When I was a TA for a large artificial intelligence class, we frequently had this complaint from students. The reason was that, due to the nature of the class, many of the problems could be worked out eventually from first principles (or from the book or notes, since these were often open book as well) by somebody who didn't understand the material, but could only be worked out efficiently by a person who knew the material. We thus tended to end up with most students finishing well before allotted time, but a noticeably minority who did very bad work right up to the end, and felt "if they'd just had another hour or two" they could have done everything right.
Our response was to explain that the test was looking not just for the ability to solve the problems, but for facility with the problem-solving techniques. Thus, speed really did count! This was also borne out by the distributions: it really wasn't a problem for the stronger students.
I would thus recommend first looking at the distribution of the grades: if the strong students mostly finished early, then you probably have a weak student who was trying to work things out from first principles, hazy recollection, or source material (if you're open book). You can then explain something similar, that you are testing not just the ability to solve problems, but facility with the concepts, and if they are weak with the concepts, it is likely to show up in slow test-taking time.
Now, it is also possible that the student has a medical impairment, such as dyslexia, that genuinely means they cannot work quickly. If this is the case, then most universities have services that can help to evaluate them and plan appropriate accommodation, which can be communicated to you for application to future exams by the responsible personnel. Don't just take a Doctor's note, because an M.D. can only detect the existence of a condition, not calibrate an appropriate educational adjustment for it.