Malpractice Liability Due to Patient Violence

Despite the commonly held perception that the world is a more dangerous place, rates of homicide and aggravated assault have significantly declined in recent years (7). Nonetheless, violence remains a major public health problem. An estimated 12 million American adults are involved in intimate partner violence each year (7). Child protective services (CPS) estimated that of the 3.4 million referrals for child abuse and neglect in 2012, 686,000 children were mistreated. Among the substantiated CPS referrals for child maltreatment, 30% experienced physical or sexual abuse (8). Approximately 16,000 homicides occur each year, making homicide a leading cause of death among young people (7, 9). The United States has the most guns per capita (89 guns per 100 people) and the greatest number of public mass shootings in the world (10). Although deaths from mass shootings make up a very small percentage of homicides, they cause inordinate fear in the public.

Although the relationship between violence and mental illness is frequently sensationalized in the media, most violence is not due to symptoms of mental illness (11, 12). Some mental illnesses are associated with an increased risk of violence (13, 14), but only about 4% of violent acts are thought to be directly attributable to mental illness (11). A meta-analysis of 48 studies of criminal recidivism found that the risk factors for violent recidivism were nearly identical among offenders who were diagnosed with a mental illness and those who were not (15).

The single greatest risk factor for future violence is past violence (14, 16). Other static risk factors for violence include previous arrests, male gender, age between 15 and 24 years, and a history of adverse childhood experiences (e.g., parental separation, being the victim of child abuse, and witnessing violence in the home; 14).

Social risk factors for violence include low socioeconomic status (13), lower educational level (17), and unstable employment (18). Disruptive behavior starting at a young age (e.g., fighting, truancy; 18, 19), antisocial personality disorder (16), and psychopathy (20) are important static risk factors for violence. Other personality factors associated with violence include a predisposition toward feelings of anger and hatred (as opposed to empathy, anxiety, or guilt), hostile attributional biases (a tendency to interpret benign behavior of others as intentionally antagonistic), violent fantasies, poor anger control, and impulsivity (14).

Two key dynamic risk factors for violence are substance use and active psychiatric symptoms. The prevalence of violence is 12 times greater among individuals with alcohol use disorder and 16 times greater among individuals with other drug use disorders (13). The majority of people who commit violent crimes are under the influence of alcohol during their violent offense (21). Worrisome mental health symptoms include psychosis, particularly when a person has a belief about a threat or control by an external force (16). Paranoid individuals are at risk of killing someone whom they misperceive as an attacker. Command hallucinations, especially when associated with a delusion, elevate the risk of violence (16). Individuals with poor insight into their illness (who are more likely to be noncompliant with antipsychotic medication) are at higher risk of violence. Posttraumatic stress disorder and previous head injury are also linked to violence (14).

Contextual factors to consider in violence risk assessments include current stressors, lack of social support, availability of weapons, access to drugs and alcohol, and the presence of similar circumstances that led to violent behavior in the past (14). For example, it would be concerning if a person with a history of attacking his unfaithful wife’s lover is faced with his wife’s being unfaithful again. Another concerning situation would be that of an angry individual with a grievance who recently suffered humiliation or a narcissistic injury. Access to lethal weapons, particularly firearms, is important because the difference between an assault and a homicide often lies in the lethality of the weapon used.

Violence risk assessments are involved in psychiatric decisions regarding discharge from a psychiatric hospital, civil commitment, and seclusion and restraint. There are three broad approaches to violence risk assessment. The first is unaided clinical judgment, in which mental health professionals estimate violence risk on the basis of their clinical experience without the use of structured instruments (22). The second approach uses actuarial instruments specifically designed to estimate the probability of violence. These actuarial instruments, such as the Violence Risk Appraisal Guide, rely on statistical calculations that gauge the likelihood of an outcome (violence) when a particular set of risk factors is present (23). Some authors favor the complete replacement of clinical judgment with actuarial methods (23). The third approach to violence risk assessment is structured professional judgment. This method uses elements of both clinical judgment and actuarial instruments. Structured professional judgment instruments assist the evaluator in identifying empirically established risk factors. Once the information is collected, it is combined with clinical judgment in decision making (23, 24). An example of a structured professional judgment instrument is the Historical Clinical Risk Management-20 Scale (23).

Although debate exists regarding whether actuarial instruments are more accurate than structured professional judgment instruments, it is well established that clinical judgment is the least accurate method (22). Risk assessment instruments offer a moderate level of accuracy in categorizing people into low versus high risk (2). They have better ability to accurately categorize individuals as low risk than as high risk, where false positives are more common (25, 26).

Many psychiatrists are not trained to use actuarial or structured professional judgment instruments for violence risk assessment, and clinical judgment remains the most commonly used method in clinical practice (22). At the present time, use of a structured violence risk assessment instrument is not required to meet the standard of care (22), although it is becoming more common in some settings (e.g., when discharging a patient found not guilty by reason of insanity from a forensic hospital).

If clinical judgment without a risk assessment instrument is used, data about violence risk factors should be collected in a systematized fashion. One useful strategy is to create a personal template that outlines static (not subject to change by intervention) and dynamic (subject to clinical intervention) risk factors for violence to guide the clinical evaluation. Clinicians should not rely on the patient’s self-report alone. Collateral sources of information should be reviewed, such as past medical records and court dockets online, in addition to interviews with family, friends, or coworkers. Psychiatrists are less likely to be found liable if they collect a reasonable amount of information and make an informed clinical decision that turns out to be wrong than if they make an incorrect judgment based on an insufficient database (1).

If mental health clinicians do not carry out their duty to warn or protect when a situation requires it (e.g., imminent threat of serious harm toward an identifiable victim), they have a significant risk of liability if the patient seriously harms someone. Clinicians should know the specific duty-to-protect requirement in their own state (see article by Appel, this issue; 27). In certain jurisdictions, clinicians have been found liable for failing to use reasonable care to prevent foreseeable violence to third parties. Mental health clinicians may face potential liability in the following scenarios: failure to hospitalize a patient with both mental illness and a high violence risk (even in the absence of threats), for example, failing to hospitalize a patient with acute paranoia who has a history of previous violence while psychotic; premature discharge of a patient from a psychiatric hospital who remains at an elevated risk of violence and in need of further treatment, for example, discharging a patient who remains hypomanic and assaultive toward peers; failure to schedule follow-up appointments at an appropriate frequency in the outpatient setting for a patient with a history of recent serious violence; discontinuing psychotropic medications without reviewing the past records of a patient’s prior symptoms and violence history; failure to conduct a violence risk assessment of an individual with homicidal fantasies who denies intent to act on the fantasies; and offering an opinion that a patient who made a serious threat is safe to return to work without a doing a complete violence risk assessment.

Mr. Jones is a 23-year-old man with a history of schizophrenia who was brought to the emergency department after his mother called the police because he barricaded himself in their basement. Mr. Jones refused to open the door because he believed that someone was trying to insert a tracking device into his body. He has been noncompliant with his antipsychotic medication for the past three months and using alcohol to self-medicate. He has a history of previously acting violently in misperceived self-defense because of paranoid delusions. Although his mother asked him to not keep guns in the home, Mr. Jones insisted on keeping a loaded pistol in the basement for self-protection.

When Mr. Jones came to the emergency department, he minimized his symptoms. He told Dr. Smith, the evaluating psychiatrist, that he locked the basement door so his mother would leave him alone while he was watching a movie. He denied hallucinations, depression, anxiety, and thoughts of hurting himself or others. He told Dr. Smith that he just wanted to go back home and watch a movie. Dr. Smith did not speak with Mr. Jones’s mother before discharging him with the recommendation to follow up with his outpatient psychiatrist within one week.

Two hours after discharge, Mr. Jones shot and killed a UPS delivery man who knocked on the front door to deliver a package because of a delusional belief that the UPS driver was a persecutor in disguise. The UPS driver’s family sued Dr. Smith for not hospitalizing Mr. Jones when he was seen in the emergency department.

Although Mr. Jones did not make any threats, it could be argued that he should have been assessed as a high risk for violence. Mr. Jones was in a paranoid psychotic state with a history of previous violence while psychotic, was using alcohol, and had ready access to a lethal weapon. Dr. Smith relied only on the patient’s account and did not gather collateral information. Dr. Smith could face potential liability for not adequately assessing Mr. Jones’s risk of violence and not initiating hospitalization. Even if Mr. Jones declined voluntary hospitalization, Dr. Smith should have considered civil commitment.

Potential liability for failure to schedule more frequent outpatient follow-up visits was raised in the Volk v. DeMeerler case (28). Mr. Jan DeMeerler, a man with a history of bipolar disorder, shot and killed his ex-fiancé, Ms. Schiering, and her nine-year-old son. He also attempted to kill her older son before he committed suicide. Mr. DeMeerler had been in outpatient psychiatric treatment with Dr. Howard Ashby for nine years before the offense. Mr. DeMeerler had a history of previous suicidal and homicidal ideas and a history of erratic behavior, but he had never identified Ms. Schiering or her children as potential victims. Dr. Ashby last saw Mr. DeMeerler in April 2010, approximately three months before the shootings. At that time, Mr. DeMeerler reported having relationship problems with Ms. Schiering, but they were working on repairing their relationship. Mr. DeMeerler had some signs of hypomania, and suicidal ideation when he felt depressed, but he did not express homicidal ideas, and he denied any intent to act on his suicidal ideas. Dr. Ashby continued Mr. DeMeerler’s medication regimen of valproic acid, risperidone, and bupropion (28).

On July 16, 2010, Ms. Schiering ended her relationship with Mr. DeMeerler for good. The next night, Mr. DeMeerler killed Ms. Schiering, her 9-year-old son, and himself. Ms. Schiering’s family filed a lawsuit against Dr. Ashby and his clinic. The trial court granted summary judgment in favor of Dr. Ashby because Mr. DeMeerler had not expressed a threat to harm Ms. Schiering or her sons, and thus Dr. Ashby appeared to be under no legal duty to warn them of potential violence. The decision was appealed. The plaintiff’s expert provided an affidavit that stated that Dr. Ashby breached the standard of care by not inquiring further into Mr. DeMeerler’s suicidal thoughts. The expert opined that line of inquiry could have revealed homicidal thinking. The expert also opined that more frequent follow-up appointments in the months before the murders could have prevented Mr. DeMeerler’s condition from deteriorating and that Dr. Ashby’s negligent treatment was a “causal and substantial factor” in the outcome (28).

The appeals court reversed the summary judgment regarding the medical negligence claim, which allowed the plaintiffs to proceed with the malpractice lawsuit against Dr. Ashby. Dr. Ashby appealed to the Washington Supreme Court, which agreed with the reversal of summary judgment. The Washington Supreme Court held that “Ashby and DeMeerleer shared a special relationship and that special relationship required Ashby to act with reasonable care, consistent with the standards of the mental health profession, to protect the foreseeable victims of DeMeerler” (28). The Washington Supreme Court indicated that whether or not the danger to the victims in this case was foreseeable was a matter of fact to be determined by the trial court (28). Ultimately, the case settled for an undisclosed amount.

This ruling caused anxiety among mental health clinicians because it broadened the scope of potential liability for a patient’s violent acts in the State of Washington despite its Tarasoff-limiting statute. In sum, the Washington Supreme Court said that Dr. Ashby might be liable for the violent acts of his patient that occurred three months after the patient’s last appointment, even in the absence of expressed homicidal ideation toward an identifiable victim.

The Volk v. DeMeerler case illustrates the ever-changing legal landscape of liability for patient violence toward third parties. Hence, it is sometimes difficult to predict when psychiatrists may be held liable and when they will not. In this section, we outline steps clinicians can take to show that reasonably prudent care has been provided. The requirements to meet the standard of care depend on the unique circumstances of each case. The recommendations offered here are aspirational and are not designed to suggest a standard of care.

Two central themes in risk management are identifying the level of risk and addressing it appropriately. The clinician should perform an adequate violence risk assessment if a patient has made threats or exhibited recent violent behavior. However, as in our hypothetical case scenario, threats and violent behavior are not the only signs of an elevated risk of violence. It is useful to gather data regarding static and dynamic risk factors for violence in a systematized fashion. Also, it is useful to inquire about any preparation the patient has made for a violent act (e.g., journaling plans, purchasing a weapon, choosing a location for an attack). The current situation should also be compared with past situations in which a patient has acted violently.

The next step is devising a treatment plan that is appropriate for the patient’s level of risk. Ideally, a treatment plan should address each modifiable risk factor for violence. For example, with a psychotic patient who is noncompliant with medication, consider using a long-acting injectable medication. For substance use disorders, consider the use of a medication to assist with sobriety (e.g., naltrexone) and referral to a substance use treatment program.

Violence risk cannot be reduced to zero during an inpatient stay because many risk factors are static (e.g., prior violence, childhood factors, antisocial attitudes, criminal history), and some dynamic risk factors will require further treatment in the outpatient setting (e.g., substance use disorders, impulsivity, and poor coping skills). However, the most critical modifiable risk factors should be addressed during the hospital stay. For example, if a patient with postpartum depression has a delusion that her baby is a demon, the delusion should be treated and resolved before discharge. The depression should also be treated in the hospital, although it may not fully resolve before discharge. If the patient also has a history of impulsivity and poor coping skills, a plan to address these issues as an outpatient is reasonable.

Once in the outpatient setting, a patient with an increased risk of violence will usually require frequent follow-up appointments to monitor her or his safety. Clinicians should avoid drastic medication changes if they have insufficient knowledge of an individual who has a history of psychosis or severe mood disorder. If the patient were to decompensate and become violent as a result of the medication changes, the psychiatrist could be found liable. This would be especially likely if previous records reveal a pattern of agitation, paranoia, and violence at lower dosages of their medication. If patients with serious risk factors do not show up for their appointments, or are lost to follow-up, reasonable efforts to reach out to them and their family to reengage them in care should be made. These efforts should be documented.

Although violence is difficult to accurately predict, psychiatrists are expected to make reasonable efforts to reduce the risk. While some states have enacted statutes to reduce mental health clinicians’ liability for the violent acts of their patients, courts have varied in their interpretation of these statutes. To reduce the risk of liability, an adequate violence risk assessment should be completed and documented. Noting disavowal of homicidal intent alone may be insufficient. It is much more difficult to prove negligence when a clinician collects sufficient information (including collateral), inquires about access to weapons, and documents the rationale for the treatment plan.