References

Byrne P, Demasi M, Jones M, Smith SM, O'Brien KK, DuBroff R. Evaluating the association between low-density lipoprotein cholesterol reduction and relative and absolute effects of statin treatment: a systematic review and meta-analysis. JAMA Intern Med. 2022; 182:(5)474-481 https://doi.org/10.1001/jamainternmed.2022.0134

Carling CL, Kristoffersen DT, Montori VM The effect of alternative summary statistics for communicating risk reduction on decisions about taking statins: a randomized trial. PLoS Med. 2009; 6:(8) https://doi.org/10.1371/journal.pmed.1000134

Diamond DM, Leaverton PE. Historical review of the use of relative risk statistics in the portrayal of the purported hazards of high LDL cholesterol and the benefits of lipid-lowering therapy. Cureus. 2023; 15:(5) https://doi.org/10.7759/cureus.38391

Gigerenzer G, Wegwarth O, Feufel M. Misleading communication of risk. BMJ. 2010; 341 https://doi.org/10.1136/bmj.c4830

Spiegelhalter D. The art of statistics: learning from data.London: Penguin Random House; 2019

Statins: the risks and statistics

09 November 2023
Volume 32 · Issue 20

The application of medical statistics has the potential to achieve life-saving outcomes. Thus, Spiegelhalter (2019: 294) noted that if a statistic called a sequential probability ratio test had been applied routinely to the accumulating recorded mortality data of the patients of murderous GP Harold Shipman, 175 lives could have been saved.

Less dramatic, but with possible far-reaching consequences for patients' health, is how statistics – especially in relation to the use of relative and absolute risk – can invite certain inferences, for example, in relation to the effectiveness of statins. Let us assume that drug X prevents heart attacks in 1% of a population given drug X, but that 2% of a population given a placebo have heart attacks. Then, in terms of absolute risk (AR) reduction, the benefit conferred on the treated population by drug X is a mere one percentage point better that not being treated. However, the relative risk (RR) reduction conferred by drug X is 50% since one is 50% of two. And as Gigerenzer et al (2010) pointed out, RRs say nothing about the baseline risk ie, does ‘twofold’ mean from one to two or from 50 to 100 in, say, 7000? It is an important question to be answered when benefits are being weighed against harms.

In a randomised trial of almost 3000 participants who were invited to respond to six summary statistics for communicating coronary heart disease (CHD) risk reduction with statins, Carling et al (2009) concluded that when presented with the benefits of taking statins as an RR reduction, participants were more likely to accept treatment compared to when the participants were given AR reduction statistics. And in a meta-analysis of 21 randomised clinical trials that considered the efficacy of statins in lowering total mortality and cardiovascular outcomes, Byrne et al (2022) found reductions in:

‘If clinicians are going to recommend statins to their patients, then their duty of care must extend to ensuring that the nature of absolute risk and relative risk is explained’

‘AR of 0.8% for all-cause mortality, 1.3% for myocardial infarction, and 0.4% for stroke in those randomized to treatment with statins compared with control, with RR reductions of 9%, 29%, and 14%, respectively.’

Given these striking differences between expressions of AR and RR reductions to describe the same outcome of taking statins, it is perhaps unsurprising that Diamond and Leaverton (2023) have cited researchers' documented misgivings about the quality of published medical research and suggested that such scepticism may be at least partly attributable to ‘decades of misleading presentations of research findings by clinical trial directors’.

Diamond and Leaverton (2023) offered an historical perspective, over the past 40 years, on how RR reduction data have been used – or, rather, misused – to report the findings from randomised controlled trials (RCTs) on CHD event monitoring and prevention. The authors further suggested that undue emphasis on RR reduction at the expense of AR disclosure when RCT outcomes are reported:

‘has led healthcare providers and the public to overestimate concerns about high cholesterol and to be misled as to the magnitude of the benefits of cholesterol-lowering therapy.’

One of the five landmark RCTs re-evaluated by Diamond and Leaverton (2023) was the JUPITER trial of rosuvastatin, which included almost 18 000 participants, and reported, inter alia, that there was a 54% reduction in fatal myocardial infarctions (MIs). However, more than 99% of rosuvastatin and placebo participants did not have a fatal MI, so how could rosuvastatin have mediated a 54% reduction in fatal MIs? Because the JUPITER data were expressed as RR reductions. Thus, the incidence of fatal MIs was 0.76% in the placebo group and 0.35% in the rosuvastatin-treated group. This means that the AR reduction was 0.41 percentage points, and 0.41 is 54% of 0.76%.

It seems reasonable to expect that if clinicians are going to recommend statins to their patients, then their duty of care must extend to ensuring that the nature of AR and RR is explained. To restrict an explanation of the benefits of statins to adducing RR data alone might amount in the eyes of some to scientific misconduct.