I don't think anyone argues that in specific incidents helmets are safer.
The question is whether or not the second order effects of compulsory helmets outweigh the benefits for the individual, namely;
- helmets make biking more inconvenient and compulsory usage of helmets suppresses cycling rates. In addition to the argument that is is a net negative because people get less exercise, there is also a second argument; because cyclists are less common, drivers are less familiar with how to interact with them, which leads to poorer judgement and more accidents per capita for the people still cycling. (In some cases, unfamiliarity also breeds anger and resentment; many a cyclist has a story of getting run off the road, drivers rolling coal at them, etc.)
- drivers are generally overestimating confidence and safety when it comes to cycling. They may generally act more dangerously around people who look like they know what they're doing (e.g. the stereotype of the Lycra-clad road biker), which leads to more and deadlier accidents.
As far as your sister's accident goes, there are two modes of thinking about it. One is to create more regulation, which generally tends to encumber the more vulnerable user (e.g. mandatory helmets), and depends on pretty much constant, vigilant enforcment; the other is to physically engineer a road environment where a high-speed intersection collision is less likely, if at all possible. (This may involve things like a tight roundabout to force slow speeds and paying attention; raising the crosswalks or even the entire intersection; narrowing the road and installing chicanes so that one cannot speed in a straight line down a road; or creating a shared space by removing all curbs and road markings so that one must pay attention because the environment is so uncertain and confusing.)