Part 1 was about what an adult ADHD evaluation has to establish. Part 2 was about running that as a process that does not eat your week. This one starts a step later, at the moment the evaluation is finished, the diagnosis holds, and you are about to write the first prescription.
Almost everything that goes wrong from here is an expectations problem or a boundary problem. Very little of it is a pharmacology problem. The patient who is disappointed at week three, the one who runs out early, the one who arrives already certain that only one specific product works, the one you are quietly dreading on your Thursday schedule: none of those situations are improved by knowing more about dopamine transporters. They are improved by having said the right things before you wrote the prescription, and by having a clear idea of what you will do when something looks wrong.
So this post is deliberately not about the drugs. Part 4 is the medication guide, and it will cover agents, dosing, duration, side effects, and the non-stimulant options. This one is about the part that surrounds the prescribing and that nobody teaches.
Set the target before you set the dose
The goal of treatment is functional improvement. Not a lower score on a rating scale, and not a patient who reports feeling sharper. Whether the person's life works better than it did. That is obvious enough written down, and it is also the thing that quietly stops being true about four months in, once the conversation has drifted into how the dose feels.
So before the first prescription, get specific about what the ADHD is actually costing this person right now.
Not "focus," and not "feel more normal." Something you could observe. She is taking two hours of charting home every night. He has dropped the same class twice because he cannot sit through a fifty minute lecture. The bills get paid in the week the late notice arrives rather than the week the bill does. That level of concrete.
This does two things. It gives you a treatment anchor, and it gives you a way to evaluate whether the medication is actually working or just being taken. Those are not the same question, and if you never named a target you will not be able to tell them apart. "Is it working?" quietly collapses into "do you like how it feels?", which is the single worst question to have built your monitoring around, and the one that makes every subsequent conversation about dose harder than it needs to be.
So how do you measure that. Mostly by deciding in advance what you are going to count, and then counting the same thing the same way every time.
A few things make it work. Pick one target, or two at the most, because a list of six is a list you will quietly stop asking about by the third visit. State it with a number attached, even a crude one: nights per week the charting comes home, how many of the last five lectures he actually sat through, whether the bills went out before the due date last month. Get that number before the first prescription, because "better than before" measured against a memory is not a measurement, and the memory is going to be colored by how the person feels on the day you ask. Then ask the same question the same way at every visit, rather than opening with "how are the meds going," which invites exactly the answer you are trying to avoid. And write the target in the note, in the patient's own words. It costs you one sentence, and it is the only durable record of what the two of you agreed to watch. Without it you are reconstructing the target from memory six months later, which is the same problem the baseline was there to solve.
A word on scales, because this is where people reach for the wrong instrument. The ADHD rating scales most of us have on hand measure symptom frequency, not function. Readministering an ASRS tells you whether someone is endorsing fewer symptoms this month, which is a related question and not the one this section is about. Symptoms can improve while the charting still comes home every night. If you want a scale here it has to be a functional one, and those exist: the WFIRS-S asks about family, work, school, life skills, self-concept, social functioning and risk rather than how often a person loses their keys. It is in the TBP assessment suite already. Used as a repeated measure at a consistent interval it tracks the thing you actually named as the target. It is still not proof of anything, and it does not replace asking.
Collateral is worth a mention here too. The people around the patient frequently notice functional change before the patient does, and what they notice is not always what the patient is reporting. If you already obtained collateral during the evaluation, one follow up question at three months costs you almost nothing.
Then the negative version, because this is where the drift actually happens.
The dose is not the measure. It is easy to start treating titration as though it were progress, so that moving from twenty milligrams to thirty feels like the treatment advancing. It is not. That is a change in what you put in, and it tells you nothing about what came out. The same error runs the other direction: a patient sitting comfortably at a low dose is not undertreated because the number is small. What is on the prescription tells you what you gave. It does not tell you whether anything got better.
Neither is it enough that the patient likes the medication and wants to keep taking it. That is worth knowing, and it is real information about tolerability, adherence, and whether they trust you. But people are entirely capable of wanting to continue something that makes them feel good and changes nothing, and that is precisely the pattern you set the target up to catch. It runs the other way too. Someone can be markedly better at the thing you were both worried about and still be fairly lukewarm about the medication itself.
Both of those things are useful. Neither one answers the question you set out to answer, which is whether the charting is still coming home.
What the medication does, and what it does not
Say this before the first prescription rather than after the first disappointment.
A stimulant lowers the cost of starting a task and of staying on it. It does not supply the task, the system, the motivation, or the reason. Patients who arrive expecting the medication to make them want to do the thing will be disappointed on a dose that is working perfectly well, and that disappointment often gets reported to you as "it isn't working."
It also does not build the systems. It lets someone use the ones they already have. Most adults arriving for an ADHD evaluation have accumulated a long list of things they know would help: the calendar, the alarms, the lists, the one friend who texts them about deadlines. The problem was never that they did not know. It was that they could not reliably execute any of it. Treatment often looks less like new capability than like existing capability finally becoming usable, which is a smaller and more accurate thing to promise.
And it does not undo years of avoidance. If someone has spent two decades steering around the thing they are bad at, the medication does not make them good at it. It makes the attempt survivable. They still have to make the attempt, and that part is work. Better to say so before they start than at month three when they are disappointed.
And then the honest part. Partial improvement is a good outcome. That is worth saying out loud, especially with adults who have waited a long time for an answer and have built the diagnosis into a fairly large story about their life. Someone arriving with the expectation that this is the thing that finally fixes it is set up to read a real, useful, meaningful response as a failure.
The first month is your noisiest data
I try not to read too much into week one.
The early period is contaminated by everything: novelty, relief at being believed, the genuine effect of finally having an explanation for thirty years of difficulty. Patients frequently describe the first few days in terms that no medication is going to sustain. That is not a warning sign. It is just what the first week is.
Which sets up the conversation you will have around week three or four, when some patients report that it stopped working. A great deal of the time, what stopped is the subjective feeling. The noticeable, felt effect can fade with time, and that is not the same thing as loss of therapeutic benefit. That can happen while real functional benefit is still there.
This is the point where having named a functional target pays for itself. The question is not "does it still feel the way it did on day two." The question is "are you still getting the charting done before you leave." If the answer is yes, you have a conversation about expectations. If the answer is no, you have a clinical question, and Part 4 is where I will take that apart.
What a red flag actually is
A red flag is a reason to look closer, not a diagnosis of misuse.
I want to be blunt about why the phrase needs that qualifier. In practice it collapses into a verdict fast. Someone runs out four days early, a note goes in the chart about concern for misuse, and from that point on every interaction with that patient is read through that lens by you and by everyone who reads your chart afterward. Meanwhile the actual explanation was a pharmacy that filled the wrong quantity, or a dose increase you made mid month without moving the fill date, or a patient who is disorganized, which is the condition you diagnosed them with.
That last one deserves more attention than it usually gets. Several of the classic stimulant red flags are also symptoms of the disorder being treated. Losing the bottle. Forgetting the appointment. Poor tracking of a small countable object across a thirty day interval. We should be slower than we are to treat those as character evidence in a population selected for exactly that difficulty.
So here is how I try to hold the common flags, with the boring explanation first, because the boring explanation is usually the right one.
Two things separate a flag that needs a conversation from a flag that needs a change in plan. The first is pattern. One event with a coherent account is an event. The same event three times with a different account each time is a pattern, and the pattern is the finding, not the individual event. The second is the quality of the account itself. How someone explains it, and whether the explanation holds when you ask a follow up question, tells you more than the original event did.
And context earns weight. Someone with ten years of stability, a documented history, and one odd result is not in the same position as a new patient with one odd result. One unexplained lab does not override everything you already know about a person. It means you need the conversation before you proceed.