Serious Ethical Violations in Medicine: A Statistical and Ethical Analysis of 280 Cases in the United States from 2008-2016

Abstract

Serious ethical violations in medicine such as sexual abuse, criminal prescribing of opioids, and unnecessary surgeries directly harm patients and undermine trust in the profession of medicine. We review the literature on violations in medicine and present an analysis of 280 cases. Nearly all cases involved repeated instances (97%) of intentional wrongdoing (99%), by males (95%) in non-academic medical settings (95%), with oversight problems (89%) and a selfish motive such as financial gain or sex (90%). Over half of cases involved a wrongdoer with a suspected personality disorder or substance use disorder (51%). Despite clear patterns, no factors provide readily observable red flags making prevention difficult. Early identification and intervention in cases requires significant policy shifts that prioritize the safety of patients over physician interests in privacy, fair processes, and proportionate disciplinary actions. We explore a series of ten questions regarding policy, oversight, discipline, and education options. Satisfactory answers to these questions will require input from diverse stakeholders to help society negotiate effective and ethically balanced solutions.

This paper represents the culmination of seven years of research on serious ethical violations in medicine. Serious ethical violations are acts that not only disregard codes of medical ethics, but also risk directly harming patients and subjecting the wrongdoer to criminal, tort, or medical board actions. In our studies, we focused on three kinds of violations: Improper prescribing of controlled substances (IPCS), sexual abuse of patients (SAP), and unnecessary invasive procedures (UIP). Though they may appear disparate, as we explain below, they involve similar dynamics. We focused on these three kinds of violations because they are among the most common causes of major disciplinary actions by medical boards such as revoking licenses (Arora, Douglas, and Dorr Goold 2014, Harris and Byhoff 2017) and, above all, in contrast say to upcoding or failures to disclose conflicts of interests, which are also violations of medical ethics, these three behaviors directly harm patients (Krause Winter 2012, DuBois, Walsh, et al. 2017). The violations we studied contributed to opioid addictions and sometimes overdose deaths; emotionally traumatized patients through sexual abuse by trusted physicians; and caused physical, emotional and financial injury to patients through the performance of unnecessary procedures such as spinal surgeries that benefitted the physicians financially.

When such events attract media attention, they also harm the reputation of medicine. When such events are prolonged or covered up—as in the recently highly publicized cases of Larry Nassar, a physician accused of molesting at least 265 young gymnasts over a 20-year period (Connor 2018, Eggert 2018) and George Tyndall, who made lewd comments and conducted inappropriate intimate exams on at least 52 women across a 26-year period at the University of Southern California despite repeated complaints against him (Medina 2018)—they also rightly diminish the trustworthiness of the healthcare professions and associated institutions. This in turn can discourage patients from seeking the care that they need or complying with physician recommendations (Khullar 2018a).

In section I of this paper, we describe prior research by others on the problem of serious ethical violations in medicine, including data on frequencies, harms to patients, and correlates of violations. These data are surprisingly limited. In section II, we explain the challenges of understanding why and how serious ethical violations occur in medicine. In light of these challenges, we describe in section III the case-based approach our team has taken over the past seven years. Whereas our past publications have each focused on one specific kind of ethical violation, in section IV we offer an analysis of our entire dataset of 280 cases to identify factors that are common to most cases, as well as factors that divide cases into distinct groups or typologies. Finally, in section V, we pose a series of questions about the policies that govern medical education, the oversight of medical practice, and responses to observed violations. In section V, we argue that the balance of moral considerations currently favors the fair treatment of physicians who are accused of ethical violations over the protection of patients, and some degree of correction is needed. We raise a series of questions about potential changes in policies and practices regarding physician oversight, discipline, and education that might reduce rates of serious ethical violations in medicine.

I. Prior Research on Ethical Violations in Medicine

Most research on serious ethical violations in medicine has focused on frequencies and demographic correlates of violations (e.g., physician gender and specialty). A smaller body of work examines the impact of serious ethical violations on patients. No studies have developed causal models that aim to understand the factors that motivate and enable such violations to occur.

Frequency

Only a minority of physicians commit serious ethical violations. According to the Federation of State Medical Boards (FSMB), 0.5% of physicians are the subject of state medical board (SMB) disciplinary action each year, of which approximately 0.1% are severe actions involving revocation, suspension or surrender of license (Harris and Byhoff 2017, Peachman 2016, Federation of State Medical Boards 2014). On the one hand, the number of severe disciplinary actions—approximately 1 in 1000 per year—is reassuringly small. On the other hand, it is similar to the annual incidence of breast cancer diagnoses, and much larger than the incidence of HIV cases (National Center for Health Statistics 2017); no one would deny that these are urgent public health concerns. Moreover, as we explain in section II, the existing frequency data yield extreme underestimations of the rate of serious ethical violations, and more radical underestimations of specific forms of ethical violations.

A small number of physicians account for the majority of IPCS. CDC data indicate that 3% of physicians account for 62% of prescribed opioids (Casturi 2013). In South Florida, 45 physicians dispensed 9 million oxycodone pills in the last six months of 2008 (Rigg and Murphy 2013). Physicians are the second leading source, after friends and family, of prescription opioids used by abusers for non-medical reasons (Jones, Paulozzi, and Mack 2014). Despite this, the majority of attention regarding opioid abuse to date has focused on addicted patients who seek out pills surreptitiously or on inappropriate prescribing of opioids more generally, rather than egregious intentional over-prescribing by a small number of physicians to a large number of patients through “pill mills.” Most state prescription drug monitoring programs (PDMPs) are used to track patients requesting opioids and pharmacy dispensing patterns, rather than as a tool to track inappropriate prescribing by physicians. In our study of 100 cases of IPCS, not one case was discovered by a PDMP, despite the fact that PDMPs were in place when a majority of the cases occurred (DuBois et al. 2016a).

Much of the available data on SAP reflects less severe actions like sexual impropriety or boundary violations (e.g., improper comments or watching a patient undress), making it difficult to tease out the most egregious actions (e.g., sodomy, child molestation, and rape). Between 2003 and 2013, 2.9% of all national practitioners database (NPDB) licensure reports were for sexual misconduct (AbuDagga et al. 2016). Much of the data on sexual violations relies on self-report of sexual contact with patients, which ranges from 3–12% for male physicians and 1–4% for female physicians (Carr 2003, Sansone and Sansone 2009). In one survey of 10,000 physicians from 1992, 9% self-reported sexual contact with patients, yet 23% of the same sample reported knowing another physician who had sexual contact with a patient (Gartrell et al. 1992). Patients rarely report sexual misconduct; only 5–10% of victims of physician sexual assault report it, a rate of reporting that is lower than sexual assaults in the general population (Tillinghast and Cournos 2000, Carr 2003). A smaller study of 49 convicted physicians in California found that nearly half were likely to be reoffenders, illustrating the ongoing and repeated nature of many of these violations (Perez 2013).

Data on UIP is the most limited. This is partly due to the challenge of disentangling intentional and egregious violations from the overall data on unnecessary tests, procedures, or interventions by physicians (such as inappropriate screening or antibiotic prescription), as they are often calculated together. In 2016, the Department of Justice opened 975 criminal investigations of health care fraud, and filed criminal charges in 480 cases, yet no data are available on how many cases involved UIP as opposed to financial fraud alone (U.S. Department of Health and Human Services and Department of Justice 2016). In most cases of UIP, patients would have no way of knowing that the procedure was unnecessary unless their charts were reviewed by another specialist; and even then, UIP sometimes involves falsifying charts to support the necessity of procedures.

Harm to Patients

We focused on ethical violations that directly harm patients, though each of the three violations we studied inflict harm in distinct ways. From 1999 to 2016, more than 200,000 people in the U.S. died from overdoses involving prescription opioids (Centers for Disease Control and Prevention 2016). Opioid addiction also has clinical and economic costs. Compared to matched control groups, opioid abusers have higher frequency of use of medical services (e.g., Emergency Department, outpatient visits, and inpatient stays), with consequent excess health care costs (Meyer et al. 2014). The problem of prescription opioid addiction is complex and the causes multifactorial. However, when physicians abandon responsible practices for prescribing opioids and other controlled substances, they prey upon individuals with addictions and risk contributing to their deaths.

SAP causes significant emotional and psychological harm. The psychological consequences include depression, anger, drug and alcohol abuse, mistrust, and posttraumatic stress symptoms (Carr 2003). Victims of sexual assault often report emotional injury manifested in feelings of exploitation and betrayal that can impair future relationships with physicians (AbuDagga et al. 2016, Gartrell et al. 1992, Federation of State Medical Boards 2006). Even when sexual activity appears “consensual,” it may still result in mental and emotional harm to the patient (Carr 2003).

UIP leads to emotional, economic, and financial harms caused by unnecessary hospital stays, follow up procedures, overcharging, and excessive recovery or rehabilitation costs (Krause 2010, Buck 2016). The economic harms extend beyond the individual to all of society through potentially higher taxes and more expensive insurance premiums (Kyriakakis 2015). At the most extreme end, UIP can lead to serious physical harm and even death. One study of 1361 insurers registry claims found that 44% of cardiac catheterization litigation cases resulted in death, while a much smaller study of 19 pancreatoduodenectomy malpractice cases found that 47.6% resulted in death (Kim and Vidovich 2013, Anandalwar et al. 2017). Kyriakakis (2015) argues that failure to account for patient harm when pursuing cases of fraud contributes to skewed punishments, inadequate restitution for patients, and the perception of fraud as only white collar crime.

Correlates of Physician Misconduct

Four factors have been associated with severe disciplinary actions against physicians across multiple studies: Being male, over 45 years old, non-board certified, and having trained outside the United States (Khaliq et al. 2005, Studdert et al. 2016, Kohatsu et al. 2004, Arora, Douglas, and Dorr Goold 2014, Goldenbaum et al. 2008). However, no data exist to explain why these factors are associated with severe disciplinary actions.

An analysis of 34 formal disciplinary actions found solo practice to be a correlate (41% of cases), but this analysis was based on a small sample (Plaut et al. 2013). Additional data are available from studies of physicians referred to remediation programs or who are subjects of less severe actions by medical boards. Some of these studies have found correlations between misconduct and psychological and personality traits, including a history of child physical or sexual abuse, trauma, depression, and substance abuse (Finlayson et al. 2013, Brooks et al. 2012, MacDonald et al. 2015, Roback et al. 2007, Samenow et al. 2012, Spickard et al. 2008). Additional studies suggest a correlation between prior grades and test scores, including MCAT, Step 2 Clinical Knowledge, and professionalism scores, and future disciplinary action (Cuddy et al. 2017, Papadakis et al. 2008, Papadakis et al. 2004). These studies address markedly different and less severe outcomes for physicians, and Prasad (2011) warns that the link between medical school behaviors and future conduct is weak and could lead to unfair and arbitrary screening.

II. The Challenge of Understanding Serious Ethical Violations

Understanding serious ethical violations in medicine is essential to preventing future occurrences. If we understand the motives for violations, in principle, we might change reward systems or identify situations that pose heightened risks to patients. If we understand the environmental factors that provide the opportunity for violations, we can change practice environments or increase monitoring. However, gaining an understanding of motives and environment is hindered by incomplete, vague, and unavailable data regarding frequency, causes, and the impact of serious ethical violations. For example, in approximately 62% percent of NPDB misconduct reports from 2010–2014, no useful information was provided on the cause of action: 20% contained no classification of the cause of action; 30.4% were classified as “not specified”; and 11.2% as “other” (Harris and Byhoff 2017). In a separate study, an estimated 70% of physicians disciplined for sexual misconduct were not classified as sexual offenders in public databanks (Ernsthausen 2016). Essentially, in the vast majority of cases, the few groups that have access to NPDB data typically learn only that action was taken against a physician; they are unable to learn why action was taken.

Moreover, the vast majority of cases of serious ethical violations are never reported to SMBs or the NPDB. Patients frequently do not report serious ethical violations. As noted above, only 5–10% of victims of sexual assault report it (Ernsthausen 2016, Tillinghast and Cournos 2000, Carr 2003). This is lower than the overall U.S. rate of reporting sexual assault or rape to law enforcement (U.S. Department of Justice 2016). In cases of UIP, patients typically do not know that the invasive procedures performed were unnecessary (DuBois, Chibnall, et al. 2017b). In cases of IPCS, patients frequently have substance use disorders and seek the very drugs that are inappropriately prescribed to them (DuBois et al. 2016a). Additionally, when institutions become aware of serious ethical violations, they do not consistently report them to SMBs or the NPDB, despite reporting requirements. For example, in the recent USC case, the misbehavior of Dr. Tyndall was treated as an internal personnel matter rather than a matter for the medical board (Medina 2018). A recent review of over 100,000 disciplinary documents dating back to 1999 found that sexual misconduct is often not reported to regulators due to the following factors: SMBs can choose to issue a private letter or enter into confidential agreements with physicians; physicians can enter into remediation programs to avoid public sanctions; boards may be reluctant to investigate a patient complaint due to time or resource constraints; and investigations may be halted when physicians voluntarily agree to surrender their licenses (Norder, Ernsthausen, and Robbins 2016). State medical boards have six-fold variation by state in disciplinary actions taken; features such as having more board members and independence from regional government lead to increased disciplinary actions by boards (Harris and Byhoff 2017).

In accord with federal regulations (45 CFR Part 60), the NPDB is not accessible to the public, which includes investigative journalists and researchers. Upon request, limited datasets may be made available. However, they are stripped of crucial information. This explains why a recent study of sexual misconduct was unable to provide even the most basic information about cases such as physician sex, specialty, or practice setting (AbuDagga et al. 2016).

To summarize, SMBs and the NPDB receive reports on a minority of cases of serious ethical violations; the data they do have are incomplete and frequently vague; and only a very limited subset of data is made publicly available. Moreover, data from other sources such as surveys are typically more than 10 years old and rely on self-report, which may not yield trustworthy estimates of behaviors that people typically deny because punishments can be severe. None of the readily available datasets provide details on the nature of the wrongdoing, the physician wrongdoer, and the practice environment; hence, no published papers have presented data-driven causal models.

III. Our Approach to Understanding Violations

As noted above, most previous studies of serious wrongdoing in medicine have focused either on determining prevalence or correlates such as physician gender, age, specialty, nation of training, and practice settings. No studies have tried to provide a causal analysis of serious wrongdoing in medicine. One reason seems obvious: It would be unethical to design a prospective trial aimed at testing hypotheses about what causes serious violations. One could not, for example, randomize physicians to solo practices or academic medical practices, and then use confederates to approach physicians with drug diversion schemes. For this reason, we developed causal theories using a mixed-methods case analysis approach.

In a series of recent papers, we reported results from our investigations into the three kinds of wrongdoing. In the first paper, we reported on 100 cases of IPCS (DuBois et al. 2016a); in the second, on 101 cases of SAP (DuBois, Walsh, et al. 2017); and in the third on 79 cases of UIP (DuBois, Chibnall, et al. 2017b). In these three papers, we provide extensive details on our methodological approach. Here we provide a brief overview, which is summarized in Figure 1.

Figure 1.

Mixed-Method Case Research Process

Our methodological approach in each of these papers involved five steps. First, we identified potential cases of serious ethical violations through systematic literature reviews in Lexis-Nexis Law, which includes not only law reviews, but records from courts, medical boards, and newspapers. Second, we conducted a new review of the literature with the physician’s name as the key search term using a broader set of databases including Google, Lexis-Nexis Law, and websites of state medical boards, state circuit courts, HealthGrades, the American Board of Medical Specialties’ Certification Matters, and the U.S. Office of the Inspector General. We consulted an average of 22 sources per case. Third, the case literature was descriptively coded using a detailed codebook. Fourth, we used a criminal law framework to identify variables that provided motive, means, and opportunity (MMO) for the ethical violation. Historical methods that move from effects (e.g., an ethical violation) to causes must be driven by theory (George and Bennett 2005, Bennett and Elman 2006); in our case, this was the MMO criminal law theory. We selected this theory because we assumed that serious ethical violations in medicine, like other criminal behaviors, are motivated behaviors that can occur only when the wrongdoer has means and opportunity. In general, our causal coding focused on motives and opportunities, rather than means, for the simple reason that all physicians—often somewhat uniquely—have the means of committing the kinds of wrongdoing we studied: They can prescribe controlled substances; they can ask patients to undress and examine them alone behind closed doors; and they can control the information provided to patients during informed consent, perform invasive procedures, and bill for these procedures. All codes for motives and opportunities are defined and illustrated with UIP examples in Table 4 and Table 5 respectively in our open-access publication on UIP (DuBois, Chibnall, et al. 2017a). Finally, within each type of ethical violation, we engaged in cross-case analysis to identify clusters or typologies of cases. Typologies are created by examining how the variables that are theoretically linked to the occurrence of cases cluster together in meaningful ways (Elman 2005). For example, using qualitative analysis and hierarchical cluster analysis, we identified three primary typologies of IPCS: Cases involving greed as the primary motive plus a lack of oversight (usually the physicians were in solo practice); cases involving physicians with substance use disorders who diverted drugs for their own use; and physicians who demonstrated poor judgement or skills (DuBois et al. 2016b).

Table 4.

Case Typologies with General Characteristics: Duration, Investigation, and Consequences

Typology	1 (n=84)	2 (n=44)	3 (n=14)	4 (n=10)	5 (n=17)	6 (n=80)	7 (n=15)	Stats
Kind of Wrongdoing	UIP or IPCS	UIP or IPCS	UIP or IPCS	UIP or IPCS	IPCS	Sex Abuse	Sex Abuse
Main Motive	Greed	Greed	Greed	Problem- solving /	Substance Abuse	Non-Consensual Sex	Sex
Main Environmental	Lack of Oversight With No Corrupt Climate	Oversight Failure	No Oversight Problems	Ambiguous Norms	Misc.	Lack of Oversight	No Lack of Oversight
Patient died	21.4% (18)	38.6% (17)	21.4% (3)	30.0% (3)	29.4% (5)	0	0	V=.38***
Duration ≥2 years	73.8% (62)	88.6% (39)	78.6% (11)	70.0% (7)	52.9% (9)	60.0% (48)	53.3% (8)	V=.25*
Wrongdoing Repeated	98.8% (83)	95.5% (42)	100%	100%	100%	95.0% (76)	100%	V=.13
Wrongdoing in >1 Environment	11.9% (10)	56.8% (25)	71.4% (10)	10.0% (1)	47.1% (8)	21.3% (17)	33.3% (5)	V=.43***
Number of victims >10	22.8% (18/79)	58.3% (21/36)	58.3% (7/12)	10.0% (1)	0	20.0% (16)	46.7% (7)	V=.38***
Criminal Investigation	90.5% (76)	70.5% (31)	50.0% (7)	50.0% (5)	76.5% (13)	93.8% (75)	86.7% (13)	V=.36***
Board Investigation	92.9% (78)	72.7% (32)	78.6% (11)	100%	100%	92.5% (74)	100%	V=.29**
Lost license	83.3% (70)	61.4% (27)	64.3% (9)	90.0% (9)	100%	86.3% (69)	93.3% (14)	V=.29**
Discontinued Medical Practice	79.8% (67)	61.4% (27)	57.1% (8)	70.0% 7)	88.2% (15)	73.8% (59)	80.0% (12)	V=.19
Prison Sentence	57.1% (48)	38.6% (17)	42.9% (6)	30.0% (3)	52.9% (9)	56.3% (45)	53.3% (8)	V=.16

Note: 16 cases (5.7%) were unclassifiable and are not represented in the table. Cramer’s V indicates strength of association between two categorical variables.

^*p < .05;

^**p < .01;

^***p < .001. Oversight failures include cases with and without corrupt moral climate. Data on number of victims unavailable in 15 cases.

IV. A Mixed-Methods Analysis of 280 Cases

In this paper, we advance our work on serious ethical violations by examining our entire dataset of 280 cases with two main goals:

1. Identifying Cross-cutting Factors. We examine factors that are prevalent across all three types of serious ethical violations (IPCS, SAP, and UIP).

2. Developing Cross-cutting Typologies. Our previous efforts to understand IPCS, SAP, and UIP involved building theories of why cases occurred by identifying MMO variables within a given sampling area. In this paper, we present typologies of cases by examining how the 280 cases group into mutually exclusive categories when sorted by primary motive and opportunity factors, and then illustrate how secondary motives, physician traits, and diverse environmental variables differ across these typologies.

What Factors Cut Across Most Cases?

The kinds of violations we studied across the 280 cases were very different. Yet, despite their heterogeneity, most cases shared certain striking features. Nearly all cases involved repeated instances (97%) of intentional wrongdoing (99%), by males (95%) in non-academic medical settings (95%), with oversight problems (89%) and a selfish motive such as financial gain or sex (90%). Over half of cases involved a wrongdoer with a suspected Cluster B personality disorder (antisocial or narcissistic) or substance use disorder (51%). In 70% of cases, the actual wrongdoing persisted for 2 or more years; 33% persisted for 5 or more years.

At least twenty percent of cases involved an ignored report of the violations, which enabled cases to last longer than necessary; and an additional 23% involved a failed opportunity to blow the whistle; that is, supervisors, peers or co-workers observed the wrongdoing and did not intervene. We caution, however, that these are likely underestimations—unlike physician sex or specialty, our data depended on such information coming to light along with the violation itself. A higher than expected percentage of physicians were not board certified (44%) and completed their medical degrees outside of the United States (32%). Table 1 compares the physicians in our dataset to national samples of physicians on key variables such as board certification and nation of training. In this regard, our study confirms what prior studies have found: Physicians who are subject to disciplinary actions, when compared to national averages, are more likely to be male, not board certified, older than 40, and trained outside the U.S. (Khaliq et al. 2005, Studdert et al. 2016, Kohatsu et al. 2004, Arora, Douglas, and Dorr Goold 2014, Goldenbaum et al. 2008).

Table 1.

Comparison of physician wrongdoer descriptive data to historical data

Variable	Historical+	IPCS (N = 100)	UIP (N = 79)	SAP (N = 101)
Non-Academic Setting	93% (The Physicians Foundation by Merritt Hawkins 2010)++	98.0%	92.4%	94.1%
Solo Practice	13% (Association of American Medical Colleges 2012)	62.0%	17.7%	38.6%
Board Certified	70% (Young et al.)	37.0%	70.9%	30.7%
Age >49	47% (Young et al.)	62.0%	49.4%	56.7%
Male	69% (Association of American Medical Colleges 2012)	88.0%	96.2%	100%
Born Outside USA	27% (McCabe 2012)	16.0%	27.8%	15.8%
Trained Outside USA	24% (Association of American Medical Colleges 2012, Young et al.)	32.0%	40.5%	25.7%
Suspected Antisocial	3.9-5.8% in general population (Compton et al. 2005) 35-47% in prison population (Fazel and Danesh 2002, Black et al. 2010)	57.0%	48.1%	31.7%

⁺Whenever possible, we provide historical data using statistics from 2010 – 2012 to mirror the average dates our cases ended.

⁺⁺According to AAMC, 84% of physicians who completed a residency work in private practice versus academic medicine (Association of American Medical Colleges 2016); however, overall—including those who did not complete residency program—the rate appears to be 93%.

Some commonly hypothesized predictors or correlates of professional violations rarely or never appeared in cases: playing conflicting roles such as physician and administrator, researcher, or educator (0%); retaliation (0%); professional ambition (1%); stress (0%); carelessness (1%); and serious mental illness such as major depression, bipolar or schizophrenia (1%) (as opposed to personality disorders such as antisocial or narcissistic personality, substance use disorders, and paraphilias such as Frotteurism or pedophilia). Very rarely were norms ambiguous, where practice guidelines or ethical codes were divided or silent on the behavior or practice that constituted the violation (2% overall; 5% in UIP).

In all of the sampling areas, the most common motives were self-centered: financial gain, sex, or drugs. In rare cases, the physician appeared incompetent (e.g., was unfamiliar with current standards for opioid prescribing or for a specific procedure) or standards were ambiguous (e.g., some respectable experts disagreed with current standards). This is, of course, a byproduct of the cases we investigated: High-profile cases of serious violations. In each of our sampling areas, physicians have the means for committing the violations: They have the right to prescribe controlled substances, they have the authority to ask another person to undress and to conduct intimate exams, and they have the authority to perform and bill for invasive procedures. Opportunity was typically afforded by a lack of oversight (e.g., practicing medicine in a solo practice) or oversight failures (e.g., administrators who ignored whistleblowers or a morally corrupt climate).

Seven Typologies of Serious Ethical Violations in Medicine

While cases had much in common, they can be grouped in different ways. One approach, which we used in our prior analyses, is to focus on the kind of wrongdoing (IPCS, SAP, or UIP). In this paper, we chose a different approach because often the causal factors that contributed to cases cut across kinds of wrongdoing. Greater insight can be gained by recognizing typologies based on the primary motive and the primary environmental factors that provided opportunity. (We defined a motive or environmental factor as secondary when it never occurred except in the presence of another motive or environmental factor that provided opportunity.) Adopting this approach, we identified seven non-overlapping typologies of cases. In developing typologies, five motives met our criteria for being primary: Greed, sex, non-consensual sex, substance use, and poor problem-solving. Cases that lacked such a motive were labeled outliers. We separated the motives “sex” and “non-consensual sex” because the literature on sexual abuse indicates that acts of child molestation, fondling, sodomy, and rape are often motivated by paraphilias or violence rather than so-called normophilic sex (American Psychiatric Association 2013). Suspected Cluster B personality disorders were never present alone as a motive, but only with greed or sex, so suspected presence of a personality disorder was excluded as a sufficient cause. We identified three primary environmental problems that provided opportunity for the violations: lack of oversight, oversight failure (with or without corrupt moral climate), and ambiguous norms. We also recognized typologies without any one type of environmental factor that played a causal role.

All other variables characterizing the physician-wrongdoer or the environment were treated as co-variates. We used the Cramer’s V statistic to determine whether and how much the variable differed across typologies. Table 2 presents the seven typologies with case examples.

Table 2.

Typologies with Sample Cases

Typology	Motive	Environment	Wrongdoing	Sample Cases – Published Allegations
1 (n=84)	Greed	Lack of Oversight with No Corrupt Moral Climate	IPCS or UIP	• Manzella, an infectious disease doctor, wrote fraudulent prescriptions to obtain thousands of oxycodone pills for a New Jersey con man who sold them on the black market. Greed appeared to be the motive. Manzella worked with a lack of oversight in a private practice with his brothers who did not know about the scam (Hall 2013). • Wasserman, a dermatologist, performed thousands of unnecessary biopsies and adjacent tissue transfers on Medicare patients in order to obtain reimbursement. Wasserman settled with the DOJ for $26.1 million. He owned his own practice and operated without peer scrutiny (U.S. Department of Justice 2013).
2 (n=44)	Greed	Oversight Failure	IPCS or UIP	• Glaser performed at least 90 unnecessary cardiac catheterizations for financial gain. Nurses complained that Glaser was sleeping in patient rooms and operating on patients 18 to 21 hours a day. Their complaints were ignored by hospital administrators (Holdren 2015, White 2015, 2016). • Buckwalter saw 80 patients a day (the average family physician sees 19). About half came in only for prescription refills (mostly opioids); many paid cash. Some flew in from out of state for prescriptions; he mailed opioid prescriptions without examining patients. His employees saw what he was doing; no one blew the whistle (Allen 2015, 2008).
3 (n=14)	Greed	No Oversight Problems	IPCS or UIP	• Makker performed medically unnecessary and excessive spinal fusion surgeries on predominately Medicare patients. His rate was nearly 10 times the national average (Carreyrou and McGinty 2011). No other parties were reprimanded or fined. • Williamson wrote oxycodone prescriptions to patients without a legitimate medical purpose. The pills were charged to Medicaid and sold by others for $30 and $40 each. The investigation began after a patient was arrested and reported Williamson to the police (U.S. Department of Justice 2012). No other parties were reprimanded or fined.
4 (n=10)	Problem-Solving, No Greed	Ambiguous Norms ^	IPCS or UIP	• McIver prescribed excessive amounts of painkillers to patients with chronic pain or addiction. Some of these patients sold their drugs. McIver took no steps to either monitor potential addiction or report suspected addicts. His patients who diverted or abused their opioids all testified that they received their prescriptions by lying to him. McIver wanted his patients to live relatively pain free, but his opioid prescribing patterns were not consistent with good practice. (Rosenberg 2007). • Usman, a family practitioner, was an advocate for the Defeat Autism Now! protocol. She turned to alternative medicine after a child patient died due to a peanut allergy. She treated children with risky infusion drugs that leach heavy metals. One 5-year old child died from the IV push being administered too quickly and the dose being too large. Usman continued to prescribe the treatment even after this patient’s death (Circuit Court of Cook County 2011, State of Illinois Department of Financial and Professional Regulation 2012).
5 (n=17)	Substance Use	Miscellaneous	IPCS	• Basham-Callaway was an osteopathic physician who practiced in two pain clinics. She regularly prescribed oxycodone and Adderall for cash and would obtain some of these drugs for her own use (Clevenger 2010, U.S. Department of Justice 2011). • Nally wrote prescriptions for patients who would then fill them and return half of the filled prescriptions to Nally for her use (Commonwealth of Kentucky 2012).
6 (n=80)	Non-consensual Sex	Lack of oversight	SAP	• Alur, a general and family practice physician, was a co-owner of the practice where he performed non-consensual oral sex on patients. He told the patients that they did not need to pay for their visits (Commonwealth of Kentucky 2015). • Pediatrician Blankenburg co-owned a practice with his brother where he gave drugs and money to adolescent boys after he performed sex acts on them to convince them to stay quiet (2012).
7 (n=15)	Sex	No Lack of Oversight	SAP	• General practitioner Hung Do fondled the breasts and vaginas of several patients; the touching had nothing to do with appropriate examinations. Some instances occurred in front of a chaperone or family member (Guisti 2009). • Walden was a prison physician who performed unnecessary prostate and rectal exams (sometimes ungloved) and fondled prisoners; the Warden ignored complaints, and the physician staffing company just transferred him to new prisons (Prieskop 2015).

IPCS=Improper prescribing of controlled substances; SAP=sexual abuse of patients; UIP=unnecessary invasive procedures. Sixteen cases (5.7%) were unclassifiable.

^{^}In taxonomy 4, cases were included if they involved either poor problem-solving or ambiguous norms; in all other taxonomies, both the motive and environmental factor were always involved.

Table 3 presents the seven typologies with the frequency of physician and practice environment characteristics found in each. A few variables did not differ significantly across the typologies—physician age, physician experiences of personal problems such as divorce or bankruptcy at the time of the violations, being in academic versus non-academic settings, nation of origin (which is distinct from nation of training), and whether the patient victims were particularly vulnerable. We assume that all patients are vulnerable due to health concerns and the differences between patients and physicians in terms of power and knowledge of medicine. Accordingly, we focused on the characteristics that the National Bioethics Advisory Council identified as creating vulnerabilities such as cognitive impairments (National Bioethics Advisory Commission 2001) or belonging to a group with special legal protections—e.g., mandatory reporting—such as children and older adults. Other factors varied significantly across the typologies. Cases that were motivated by poor problem-solving or by the physician’s substance use disorders involved more frequent descriptions of physicians with poor skills and greater gender diversity. Women were far more likely to be targeted as victims in both typologies involving sexual abuse. Cluster B personality disorders (e.g., antisocial or narcissistic) were suspected more frequently in cases motivated by greed than in other cases. Cases motivated by greed and enabled by oversight failures were more likely to involve accomplices rather than lone actors. Primary care physicians were more likely to be involved in non-consensual sex abuse cases and cases involving diversion of drugs for their own use. Financial conflicts of interest—of the sort one would need to report to a financial conflict of interest committee, such as relationships to industry—were theorized to play a contributing, causal role in only 3.9% of cases overall (n=11) and were more common in cases involving UIP.

Table 3.

Case Typologies with Physician and Environmental Characteristics

Typology	1 (n=84)	2 (n=44)	3 (n=14)	4 (n=10)	5 (n=17)	6 (n=80)	7 (n=15)	Stats
Kind of Wrongdoing	UIP or IPCS	UIP or IPCS	UIP or IPCS	UIP or IPCS	IPCS	Sex Abuse	Sex Abuse
Main Motive	Greed	Greed	Greed	Problem- solving /	Substance Abuse	Non-Consensual Sex	Sex
Main Environmental	Lack of Oversight With No Corrupt Climate	Oversight Failure	No Oversight Problems	Ambiguous Norms	Misc.	Lack of Oversight	No Lack of Oversight
Male	91.7% (77)	100%	92.9% (13)	70.0% (7)	82.4% (14)	100%	100%	V=.32***
Age >49	57.8% (48/83)	55.8% (24/43)	50.0% (7)	60.0% (6)	47.1% (8)	59.5% (47/79)	42.9% (6/14)	V=.09
Accomplice	46.4% (39)	77.3% (34)	35.7% (5)	30.0% (3)	47.1% (8)	1.3% (1)	0	V=.58***
Suspected Personality Disorder	70.2% (59)	59.1% (26)	35.7% (5)	0	0	36.3% (29)	13.3% (2)	V=.46***
Primary Care	40.5% (34)	15.9% (7)	7.1% (1)	40.0% (4)	58.8% (10)	58.8% (47)	53.3% (8)	V=.35***
Financial Conflict of Interest	3.6% (3)	11.4% (5)	21.4% (3)	0	0	0	0	V=.29**
Board Certified	47.6% (40)	59.1% (26)	64.3% (9)	50.0% (5)	52.9% (9)	26.3% (21)	40.0% (6)	V=.26**
Trained Outside US	35.7% (30)	38.6% (17)	42.9% (6)	10.0% (1)	35.3% (6)	20.0% (16)	60.0% (9)	V=.24*
Poor Skills	16.7% (14)	13.6% (6)	14.3% (2)	40.0% (4)	35.3% (6)	6.3% (5)	13.3% (2)	V=.24*
Other Illegal Behavior	27.4% (23)	9.1% (4)	14.3% (2)	0	41.2% (7)	20.0% (16)	20.0% (3)	V=.22*
Personal Problems	13.1% (11)	4.5% (2)	21.4% (3)	20.0% (2)	23.5% (4)	7.5% (6)	0	V=.20
Vulnerable Group Targeted	63.1% (53)	56.8% (25)	35.7% (5)	90.0% (9)	52.9% (9)	52.5% (42)	73.3% (11)	V=.20
Women Targeted	1.2% (1)	0	7.1% (1)	0	17.6% (3)	88.8% (71)	86.7% (13)	V=.87***
Private Practice /Non-Academic	97.6% (82)	90.9% (40)	92.9% (13)	100%	100%	92.5% (74)	100%	V=.15
Born Outside US	20.2% (17)	22.7% (10)	28.6% (4)	10.0% (1)	17.6% (3)	13.8% (11)	33.3% (5)	V=.14

Note: 16 cases (5.7%) were unclassifiable and are not represented in the table. Cramer’s V indicates strength of association between two categorical variables.

^*p < .05;

^**P < .01;

^***p < .001. Primary care = Pediatrics, family, internal medicine, geriatrics. Oversight failures include cases with and without corrupt moral climate. Age data unavailable for 4 cases.

Table 4 presents the seven typologies with case duration, consequences, and disciplinary actions. In cases involving UIP or IPCS, patient deaths occurred in 21–39% of cases, and never occurred in the sex abuse cases. Across the seven typologies, there were no differences in prison sentences, discontinuing the practice of medicine, or whether the wrongdoing was repeated. All typologies involved a majority of cases that lasted 2 or more years, but those that included UIP were even more likely to last 2 or more years. The physician was less likely to be criminally prosecuted when the institution had oversight failures or when practice norms were ambiguous. Cases involving ambiguous norms or that were motivated by the physician’s substance use were less likely to have more than 10 patient victims.

Limitations

Our project adopted a mixed-method, case-based approach using publicly available cases of serious ethical violations by physicians. This approach uniquely enabled both a rich description of cases and the development of theory-guided causal modeling across different types of cases. Nevertheless, such an approach has limitations. Our 280 cases represent either 100% (SAP and UIP) or a random selection (IPCS) of all cases publicly available through our search approach; however, it is not possible to obtain a representative sample of cases for reasons outlined above. Thus, while the project identified seven valid typologies, we can make no inferences regarding how frequently they occur, even relative to each other. Experimental methods are not appropriate for the study of serious ethical violations; in the absence of controlled research, it is difficult to account for all potentially explanatory and confounding variables. Also, because we relied on published accounts of cases, our determination that a variable was absent or present generally depended on whether the variable was mentioned in accounts. Some information was routinely available through public databases, e.g., a physician’s gender, specialty, and status as board certified. But other variables—such as the number of victims or whether there were failed attempts to blow the whistle—were almost certainly underreported in the literature and accordingly underrepresented in our data.

V. Exploring the Ethical and Policy Implications of Data on Violations

In a seminal article, “Public Health Ethics: Mapping the Terrain,” Childress and colleagues note that “no ethical principle can eliminate the fact that individual interests must sometimes yield to collective needs” (p. 175). We would argue that there is a collective need to protect patients from serious ethical violations in medicine. While such violations are rare, as we noted above, documented cases are approximately as common as new incidences of breast cancer and far more common than HIV, and we know cases of violations are significantly underreported. At present, the balance of moral considerations has favored the fair treatment of physicians over the protection of patients. It errs on the side of physicians in many ways: by providing second chances to those who commit offenses; by protecting the confidentiality of records documenting past violations; by permitting physicians to practice in isolation from peers, even when their training and demographics suggest increased risk of wrongdoing; by permitting physicians to self-regulate, with physicians constituting the majority of members of state medical boards (Arora, Douglas, and Dorr Goold 2014, p. 53). We would argue that the data presented in this paper suggest that the field of medicine has self-regulated in a manner that protects self-interests above patient interests. To be clear, the self-interests are legitimate interests; this is why we face a serious ethical dilemma that requires engagement from bioethicists and a broad array of stakeholders including physicians and the general public.

Policies that effectively reduce the frequency of serious ethical violations of medicine will require some level of infringement on physicians’ interests in protecting their reputation from the damage that can be done by false accusations, ensuring due process, and preserving professional autonomy. In part, this is because it is so difficult to prevent initial instances of serious ethical violations. Although some physician and practice characteristics were far more common in our sample than in the general population of physicians, there is no way to identify a perpetrator through profiling because the vast majority of physicians who might fit a profile (e.g., male, over 40, not board certified) never commit a serious ethical violation, and some who do not fit the profile nevertheless offend. Moreover, Cluster B personality disorders and the paraphilias that can motivate some forms of sexual abuse are not always evident to colleagues and oversight officials. Rather, they manifest through patterns of behavior, patterns that may only become apparent by encouraging reporting and tracking behavior across diverse training and practice settings. Nevertheless, it is crucial to identify perpetrators efficiently and to take decisive action because the scope of damage to patients from serious ethical violations in medicine is often substantial. The recent case involving Larry Nassar provides a powerful illustration of this.

Additionally, some measures that would likely help to reduce repeat instances of serious ethical violations might conflict with the privacy interests and emotional needs of patients who have been victimized. Here the collective needs of future patients potentially conflict with moral commitments to treat patients who have been harmed by physicians in a manner that is sensitive and prioritizes their needs during times of difficulty.

In what follows, we raise a series of questions about possible responses to the problem of serious ethical violations in medicine. In general, these questions explore ways of shifting the balance from protecting physician interests to protecting patient interests; however, some questions (7 – 9) explore shifting the balance from a current set of patient interests to new patient interests. We know that some of the options raised for consideration will be controversial, while others will be difficult to implement due to resource limitations as well as logistical or political reasons. All must be vetted and carefully examined to determine if they are prudent and practicable.

1. When do physicians who engage in serious wrongdoing deserve a second chance?

As our study demonstrated, following a finding of serious wrongdoing, disciplinary actions range from mild (such as attending a continuing medical education program on boundaries or on proper prescribing) to severe (such as permanent loss of licensure or imprisonment). In general, the most severe action a medical board can take is to permanently strip physicians of their license. This is a drastic measure that should not be taken lightly. Apart from the impact on the individual physician, the U.S. expects a shortage of 40,000–105,000 physicians by 2033 (Dall et al. 2017). Moreover, the U.S. government invests approximately $11 billion per year in residency training of physicians (Heisler et al. 2016). At the same time, one study found that those who have been subjected to severe sanction over the past 5 years are 32 times more likely to receive a severe sanction than their peers over the subsequent 5 years (Grant and Alfred 2007). Are there some one-time offenses that deserve a permanent revocation of a medical license? In some of the most egregious cases, such as rape or Medicare fraud, criminal convictions often effectively remove the physician from the field. Even when criminal charges are not brought forward, egregious cases are easy: Removing a license seems justified when it will prevent repeated serious harm to patients. The more difficult decision pertains to repeat minor offenses, which frequently precede more serious violations (DuBois, Walsh, et al. 2017). Yet, the biggest obstacles to removing a physician’s license appear not to be philosophical, rather structural: There is a six-fold variance in the rates of serious disciplinary actions by state medical boards, with boards taking serious disciplinary action more frequently when they are well-funded, well-staffed, independent of state medical societies, and able to use data from diverse sources and a preponderance of the evidence standard (rather than a beyond a reasonable doubt standard) (Wolfe, Williams, and Zaslow 2012).

2. How can we ensure a second chance is not the 265th chance?

Publicly available evidence indicates that Larry Nassar abused at least 265 young women and children. Thus far, it appears that at least eight women reported his behavior to someone (Mack and Lawler 2017); yet it persisted. If the Nassar case was handled discretely without involvement from the press far fewer cases would have come to light, and depending on the details of the particular accusation, he might have been sent for training and asked to accept additional oversight.

Currently the NPDB guidebook states that, “The information in the NPDB should serve only to alert eligible entities that there may be a problem with the performance of a particular health care practitioner, entity, provider, or supplier” (Health Resources and Services Administration 2015, p. A7). That is to say, the NPDB admits that it is not a reliable source of detailed information on a physician’s history of wrongdoing. Even if Nassar ended up in the NPDB, his violations might not have been categorized, and if they were categorized, they might have been categorized as “not specified” or as “other” (Harris and Byhoff 2017). But in many cases, there is no more reliable, detailed, or comprehensive source of information. For institutions and boards investigating an accusation against a physician, it would be helpful to see not only past actions against a physician’s license but also past accusations.

A complete database of accusations against physicians, from the time they take their Step 1 licensing exam as medical students until they retire, would be extremely useful in addressing the problem of serious ethical violations because patterns only emerge across time and work settings. Such a tracking program could be stigmatizing for physicians and could be abused. Accusations are not the same as findings of guilt. Accordingly, some information might be restricted access, as is the NPDB at present, while other information—such as severe sanctions—might be publicly available to inform patients as they choose physicians.

Those who choose to serve their country in roles that provide access to highly sensitive government documents are required to accept background checks and surveillance of their activities. Perhaps it is not unreasonable to accept that long-term tracking of accusations is part of the job package when you have the professional right to conduct intimate exams on children, to prescribe controlled substances, and to receive significant payments for cutting into a patient.

A related policy question pertains to institutions that have reporting requirements—and would have increased requirements in this model: What should be the consequences to an institution when complaints or concerns are expressed but not investigated, or when investigations indicate reportable violations, but the institution remains silent?

3. Given that a strong predictor of serious wrongdoing—and one of the few observable red flags—is a history of professionalism violations, at what point should physician training programs remove students or residents?

Medical schools in the U.S. graduate approximately 97% percent of medical students within 8 years of starting medical school (Association of American Medical Colleges 2014). The 3% who do not complete includes those who leave due to mental or physical illness, death, changed vocational plans, or removal from a program due to poor academic performance or behavioral problems. Medical schools are extremely selective in their admissions yet perhaps a higher bar be should be set for graduating from medical school. If our primary goals are to ensure fairness to physicians in training, to protect society’s investment in physicians, and to increase the physician pipeline, then the 3% mark may be acceptable. But if our primary goal is to protect patients, perhaps this figure is too low. At present, we have seen no data on rates of serious ethical violations among medical students, so it is difficult to assess the impact of such behavior on future practice. A study by Papadakis and colleagues found that unprofessional behavior during medical school was associated with a threefold increase subsequent disciplinary action by a medical board (Papadakis et al. 2005). This relationship was strong despite the fact that “unprofessional behavior” include indications such as immaturity, lack of initiative, and “Unprofessional behavior associated with anxiety, insecurity, or nervousness” (p. 2679)—that is, relatively minor problems. We expect that more serious offenses (e.g., practicing while impaired or sexually harassing classmates) would predict future disciplinary actions to a much higher degree.

4. Should higher qualifications be required to work in solo practice?

Missouri recently implemented an assistant physician program, which allows individuals who complete medical school and pass the first 2 steps of the licensing exam to practice largely independently without ever completing a residency or internship year (Crane 2015). The purpose is to increase healthcare access in medically underserved areas. But according to our data, not being board certified, much less fully licensed, serving in a solo practice, and serving vulnerable patients are all risk factors for serious ethical violations, particularly IPCS and the most egregious forms of SAP. Our data suggest that accepting lower qualifications to work in solo practice, particularly in medically underserved areas, may sacrifice patient safety for small gains in increased access.

5. Should federal rules guide the composition and authority of SMBs as well as minimum actions for specific violations?

As noted above, state medical boards vary significantly in the rates with which they take serious disciplinary actions such as revoking a license—from 1.3 to 6.8 physicians per 1,000 (Wolfe, Williams, and Zaslow 2012). Features such as having more board members, independence from regional government and state medical societies, adequate budget, and reasonable rules of evidence lead to increased disciplinary actions by boards (Harris and Byhoff 2017, Wolfe, Williams, and Zaslow 2012). Federal evidence-based guidance can increase the likelihood that SMBs will properly investigate cases of serious ethical violations and respond with appropriate, effective disciplinary actions. If authority over SMB rules remains at the level of states, it is important to take actions that facilitate adoption of best practices.

6. Should the National Practitioners Database (NPDB) gather detailed information and more broadly share data on physician disciplinary actions?

Our experience of researching serious ethical violations over the past seven years has reinforced the findings of others: Obtaining data on cases is extremely difficult; obtaining representative data that are rich in details about individuals and their practices settings is impossible. Human subjects protection guidelines support broad sharing of detailed and sensitive medical records when data are de-identified. These same principles should apply to the collection and sharing of sensitive physician data. The NPDB could serve as a central data repository and consult with researchers to identify a series of variables that should be collected in every case report. Using standards similar to HIPAA safe harbor rules, such data could be shared. While such data are sensitive, they are surely no more sensitive than, say, a patient’s HIV status, substance use patterns, or mental health.

7. When should medical oversight bodies share information with law enforcement, and what voice should patients have in such decisions?

The Federation of State Medical Boards’ document, “Addressing Sexual Boundaries: Guidelines for State Medical Boards,” nowhere mentions the obligation to or even the possibility of involving law enforcement in such cases (Federation of State Medical Boards 2006). It does not remind physicians that they are mandatory reporters of child and elder abuse. It does not recommend informing patients who have been abused that the board is incapable of prosecuting criminal behavior and that they have the option of involving law enforcement if they wish to pursue criminal charges. It does not provide any guidance for assisting a patient in reporting to law enforcement. Increasing reporting to law enforcement will not only reduce the self-regulation of medicine, but risks infringing on patient preferences and increasing the trauma patients experience. Efforts to address this important topic would benefit from input from patients, law enforcement, and medical societies.

8. How can we advise patients to report violations without engaging in victim blaming and shifting responsibility from the physician community?

When the Atlanta Journal-Constitution completed its investigative reporting series on sexual abuse of patients by physicians, they created a resource page for patients (The Atlanta Journal-Constitution 2016). Similarly, Consumer Reports has produced a Safe Patient Project website (Consumers Union 2018). Both of these resources aim to empower patients, and they recommend that patients involve law enforcement when they suspect they have been abused by a physician. Yet this raises important questions about competing values and concerns. On the one hand, many if not most cases of serious ethical violations would never be investigated if not for patients who report the behavior to someone in authority. On the other hand, the reporting process can range from uncomfortable to traumatizing and outcomes are uncertain; accordingly, it is problematic to charge patient-victims with the responsibility to solve the problem of serious ethical violations. It is important that different reporting and prosecutorial options are clear to those who are willing to report. Similar to Title IX investigations, some SMBs use a preponderance of the evidence standard rather than a “clear and convincing” or “beyond a reasonable doubt” standard. This can make it easier for a SMB to take disciplinary action than a criminal or civil court (Wolfe, Williams, and Zaslow 2012). It may also be more private and less traumatizing for victims. However, it is unlikely to lead to incarceration, which may be appropriate in some cases. It is important to empower patients to make informed decisions regarding various courses of action. Efforts to address this important topic would benefit from input from patients, law enforcement, and medical societies.

9. How can we empower patients to protect themselves without undermining trust in medicine?

Related to the above point, efforts to educate the public about serious ethical violations in medicine and possible warning signs is a two-edged sword. On the one hand, patients have a right to know this information. Patients should know that some physicians do engage in serious misconduct and they should know, for example, when an intimate examination is indicated, that they have a right to a chaperone, or where they can find information about unnecessary procedures. Knowledge is a key to empowerment. Nevertheless, sharing statistics on serious ethical violations in medicine runs a risk of harming patients if it further erodes trust in medicine, particularly within groups that already access medical care less frequently, and adhere to physician recommendations less strictly, due to mistrust (Ferrera et al. 2015, Blendon, Benson, and Hero 2014, Khullar 2018b). Problems may be particularly acute in medically-underserved areas where patients may have few choices among physicians; in such cases, is it beneficial to promote concerns about physicians who fit a common profile in our studies (e.g., are trained outside the U.S., are not board certified, and are male), when patients may have no choice to change physicians and when the majority of physicians who meet this profile never generate complaints with their state medical board? Withholding information on serious ethical violations would seem overly paternalistic, but promoting information about these matters could have unintended consequences for patients.

10. What can be done to address demographic risk factors for serious ethical violations?

Within many of our taxonomies of wrongdoing, certain physician characteristics were far more common in our sample than in the general population of physicians, and other studies have identified similar patterns: Physicians who are male, trained outside of the US, not board certified, and over 50 appear more likely to offend. In general, our society frowns on profiling or discriminating against employees using traits such as race, sex, or age. Profiling would be discriminatory in part because the vast majority of physicians who might fit a profile never commit a serious ethical violation. Importantly, some physicians who do not fit the profile nevertheless offend. However, it may be appropriate to consider physician characteristics in at least two ways. First, characteristics related to training are not intrinsic to individual persons. It is fair to expect that internationally trained physicians, and non-board-certified physicians, remedy any deficits of training that might explain differences in professional behavior. Unfortunately, it is not clear what specific aspects of training might explain these differences, nor whether differences are actually due to self-selection—who chooses to study internationally and who choose not to seek board certification. Further research on these topics will be important to guiding decisions regarding the qualification of physicians. Second, some data indicate that women in senior leadership positions are rated higher on integrity and honesty by their peers, relative to male counterparts (Zenger and Folkman 2012). Yet women continue to lag behind men in senior leadership positions within medical centers (Rochon, Davidoff, and Levinson 2016). We think it is time to encourage medical centers to increase gender diversity in senior leadership, in part, to help combat the problem of serious ethical violations.

In this paper, we have focused on policy questions that involve a conflict of moral considerations. Not all responses to the problem of serious ethical violations in medicine involve such difficulties. For example, medical ethics courses might increase their focus on teaching—and even practicing through role playing—appropriate behaviors and actions to take when one observes wrongdoing among peers or is in a position to investigate patient complaints. This could include cases involving not just egregious wrongdoing, but also the more minor behaviors that can serve as forms of grooming patients for abuse or testing the oversight within an institution. Such training can contribute to an institutional culture that encourages speaking up and protection for those who do report wrongdoing. Similarly, in this paper, we have not explored questions pertaining to very specific forms of wrongdoing such as UIP and IPCS, which are frequently motivated by greed and enabled by current reimbursement systems. Nevertheless, such questions deserve serious attention.

Conclusion

It is challenging to balance a concern for the fair treatment of an accused physician or medical student with the protection of patients. We believe that the balancing act has thus far favored the fair treatment of physicians who are accused of ethical violations. Some degree of correction is needed in how we balance the complex and legitimate concerns of physicians, patients, family members, and society. Admittedly, some of the measures that would almost certainly decrease abuse of patients would also diminish the privacy of accusations, hearings, and actions against physicians, and increase the consequences of repeated minor violations even though they might never escalate to serious ethical violations. Other measures might increase pressure on patients who have been harmed to report physician wrongdoing and cooperate with medical boards or law enforcement.

We are not sure what precisely should be done, but we are sure that significant reform is needed. In 2002, in response to the problem of sex abuse by clergy, the United States Conference of Catholic Bishops (USCCB)—after years of inaction (Childress 2014)—publicly adopted several measures to prevent and respond to abuse, including: ordering an independent study of the extent of the problem and contributing factors (John Jay College Research Team 2011); developing a zero tolerance policy and policies for addressing credible reports, including involving law enforcement (United States Conference of Catholic Bishops 2002/2011); requiring annual external audits of diocesan compliance with policies and procedures (United States Conference of Catholic Bishops 2018a); mandating training for all clergy and laypeople who work with children (National Catholic Risk Retention Group 2018); and establishing clear mechanisms for reporting suspected abuse (United states Conference of Catholic Bishops 2018b). Even after extensive reporting of serious ethical violations in medicine (Teegardin et al. 2016, Kohler and Bernhard 2010), and a slow but steady decline in public trust in medicine (Blendon, Benson, and Hero 2014, Khullar 2018b), leading physician organizations have not established new policies and procedures to address the problem.

These issues urgently require the attention of stakeholders, policy-makers and ethicists as competing policies are considered in terms of their effectiveness in protecting patients and the proportionality and necessity of the infringement they require on physician interests (Childress et al. 2002). We hope that our study of 280 cases of serious ethical violations, and the series of action-item questions we have posed, will stimulate vigorous discussion aimed at reform.