Developmental History

Reconstructing Childhood

DSM-5 requires that several ADHD symptoms were present before age 12. In an adult evaluation, that criterion cannot be observed — only reconstructed. Which makes it a problem of evidence quality, not a memory quiz.

Shafi Lodhi, MD · September 2026

DSM-5 requires that several ADHD symptoms were present before age 12. When the patient in front of you is thirty-five, that criterion cannot be observed. It can only be reconstructed, and reconstruction is an evidence problem. The practical answer is to stop asking whether the patient was inattentive as a child and start collecting scene-level material from named periods of their schooling; to revisit the same functional domain at two or three separate ages; to pursue collateral in a deliberate order — parent informant, contemporaneous school records, old report cards, siblings; and to treat contradictions as findings rather than noise. The standard you are working toward is a consistent longitudinal pattern supported by specific examples. Not certainty. Certainty is not available.

A criterion no one in the room can observe

Every other ADHD criterion is available to me during the evaluation. I can hear how a patient organizes a narrative. I can ask for last week’s concrete failures and see whether they arrive with detail or with adjectives.

The onset criterion is different in kind. It asks me to make a finding about a child I never met, in a classroom that no longer exists, on the basis of testimony from someone with an obvious stake in the outcome and a twenty-five-year gap in the record. DSM-5 moved the threshold from age seven to age twelve, partly in acknowledgment that adults asked to remember age seven are being asked for something memory does not reliably supply.

Forensic work trains you to see this clearly, and the method transfers cleanly to the clinic. When the only evidence is testimony about the distant past, you do not ask the witness whether the thing happened. You ask what they remember, in detail, from specified times and places, and then examine whether the details hold together.

“Were you like this as a child?” collects nothing

The question is leading, and it is answered before it is finished. The patient across from me has usually spent weeks reading about ADHD, has often been told by a friend or a sibling or an algorithm that this is what they have, and arrives with a narrative already assembled. Asking whether they were like this as a child invites them to ratify it. “I think so” is the answer I hear most often, and it tells me the least.

“I think so” is not a weak yes. It is an absence of evidence wearing the costume of one.

Retrospective report also has a directional pull. An adult who is failing now tends to remember a childhood that was already failing; present distress recruits congruent memories and dims the rest. This is not malingering. It is how autobiographical memory behaves under a mood state and a leading question.

Retrospective childhood rating scales inherit the flaw of symptom checklists generally: they convert a recall task into a recognition task. When the item supplies the content — as a child, did you often lose things necessary for tasks? — the patient does not have to generate anything from their own life. They only have to find the item plausible, and most adults can. I use those scales as a rough screen, never as the evidence for onset. The instrument comparison takes this up in more detail.

Anchoring to scenes

Recall improves when you stop asking about traits and start asking about places. The most productive question I have in this part of the interview is about homework at the kitchen table. Where did it happen, and who was in the room. How long did a twenty-minute worksheet take. Was a parent sitting beside you, for how many years, and what happened when they stopped. Did it end in tears — yours or theirs. Patients who cannot answer “were you inattentive” will give me forty-five minutes on the kitchen table.

Then report-card day — not the grades, the physical moment. Did you want to hand it over or hide it. What did the comment section say, and can you get close to the wording. The phrases that come back are stereotyped: not working to potential, capable but careless, talks too much, needs to slow down, doesn’t finish. A patient who spontaneously produces one of those has handed me something closer to a contemporaneous document than a memory — it was written by an adult who was watching them and had no interest in a diagnosis.

Then the backpack: the crumpled permission slip at the bottom, the library book paid for twice. Then the difference between third grade and seventh, which matters more than any other transition in the developmental history. One teacher and one room become six teachers, six syllabi, a locker, and a schedule you are expected to manage yourself. That is where external structure thins out, and where a great many real childhood impairments first become visible to the adults around them.

Temporal triangulation

One scene is an anecdote. The finding you need is a pattern, and a pattern requires the same domain examined at more than one age.

So I pick a domain — completing written work, keeping track of belongings, sitting through something boring — and I return to it at elementary school, at middle school, at high school, and then at the first job. Did the difficulty with written work exist at eight, or appear at fourteen? Was the disorganization present when a parent was still managing the backpack, or did it emerge when the parent stopped? Asking the same question across epochs is how you distinguish a neurodevelopmental trajectory from an adolescent-onset problem with a different explanation, and how you catch the patient whose symptoms all date from a single year.

Examiners intend to do this and then run out of interview. The LODHI-A organizes the developmental history by epoch and domain rather than by criterion, so the revisiting is structural.

Let the inconsistencies be data

A patient tells me elementary school was a catastrophe. Ten minutes later, walking through the same years by grade, he mentions he was in the accelerated math group and that no teacher ever raised a concern. That contradiction is one of the more useful things I will get all day, and smoothing it over is the wrong response.

It has two honest readings. The impairment was real and buffered by ability and structure, in which case the accelerated math group is evidence of the buffering rather than evidence against the impairment. Or the catastrophe is retrospective — a current failure written backward over a childhood that was, at the time, unremarkable. Both happen. You distinguish them by asking what the good years cost: hours, tears, a parent’s labor, an all-nighter at age eleven. Effortless performance and expensive performance look identical on a transcript.

Collateral, in order of value

First, a parent or whoever supervised homework, interviewed directly and asked about the same scenes the patient was asked about — not asked whether their child had ADHD. Parents minimize out of guilt. Parents with undiagnosed ADHD of their own have a shifted baseline and will tell you every child is like that, his father was the same — simultaneously a minimization and a family history.

Second, contemporaneous school records: report-card comments, attendance, disciplinary notes, any early intervention, IEP, or 504 documentation, any retained grade. These are dated, written by observers with no stake in the present question, and they outrank everybody’s memory including mine.

Third, the old report cards in the box at the parents’ house. Ask the patient to go look; they exist more often than you would expect. Fourth, siblings — an independent vantage on the same household, but same-generation memory, and family narratives leak between siblings in a way they do not leak from a 1997 report card.

When none of it exists — the parents have died, the district purged its files, the patient emigrated at nineteen — you do not manufacture a substitute. You write down what you have and what you do not, and you say plainly that the onset criterion rests on uncorroborated retrospective self-report. That sentence is not a failure of the evaluation. It is accuracy, and it matters when a controlled substance sits at the end of the decision.

Two histories that scored the same

A composite, drawn from many evaluations. Two patients, both thirty-four, both self-referred after a colleague was diagnosed, both endorsing nearly every item on the screening measures they arrived with.

The first, asked about homework, gives me the kitchen table from four in the afternoon until nine at night, her mother beside her, one worksheet, the same green folder lost and replaced three times in fifth grade. She remembers a sixth-grade comment — bright but scattered — and she remembers seventh grade as the year it fell apart, when one teacher became six. Her mother, interviewed separately, describes the same five-hour evenings unprompted.

The second tells me he was a normal kid, he thinks. Homework: he doesn’t really remember. Report cards: fine, he guesses. Teachers: nothing comes to mind. Nothing comes at eight, at twelve, or at sixteen. What does come, in detail, is the last eighteen months — a promotion into managing six people, an inbox he cannot hold, sleep that broke somewhere in there.

He may well have ADHD. But I cannot say so yet, and I tell him that, and I tell him what would let me say it. The difference between these two evaluations is not symptom count; their symptom counts were identical.

When the impairment was real and invisible

The late-recognized case deserves its own caution, because the reflex to treat a thin childhood history as a negative history will cause you to miss it. A bright child in a small, orderly classroom, with a mother functioning as an external executive system, may cross no one’s threshold for concern until the scaffolding is removed — freshman year of college, or the first job with no supervisor assigning the work. The correct formulation is not that the disorder began at nineteen, but that impairment became visible at nineteen, when the structure absorbing it was withdrawn. Evidence for the earlier years, when it can be found, is usually in the cost: the hours, the anxiety, the child who was tutoring herself. Girls with predominantly inattentive presentations are the commonest version of this story, and the methodology notes address that pattern specifically.

The honest limit

Memory is reconstructive. Every retrieval is an act of assembly, shaped by mood, by the narrative the person currently holds about their life, and by the questions I ask. That last clause is the uncomfortable one: my interview is itself one of the forces acting on the evidence I am collecting. I cannot make a thirty-five-year-old’s fourth grade available. I can prefer scenes to judgments, the patient’s words to my examples, a dated document to either — and state explicitly which of those tiers each conclusion is standing on.

What satisfies the onset criterion is not a confident yes. It is a longitudinal pattern of specific examples, drawn from more than one epoch and where possible more than one source, that holds together when you press on it — and a written record of how much weight that evidence can bear.

The developmental history section of the LODHI-A is the part I have rewritten most often, because it is where the evidence is thinnest and the stakes highest. It is free for licensed clinicians; you can request the instrument and use whatever portion of it is useful to you.

More articles · FAQ · Request the instrument