Spotlight on the Workplace Violence Bill: The ‘dangerous patient’ list provision

Imagine arriving at an emergency room frightened, in pain, or experiencing psychiatric distress.

Before anyone speaks with you, a warning identifying you as a potential safety threat appears in your medical record. It may be based on an encounter at another hospital, a police report, information held by a state agency, or someone else’s account of an incident that occurred years earlier.

You may not know what the warning says, who entered it, whether it is accurate, or how long it will remain. Yet it may shape whether workers approach you with compassion or suspicion, whether security is called, whether you are searched, whether your pain is believed, whether you are allowed to leave—and whether police become involved.

Massachusetts lawmakers have not explicitly proposed creating a public registry of “dangerous patients.” Both versions of the healthcare workplace-violence bill would, however, direct state health and public-safety officials to develop recommendations for expanded information sharing involving people with psychiatric, behavioral-health, or developmental-disability diagnoses who have become involved with the criminal legal system.

Neither version literally creates a “dangerous people list.” But both could help build the legal and technological architecture for information about identified people to travel among healthcare, law-enforcement, and public-safety systems.

That deserves considerably more attention than it has received.

The bills’ proposed reports on interagency data sharing

Section 8 of H.4767 would require the Executive Office of Health and Human Services to work with the Executive Office of Public Safety and Security to recommend improved data sharing, communication, and collaboration among healthcare facilities, public-safety agencies, and law-enforcement entities.

The House report would specifically consider allowing healthcare facilities to access “reports on individuals” maintained by agencies within those executive offices through electronic medical records, health-information exchanges, or similar systems. It would also consider integrating that information into people’s medical records for treatment or diagnosis.

The provision is directed specifically toward “criminal justice-involved patients” with psychiatric or behavioral-health diagnoses.

Section 9 of S.3184 requires a similar report concerning interagency data sharing. The Senate expands the population under discussion to include people with intellectual or developmental disabilities, including autism, but removes the House’s explicit proposal that healthcare facilities gain access to government reports about individuals.

Instead, it calls more generally for improving the exchange of information among agencies to support people’s treatment and diagnosis.

Both versions require the resulting recommendations to address:

limiting access to the minimum information needed for a defined purpose;
using de-identified or aggregated information whenever practicable;
sharing identifiable information only when considered essential for treatment, diagnosis, or immediate safety;
prohibiting its use to initiate or support criminal, civil, regulatory, licensing, or administrative actions except as expressly authorized by the legislation;
tracking and auditing access;
imposing penalties for misuse; and
preserving existing protections under HIPAA, federal substance-use confidentiality rules, and other laws.

These proposed safeguards acknowledge some of the system’s obvious dangers. They do not resolve the underlying problem.

The bills would require recommendations, not establish an operational information-sharing system. Those recommendations could, however, lead to a system through which healthcare and public-safety institutions collect, circulate, and act upon information about identified people. Access controls do not remove institutional bias. Audits do not undo a coerced injection, an unnecessary restraint, an avoidable arrest, or care denied because someone was approached as dangerous. An appeal process does little for someone who never knew the information existed until after it shaped an encounter.

The question is not simply whether such a system could be designed more carefully. It is whether institutions with the power to confine, forcibly treat, arrest, and prosecute people should create it at all.

What information could be shared, and who could challenge it?

Neither version explains:

What kinds of agency reports could be shared.
Whether information would have to concern proven conduct or could include unverified allegations.
Whether an arrest, conviction, or formal complaint would be required.
Whether reports about involuntary commitment, attempts to leave, resistance during restraint, or other disability-related behavior could be included.
Who would decide that the information was relevant.
Whether people would be notified when identifiable information about them was entered into or shared through the system.
Whether they could review the information, challenge its accuracy, add context, or have it removed.
How long the information would follow them.
Whether workers would see the original account or merely a warning label.
How mistaken identity would be addressed.
Whether information could influence security, commitment, custody, or forced-treatment decisions.
Whether a patient’s allegations of staff abuse or misconduct would appear alongside allegations made against the patient.

These are not minor implementation details. They reveal how much power the proposal could give institutions while leaving the people being labeled with little control over what is said about them, how far it travels, or how it shapes their care.

Answering these questions would not make the underlying proposal acceptable. Their absence simply makes an already dangerous idea even more alarming.

The Senate’s language is less explicit than the House’s, but that does not eliminate the concern. The Senate’s broader language leaves the form of any proposed information-sharing arrangement to the agencies’ recommendations, rather than specifying it in the bill.

What about Homeland Security and immigration enforcement?

The Senate directs the report to the Legislature’s Joint Committee on Public Safety and Homeland Security. Despite its name, this is a Massachusetts legislative committee—not the federal Department of Homeland Security, which oversees Immigration and Customs Enforcement and Customs and Border Protection.

Neither bill expressly authorizes disclosure to ICE or other federal immigration authorities. Still, both involve the Executive Office of Public Safety and Security, which oversees the State Police and the Commonwealth Fusion Center. Given those agencies’ existing information-sharing relationships, there are reasonable concerns about where information could ultimately travel.

The possibility of immigration-related use adds another layer of danger, but an explicit prohibition on ICE access would not cure the larger problem. Information does not need to reach immigration authorities to harm immigrants. The expectation that healthcare records may communicate with public-safety systems can deter people from seeking care, intensify fear during emergencies, and make encounters more dangerous.

The appropriate response is not to construct the system and promise that one particularly harmful agency will be excluded. It is to remove the interagency data-sharing provisions.

A warning attached to a person is never “just information”

Hospitals already use electronic behavioral or safety flags to alert workers about patients considered potentially aggressive. These systems are generally created by individual healthcare organizations rather than through a statewide healthcare and public-safety network.

Information can sometimes help prevent a harmful encounter—but that does not require a dangerousness list or the exchange of government reports.

A voluntary plan developed with a person might tell workers not to approach from behind—a good practice all around—to provide additional space, arrange an interpreter, avoid known trauma triggers, or follow strategies the person has identified as helpful.

That is information shared with someone’s participation and directed toward changing what the institution does.

It is fundamentally different from attaching a warning to someone based on allegations or institutional records they may never have seen. The occasional usefulness of the first does not justify creating the second.

A label that communicates “dangerous” changes how workers perceive and approach someone.

It may cause staff to enter with additional personnel, visible security, gloves, restraints, suspicion, or force. A patient may then react to being surrounded or treated as dangerous, only to have that reaction interpreted as confirmation that the warning was correct.

The label can create the very escalation it supposedly predicted.

Racial disparities in existing hospital behavioral flags

A 2023 study published in JAMA Network Open examined behavioral flags across more than 426,000 emergency-department visits within one healthcare system.

Black patients received behavioral flags at higher rates than white patients. After adjustment for other measured factors, visits by Black patients were associated with more than two-and-a-half times the rate of behavioral flags.

The differences did not end with the label. Among patients with flags, Black patients waited longer to be placed in a room and to see a clinician, had shorter overall stays, and received fewer laboratory and imaging tests than white patients with flags.

The study could not establish that flags caused each difference in care. But it demonstrated exactly why these systems cannot be treated as neutral collections of objective facts.

Someone must decide whose behavior appears threatening, whose fear is understandable, whose anger is tolerated, and whose resistance becomes evidence of dangerousness. Those judgments are made within systems already shaped by racism, ableism, transphobia, psychiatric stigma, and unequal distributions of institutional power.

Sharing the resulting label more widely does not remove that bias. It multiplies its reach.

What emergency nurses report about behavioral flags

In a separate study examining emergency nurses’ experiences with behavioral flags, some workers said alerts helped them prepare for difficult encounters. Some believed a flag had occasionally helped avert harm by allowing workers to plan ahead.

But most participants said the flags themselves did not prevent violence. Workers also raised concerns about biased placement, subjective narratives, outdated information, and warnings that remained in people’s records years after an isolated incident.

Some nurses acknowledged that workers occasionally escalated encounters and that burnout, understaffing, long waits, untreated pain, and confrontational approaches could contribute to the behavior producing a flag.

Once the flag exists, however, that context can disappear.

“Patient attempted to bite staff” may follow someone indefinitely.

“Patient was held down by five people, struggled to breathe, and bit someone while attempting to escape” tells a considerably different story.

So does: “Patient repeatedly asked what was happening, received no interpreter, and panicked when security entered.”

A system developed from these recommendations could turn one institution’s incomplete account into an enduring judgment accepted by every institution that encounters the person afterward.

Who gets to tell the official story?

This problem becomes even more serious when the information originates with police or another government agency.

Police reports generally document allegations and officers’ observations. They are not judicial findings, and they are not necessarily complete or accurate accounts of everything that occurred. An arrest does not establish guilt. A call for assistance does not establish dangerousness. An involuntary commitment does not establish that someone harmed anyone.

Even a documented act of aggression does not establish that the person remains a risk years later or would behave similarly in a different environment with appropriate communication and support.

Yet once information enters a medical record, it can acquire the appearance of clinical truth. Future workers may repeat it without knowing its source, context, age, or reliability.

A police allegation can become a medical warning. The medical warning can influence a future encounter. That encounter can generate another police report, which further strengthens the warning.

That is not simply information sharing. It is a feedback loop between healthcare and criminalization.

Lists have a way of growing

Systems built to identify supposedly dangerous people rarely remain confined to the narrowest examples used to justify them.

Definitions expand. More agencies receive access. Information collected for one purpose becomes useful for another. An allegation becomes a data point; several data points become a risk designation; and the designation begins influencing decisions that lawmakers never publicly debated.

Errors can also become extraordinarily difficult to correct because each agency points to information supplied by another. Responsibility becomes dispersed while the consequences remain concentrated on the person labeled.

Those consequences may never appear as a formal criminal or administrative action. They may look like:

Security waiting when the patient arrives.
Police being called earlier than they otherwise would have been.
Workers entering the room in a group.
Routine requests being interpreted as manipulation or escalation.
Pain complaints being treated as drug-seeking.
Disability-related behavior being interpreted as threatening.
A patient being denied choices routinely offered to others.
Greater willingness to restrain, medicate, or isolate.
Clinicians discounting the patient’s account of abuse by workers.
People avoiding healthcare because they expect hostility, force, or arrest.

A statutory prohibition against using shared information to initiate a criminal or administrative action would not prevent many of these outcomes.

Existing research on behavioral flags shows limited and mixed perceived benefits alongside documented risks of racial disparity, outdated information, institutional bias, and differences in the care people receive. Neither reporting provision expressly requires an evaluation of whether expanded interagency information sharing would reduce workplace violence.

More importantly, this is not a problem that can be solved by finding the perfect retention period, appeal process, privacy policy, or definition of dangerousness.

Lists created by institutions with coercive power do not remain neutral collections of facts. They reflect the judgments and prejudices of the people entering the information. They remove events from their context. They travel beyond their original purpose. They shape future decisions without the labeled person in the room. And their harms fall most heavily on people already subjected to racism, ableism, psychiatric oppression, transphobia, poverty, and criminalization.

Any list may occasionally provide information that someone finds useful. That limited possibility does not justify constructing a system whose foreseeable harms include more suspicion, more force, more police involvement, unequal care, and avoidance of healthcare altogether.

Patients deserve not to be secretly branded by allegations they cannot see, challenge, contextualize, or escape.

Massachusetts should not study how to build this system more carefully. The conference committee should remove the interagency data-sharing provisions from the bill.

We do not need a better dangerous-person list.

We need to reject the premise that creating one will keep us safe.