Calculating RR, ARR, NNT
In a trial RR = Event rate (eg # of people with one stroke/ total people) in treatment group/event rate in the control group. ARR = Event rate in control group minus the event rate in the treatment group. NNT = 100/ARR as a percentage.
In an meta-analysis RR is calculated using RevMan. ARR is the same as RD but can also be calculated from event rate in control group times (1 minus RR). See example.
ARR = 1299/11166 (0.1) (100) =1.2%, NNT=83 for 5 yrs.
ARR = 4.3%, NNT 24 for 5 years
Statins, the largest human drug experiment to date November, 2014 James (Jim) Wright jim.wright@ti.ubc.ca
Cholesterol hypothesis the almost universally accepted cholesterol-diet-chd hypothesis (1999). elevated plasma cholesterol levels cause CHD, that diets rich in saturated fat (animal fat) and cholesterol raise cholesterol levels, and that the lowering of cholesterol levels reduces CHD risk.
Epidemiology LDL correlates positively with coronary heart disease events in middle aged adults. HDL correlates negatively with coronary heart disease events in middle aged adults. But correlation does not equal causation.
Against the cholesterol hypothesis WHI trial 16,608 women randomised to hormone therapy or placebo and followed for 5 years. LDL reduced by 11%, HDL increased by 10%. Total CHD RR 1.28 [1.01-1.62].
Against cholesterol hypothesis Heart Protection Study randomised 20,536 people with cardiovascular disease to 40 mg simvastatin or placebo. Benefit was completely independent of baseline LDL level. Benefit was the same for patients with the greatest reduction in LDL as for the patients with the lowest reduction in LDL
Statins Best selling class of drugs of all time > $20 billion/year. Most prescribed class of drugs of all time. Utilization is continuing to increase. One statin, cerivastatin, has been withdrawn from the market because of deaths from rhabdomyolosis.
What is the net health effect of statins for secondary prevention? What is secondary prevention? Patients who have had an event: MI, stroke, TIA, leg amputation. Patients who have occlusive vascular disease: angina, peripheral vascular disease. Patients with documented occlusive vascular disease by angiography or ultrasound.
Secondary prevention trials? 4S simvastatin CHD. CARE pravastatin CHD. LIPID pravastatin CHD. HPS simvastatin CHD, stroke or PVD LIPS fluvastatin CHD ASPEN atorvastatin, diabetes and CHD SPARCL atorvastatin, TIA or stroke CORONA rosuvastatin, CHF GISSI-HF rosuvastatin, CHF
Do statins reduce total mortality in secondary prevention populations? All trials report this outcome. Yes. RR 0.87 [0.84 0.91], ARR 2%, NNT 50 over 5 years.
Do statins reduce total cardiovascular events for secondary prevention Yes RR 0.77 [0.74 0.80], ARR 4.4% NNT 23 over 5 years.
Do any secondary prevention trials report total SAEs? Yes. Two of the most recent do. RR 0.98 [0.95 1.02]
What is a primary prevention patient? A person who has no history of or documented occlusive vascular disease? They can have risk factors including diabetes mellitus, high blood pressure, high cholesterol, smoking etc.
Primary prevention trials WOSCOP pravastatin men high risk AFCAPS lovastatin low HDL ALLHAT-LLT pravastatin high risk PROSPER pravastatin in elderly ASCOT-LLA atorvastatin hypertension ALERT fluvastatin renal transplant CARDs atorvastatin diabetes ASPEN atorvastatin diabetes JUPITER rosuvastatin increased CRP Mega pravastatin Japanese, high cholesterol 3 other small trials
Do statins reduce mortality in primary prevention populations?
Total mortality
Do statins reduce mortality in primary prevention populations? Not if you exclude the trials that failed to account for all patients or were stopped early for benefit.
Total mortality less biased RCTs
Which primary prevention trials reported total SAEs? Seven, representing > 49,000 people and 70% of population studied. WOSCOP AFCAPS PROSPER CARDS ASPEN JUPITER MEGA
Do statins effect total SAEs in primary prevention populations?
Total serious adverse events
Do statins effect total SAEs in primary prevention populations? No. Relative risk (RR) 1.00 [0.96 1.04] In these same 7 trials statins reduced total cardiovascular SAEs, RR 0.72 [0.66 0.79] ARR 1.2% NNT 83 over 5 years. Why is the reduction in CV SAEs not reflected in total SAEs?
What have we done to get the SAE data from the missing trials? We have personally and repeatedly requested in writing the data from all the primary prevention trial authors and manufacturers. In some cases we received no response. In other cases we received friendly replies and pages of data. However, in no case did we receive the SAE data that we were requesting.
Questions? Why will the authors and manufacturers not provide the total SAE data from all the statin trials? What can be done to pressure the authors, companies or regulators to make this information available? If we are correct, what needs to be done next? Can statins cause serious harms that could negate the benefit?
Therapeutics Letter #89 April - May 2014
TL 89 Conclusions The action of statins to reduce many compounds in addition to cholesterol is problematic. Harms with statins are often subtle, usually dose related, sometimes serious and require vigilance to detect. The magnitude of most statin harms remains uncertain at this time. It is essential to weigh the potential benefits and the potential harms in all patients taking or being considered for statin therapy.
Who should be offered statins at the present time? Patients with proven occlusive vascular disease and total cholesterol >3.5 mmol/l (>140 mg/dl). Tell them that the drug reduces their chance of heart attack or stroke by about 4 to 5% and death by about 2% over 5 years. One CV event will be prevented for every 20 to 25 people treated for 5 years. One death will be delayed for every 50 people treated for 5 years. The patient decides whether this benefit is worth it to them.
Who should not be offered a statin? Patients without proven occlusive vascular disease regardless of their risk. Risk assessment tools are inaccurate for individuals and mostly assess non-modifiable risks such as age and sex. This includes people with diabetes, hypertension, smokers, hypercholesterolemia, etc. You can tell them that the trials have not demonstrated a reduction in total death or total hospitalization.
Issues specific to the elderly Magnitude of dose related harms is probably higher. Subtle effects such as effects on energy, sleep quality and cognition are even more difficult to detect. Muscle weakness could lead to falls. In the PROSPER trial, average age 75, there was no benefit in the primary prevention population (total CV events, RR 0.94 (0.77-1.15).
What is the net health effect of the present use of statins? 1. Large benefit (decreased cardiovascular disease). Easily measured. 2. Modest benefit. (difficult to detect) 3. Neutral effect (benefits = harms) 4. Modest harm. (difficult to detect) 5. Large harm. (growing number of severely damaged people).
Conclusions Demand evidence of a net health benefit before recommending any therapy. A net health benefit means a magnitude of reduction in total serious adverse events or total mortality that is meaningful to the patient. Statins have not been proven to have a net health benefit in most patients taking them today (>70%).
Conclusions (cont) Statins are powerful drugs with both benefits and harms. The extent and magnitude of the long term harms of statins are mostly unknown at the present time. Consider stopping statins in patients with short life expectancy.
Conclusions (cont) Further reading: Abramson JD, Rosenberg HG, Jewell N, Wright JM. Should people at low risk of cardiovascular disease take a statin? BMJ2013;347:f6123 doi: 10.1136/bmj.f6123 (Published 22 October 2013)
Questions???