Behavioral HealthHealthcare Data

Behavioral Vital Signs: Reading the Data Before the Crisis

By Bibhas Singla, MD, MBA, Chief Medical Officer, Sanford Behavioral Health

Every crisis I have seen in behavioral health had a prologue. The suicide attempt, the psychotic break, the emergency department visit at two in the morning – none of these arrive out of nowhere. They are preceded by weeks of small, unnoticed, and unaddressed changes. The misfortune of behavioral health is that these important signals are visible, recorded, time-stamped, and yet sitting in unmonitored databases.

Behavioral health remains the last major specialty practicing snapshot medicine. A cardiac patient gets continuous telemetry; a psychiatric patient gets a questionnaire every few months if they show up to an appointment, and an instruction to call if things get worse. We intervene when a symptomatic threshold is crossed, and then we wonder why outcomes lag behind. Meanwhile, the industry’s reflexive answer has been more data: wearables, apps, portals. Lack of data has never been the problem. Data without analytics is inert, and analytics without a clear owner for the response is just administrative burden and legal risk.

The earliest warning signals we need are not exotic. They do not require a new sensor or a new app. They are already flowing through the ordinary operational systems every health organization and patients are using – the prescription fill data, the scheduling systems, and the claims data. The behavioral vital signs already exist. What is missing is the tooling that continuously analyzes them and alerts a case manager, who then reaches out to the patient and addresses their needs before it becomes a crisis.

Let’s start with the prescription fill data. When a patient on an antidepressant or antipsychotic misses a refill, it is a clinical event that is often the first step toward psychiatric decompensation. Research has consistently shown that a gap of about two weeks in medication supply identifies most patients on their way to discontinuing treatment, and that longer gaps in antipsychotic therapy multiply the risk of psychiatric hospitalization several times over. It requires nothing at all from the patient. It requires a predictive analytics tool that flags a change in patterns, alerts a team member, and that alert triggers a response from the frontline team.

A refill that never happens is a vital sign. The question is whether anyone is responding to it.

The second vital sign hides in plain sight: the visit that is not about mental health at all. Before a behavioral crisis, patients often show up more, not less, but often not at their psychiatry clinic. Sleep complaints, anxiety-related chest pain, gastrointestinal symptoms, and a third urgent care visit in six weeks. All these visits are recorded as claims data, but with no one connecting the dots and triggering an intervention. The declining mood and anxiety behind these visits go unaddressed, leading to a crisis in the next few weeks.

The opposite matters just as much: the missed appointment. A no-show is usually treated as a scheduling annoyance to be backfilled. In behavioral health, it indeed is data. Rising utilization for somatic complaints, or a quiet pattern of cancellations, is precisely the kind of shift that should route a patient to a case manager for a simple, human check-in long before anything appears on a problem list.

Now imagine these vital signs connected and analyzed together. A care manager can review a handful of charts in a day; an analytic engine can scan millions of data points continuously. Connect the three vital signs above, add dozens more, and anchor the models in validated studies that quantify the probability of an outcome, and risk stops being an impression formed at an appointment. Instead, it becomes a number refreshed every night, for every patient in a population at once.

This is not hypothetical. The Veterans Health Administration (VHA) already scans routine records every month to flag its highest-risk patients, and every flag lands with a named coordinator whose job is to reach out.

So why is this not everywhere? Because no single entity holds the whole picture. Payers hold the claims and pharmacy data, health systems hold the clinical record, and EHR vendors control the workflow where alerts must appear, while privacy protections that surround mental health records slow the sharing.

That fragmentation is not a dead end though, and practical steps exist today. The simplest starting point is the data an organization already owns outright (eg. scheduling, utilization, portal activity) which requires no sharing agreement at all. Value-based contracts can provide the financial argument, while collaborative care arrangements and modern interoperability standards provide legal channels. The strategy is sequencing: start inside your own walls, prove the response works, then extend.

And all of this is only the edge of the frontier. Refill gaps, somatic visits, and no-shows are the crude, first-generation vital signs that will one day look primitive. Language-based markers, digital phenotyping, and continuous measurement are coming quickly. But we do not need to wait for any of it. The first generation of behavioral vital signs is already sitting in the systems we control. The question every behavioral health leader should ask this year is simple: which signals are we already collecting, how are those analyzed, and who is assigned to respond?