Behavioral Health and Primary Care Integration Instruments: The Next Generation

Gerald Hurowitz, MD
6 min readAug 10, 2021

After years of steady shoegazing, Behavioral Health (BH) Integration is finally on a heads-up march. What happened to bring about this development? The idea that improving BH and integrating its treatment into Primary Care (PC) has been around for some time. Why the sudden forward movement, and is it now just a matter of time before all of us — patients, providers and payers — reap the benefits?

Some analysts see the advent of Electronic Health Records (EHRs) and the general move toward big data as the catalysts. Others point toward the government, pushing healthcare in the direction of a digitalized future where Healthcare inflation will be tamed. EHRs hold the promise of improving productivity by streamlining the transfer of the right kind of data from ailing patients to caring clinicians, and the delivery of the right sort of information and treatment back in return. But, let’s examine more closely this “streamlining” notion and see how it can bring on this new day in Healthcare.

Adding Diagnoses, Adding Dimensions

For BH Integration, the specific challenge is to curate and deliver to the clinician the information they need to identify cases, identify symptoms, diagnose, and provide a basis for monitoring and guiding treatment progress. This challenge has awakened many stakeholders to the fact that, in choosing a BH screening tool, the instrument selected should not be limited to a single diagnosis. The one-dimensional PHQ9, long the standard for eliciting mental health symptoms generally, in fact serves only to uncover cases of depression. And as common and as destructive as depression is, it is not the only nor the most prevalent BH diagnosis.

Realizing this shortcoming in the PHQ9, many providers are on the lookout for instruments with the capacity to uncover a range of common BH problems. These tools need to be multi-diagnostic. And at the very same time, epidemiological and basic science research have been elucidating the dimensional aspect of mental illness. Conditions fall along a spectrum and often overlap and coexist. They develop along a trajectory from childhood to adulthood, but also morph over much shorter time scales in many individuals. In putting a mirror to Nature, in order to more accurately reflect real illness in real patients, the ideal BH instrument, therefore, must be multi-dimensional as well as multi-diagnostic.

Streamlining Data to Create an Expert System

To provide the right sort of data to clinicians — quickly, reliably, and in a useful and meaningful form — a properly designed screen needs to facilitate the learning style and behavior of expert clinicians. Pattern recognition is the hallmark of good clinical decision making. Every medical student learns that a differential diagnosis is the clinician’s system for organizing relative risks from among a list of possible causes of their patient’s symptoms. The suspected diagnoses on that list persist until evidence rules in or out each item one by one, until the final diagnosis is reached.

Rushing to closure on a particular diagnosis can be bad for a patient suffering from a rarer illness or from more than one condition. Seasoned clinicians bear in mind and continue to weigh the competing possibilities from their differential diagnosis while the illness course and treatment proceed; all the while keeping an eye on their patient’s overall status.

How do we train clinicians who are not experts to manage BH problems? The approach should model the know-how of the experts. That means we provide a system that fosters pattern recognition and the recurrent updating of the differential diagnosis as treatment proceeds. Combining the diagnosis-focused categoricalwith the holistic dimensional approach supports pattern recognition and comes closest to the thought processes of the expert. Data points need to provide meaningful feedback, and the recurrent updating of progress should facilitate clinical adjustments without premature closure. Such an approach leads ultimately to superior outcomes.

Let’s look at three instruments from the new generation of screening tools that improve on the one-dimensional PHQ approach. Each in its fashion provides for a dimensional approach to BH symptom curation.

The DASS 21:

A 21-item questionnaire intended to provide a multi-dimensional snapshot of an individual’s mood and anxiety state, the DASS 21 limits those dimensions to depression, anxiety, and stress. And while ‘stress’ is a useful concept that is understood by most everyone, it is not per se a diagnostic category. A stress “score” may shed light on the strain that a patient’s life stressors may be causing, and such stress itself may worsen the co-existing depressive and anxiety symptoms the tool is also measuring. Yet, the DASS 21 overlooks entirely a patient’s risk of Bipolar and Post-Traumatic Stress Disorders. These two conditions are not only prevalent, but their misidentification can have grave consequences for affected individuals. In particular, in the context of antidepressant treatment, burgeoning bipolar symptoms often go unrecognized until the condition is made considerably worse and more resistant to treatment.

For clinicians laboring to treat real patients, the dimensionality delivered by the DASS 21 simply fails to provide the information that makes the right sort of difference. With two major conditions essentially “invisible” to its measurement, conditions that in fact frequently co-occur with anxiety and depression, the DASS 21 simply cannot be considered an adequate tool for true BH integration.

The Cross-Cutting Scale (CCS):

In the case of the CCS, we have an over-abundance of categories without the delivery of a true dimensional perspective. The 25-question CCS scores an individual’s risk from among thirteen different clinical problems. Then, through a second level of tests — one each for depression, anger, mania, anxiety, somatic symptoms, sleep disturbance, obsessive-compulsive disorder, and substance abuse — the clinician is to assess the degree to which these conditions affect an individual’s quality of life. Four additional problems that round out the thirteen — psychosis, suicidality, dissociation and “personality functioning” — have no associated second level test. This hodgepodge of categories, some of them true disorders and others associated directly or indirectly with disorders, despite its breadth is not a true multi-dimensional tool.

The problem here is that the information is unwieldy and therefore the CCS, having collected all this information, is still a poor tool for adjusting to the clinical trajectory while keeping that eye on the whole patient.Despite the surfeit of clinical data delivered by the CCS, surprisingly, there is no provision for PTSD to be assessed on either level. And while it is useful to have second level tests made available to patients when the clinical picture is unclear, and these additional instruments may then be used to track treatment response, the lack of a single measure of illness severity fails to track the patient’s overall status and progress.

The M3:

Combining the categorical with the dimensional in the most organic way, the same way trained clinicians look at their patients, the M3 first of all provides an overall single score indicating how the patient is doing generally. Then, broken down into the four most prevalent BH conditions — anxiety, depression, bipolar disorder and PTSD — it provides a way to graphically keep track of trends along each of these clinically meaningful dimensions. At the very same time, it queries for alcohol and substance use patterns in the context of ongoing symptomatology.

The M3 also has the capability to add on second level instruments, but these may be chosen at the discretion of the clinician. Therefore, add-on tests are brought into use as the clinician feels the need.

Designed with PC practices in mind, available in English and Spanish at a fifth-grade reading level, the M3 has the added advantage of being cloud-based, which means there is no need to install software. The M3 is easy even for seniors to complete, and it typically takes just 3 to 5 minutes. It is also available online and as a mobile app, and it secure, encrypted, and fully HIPAA compliant.

Built to get the most out of a computer-based interface, the M3 is a tool for the 21st century. It is by virtue of its true multi-dimensional approach that the M3 is the only multi-dimensional electronic screening tool for use in BH Integration.

Making Traction by Making Sense

Providing the right sort of information is the key to streamlining BH Integration into Primary Care. Data needs to reflect the most relevant aspects of the clinical entities it purports to measure, it must be reliable, and it must be useful. The usefulness criterion means that the feedback provided by the instrument must empower clinicians to adjust in a timely fashion to the clinical trajectory of the patient. BH conditions are categorical and dimensional. All of the most common BH conditions need to be monitored for this to work, and so any instrument that is “blind” to Bipolar Disorder or to PTSD, and which does not track with a single measure the patient’s overall status, cannot be expected to deliver the kind of effective integration we all need and deserve. The M3 is currently the only state-of-the-art instrument that meets these criteria.



Gerald Hurowitz, MD

Assistant Professor of Clinical Psychiatry at Columbia University. Co-founder and Chief Medical Officer M3. Psychopharmacologist and Neuropsychiatrist.