When looking at data, an otherwise healthy 20 year old, and an 80-something with multiple chronic and/or terminal Dx's, would stand out.
The challenge in looking at population-level data is that the Dx may well not be available, though other indicia might suggest this (e.g., residence is an assisted-living or long-term care facility). In practice, studies are frequently limited to available data, and proxies are frequently used. These may not be appropriate for specific individuals, but at scale, the law of large numbers and general correlation are your friends.