In this opening talk at the TIPS 2017 conference, Alik Widge frames why psychiatry urgently needs a new, quantitative science of diagnostics. He outlines three fundamental reasons: (1) current treatments underperform because the front‑end diagnostic process is poor—DSM‑5 categories are highly overlapping and heterogeneous, as shown by field‑trial data; (2) patients are unreliable reporters of their own symptoms, illustrated with the Green Eggs and Ham exposure‑therapy analogy and the contrast between limited clinic visits and real‑world behavior observed by caregivers; (3) objective, data‑driven outcome measures are essential for value‑based healthcare, enabling clinicians to demonstrate treatment efficacy to insurers in a manner analogous to diabetes A1C metrics. Widge critiques the traditional approach of grouping patients by diagnostic labels and then seeking statistical signals in MRI/EEG studies, arguing that objective, quantified measures and data‑driven clustering are needed to move beyond the current “overlap mishmash.” He concludes that proving value through measurable outcomes will be transformative for mental‑health reimbursement and care quality.
Chapters
So I am going to kick this off with about 5, 10 minutes of quick stage setting, and then turn it over to the panel. We're going to have each of them talk for about 10, 15 minutes, and then use the remainder of the time for what I hope will be a lively discussion, and what I intend to provoke into a lively discussion, if at all possible. So I wanted to start by just framing, and really, in many ways, given what we just heard for the past hour, reframing this question. Okay, why do we need this new science of diagnostics anyway? And most of you in this room know this, but I think it's important for us to get this down, operationalize it so we can share it more clearly, and that there are 3 fundamental reasons. One, because the reason our treatments don't work well enough is because we're not doing well enough on the front end of figuring out what's wrong in the first place. Yeah. Two, because patients themselves are terrible at telling us anything. And then number three, something I don't think I've heard yet brought up today, but it's going to become critical, because this is the pathway towards value-based healthcare in psychiatry. And so we all know this. We've heard this many times even throughout this day, that the fact is, depending on how you slice the data, the majority of people who present for mental health treatment will ultimately not get the relief they're really seeking, will be left with substantial life-impairing symptoms. And that this is in part because these disorders, although they are brain disorders, they are disorders of the brain in way— in subtle ways of information flow that doesn't show up easily on an MRI scan, that has not been simple for us to track or classify We are still building the circuit neuroscience of psychiatry that will let us get those treatments. But part of the reason it's taken us so long is this. And this is a slide from the field trials of DSM-5, the Diagnostic and Statistical Manual, a.k.a. the Bible of psychiatry. And aside from reminding you of NBC's Peacock logo, what it should point out to you is two things. One— The majority of patients with a given disorder have multiple disorders, and number 2, there's this huge cone down at the bottom— I'm not even gonna bother with the pointer, you can see it— that is just this overlapping mishmush that you ask 3 psychiatrists and you'll get 4 different answers of what the patients have. And the way we've been doing our neuroscience— hopefully Tom is also gonna pound on this point when he gives his talk— is to take people with one of these categories, stick 'em in an MRI scanner or an EEG cap on them, and hope that we get a statistically significant result that hasn't been working. The kinds of objective, quantified ta— measures that we're talking about today, and especially the ability to identify these natural clusters that emerge from them when you've got enough data, are what's going to get us away from this to something separable. So that's thing one. Thing two, even bigger, and I think this was really highlighted in, for example, some of the previous slides about detecting relapse. Patients don't actually know that well. With all respect to what was said about listen to the end users, that is true, but one of the first things you learn in psychiatry is that patients are actually terrible at telling you what's going on with them and what they actually need to get better. And honestly, I'm a parent. I think about a lot of these things through the lens of children's literature. So for anybody who has not yet been exposed to Green Eggs and Ham, the fundamental point is an entire book of a gentleman repeatedly protesting that he does not like green eggs and ham and that please would you go away and stop offering them to him, right up until he tries them, at which point he likes them. And anybody who's ever tried to do exposure therapy with someone with an anxiety disorder knows exactly what I'm talking about. And the repeated protestations that "This will never help me," right up until it gets them better. This is the nature of psychiatric disorders, and closely tied with it Is that— we see our patients, if we're lucky, we get to see them an hour a week. That's a terrible slice. So now this is my own children. On your left is what their daycare providers, who see them a few hours a day, would tell me that they are, namely sweet innocent angels who play with each other so well. And on the right is what actually happens, which is my older daughter using her younger sibling as a crash test dummy for her wagon. So if you actually can get to where the patient lives and what they're doing when they're not in your office, you'll get a very different picture of what's actually happening, and we might find out that we don't know our patients nearly as well as we think we do. And then finally, and this is one that I hope we'll talk about during the discussion, outcomes. Outcomes. And value. Because psychiatry is late to the game in this, but healthcare in general in America has been going this way for a decade, and that train is not going to stop barreling down on us. That more and more we're seeing capitated payment models, we're seeing bundled payments, we're seeing everybody who's got the dollars that actually pay for clinical care saying one thing. I want the best results for patients. I want the biggest numerator, and I want the smallest cost for it. I want the smallest denominator. Now, there are limits to how small you can make that denominator. You actually do have to pay people, especially in Boston. But boy, we could sure do a lot better on the numerator, and especially we could do a better way of proving we're actually getting somebody better beyond just saying— Yeah. Trust me, the technology— so the technologies we are talking about today are exactly what's going to support being able to go to insurer and do the same thing that currently my endocrinologist colleagues do with their diabetes patients, where they can say, hey, look, I manage A1Cs better than 90% of the people in the Boston area. You should be paying me 20% more than the rest of them. If we can do that, if we can actually demonstrate the value of mental health care, Just that. Just demonstrating that what we do works. That alone is transformative. Forget all the stuff that we're going to do about creating new treatments that work even better.
