All-Cause Mortality

Fifteen trials reported all-cause mortality (Table 3; Appendix C1).^{52,54,60,63,65,66,69,73,74,82,83,93,95,96, 103} Absolute event rates ranged from 0 to 5 percent in the statin groups and 0 to 6 percent in control groups. Statins were associated with statistically significant reduction in risk of all-cause mortality versus placebo in two trials. The large JUPITER (Justification for the Use of Statins in Prevention: An Intervention Trial Evaluating Rosuvastatin) trial⁷⁴ (n=17,802; 2 years followup), which enrolled patients with elevated CRP levels and LDL-C levels of less than 130 mg/dL, reported a hazard ratio (HR) of 0.80 after 2 years of statin therapy (95% CI, 0.69 to 0.97; ARD, -0.6%). The smaller ACAPS (Asymptomatic Carotid Artery Plaque Study) trial (n=919; 3 years followup),⁵² which enrolled persons with early-onset cerebrovascular disease, also found reduced risk of all-cause mortality with statin therapy, though the estimate was less precise (RR, 0.12 [95% CI, 0.02 to 0.99]; ARD, -0.02%). Pooling evidence from all trials resulted in a very similar risk estimate to that in the JUPITER trial (RR, 0.86 after 1 to 6 years [95% CI, 0.80 to 0.93]; I²=0%; ARD, -0.40% [95% CI, -0.64 to -0.17]; I²=4%) (Appendix D1). Across studies, the NNT ranged from 47 to 294 over 2 to 6 years in nine trials, and six trials reported no benefit from statins; the pooled NNT was 250. The risk estimate was heavily influenced by the JUPITER and ASCOT-LLA (Anglo-Scandinavian Cardiac Outcomes Trial–Lipid-Lowering Arm) studies, both of which were stopped early and together accounted for about 40 percent of the total sample and more than 35 percent of mortality events. The point estimate and ARD from ASCOT-LLA (3.6% vs. 4.1% after 3 years; RR, 0.87 [95% CI, 0.71 to 1.05]; ARD, -0.55%), which focused on patients with hypertension, was similar to the point estimate from JUPITER.

Table 3

Clinical Outcomes and Pooled Risk Estimates From Randomized Trials of Statins Versus Placebo.

Results were similar in sensitivity analyses (Table 4). Excluding results from JUPITER and both JUPITER and ASCOT-LLA had little effect on pooled estimates (RR, 0.88 [95% CI, 0.80 to 0.96]; I²=0% and RR, 0.88 [95% CI, 0.80 to 0.97]; I²=0%, respectively). Restricting the analysis to good-quality studies^{69,74,82,96,103} also did not affect estimates (RR, 0.85 [95% CI, 0.77 to 0.94]; I²=0%), and results were similar when trials were stratified according to duration of followup of 3 years or less (RR, 0.83 [95% CI, 0.72 to 0.94]; I²=0%)^{52,60,65,66,74,82,93} versus more than 3 years (RR, 0.88 [95% CI, 0.80 to 0.98]; I²=0%).^{54,63,69,73,83,95,96,103} There were also no differences in estimates when three trials^60,82,95 that included patients with prior CVD were excluded (RR, 0.86 [95% CI, 0.78 to 0.94]; I²=4%) or when two trials^63,74 that enrolled patients with mean baseline TC levels of less than 200 mg/dL were excluded (RR, 0.87 [95% CI, 0.79 to 0.95]; I²=0%). Results were also similar when trials were stratified according to baseline LDL-C level of less than 160 mg/dL (RR, 0.87 [95% CI, 0.80 to 0.95]; I²=0%) versus 160 mg/dL or greater (RR, 0.79 [95% CI, 0.62 to 1.01]; I²=0%).

Table 4

Sensitivity Analysis: Pooled Estimates for Statins Versus Placebo.

Cardiovascular Mortality

Cardiovascular mortality was reported in 10 trials (Table 3; Appendix C1).^{52,54,60,64,74,82,83,95,96,103} The effect of statin use on cardiovascular mortality was somewhat inconsistent. Although the large WOSCOPS (West of Scotland Coronary Prevention Study) (n=6,595)⁹⁶ trial found a statistically significant difference between statins versus placebo and risk of cardiovascular mortality (RR, 0.68 [95% CI, 0.48 to 0.98]), the JUPITER (n=17,802),74 AFCAPS/TexCAPS (Air Force/Texas Coronary Atherosclerosis Prevention Study) (n=6,605),⁵⁴ and MEGA (Management of Elevated Cholesterol in the Primary Prevention Group of Adult Japanese) (n=7,832)⁸³ trials reported similar point estimates that did not reach statistical significance (0.3 % versus 0.4 % after 2 years; RR, 0.78 [95 % CI 0.48 to 1.27 ]0.5% vs. 0.8% after 5 years; RR, 0.68 [95% CI, 0.37 to 1.26] and 0.3% vs. 0.5% after 5 years; RR, 0.63 [95% CI, 0.30 to 1.33], respectively), and the ASCOT-LLA (n=10,305)⁶⁰ and HOPE-3 (Heart Outcomes Prevention Evaluation-3) (n=12,705)¹⁰³ trials found no effect (1.4% vs. 1.6% after 3 years; RR, 0.90 [95% CI, 0.66 to 1.23] and 2.4% vs. 2.7% after 6 years; RR, 0.90 [95% CI, 0.72 to 1.11], respectively). In pooled analysis, statin therapy was associated with decreased risk of cardiovascular mortality (RR, 0.82 after 2 to 6 years [95% CI, 0.71 to 0.94]; I²=0%) (Appendix D2). The pooled ARD was -0.20 percent (95% CI, -0.35 to -0.05; I²=11%) and the pooled NNT was 500 (range, 76 to 1,111 in 8 trials; 2 trials found no benefit with statin therapy).

Findings were similar in sensitivity analyses (Table 4). Restricting the analysis to good-quality trials^{64,74,82,96,103} resulted in a similar risk estimate (RR, 0.82 [95% CI, 0.44-69 to 0.98]; I²=0%). The point estimates were similar when studies were stratified according to duration of followup of 3 years or less (RR, 0.85 [95% CI, 0.65 to 1.10]; I²=0%)^{52, 60,74,82} or more than 3 years (RR, 0.81 [95% CI, 0.68 to 0.95]; I²=0%). Removing three trials^60,82,95 that included a small proportion of persons with prior CVD events also did not affect the risk estimate (RR, 0.80 [95% CI, 0.68 to 0.93]; I²=0%). Excluding the JUPITER trial,⁷⁴ which enrolled persons with baseline TC levels of less than 200 mg/dL and was stopped early, also resulted in a similar pooled estimate (RR, 0.82 [95% CI, 0.71 to 0.95]; I²=0%). The estimate was also similar when excluding both JUPITER⁷⁴ and ASCOT-LLA⁶⁰ (RR, 0.80 [95% CI, 0.68 to 0.95]; I²=0%).

Stroke

Thirteen trials reported incidence of fatal and nonfatal stroke (Table 3; Appendix C1).^{52,60,63,64, 69,72,74,82,83,92,95,96,103} One trial reported results separately for nonhemorrhagic and hemorrhagic stroke;⁸³ the other trials did not clearly specify the type of stroke. Results from individual trials generally favored statin therapy over placebo or no statin, though estimates were not always statistically significant. Although four trials enrolled patients with mild cerebrovascular disease at baseline, none were designed to evaluate effects of statins on risk of stroke, given relatively small sample sizes (n=250 to 919) and relatively short duration of followup (6 months to 3 years).^52,65,67,92 Two^52,92 of these trials reported stroke events, though one trial only reported one event.⁹²

Statins were associated with decreased risk of fatal or nonfatal stroke (RR, 0.71 after 6 months to 6 years [95% CI, 0.62 to 0.82]; I²=0%) (Appendix D3). The pooled ARD was -0.38 percent (95% CI, -0.53 to -0.23; I²=0%) for a NNT to prevent 1 fatal or nonfatal stroke of 263. Excluding one trial that reported a NNT of 11, the NNT ranged from 92 to 625 in 10 trials after 1 to 6 years; two trials reported no benefit with statins. A good-quality systematic review reported a similar risk estimate (10 trials; RR, 0.78 [95% CI, 0.68 to 0.89]; I²=26%).¹⁰⁵

Findings were similar in sensitivity analyses (Table 4). There were no clear differences in pooled estimates when one poor-quality trial⁷² was excluded from the analysis (RR, 0.72 [95% CI, 0.62 to 0.83]; I²=0%), when the analysis was restricted to good-quality trials (RR, 0.68 [95% CI, 0.56 to 0.83]; I²=0%), when one trial with 6 months of followup was excluded (RR, 0.71 [95% CI, 0.62 to 0.82]; I²=0%), and when studies were stratified according to duration of followup of 3 years or less (RR, 0.64 [95% CI, 0.51 to 0.80]; I²=0%) or more than 3 years (RR, 0.77 [95% CI, 0.64 to 0.92]; I²=0%). Removing three trials^60,82,95 that included persons with prior CVD events (RR, 0.70 [95% CI, 0.60 to 0.83]; I²=0 %) or two trials^63,74 that enrolled patients with mean baseline TC levels of less than 200 mg/dL also did not affect the estimate (RR, 0.73 [95% CI, 0.61 to 0.85]; I²=0%). Estimates were also similar when trials were stratified according to baseline LDL-C levels of less than 160 mg/dL versus 160 mg/dL or greater (RR, 0.70 [95% CI, 0.60 to 0.81]; I²=0% vs. RR, 0.83 [95% CI, 0.58 to 1.19]; I²=0%, respectively). Estimates were also similar when JUPITER (RR, 0.75 [95% CI, 0.63 to 0.89]; I²=0%) and both JUPITER and ASCOT-LLA (RR, 0.75 [95% CI, 0.63 to 0.90]; I²=0%) were excluded.

When stratified by fatal and nonfatal stroke, statins were associated with decreased risk of nonfatal (3 trials; RR, 0.57 [95% CI, 0.41 to 0.81]; I²=0%; ARD, -0.32% [95% CI, -0.52 to -0.12])^69,74,92 and fatal stroke (2 trials; RR, 0.38 [95% CI, 0.12 to 1.22]; I²=0%; ARD, -0.11% [95% CI, -0.38 to 0.15]),^69,74 although few trials reported separate results for fatal and nonfatal stroke, estimates were imprecise, and the difference in risk of fatal stroke was not statistically significant.

MI

Twelve trials reported incidence of fatal and nonfatal MI (Table 3; Appendix C1).^{52,54,60,63,64,67, 69,74,82,83,96,103} Results from individual trials were mixed but most large trials found that statin use was associated with a significant reduction in risk of MI. For example, risk estimates in the AFCAPS/TexCAPS (2% vs. 3%; RR, 0.60 [95% CI, 0.43 to 0.83]), ASCOT-LLA (1.7% vs. 2.9%; RR, 0.67 [95% CI, 0.53 to 0.84]), HOPE-3 (0.7% vs. 1.1%; RR, 0.65 [95% CI, 0.45 to 0.95]), JUPITER (0.3% vs. 0.7%; RR, 0.45 [95% CI, 0.29 to 0.69]), MEGA (0.5% vs. 0.8%; RR, 0.53 [95% CI, 0.29 to 0.95]), and WOSCOPS (5.3% vs. 7.5%; RR, 0.69 [95% CI, 0.56 to 0.86]) trials all favored statin use. Differences between statin and placebo groups in smaller trials such as ACAPS (1.1% vs. 1.1%; RR, 1.00 [95% CI, 0.29 to 3.42]), ASTRONOMER (Aortic Stenosis Progression Observation: Measuring Effects of Rosuvastatin) (0% vs. 2.2%; RR, 0.14 [95% CI, 0.008 to 2.76]), CAIUS (Carotid Atherosclerosis Italian Ultrasound Study) (1.3% vs. 1.3%; RR, 1.02 [95% CI, 0.15 to 7.15]), and KAPS (Kuopio Atherosclerosis Prevention Study) (1.4% vs. 3.8%; RR, 0.36 [95% CI, 0.09 to 1.39]) were not statistically significant. In pooled analysis, statins were associated with decreased risk of MI (RR, 0.64 after 2 to 6 years [95% CI, 0.57 to 0.71]; I²=0%; ARD, -0.81% [95% CI, -1.19 to -0.43]; I²=70%) (Appendix D4). The pooled NNT to prevent 1 MI was 123; the NNT ranged from 45 to 263 in 10 trials, and two trials reported no benefit with statin therapy. Results based on six good-quality trials were consistent with the overall pooled estimate (RR, 0.61 [95% CI, 0.51 to 0.72]; I²=8%).^{64,69,74,82,96,103}

Findings were similar in sensitivity analyses (Table 4). Restricting the analysis to the seven trials^{54,63,64,69,83,96,103} with more than 3 years of followup did not affect the estimate (RR, 0.65 [95% CI, 0.56 to 0.74]; I²=0%). Excluding two trials^60,82 that enrolled some participants with a history of CVD events (RR, 0.63 [95% CI, 0.55 to 0.72]; I²=0%), excluding two trials^63,74 that enrolled patients with baseline TC levels of less than 200 mg/dL (RR, 0.64 [95% CI, 0.57 to 0.73]; I²=0%), and restricting the analysis to trials that enrolled patients with baseline LDL-C levels of less than 160 mg/dL (RR, 0.61 [95% CI, 0.54 to 0.70]) had little effect on estimates. Estimates were also similar when JUPITER (RR, 0.68 [95% CI, 0.58 to 0.73]; I²=0%) and both JUPITER and ASCOT-LLA (RR, 0.65 [95% CI, 0.57 to 0.74]; I²=0%) were excluded.^60,74

Seven trials reported separate results for fatal and/or nonfatal MI.^{52,54,67,74,82,83,96} When analyzed separately, estimates for fatal MI (RR, 0.70 [95% CI, 0.50 to 0.99]; I²=0%; ARD, -0.16% [95% CI, -0.42 to 0.11]) and nonfatal MI (RR, 0.64 [95% CI, 0.46 to 0.91]; I²=50%; ARD, -0.46% [95% CI, -0.90 to -0.02]) were similar.

Revascularization

Incidence of revascularization was reported in seven trials (Table 3; Appendix C1).^{54,69,74,82,83,96, 103} The five largest trials, AFCAPS/TexCAPS,⁵⁴ HOPE-3,¹⁰³ JUPITER,⁷⁴ MEGA,⁸³ and WOSCOPS,⁹⁶ all found that statins were associated with reduced risk of revascularization (RR estimates ranged from 0.54 to 0.68). The two smaller trials^69,82 reported similar risk estimates (RR, 0.70 [95% CI, 0.42 to 1.17] and RR, 0.79 [95% CI, 0.22 to 2.91]), though differences were not statistically significant. When results were pooled, statins were associated with reduced risk of revascularization (RR, 0.63 after 2 to 6 years [95% CI, 0.56 to 0.72]; I²=0%) (Appendix D5). The ARD was -0.66 percent (95% CI, -0.87 to -0.45; I²=8%; NNT range, 65 to 244; pooled NNT, 152). Findings were similar in sensitivity analyses (Table 4). Restricting the analysis to the five good-quality trials did not affect this estimate (RR, 0.62 [95% CI, 0.52 to 0.74]; I²=0%).^{69,74,82,96,103} Excluding two trials^74,82 that had followup of 3 years or less resulted in a similar estimate (RR, 0.66 [95% CI, 0.57 to 0.77]; I²=0%). Results were similar in the subgroup of five trials^{54,69,74,83,103} in which the mean baseline LDL-C level was less than 160 mg/dL (RR, 0.63 [95% CI, 0.55 to 0.73]; I²=0%) (Table 3).

Composite Cardiovascular Outcomes

Thirteen trials reported on composite cardiovascular outcomes (Table 3; Appendix C1).^{52,54,60,63, 65,69,72-74,83,95,96,103} In two trials, the composite outcomes were not well defined,^65,72 and the composite outcome definition varied in the remainder of the studies (Appendix C1). In general, statin therapy was associated with decreased risk of composite cardiovascular outcomes versus placebo or no statin. Despite the variability in how cardiovascular outcomes were defined, we pooled rates of composite cardiovascular outcomes, as event rates for some individual outcomes were low in many trials. When pooled, statin therapy significantly reduced incidence of composite cardiovascular outcomes compared with placebo (RR, 0.70 [95% CI, 0.63 to 0.78]; I²=36%) (Appendix D6). The ARD ranged from -2.26 to -0.35 percent over 1 to 6 years of followup, and the pooled ARD was -1.39 percent (95% CI, -1.79 to -0.99; NNT range, 8 to 286; pooled NNT, 72). Excluding JUPITER (RR, 0.72 [95% CI, 0.64 to 0.80]; I²=29%)⁷⁴ and both JUPITER and ASCOT-LLA (RR, 0.71 [95% CI, 0.63 to 0.81]; I²=35%)⁶⁰ resulted in similar estimates, as did restriction to good-quality trials (RR, 0.69 [95% CI, 0.61 to 0.78]; I²=28%) or trials in which mean baseline LDL-C levels were less than 160 mg/dL (RR, 0.70 [95% CI, 0.61 to 0.79]; I²=46%) (Table 4).

Assessment for Publication Bias

We did not identify funnel plot asymmetry based on funnel plots for all-cause mortality, fatal and nonfatal stroke, and fatal and nonfatal MI (Appendixes D7-11). Funnel plot asymmetry was present for cardiovascular mortality (p for Egger test=0.049), but few small trials were available (Appendix D8).

Demographic Characteristics

Clinical Characteristics

Lipid Parameters

Six trials (AFCAPS/TexCAPS, ASCOT, HOPE-3, JUPITER, MEGA, and WOSCOPS) reported effects of statin treatment on cardiovascular outcomes in subgroups defined by baseline lipid levels.^{54,60,83,103,106,107} Relative effect estimates favored statin therapy in all lipid subgroups, with no clear pattern suggesting differential risk estimates according to baseline TC, LDL-C, HDL-C, or TG levels in five of the trials (Table 6). The fifth trial (MEGA)⁸³ found no difference in risk of CHD events between statins versus no statins in patients with baseline LDL-C levels of less than 155 mg/dL (HR, 0.90 [95% CI, 0.56 to 1.44]) and decreased risk in patients with baseline LDL-C levels of greater than 155 mg/dL (HR, 0.54 [95% CI, 0.35 to 0.81]), but the interaction was not statistically significant (p=0.06).¹⁰³

Table 6

Statins Versus Placebo: Effects in Subgroups Based on Clinical Characteristics.

We also found no clear differences in risk estimates in trials of statins versus placebo in sensitivity and stratified analyses based on baseline TC, HDL-C, or TG levels, though statistical heterogeneity was reduced in some cases (see Key Question 1a).

Other Characteristics

The AFCAPS/TexCAPS trial stratified results according to baseline LDL-C and CRP levels in a post-hoc analysis.¹⁰⁰ Among patients with an LDL-C level of less than 149 mg/dL, statin therapy was associated with decreased risk of acute major coronary events in those with a CRP level greater than 0.16 mg/dL (RR, 0.58 [95% CI, 0.34 to 0.98]) but not in those with a CRP level less than 0.16 mg/dL (RR, 1.08 [95% CI, 0.56 to 2.08]); although the interaction among statin therapy, baseline lipid level, and CRP level did not reach statistical significance (p=0.06) (Table 6).¹⁰⁰ Among patients with an LDL-C level of 149 mg/dL or greater, statin therapy was associated with reduced risk of major coronary events in patients with a CRP level less than 0.16 mg/dL (RR, 0.38 [95% CI, 0.21 to 0.70]) and a CRP level greater than 0.16 mg/dL (RR, 0.68 [95% CI, 0.42 to 1.10]). Subsequently, the JUPITER trial, which enrolled patients with a CRP level of 2.0 mg/L or greater at baseline (median, 4.2 to 4.3 mg/L) and an LDL-C level of less than 130 mg/dL (median, 108 mg/dL), found that statin therapy was associated with decreased risk of all-cause mortality (RR, 0.80 [95% CI, 0.67 to 0.96]), cardiovascular mortality (RR, 0.53 [95% CI, 0.41 to 0.69]), and other cardiovascular outcomes versus placebo.⁷⁴ The more recent HOPE-3 trial (mean baseline LDL-C level, 128 mg/dL; median CRP level, 2.0 mg/L)¹⁰³ reported findings somewhat discordant from AFCAPS/TexCAPS. HOPE-3 found no difference in effects of statins on composite cardiovascular events when patients were stratified according to a CRP level of 2.0 mg/L or less (HR, 0.82 [95% CI, 0.64 to 1.06]) or greater than 2.0 mg/L (HR, 0.77 [95% CI, 0.60 to 0.98]; p=0.70 for interaction).

Three trials reported no interaction between effects of statins versus placebo and body mass index (BMI).^60,80,87 The MEGA trial also reported no interaction between effects of statins and smoking status (smokers: HR, 0.69 [95% CI, 0.42 to 1.13] vs. nonsmokers: HR, 0.64 [95% CI, 0.43 to 0.96]).⁸⁷ JUPITER found similar effects of statin therapy on the primary composite cardiovascular endpoint in the subgroup of patients with an elevated CRP level and no other risk factors other than increased age (HR, 0.63 [95% CI, 0.44 to 0.92]) and the overall sample (HR, 0.56 [95% CI, 0.46 to 0.69]).⁷⁴

No trial reported stratified results for patients with or without familial hypercholesterolemia.

Withdrawal Due to Adverse Events

Nine trials reported withdrawal due to adverse events (Table 7).^{52,54,82,83,92,93,95,102,103} Seven trials found no difference between statins versus placebo in rates of withdrawal due to adverse events. In the MEGA trial, patients who received statins were more likely than patients receiving placebo to withdraw due to adverse events (11.0% vs. 8.4%; RR, 1.31 [95% CI, 1.15 to 1.51]),⁸³ while in the HOPE-3 trial, fewer patients taking statins than placebo withdrew due to adverse events (8.5% vs. 9.1%; RR, 0.70 [95% CI, 0.62 to 0.79]).¹⁰³ The pooled estimate showed no difference in risk (9 trials; RR, 0.95 [95% CI, 0.75 to 1.21]; I²=86%; ARD, 0.02% [95% CI, -1.55 to 1.60]) (Appendix D12).

Table 7

Harms of Statins Versus Placebo in Randomized Trials.

Serious Adverse Events

Eight trials reported risk of serious adverse events (Table 7).^{54,64,66,73,74,93,102,103} There were no significant differences between treatment and placebo groups reported in any trial or when trials were pooled, based on seven trials with poolable data (RR, 0.99 [95% CI, 0.94 to 1.04]; I²=0%; ARD, 0.07% [95% CI, -0.29 to 0.42]) (Appendix D13). Rates of serious adverse events with statin therapy varied substantially between trials (0.9%⁹³ to 34%⁵⁴) due to variability in how serious adverse events were defined, methods used to ascertain adverse events, duration of followup, and other factors.

Cancer

Eleven trials (in 12 publications) reported risk of cancer (Table 7).^{52,54,64,65,67,69,74,82,83,96,102,103} Ten trials reported any incident cancer, with none finding significant differences between statins and placebo in risk.^{54,64,65,67,74,82,83,96,102,103} Rates of any cancer with statin therapy ranged from 0.5 to 7.6 percent. Incidence of fatal cancer was reported in five trials.^{52,54,69,74,103} The JUPITER trial found that statins were associated with lower risk of fatal cancer versus placebo (0.4% vs. 0.7%; RR, 0.60 [95% CI, 0.40 to 0.92]).⁷⁴ The other four trials reported no differences.

In pooled analyses, there were no difference between statin therapy and placebo or no statin in risk of any cancer (10 trials; RR, 1.02 [95% CI, 0.90 to 1.16]; I²=43%; ARD, 0.11% [95% CI, -0.39 to 0.60]) (Appendix D14) or fatal cancer (5 trials; RR, 0.85 [95% CI, 0.59 to 1.21]; I²=61%; ARD, -0.17% [95% CI, -0.50 to 0.16]) (Appendix D15).

New-Onset Diabetes

Four trials (in 5 publications) and two observational studies reported risk of new-onset diabetes (Table 7).^{60,74,101,103,109-111} Unpublished data on risk of diabetes from two other trials of statins in adults without prior cardiovascular events (MEGA and AFCAPS/TexCAPS) were also reported in a systematic review.¹¹² Based on a pooled analysis of published and unpublished trial data, there was no difference in risk of diabetes (6 trials; RR, 1.05 [95% CI, 0.91 to 1.20]; I²=52%; ARD, 0.12% [95% CI, -0.31 to 0.54]) (Appendix D16). Analysis using the profile likelihood method resulted in a similar estimate (RR, 1.06 [95% CI, 0.93 to 1.18]). Results from these studies were inconsistent. The JUPITER trial found an increased risk of diabetes with statin use (3.0% vs. 2.4%; RR, 1.25 [95% CI, 1.05 to 1.49]).⁷⁴ In stratified analysis of JUPITER data, participants with one or more diabetes risk factors (including metabolic syndrome, impaired fasting glucose, BMI >30 kg/m², and a hemoglobin A1c level of >6.0%) were at higher risk of incident diabetes than those without diabetes risk factors (HR, 1.28 [95% CI, 1.07 to 1.54] vs. HR, 0.99 [95% CI, 0.45 to 2.21]).¹⁰⁹ The other five trials found no clear association between statin use and increased risk of diabetes. The WOSCOPS trial found that statin use was associated with reduced risk of diabetes (1.9% vs. 2.8%; HR, 0.70 [95% CI, 0.50 to 0.98]),¹⁰¹ and the ASCOT-LLA (3.5% vs. 3.6%; RR, 1.02 [95% CI, 0.86 to 1.23])⁶⁰ and HOPE-3 (3.0% vs. 2.6%; RR, 1.15 [95% CI, 0.91 to 1.44], respectively)¹⁰³ trials found little difference in risk. Both trials (MEGA and AFCAPS/TexCAPS) with unpublished data on risk of diabetes found no association between statin use and diabetes (5.7% vs. 5.3%; RR, 1.07 [95% CI, 0.87 to 1.32] and 2.3% vs. 2.3%; RR, 0.98 [95% CI, 0.71 to 1.35]).¹¹²

Potential reasons for the discrepancy in estimates of diabetes risk include differences in the methods used to diagnose diabetes and differences in the potency of the statins evaluated. In JUPITER, diagnosis of diabetes was based on physician report.¹⁰⁹ In WOSCOPS,¹⁰¹ diagnosis of diabetes was based on a fasting plasma glucose level of greater than 126 mg/dL on at least two occasions, with an increase of at least 36 mg/dL from baseline; in ASCOT-LLA,⁶⁰ as a fasting plasma glucose level of greater than 126 mg/dL; and in HOPE-3, as a fasting plasma glucose level of greater than 126 mg/dL or a hemoglobin A1c level greater than 110% the upper limit of normal.¹⁰³ Methods for diagnosing diabetes in the MEGA and AFCAPS/TexCAPS trials were physician report, use of medication, or fasting plasma glucose of level of greater than 126 mg/dL.¹¹² The pooled estimate was similar in a sensitivity analysis in which WOSCOPS diabetes incidence was based on less stringent alternative criteria for diabetes¹¹² that excluded the requirement for an increase of at least 36 mg/dL from baseline (RR, 1.07 [95% CI, 0.95 to 1.19]; I²=33%). JUPITER was the only trial to evaluate use of a high-potency statin (see Key Question 3).

Two large observational studies also found mixed evidence on statin use and diabetes. A matched case-control study that used the U.K. General Practice Research Database to identify 588 diabetes cases and 2,063 matched controls (patients with prior MI excluded) found an odds ratio (OR) of 1.01 (95% CI, 0.80 to 1.40) with statin use versus nonuse, after adjustment for BMI, hypertension, steroid use, smoking history, and number of visits to a general practitioner within 3 years.¹¹⁰ However, an analysis from the Women's Health Initiative of 10,834 postmenopausal women using statins and 143,006 women with no statin use and no history of self-reported CVD found that statin use significantly increased risk of incident diabetes (adjusted HR, 1.48 [95% CI, 1.38 to 1.59]).¹¹¹ The WHI results included multivariate adjustment for age, race/ethnicity, education, smoking history, BMI, physical activity, alcohol use, energy intake, family history of diabetes, and use of hormone therapy. The studies used slightly different methods to determine presence of diabetes. The U.K. General Practice Research Database used computerized medical records of two or more prescriptions of insulin or an oral hypoglycemic or at least three recorded entries of diet management for diabetes.¹¹⁰ Cases with a new diabetes diagnosis within 90 days of first treatment for hyperlipidemia were excluded. WHI relied on self-reported new diabetes diagnosis based on patient questionnaires.¹¹¹

Muscle-Related Harms

Myalgia was reported in seven trials,^{54,65,66,82,93,96,102} rhabdomyolysis in eight trials,^{54,60,66,74,83,93, 102,103} and myopathy in four trials (Table 7).^{54,74,102,103} One small trial found that statins were associated with decreased risk of myalgia versus placebo (RR, 0.53 [95% CI, 0.31 to 0.90]), though it did not report how myalgia was defined;⁶⁵ the other six trials reported no difference between groups (7 trials; RR, 0.96 [95% CI, 0.79 to 1.16]; I²=42%; ARD, 0.03% [95% CI, -0.53 to 0.60]) (Appendix D17). Rates of myalgia with statin therapy ranged from 0.3 to 22.8 percent.

There was also no increased risk of myalgia in two trials that evaluated high-potency statin therapy (RR, 1.03 [95% CI, 0.97 to 1.11]⁷⁴ and RR, 1.05 [95% CI, 0.73 to 1.52]⁹³).

None of the trials found a significant difference between statins versus placebo in risk of rhabdomyolysis, although the number of events was very small (3 events in one study,⁵⁴ 1 event in three studies,^60,74,103 and none in four studies).^66,83,93,102 The pooled estimate for rhabdomyolysis showed no difference but the estimate was imprecise and based on only four trials that reported events (RR, 1.57 [95% CI, 0.41 to 5.99]; I²=0%; ARD, 0.01% [95% CI, -0.02 to 0.03]) (Appendix D18). Three trials found no difference between statins versus placebo in risk of myopathy (RR, 1.09 [95% CI, 0.48 to 2.47]; I²=0%; ARD, 0.01% [95% CI, -0.05 to 0.06]) (Appendix D19),^74,102,103 and another trial reported no cases of myopathy in either group.⁵⁴

Liver-Related Harms

Eleven studies reported no difference between statin therapy versus placebo in risk of elevation in alanine or aspartate aminotransferase, although the definitions varied (degree of elevation, aspartate and/or alanine aminotransferase, single or repeatedly elevated levels) (Table 7).^{52,54,64-66,69,74,82,83,93,96} There was no difference between statin therapy versus placebo or no statin in risk of aminotransferase elevation based on any definition (11 trials; RR, 1.10 [95% CI, 0.90 to 1.35]; I²=0%; ARD, 0.08% [95% CI, -0.04 to 0.19]) (Appendix D20) or when the analysis was restricted to trials that reported risk of an alanine aminotransferase level greater than 3 times the upper limit of normal, which was the most consistently used definition (5 trials; RR, 1.11 [95% CI, 0.78 to 1.57]; I²=0%).^{64,65,69,74,82,93,96} One trial reported no difference between statins versus placebo in risk of (undefined) hepatic disorders (RR, 1.16 [95% CI, 0.96 to 1.41]).⁷⁴ Very few serious liver-related harms were reported.

Other Harms

Two trials of primary prevention populations reported no difference between statins (one using high-intensity rosuvastatin⁷⁴ and one using moderate-intensity atorvastatin⁶⁰) versus placebo in risk of renal impairment (HR, 1.29 [95% CI, 0.76 to 2.19]⁶⁰ and RR, 1.11 [95% CI, 0.99 to 1.26]⁷⁴). One trial reported the effect of statin treatment on a series of cognitive tests.⁹² The study found that statin-treated patients showed less improvement on tests previously shown to be sensitive to statin treatment (group difference in mean change of summary z-scores, 0.18 [95% CI, 0.07 to 0.29]; p=0.002) and on several other tests (group difference in mean change of summary z-scores, 0.17 [95% CI, 0.05 to 0.29]; p=0.007) but not on tests previously shown to be statin-insensitive (group difference in mean change of summary z-scores 0.02 [95% CI, -0.07 to 0.10]; p=0.72), although the clinical importance of these findings is difficult to interpret (Table 7).

The HOPE-3 trial found that statin use increased risk of cataract surgery, which was unanticipated and not a predetermined outcome of the trial (3.8% vs. 3.1%; RR, 1.24 [95% CI, 1.03 to 1.49]).¹⁰³ None of the other primary prevention trials reported this outcome.

Benefits

Direct evidence on clinical outcomes associated with differential intensity of statin therapy is extremely limited. The two trials of statin therapy at different intensities were underpowered to evaluate clinical outcomes.^66,92 One trial of women (n=485 randomized to statin therapy) with moderate hyperlipidemia reported no deaths in women randomized to either atorvastatin 10 or 20 mg/day (moderate-intensity) or 40 or 80 mg/day (high-intensity).⁶⁶ The other trial, which enrolled men or women (n=206 randomized to statin therapy) with moderate hyperlipidemia, reported no stroke events in patients randomized to simvastatin 10 mg/day (low-intensity) and one event in patients randomized to 40 mg/day (moderate-intensity).⁹² A third trial, which initially randomized patients to lovastatin 20 mg/day (low-intensity), did not report on differences in clinical outcomes between patients who remained on low-intensity therapy (n=1,647) versus those who were titrated to 40 mg/day (n=1,657) (moderate-intensity therapy).⁵⁴ It also found no difference in risk of aminotransferase elevation more than 3 times the upper limit of normal (0.7% vs. 0.4%; RR, 1.64 [95% CI, 0.64 to 4.23]).

Indirect comparisons of trials of statins versus placebo or no statin stratified according to the intensity of therapy were also limited. For all-cause mortality, risk estimates were similar in trials of low-intensity (RR, 0.72 [95% CI, 0.52 to 1.00]; I²=0%; ARD, -0.55% [95% CI, -1.10 to 0.00]), moderate-intensity (RR, 0.88 [95% CI, 0.80 to 0.97]; I²=0%; ARD, -0.55% [95% CI, -0.97 to -0.13]) and high-intensity statins (RR, 0.80 [95% CI, 0.67 to 0.97]; I²=0%; ARD, -0.44% [95% CI, -0.70 to -0.18]). For other clinical outcomes, there were too few trials of low- and high-intensity statins for meaningful comparisons.

An analysis on the association between degree of lipid reduction achieved and clinical outcomes may also provide some indirect evidence about effects of statin therapy intensity.⁴⁸ Based on data from 22 trials of statins versus placebo or no statin (including some trials that included patients with prior cardiovascular events), the Cholesterol Treatment Trialists' Collaboration found that LDL-C reduction with a statin was associated with decreased risk of all-cause mortality (RR, 0.91 [95% CI, 0.88 to 0.93] per 36 mg/dL reduction in LDL-C) and a composite outcome of major cardiovascular events (nonfatal MI, CHD death, stroke, or coronary revascularization) (RR, 0.79 [95% CI, 0.77 to 0.81] per 36 mg/dL reduction in LDL-C). The estimate was similar when the analysis was restricted to participants without a history of vascular disease (RR, 0.75 [95% CI, 0.70 to 0.80]). Estimates were also consistent for specific cardiovascular outcomes (including major coronary events [nonfatal MI and CHD death], fatal or nonfatal stroke, and coronary revascularization).

Harms

Evidence on how harms of statin therapy vary according to statin potency is limited. JUPITER, the only study among those that reported diabetes incidence to evaluate high-intensity statin therapy (rosuvastatin 20 mg/day), reported a significantly increased risk of diabetes with statin use.^74,109 There was no increased risk of diabetes with moderate-intensity statin use in the ASCOT-LLA, HOPE-3, and WOSCOPS trials (atorvastatin 10 mg/day, rosuvastatin 10 mg/day, and pravastatin 40 mg/day, respectively) (RR, 0.96 [95% CI, 0.75 to 1.22]; I²=67%).^60,96,103 The MEGA trial, which used low-intensity statin therapy (pravastatin 10 to 20 mg/day),⁸³ and the AFCAPS/TexCAPS trial,⁵⁴ which used low- to moderate-intensity statin therapy (lovastatin 20 to 40 mg/day), also found no association between statin therapy and increased risk of diabetes.

Analysis of patient-level data from primary prevention trials found no association between the degree of LDL-C reduction and risk of cancer or cancer mortality.⁴⁸