Skip to main content

Welcome to a Grasp lesson

Create a free account to keep studying and create your own study path.

Lesson 7

Interpreting Diabetes Risk from Statin Intensity and Baseline Glucose

Recommended resources

Hello. The previous lesson showed why a blinded comparison matters when attributing muscle symptoms to a statin. The same principle helps with a different question: do statins increase newly diagnosed diabetes, and by how much?

This lesson uses randomized, blinded trial evidence to separate a relative increase from the additional diagnoses per 1,000 people. It then examines why baseline blood glucose and the frequency of diabetes testing change the absolute numbers you see.


Start with the statin–placebo comparison

The Cholesterol Treatment Trialists’ Collaboration combined individual participant data from large, double-blind statin trials. Its analysis counted new diabetes diagnoses among participants without diabetes at baseline, using recorded diagnoses, new glucose-lowering treatment, or qualifying blood-test results. Random assignment makes the statin–placebo difference more informative about causation than a report that someone developed diabetes after starting a statin.

Read the opening findings before comparing the numbers below.

Effects of statin therapy on diagnoses of new-onset diabetes and worsening glycaemia in large-scale randomised blinded statin trials: an individual participant data meta-analysis

Read the Cholesterol Treatment Trialists’ Collaboration analysis for its direct statin–placebo estimates and its warning about how testing affects absolute diagnosis rates.

In “Summary,” under “Findings,” read the trial findings. Start with the description of the participating trials and stop after the explanation of why the absolute excess differed between trial groups. Track the placebo annual rate, the statin annual rate, and the rate ratio for each intensity.

Statin versus placebo Placebo diagnoses Statin diagnoses Relative effect Estimated absolute excess
Low- or moderate-intensity About 1.2% per year About 1.3% per year Rate ratio 1.10 0.12 percentage points per year, or about 1.2 additional diagnoses per 1,000 people per year
High-intensity About 3.5% per year About 4.8% per year Rate ratio 1.36 1.27 percentage points per year, or about 12.7 additional diagnoses per 1,000 people per year

A rate ratio of means a 10% proportional increase over the comparison rate—not that 10% of treated people develop diabetes. Similarly, means a 36% proportional increase. The estimated differences provide the more tangible scale: the analysis found about 1.2 and 12.7 additional diagnoses per 1,000 people during a year in the respective trial groups. The displayed statin and placebo rates are rounded, so subtracting them does not reproduce the more precise excess estimates exactly.

The 95% confidence intervals for the rate ratios were – for low- or moderate-intensity treatment and – for high-intensity treatment. Both lie above , supporting a real increase in new diagnoses rather than an inconclusive difference. These are annual rates, however, not a promise that any particular person will receive a diagnosis, nor a five-year risk obtained simply by multiplying a row by five.



Why starting blood glucose changes the absolute risk

Glycaemia means the level of glucose in the blood; HbA1c is a measure reflecting blood glucose over a longer period. A diabetes diagnosis often depends on whether a measured value crosses a defined threshold. In this analysis, statin allocation produced a small average upward shift in glucose and HbA1c. For participants without diabetes at baseline, the mean glucose difference from placebo was mmol/L with either intensity category; the mean HbA1c difference was percentage points with low- or moderate-intensity statins and percentage points with high-intensity statins.

A small average shift can lead to additional diagnoses when many people begin close to a diagnostic threshold. It does not mean that everybody’s result rises by exactly the average amount, or that crossing the threshold is merely a meaningless change of label. It means the starting distribution of blood-test values is crucial to interpreting the number of additional diagnoses.

Read how the investigators linked that starting distribution to the excess cases.

Effects of statin therapy on diagnoses of new-onset diabetes and worsening glycaemia in large-scale randomised blinded statin trials: an individual participant data meta-analysis

Return to the trial analysis to connect its small average blood-test changes with the distribution of additional diabetes diagnoses.

In “Results,” read the glycaemia results. The passage begins with glucose and HbA1c measurements and ends with the comparison between baseline-glycaemia quartiles and a broader diabetes-risk score. Focus on excess cases, rather than treating the highest quartile as a prediction that most people in it will develop diabetes.

The figure makes the absolute-risk pattern visible. Each panel divides participants into four groups, or quartiles, using either baseline glycaemia or a predicted five-year diabetes-risk score. Blue shows the annual diabetes diagnosis rate in the placebo group. Red shows the estimated additional annual rate associated with statin treatment.

Four panels compare annual new-diabetes diagnosis rates across quartiles of baseline glycaemia (left) or predicted diabetes risk (right). Blue bars show placebo-group rates; red segments show estimated statin-associated excess rates calculated using the overall rate ratio for that intensity, rather than directly observed statin–placebo differences within each quartile.

Read the red labels as annual percentage-point excesses, not as the total height of each bar. For low- or moderate-intensity treatment, the baseline-glycaemia panels show estimated excesses ranging from 0.06% per year in the lowest quartile to 0.48% per year in the highest. Those correspond to approximately 0.6 versus 4.8 additional diagnoses per 1,000 people per year. For high-intensity treatment, the corresponding estimates are 0.23% and 1.78% per year, or about 2.3 versus 17.8 per 1,000 per year.

Why does the red segment grow so much? The blue placebo rate is already much higher in the top quartile. Applying an approximately similar proportional increase to a higher starting rate produces a larger absolute excess. The investigators estimated that about 62% of excess new-diabetes cases occurred among participants in the highest quarter of baseline glycaemia. That does not mean 62% of people in that quarter developed diabetes. Nor does it mean a person with a low starting value has zero risk.



Intensity matters—but so does how often trials looked

At first glance, the overall table suggests that high-intensity statins produce roughly ten times the annual absolute excess seen in the low- or moderate-intensity trials. That would be too simple an interpretation. The high-intensity and lower-intensity figures come from different sets of trials. In the high-intensity trials, follow-up HbA1c testing was much more common. More testing finds more people whose results meet a biochemical definition of diabetes—including in the placebo group. This largely explains why its placebo diagnosis rate was 3.5% per year, rather than 1.2% per year.

Read the investigators’ account of that difference, followed by their more direct comparison of treatment intensities.

Effects of statin therapy on diagnoses of new-onset diabetes and worsening glycaemia in large-scale randomised blinded statin trials: an individual participant data meta-analysis

Use the detailed Results to distinguish a genuine intensity-related effect from differences in how trials detected diabetes.

In “Results,” begin with the paragraph opening “The placebo event rate for new-onset diabetes was substantially higher” and read the testing comparison through its explanation of follow-up HbA1c and glucose measurements. Then read the intensity comparison in the next paragraph, which begins “Further information on the risks.” Stop after the comparison of the indirect and direct high-intensity estimates. Notice that the second passage compares more with less intensive statin treatment, rather than comparing two unrelated placebo groups.

In trials that randomized participants to more versus less intensive statin therapy, the more intensive treatment had a rate ratio of 1.10 for new diabetes and an estimated excess of 0.22 percentage points per year—about 2.2 additional diagnoses per 1,000 people per year compared with less intensive treatment. That comparison supports an intensity-related increase. It does not imply that the large gap between the two placebo-controlled rows in the first table is caused solely by dose: the populations and, especially, the opportunities to detect diabetes differed.

The practical reading rule is to keep three pieces together: treatment comparison, starting diabetes risk, and testing method. A single “statins raise diabetes risk by 36%” headline omits the starting rate needed to understand the absolute effect. A claim that “high-intensity statins cause 13 extra cases per 1,000 every year for everyone” wrongly transfers one trial group’s annual excess to people with different baseline values and testing schedules.

There is a further distinction between a new diagnosis and later complications of diabetes. This analysis establishes an increase in diagnoses and small average glycaemic changes. It could not directly determine the long-term effect on serious diabetes-related complications, which can take many years to appear. Neither dismissing the diagnosis nor assuming each additional diagnosis will lead to such a complication follows from these results.



Takeaway

Randomized, blinded trials support a real increase in newly diagnosed diabetes with statin treatment, with a greater proportional increase for high-intensity treatment. The additional diagnoses are concentrated among people starting closer to diabetes-range blood-test values; for a similar relative effect, a higher baseline rate produces a larger absolute excess. Annual rates from different trials also depend on how often diabetes was sought, so the striking difference between their absolute estimates is not a clean dose comparison.

Diabetes risk belongs alongside expected cardiovascular benefit in a discussion using comparable populations and time periods—not as a reason to change treatment independently. The next module turns to rare serious harms and the limits of what safety studies can rule out.