Hello. This module brings together the course’s work on cardiovascular benefit and statin harms. The aim here is not to produce a single “benefit minus harm” score. It is to make a comparison you can inspect: for people at different starting risks, how many cardiovascular events might be prevented, what harms might occur, and whether the numbers describe comparable groups and periods.
By the end of this lesson, you should be able to build that comparison for lower- and higher-risk examples while keeping heart attacks, strokes, muscle symptoms, diabetes, and serious muscle injury distinct.
Start with the same denominator—and a defined outcome
A statement such as “statins reduce cardiovascular events by 25%” gives a relative change. To estimate an absolute benefit, you also need the chance of an event without treatment. You must specify what counts as an event and over how many years.
Watch this short illustration before working through the figures.
Before You Take A Statin, Watch This...
In “Before You Take A Statin, Watch This...,” Doctor Alex uses two starting risks to explain why one relative reduction can produce different absolute benefits.
Watch the risk examples. Focus on the comparison of four events versus three per 100, then 20 versus 15 per 100. Treat these as arithmetic illustrations, not predictions for a particular person.
The calculation behind both examples is:
The untreated risk and relative reduction must refer to the same outcome and period. A 10-year estimate for “heart disease or stroke,” for example, should not be silently relabelled a 5-year estimate for “death.”
Two cardiovascular-risk examples over ten years
The article Communicating statin evidence to support shared decision-making illustrates absolute benefits using an assumed 25% relative reduction in combined fatal and nonfatal cardiovascular events. Its Table 2 supplies untreated risks and modelled treated risks across several starting-risk levels. It is a useful worked model, not a forecast that everyone taking any statin dose will receive exactly that reduction.
Communicating statin evidence to support shared decision-making
Read the article’s explanation of how starting risk changes absolute benefit, then use Table 2 for the examples below. Its harms section supplies a ten-year estimate for serious myopathy.
In the discussion immediately before Table 2, read the benefit explanation. In Table 2, “Chances of cardiovascular event with or without statin over 5-year and 10-year spans,” inspect the 7.5% and 20% 10-year risk columns; the note beneath the table identifies the assumed relative reduction. In “Harms,” read the opening paragraph from its discussion of muscle effects through the ten-year estimate. Notice that this serious-injury estimate is not a count of all muscle aches.
Imagine two groups of 1,000 people, each followed for ten years. Neither group has one uniform fate: these are expected frequencies across comparable people, not a way to identify who personally benefits.
| Illustrative starting risk | Cardiovascular events without a statin | Events with a statin, assuming a 25% relative reduction | Cardiovascular events prevented |
|---|---|---|---|
| Lower risk: 7.5% over ten years | 75 per 1,000 | About 56 per 1,000 | About 19 per 1,000 |
| Higher risk: 20% over ten years | 200 per 1,000 | 150 per 1,000 | 50 per 1,000 |
For the lower-risk group, the unrounded calculation is events prevented per 1,000. For the higher-risk group, it is . The relative assumption stays fixed; the absolute benefit grows with starting risk. People taking a statin can still have an event, while many who do not take one would not have had an event anyway.
The article also cites an estimate of approximately one additional case of serious myopathy per 1,000 statin users over ten years. That supplies a harm frequency on the same ten-year scale as this benefit model. If used alongside either row, however, it needs a qualification: it is an overall estimate from different evidence, not a prediction tailored to either example group or necessarily to the same statin regimen. Serious myopathy and a combined cardiovascular event are also different outcomes. “About 19 or 50 events prevented versus about one serious myopathy case” describes frequencies; it does not establish an exchange rate between their consequences.
Nor are “cardiovascular events prevented” automatically “lives saved.” A combined outcome can include nonfatal events. To compare deaths, you would need a separate estimate of deaths with and without treatment over a stated period.
Add harms without erasing their timing or severity
The NICE patient decision aid helps separate symptoms experienced while taking a statin from symptoms caused by it. It also gives a diabetes estimate from studies with a shorter follow-up than our ten-year benefit table.
CG181 Patient decision aid on should I take a statin?
Read the NICE patient decision aid’s side-effect explanations for muscle pain, severe muscle damage, and newly diagnosed type 2 diabetes. Concentrate on the different outcomes and study periods, rather than transferring every number into the ten-year table.
In “What are the possible side effects of statins?” (printed pp. 5–6), begin at “Muscle pain” and continue through “Diabetes,” stopping after its “about 1 person” estimate, just before “Other problems.” Read these comparisons. Notice that the aid says statin-caused muscle pain tends to occur in the first year, whereas the diabetes figures come largely from studies lasting about four or five years.
The decision aid’s muscle-pain illustration says that, among 100 statin users, about 26 would have experienced muscle pain anyway and about two experience it because of the statin. The 26 are not “imagining” pain; pain occurring during treatment does not, by itself, establish its cause. The aid separately discusses severe muscle damage. Neither ordinary muscle pain nor serious injury should be substituted for the other in a comparison.
For type 2 diabetes, the aid estimates roughly one additional diagnosis per 100 statin users in studies mostly lasting four or five years, alongside about four diagnoses per 100 that would have occurred anyway. That is about 10 additional diagnoses per 1,000 over the studies’ approximate follow-up. It is a group average, not a fixed risk for every person: starting likelihood of diabetes matters.
We therefore cannot put “10 additional diabetes diagnoses” into the ten-year table and call it a ten-year harm. A closer, though still imperfect, comparison uses the article’s five-year benefit model. It assumes that the five-year cardiovascular risks are half the ten-year risks:
| Illustrative five-year cardiovascular risk | Events without a statin | Events with a statin under the 25% model | Events prevented |
|---|---|---|---|
| Lower risk: 3.75% | 37.5 per 1,000 | About 28.1 per 1,000 | About 9.4 per 1,000 |
| Higher risk: 10% | 100 per 1,000 | 75 per 1,000 | 25 per 1,000 |
Expected frequencies can contain fractions because they describe averages across groups. Set beside this table, NICE’s about 10 additional diabetes diagnoses per 1,000 over roughly four or five years offers a near-matched, not precisely matched, time window. The estimates also come from different populations and treatment assumptions. In particular, the table’s five-year starting risks are modelling shortcuts, and the average diabetes increment should not be assigned identically to both example groups as a personal prediction.
The lower-risk row illustrates why a comparison can remain a genuine decision even when two frequencies look similar: a cardiovascular event and a diabetes diagnosis are not interchangeable. In the higher-risk row, the modelled number of events prevented is larger, but the same cautions about outcome severity, individual diabetes risk, and mismatched evidence still apply.
What a usable comparison says—and does not say
A sound benefit–harm summary keeps separate lines for cardiovascular events prevented, muscle pain attributable to treatment, newly diagnosed diabetes, and rare serious muscle injury. Each line names its denominator, follow-up period, outcome, and the population or treatment assumptions behind it. Where these do not align, the mismatch stays visible rather than being concealed by arithmetic.
There is no justified step in which you simply subtract diabetes diagnoses or muscle symptoms from cardiovascular events prevented to obtain “net people helped.” Outcomes differ in severity and duration, and the same person may experience more than one outcome. Estimates also have uncertainty: a modelled 25% reduction is not a guaranteed personal effect, while rare-harm estimates may be less precise than their tidy appearance suggests.
Key takeaway: absolute benefit depends strongly on starting cardiovascular risk. In the ten-year model, it is about 19 versus 50 cardiovascular events prevented per 1,000 at the two starting risks, alongside a separately estimated rare serious-muscle harm on a ten-year scale. More common symptoms and diabetes belong in the discussion too, but their definitions and observation periods must be stated honestly. These examples support a clinician conversation, not a medication change made on your own.
The next lesson will apply this disciplined comparison to a conflicting statin headline: what was actually studied, how large the absolute effects were, how uncertain they are, and whether the findings fit the people described.