for statins it runs around a fifth per 1 mmol/L of LDL lowering. Absolute benefit depends on the starting risk. A 21 percent cut on a high baseline risk prevents many events; the same 21 percent cut on a low baseline prevents few. That is why the number needed to treat is small in high-risk patients and large in low-risk ones.

Are the muscle aches people report from statins real?

Some are, but blinded trials complicate the picture. In the SAMSON n-of-1 trial, patients reported nearly as many symptoms on placebo as on the statin, with about 90 percent of the symptom burden present on placebo (Wood et al., NEJM, 2020). Serious muscle injury from statins is real but uncommon. The evidence suggests much of the routine aching attributed to statins is not caused by the drug.

Do statins cause diabetes?

Trials show a modest increase in new-onset type 2 diabetes, roughly 9 percent in a large meta-analysis (Sattar et al., Lancet, 2010). In absolute terms this was a small number of extra diagnoses compared with the cardiovascular events prevented in the trial populations, and safety reviews judged the net effect favorable for people who meet treatment criteria. That is a population finding, not a personal guarantee.