Background

The threshold for involuntary commitment, called the dangerousness standard, is defined in Section 302 of Pennsylvania’s Mental Health Procedures Act of 1976 (MHPA) and is met when a patient “poses a clear and present danger of harm to others or to himself.” ¹³ The dangerousness standard for involuntary commitment was broadly enshrined in state laws across the country in the 1970s and 80s and reserves significant latitude for provider and judicial judgment.Reference Datlof ¹⁴ Application of the standard has not been uniform because “danger” is not precisely definedReference Brooks ¹⁵ and implementation may be affected by other factors,Reference Applebaum ¹⁶ like a concurrent criminal offense.Reference Levitt ¹⁷ Since its widespread adoption, numerous critiques have been levied against the utility of the dangerousness standardReference Miller and Hanson ¹⁸ due to its under-application,Reference Treffert ¹⁹ over-application, inability to accurately predict future dangerous behavior, and lack of support by social science data.Reference Dix ²⁰

In Pennsylvania, a person typically meets the dangerousness standard if they exhibit suicidal, homicidal, or self-injurious behavior/intent. Any individual familiar with the person’s dangerous behavior can initiate an involuntary commitment petition; if the petition is approved by the county administrator or their delegate, the person will be involuntarily brought to the emergency department by police.Reference Lazar ²¹ Involuntary commitment petitions can also be self-authorized by a physician or police officer. The initial involuntary hold lasts up to 120 hours and may be extended with a separate petition. A commitment may be appealed in court with an attorney and the committing physician present.Reference Whitehill ²²

Surveying the Data Landscape

Structured internet searches pointed to several potential sources of involuntary commitment data.Reference Barnard ²³ The searches included a county name plus “involuntary commitment” or “involuntary commitment dashboard” or “involuntary commitment data” or “mental health taskforce” for each of Pennsylvania’s 67 counties. These searches did not lead to useful state- or county-level data, except in Allegheny and Centre counties. Other counties occasionally provided rough counts of commitments in block grant reports or newspaper articles, but these records were often confusing or conflicting.Reference Harvie ²⁴ The majority of newspaper articles and county webpages identified through the structured searches detail Pennsylvania’s involuntary commitment process, direct users to resources, and advocate for funding. While researchers found that some counties have called for tracking of commitment services as a way of including stakeholders, these data were not publicly available. ²⁵ Ultimately, the structured internet searches failed to identify data that would lead to a useful state-wide count of involuntary commitments.

Public Dashboards and Data-Sharing Initiatives

The structured internet searches identified a significant amount of media, mostly consisting of newspaper articles and county reports, from Allegheny and Centre counties. In both counties, the violent death of a person with a mental illness brought attention to the issue of involuntary commitment and led to some data transparency. Allegheny County produced lasting visibility for the commitment population while Centre County did not.

The stories behind data transparency are illustrative. In 2012, a man with mental illness diagnoses and a history of involuntary commitments entered the Western Psychiatric Institute in Allegheny County with a firearm, killing one person and injuring several others before being shot by police.Reference Gurman ²⁶ The event became part of a larger discussion about guns, mental illness, and involuntary commitment.Reference Ruffalo ²⁷ The initial media responses to the shooting typify the link between mental illness and dangerousness, focusing on the ways the mental healthcare system had failed to protect the public. The media quickly reported on the individual’s history of violenceReference Begos ²⁸ and identified systemic failures that might have identified his dangerousness to the public, making him a candidate for involuntary commitment. ²⁹ The County Comptroller responded to public attention ³⁰ with an audit of mental health services, leading to several substantive changes, including a new dashboard for involuntary commitment data. ³¹ The dashboard is updated daily by Allegheny County’s Department of Human Services (DHS) and reports the number of filed/upheld involuntary commitment petitions, location information, and who initiated/authorized the petition. ³²

Allegheny County DHS further links institutional data with county administrative data for treatment oversight and coordination ³³ in compliance with the Health Insurance Portability and Accountability Act (HIPAA) and the MHPA. ³⁴ Allegheny County’s data have been critical for research and policy development. For instance, their data have been used by journalistsReference Winger ³⁵ and researchers, who found that Black young adults are about 2.5 times more likely to experience an involuntary commitment petition than their White peers. ³⁶ Demographic injustices were also found among older members of the commitment population, where “[M]en and Black individuals are disproportionately likely to be subjected to an evaluation for involuntary commitment” with no substantial difference in the likelihood of that commitment being upheld.Reference Welle ³⁷ Most strikingly, within five years of their first involuntary commitment evaluation, researchers found that 20% of the involuntary commitment population had died. ³⁸

Allegheny County DHS has planned further analyses and policy reviews using their data, suggesting that the project is relevant and worth sustaining. ³⁹ However, there was no uptake when they offered the tracking system technology to other Pennsylvania counties. ⁴⁰

Conversely, the death of Osaze Osagie in Centre County attracted similar attention to involuntary commitments without leading to lasting data transparency. On March 20, 2019, State College police served an involuntary commitment warrant on Osagie, a Black man living with schizophrenia whose father had initiated an involuntary commitment petition. Osagie rushed three officers with a serrated knife; they responded by discharging a Taser with no effect and then shot Osagie several times, killing him.Reference Cantorna ⁴¹

This event caught the attention of local media and the public: community members attended county board meetings, formed an advocacy group, and staged protests that connected Osagie’s death with the Black Lives Matter movement.Reference Manno ⁴² The County responded to the incident with the “Final Report,” ⁴³ best practices guidelines, ⁴⁴ and a report by the district attorney. ⁴⁵ These documents addressed community concerns, offered accountability, and advocated for those experiencing mental health crises.

The County’s “Final Report” includes some limited commitment data, including the number of involuntary commitment petitions that were submitted, approved, or denied from January to September 2020. ⁴⁶ The report recommended that the county build a data collection and sharing system. But, at the time of writing, no public data-sharing system could be found, and it was impossible to verify through public sources, communication with county administrators,Reference Arbogast and DeFilippis ⁴⁷ or by attending advisory board meetings if any concrete steps had been taken toward this recommendation. ⁴⁸

In both Allegheny and Centre counties, the violent death of a person with a mental illness who was a candidate for involuntary commitment attracted public attention and motivated officials to make the commitment population visible through data. While reports produced by Allegheny County DHS demonstrated that data-sharing initiatives can play an important role in shaping policy, there was no comparable lasting change in Centre County.

Administrative Documents and Right-to-Know Requests

The MHPA requires the collection of several administrative documents during an involuntary commitment. The path of two administrative documents, the “Application for Involuntary Emergency Examination and Treatment” and “Notification of Mental Health Commitment,” offered two potential avenues for recovering state-wide involuntary commitment data. Right-to-Know (RTK) requests submitted to state and county offices did not locate useful commitment data based on these forms. ⁴⁹

During an involuntary commitment petition, the first part of an “Application for Involuntary Emergency Examination and Treatment” is completed, including a record of the individual’s demographic information, the identity of the petitioner, and the dangerous behavior that supports the petition. In the emergency department (ED), additional sections of the “Application” are completed, including information on whether it was upheld or denied by the physician. If the petition is upheld, the completed form is submitted to the county mental health administrator.

The way the “Application” is handled by county mental health administrators differs across Pennsylvania. In Centre County, for example, the forms are received by an administrator weekly and reviewed to ensure procedural compliance and the protection of patient rights. ⁵⁰ An administrator claimed that there was a system for tracking statistics and a mechanism for reporting these to the state DHS, but this could not be verified. ⁵¹ Researchers submitted RTK requests to assess if Centre County or DHS could provide de-identified information on commitments. DHS responded that “the department does not know of or track all involuntary commitments.” Similarly, Centre County found no records responsive to the request. ⁵²

Another form, the “Notification of Mental Health Commitment,” must be submitted to the Pennsylvania State Police (PSP) within seven days of an involuntary commitment. The PSP maintains a database of commitment notifications because it is unlawful for a person who has been involuntarily committed to possess a firearm. ⁵³ This database feeds into the Pennsylvania Instant Check System (PICS), a statewide system for conducting background checks. ⁵⁴

An RTK request was submitted to the PSP to recover these data. The PSP responded to the request with the aggregate number of “Notifications” received in every county since 1976. ⁵⁵ Although the “Notification” requires the collection of demographic information, researchers were informed that “the PSP does not break down Mental Health Commitments by de-identified demographics.”

A subsequent RTK request was submitted to the PSP seeking the number of “Notification[s]” received by Allegheny, Centre, and Luzerne counties during 2022. The PSP provided a report with the date of involuntary commitment, the date the form was received by the PSP, and the type of commitment. The authors found numerous discrepancies in these data. In Allegheny County, for instance, the PSP dataset reports receipt of 3,306 involuntary commitment notifications, while the public dashboard notes 3,379 upheld involuntary commitments during the same period. The PSP records from Allegheny County contained 257 entries without information on the type of commitment and 384 petitions from outside the requested date range — the oldest inexplicably from 1935. ⁵⁶ In general, these data have many incomplete or conflicting records.

This survey of extant data reveals that state and county agencies have a paucity of accurate records on involuntary commitments. Public documents, from block grants to county reports, have little useful data while simultaneously reflecting the state’s understanding that data are useful for oversight and transparency. When records were accessible, they lacked demographic information or had questionable reliability. This gap in useful, reliable data exists in concert with calls for accountability and transparency, highlighted in Centre and Allegheny counties, where public attention followed violence to/by the mentally ill.

Undone Science and Agnotology

In Pennsylvania, commitment data are characterized by a tracking system motivated by firearm regulation and countervailed by the absence of a patient-focused database. The language of undone science and agnotology clarify the implications of this gap. Undone science reads the gaps between knowns and unknowns as “historical products of disciplinary practices”Reference Frickel and Bess Vincent ⁶⁵ located in “the institutional matrix of governments, industries, and social movements.”Reference Frickel ⁶⁶ This theory suggests that the absence of a measurement, in this case patient-focused commitment data, is not a neutral category — instead, the presence/absence of “indicators are inevitably political, rooted in particular conceptions of problems and theories of responsibility. They represent the perspectives and frameworks of those who produce them, as well as their political and financial power. What gets counted depends on which groups and organizations can afford to count.”Reference Merry ⁶⁷ The association between undone science, the state, institutions, and laws also benefits from a discussion of agnotology, a program of study interested in the norms that create ignorance and the absence of knowledge.Reference Proctor and Schiebinger ⁶⁸

This characterization of present/absent data requires a description of the attributes of ignorance, including granularity, scale, and intentionality:Reference Croissant ⁶⁹

1. (1) Granularity measures the quantization of knowledge, from “atomic propositions” ⁷⁰ and “[a]bsent bits of information in communication streams” ⁷¹ to specific redactions, classes of documents, and whole domains of knowledge. Granularity frames the PSP data, which includes some relevant information but lacks demographics. The available quantitative data have low granularity, are unreliable, and do not relate any qualitative information about an individual patient’s subjective experience.

2. (2) Scale pertains to “the components and systems in which that knowledge, whatever its granularity, might circulate.” ⁷² The scales of ignorance appear in the individual, middle, and upper ranges. In the setting of involuntary commitment, the state recognizes the relationship between rights and accountability for the commitment population in the individual range. For the patient in the clinic or ED, the state explicitly defines benefits, harms, responsibilities, and modes of redress.Reference Lundeen ⁷³ The middle range contains pathologies and cultures around who maintains data, bears responsibility, and can respond to injustices. The commitment population becomes less recognizable in the middle range, where intra-/inter-organizational actors — hospital systems, offices of county administrators, and state departments — experience intense variation in available information about the population. The upper range consists of “frameworks which articulate deep epistemological rifts in knowledge,” ⁷⁴ places where ignorance is inflected in general theories of understanding. This includes the common language for describing illnessReference Sontag ⁷⁵ and the frames used to communicate about mental illness. ⁷⁶

3. (3) Intentionality maps whether absences or omissions are conscious, intentional, and willful, or unconscious, surprising, and unknowable. ⁷⁷ These vertices include knowledge, knowledge that is known not to exist, and unknowns. Commitment data is knowledge that is known not to exist. This type of absent knowledge may be produced by a privileged group that controls the discourse, throttles the spread of information, or simplifies complex stories in the service of select interests.Reference Shotwell ⁷⁸ In cases where it is known that knowledge does not exist, it is possible to theorize a process of “active production” that reveals disciplinary practices and the function of power.Reference Tuana ⁷⁹

Granularity in particular helps make sense of extant involuntary commitment data. Middle range actors have the capacity to collect meaningful data on commitments as evidenced by statute, crisis reporting, ⁸⁰ county efforts, ⁸¹ and anecdotal evidence. ⁸² But the only statewide commitment database belongs to the PSP — and the granularity of that data is extremely low. The absence of reliable data, or even a reliable count of the commitment population, is thus a meaningful enumerative omission.Reference Cheong ⁸³ This classification gives shape to a population that might be counted but is not, further “underscore[ing] the positive, active political and symbolic work being done by the ostensibly negative, inactive state behaviors of not registering, not documenting, not counting, and otherwise not recognizing.” ⁸⁴

Even if a rigorous commitment database, like Allegheny County’s, were built using the administrative documents currently collected by the state, an analysis of granularity reveals another omission. The lack of patient perspective in administrative documents means that the data are complicated by a kaleidoscope of imbalanced power relations. In the current system, patients must wait to seek redress in court if they disagree with the physician’s judgment — and, notably, it is the physician’s perspective that is the focus of the “Application for Involuntary Emergency Examination and Treatment.” Very little information about the patient’s subjective experience is collected on the “Application.” Similarly, tracking systems that only record quantitative data are unable to relate the richer qualitative data that may provide a better picture of the patient’s experience of involuntary commitment.Reference Mekel ⁸⁵ The function of these omissions is that a patient subject to an involuntary commitment may be seen by a physician in the ED and never be fully seen by the state.

The MHPA and Carceral Censuses

State regulatory regimes, like the MHPA, influence research priorities and the way the state sees its residents.Reference Scott ⁸⁷ Regulatory omissions fall into two broad categories: formalized omissions, which are “enacted in the letter of the law or policy,” and discretionary omissions, which rely on subjective interpretations of policy “in ways that lead to the denial of persons from being registered or enumerated.” ⁸⁸ The MHPA operationalizes both types of omissions by formally limiting access to patient records to the patient, provider, county administrator, court, and payors, ⁸⁹ while making a discretionary allowance for disclosing de-identified information.Reference Moskop ⁹⁰ The state’s intentional or unconscious choice to not exercise this discretionary allowance is significant because data transparency and enumeration would provide a route for the commitment population to lobby for recognition, accountability, and policy remedies.

While visibility through enumeration does not inherently affirm rights, it provides a scaffolding for individuals to assert recognition through their collective carceral, racial, or health identities. ⁹¹ Disciplinary systems and prison reform movements have historically been shaped by the visibilityReference Foucault and Sheridan ⁹² of those subject to the state’s power: the population must first be seen before the system can change.Reference DePuy ⁹³

Broadly, involuntary commitment and criminal detention share a loss of autonomy, stigma, and restriction of rights — but the prison population is visible through enumeration. The US Census Bureau counts national prisoner “admissions and releases by type and sex during each calendar year, as well as the number of prisoners in custody by sex and sentence length.”Reference Minor-Harper ⁹⁴ Over time, the National Prisoner Statistics Program has adapted its data collection system to meet the changing needs of the incarcerated population, researchers, and the government. ⁹⁵ It is striking that the commitment population is not afforded the same visibility. As Morris notes, “[c]ivil commitment is not equivalent to criminal justice in structure or function, and criminal justice data reporting is not perfect; still, it is difficult to imagine a similar situation in which such basic information about the criminal justice system, such as the numbers of people incarcerated annually, remain a mystery.” ⁹⁶

The responsiveness to community informational needs appears at the intersection of policing and involuntary commitment in Centre County. The State College Police Department (SCPD) began tracking involuntary commitment warrant services after Osagie’s death in 2019. An SCPD Assistant Chief emphasized that the department did not previously recognize the utility of tracking involuntary commitment warrant services and suggested that the data may be useful for understanding system utilization and identifying officer bias. ⁹⁷ The Assistant Chief likened this change to the addition of a racial demographics question to forms related to traffic stops, a methodological shift that became widespread in the early 2000s to make discriminatory policing visible.Reference Ramirez ⁹⁸

Most middle range actors in Pennsylvania have not demonstrated a similar willingness to look for methodological inconsistencies across the commitment population. This is, in part, because the MHPA protects the rights of individuals and never resolved the numerous gaps, vague language, and varying interpretationsReference Datlof ⁹⁹ that affect the population when they interact with middle range institutions. ¹⁰⁰ The emphasis on preserving individual rights became the only possible check in the system, licensing an active production of different practices under the same standards. For instance, “an individual can be committed for a specific set of behaviors in one community while not committed forty miles away in another community.” ¹⁰¹ This is a middle range implementation failure where the institutional cultures and legal strictures intended to produce homogeneous treatment of individuals fail to secure uniformity. ¹⁰² When the state uses porous laws and norms to protect the rights of individuals, it may create issues that only appear when the population is made visible and enumerated. If there is no population-level visibility, the state creates the appearance of justice at the individual range while giving rise to suspected, but largely unknown, injustices in the middle range.

Privacy, Confidentiality, Visibility, and Dangerousness

Because mental health is sensitive, might the absence of data and patient invisibility be understood as a necessary deference to privacy? The ethical territory is contested and, perhaps, some definitions are in order. Privacy is classically understood as “the right to be let alone.” ¹⁰³ There are three principal usages of privacy: physical privacy, informational privacy, and decisional privacy.Reference Allen and Reich ¹⁰⁴ While related, confidentiality typically refers to the way private information travels once shared.Reference Donaldson and Lohr ¹⁰⁵ Privacy and confidentiality have been hallmarks of medical care since antiquity and are an integral part of medical ethics. ¹⁰⁶

In Centre County, it was hard to tell whether individuals or institutions were principally served by deferring to informational privacy and confidentiality. When prompted, officials repeatedly cited privacy as their rationale for the continued absence of involuntary commitment data, arguing that it is impossible to produce meaningful data without compromising patient privacy. ¹⁰⁷ These officials share a position with commentators who favor an absolutist view of privacy which, they suggest, is a necessary precondition for medical autonomy.Reference Allen and Zalta ¹⁰⁸ The officials’ statements contrast with the aspirations of Centre County’s post-2019 data-sharing initiatives and effectively shields the county from public oversight.

Non-sharing is problematic, in part, because absolute privacy protects the individual at the expense of the population if information about the individual has relevance to the group. This departs from the classic principles of bioethics — respect for autonomy, nonmaleficence, beneficence, and justiceReference Beauchamp and Childress ¹⁰⁹ — which center the individual patient and their web interactions with physicians “without considering the collective processes of health. This focus on the individual and interpersonal then allows processes that produce inequality to avoid consideration, critique, confrontation and change. In essence, the focus on the individual allows the engines producing inequality to continue unchallenged.”Reference Breilh ¹¹⁰

Although a patient experiences an involuntary commitment as an individual, information about their care has utility to the commitment population. The point is chiasmatic: social utility alone is an insufficient rationale for the state to override the right to privacy, but rights should not necessarily function as an “absolute barrier to the pursuit of collective goals.” ¹¹¹

The modern public health apparatus regularly balances privacy and collective goals through a far-reaching medical surveillance infrastructure that manages diseases, illnesses, and public health problems by making them visible without authorization. In cancer studies, for instance, authors have long “indicate[d] the social harm that could result if use of multiple sources in cohort epidemiology were curtailed. For example, insistence on prior permission from all members of a cohort, would render unthinkable most of the present undertakings.”Reference Newcombe ¹¹² Involuntary commitment is an outlier insofar as institutional preference skews toward strict observation of decisional and informational privacy. Large healthcare reporting centers and county mental health offices collect a wide variety of sensitive health data but remain unwilling or unable to collect and share useful data on involuntary commitments. ¹¹³

Absolute privacy and strict confidentiality are not written into commitment laws. State and federal policies permit breaches in privacy for select collective, socially important reasons beyond an individual’s medical treatment. The MHPA outlined acceptable uses of “anonymous” patient information long before the national standards for using patient information were codified by HIPAA in 1996, carving out specific acceptable uses of de-identified patient information. ¹¹⁴ HIPAA allows institutions to enter into agreements with researchers to disclose “limited data sets” of healthcare information for research purposes. ¹¹⁵ Public health laws do not require reporting of involuntary commitment data but also do not specifically limit public access to de-identified individual or aggregate data. ¹¹⁶

For involuntary commitment patients, the circumstances that typically lead to breaches of privacy are bound up in the issue of public safety, not with the goal of improving healthcare. The Code of Ethics of the American College of Emergency Physicians states that “[e]mergency physicians shall respect patient privacy and disclose confidential information only with consent of the patient or when required by an overriding duty such as the duty to protect others or to obey the law.” ¹¹⁷ The duty to protect others compels a provider to violate patient privacy and medical record confidentiality when a patient exhibits violent intent, in line with the dangerousness standard. The duty reappears in Kendra’s Law, Laura’s Law, and the Tarasoff Rule. These laws were born out of shocking deaths and curtail the confidentiality of mental health patients, favoring information sharing and public safety under certain circumstances. The thrust is that laws and norms lead to information sharing when an individual could be dangerous to the public, not for the benefit of the patient.

While these laws and norms might achieve public safety in certain limited cases, the unfortunate fact is that the dangerousness standard’s language of “danger of harm to others or to himself” is unintentionally revealing. ¹¹⁸ For mentally ill individuals, especially those who might be violent, protecting “others” comes before protecting “self.” This aligns with the emphasis on public safety that makes patient-centered information sharing and visibility a secondary aim.

Mental health laws are often rationalized as beneficent policies that preserve privacy, but the predictable outcome of this deference to privacy is invisibility. These laws exist in concert with middle range implementation failures and upper range social frameworks/stigmas that favor invisibility and control.Reference Mattelart ¹¹⁹ This logic is reflected in the current administration’s “compassionate” policies ¹²⁰ toward the unhoused and mentally ill individuals who will be rendered invisible through forcible removal from public spaces. ¹²¹ Acts of control in the name of compassionate care regularly target the unhoused by using mental illnessReference Schweik ¹²² and disability as proxies for removal. ¹²³ A similar visibility logic is displayed in the Supreme Court’s recent ruling in City of Grants Pass v. Johnson, which reversed decades of precedent by permitting the criminalization of sleeping in public. ¹²⁴ In both cases, the state favors solutions that turn a visible condition into an invisible one ¹²⁵ — being unhoused or mentally ill in a shelter bed is preferable to being visible in public. ¹²⁶

Moving Forward

It is challenging to identify root causes and perhaps unwise to definitively resolve the specific motives behind involuntary commitment’s numerous invisibilities and undone sciences. One must be cautious about assigning blame when knowledge doesn’t exist because “there is often no ‘smoking gun,’ or more correctly no smoking memo, pointing to specific guilty political actors.”Reference Michalowski, Chambliss and Kramer ¹²⁷ In part, this is because a state may violate laws, conduct norms, or cause social injury through the complex, interwoven “consequences of an economics-politics-culture nexus that produces a normalized exercise of bio-power that is frequently deeply injurious to particular segments of the society, or even the society as a whole.” ¹²⁸ To name the exact historical forces behind the absence of data would require a fluid investigation that may find logic and rationality in processes where there was none.

Furthermore, such a history would be insufficient — working on an exhaustive institutional history of involuntary commitment policies would not necessarily affect the much-needed changes to policy and norms. The status quo, which defers to privacy and confidentiality, maintains numerous gaps, injustices, and indignities independent of intentionality — leaving researchers, patients, and the public to grapple with the consequences. Considering the administrative documents that must be produced during each involuntary commitment, available technology, legal points of data capture, and the value to care and health policy, the state should move to collect and analyze involuntary commitment data. A decades-old observation about medical records applies: “the challenge is no longer technological, as the technology has already been developed. Rather, the issues are moral and political.”Reference DeCew ¹²⁹