Introduction
Vascular disease in the elderly (age ≥70 years) accounts for a high global burden of disease as risk of atherosclerotic vascular events and their case-fatality rate increase exponentially with age [
1‐
3]. Older patients who survive a major adverse cardiovascular event (MACE) are commonly chronically disabled because of heart failure, cardiac arrhythmia or neurologic deficits. Population aging and the chronic complications of vascular events that physicians encounter in the elderly have triggered a debate about the benefit of cardiovascular risk management in these patients. The “PROspective Study of Pravastatin in Elderly at Risk” (PROSPER) trial found a 15 % reduction in myocardial infarction, stroke and coronary heart disease death with pravastatin treatment in elderly subjects [
4]. Older patients are underrepresented in other trials evaluating cardiovascular prevention strategies and it is, therefore, uncertain whether statins are effective in the elderly. They have lower life-expectancy in general than middle-aged individuals which could potentially limit their benefit from statins. The benefit of statins is particularly uncertain and debated in those with limited life expectancy due to nonvascular diseases [
5].
Statin therapy is recommended for the secondary prevention of MACE in those who have vascular disease, unless comorbidity and polypharmacy confound management [
6‐
8]. For the primary prevention of MACE, the European guideline states that ‘statin therapy may be considered particularly in the presence of at least one other risk factor than age’ and the National Institute for Health and Care Excellence (NICE) guideline advices statin therapy in those with an estimated 10-year risk of MACE ≥10 % [
7,
9]. Since the QRISK2 score estimates a 10-year risk ≥10 % for every patient aged ≥70 years and almost all older patients have at least one vascular risk factor besides age, statin therapy would be indicated for nearly all elderly patients [
10]. However, under-prescription of statins in daily practice may in part reflect uncertainty about the extent to which elderly patients may benefit from a statin [
11]. As the absolute risk for vascular disease and the absolute risk reduction (ARR) caused by a statin are influenced by individual patient characteristics, there is a potential range in the benefit received from a statin. Therefore, we aimed to estimate the absolute treatment effect of statin therapy on MACE for the individual elderly patient by developing and validating a prediction model based on individual patient characteristics [
12,
13].
Discussion
Risk for MACE and the absolute treatment effect of a statin on MACE for individual elderly patients can be estimated with a clinical prediction model containing simple, readily available patient characteristics. There is a wide distribution of MACE risk in elderly patients with and without vascular disease. For secondary prevention of MACE, treating all patients is most beneficial since predicted absolute risk reductions are almost invariably high. With the use of a prediction model that quantifies an individual’s expected absolute risk reduction by statin treatment, those who benefit meaningfully from statin therapy in absolute terms in the primary prevention setting can be identified.
Current guidelines recommend statin treatment for the secondary prevention of MACE in general, but where possible treatment decisions should ideally be made per individual taking into account expected absolute treatment effect, adverse events and patient preferences [
6‐
8]. High absolute treatment effects found in our study underline this recommendation, and moreover enable physicians to estimate the individual absolute treatment effect for a patient. There seems to be a maximal absolute risk reduction that can be achieved by a statin in this high-risk population, as a statin may delay recurrent MACE in patients at very high risk (≥70 % 10-year MACE risk) rather than prevent it from happening during a lifetime. In these individuals, the benefit from statin therapy in recurrent MACE-free life years might be limited. In general, treating elderly patients with vascular disease with a statin seems beneficial as the large expected benefits are very likely to outweigh potential harms. These include adverse events like myopathy and incident type 2 diabetes, drug–drug interactions and the inconvenience of polypharmacy which impair quality of life in elderly patients in particular [
27,
28]. An example of a drug–drug interaction that increases the risk of adverse events comes from the United States of America where 83 % of patients with dyslipidemia is treated with a CYP3A4-metabolized statin of whom 25–30 % concomitantly use a CYP3A4-inhibitor [
29]. Even though there might be a higher risk of serious adverse events in the elderly, there is no conclusive evidence for a higher incidence of rhabdomyolysis, cognitive deterioration, liver or kidney injury [
30].
For primary prevention of vascular disease, current guidelines advise to treat those at high risk which means that practically everyone aged ≥70 years would be given a statin since age dominates risk scores [
6,
7,
9]. However, in clinical practice statin treatment rates for elderly patients are low presumably reflecting ambiguities about the absolute benefit of statin treatment for the primary prevention of vascular disease in the elderly [
11]. Moreover, the incidence of severe comorbidities increases with age and emphasis might be placed on treating these inter-current illnesses. Our prediction model shows that the absolute effect of a statin on MACE is influenced by individual patient characteristics. With the use of this prediction model those individuals who benefit most from statin treatment can be identified. A patient’s advantage of statin therapy in terms of reduction of absolute MACE risk can be estimated and weighed against potential harms of treatment and the costs of statin therapy, even though these are low, in making a treatment decision. A potential harm of statin therapy found in the PROSPER trial was an increase in cancer incidence [
4]. However, a meta-analysis of 35 large randomised controlled trials found an equal risk of cancer in those with and without statins [
31]. Even so, there was no increased cancer risk in statin users during the extensive 8–11 year follow-up of both the PROSPER and ASCOT-LLA trial [
32,
33]. In patients aged >70 years from the ASCOT-LLA population (
n = 2415), atorvastatin did not raise cancer risk (sHR 0.83, 95 % CI 0.58–1.20). Therefore, it is likely that statin therapy does not increase cancer incidence. For adequate estimations of MACE risk and the absolute risk reduction with a statin, death due to cancer was taken into account as a competing event.
Apart from the estimation of individual absolute statin treatment effects, these models inform physicians and patients about an individual’s 5-year or 10-year risk for MACE. Thereby the need for preventive medical and life-style interventions could be established. Informing patients about their risk and engaging them in treatment decision-making might stimulate treatment adherence [
34]. Vascular risk estimation in elderly patients has been challenging and the Systematic Coronary Risk Evaluation (SCORE), QRISK2 and Framingham/Pooled Cohort Equations risk charts are not validated for patients >65 years, >74 years and >79 years of age, respectively [
10,
35,
36]. Furthermore, they do not take into account that many elderly patients die from a nonvascular cause. One risk score for patients aged ≥65 years accounted for competing events like the risk score in this study, but that study included only patients without vascular disease and the outcome was coronary artery events instead of MACE as in this study [
17]. Interestingly, LDL-cholesterol was a weak predictor for MACE in both patients with and without vascular disease. Other risk factors contribute more to risk prediction in the elderly. As our aim was to predict individual absolute benefit from statin therapy we did not assess causality. Previous studies showed no or an inverse association between LDL-cholesterol and all-cause mortality [
37]. In the PROSPER trial pravastatin lowered MACE risk with 15 % whereas it had no effect on all-cause mortality, which implies that the causal association between LDL-cholesterol and MACE may differ from the association between LDL-cholesterol and mortality.
Strengths and limitations
A strength of this study is that the prediction models were derived and externally validated in an elderly population. Moreover, the variable ‘number of medications’ was added to the models as a proxy for comorbidity [
38,
39]. Also, we accounted for competing events (death due to a nonvascular cause) in our statistical analysis. Furthermore, the model for those with vascular disease was validated in a patient population from a cohort study. Thereby, we show that our model is generalizable to a broad elderly population and not restricted to relatively healthy patients in trials. There are some limitations of this study. Overall discriminative ability of these models was moderate, and low in the ASCOT-LLA population. This could be explained by the homogeneity of trial populations in general and of the ASCOT-LLA population in particular with a small range in MACE risk [
40]. The adequate calibration of these models may be more important in assessing model validity, as we aim to accurately predict MACE risk and the absolute statin treatment effect for the individual elderly patient. In the PROSPER and ASCOT-LLA trial, fixed statin doses were given and dosing was not titrated to a specific LDL-cholesterol target. Patients in the SMART study used different statins and dosages. It could be that treatment effects for more potent statins or dosages are underestimated with the current model [
41]. In sensitivity analyses we established what the individual absolute risk reduction might be with atorvastatin 20 mg or the treatment effect from a meta-analysis in elderly subgroups from statin trials. These results should be interpreted with caution as the meta-analysis was performed in slightly older patients (≥75 years) for an LDL-reduction of 1 mmol/l, for a different vascular outcome and not taking competing risks into account [
25]. Finally, our results cannot be extrapolated to the very old (≥85 years) and to patients with chronic kidney disease stage IV or V (eGFR <30 ml/min), since they were not enrolled in these studies.
Acknowledgments
We gratefully acknowledge the contribution of the SMART research nurses; R. van Petersen (data-manager); B. G. F. Dinther (vascular manager) and the participants of the SMART Study Group: A. Algra MD, PhD; Y. van der Graaf, MD, PhD; D. E. Grobbee, MD, PhD; G. E. H. M. Rutten, MD, PhD, Julius Center for Health Sciences and Primary care; F. L. J. Visseren, MD, PhD, Department of Internal Medicine; G. J. de Borst, MD, PhD, Department of Vascular Surgery; L. J. Kappelle, MD, PhD, Department of Neurology; T. Leiner, MD, PhD, Department of Radiology; P. A. Doevendans, MD, PhD, Department of Cardiology. No compensation was received for these contributions. This work was financially supported by ZonMw, the Netherlands Organization for Health Research and Development (Grant No. 836011027).
Compliance with ethical standards