Don't risk that, there is every chance this person does work in a way your metrics aren't showing. Or at the very least that they can craft an explanation that makes sense like that. The most effective method is light touch here. For example, compare their stats directly to a similar coworker, and do not frame it as a workload comparison, show it incidentally as part of your investigation into how operational systems are. "We can see this user is responsible for the majority of systems outage reports, and they were only able to work 10-12 hours this week. The reported systems issue only affected this user, statistics show another 20 users were able to work 40 hours without issue using the same system. We will be working directly with this user to see if a system issue is somehow impacting only their ability to work." and provide that professionally, and without judgement. Then after several weeks where this same thing happens, and management keeps getting stats showing only one person is affected by "outages" the management will start to get the point. Additionally, include ticket or email metrics showing how often others communicate concerns compared to that end user.

We had some really troublesome people in call centers. People who generated 90% of tickets between only a few call center reps, out of hundreds. One person generated 40% of requests alone for a few months, and refused to accept the issue was their home network until their manager saw the truth and had them come in to the office to work. Do not attack. Do not openly judge. Just report data showing the truth.

And if you discover along the way that your assessment of "10-12 hours a week" was not entirely accurate, and most other users have similar logs, learn from that. It's very easy to get the metrics interpretation wrong when you're not watching them directly throughout a workday.