The impact of the national practitioner data bank on licensing actions by state medical licensing boards.
|Abstract:||The United States Congress mandated the establishment of the National Practitioner Data Bank in large part to decrease the likelihood that errant individuals might be able to avoid detection by licensing boards and practice medicine. We use a decade of longitudinal data (1985-94), for each of the 50 states, to evaluate the Bank's impact on state licensing board actions, during the four years following its 1990 birth. The results of a pooled, time-series analysis reveal that medical board restrictions on physicians' practices increased substantially following the creation of the Data Bank. We conclude that the increase was likely due to the licensing boards taking actions against delinquent physicians who had previously slipped through cracks in the regulatory system or who had earlier received warnings or administrative fines.|
Administrative agencies (Licensing, certification and accreditation)
Administrative agencies (Laws, regulations and rules)
Physicians (Licensing, certification and accreditation)
Physicians (Laws, regulations and rules)
|Publication:||Name: Journal of Health and Human Services Administration Publisher: Southern Public Administration Education Foundation, Inc. Audience: Academic Format: Magazine/Journal Subject: Government; Health Copyright: COPYRIGHT 2010 Southern Public Administration Education Foundation, Inc. ISSN: 1079-3739|
|Issue:||Date: Summer, 2010 Source Volume: 33 Source Issue: 1|
|Topic:||Event Code: 930 Government regulation; 940 Government regulation (cont); 980 Legal issues & crime Advertising Code: 94 Legal/Government Regulation Computer Subject: CD-ROM catalog; CD-ROM database; Database; Government regulation|
|Product:||Product Code: 8011000 Physicians & Surgeons NAICS Code: 621111 Offices of Physicians (except Mental Health Specialists)|
|Geographic:||Geographic Scope: United States Geographic Code: 1USA United States|
In this article we explore the impact of the National Practitioner
Data Bank (NPDB), which was designed in large part to decrease the
likelihood that errant physicians might be able to avoid detection by
licensing boards and practice medicine. The Bank opened for business in
September, 1990. The intent of the U.S. Congress in establishing it was
to assist the states' physician (& dentist) licensing boards by
providing a clearinghouse for information. We use a decade of
longitudinal data to evaluate the Bank's impact on state licensing
Physician self-regulation and the creation of the NPDB
The need for the creation of the NPDB stemmed, in large part, from problematic self-regulation by the medical profession. Physicians conventionally claim that only they have the specialized knowledge and medical wisdom that is needed to understand and resolve problems in their ranks. Medical services usually cannot be evaluated by outcomes and acceptable process matters are not easily recognized by laypersons. Self-regulation was deemed appropriate for physician licensing boards, given that physicians typically are the only persons truly qualified to evaluate medical services (Federation of Medical Regulatory Authorities of Canada and the Federation of State Medical Boards, 2008).
The U.S. Congress, in an early attempt to improve care and curb rising costs in Medicare, authorized a plan for self-review by the medical community. The Social Security Act Amendments of 1972 (Public Law 92-603) established Professional Standards Review Organizations (PSROs) to monitor services being delivered to hospitalized Medicare patients. The legislation stipulated that the individuals who were to conduct these reviews would be physicians and not government bureaucrats (Smits, 1982). PSROs were empowered to refuse payments for unacceptable treatment and to exclude doctors from participation in the benefit program. But only rarely did PSROs use these sanctions. During the eleven-year experiment with peer review boards, from 1973 to 1984, the PSROs formally disciplined a total of seventy hospitals and physicians (Wallis, 1986). The review panels preferred to educate errant physicians with bulletins and explanations of guidelines, during informal meetings or small conferences (see Jesilow, Pontell & Geis, 1993 for a more comprehensive discussion of the PSROs).
Doctors of the time were notoriously unwilling to label the practices of fellow physicians as deviant. Elliott Freidson (1975) illustrated the process by which doctors excused the deviant behavior of colleagues in his classic work, Doctoring Together. Freidson studied a group medical practice and concluded that each physician was allowed to practice as the individual saw fit, as long as the behavior was not filled with gross deficiencies. Aberrant behavior was excused by peers as something any doctor might reasonably do given the circumstances. Doctors were a delinquent community, he concluded, when it came to monitoring the conducts of its members (Freidson, 1975; see also Peters, 1972; Richardson, 1972; Swazey, 1991 regarding physician peer review and monitoring).
The failure of self-regulation played a major role in sparking creation of the NPDB. The U.S. Congress disapproved of the practices of many hospital quality committees to not report misdeeds to licensing boards, particularly if the errant individuals left their current hospital positions (House Report, 1986). It was possible for misbehaving physicians to escape the full impact of past negative evaluations of their practices by simply not informing new employers of their existence. Moreover, physicians who had restrictions placed on their practices by some state boards could continue the conduct that had gotten them into trouble by moving to other states. They already held licenses in the jurisdictions to which they relocated and they could open shop immediately upon arrival, without revealing their embarrassing past (Jesilow et al., 1993; see Kusserow, Handley & Yessian, 1987 for an overview of state medical discipline at the time). The NPDB was designed to minimize such loopholes (Brennan, 1998).
National Practitioner Data Bank
The enabling legislation for the National Practitioner Data Bank (the amended Health Care Quality Improvement Act of 1986) grants immunity from lawsuits to mandated providers of information. Hospitals are required to report doctors who resign after an investigation of them was started as well as any actions that affect a physician's clinical privileges for at least 31 days. Hospitals are obligated to consult the registry at least once every two years regarding present and prospective personnel. State licensing boards, insurance companies, professional societies, and healthcare institutions are mandated to report a variety of information to the Data Bank, including malpractice judgments and settlements, sanctions imposed by medical boards, losses of membership in professional societies, and actions taken by hospitals and other healthcare units against physicians and dentists. The NPDB fulfills its clearinghouse responsibility by making this information available to authorized entities. The establishment of the NPDB was based on the beliefs that, once armed with information, medical boards would be less likely to license inappropriate individuals and be more likely to take actions against doctors who had severe shortcomings (U.S. Department of Health & Human Services [hereafter HHS], 2007).
The National Practitioner Data Bank has been a lightning rod for criticism throughout its history, despite a dearth of information on its impact (Robertson, 2001; Ryzen, 1992; Satiani, 2004; Todd, 1995; Waters, Warnecke, Parsons, Almagor & Budetti, 2006). The critiques have often come from medical quarters, probably because the existence of the NPDB implies a lack of confidence in the integrity of the medical profession and challenges its position that it can monitor its own. James Todd (1995), former president of the American Medical Association, for example, criticized the NPDB for doing very little to improve matters. He concluded that "[government's role should be to set the standards to which the profession should be held accountable, leaving it to the profession and those it serves to decide how close the practitioner or entity approaches those standards (p. 378)." Studies of the NPDB have focused on process matters (e.g. reporting and inquiries to the NPDB and, in particular, its use in the credentialing process) (Baldwin et al., 1999; Neighbor, Baldwin, West & Hart, 1997; Office of Inspector General [hereafter OIG], 1995; Oshel, 1995; Waters et al., 2006). These reports have generally found that the Data Bank is being consulted. The authors of a recent study, for example, concluded that "[m]ost institutions make timely NPDB inquiries that facilitate widespread use of the information in credentialing activities (Waters et al. 2006, p.30)." Reporting of errant physicians to the NPDB is another matter.
The OIG for Health and Human Services reported that from its inception in "September 1, 1990 to December 31, 1993, about 75 percent of all hospitals in the United States never reported an adverse action to the" NPDB (Brown, 1995, p. i). The OIG was also troubled by the vast variation of reporting between states and raised concern about the suggested differences "in the capacity or willingness of hospitals to submit reports to the Data Bank (Brown, 1995, p. ii)." A later OIG report noted that during the NPDB's first nine years of operation, managed care organizations informed it of only 715 adverse actions and that eighty-four percent (1,176 of 1,401) of the organizations had not reported even one (Yessian, Greenleaf, Hereford, Han & Levine, 2001).
A few studies suggest that hospital committees have avoided the law's reporting requirement by utilizing penalties that legally do not need to be conveyed to the NPDB. Increased monitoring of a physician's professional activities or requiring the physician to attend continuing medical education, for example, are actions that do not need to be reported. Hospital administrators admit using these penalties rather than restricting a doctor's clinical privileges, which would require a report from the hospital to the NPDB. Hospitals (& errant physicians) also avoid mandated reports by allowing the problem practitioners to resign or voluntarily surrender their clinical privileges (Baldwin et al., 1999; Neighbor et al., 1997; HHS, 1995).
Another likely consequence of the Data Bank has been an increase in the number of physicians fighting malpractice charges, instead of settling with plaintiffs. The only quasi-experimental study (pre- post- design) of the NPDB that we found in the literature concluded that physicians and their insurers were less likely to settle malpractice cases following its introduction and that this "appears to have decreased the proportion of questionable claims receiving compensation (Waters et al., 2003a, p. 283)." Medical boards often take malpractice payments as triggering events for further investigation of physicians (Bovbjerg & Petronis, 1994; Fellmeth & Papageorge, 2005; Studdert et al., 2006; see Grant & Alfred, 2007 for a recent discussion of the operation of state medical boards). The suggestion is that, following the introduction of the Data Bank, physicians chose to fight questionable charges rather than allow their insurers to settle claims, which would have then been reported to the NPDB and eventually made known to licensing boards. Of course, whether one sees such matters as positive or negative outcomes depends on your viewpoint. The former AMA president we have already quoted labeled as an "adverse effect" a 1993 finding by the Physicians Insurers Association of America "that physicians were less willing to settle claims as a result of the NPDB (Todd, 1995, p. 377)." The lack of experimental evidence regarding the NPDB's impact has left a substantial gray area concerning its usefulness, which has allowed critics sufficient space to condemn its operation or to call for substantial changes. This paper partially corrects this situation by analyzing state licensing actions before and after the enactment of the Data Bank. Congress expected that the creation of the NPDB would decrease the likelihood that errant physicians could continue their conduct by withholding information from new employers or by moving to another state. We hypothesized, controlling for other matters, a statistically significant increase in the percentage of physicians receiving a sanction from state boards following the introduction of the NPDB. Such a finding would suggest that fewer inappropriate physicians are being allowed to practice and that Congress's will is being fulfilled.
We employed a quasi-experimental design to determine whether the National Practitioner Data Bank may have affected physician sanctioning by state licensing boards. Data on state licensing actions by year are available from the Federation of State Medical Boards (FSMB) website for the fifty states (Federation of State Medical Boards, n.d.).
We decided to only study the period, 1985-1994. Our decision to study this interval was guided by a number of factors. First, we needed accurate data from years prior to the establishment of the NPDB in order to determine its impact. Reporting of the number of licensing actions by individual state boards to the FSMB was inconsistent before 1985 (Galusha & Breaden, 1989); this date then became the earliest point for our study.
Our decision to end our study period at 1994 was based on two matters. The first for us was more significant and needs a bit of explaining. It stems from the punishment philosophies of deterrence and incapacitation. The threat of malpractice suits and potential licensing actions, according to the theory, act as general deterrents, in that physicians will avoid risky procedures in order to prevent future legal and financial obligations. In addition, physicians may be encouraged to take precautions during procedures to avoid potentially costly accidents and loss of license (Cane, 1982; Holmes 1873 & 2005/1881). The suggestion is that increased sanctions by individual state licensing boards may deter some risky physician behaviors and result in fewer licensing actions in later years. Incapacitation is expected to also diminish the number of licensing board actions against physicians. Confiscating the licenses of aberrant physicians assures boards that at least the most negligent physicians are removed from practice. The difficulty for us is determining whether declining (or leveling) rates of licensing actions after 1994 are due to a diminishing of the impact of the NPDB as a clearinghouse for information, or due to deterrence (physicians avoiding risky procedures) and incapacitation (negligent physicians losing their licenses or having them limited). An analogy might help clarify this matter.
A state may implement sobriety checkpoints to catch impaired drivers. Initially, the checkpoints result in a substantial rise in arrests, but as time goes by the number of arrests diminish. Is the reduction a result of the failure of the checkpoints to snare drunk drivers? Or is it a result of the success of the checkpoints; potential drunk drivers are in jail and unable to drive (incapacitated), while others have been deterred from getting behind the wheel? Does a lack of arrests indicate that the police are doing a poor job, or does it reflect the fact that there is no one to arrest? For our purposes, might leveling rates of licensing actions after 1994 be due to a failure of the NPDB as a clearinghouse for information, or might they be a consequence of negligent physicians losing their right to practice medicine combined with physicians avoiding risky actions? Determining the answers to these questions is not easy and may require individual level data that is not currently available (e.g., see Reuter & Bushway, 2007).
The second reason we ended our study period at 1994 stems from changes in various states tort laws that complicated matters after 1994. We expected that individual state tort reforms might have an impact on licensing actions, independent of any impact of the NPDB (Lavenant, Hayward & Jesilow, 2002). The statistical model we used in the current study was able to consider such matters. But new rounds of tort reforms beginning in 1995 were problematic. More than twenty states passed reforms that their courts then ruled unconstitutional and, as a result, their use was curtailed (Center for Justice & Democracy, n.d.). Inclusion of years after 1994 in our analyses would compromise the results; many states had periods when the reforms were momentarily on-the-books, but it is impossible for us to determine the times when the temporary tort reforms might have affected licensing board actions. State licensing board actions occur much later than the malpractice filings that may eventually trigger them. Put simply, there are periods when the reforms were legislated, but it is not possible to tell if the laws were in force. We are not alone in dealing with such concerns. Similar problems involving legal modifications to state tort laws led the authors of a recently published study to limit the years of their inquiry (Guirguis-Blake, Fryer, Phillips, Szabat & Green, 2006). We decided to curtail our study data at 1994 to minimize the impact of such matters on our dependent variable.
We ran models with two dependent variables, "serious sanctions" and "other prejudicial actions." The serious sanction variable includes a combination of probation, license revocation, and license suspension. As the term implies, the variable consists of the most serious sanctions state medical boards can impose on medical practitioners. Matters that are included within these three categories need not be clear-cut. Revocation involves losing one's license; suspension may involve the temporary loss of one's license or specific parts of one's practice, while probation may restrict a license. Under certain circumstances, suspension and probation may be similar (e.g. when a male physician is suspended from seeing female patients without a nurse present). We also combined the sanctions of probation, license revocation, and license suspension for the purpose of analyses because during the period of study many states often had no actions (or only one or two) in any single category. Combining the sanctions increased the potential for finding a statistical change. Moreover, the purpose of Congress in establishing the Data Bank was to diminish the likelihood that errant physicians would slip through loopholes. We reasoned that any of the severe sanctions indicated that the physician had been snared.
The "other prejudicial actions" dependent variable includes less severe disciplinary responses that boards can employ, such as consent orders, fines, and letters of admonishment. We assumed that the use of these lesser sanctions was in part due to the level of information that licensing boards had about physicians. The specific sanction a physician receives is likely tied to a number of factors, including the quality of the evidence and the resources of the board to conduct investigations (see generally Fellmeth & Papageorge, 2005). The extent of the information a board had about an errant physician might indicate use of the lesser sanctions. The introduction of the Data Bank, we reasoned, would increase the extent of information boards could obtain about errant doctors and result in a rise in the use of serious sanctions and a concomitant decline in the use of other prejudicial actions.
To standardize the dependent measures, information was collected on the number of medical doctors in each state. Physician populations were obtained from the American Medical Association's Physician Characteristics and Distribution in the United States (Randolph et al. 1995). Data were collected for the same years as the FSMB data, 1985-1994.
The inclusion of independent variables in the analyses was needed to determine if factors, other than the NPDB, were at work. We collected information on a number of matters that we believed might affect a state licensing board's sanctioning rate, including tort reforms, urban population, political ideology, religiosity, punitive ideology, and population health. We derived our hypotheses from the fields of health economics and criminology.
We have already discussed our logic for the inclusion of the state tort reforms in the model; there is evidence that they may impact the number of licensing actions (Lavenant et al., 2002). Changes in tort laws may be an alternative explanation for changes in a state's sanctioning rate.
The American Medical Association's Tort Reform Compendium (Bannon, 1989) outlines ten medical law revisions each state undertook between 1975 and 1988 (addendum clause, arbitration, attorney fee regulation, collateral source rule, frivolous lawsuit penalties, joint and several liability, limits on recovery, patient compensation funds, periodic payments of damages, and pretrial screening panels). We updated this information to 1994 with data from the American Tort Reform Association, which provides state reforms on its website (ATRA, 1995).
The extent of a state's urban population may affect the impact of the NPDB. The OIG noted that hospitals in rural states were "heavily represented among those with the highest level of nonreporting" to the NPDB, while hospitals in urban states were among those with the highest level of reporting (Office of Inspector General, 1995: iii). As a result, medical boards in urban states may be more influenced by the establishment of the Data Bank.
The extent of a state's urban population may also be related to the influence of the NPDB, because "urbanicity" may impact the probability that victims will bring civil suits. Malpractice judgments and settlements are required to be reported to the NPDB. Patricia Danzon (1984) found that a state's degree of urbanization was positively associated with the rate of malpractice claims in the state. The suggestion is that the NPDB will have a greater impact in urban states.
"Urbanicity" may also influence characteristics of medical board members, who sit in judgment on doctors, and impact licensing decisions. The extent of a state's urban population is a common independent variable in studies of court sentencing because it affects judicial characteristics (Bullock, 1961; Meyer & Jesilow, 1997; Myers, 1987; Weber, 1954). Similarly, licensing board members in urban states, because of their background, may see things differently than board members in predominately rural ones and this may affect the likelihood of sanctioning.
We included as an independent measure in our study the percentage of a state's population that lived in urban zones. We hypothesized that states that were more urban would be more impacted by the NPDB. Urban, nonurban, farm, and other measures of the total population of each state are from the 1990 Census (U.S. Bureau of the Census [hereafter Census], 1990).
Physician sanctioning may also be affected by the political ideology of a state. Boards have a wide array of punishments available to them, including warning letters, probation, education, rehabilitation, and revocation (Grant & Alfred, 2007). Errant physicians who are sanctioned by liberal boards may receive dispositions that differ from wrongdoers who face more conservative board members (Davis, Severy & Kraus, 1993). Research with judges has demonstrated that Democrats were more inclined to take a liberal posture in decision making, while Republicans were more conservative (Nagel, 1961; Tate, 1981). We reasoned that Republican-dominated licensing boards would be more likely to act on the information derived from NPDB reports to sanction errant physicians and, as a result, the NPDB would have a greater impact in these states.
We attempted to collect the political party affiliation of board members during the study period by writing the state licensing agencies, but were not successful. Lacking a direct measure of political ideology, we posited that a licensing board would reflect the ideology of the state within which it resided.
We collected information on each jurisdiction's presidential voting for 1988 (Reagan elected President) and 1992 (Clinton elected President). States that each year supported the Republican candidate were considered conservative. States that each year supported the Democratic candidate were considered liberal. States that voted Democratic one year and Republican another were considered moderate.
Religion is a common independent variable in social studies and is considered to impact judicial decisionmaking (Idelman, 1993; King & Hunt, 1984). Once again, we equated medical board members with judges and reasoned that religion would play a role in the likelihood of errant physicians being sanctioned. That is, we hypothesized that "more religious" states would vary from "less religious" ones with respect to the impact of the NPDB.
Rather than basing our measure on distinctions between religions, we decided to use a measure that reflected religious activity. A rate of church membership was used for each state, based on 1990 census figures. These data are available from the American Religion Data Archive web site, maintained at Pennsylvania State University and include church members as reported by 133 Judeo-Christian church bodies (ARDA, n.d.).
We thought that a state's punitive ideology might also affect its medical licensing board's activities and influence any impact of the NPDB; that is, some states may prefer harsh punishments, while other states may favor milder ones. Our measure of a state's punitive ideology was the percentage of the state's population who were incarcerated. The prison population is driven by the willingness of citizens to sentence petty offenders to long prison terms and to pay for housing them. California's "Three Strikes" law is illustrative. It caused the state's prison population to rise dramatically at the same time that measured crime in the state was declining.
To measure the state's punitive ideology, we began with the number of prisoners under state or federal jurisdiction (December 31st year-end total) annually from 1985-1994. These data are available as a downloadable spreadsheet (Hill & Harrison, 2005). The 1990 state census population was used to calculate the percentage of the state's population who were incarcerated, which was our measure of a state's yearly punitive ideology.
The general health of a state's population may affect opportunities for malpractice, which should ultimately affect state licensing board actions. A relatively ill population in any given year might temporarily overload the capacity of the medical profession to safely treat patients and result in improper care and physician sanctioning.
We computed the number of inpatient hospital days per state resident as a measure of the relative health of a state. The annual inpatient census is available from the Annual Survey of the American Hospital Association (American Hospital Association, 1984-1993). We used the 1990 Census to ascertain each state's population.
Table 1 provides the descriptive statistics for the additional independent variables used in the pooled time-series analysis. Table 2 presents the bivariate correlations between states' sanctioning rates and the additional independent variables.
We used a panel analysis to ascertain the impact of the establishment of the NPDB on state licensing board actions. The method allows us to also determine the impact of the passage of the individual tort reforms (which were passed in different years for different states) and our additional explanatory variables.
We resolved that a random effects regression model would be a good tool for analyzing our data. Random effects models, compared to fixed effects, allow for estimates of both between-state differences as well as within-state changes over time (Hsiao, 1986). Fixed effects models, in contrast, cannot include time-invariant predictors. Such time-invariant predictors, however, are of interest in this study. These include some state-specific effects, such as the extent of urban population, for which we do not have yearly measures during the study period.
Our longitudinal data are arranged into a pooled time-series of 50 states during 10 years. Each state has 10 records of data, equivalent to a survey panel data set with 10 waves. An Ordinary Least Squares (OLS) solution is inappropriate because OLS use assumes that observations are independent of one another (Johnson, 1995). Our records are dependent because each state contributes 10 records to the data set (Allison, 1994). A generalized least squares (GLS) solution for the model might solve the dependence problem by assigning weights "based on the components of variation that fall between and within individuals in the sample (Johnson, 1995: 1070)." The distributions of our dependent variables, however, are positively skewed (serious sanctions per 1000 physicians: mean=3.508, SD=2.450, variance=6.005, range =0 to 18.02, skewness=1.599, estimate of dispersion = 6.005/3.508 = 1.712; other sanctions per 1000 physicians: mean=2.393, SD=2.263, variance=5.120, range=0 to 14.12, skewness=1.863, estimate of dispersion=2.140). If a dispersion estimate (variance divided by mean) is greater than 1, then the data may be overdispersed; if less than 1, then data may be underdispersed. If the value is within the typically-acceptable 0.8 to 1.2 range, the model can be considered to be correctly specified (Hilbe, 1994; SAS Institute, Inc., 1993). The likelihood ratio test for overdispersion for serious sanctions and other prejudicial sanctions was calculated for each regression model and resulted in Chi-squared values that were significant (p=.000). The statistically significant evidence of overdispersion indicates that the negative binomial regression model, which we used, is preferred to the poisson regression model. Figures 1 and 2 display the histograms for the two dependent variables.
[FIGURE 1 OMITTED]
[FIGURE 2 OMITTED]
The statistical design is a random effects, negative binomial regression model for pooled time series panel data for 50 states during 10 years, 1985-1994. The two dependent variables are the number of serious sanctions and the number of other prejudicial sanctions, standardized for the population of physicians in each state (in STATA statistical program, the natural log of the physician population was used as the "offset" function; this strategy makes use of the correct probability distribution of the dependent variables). The tort reforms are represented by a series of dummy dichotomous variables, "0" for years before the tort was altered and "1" for the first full year the tort reform was in effect and all subsequent years. The tort reforms are lagged by two and three years in the models to reflect our belief that it would take a minimum of two years for a reform to alter the pattern of tort cases. Our model includes 10 different events (the timing of the passage of the individual tort reforms in each state), five additional control variables (characteristics of states), and period effects. The model includes both time-invariant and time-variant variables. The predictor variables that vary over time include punitive ideology and population health. The variables that do not vary over time include urban population, political ideology, and religiosity.
The basic regression model is: SERIOUSst or OTHERst= [beta]0 + [summation][beta]kTORTkst + [summation][beta]kSTATEkst + [summation][beta]kPERIODkst + as + [epsilon]st.
In this equation, SERIOUS is the number of serious sanctions (the first dependent variable), OTHER is the number of other prejudicial sanctions (the second dependent variable), s is the number of states, t indexes the time-series (from 1985 to 1994), [beta]0 is an overall constant, [beta] represents the regression coefficients, k is the number of measured independent variables, TORT indexes the individual tort reforms, STATE indexes the control variables (state characteristics), and PERIOD indexes the dummy variables representing period effects. a is a constant effect for individual states that is treated as a random variable in the model and is assumed to be uncorrelated with the independent variables (Hsiao, 1986; Johnson, 1995). [alpha] captures the effects of all unmeasured time-invariant variables of states (Maddala, 1987; Petersen, 1993). [epsilon] is an error term, the effect of unobservable variables that vary across states and over time. Not represented in the equation are the standardization of the dependent variables, the lag terms, and an error term accounting for overdispersion in the sanctioning rates. The method of estimation used is maximum likelihood, and the estimated [beta]s are the weighted average of the between-state and within-state estimators.
We ran several preliminary models to disentangle period effects and the effects of the independent variables on the sanctioning of physicians by state licensing boards. The impact of the NPDB is evident in Model 2 that simultaneously considered the impact of all of our independent variables (see Tables 3 and 4). Beginning with the NPDB's first full-year of operation (1991), there was a significant spike in states' serious sanction rates (IRR ranged from 1.39 to 1.62, p<.001). In the years prior to 1991, the average state serious sanctioning rate never rose above 3.3 physicians per thousand (range 2.4 to 3.3). Beginning in 1991, the rate never fell below 4 per thousand (range 4.0 to 4.9). The average state rate for other prejudicial actions, in contrast, steadily rose at a statistically significant rate prior to the establishment of the NPDB, cresting in 1990 at 3.8 per thousand, but fell precipitously during the following year (p<.05), reaching a low of 1.4 per thousand in 1994. Table 5 compares by year the average state serious sanctioning rate with the average state other prejudicial actions rate.
We hypothesized that matters other than the introduction of the NPDB might impact sanctioning by states' medical boards and we included these in our model. We elsewhere discuss these in some detail (Jesilow & Ohlander, in press), but our focus in this piece is on the impact of the Data Bank. We do note here, however, that several of these independent variables were associated with a state's serious sanctioning rate. Our results suggest that alterations to states' joint and several liability rules and the establishment of penalties for frivolous lawsuits may have increased the use of serious sanctions by medical licensing boards, while instituting arbitration and attorney fee regulation may have decreased their use. Our analyses also revealed factors of a state's population that were associated with severe sanctions being levied. Rural, conservative states that were relatively punitive had higher severe sanctioning rates than states that were more urban, politically moderate or liberal and less punitive (see Table 3).
CONCLUSIONS AND DISCUSSION
The National Practitioner Data Bank, which was designed to minimize the chances that a negligent physician would be able to continue practicing, seems to have had an impact. Its implementation was associated with statistically significant increases in the rates of physicians being placed on probation, or having their licenses revoked or suspended. State licensing boards, on average, levied serious sanctions against 2.40 to 3.20 physicians out of every 1000 licensed medical doctors during the years prior to the implementation of the NPDB. That rate rose dramatically to between 3.92 and 4.77 during the years immediately following full implementation of the Data Bank.
It seems unlikely that the rise in the serious sanctioning rate was due to sudden increases in inappropriate behavior by physicians. A more likely explanation for the sharp increase following the establishment of the NPDB is that licensing boards were now taking actions against physicians who previously were able to avoid detection or who received milder punishments that allowed them to continue practicing. The use of other prejudicial actions increased each year prior to the establishment of the Bank, but decreased each year thereafter. The suggestion is that some cases, which prior to the establishment of the NPDB would have resulted in mild punishments, were treated more severely following the establishment of the Data Bank. Required reports from hospitals, insurance companies, professional societies, healthcare institutions and state licensing boards to the Data Bank made it easier for individual boards to learn of misdeeds by physicians and take corrective action. This conclusion is supported by a survey of organizations that query the NPDB; more than five percent of credentialing decisions were altered because of the information the NPDB provided (Waters et al., 2003b).
The NPDB, similar to any public policy, may not be perfect and there are areas of concern. The Data Bank, for example, relies on reports from hospital committees and there is evidence that physicians do not support the Bank and take steps to avoid reporting requirements (Baldwin et al., 1999; Fellmeth & Papageorge, 2005; Neighbor et al, 1997; HHS, 1995). There is some indication from hospital administrators that they are generally satisfied with "the accuracy of reports, the timeliness of responses to queries, and the completeness of reports," and have some dissatisfaction with "the fee for querying, effect on staff workload, billing procedures, and clarity of requirements and procedures (Neighbor et al, 1997, p. 664)." These matters, however, do not seem serious enough to warrant hospital committees avoiding the reporting requirements of the law.
It may not necessarily be the activities of the Data Bank that physicians oppose; the NPDB is merely a clearinghouse for information. It may be that what irks the hospital committees is that which the NPDB represents: a questioning of their ability to police themselves and increased government regulation of the profession. No longer is the medical profession an autonomous group and its members resent the reduction in their influence. This issue will likely take on a greater presence in the public dialogue as the U.S. moves closer to universal healthcare and physicians confront new government schemes to control their behavior.
Allison, P. D. (1994). Using panel data to estimate the effects of events. Sociological Methods and Research, 23:174-99.
American Hospital Association. (1984, 1985, 1987, 1988, 1989, 1990, 1992, 1993). Annual survey. Chicago: American Hospital Association.
ARDA. The Association of Religion Archives. (n.d.). Churches and church membership in the United States. Accessed March 17, 2006, http://www.thearda.com/Archive/Files/Descriptions /CMS90ST.asp.
ATRA. The American Tort Reform Association. (1995). State and federal reforms. Accessed June 25, 2006, http://www.atra.org/reforms/.
Baldwin, L. M., Hart, L. G., Oshel, R. E., Fordyce, M. A., Cohen, R., & Rosenblatt, R. A. (1999). Hospital peer review and the National Practitioner Data Bank: Clinical privileges action reports. Journal of the American Medical Association, 282, 349-355.
Bannon, N. K. (1989). Tort reform compendium. Chicago: American Medical Association.
Bovbjerg, R. R., & Petronis, K. R. (1994). The relationship between physicians' malpractice claims history and later claims. Does the past predict the future? Journal of the American Medical Association, 272, 1421-6.
Brennan, T. A. (1998). The role of regulation in quality improvement. The Milbank Quarterly, 76, 709-731.
Brown, J.G. (1995). Hospital reporting to the National Practitioner Data Bank, Rockville, Md: Office of General, U.S. Department of Health and Human Services. Publication OEI-01-94-00050.
Bullock, H. (1961). Significance of the race factor in the length of prison sentences. Journal of Criminal Law, Criminology, and Police Science, 52, 411-17.
Cane, P. (1982). Justice and justifications for tort liability, Oxford Journal of Legal Studies, 2, 30-62.
Center for Justice and Democracy. (n.d.). Tort Reforms' are unconstitutional. Accessed March 2007, http://www. centerj d.org/free/mythbustersfree/MB_Unconstitutional.htm..
Danzon, P. M. (1984). The frequency and severity of medical malpractice claims. Journal of Law & Economics, 27, 115.
Davis, T. L., Severy, L. J., & Kraus, S. J. (1993). Predictors of sentencing decisions: The beliefs, personality variables, and demographic factors of juvenile justice personnel. Journal of Applied Social Psychology, 23, 451-477.
Federation of State Medical Boards. (n.d.). Accessed March 2007, http://www.fsmb.org/fpdc_basummaryarchive.html.
Federation of Medical Regulatory Authorities of Canada and the Federation of State Medical Boards. (2008) Medical regulatory authorities and the quality of medical services in Canada and the United States. Milbank Memorial Fund. Accessed June, 2008, http://www.milbank.org/reports/0806MedServicesC anada/0806MedServicesCanada.html
Fellmeth, J. D'Angelo., & Papageorge, T. A. (2005). Final report Medical Board of California enforcement program monitor. University of San Diego School of Law: Center for Public Interest Law. Accessed January 29, 2006, http://www.medbd.ca.gov/publications/enforcement _report.html.
Freidson, E. (1975). Doctoring together: A study of professional social control. New York: Elsevier.
Galusha, B. L., & Breaden, D. G. (1989). Official 1987 Federation summary of reported disciplinary action. Federation Bulletin, 275.
Grant, D., & Alfred, K. C. (2007). Sanctions and recidivism: an evaluation of physician discipline by state medical boards. Journal of Health Politics, Policy and Law, 32, 867-885.
Guirguis-Blake, J., Fryer, G. E., Phillips, R. L., Jr., Szabat, R., & Green, L. A. (2006). The U.S. medical liability system: Evidence for legislative reform. Annals of Family Medicine, 4, 24-246.
Hill, G., & Harrison, P. (2005). Sentenced prisoners under state or federal jurisdiction. Bureau of Justice Statistics. Accessed January 17, 2006, www. ojp.usdoj. gov/bjs/data/corpop02.csv.
Holmes, O. W., Jr., (1873). The theory of torts. The American Law Review, 7, 652-663.
Holmes, O. W., Jr., (2005). The common law. New Brunswick: Transaction Publishers. (Original work published 1881).
House Report. (1986). H. Rep. No.903, 99th Cong., 2d Sess. 3. Reprinted 1986 U.S. Code Cong & Admin. News 6384, 6385.
Hsiao, C. (1986). Analysis of panel data. Cambridge: Cambridge University Press.
Idleman, S. C. (1993). The role of religious values in judicial decision-making. Indiana Law Journal, 68, 433-4.
Jesilow, P., Pontell, H., & Geis, G. (1993). Prescription for profit: How doctors defraud Medicaid. Berkeley: University of California Press.
Jesilow, P. & Ohlander, J. (in press). The impact of tort reforms on the sanctioning of physicians by state licensing boards. Journal of Empirical Legal Studies.
Johnson, D. R. (1995). Alternative methods for the quantitative analysis of panel data in family research: Pooled time-series models. Journal of Marriage and the Family, 57, 1065-1077.
King, M. B., & Hunt, R. A. (1984). Measuring religion: A comment on Morgan. Social Forces, 62, 1087-8.
Kusserow, R.P., Handley, E.A., & Yessian M.R. (1987). An overview of state medical discipline. Journal of the American Medical Association. 257: 820-824.
Lavenant, C., Hayward, C., & Jesilow, P. (2002). Tort reform and physician sanctioning. Law and Policy, 24, 1-16.
Maddala, G. S. (1987). Limited dependent variable models using panel data. The Journal of Human Resources, 22, 307-338.
Meyer, J., & Jesilow, P. (1997). Doing Justice in the people's court: Sentencing by municipal court judges. Albany: State University of New York Press.
Myers, M. A. (1987). Economic inequality and discrimination in sentencing. Social Forces, 65, 746-766.
Nagel, S. S. (1961). Political party affiliation and judges' decisions. The American Political Science Review, 55, 843-850.
Neighbor, W. E., Baldwin, L. M., West, P. A., & Hart, L. G. (1997). Rural hospitals: Experience with the National Practitioner Data Bank. American Journal of Public Health, 87, 663-666.
Oshel, R. E., Croft, T., & Rodak, J., Jr., (1995). The National Practitioner Data Bank: The first 4 years. Public Health Reports, 110, 383-394
Peters, E.N. (1972) Practical versus impractical peer review, Medical Care. 10: 516-21.
Petersen, T. (1993). Recent advances in longitudinal methodology. Annual Review of Sociology, 19, 425454.
Randolph, L., Seidman, B., & Pasko, T. (1995). Physician characteristics and distribution in the United States. Chicago, IL: American Medical Association.
Reuter, P. & Bushway, S. D. (2007). Revisiting incapacitation: Can we generate new estimates? Journal of Quantitative Criminology, 23, 259-265.
Richardson, F. (1972). Peer review of medical care. Medical Care 10, 29-39.
Robertson, W. O. (2001). National Practitioner Data Bank--still not effective after all these years? The Western Journal of Medicine, 174, 149-150.
Ryzen, E. (1992). The National Practitioner Data Bank: Problems and reforms. The Journal of Legal Medicine, 13, 409-462.
Satiani, B. (2004). National Practitioner Data Bank. Journal of the American College of Surgeons, 199, 981-986.
Smits, H. (1982) PSROs: Origins, directions, and changing assumptions. Bulletin of the New York Academy of Medicine, 58, 11-18.
Studdert, D. M., Mello, M. M., Gawande, A. A., Gandhi, T.K., Kachalia A., Yoon, C., et al. (2006). Claims, errors, and compensation payments in medical malpractice litigation. New England Journal of Medicine, 354, 2024-33.
Swazey, J.P. (1991) Are physicians a "delinquent community"?: Issues in professional competence, conduct, and self-regulation. Journal of Business Ethics. 10, 581-590.
Tate, C. N. (1981). Personal attribute models of the voting behavior of U.S. Supreme Court Justices: Liberalism in civil liberties and economics decisions, 1946-1978. The American Political Science Review, 75, 355-367.
Todd, J. S. (1995). National Practitioner Data Bank; Just numbers or knowledge? Public Health Reports, 110, 377-378.
U.S. Bureau of the Census. (1990). Statistical abstract of the United States. Washington, DC: Government Printing Office.
U.S. Department of Health and Human Services. (1995). National Practitioner Data Bank: User satisfaction with reporting and querying and usefulness of disclosure information for decision making, 19921994. Rockville, Md: Health Resources and Services Administration.
U.S. Department of Health and Human Services. (2007). Fiscal year 2008 justification of estimates for appropriations committees. Accessed November 14, 2007, http://www.hrsa.gov/about/budgetjustification08/nat ionalpractitionerdatabank.htm.
Wallis, C. (1986). Weeding out the incompetents. Time, 26 May, 57-58 .
Waters, T. M., Studdert, D, M., Brennan, T. A., Thomas, E. J., Almagor, O., Mancewicz, M., et al. (2003a). Impact of the National Practitioner Data Bank on resolution of malpractice claims. Inquiry, 40, 283294.
Waters, T. M., Parsons, J., Warnecke, R. B., Almagor, O., & Budetti, P. P. (2003b) How useful is the information provided by the National Practitioner Data Bank? Journal on Quality and Safety, 29, 416-424.
Waters, T. M., Warnecke, R. B., Parsons, J., Almagor, O., & Budetti, P. P. (2006). The role of the National Practitioner Data Bank in the credentialing process. American Journal of Medical Quality, 21, 30-39.
Weber, M. (1954). Max Weber on law in economy and society. New York, NY: Simon and Schuster.
Yessian, M.R., Greenleaf, J.M., Hereford, R.W., Han, N.J., & Levine, A. (2001). Managed care organization nonreporting to the National Practitioner Data Bank: A signal for broader concern. Rockville, Md: Office of Inspector General, U.S. Department of Health and Human Services. Publication OEI-01 99-00690.
University of California, Irvine
Table 1. Descriptive Statistics for State Characteristics (n = 50 jurisdictions * 10 years = 500 cases, missing = 0) Dependent variables Mean SD Range SSR: Rate of serious sanctions 3.5 2.5 0.0 to 18.0 per 1,000 physicians OSR: Rate of other sanctions 2.4 2.3 0.0 to 14.1 per 1,000 physicians Independent variables Urban population: % of state's 68.2 14.5 32.2 to 92.7 population that is urban (1990) Political ideology: liberal=0, 1.2 0.7 0 to 2 moderate=1, conservative=2 Religion: # of church adherents 54.6 12.8 32 to 80 per 100 state population Punitive ideology: # prisoners 2.5 1.2 0.5 to 7.0 under state or federal jurisdiction per 1000 state population Population health: # of in-patient 1318.2 1073.1 129.3 hospital days per 1,000 population to 16,520.1 Table 2. Bivariate Correlations between State Characteristics and Sanctioning Rates 1 2 3 4 5 6 7 Serious 1.00 Sanction Rate (1) Other +.24 1.00 Prejudicial Action Rate (2) Population -.02 -.02 1.00 health (3) Urban -.16 -.05 -.02 1.00 population (4) Political +.06 +.20 -.04 -.16 1.00 ideology (5) Religiosity -.00 +.08 +.12 -.02 +.15 1.00 (6) Punitive +.23 -.06 -.00 +.24 +.29 -.10 1.00 ideology (7) Table 3 Incidence-Rate Ratios for Random Effects Negative Binomial Regression of Serious Sanctions (n=500) Model 1 Model 2 Ad Damdum Clause 0.88 0.87 Arbitration 0.77 ** 0.77 ** Attorney Fee Regulation 0.79 * 0.78 ** Collateral Source Rule 0.92 0.92 Frivolous Law Suit 1.13 * 1.14 * Joint & Several Liability 1.16 * 1.10 Reimbursement Limits 1.12 1.14 Patient Comp. Board 1.19 1.18 Periodic Payments 0.96 0.99 Pre-trial Screening 0.85 0.82 Urban Population 0 98 *** 0 98 *** Political Ideology 1.26 * 1.24 * Religiosity 0.99 * 0.99 * Punitive Ideology 1.08 * 1 12 *** Population Health 1.00 1.00 National Practitioner Data Bank 1 27 *** 1986 (1985 reference) 1.10 1987 1.17 * 1988 1.24 ** 1989 1.19 * 1990 1.04 1991 1 42 *** 1992 140 *** 1993 1 39 *** 1994 1.62 *** Estimated overdispersion 1.88, 1.81, 2.74 coefficients: ln_r, ln_s 2.76 Likelihood ratio test of 267.25 (.000) 265.62 (.000) overdispersion=0: Chi-squared statistic (p-value) Full model LL -1810.64 -1820.39 (constant LL=-2138.49) McFadden's pseudo [R.sup.2] .15 Table 4. Incidence-Rate Ratios for Random Effects Negative Binomial Regression of Other Prejudicial Actions (n=500) Model 1 Model 2 Ad Damdum Clause 0.84 0.88 Arbitration 0.85 0.90 Attorney Fee Regulation 0.85 1.00 Collateral Source Rule 1.06 1.10 Frivolous Law Suit 0.91 0.99 Joint & Several Liability 1.09 1.22 * Reimbursement Limits 1.14 1.21 * Patient Comp. Board 1.38 * 1.42 * Periodic Payments 1.09 1.14 Pretrial Screening 0.93 1.03 Urban Population 0 98 *** 0 97 *** Political Ideology 1.22 1.17 Religiosity 1.00 1.00 Punitive Ideology 0.90 1.01 Population Health 0.99 1.00 National Practitioner Data 0.47 *** Bank 1986 (1985 reference) 1.05 1987 1.23 * 1988 1.35 ** 1989 1 53 *** 1990 1 78 *** 1991 0.71 * 1992 0.90 1993 0.79 1994 0.78 Estimated overdispersion 1.26, 1.16, 2.30 coefficients: ln_r, ln_s 2.33 Likelihood ratio test of 324.22 (.000) 304.85 (.000) overdispersion=0: Chi-squared statistic (p-value) Full model LL -1714.68 -1731.56 (constant LL=-2060.13) McFadden's pseudo [R.sup.2] .17 Table 5. Average State Serious Sanctioning Rate (SSR) and Other Prejudicial Actions (OPA) Rate by Year 1985 1986 1987 1988 1989 SSR 2.4 2.5 3.1 3.3 3.1 OPA 2.2 2.1 2.4 3.2 3.4 1990 1991 1992 1993 1994 SSR 2.9 4.6 4.4 4.0 4.9 OPA 3.8 1.7 1.9 1.6 1.4
|Gale Copyright:||Copyright 2010 Gale, Cengage Learning. All rights reserved.|