Should I take a cholesterol tablet? If so, which one?

here is a summary that Ben Ewald did for discussion in the RACGP Red Book editorial committee. Unfortunately the graphs do not copy and paste and I cant see how to upload the word doc.

What evidence supports screening for lipids in general practice patients without CVD?

The kinds of evidence that would support this practice:
• That finding elevated LDL leads to treatment that reduces the risk of death.
• Finding elevated LDL leads to treatment that reduces the risk of non fatal CVD.
• That treatment of elevated LDL with statins reduces the risk of fatal and non fatal CVD.
• That elevated LDL denotes CVD risk beyond what is available from history and physical examination.

• Frequency of testing: What is the pre test probability that normal lipids one year may be followed by elevated lipids in a subsequent year?

The previous editions of the RedBook and much literature supports the concept of addressing overall CV risk rather than a single factor. The predominant risk factor for CV death in Australia is waist hip ratio, as shown by Welborn in the analysis of the 1989 Aust Risk Factor Prevalance study 11 year follow up. Waist:Hip ratio is not included in the Framingham equations, but has a much more powerful effect than cholesterol levels.

(graphs showing the predictive power of waist hip ratio, diastolic BP, trigs, and total chol decline in that order).

Waist:Hip ratio can be easily measured in the clinic, is quicker and cheaper than measuring cholesterol, and denotes risk more strongly so why dont we use it?

Does primary prevention with statins reduce mortality, and is any benefit related to baseline lipid levels?
This question is the topic of many trials and several systematic reviews and meta analyses. The main methodological problem is that many prevention trials have been done on samples that include patients with existing CHD. Here I have summarised 2 reviews:

1) Statins and all cause mortality in high-risk primary prevention. K.Ray Arch Int Med 2010; 170 (12):1024-1031 11 trials, cut date May 09.

A 2010 meta analysis by Ray, has solved this problem by obtaining the data from original trials stratified by prior CHD status. Re analysis shows that primary prevention with statins gave RR 0.91 (ci 0.83-1.01) so a non significant result, based on 65 229 participants and 244000 person years of follow up. One large trial that favoured treatment with statins “Jupiter” specifically recruited people with normal lipids but elevated CRP. [Authors of this trial also have extensive conflicts of interest, such as holding patents on a method for high sensitivity CRP testing]

The paper by Ray also examined the effect of baseline lipids on risk reduction. Figure 4, p 1029. This convincingly shows that the degree of benefit has no relationship to the baseline cholesterol level.

Ray did not report the effect on CVD mortality but this was later calculated by Dentali showing RR 0.89 (0.80-0.99). Now statistically significant but a small effect.

In summary there was no statistically significant effect despite the huge sample size. If we accept the observed RR and apply this to a 60 year person with mortality of 10/1000 person years as observed in these studies, NNT is 1110 for 1 year to prevent one death, or 990 using the number for CVD deaths only.

2) F. Taylor , Cochrane review, Statins for the primary prevention of cardiovascular disease. 14 trials, cut date march 07

Taylor accepted trails with up to 10% secondary prevention patients. This set includes neither JUPITER or ALLHAT, the two biggest contributors to Ray. They wanted to examine the effect by baseline LDL level but the information was missing from many reports.
Effects seen:
All cause mortality RR 0.83 (0.73-0.95) NNT as above, 588-1 yr.
Fatal + non fatal CHD events RR 0.72 (.65-.79)
Fatal + non fatal stroke RR 0.78 (.65-.94)
Revascularisation RR 0.66 (.53-.83)

In summary, the effects of statins on all cause mortality are small, if present, and are not related to baseline LDL level. The effects on non fatal CHD, stroke, and revascularisation are larger. While we could accept that statins have a small benefit in primary prevention it is not based on the baseline LDL level.

The Redbook 7th edition lists testing for lipids in those aged 45 as level A evidence, referencing a NHF position statement, however the position statement does not examine screening, only treatment.

Should lipids be re-measured?
Evidence from child-to adult cohort studies such as the Bogalusa heart study shows considerable tracking of lipid levels. People tend to stay in their quartiles. I could not find any reports of the natural variation in untreated lipid levels over time but the evidence must exist.

Proposed changes to guidance
Measure lipid levels once at age 35, to detect those with familial hypercholesterolaemia, (prevalence about 0.2%) then not again unless people are high risk on other grounds.

BMI, Waist,or Waist:Hip ratio?
Welborn examined the predictive values of various ways to measure obesity, as shown in the figure below, showing that Waist:Hip ratio is much more predictive than BMI, or Waist circumference.