It’s 7:30 A.M. on a frosty Wednesday in December. I’m waiting impatiently in the lobby of the Anlyan Center, a large research building at Yale University. I’m waiting for a man named William Perry.* After two frustrating weeks of phone tag that involved his mother and her actual landline, William is coming in to be screened for my study of bipolar disorder.

I’m sipping my coffee when a man in jeans, two hoodies and a beanie walks through the door.

“Dr. Barron?” He extends an icy handshake. His warm smile makes me feel guilty for being grumpy.

We walk across the lobby’s curved limestone arcade, which at this hour is full of natural light. William’s sneakers shh-shh-shh across the granite, his feet never quite leaving the ground. In the elevator, I notice that he hasn’t shaved or showered in a few days. He stoops forward, resembling an upside-down J.

William isn’t my patient. He goes to a busy local clinic where he heard about my study. My goal this morning is to decide whether William has bipolar disorder according to the fifth edition of the Diagnostic and Statistical Manual (DSM-V), which is sold by the American Psychiatric Association (APA).

We take the elevator to the basement. In my right hand, I have a manila folder with the APA’s Structured Clinical Interview for the DSM-5 (SCID, pronounced “skid”). The SCID is a step-by-step series of questions that I’ll use to ask William about his past and current psychiatric symptoms. In theory, if I SCID (yes, also a verb), I’ll arrive at what is called a “reliable” DSM-5 diagnosis, which means that if William was SCID-ed by two other clinicians, all three of us would reach the same diagnosis.

Ironically, as useful as the SCID sounds, I’ve never used it in a clinical setting—and at no point in my residency was I trained to use it. Some of my attendings roll their eyes at it; one actually forbade me from using it during my practice diagnostic interviews. The psychoanalysts don’t like it because it ignores the “unique milieu of individuality.” The more biologically oriented clinicians don’t like it because it ignores biology. The DSM lacks “validity,” they say. A diagnosis based on a combination of symptoms is, they might argue, like a constellation of stars—sure, you could reliably identify the Big Dipper, but no one would argue that the Big Dipper is a valid interstellar system. It’s just a name.

But I have to SCID William if I want to run a rigorous clinical study. Otherwise, people might not know what I mean by “bipolar disorder.”

So, William and I make our way to the interview room, a white-walled, windowless nook with a desk and chairs. I tell him that the SCID is like a choose-your-own-adventure; how he answers will guide how I skip around in the packet. I don’t go into the details of reliability and validity but, with a grin, I warn him that I’ll read straight from the packet to make sure I do the test correctly. “Don’t worry, William,” I say, “it may be boring, but I’m happy to say that I’ve never lost a patient doing this exam.” He laughs and settles into his chair.

I’m glad he laughs. Patients who laugh are often doing well. And I like thinking I’m funny.


I turn to Overview and run my finger down the page to my first line.

“I’m going to be asking you about problems or difficulties you may have had,” I read, “and I’ll be making some notes as we go along. Do you have any questions before we begin?”


“How old are you?”


“With whom do you live?”

“With a roommate in subsidized housing. He’s not doing well.”

I nod, wondering why his mother was so involved in our phone tag if he doesn’t live with her. I scribble his answers on my packet. William has lived in the same government housing for years, but his roommate is relatively new and stays in his room chain-smoking cigarette after cigarette, all day, every day.

“Have you ever been a patient in a psychiatric hospital?”

“Oh yes, many times.”

My eyes scan below: IF YES: What was that for?

“What was that for? Could you please describe when, why and for how long?”

“The first time was when I was 20. Someone put something in my drink at a club, and I flipped out. I was at CVH [Connecticut Valley Hospital] for six or seven months.”

If an inadequate answer is given, challenge gently.

“What do you mean you flipped out?” I challenge gently.

“I started hearing and seeing things.”

“What sorts of things?”

“You know, just sounds and shadows.” I nod even though I don’t know. “Never really voices at that time. They started me on haloperidol and thorazine. Now, well, now I’m on a very low dose of risperidone,” he says proudly. I realize I’ve never had a patient be proud of being on a low dose of anything.

Haldol, thorazine, and risperidone are antipsychotics, medicines commonly given to patients with thought disorders like schizophrenia and, less commonly, to patients with mood disorders like bipolar disorder. I think how I wouldn’t be surprised if William had a handful of mental illnesses.

The co-occurrence of mental illnesses is well established. In 2005, a group of researchers at Harvard University reported that across 10,000 people, one in four met criteria for a single disorder, of these nearly half had two disorders, and over a quarter had more than three disorders. People with, say, PTSD, depression and a substance use disorder aren’t at all unusual.

That so many patients met criteria for multiple disorders astonished researchers (the paper has nearly 11,000 citations) and led to a large discussion over what we want a diagnosis to accomplish. Do we really want an infinite number of diagnoses that describe an infinite number of symptom combinations? If many diagnoses are required to describe your symptoms, they’re meaningless; if you have every disorder, you have no specific disorder.

Diagnosis is a messy business. In the 1940s, it became clear that psychiatric diagnoses were neither reliable nor valid. People received contradictory diagnoses by different clinicians—one study reported that clinicians only agreed on a diagnosis 20% of the time. The first effort to standardize diagnosis was by the military so it could reliably screen recruits. In fact, the first DSM was published in 1952 and was based on this military document; it included 106 diagnoses. In 1968, the APA released its second DSM, which included 182 diagnoses. But DSM-I and -II were largely ignored by the psychiatric community because the predominant (psychoanalytic) belief was that a person’s illness stemmed from his or her unique unconscious conflicts, which couldn’t be packaged into unwieldy diagnostic boxes.

So, in 1980, DSM-III made the boxes smaller by offering 265 diagnoses. DSM-III also provided a different approach to diagnosing. Instead of diagnosing illnesses based on their cause (something the science simply wasn’t mature enough to sort out), DSM-III provided a checklist of symptoms that needed to be present to justify a diagnosis. For people to have schizophrenia, they had to have a specific constellation of symptoms common in other patients with schizophrenia. DSM-IV and -V (each of which has approximately 300 diagnoses) carry on the assumption that reliable diagnoses can be made by tallying up symptoms. Another assumption is that these diagnoses are valid.

William had been hospitalized at least six times (he couldn’t remember precisely), always with hallucinations. Throughout his life, William counted a total of 35 months spent in a psychiatric hospital, with an average stay of just under six months. William was a very sick man.

“Have you ever tried to kill yourself?”


IF YES: “What did you do?”

“I slit my wrist when I was in jail for beating my father in the head with a hammer.”

“Excuse me?” I clear my throat. We were off script.

“I hit my father in the head with a hammer after he was talking crazy in front of my girlfriend. But I just knocked him out.”

“Just knocked him out?” I ask curiously, even though this was not the time to ask about violent behavior.

“Yeah, he didn’t die or anything.” William’s deadpan voice betrays his boredom telling this story.

“Oh, good.” The corners of my mouth twitch upwards. I look at the door.

“In the past month, have you been using any illegal or recreational drugs?”

He hasn’t, but he first smoked marijuana when he was 10 years old. He first used crack when he was 14. He’d been sober from marijuana, crack and alcohol for the last three years.

“I only smoke about half a pack a day now,” he tells me, sitting up in his chair with another warm smile.

“That’s wonderful, I’m sure that was a lot of work,” I say, winging it.


A1: “In the past month, since November, has there been a period of time when you were feeling depressed or down most of the day, nearly every day?”


“What has it been like?”

“I’ve felt really sad lately. I thought I had a friend. You know, a friend?” I nod. “But then he choked me and stole my cell phone. I’ve felt kind of, well, kind of lost since then. That was three weeks ago.”

“Oh, god, I’m sorry to hear that.” I recall our phone tag. “How long had you been friends?”

“Five years.”

“Do you have any other friends?”

“No. I don’t think my roommate is my friend.” I pause for a moment then return to my script.

A7: “Have you had trouble thinking or concentrating?”

“The whole thing shocked me, so I guess so.” I write down shocked me and wonder if I’d feel “shocked” if my friend strangled me. I don’t know.

A10: At least five of the above criterion A symptoms (A1-A9) are rated “+.” I circle “NO” because William answered “No” to A2, A3, A4, A5, A6, A8 and A9.

Continue with A15 (Past Major Depressive Episode), page 13. I flip forward and William describes multiple past depressive episodes.

A29: “In the past month, since November, has there been a period of time when you were feeling so good, ‘high’, excited or ‘on top of the world’ that other people thought you were not your normal self?”

“No.” He’d felt depressed since his friend strangled him.

A54: “Have you ever had a period of time when you were feeling so good, ‘high’, excited, or ‘on top of the world’ that other people thought you were not your normal self?”

William nodded. “Cocaine.”

“What was it like?”

“The greatest feeling of my life. Definitely high on life.” His eyes widen and he raises his fists victoriously, grinning.

A55: “How long did this last?”

“Four or so days.”

Four days is less than one week, so I Continue with A66 (Past Hypomanic Episode), page 26.

A66: “Did that period when you were ‘high on life’ last for at least 4 days?”

William looks at me confused since he just answered that question. “Four or so days.”

We work our way through the next bundle of questions, all of which he answers “Yes.”

Even though William had all the symptoms of mania, because his symptoms didn’t last for over one week, this was technically hypomania. And because he was also on drugs, he wasn’t technically even hypomanic. He was just on drugs.

IF NOT ALREADY ASKED: “Has there been any other time when you were ‘high on life’ and you weren’t using those drugs?”


I shift in my chair and circle “YES” three times on my packet. Finally, some evidence of bipolar disorder.

IF YES: Go back to A66, page 26, and ask about that episode. I skip A66 since I already know the answer.

A67: “During that time, how did you feel about yourself?”

“I was on top of the world.”

Did you sleep less than usual? Yes. Were you more talkative than usual? Yes. Thoughts racing through your head? Yes. Easily distracted? Yes. How did you spend your time? Just busy. Did other people notice the change in you? My mom did. Oh, and my girlfriend told me I wasn’t right in the head.

I wonder if this was the girlfriend who saw William beat his father with a hammer. I circle YES then read: PAST HYPOMANIC EPISODE Continue with B1 (Psychotic Symptoms), page 31.


B1: “Has it ever seemed like people were talking about you or taking special note of you?”


Did you ever have the feeling that something on the radio, TV, or in a movie was meant especially for you? Yes. What about anyone going out of their way to give you a hard time, or trying to hurt you? Yes. Did you ever feel like you were being followed, spied on, manipulated, or plotted against? Yes, I often feel like people are watching me through the TV.

B3: “Have you ever thought that you were especially important in some way, or that you had special powers or knowledge?”

“Yes, I can see the stars through the ceiling in my bedroom.”

Paranoias? Religious delusions? People putting thoughts into your head? Unusual sensations? Yes, yes, yes, and yes.

“The other day before it rained, I saw rain even though the sun was out. Things like that used to bother me,” he says with a comforting, paternal smile, “but then I finally accepted that God is mysterious. It doesn’t bother me anymore when I see the rain.”

He often hears voices. This bothered him until he read in the Bible that God speaks to people, so now the voices are a gift from God. Throughout our conversation, William is calm and pleasant.

Continue with C1 (Differential Diagnosis of Psychotic Disorders), page 37.

As I flip the pages, I realize that William’s diagnosis of bipolar disorder is being buried in the SCID’s algorithm. In my mental tally, he has symptoms of bipolar disorder, schizophrenia, and substance use disorder—perhaps even PTSD. William told me he’d been diagnosed with bipolar disorder and, evidently, the medications he’s taking are working. So, does it really matter if the SCID tells me William doesn’t have bipolar disorder?

Last year, psychologists Avshalom Caspi and Terrie Moffitt, a husband and wife research team at Duke University, addressed “the crisis of confidence” in the way that mental illnesses are classified and diagnosed. Instead of splitting patients based on their unique combination of symptoms, Caspi & Moffit looked for patterns across the many combinations of symptoms that about 1,000 people in their longitudinal study developed over nearly forty years.

The “p factor” is kind of like an S&P 500 for psychiatry; it sums up someone’s propensity to develop a mental illness across multiple domains. The higher someone’s “p” score, the worse they fare from childhood to adulthood.

Even without calculating William’s “p,” I wager his score is very high. But this doesn’t tell me anything about his specific illness or what medications he would benefit from. Or whether he can participate in my study.


Section C is answered mostly behind the scenes, based on previous answers William has given. Because William has psychotic symptoms even when he isn’t depressed or hypomanic, I move to C2. Because William has delusions and hallucinations, I move to C3. Because William is only depressed or hypomanic when he is also psychotic, I flip forward a couple pages to C9, Criteria for Schizoaffective Disorder.

I confirm that every time William was depressed or hypomanic he was also psychotic and move to C10.

C10: “Thinking about your whole life from the time you first became ill until now, has there been any time when you had delusions or hallucinations when you were not depressed or feeling ‘high on life’?”


C11: “How often would you say you have these delusions and hallucinations along with feelings of depression or being ‘high on life’?”

“Oh, I don’t know, more often than not.” I assume this means more than 50 percent of the time and circle YES.

C12: “Just before these episodes began, were you using any street drugs?”

“Well, yes. Except for the last three years, I’ve been using drugs since I was 10 years old.”

Diagnose: Substance-Induced Psychotic Disorder.

I sigh. Not bipolar disorder.

We wish each other a warm holiday. I walk William to the exit and thank him again for his time. “I hope the wind doesn’t blow you away!” I say, happy to be funny again. William chuckles and shakes my hand and returns to the cold.

As I walk to my desk, I wonder exactly why William didn’t have bipolar disorder. Did hallucinations make his mood less “bipolar”? Or perhaps, was he less schizophrenic because of his mood swings? And does the fact that he used drugs most of his life diminish his bipolar or schizophrenia-ness? My field seems to think so. But if you eat too many cheeseburgers and one day have a heart attack, you don’t have a “cheeseburger-induced heart attack.”

The SCID gave me the feeling of precision and certainty and order. I enjoyed it and I think William did as well; it felt like were doing something useful. For about an hour, I’d used the SCID to trace William’s mental life as I might use a nautical chart to confidently trace a coastline. Yet symptoms don’t reflect the brain in the way that a curved line reflects a harbor. And as I never measured how ill he was, I was unable to plot which treatments had brought William closer to or farther from a sane harbor.

I stare at the manila folder on my desk and long for numbers. What I needed was not names but measurements; fewer constellations, more GPS.

* Weeks after Mr. Perry’s involvement in my study ended, he and I met to discuss this article. After I explained the topic of this article, he voluntarily gave me written consent to use his name and describe his clinical history. He now has a cell phone.