I find it's easier to think of your Box2D simulation as acting in a different coordinate system. So you might have boxes that are "1 box2d unit" wide, but then in your renderer that might correspond to boxes that are 3 "render units" wide, which is transformed by the sf::View to 40 pixels wide (say) depending on the zoom level.
But yes, you could have one coordinate system where 1 box2d unit corresponds directly to a sf::Rect of width 1, which would then be transformed by the sf::View to the screen coordinate system.