I disagree with the premise the question.
Some of the most popular casual games in the world use the classic points systems that have been used since the 90s. Angry Birds, Plants vs Zombies, Bejeweled, Subway Surfer are played by millions of players worldwide and they all use a classical points system. Also, many users regularly share their scores on social networking platforms and compare/compete with others. So, points systems are not on their way out as far as the non-hardcore gaming audience is concerned.
Secondly, game systems are far less prone to change than gamified systems. Stack Exchange gets updates and changes every few weeks. Games do not undergo significant changes for months or years. This means that points as a metric serves a very different role in gamified systems than in games.
In gamified systems, points can be a metric for various things ranging from participation, skill, engagement, quality of participation, social status, connections and more (discussed in Gamification By Design, Chapter 3). The same cannot be said for conventional modern games.
Third, one of the reasons for the success of gamified systems at present is the familiarity with game mechanics that the 80s and 90s generation had. To be specific, points systems, since they were the one consistent game mechanic in that era of games along with levels. This familiarity is what makes adoption of gamified systems easy for most audiences, and hence, they serve as more than just a performance metric.
Have their been any studies regarding the efficacies of points in a
gamified system or their performance against any alternatives?
While there are implementations of systems that do not use points, there does not seem to be any research on the effectiveness of points systems as a whole.