[dredmorbius]

Extreme agers are rare enough that they're highly unlikely to show up in median statistics. If there are people living exceptionally long, they're still a small proportion. You might get some mileage from percentile analysis (95%ile, 99%ile), and there may be robust research methodology along those lines (I'm not tapped into that).

Which actually does leverage your concept somewhat (increase the sample size to include more members, and reduce extreme outliers). There is also plenty of analysis which does benefit by median (vs. mean or maximum) analysis, or by looking closely at observed distributions and outlier characteristics.