Article Text

Download PDFPDF

Variations by state in physician disciplinary actions by US medical licensure boards
  1. John Alexander Harris1,
  2. Elena Byhoff2,3
  1. 1Department of Obstetrics and Gynecology, University of Michigan, Ann Arbor, Michigan, USA
  2. 2Department of Medicine, University of Michigan, Ann Arbor, Michigan, USA
  3. 3Department of Veterans Affairs, Ann Arbor, Michigan, USA
  1. Correspondence to Dr John Alexander Harris, Department of Obstetrics and Gynecology, University of Michigan, Bld 14 G100-19, 2800 Plymouth Rd, Ann Arbor, MI 48109, USA; harrja{at}med.umich.edu

Abstract

Objective To investigate the variation in the rate of state medical board physician disciplinary actions between US states.

Methods Longitudinal study of state medical board physician disciplinary action rates using the US National Practitioner Data Bank and American Medical Association estimates of physician demographics across all 50 states and the District of Columbia from 2010 to 2014. Results were reliability adjusted using a multilevel logistic model controlling for year of disciplinary action, physicians per capita in each state and the rate of malpractice claims per physician in each state.

Results From 2010 to 2014, there were a total of 5046 506 physician licensure years present. Medical boards reported a total of 21 647 disciplinary actions, of which 5137 (23.7%) were major disciplinary actions involving revocation, suspension or surrender of licence. The mean, reliability-adjusted rate of all disciplinary actions was 3.76 (95% CI 3.21 to 4.42) with a significant variation between states. State rates ranged from 2.13 (95% CI 1.86 to 2.45) to 7.93 (95% CI 6.33 to 9.93) actions per 1000 physicians. The mean rate of major disciplinary actions was 2.71 (95% CI 1.93 to 3.82), ranging from 0.64 (95% CI 0.53 to 0.76) to 2.71 (95% CI 1.93 to 3.82) actions per 1000 physicians. The correlation between the rate of major disciplinary action and minor disciplinary actions was 0.34.

Conclusions There is a significant, fourfold variation in the annual rate of medical board physician disciplinary action by state in the USA. When indicated, state medical boards should consider policies aimed at improving standardisation and coordination to provide consistent supervision to physicians and ensure public safety.

  • Accreditation
  • Governance
  • Health policy
  • Patient safety

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Introduction

In the USA, physicians self-regulate their profession through medical licensure and disciplinary boards, generally at the state administrative level. These US state medical boards serve to protect the safety of the public through physician licensure, licensure surveillance, misconduct investigation and subsequent discipline.1 ,2 The public expects and desires fair, equitable, consistent standards and processes that ensure safe and ethical medical care across the country. Yet medical boards are state or territory specific, and consequently vary in regulations, processes and resources.3 ,4 If the rate of physician misconduct is comparable across the USA, there may be an assumption that medical boards would equitably enforce disciplinary actions and that enforcement would be similar across all states. However, the number of regional regulatory medical bodies in the USA and the variation in regulations, processes and resources raises the question of whether the rates of discipline actions for physician misconduct are indeed similar across states.

In the USA, a public safety advocacy group, Public Citizen, has shown a 8- to 18-fold variation in yearly rates of physician discipline by state from 1999 to 2012 using summary data available from the Federation of State Medical Boards.5 However, the reliability of estimated individual state disciplinary rates may vary significantly due to the small number of disciplinary actions each year. We aimed to provide a more accurate estimate of disciplinary rates using a national, government-administrated database of medical board disciplinary actions as well as national databases that compile the number of licensed and practicing physicians over a period of 5 years from 2010 to 2014. We employed Bayesian statistical methods to estimate reliability-adjusted rates of medical board physician discipline actions by US state and controlled for temporal trends, physicians per capita in each state and a measurement of the malpractice climate in each state. We investigated between-state variation in disciplinary actions under the hypothesis that known variations in regulations, processes and resources in state-specific medical boards would be associated with a significant between-state variation in the rate of yearly medical board disciplinary.

Methods

US National Practitioner Data Bank

We collected all medical board adverse actions from all 50 US states and the District of Columbia using the Public Use Data File of the National Practitioner Data Bank (NPDB) from January 2010 to December 2014.6

The NPDB was mandated in the US Health Care Quality Improvement Act of 1986, which requested the US Department of Health and Human Services to collect all physician misconduct actions and malpractice claims in the USA.7 The primary purpose of the databank was to collect and distribute information about physicians to improve healthcare quality, protect the public and reduce healthcare fraud or abuse.8 Hospitals, healthcare plans and medical boards routinely query the NPDB to learn of previous adverse findings against a physician that they are considering giving privileges, allowing submission of medical claim bills for reimbursement, or licensing in a new state. Actions that must be reported to the NPDB by a medical board include adverse licensure actions (limitations, suspension, surrender or revocation of a medical licence), adverse clinical privileging actions (loss of hospital or clinical privileges), adverse professional society membership actions (loss of membership due to misconduct), exclusion from federal or state healthcare programmes or plans, and any civil judgements or criminal convictions. Events that are not mandated to be reported to the NPDB would include claims of misconduct reported to a hospital, health plan or medical board that were either not investigated or were investigated but did not result in a formal action against the physician. Also, events would not be reported to the NPDB if they involved civil or criminal legal cases that were dismissed or the ruling was in favour of the physician. While medical boards consistently report licensure actions to the NPDB, there have been reports by the US General Accounting Office and public advocacy groups suggesting wide variation in the rate of non-medical board institutions, such as hospitals, reporting adverse events to the NPDB.9–12 Despite limitations, it is a useful source of information on disciplinary actions and malpractice payments in the USA.13–15 The present analysis limited the adverse events of interest to only licensure actions by medical boards. Major actions were defined as revocation, suspension or surrender of licence. Minor actions include actions that were defined as a restriction of clinical practice, a mandatory clinical proctoring or a monetary fine.

Sources of other covariate data

To estimate rates of medical board disciplinary actions controlling for the number of physicians in a state, we used a yearly demographic report by the American Medical Association (AMA) to collect the current total number of physicians practicing in each state.16 As the number of physicians practicing in a state varies from year to year, we used reported data from 2010 to 2012 and created a linear trend model of the physician population data from 2007 to 2012 to predict the total physicians in each state for 2013 and 2014. We also collected data on state-level physician labour supply to determine the number of physicians per total state population.17 We examined data from three dates during the study period, but as there was no significant change in the values over time, we used 2013 values as the covariate for physician labour supply. To estimate the malpractice climate in each state, we counted the number of malpractice payments reported to the NPDB per number of physicians in a state during the study period.

Primary outcome and predictor variable

The outcome of interest was the yearly state rate of all medical board disciplinary actions per 1000 physicians.16 We combined data on disciplinary actions from the NPDB with data on the number of practicing physicians per state from AMA data to create a combined dataset with individual-level observations so that we could disaggregate state-level data to report results per 1000 physicians. The primary predictor variable was US state where the disciplinary action was reported.

Statistical analyses

We modelled disciplinary actions using a multilevel logistic regression model controlling for state physician supply, year-to-year trends in disciplinary actions during the study period and malpractice climate as fixed effects and state-level variation as a random effect.17

If the variation in state disciplinary patterns is assessed using the average disciplinary rate in each state, the analysis will overstate the true range of the state variation.18–20 To account for the small amount of actual state-level variation present in the rates of physician discipline, a ‘shrinkage factor’ may be used to adjust based on the reliability, the signal-to-noise ratio, of the data. The shrinkage factor helps to statistically adjust the outcome measure for the level of reliability in the outcome measure for each state. For instance, a state with relatively few disciplinary actions will have a low reliability and therefore the estimate of the state disciplinary rate should be shrunk towards the average rate for all states.19

Unadjusted rates of disciplinary actions per state were reported using the full combined dataset. Using the full dataset, the multilevel logistic regression model was unable to provide valid output estimates due to failure of statistical intermediate analytical derivative calculations, due to large number of observations and quantitative complexity of the model. However, using a stratified (by all disciplinary actions, major disciplinary actions, year and state) random 25% sample, the model was able to provide estimated disciplinary rates. A descriptive comparison of the full and 25% sample is available in the online supplementary appendix. We used the model to predict the reliability-adjusted probability of physician disciplinary action per physician by state and calculated 95% Bayesian CIs.

Supplementary appendix

Comparison of total disciplinary actions between full study sample and 25% sample

We estimated the reliability-adjusted correlation between rates of major and non-major disciplinary actions using the Spearman–Brown formula with double correction for attenuation.21 The study was deemed exempt by the institutional review board.

Results

The NPDB from 2010 through 2014 included 274 880 reported adverse events: 259 239 disciplinary actions and 15 641 malpractice payments. Of these disciplinary actions, 220 159 entries were actions by state licensure organisations. A total of 194 126 actions involved non-physician medical providers and 26 033 actions involved physicians. After excluding reciprocal actions (where a physician licensed in multiple states receives the same disciplinary action in all states during the same year), there were 21 647 unique physician misconduct state medical board actions was among 5046 506 physician years. Of those medical board actions, 5137 were major actions. A comparison of the number of disciplinary actions as well as covariates by state is presented in table 1.

Table 1

Description of US state characteristics disciplinary actions and related factors

For the unique physician misconduct state medical board actions, there were 26 804 reasons specified for the action. A total of 5242 actions had no information on the reason for the action. When a reason was described, the basis for these actions was most often classified as ‘Not specified’ 38% (n=9977), ‘Illegal activity’ 8% (n=2053), ‘Unprofessional conduct’ 4% (n=1076), ‘Licence action’ 7% (n=1854), ‘Negligence’ 9% (n=2443), ‘Substance abuse’ 4% (n=1068), ‘Sexual or boundary misconduct’ 2% (n=430), ‘Failure to comply with medical board’ 6% (n=1600), ‘Fraud’ 6% (n=1525), ‘Failure to maintain adequate records’ 2% (n=631), ‘Immediate threat to health and safety of public’ 1% (n=312) and ‘Other’ 14% (n=3835).

Using the full sample, the unadjusted mean rate of yearly disciplinary actions was 4.29 (95% CI 4.23 to 4.35) and ranged between states from 1.74 to 10.27 per 1000 physicians from 2010 to 2014. There was a sixfold difference between the lowest and highest rates of physician disciplinary actions before reliability adjustment.

Using the stratified random sample, after adjustment for measurement reliability, supply of physicians, malpractice environment and year, the mean rate of disciplinary actions was 3.76 (95% CI 3.21 to 4.42) with a significant variation between states, with state rates ranging from 2.13 (95% CI 1.86 to 2.45) to 7.93 (95% CI 6.33 to 9.93) actions per 1000 physicians. There was a fourfold variation in the total rate of physician misconduct between states with the lowest and highest rates. A full comparison of rates of physician misconduct disciplinary actions between states is presented in table 2. Figure 1 presents the rank order of states by their respective rates of physician misconduct disciplinary actions.

Table 2

Reliability and confounder-adjusted yearly disciplinary actions per 1000 physicians with 95% CIs

Figure 1

Any physician disciplinary actions per year by US state.

For major disciplinary actions involving revocation, suspension or surrender of licence, there was also a significant variation by state, ranging from 0.64 (95% CI 0.53 to 0.76) to 2.71 (95% CI 1.93 to 3.82) actions per 1000 physicians with a mean rate of 1.15 actions (95% CI 0.87 to 1.55) per 1000 physicians (figure 2). There was a fourfold variation in the rate of major physician disciplinary actions between the lowest and highest rates. The correlation between the rate of major disciplinary action and minor disciplinary actions was 0.34.

Figure 2

Major physician disciplinary actions per year by US state. Major physician disciplinary actions are defined as licensure actions involving suspension, surrender or revocation of a medical licence.

Discussion

Primary findings

In the USA, there is a significant fourfold variation in the annual rate of medical board physician disciplinary actions between states even after controlling for data reliability, year-to-year variation, physician labour supply and malpractice climate in each state. Additional sensitivity analysis restricted to only major disciplinary actions found a similar level of between-state variation. There is poor correlation between states that have higher rates of major and minor disciplinary actions. Overall, there is significant evidence of large variation in the rate of disciplinary actions between states.

Strengths/weakness of the study

Strengths

The major strengths of our study are the inclusion of data from multiple large national databases over five recent years to evaluate differences in the rates of medical board disciplinary actions. We address the variation and risk of bias in the data by using multilevel regression models and Bayesian estimates of the reliability-adjusted rates; we have provided accurate rank-order state-level estimates controlling for variation by year, malpractice environment and physician supply.

Weaknesses

Our study has several limitations. First, our primary outcome of interest was total disciplinary actions, rather than other important outcomes such as complaints filed or investigations completed. Therefore, our analysis does not include misconduct complaints that did not result in a disciplinary action. Because of this, we are unable to evaluate important mediators of the disciplinary action process including the ‘claim-to-investigation’ ratio or the ‘investigation-to-disciplinary action’ ratio. Our analysis cannot, therefore, be used to make claims regarding actual physician misconduct, or the rates of which this misconduct is addressed between states. Next, there are multiple available estimates of physician population. We chose to use the AMA values because these values are publicly available each year and include all physicians whether they are presently licensed in the state or not practicing. Some physicians had multiple actions reported in different states on the same year. While we assigned the action to only state which we believed to be the state of the original action, it is possible that individuals could have disciplined for separate acts in different states. We did not control for a multitude of state medical board level factors such as number of individuals on a medical board, the presence of non-physician board members, the differences in board authority or responsibility, or factors related to how the board functions (for instance, how it determines the need for disciplinary action). While these are important factors in how state boards function, these factors are only approximations of the central concepts that may affect disciplinary rates such as effectiveness of investigations, independent and unbiased proceedings, and consistency of medical board processes over time.

Other studies

Public Citizen, the US non-profit consumer rights organisation, has reported overall and state-specific disciplinary rates from 1999 to 2012.22 Using estimates of physician discipline in each state from Federation of State Medical Board summary reports, they have consistently noted that serious disciplinary actions vary by state, ranging from 8- to 18-fold variation depending on the year. Public Citizen also reported 3-year average rates for states that produced smaller estimates of variation between fourfold and sevenfold. A comparison of the Public Citizen rankings and NPDB summary data noted a significant association between the two ranking systems, but noted that there were significant differences in rankings depending on the data souce.23 While these analyses introduced the issue of between-state variations in physician discipline into the public discussion, the lack of reliability adjustment and use of standardised population estimates led to likely overestimated effect sizes.18 ,19 ,24

Comparisons with other publications

Research on physician discipline has focused on two types of factors leading to physician discipline: physician characteristics and medical board characteristics. Individual physician characteristics associated with disciplinary actions include male gender, lack of board certification, older age, certain specialties (family practice, general practice, obstetrics and gynaecology, and psychiatry) and a history of unprofessional behaviour in medical school.25–29 Studies on disciplined physicians in the USA, Australia, New Zealand and Canada have noted similar trends in individual physician risk factors across countries.27 ,29–32 Besides known unprofessional behaviour in medical school, many of these factors are not modifiable and are not in themselves reasons to deny licensure or deserve greater scrutiny from medical boards. Many of these factors such as the distribution of gender, age and specialties among regions are stable, and likely to not be a key factor in disciplinary rate variation between states.

Medical board characteristics that are associated with disciplinary actions include presence of non-physicians on medical board, increased numbers of board members and board independence from regional government.1 ,4 ,33 None of these studies examining medical board characteristics have discussed the overall variation of disciplinary action rates between states. In this study, we did not find that malpractice environment, physician supply or the year of discipline were significantly associated with the rate of physician discipline. Finally, there is little correlation between the rate of major and minor disciplinary actions, which suggests, in one scenario, that some states may have higher rates of major actions while not having higher rates for minor action. Disciplinary practices are likely not as simple as high disciplinary rate boards versus low disciplinary rate medical boards: some medical boards may be more likely to use ‘minor actions’ like practice restrictions or fines, while other state medical boards use ‘major actions’ licensure suspension or revocation.

Beyond the work by public safety advocacy groups, few investigators have focused specifically on medical board composition and how variation between medical boards may affect the disciplinary process. In a regulatory analysis of a wide variety of different characteristics of medical boards related to higher rates of disciplinary actions, including independence from state government, board members per total doctors in the state, full-time staff per total doctors in the state, percentage on non-physician board members, the receipt of government funding and state reporting requirements, only the ratio of board members per total physicians in the state was significant in all analyses.4 In another analysis comparing medical board behaviour from 1960 to 1977, they noted that ‘the principle determinant of board vigilance was the degree to which [medical boards] were not dominated by physicians’.33 While little work has been completed in the area of medical board effectiveness, the central concerns about the varying resources (the relative size of the medical board compared with the population of physicians) and independence (the presence on non-physician members) suggest that different board characteristics could be a contributing factor to the between-state variation of disciplinary rates we observed.

Possible explanations

To best understand the contributing factors in interstate variation in disciplinary actions, we propose a framework to outline the process from misconduct to discipline where opportunities for variation to occur (figure 3). At each step in the process, there are several reasons why a particular misconduct event may not proceed to the next step towards disciplinary action taken. Once a misconduct event occurs, it must first be recognised by participants as an act of misconduct. Next, the individual involved must understand how and feel empowered to report the event as misconduct. Once a possible misconduct event is reported, it must have enough detail and specificity to make an investigation of the claim possible. Many complaints to medical boards, 80% in one study, are too vague or difficult to corroborate to lead to a more extensive investigation.29 Once a claim of misconduct is investigated, it may lead to a decision of fault or innocence depending on the available evidence and interpretation. The final step in the pathway of misconduct to discipline depends on each individual board's interpretation of wrongdoing.

Figure 3

Physician misconduct conceptual model.

At each step in this process, different decisions by a medical board may make it more or less likely that a misconduct event leads to the appropriate outcome. Medical boards may vary in public status in the community; in one state an aggrieved individual may quickly report an event, while in a different state the same individual may have difficulty in finding the appropriate regulatory body to report an event. Depending on the resources and time of each board, they may be less or more likely to initiate investigations into wrongdoing. Finally, standards for judgement and the standards for disciplinary action may vary from medical board to medical board. This multistep process offers many opportunities for variations in the final rate of disciplinary action especially with the known variation in state-specific procedures, policies and resources.

Future research

Future research may include focused comparisons of medical board characteristics or processes in high and low disciplinary rate states. International comparisons may be helpful as the structures of medical boards vary greatly in composition and oversight between nations. Also, the nationalisation of medical board in Australia in 2010 proves a helpful natural experiment to examine the effect of a national medical board compared with regional ones.

Conclusions

There is a significant amount of research that identifies physician factors associated with misconduct and disciplinary action. Further examination of the variations in US medical boards or comparisons with nationalised medical boards may yield helpful insights into how to optimise this important system that protects the public and advances the profession.

References

Footnotes

  • Contributors JAH participated in the study concept and design, acquisition of data, analysis and interpretation of data, drafting of the manuscript, critical revision of the manuscript for important intellectual content, statistical analysis and study supervision. EB participated in the study concept and design, analysis and interpretation of data, drafting of the manuscript and critical revision of the manuscript for important intellectual content.

  • Funding US Department of Veterans Affairs, Robert Wood Johnson Foundation.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Data sharing statement All data are publicly available.

Linked Articles