Table 5.

Change in risk of stroke (relative odds reduction) for 1.0 mmol/l reduction in LDL cholesterol in randomised trials and estimates from analysis of cohort studies (fig 2) according to whether trial participants had known vascular disease on entry and whether stroke was thromboembolic or haemorrhagic and fatal or non-fatal

	Randomised trials
Category	No of trials	No of events	Estimated % change in risk (95% CI)	Estimated % change in risk in cohort studies†(95% CI)
All stroke	41	3319	−20* (−14 to −26)	—
All stroke in people with known vascular disease	32†	2311	−22* (−28 to −16)	—
All stroke in people without known vascular disease	7†	752	−6 (−22 to 14)	−6‡ (−12 to 1)
Thromboembolic stroke	8	1204	−28* (−35 to −20)	−15* (−21 to −6)
Haemorrhagic stroke	8	149	−3 (−35 to 47)	19* (10 to 29)
Fatal stroke	56	678	−2 (−17 to 16)	0‡§ (−6 to 6)
Non-fatal stroke	40	2519	−23* (−29 to −16)	−7‡ (−13 to −1)

^*P<0.001.

^†Trials in which there were similar numbers of strokes in patients with and without vascular disease on entry are omitted, except one in which separate numbers were available.^w168

^‡From data in figure 2 (taking account where necessary of proportions of non-fatal and all strokes that are thromboembolic and haemorrhagic).

^§Meta-analysis of 45 cohort studies¹¹ (in which nearly all strokes recorded were fatal) gave similar result (2, −1 to 6).