I read different conclusions from the study. "Bicycle-specific infrastructure was associated with reduced injury risk at non-intersection locations. Cycle tracks (physically separated lanes alongside major streets ... were especially protective".
At intersections, "bike or pedestrian infrastructure was not associated with increased or decreased injury risk at intersections and was therefore not included in this model."
The conclusions I saw in the study were that traffic speeds below 18 kph reduced risk for cyclists, and places with more cycling traffic had increased risk. Traffic circles and riding against traffic increased risk. None of which was really surprising.