Modern roundabouts are the safest form of intersection in the world - the intersection type with the lowest risk of fatal or serious injury crashes - (much more so than comparable signals). Modern roundabouts require a change in speed and alter the geometry of one of the most dangerous parts of the system - intersections.
A 2017 roundabout study out of Minnesota found that the dual lane roundabouts saw a 6% increase in injury crashes and a 212.5% increase in PDO crashes. If we use FHWA’s comprehensive crash costs by injury severity level to perform a cost/benefit analysis, we see that these dual-lane Minnesota roundabouts increased crash costs by $1,664,300 over the comparable intersections they replaced. Even though there were 3 less A-level crashes at the roundabouts, there were 7 more C-level crashes and a whopping 270 more PDO crashes.
http://www.dot.state.mn.us/trafficeng/safety/docs/roundaboutstudy.pdf BEFORE ROUNDABOUT CRASH COSTS = $4,882,600Fatality (K) 0 X $4,008,900 = $0
Disabling Injury (A) 3 X $216,000 = $648,000
Evident Injury (B) 15 X $79,000 = $1,185,000
Possible Injury (C) 46 X $44,900 = $2,065,400
PDO (O) 133 X $7,400 = $984,200
AFTER ROUNDABOUT CRASH COSTS = $6,546,900Fatality (K) 0 X $4,008,900 = $0
Disabling Injury (A) 0 X $216,000 = $0
Evident Injury (B) 15 X $79,000 = $1,185,000
Possible Injury (C) 53 X $44,900 = $2,379,700
PDO (O) 403 X $7,400 = $2,982,200
Instead of constructing complex multi-lane roundabouts that are prone to crashes (and increases in injury crashes) maybe agencies should consider innovative signal designs that eliminate direct left turns. Median U-turns have been implemented in Michigan for over 50 years and have proven safety benefits with the FHWA estimating that they reduce injury crashes by 30%
https://safety.fhwa.dot.gov/intersection/alter_design/pdf/fhwasa14069_mut_infoguide.pdf. A 30% decrease in injury crashes sounds pretty good compared to a 6% increase in injuries these complex roundabouts are seeing.